-------
71
CS! ^-< — « u"3 CO CD CSJ
--
CO "•*• CO •
--
CD —1 —i CD —t CD CD ro CM CD —> —t CD ro —i
CDCDCDCDCDCDCDCD CD CD CD* CD* CD CD CD
LO CD LO CD CD
tOC---W
— .CSJ— «
O->CSl
_ -, CSI
CM CSJ
> I*-. CM ro •«-
CSI CM —i CM
£ 3
CD LU
; at' "~
ill
CD CD O11
f*^ CD CM
-*S-i-x,LO
— i CM CM
_ _
COCZCD
,
— -i — ~t CSJ •— « CSI
»o
h*5
i cr-- r
CD
CS'^OLO —
- - -
.
, CM CD o--
ro LO r-v. ro co o> so
—. ro — ro
CD CD' •—• -—- -—- -—- -—-
CD CD LO CD CD CO O1-
. ,
• -V CD CM C^ CSJ -C-
' C*- CO CO
o-> co ro co
* "
— i — — 4 CM
co co co o-- —i r^. CD co —• CM o~- co ro co
— —•—t CM — CM CMCMCSJ
: g ^
a
- ^o ^c co
f C"~ — « IO
-
CD—. — CDCD — CD* CD* CD — CDCD—I •
ro •«• r^ co CM i-o csj
—H csj -w CD -«r co —
•w -«9- co LO LO r^. co
—' CD CD CD CD CD CD
> CD CD O- CO
— — CD CD* — — CD* CD
-
— — .CSI —.CM
CD CSJ r-o CD r-^ CM CCt ro
CO CO f-O '
CMCM— tf
co co r-* CM i
Si
sOCOCf-C^CSl
a
CMCM—iCDLOLOCDLOCDO
—, — —, CM — — CM —i CM —4
f*- CD CM CM I-O
r^. *a —« CD LO
LO os o- r-*. LO
CD CD •—. CD
CD CD CD CD
CDCDCDCD-OCDCDCDCDCDCDCDCD
CDO--CD Cf. u*3 CO CSj
.^--sC'Cr- CS) C*- -*J- x£.
I^^C^-CC LOCS--3-— H
ro cf. LO • csjc">-«-rT>
CS) hOCSJCSI
^LO-^r^-w-'CD'vO — r^.CMtOsOOs'^.COrO CC-O-O-CDCO-^-r^,
— — - - CM —< —H —^ CD ro ro —i CM CD fo
co — ro-WNorocD^s-CMrocNj rocoCD-*-CDo-co —iNOONr^cMror-v,
£S^i?S;'~1S^vC'!i20^*:^^C' CoCo^NOC^Scor-tJ^CMCDLONO —
CD ON ON CO —-. CD ON ro r*. f*-. CD CD CM «—i »O CM LO O- fO ro CD CO U"3 CM
CM— —. CM CM —• CM i—i —i —'"ocsirocMCMCsjro , rocsjcsiro
= = s
ro —« CD CO CD ^9- LO «—' Csj CM CM M5 NO CO N£> -^- -^- -^ s^ to r-^
S ^5 S iS 1^ trt S « "* "^ "* ^* L/^ crs *~c' **** f*>* ^^ ^
co-NOCDr-.csiLO(— sor-*co
I I 1 I I I I I I I
NO »O
CO CD
CO CM f"5
S s -
-sO
CO I
..
CD — — ON LO lO CD fO O- CSJ Os CJ"' "-^ CO >O -
- "^ • S R: — ev'~ -
WO —, CM i
> -*T LO CM NO CD r^. CM CfN LO
i -V- •«- C«J Cf- C^ LO CO CO ON
CO •«• ro —-t CM CM LO UO LO
. QS co NO LO so trt —* LO -v-
CM —1 tO 1 CM
^0. ar :
-r LU
: CM CM -^ -w
—-i CM ro -^ to so r^. co o~> CD —« <
B
£5^3
3E -^ *—» 3T
S^ •—t uj a: a= ac
SC r-sj 31 LU LU Uj
•^ so uj :c''Z: Z: ac
CD —I 03 Q- LU UJ UJ
££££§!!
Ssiiiis
-
CO 3C= OS U- CD CD
. co os CD — < csj
ro -w LO so r^
CO
LO d LU
-------
72
*» SO l^". tO
«—i CM CC CM «-H OS CD —' CD CD CD CD
cc= CM '
CO 3
r*. CD e*.
£S K
t irt ez>
lO t/? *O iO CD CD CD CD tO
NO t/? rC to to c2 CM CM »••; -i
£ ::
CD CD CD LO CD
_ . CD CD >O
CM so —4 -W NO
< IO LO vO CD fcO LO CD
i LO VO tO CD CD tO CD
CD tO LO tO lO I
—H <—i SO NO CM i
CM —' CO ^- £?•• SO tO SO '^
so -w NO NO NO CD r*- *—i •«•
KtO
—i
—* CM
g"""~ "CD"
•%o
CD tO CD LO LO
1=0 ~£> CM NO —
CD LO LO CD CD
f-» vO I*O—* Cf-
|sO ^T O-- CM CD
i «, KM CM fO —t — •
I £3 —*. CD —- —-• O ^^ —— •—• <
LO lO
—i C>-. CM
vO so IO
LO tO CD tO tO
CM ro iO
CD to r^
1o "CD" to
ro -w ir>
^o CM c?* CM to CM «—'
r-o LO i— to co 10 CD
CD CD CD CD tO CD CD
CM CO -v —'
f-^ Crv sO
s ss
CM C^-
—i CM CO
£g
2 — — C^ — — ^° '
iO tO cy- CD -~ Cf« CM CM —* CT»
er't CD CD* CD* « -^ CD CD* CD —^ CD
gg.ggg
Sfs, i— r», p-^ LO CM r^.
CD CD CD CM so r^ CM
>•••-
. CD CD ^O CD
CD SO CD
CO O^
CO • CM
C3 CD CD CD —«
—t—H —.CM —< CM
o-, r*, cc co
CSCDCDCD
8
fO CS CM
r^ — so
CMCO-W
O)
o
sa
R S § 3 S 5 R: S S E? S= ° S 3 §IS
fOrocM rocM CM — — — —*CMfoCM—*
O\ I^« UO O* CO CT1- *O CM •«• sC> C^ ^ —' '*O LO O~-
ro _ co to 1/5 co CM CM ^ r-^ +o o*- r«o CM ^ co
CD r^o CM i/^ co i
CM-«J- i— NO c»-- *o •«-•*• CM ec* (»*> CP-
—H (-^ LOiOCO^OrocDuO-WCO'^.
*
CM MI ho CM
C=COCM>OCD -«3-—«LO—^
p-^ esj 10 ^- r--. ^g- —* ^ so
^ _-*__«_4 -HCMCM—<
—* C^ r>, ^- co NO
. O1* •so r^. LO
—- —t CM —< ro ^- ^ '
o_ a_ rs ca_ z^ :=> =3
CO CO CM I
CD to *O r
i—t —4 CM
fO LO CO —<
M3 «*^ CO CO
CM CM CM 1
*,
—« C"d CM CM —< CM CM
I CM CM CM
i^^y^^H^^^K^9^M?^^^^^^^^p^^^^^^cI^JvOfT)Crj?
LocOl^C^sD^CDioCD^^-iCM^-lpwO-VuO
. —* r*o —H co ^ to 0-* r~* LO i/^ —•+ ~-~* c^
s : s
.s.OLOLOrOCDCD'^-'WCMCDONCMsOC^sC-WvOlOCDeO
i CO —-•* OO CM O NO ^ lO CD CD 0s! C'"' NO C*~- tO CM LO C^- —
I— NO h-7 Os f^O —< —t SO —^ CO — ""•
r^. CM so —•*
LO •—< '^^ —t
I —* LO CD C3 CO
_--..jr^cocMtocDi
—or^« —4 ov -^
CM I»O NO
:: S S i S §
*— n: ze
CD Q_l—
s
! LU UJ 5
^ S ^ LLJ ^
z
fei LU
I U-l «—'
:S 5
OC CD —I
C_3 5C O
g % :
co eo
to >— ac >— ,
S -« OC *t OS C>
eo re LLJ ae CD o=
CD t— o_ *o eo u.
a: nr »—• ee LU -i
o_ a- a. o- o= to
-------
73
CD CD CD CD CD CD
CO sO CSJ PO CO
—, __j CSJ —* Cs!
CD CD CD CD CD
CSJ UO CF- SO CD '^ SO
CD CD CD CD CD CD CD
CD CD CD O CD CD CD
CD CD tO
25 3
-^ u-> PO
SS • ~
LO UT> W> IO WD
|>O ro -w PO cs;
•*9- Cf' ^ >O PO
CS/ 0-- •*!- CO CO
O- CD —
CD u-> C? CD uo"'
PO CS) —i CC r-.
CD CD ul CD
. u7> —i C*- —<
UT) CD CD U1
CD uT. CD CD u-j CD
CD PO CD CD ^C CD
— CXJ CNi
r--J CSI O- •«• —< i-O CD
•*O- CSI PO -^- P"^ PO I"*.
— i CT-
"
— CD CD C2 O
CD CD CD CD C3
CD
-o
^ .
CO CSJ CO
^-5 CD CSJ CO PO ^O —«
O^ UT> Cfi SO ^C* •?•• S&
CD CO sd CD CD •«• »O
~-COCOLOCOsOsOCi--Lrt-
CD C1"- -O CO CD CSi r~~. CSl •
r^j —i csi —' csi PO csi csr
•^_ CO CSJ CT' IO CP -y- -^ i-^> CD -*O- CO
["""t^PO PO CSJ CSI CSj CSl —1
CO CO CD PO CO
fcO CSJ O->
»-
£ '£
•<: o_
cr> CD •— «
_J CD UJ
ZD a: —i
• 3^ —' LJ.
, S -r =
>— l_k. e*
oc —i o:
•- PO
l-O _J L4-J *
-cjL-cj-ttzSVS^csi^ar^^ujuj-^
- --c-«caeijj_iirt_jLij-—• uj •« =c e=» ^ ag S"
:o=CD ,=.=><= -u-ZfOtljceP--.!^^
*-* (U O
o> a*
in >-
*- W CD
-
I CSJ CSJ CSI CSI
-------
CO
a>
•a
II
II
II
II
II
II
II
II
II
II
II
CO II
HH II
CO II
S* II
Ha II
< II
Z H
•< II
II
CO II
E ii
I it
o n
O li
n
0 II
H II
II
Cd II
i-a II
05 II
< II
Z II
cd ii
E H
•< II
II
EH H
0 II
Z II
II
>< II
Ha II
E-< II
Z H
< M
0* II
< II
OH M
OH H
< H
II
CO II
O II
Z II
£3 II
0 II
OH II
E ii
O II
O H
ll
Cd H
Ha II
M II
EH II
< II
Ha II
0 II
> II
M II
E n
Cd n
CO II
n
n
n
ll
ll
n
n
n
ii
n
n
n
n
n
n
n
n
n
n
n
u
^~> II
• CO II
CQ <— n
M ca ii
1-3 Cd II
CQ II
CO £ II
£ O II
SE ll
II
II
II
II
II
II
II
II
II
II
ca n
< CO 11
OS CQ II
O £ ll
OS » II
Z II
n
u
n
u
n
u
u
u
u
u
• ii
O II
Z II
u
CO II
< II
U M
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
Cd II
U II
Z II
< II
EH ||
CO II
CQ II
C3 II
CO II
II
II
II
II
II
II
II
II
II
II
II
II
• II
0 II
Z II
in
co
CM
CO ' PJ CM >— 1
COCM
COCOO09COCOO)COOrHO«<
^^ £^ !••{ ^^f ^^ ^ft p^ £^ QQ Q^ C^J QO
i-H CO >-l CM "-H CM CO CM
r-HCOC^'-HOCOi-HO
OCOCMOCO'-ICO'—I
CO inCOCOCOt>t>C^Cnt-OO«S'CO'-IO
O CM'-Hi-HOr-H'-HrHOCMCMi-HOr-lCOO
I I I I I I I I I I I I I ! I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I I I I I I I I I I I I I I
^O f™H ^*H ^f* C^ lA CO CO C^ CO ^tf* C^ C^ ^""^ ^^ p""t lO C73 'd* 1^5 CO CO ^^ C75 CO C^ 00 ^H
^M CO ^3 CD IO Oi CO CO C^ O5 tf5 1^5 ^^ C3 ^^ O5 tf5 U) 00 f^^ C5 00 00 00 Oj l«O ^^ 00
i-HrHOOCO n >-t t^rH I-COCO"-! CD"-H
(». CO CO O5
Cd
z
1-4
Cd
z
Cd
z
o
Cd
z
Cd
z
Cd
35
OH
I
CM
Cd
Z
Cd
ca
<
OS
o
>•
Cd
CH
< < Q
0 K O
35 O
Ha M O
>• X M
Z EH Z
Cd HH
< Q
I ^^
H3 Z
> <
B a
Cd O
£ co
HH O
O K
I EH
Z I-H
>z
Z !
CO I
HH O
CQ OS Cd
Cd EH Z
Cd
M
Z
<
B
< EH
< Cd Cd
Cd K OH
03 S d
>• B B <
B EH EH CO
H Cd Cd OS
Cd
EH
•<
B
O.
CO
o
OH
O
OH
>-
OH
O
OS Cd Z E HH | | i O
O H-1 I <
i-a >• z
0, Z Z Z M
Cd I I I B
Z
O
Cx
Cd
EH
B
OH
CO
o
B
OH
CQ
On
HH EH i-a O O B
— Cd" Z
CO HH
Cd OS
OS <
O CK
1-^ r—i ^^ toi^ ^^ ^* -—< — ^ —^ ^f fc • fc •
Ezcdos i >-HH -EHCOWW
1 |
I I I I
O < Cd B OS OB
CO t> 00
CM CM CM
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
M
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
74
U
CO
tn
m
oo
S
CO
CD
03
CO
c
o
•H
n
u
CO
03
CJ
0)
n
CO
CQ
CO
•P
CO
ca
CO
k,
o
o
CO
CO
n
co
£
co
o
OS
CO
z
I
B
HH
Z
OH
Cd
-------
75
Slide 19
SEMIVOLATILE COMPOUNDS FOR WHICH FURTHER GC-MS DATA
REDUCTION AND EVALUATION ARE REQUIRED
NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
SUBSTANCE
ACRYLAMIDE
3-AMINO-9-ETHYLCARBAZOLE
l-AMINO-2-METHYLANTHRAQUINONE
5-(AMINOMETHYL)-3-ISOXAZOLOL
ANILINE
AURAMINE
AZINOPHOS-ETHYL
4-CHLORO-l , 3-PHENYLENEDIAMINE
l-(2-CHLOROPHENYL)THIOUREA
p-CRESIDINE
CUPFERRON
CYCLOPHOSPHAMIDE
DEMETON
2,4-DIAMINOANISOLE SULFATE
1,2:5,6-DIBENZACRIDINE
DiBENZo(A, DPYRENE
DIETHYLSTILBESTROL
DIHYDROSAFROLE
DIISOPROPYL FLUOROPHOSPHATE
EPINEPHRINE
2-FLUOROACETAMIDE
HEXAETHYL TETRAPHOSPHATE
LASIOCARPINE
MALEIC HYDRAZIDE
METHYL METHANESULFONATE
METHYLTHIOURACIL
1 , 5-NAPHTHALENEDIAMINE
l-NAPHTHYL-2-THIOUREA
NICLOS AMIDE
NITROGEN MUSTARD
N-NITROSODIETHANOLAMINE
p-NITROSODIPHENYLAMINE
N-NITROSOMORPHOLINE
OXYDEMETON-METHYL
PHENAZOPYRIDINE HYDROCHLORIDE
1,2-PHENYLENEDIAMINE
1 , 3-PHENYLENEDIAMINE
1,3-PROPANE SULTONE
4,4'-THIODIANILINE
0,0,0-TRIETHYL PHOSPHOROTHIOATE
2,4,5-TRIMETHYLANILINE
1,3, 5-TRINITROBENZENE
CAS NO.
79-06-1
132-32-1
82-28-0
2763-96-4
62-53-3
492-80-8
2642-71-9
5131-60-2
5344-82-1
120-71-8
135-20-6
50-18-0
8065-48-3
39156-41-7
226-36-8
189-55-9
56-53-1
56312-13-1
55-91-4
51-43-4
640-19-7
757-58-4
303-34-4
123-33-1
56-04-2
2243-62-1
86-88-4
50-65-7
51-75-2
1116-54-7
156-10-5
59-89-2
301-12-2
136-40-3
95-54-5
108-45-2
. 1120-71-4
139-65-1
126-68-1
137-17-7
99-35-4
RCRA
NUMBER
U253
U265
P007
U014
P150
U305
P026
U262
U290
U058
P155
U307
U064
U086
U090
P043
P042
P057
P062
U143
U164
U298
P072
U321
P132
U173
U287
P157
U320
U193
U258
U259
U234
MS LIB.
NUMBER(a)
204
840
31487
5213
13317
34981
2041
16231
15632
1744
3126
a) EPA/NIH-NSRDS Mass Spectral Data Base,1983 version; 39,763 mass spectra
-------
CO
QJ
"o.
00
ai
76
ai
•a
QJ
00 li-
CU I ro
4J C
ro,
c 1-
r—
01 3
r— E CO
-C X r—
^° *«
ro
S—
O
•—
JT
U
I
ro o
j= s.
o
QJ •—
o x:
s- o
O T-
r— "O
x: i
o n
i -
CM .—<
4J 4J •.-
CU QJ !_
f—' O O TTJ
5°°!*!
QJ x: x: i. QJ
E o o o o
O I I r- ,0
S- CM CM JET o
O i i o E
I— CO 00 •!— O
x: .1- .,- CL t.
O X) X) QJ _O
QJ
Q.
O
1-
a.
O
S-
o
E S 4J
I =J C
LO 4-> ro
O- 3
o > s-
O O
QJ _J
1- E .—
•^-. ro
6-S Co :3
O ra cr
• 0}
o o
O <4-
o
i — TD
ro C
C CD
QJ t±
(/)
00 4->
LU re
CO
o
a.
e
o
o
QJ
73 ••
i.
O O
OJ T3
3 QJ
•O >
i.
CO QJ
4-> 00
QJ
O O
O t—
O- D_
QJ
XT
4-> Ql
QJ C
ro
C OJ
••- o
> S_
o
jr o
4J ro
QJ 4->
o c
S- OJ
O CL
o
CU
•a
O
co
g
o:
o
CO
LU
00
QJ
4-* *• ' *
>>TTJ
i — QJ
ro z>
c c
•a: T-
4->
IO C
"3- 0
CO 0
CO
00
cu
4-^
>}
^—
ro
c.
^c
*>o
^"
CO
3
CO
CU QJ
TTJ C
•r— CU
QJ S- 3
QJ i — O
M -e~ 4->
COO
QJ S_
XI i— O
0 >,.-
E N x:
o c o
1- QJ 1
X) X! CM
S_
QJ
X:
4-J
CU
^~
>1
^*
4->
QJ QJ
•a E QJ
•r— "O
S- 1 •!—
O ^> E
•— -c o
-C 4-> S-
O QJ XJ
E
i — O i —
>,$->,
-E: 0 -c
4—) r~~ 4—^
QJ -C QJ
E 0 E
1-
QJ
XI
4J
Q)
*• — *
r—
>)
CU QJ
O C
S- CU
O- N
O C
CO QJ
•r- X>
O O
S- S-
o o
J- f
o o
1 *l~~
CM -a
* — t
co co
•r- •*
X) ~->
QJ
C
03
XT
4->
QJ
E
0
S-
QJ 0
"O 3
•r— r—
t- *4—
0 ••-
r— -a
xz o
0 S-
o
^— ^*
>^ >£Z
C O
•1 — 'f—
> TD
QJ
C
ro
4->
Q)
£
QJ^-— QJ
C >, C QJ
QJ X QJ T5
NJ O M >i
C -C C XI
QJ 4-> QJ QJ
XI QJ XI TT*
O O O i —
S_ I- S- ro
O O O 4->
x: x: x: o
o o o ro
•i- 1 -r- O
-a CM TTJ s-
1 I O
CM CO ^^ r~~
f< tf— ft e~
t— 1 X3 t— 1 O
CU
C
ro QJ
-ET C
QJ 4-> QJ
TTJ QJ i —
•r- E >,
s- o x:
QJ O S- 4->
"O ^~ O QJ
•r- X: =3 O
!- O r— S-
O M- 0
i— QJ O .—
XT C i- XI
O QJ O O
i — r— pt—
r— S-,X: T3
>>x: o i
x: 4-> •!- t-i
4-> QJ S- •>
QJ E 4-> ^-<
QJ
QJ
r—
^^
f~
4-*
QJ
QJ O
£1 1-
ro o
XT ^~
4J J=.
QJ O
O ••-
s- -o
0 I E
^- CM S-
x: •< o
O t— 1 «-
•r- 1 O
T3 00 S-
1 C O
r—troi —
• s- x:
•—I 4-> O
QJ
c
ro
QJ x:
C 4->
ro QJ
x: o
4-> QJ S-
QJ C o
o ro r—
s- x: x:
O 4-> O
•— QJ -r-
x: E i-
O O 4->
••- E 1
T3 O r~J
1 S- «
CM XI «— I
ft .r— »>
1 — 1 "O « — 1
OO
c
o
*l»
4_)
•r-
T3
TQ
X
•^>
7rj
c
QJ
Q.
Q.
<:
QJ
•o
• r-
J_
O
j^;
O
ro
i-
4_>
QJ
4_>
C
O
c
ro
o
QJ
ro QJ
CL-O
o •<—
QJ QJ O-r-
i — C i — O Z3
- •""• QJ O S— OO
S*- QJ 4-> C O
4-> "O 3 ro QJ i — i —
•r- -i- x> a. c x: >,
c i — c E'orooxr
O QJO •!— OC-JS_xri4->
••-QJ-OC S- ^ICL4JCOQJ
CLTTJ •«— ro "OQJX)OOQJ 1 O
O ••— S CjC'i— i-t-OOS-
s-"ao4->cx:o-ooos- EO
o •<— x: o i- i-ai ojxrxit— s_x:
!- Ol-O-OOCCJUJCXlO
o i — o o i — ro QJ •(—•«— o «r— i
t— >>•— r— s-x: or— -OTD ro-ocM
x: x: >,x: o o E >i i i *-> i i
O4->i — Oi — •!— OXT^I-CO CCM co
1 QJ I — IX:CLS_4-> « «»QJ *«T—
COEroCMOQJXlQJ'— Ir-lCLt— 1X1
S- QJ QJ
QJ C C
x: ro ro
QJ QJ QJ 4-> 4->
QJ C QJ CT QJ QJ
CQJQJ cro i — o QJO
rocr- rox: >>s_ cs_
xrro>>Qj-C4-> co QJO
4JO-CLC4-JQJ -r-i — i — i —
QJOOQJQJOQJ>X: 5-.X:
ES-i-i — ES-4-> o x:o QJ
oo.Q->,ooror— ro 4->roojc:
s-ooxrs-r— s_>-,s- ajt-cro
os-s-4->ox:-ax:4-' O4->ojx
i — O O CU i — O >>4J QJ S- QJ IM QJ
x: r— r- o J= ••- J= QJ+J e o 4-> ex:
OXTXTS-Oi- Olt-r— IQJO
•<— OOOO4->r— i-CMOXICMXl S-
•a ••-••- f— E i roo -H-O -oo
o-o-ox: oog i_^.^-< o rocM s_r-
Ei los- -ox: -ES- -ox:
O CM CO -r- XJ t — 1 i — O « — • O 4-> < — ' i — O
X> .— < i— 1 +-! TD •— 1 O CM i— I X> 4-> r— I o i— I
-------
77
UJ
_J
CO
z:
UJ
jr
«o:
i
0
•z.
co to
CXJ LO
| —
CU >-
-0 _l
•r- ec.
T— 21
CO -.
1
e£
•a:
Q
| —
CL
O
*
>>
S c
o S
ll
Q
1 —
Q.
O)
">!,
r—
ro
C
Ln
Q O O Q O LT>
CD O .— < LO CO CTi CT>
CO
CU
4_i
JD *O ,—
>•} .C «r—
X 0 S_
O CU U Ol CU 4->
CLi — i — C. i — Ol -i-
O) Ol •'— roOCU -i- i — C
T3-I—S-CU +-> c: i — s- -I— o
^"•j "O +J "TZ? r~~ O ra O 4-3 t- ^—
.c i T- -r- >, (o x c: •!— 4-> >>
0) «* C E m,— o re C -r- s-
"n> " O ra i- O *r- 4-S o 5T O
' — CO Ci — fa-i--O 3-i-O ro
fa • • o >5 O- Q. i jo Q. +J jr
i-CSJr— i-OO'St-OOO)4->
rO "roOt-S- '•coi-OCU
Q-— i E ro CLO..— i-r- CXra E
CU
•• s
—1 ra r-"
e 4-- o
*-. CO C
CD c ra
CO CO CU Ol
0-r- _0 E
•^r co.
>•> o ^:
• •* r— 4-^ 4->
C ra .^
o c: -o 5
•i— ro CU
•>-> s_ -o
O Q ra O)
CU 1 — CU 4->
••-3 Q- CL 3
C D. , —
•"— S- ro CU
_ o o
E «*- c o
3 ra
•«-> CU •>-> CU
ex,— a. -o
CU CL ra -r-
to E U • x
ro t- Q O
CO CO CU 1 —
3 E O- O)
CO _J f-
S- E ,- T3 0)
CU 1 >i S_ , —
^* LO jr ra >%
•(-> 2 J=
Ol O •(->
* s: 4-> LU
CU
IS)
2^
o
t— t
h-
un
o
CO
UJ
s:
0
U-
CO
UJ
1—
>-
— 1
c^
^"
^c
1— 1
1 — 1
•— ' CO
>• c
o
X -r-
•i — 4_>
TD -i-
c -a
cu -o
O «^"
Q.
«^
CO
Ol
4_3
>^
^—
ra
f~
^c
10
^3-
CO
3:
CO
CU
• — 4-> ro
O tO r—
f~ J— >^
0 >, S-
r— O 0) J- 0
O i — C CJ ra
-C ra O rO -C
o i— -!-> x: 4->
O t— O O •!-> CU
i — >-> C ra CU E
ra CD ro r— E
1- 4-> O i —
r— ro 3 ••— r— >•)
>•} r") r"> Q_ >^ j^
•— O O 0 -C 4->
i — S- CO S- 4-> CU
fO CL-r- CL CU E
Ol
o
O) •!->
c: 01
o ^
Ol •(->
"O CU r—
•r- ^ >,
M- S- *->
i — Ol i — 3
3 x: >,_o oi
Ol co 4-> x: o "O
-O -i— Ol -M CO >-)
•r- -o cu -i- -c:
Er— Ol
ra C >->i — i — T3
<— 0 -C >, >,t—
4_ S- 0) +J 4-> S-
O ro -i— OJ CU ro
ra O "O E E CL
01
C 01
ra i —
3 Ol t-
J3 i — +J C
>-j »r— »f— rO
01 X J_ C 4->
T3 O CU CU 4-» O CL
•I— CL CU r— i — *r- •»— rO
cu x oil — •!-••- c: CLO
CO-r--r-S_S-OOS-
rO "O i- 4-* 4-1 r— t- O)
X Ol 1 •(-> -I- -r- >, CL E
oe«i-"-ces-:>»
•r-OI "COOOXi—
-Oi — coo---c:rao>>
1 >> • • 4-> CL O -C S- JT
Tt-JTCvJ CU Oi — •f-J'O4-»
•>+-> "O5_roO)>)OI
.— la'T-'racLeE.cE
-------
78
i.
ID O
CM 3
CU CU
"^3 i-
•r— 13
i— 4-J
Ll_
Ol
^~
**"
4-^
03
O
••-
CU
to
c
o
T3
3
to
O
C
cu
•r—
O
u-
cu
c
o
4-5 CU
0 4->
03 >,
e ss
CU LlJ 03
^_ &
O3
CU
zr
4-"^
o
cr
to
cu
1>>
03 Q
03 Q.
CU CU
r- S-
••- 3
1 ^ 1 %
03 03
0 CU
> O.
p—
tz
'— CU
O 4-5
q-
>> O
-a o
3 S_
to :>,
o
1- -o
O_ CU
t
Heated
recovet
•
c
cu
E
Q.
O
cu
CU
•o
o
C"
cu
01
•r— .
to
03
03
Q
Q.
C_3
CJ3
£>
Q. £=
O
_1 •!-
Q- 3
rc to
V
D?
O
o
•o
o
4-> 01
CU CU
E 0
= S-
sl E
Ol
c -o
CU r-
cn cu
s -r—
O.
3 i.
1 0
C "-
CU 4->
i — cr
o cu
t=
40 .ex
0 O
03 r—
5- CU
X CU
LU -O
^
s_
0
Q-
o
2:
o
1 — 1
1—
^£
<^**
1 — 1
_J
•a:
=>
o
CM
o
'Jf CO
CM
O
cu o
•a p=
•r- 1—
i — LU
CO S
<")-^
0
Lu.
CO
LU
K-
>-
_1
• C
o
X -r-
•i^ 4->
T3 -r-
C T3
CU "O
o.«a:
Q.
<:
(/)
QJ
*^L
^^*
frt
c
^
10
oo
-
-*
t/J
cu
r~ ^~
CU -r- O
C i— CU C
•i- O C CU
•O O CU Q-
•1- •!- i- O
S- OL >,••-
>> 1 4J J=
CLCM 01 4->
CU CU CU
C C C
cu cu cu
ISI IM Nl
C C C
cu cu cu
qi o o -O qi
C= O O O C
CU i- S- S- CU
N O O O N
C i — i — i — C
CU JH J^ -C CU CU CU CU
cuj3Oooj3cuc:cc:
C O -I-T- T- C CU CU CU
CU S-T3T3T3i — CUi — i — • —
M O I I 1 >, 3 >> >j >-,
Ci — CMCO«d-JTr— X X X
CU J^ « *> •» 4-) O 1 1 1
J3O»—
CU >>
-T-J 1
.;: 03
r— c:
CO ^^
o
CM
f""^
CO
•o
o
4-9
cu
s:
r— 3
03
Q 0
1— CU
Q- s- sd-
CU CO
01 JD cu cu
•t- 01 £Z
01 T3 O
>, CU 1 4->
r- S- 1C
03 CU CU
C > •— CO
•a: o o
O C S«
Q CU CU LO
1— i. JZ |-^
Q. CL .
4J O i —
O O V- ^
4-> c jr o
4-> CD
1 CM
1 -0 r—
C D_
CU 03 OO
•i- . CU 6-S
• — c un
O -r-
o -o c
•I- •!- O
0- S_
1 >,
CM O-
-------
79
MR. TELLIARD: On our
second half of our morning session, our next speaker
is Diane Kocurek. She's going to talk again about
our favorite subject right now, which is the 304(h)
and the SW-846 methodology as a' comparison. Diane.
-------
80
COMPARISON OF SW-846 AND 304(H) METHODS
FOR ANALYSIS OF APPENDIX VIII ORGANIC COMPOUNDS
Chemical Manufacturers Association
Washington, D.C.
by
Dianna S. Kocurek, P.E.
Lial F. Tischler, Ph.D., P.E.
Tischler/Kocurek
116 East Main
Round Rock, Texas 78664
Presented at
Ninth Annual
U.S. EPA Symposium on the
Analysis of Pollutants in the Environment
Norfolk, Virginia
March 19-20, 1986
-------
81
ABSTRACT
A study was designed to compare analytical performance of gas chromatography/mass
spectrometry (GC/MS) 304(h) methods 624 and 625 as alternatives to SW-846 GC/MS methods 8240 and
8270 for the analysis of organic compounds in the Appendix VIII list in 40 CFR 261. Three prominent
laboratories participated in the study. A relatively simple groundwater matrix was spiked with various
combinations and concentrations of 24 volatile (11 priority pollutants) and 24 semivolatile compounds (8
priority pollutants) from the Appendix VIII list. Based on an evaluation of precision, accuracy, and false
negative and false positive observations, the use of the promulgated 304(h) GC/MS methods for the
analysis of pollutants in groundwater is as effective as using SW-846 methods. The list of compounds
amenable to GC/MS, however, is somewhat shorter than the Appendix VIII list as demonstrated by the
large number of false negative observations reported by all three laboratories. GC/MS methodology is
more amenable to the analysis of Appendix VIII compounds which are priority pollutants. In general, false
negative observations are more likely to occur at the lower concentration levels. The data user should be
aware of the limitations of the analytical methodology and interpret analytical data with caution due to the
occurrence of both false negative and false positive observations. Additional studies should include a
GC/MS (304(h)) method validation for those compounds on the Appendix VIII list of pollutants which can
be analyzed adequately by GC/MS methodology.
-------
82
COMPARISON OF SW-846 AND 304(H) METHODS
FOR ANALYSIS OF APPENDIX VIM ORGANIC COMPOUNDS
INTRODUCTION
The U.S. Environmental Protection Agency (EPA) published "Test Methods for
Evaluating Solid Wastes, Physical/Chemical Methods" (SW-846) to serve as a methods manual for
the sampling of ground water or leachate and analyses of compounds listed in Appendix VIII of 40
CFR 261. In October 1984, EPA proposed to make SW-846 mandatory for all testing and
monitoring activities required under Subtitle C of the Resource Conservation and Recovery Act
(RCRA). EPA stated that SW-846 contained the analytical methods for all Appendix VIII
compounds, excluding exotics and water reactive compounds.
A recent study by the Chemical Manufacturers Association (CMA), "Inter- and
Intralaboratory Assessment of Selected SW-846 Methods for Analysis of Appendix VIII
Compounds in Ground Water" (April 1985), cited several reports which revealed that SW-846 was
inadequate to provide proper guidance to analytical laboratories due to lack of sufficient
information and details, technical inaccuracies, and inconsistencies, and pointed out numerous
problems associated with analysis compounds in the Appendix VIII list. The CMA study confirmed
the findings of the other reports. CMA concluded that SW-846 does not contain analytical
methods for all Appendix VIII compounds, excluding exotics and water reactive compounds.
The present study by CMA was an extension of the earlier CMA study. One of the
purposes of this study was to compare analytical performance of gas chromatography/mass
spectrometry (GC/MS) alternative methods to GC/MS methods of SW-846 for the analysis of
organic compounds in the Appendix VIII list. This study, as well as the previous CMA study, was
designed to evaluate those compounds which are amenable to GC/MS methods. The alternative
methods selected for comparison were Clean Water Act Section 304(h) methods 624 (volatiles)
and 625 (semivolatiles). In addition, this study was designed to compare the 304(h) and SW-846
methods for priority pollutants in the Appendix VIII list versus other compounds in the list.
Comparisons were based on precision, accuracy, and false positive and false negative
observations.
-------
83
STUDY DESIGN
GROUND WATER MATRIX
The water samples used in this study were ground water spiked with various organic
compounds. The same ground water matrix which was used in the previously referenced CMA
study was used for the spiked samples in this study. A description of the ground water as taken
from the previous study is provided here for convenience.
The ground water was collected from a monitoring well in the coastal plain region of Texas.
Physical and chemical analyses, including organics, were conducted on the ground water. The
physical and inorganic chemical analyses are shown in Tablel. No organic compounds were
reported as being present in the ground water as determined by GC/MS analysis. Therefore, for
the purpose of calculating percent recovery (accuracy) in this study, the spiking value in the
sample was taken to be equal to the true value.
SAMPLE PREPARATION
A total of 24 volatile and 24 semivolatile compounds from the Appendix VIII list of 375
were selected for spiking into the ground water matrix. Of the 24 volatile compounds, 11 were
priority pollutants. Of the 24 semivolatile compounds, 8 were priority pollutants.
Four spiked samples were prepared from the ground water matrix. The spiking
arrangement was such that each compound was spiked in three of the four samples at three
different concentration levels. The list of spike compounds and concentrations for each sample
are presented in Table 2 for the volatiles and in Table 3 for the semivolatiles.
As noted in Table 2, paraldehyde is included in the list of volatile compounds.
Paraldehyde is included in the volatile group since EPA has listed this compound as amenable to
the purge and trap methodology. Analytical experience has indicated, however, that paraldehyde
is not purgeable.
ANALYTICAL LABORATORIES
-------
84
Three prominent laboratories participated in this study, and were identified in this report
as Labs A, B, and C. Lab A utilized SW-846 methods 8240 (GC/MS for volatiles) and 8270
(GC/MS for semivolatiles) for analysis of the spiked samples. Lab A also employed additional SW-
846 methods for all Appendix VIII organic compounds. Labs B and C utilized 304(h) methods,
specifically method 624 (GC/MS for volatiles) and method 625 (GC/MS for semivolatiles). None of
the laboratories were asked to analyze for metals.
The laboratories using the 304(h) methods were requested to look for other compounds
that may be present, but not priority pollutants. Both laboratories offer this service to customers
on a routine basis for a slight additional cost. The cost for analysis at Labs B and C was
approximately one-fourth to one-third the cost of analysis at Lab A.
DATA ANALYSIS METHODS
QUALITATIVE DATA
Qualitative data are considered in this study to be data reported as less than some value,
below the method detection limit (BDL or BMDL), and nondetected. These data were reported
differently by each of the three laboratories in this study. No data were reported as greater than
some value.
Lab A reported qualitative data in its individual sample reports as nondetect (ND) or BMDL.
The laboratory report defined ND as the absence of any detectable concentration of a compound
and BMDL as a detectable concentration below the method detection limit (MDL). Lab A's sample
summary report, however, for all four samples was in conflict with this terminology. In the summary
reports, both NDs and BMDLs were reported as less than the MDL (for example, < 4.7 micrograms
per liter (ng/l)). This resulted in an initial misinterpretation of the "less than" summary data as
positive observations of these compounds, or BMDLs. A review of the individual sample reports
corrected this misinterpretation and as a result, most of the less than values in the summary
reports were actually found to be NDs.
Lab B used below detection limit (BDL) to report compounds which were both
nondetectable and those which were detectable, but below the method detection limit. Lab C
reported qualitative data as NDs for compounds not detected at the limit of quantitation (LOQ),
and detected (D) less than the LOQ (for example, D,<10 |o.g/l).
3
-------
85
DATA REJECTION
Outliers
Outlier tests are normally performed on a given sample for each individual compound. For
example, one might test if any of the laboratories' results for a sample should be considered
unusual enough that it would be outside of the expected range due to normal variance.
Two different outlier tests were considered for this study, Student's t-test based on the
interlaboratory mean and standard deviation, and Dixon's ranking test. Since the study design
was limited in the number of samples and participating laboratories, neither test was considered
practical. Therefore, no data in this study were rejected as outliers.
Other Data Handling
Lab C reported the presence of 2-chlorophenol and 3-chlorophenol where the samples
were spiked with 2-chlorophenol and 4-chlorophenol. The 3-chlorophenol data were not rejected
and it was assumed that this laboratory had "detected" 4-chlorophenol, but had misidentified the
specific isomer.
Lab C also reported results for 1,2-dichlorobenzene and 1,3-dichlorobenzene from both
methods 624 and 625. Although these compounds are listed as semivolatiles, the method 624
(volatiles) results were used because they had higher accuracies (recoveries).
Methylmethacrylate, also listed as a semivolatile, was reported by Lab C from method 624. These
data were used since they were the only data for this compound from this laboratory.
PRECISION
This study was not designed to include replicate analyses for the purpose of comparing
precision between the 304(h) and SW-846 methods. Replicate analyses were only reported by
Labs A and C as part of their quality assurance/quality control programs. These data were
insufficient for method precision comparison even though both methods were represented (Lab
-------
86
A used SW-846, Lab C used 304(h)) since Lab A had replicate analyses on only one sample.
Therefore, method precision was compared by an alternate means.
Since the 304(h) methods in this study were methods 624 and 625, the overall
interlaboratory precision equations published as interim final for these methods (49 FR 43234,
October 26, 1984) were used to calculate the acceptable interlaboratory range in reported
concentrations for a given spiked compound in each sample. It was then determined whether
each laboratory's result was within the acceptable range. The number of times that a laboratory
failed to meet this criterion was totaled and the three laboratories were compared over all samples
and applicable compounds. The applicable compounds were the priority pollutants in this study
since the precision equations are limited to this group for methods 624 and 625.
Although this precision comparison is based on calculations for the 304(h) methods, it
provides a very simple qualitative assessment of whether the 304(h) and SW-846 methods
perform similarly. An example of the calculations performed is given below.
Sample 1. benzene
Equations are taken from Table 6,49 FR 43379.
S1 =0.25X-1.33
where S' is the overall precision (ug/l)
X is the accuracy as calculated below
X = 0.93C + 2.00
where C is the true concentration (u.g/1)(spike level assumed to equal true concentration)
Spike concentration of benzene in Sample 1 was 35.6 ug/l.
X . 0.93 * 35.6 + 2.00
= 35.108 p.g/1
S1 =0.25*35.108-1.33
= 7.447
Assuming a 95 percent confidence limit (t = 1.96), then the range in interlaboratory
concentrations is:
R =X±1.96*S'
= 35.108 ±1.96* 7.447
= 20.5, 49.7
-------
87
With reported values of 22.2, 35.0, and 26.0 u.g/1 for Labs A, B, and C, respectively, each
laboratory was within the interlaboratory range.
ACCURACY
Accuracy was calculated as a percent recovery of the spike concentration for each
compound in a sample. Average percent recoveries for each compound and each laboratory
were calculated as simple arithmetic means. The standard deviations of the average recoveries for
grouped compounds for each laboratory were calculated by the standard method, assuming a
normal distribution on an arithmetic scale.
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
A false positive observation is defined as a reported positive analytical result for a
compound where the compound is not actually present in the sample. Conversely, a false
negative observation is an analytical result that indicates that a compound is not present when in
actuality it is.
In the evaluation of false positive and false negative observations in this study, analytical
results reported as ND were taken as zero concentrations and therefore a negative observation.
Where Lab A reported BMDL data, indicating a positive observation, but below the method
detection limit, these data were taken as positive observations. Results for Lab C which were
reported as "detected", but less than some value, were handled as positive observations. Since
Lab B reported data as BDL, and did not distinguish between nondetects and results below the
method detection limit, it was assumed that the BDL data were all NDs. This assumption was
based on the performance of the other two laboratories which reported more NDs than "detects"
when the compound was not actually present in the sample.
COMPARISON OF METHOnS
PRECISION
-------
88
The comparison of the 304(h) and SW-846 method interlaboratory precision was based
on the overall precision equations available for methods 624 and 625 (304(h) methods) since
replicate analyses in this study were too limited for this purpose. A discussion of these equations
is given in "Data Analysis Methods."
The results of the precision performance comparison are presented in Table 4. Results
are only shown for the priority pollutants in this study since the methods 624 and 625 precision
equations have not been determined for the other compounds in Appendix VIM. No extreme
differences are seen in the overall performance of the three laboratories. Lab A (SW-846) was not
within acceptable concentration ranges a total of 7 times (out of 54), Lab B (304(h)) was not within
range 3 times, and Lab C (304(h)) was not within range 5 times.
Based on this comparison, one can conclude that the 304(h) methods perform at least as
well as SW-846 methods with respect to interlaboratory precision for analysis of priority pollutants.
ACCURACY
The accuracy or percent recovery of each compound spiked into the groundwater
samples was calculated for each sample and laboratory. A summary of these data as average
percent recovery for each compound and laboratory is shown in Tables 5 and 6.
Of the 11 volatile priority pollutants that were spiked into the samples, there is little
observed difference between results when the 304(h) or SW-846 methods are used, as shown
by the average percent recovery for all 11 compounds among the three laboratories in Table 5.
Based on the standard deviation for each laboratory, there is no statistical difference at the 5
percent significance level between any of the laboratories. This lack of difference is further
illustrated in Figure 1, where the average percent recoveries among the laboratories do not show
any strong trends or relationships. Lab A (SW-846), however, is shown to have the highest
standard deviation and the highest number of recoveries that exceed 100 percent (4 out of 11
compounds), a fact that causes its overall average percent recovery to be greater than the other
two laboratories using 304(h) methods.
Most of the other volatile Appendix VIII compounds which are not priority pollutants were
not detected in any of the samples into which the compounds were spiked, shown by the many
zero recoveries in Table 5. Only 3 of these 13 volatile compounds which are not priority pollutants
7
-------
89
were detected by at least one laboratory and the recovery results are mixed. Only Lab A (SW-846)
detected methyl ethyl ketone, and with an average recovery of 120.7 percent. Labs B (304(h))
and C (304(h)) had fairly similar recovery results (37.0 and 46.2 percent) for 1,2,3-
trichloropropane, but experienced much poorer recovery than Lab A (SW-846) at 84.1 percent.
The recovery results for carbon disulfide were widely divergent among the laboratories (187.7,
325.6, and 31.2 percent). Since the non-zero recovery data for the other volatile Appendix VIII
compounds were so limited, statistical analyses and recovery plots were not made.
The recovery data for semivolatile priority pollutants in Table 6 appear to show that Lab A
(SW-846) performed better than the other two laboratories utilizing the 304(h) methods. The
average percent recovery for Lab A (SW-846) was 78.8 percent as compared to 67.1 and 70.2
percent for Labs B and C (304(h)), respectively. However, statistically, the performance of Lab A
(SW-846) is not shown to be different from the other two laboratories at the 5 percent significance
level. Again, as with the volatile priority pollutants, Lab A (SW-846) had the highest number of
recoveries greater than 100 percent (3 of 8 compounds spiked).
The plotted data for the semivolatile priority pollutants, in Figure 2, do not show a strong
predominance of higher percent recovery for Lab A (SW-846). The data in the figure also show
that percent recovery for these 8 pollutants varies dramatically for each of the laboratories. Both of
these observations are in agreement with the lack of statistical difference among the laboratories.
Only 9 of the 16 other semivolatiie Appendix VIII compounds were detected by at least
one laboratory. The comparison of percent recoveries in Table 6 for other semivolatile Appendix
VIII compounds is similar to semivolatile priority pollutants. Lab A (SW-846) had the highest
average recovery for these nine compounds (46.3 percent); Labs B and C (304(h)) had 33.9 and
32.6 percent recovery, respectively. Statistically, Lab A (SW-846) is not different from Labs B and
C (304(h)) at the 5 percent significance level. Figure 3 illustrates this relationship.
In summary, Lab A, using SW-846 analytical methods, had the highest overall average
recoveries for the volatile and semivolatile groups for the priority pollutants and other Appendix
VIII compounds. This laboratory also had the highest number of calculated recoveries exceeding
100 percent. Despite the higher overall average recoveries attributable to the laboratory using
SW-846 methods, there is sufficient variation in the individual compound recoveries among all
laboratories that there is no significant difference in overall recoveries among laboratories using
the 304(h) or SW-846 methods.
8
-------
90
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
The false positive and negative observations occurring for volatile analyses are shown in
Table 7 for individual compounds. Among the three laboratories, only three false positive
observations were reported, once by Lab B (304(h)) and twice by Lab A (SW-846).
A large number of false negatives were observed, however, by all three laboratories.
False negative observations occurred for 3 of the 11 volatile priority pollutants, and for 11 of the
13 other Appendix VIII volatile compounds. Ten of these 13 other Appendix VIII compounds
were never detected by any laboratory. Lab A alone, using the SW-846 method, did not detect
ethylbenzene in any of the 3 samples spiked with the compound, but this same laboratory was the
only laboratory of the three which was able to detect methyl ethyl ketone in the spiked samples.
The false positive and negative observations for the semivolatile compounds are shown
in Table 8 for individual compounds. Lab B (304(h)) reported 2 false positive observations for
naphthalene and methoxychlor; Lab A (SW-846) reported 1 false positive observation for
methoxychlor.
Fewer false negative observations were reported for the semivolatile compounds as
compared to the volatile compounds in Table 7. Lab C (304(h)) reported the fewest false negative
observations (25). Labs A (SW-846) and B reported 30 and 34 false negative observations,
respectively. Seven of the other Appendix VIII compounds were never detected by any
laboratory. Diphenylamine and 4-nitrophenol were only detected by Labs B and C. Lab B was
never able to detect methoxychlor, but reported it as a false positive in the one sample not spiked
with the compound.
As mentioned in the section on data analysis methods, Lab C (304(h)) reported the
presence of 2-chlorophenol and 3-chlorophenol (the samples were spiked with 2-chlorophenol
and 4-chlorophenol). It was assumed for this study that the laboratory had "detected" 4-
chlorophenol, but had misidentified the specific isomer. This assumption does not change the
overall conclusions of the false positive and negative observations evaluation.
A summary of the false positive and negative observations is shown in Table 9 for the
volatile and semivolatile compounds. Separate results are also shown for priority pollutants versus
other Appendix VIII compounds. The total number of false analytical observations was nearly the
9
-------
91
same for all three laboratories. Lab A (SW-846) had slightly more and fewer false negative
observations for volatile priority pollutants and other Appendix VIII compounds, respectively, than
Labs B and C which used the 304(h) methods. Lab C had fewer false negative observations
reported for the semivolatiles (25), however, Lab B which used the same methods reported 34
false negative observations. Lab A using the SW-846 methods reported 30 false negative
observations. Lab C was the only laboratory not reporting any false positive observations.
As shown in Table 9, 72 is the maximum number of false negative observations which
could be reported by any one of the laboratories for either the volatile or semivolatile compound
groups in this study. Since 3 of the 4 samples were spiked with 24 volatile compounds and 24
semivolatile compounds, the maximum is 3 times 24, or 72, for each group. To break it down
further, based on the number of priority pollutants and other Appendix VIII compounds in the
volatile and semivolatile groups which were spiked in the samples, the maximum number of
possible false negative observations are: 33 volatile priority pollutants, 24 semivolatile priority
pollutants, 39 volatile other Appendix VIII compounds, and 48 semivolatile other Appendix VIII
compounds.
Based on the above maximums, the range in false negative observations among the
laboratories was 9-18 percent of the total volatile priority pollutant analyses from all the spiked
samples, 0-13 percent for semivolatile priority pollutants, 77-85 percent for other volatile
Appendix VIII compounds, and 52-71 percent for the other semivolatile Appendix VIII
compounds. For all priority pollutants, the total false negative observations represented 3
percent of the combined 432 analyses for all Appendix VIII compounds (48) in this study (48
compounds x 3 samples x 3 laboratories). The overall total false negative observations for other
Appendix VIJI compounds which were not priority pollutants was 42 percent.
The number of possible false positive observations is theoretically unlimited. That is, a
laboratory could report innumerable compounds which were not actually present in a sample.
Therefore, a calculation similar to the percent false negative observations was not attempted for
the percent false positive observations.
Based on the data in Tables 7 through 9, there is no significant difference in false positive
or negative observations among the laboratories and compounds. The same can be concluded
for the difference in the SW-846 and 304(h) methods. The data do indicate that many more false
negative observations occur for other Appendix VIII compounds than priority pollutants. One
explanation is that the priority pollutants are more commonly analyzed for, operators are more
10
-------
92
familiar with these compounds' identification and quantification, and standards are more readily
available for definitive identification. The data here are not conclusive on this point, however,
since the number of compounds used in this study was limited.
The false negative observation data do suggest a concentration effect. In Tables 7 and 8,
where a laboratory did not report false negative observations at all spike levels (indicated by 1 or 2
in the column), the false negative observations generally occurred at the lower spike levels. For
example, there are four instances in these tables where 1 or 2 false negative observations are
reported. In three of these four cases, it was at the lowest spike levels that the false negative
observation occurred.
CONCLUSIONS AND RECOMMENDATIONS
Based on an evaluation of precision, accuracy, and false negative and false positive
observations, the three laboratories in this study, one using SW-846 GC/MS methods, and the
other two using 304(h) GC/MS methods (methods 624 and 625), performed equally well.
Therefore, the use of the promulgated 304(h) GC/MS methods for the analysis of pollutants in
groundwater is as effective as using SW-846 methods and the analytical costs are approximately
one-fourth to one-third the cost of analysis by SW-846 methods.
GC/MS methodology is the current state-of-the-art for analysis of pollutants in
groundwater, however, the list of compounds amenable to GC/MS is somewhat shorter than the
Appendix VIII list. This is demonstated by the large number of false negative observations that
were reported by all three laboratories. Overall, 45 percent of the analyses for the spiked
compounds were reported as false negative observations.
GC/MS methodology is more amenable to the analysis of Appendix VIII compounds which
are priority pollutants. This is shown by the frequency of false negative observations reported by
all three laboratories for the Appendix VIII compounds which are not priority pollutants. Forty-two
percent of the analyses for the non-priority pollutant Appendix VIII compounds were false
negative observations; only 3 percent false negative observations were reported for priority
pollutants. Seventeen of the 48 compounds which were spiked into the samples were never
reported by any of the laboratories; all seventeen compounds were other Appendix VIII
compounds which were not priority pollutants.
11
-------
93
Data in the study suggest a concentration effect in the false negative observations. In
general, false negative observations were more likely to occur at the lower spike levels.
From a data user's point of view, analytical data must be interpreted with caution. The user
should be aware of the limitations of the analytical methodology. For instance, an analytical result
of "not detected" may be really a false negative observation and the compound may be actually
present in the sample. There is also a problem with false positive observations. Compounds not
present in the samples are actually being reported as being present. False positive observations
are likely to be reported as very low concentrations (near the method detection limit), and
therefore would be a particularly difficult problem in compliance monitoring where limitations are
set near or equal to the method detection limit.
Since the three laboratories participating in this study were aware of its scope and
objectives, it can be assumed that each laboratory made every effort to maximize the quality of its
performance. Thus, it is reasonable to assume that the results of this comparison represent the
"best" performance achievable by these analytical methods.
Additional studies should include a GC/MS (304(h)) method validation for those
compounds on the Appendix VIII list which can be analyzed adequately by GC/MS methodology.
Reagent water as well as groundwater matrices should be tested. GC/MS reference spectra
generated from such a study should be considered for publication and incorporated into a
reference spectra library similar to the priority pollutant spectra library. This would improve the
qualittative aspects for the analysis of those compounds that are really amenable to GC/MS
analysis.
REFERENCES
1. Stanko, G. H. and Fortini, P.E., "Inter- and Intralaboratory Assessment of Selected SW-846
Methods for Analysis of Appendix VIII Compounds in Ground Water," presented at the U.S. EPA
Symposium on the Analysis of Pollutants in the Environment, Norfolk, Virginia, April 1985.
12
-------
94
TABLE 1
PHYSICAL AND INORGANIC CHEMICAL PROPERTIES
OF THE
GROUND WATER MATRIX
USED FOR SPIKED SAMPLES
Property
Description
Appearance
pH
Total suspended matter
Total dissolved solids*
Chloride, Cl
Hardness, as CaCO3
Very turbid, sandy solids
6.9
1060 mg/l
580 mg/l
70 mg/l
318 mg/l
*Filtered through 0.45u. membrane
Analysis by GC/MS demonstrated that the ground water matrix was free of organics.
-------
TABLE 2
SPIKE CONCENTRATIONS
IN SAMPLE SOLUTIONS (micrograms per liter)
VOLATILES
95
Compound
Sample Number
Priority Pollutants
acrolein
benzene
bromoform1
chlorobenzene
chloroform
1,1-dichloroethane
1 ,2-dichloroethane
1 ,2-dichloropropane
ethylbenzene
toluene
1,1,1-trichloroethane
Other Aopendix VIII Compounds
acetonitrile
bromoacetone
carbon disulfide
chloroacetaldehyde
1 ,2,3,4-diepoxybutane2
1 ,4-dioxane
hexachloropropene
iso-butyl alcohol3
malononitrile
methacrylonitrile
methyl ethyl ketone4
paraldehyde**
1 ,2,3-trichloropropane
it
35.6
*
24.9
*
62.4
*
47.8
78.4
16.4
12.8
36.4
22.7
69.6
36.6
18.0
*
35.7
72.3
59.8
16.4
31.1
30.0
*
64.4
*
16.2
*
65.3
15.6
62.5
*
15.7
32.9
76.6
12.2
11.3
* .
48.8
35.9
102.0
47.6
24.1
It
32.8
*
20.0
21.4
32.2
11.9
32.4
49.9
32.7
*
31.3
15.9
*
65.7
38.3
72.9
34.0
46.4
, *
*
68.2
*
*
39.9
49.2
15.6
40.0
32.2
48.3
59.4
64.7
12.5
16.3
31.2
15.6
31.8
47.0
*
*
*
*
61.9
24.4
53.9
34.1
23.8
48.2
19.9
*
62.2
*
53.6
*Not spiked in sample.
"Paraldehyde is included in the volatile group since EPA has listed this compound as amenable
to the purge and trap methodology. Analytical experience has indicated, however, that
paraldehyde is not purgeable.
1tribromomethane
2dl-1,3-butadiene diepoxide
32-methyl 1-propanol
42-butanone
-------
TABLE 3
SPIKE CONCENTRATIONS
IN SAMPLE SOLUTIONS (micrograms per liter)
SEMIVOLATILES
96
Compound
Sample Number
Priority Pollutants
2-chlorophenol
1,2-dichlorobenzene
1,3-dichlorobenzene
di-n-octylphthalate
naphthalene
4-n'rtrophenol
p-chloro-m-cresol1
2,4,6-trichlorophenol
Other Appendix VIII Compounds
aniline
benzenethiol2
4-chlorophenol
3-chloropropionitrile
di-epinephrine
diethylstilbesterol
3,3'-dimethyIbenzidine
di-n-propylnitrosoamine3
diphenylamine
ethylenethiourea4
hexachlorophene
methoxychlor
methylmethacrylate
methyl parathion
phenacetin5
thioacetamide
92.0
*
157.0
37.3
231.0
121.0
*
32.7
93.4
*
*
59.3
87.6
63.2
39.7
146.0
it
39.4
It
181.0
81.9
77.0
178.0
88.4
*
158.0
39.3
74.7
46.2
*
277.0
196.0
31.1
48.2
161.0
*
117.0
31.6
79.4
*
39.6
78.7
162.0
*
*
51.3
59.3
118.0
30.7
79.1
*
149.0
*
40.4
185.0
98.0
187.0
72.3
80.3
119.0
*
94.8
*
97.3
79.2
118.0
81.1
120.0
40.9
103.0
it
*
153.0
39 6
W%/» \J
78 6
/ \J*\J
*
139 0
1 ww * \J
so a
\j\j t\j
92.3
*
*
121 n
1 f~ I • \J
40 2
"W»C.
29 7
fm\J t 1
58.4
*
1190
1 1 W . W
48 6
"w*U
1*58 n
1 \J\J* \J
*
122 0
1 CnC_> \J
1RO n
i \J\Jt \j
164 0
I \J^ • \J
*
1 IQ n
I 1 *7. \J
58.9
*Not spiked in sample.
M-chloro-S-methylphenol
2thiophenol
3n-nitrosodi-n-propylamine
42-imidazolidinethione
5p-acetophenetidide
-------
97
TABLE 4
PERFORMANCE COMPARISON OF LABORATORIES
BY INTERLABORATORY PRECISION FROM METHODS 624 AND 625 (304(H))
PRIORITY POLLUTANTS
Intel-laboratory Range*(|og/l) Number of Times Outside of Range
Sample 1
Lab A LabB Lab C
(SW-846) (304(h)) (304(h))
Volatiles
acrolein
benzene
bromoform
chlorobenzene
chloroform
1,1-dichloroethane
1 ,2-dichloroethane
1 ,2-dichloropropane
ethylbenzene
toluene
1,1,1-trichloroethane
— not given for Method 624 — ...
21-50
**
17-37
**
44-87
It*
6-90
42-116
14-23
9-20
**
8-25
**
39-83
11-23
38-90
**
12-24
23-46
50-117
9-17
21-50
29-73
20-42
**
20-45
2-30
**
41-92
26-59
32-83
47-101
11-18
10-21
22-44
10-22
4-60
27-70
**
**
1
0
0
0
1
0
0
3
0
1
0
2
0
0
1
0
0
0
0
0
1
2
0
0
0
0
0
0
0
0
Semivolatiles
31-113
2-chlorophenol
1,2-dichlorobenzene **
1,3-dichIorobenzene 27-248
di-n-octylphthalate
naphthalene
4-nitrophenol
p-chloro-m-cresol **
2,4,6-trichlorophenol 13-46
5-51
74-280
4-142
66-187
6-61
13-101
16-57
**
98-368
98-259
9-39
33-94
**
33-225
**
0-50
65-247
47-131
Total 6
52-187
16-48
13-124
**
46-169
0-96
31-125
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
"Calculated from equations in Table 6, 49 FR 43379
**Not spiked in sample
Total 1
Grand
Total 7
-------
98
TABLE 5
COMPARISON OF ACCURACY
FOR LABORATORIES USING 304(H) AND SW-846 METHODS
VOLATILES
Average Percent Recovery
Lab A(SW-846) Lab B(304(h)) Lab C(304(h))
Priority Pollutants
acrolein
benzene
bromoform
chlorobenzene
chloroform
1,1-dichloroethane
1,2-dichloroethane
1,2-dichloropropane
ethylbenzene
toluene
1,1,1-trichloroethane
Average
0.0
64.2
101.3
107.9
95.5
144.3
113.2
87.8
0.0
89.4
79.9
80.3
Standard Deviation 44.6
Other Appendix VIII Compounds
acetonitrile
bro mo acetone
carbon disulfide
chloroacetaldehyde
1,2,3,4-diepoxybutane
1,4-dioxane
hexachloropropene
iso-butyl alcohol
malononitrile
methacrylonitrile
methyl ethyl ketone
paraldehyde
1,2,3-trichlororopane
0.0
0.0
187.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
120.7
0.0
84.1
0.0
97.8
57.6
81.5
87.8
64.6
105.6
81.7
93.2
91.8
83.0
76.8
29.0
0.0
0.0
325.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
37.0
71.4
48.5
40.7
82.2
92.9
89.7
98.7
81.4
82.2
80.2
96.1
78.5
18.6
0.0
0.0
31.2
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
46.2
-------
99
TABLE 6
COMPARISON OF ACCURACY
FOR LABORATORIES USING 304(H) AND SW-846 METHODS
SEMIVOLATILES
Average Percent Recovery
Lab A(SW-846) Lab B(304(h)) Lab C(304(h))
Priority Pollutants
2-chlorophenol
1,2-dichlorobenzene
1,3-dichlorobenzene
di-n-octylphthalate
naphthalene
4-nitrophenol
p-chloro-m-cresol
2,4,6-trichlorophenol
Average
Standard Deviation
Other Appendix VIII Compounds
aniline*
benzenethiol
4-chlorophenol*
3-chloropropionitrile
di-epinephrine
diethylstilbesterol
3,3'-dimethylbenzidine
di-n-propylnitrosoamine*
diphenylamine*
ethylenethiourea
hexachlorophene*
methoxychlor*
methylmethacrylate*
methyl parathion*
phenacetin*
thioacetamide
Average*
Standard Deviation*
86.2
101.1
101.4
96.1
56.3
0.0
109.2
80.0
78.8
35.9
54.6
0.0
0.0
0.0
0.0
0.0
0.0
86.6
0.0
0.0
10.9
59.7
37.4
109.2
58.5
0.0
46.3
38.1
70.7
62.8
53.5
81.3
70.2
30.7
97.5
69.7
67.0
19.6
40.0
0.0
0.0
0.0
0.0
0.0
0.0
74.5
115.0
0.0
0.0
0.0
0.0
41.8
33.9
0.0
33.9
40.2
68.7
107.9
89.5
6.2
47.4
57.1
87.7
97.3
70.2
33.0
18.7
0.0
47.9
0.0
0.0
0.0
0.0
54.1
102.4
0.0
0.0
21.1
14.6
10.4
24.0
0.0
32.6
31.3
*Average and standard deviation of nine compounds noted.
-------
100
TABLE 7
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
BY COMPOUND AND LABORATORY
VOLATILES
Observations
False Negative*(False Positives)
Lab A
SW-846
LabB
304(h)
LabC
304(h)
Priority Pollutants
acrolein
bromoform
chloroform
ethylbenzene
toluene
Total
3
0
(1)
3
(1)
6(2)
3
0
0
0
0
2
1
0
0
0
Other Appendix VIII Compounds
acetonitrile
bromoacetone
carbon disulfide
chloroacetaldehyde
1,2,3,4-diepoxybutane
1,4-dioxane
hexachloropropene
iso-butyl alcohol
malononitrile
methacrylonitrile
methyl ethyl ketone
paraldehyde
Total
Total for All Compounds
3
3
0
3
3
3
3
3
3
3
0
3
30
36(2)
3
3
(1)
3
3
3
3
3
3
3
3
3
33(1)
36(1)
3
3
0
3
3
3
3
3
3
3
3
3
33
36
Summary
• Ten of the total 24 compounds were never detected by any laboratory. All 10 were other
Appendix VIII compounds.
• Ethylbenzene was never detected by Lab A.
• Methyl ethyl ketone was never detected by Labs B and C.
• Only 3 false positive observations were reported.
'Maximum number of false negative observations possible for any laboratory was three for each
compound since each compound was spiked in 3 of 4 samples analyzed by each laboratory.
-------
101
TABLE 8
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
BY COMPOUND AND LABORATORY
SEMIVOLATILES
Observations
False Negative*(False Positives)
Lab A
SW-846
LabB
304(h)
LabC
304(h)
Priority Pollutants
naphthalene
4-nitrophenol
Total
0
3
(1)
0
(1)
0
0
Other Appendix VIII Compounds
aniline
benzenethiol
4-chlorophenol
3-chloropropionitrile
di-epinephrine
diethylstilbesterol
3,3'-dimethylbenzidine
diphenylamine
ethylenethiourea
hexachlorophene
methoxychlor
methylmethacrylate
thioacetamide
Total
Total for All Compounds
0
3
3
3
3
3
3
3
3
0
(1)
0
3
27(1)
30(1)
1
3
3
3
3
3
3
0
3
3
3(1)
3
3
34(1)
34(2)
0
3
0
3
3
3
3
0
3
3
0
1
3
25
25
Summary
• Seven of the total 24 compounds were never detected by any laboratory All 7 were other
Appendix VIII compounds.
• Diphenylamine and 4-nitrophenol were never detected by Lab A.
• Methoxychlor was detected by Labs A and C. Lab B reported it as a false positive obervation
• Only 3 false positive observations were reported.
'Maximum number of false negative observations possible for any laboratory was three for each
compound since each compound was spiked in only 3 of 4 samples analyzed by each laboratory.
-------
102
TABLE 9
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
SUMMARY
Laboratory
Number of Observations
False
Positives
(PP/App.VIII)
False
Negatives*
(PP/App.VIII)
Volatiles
A(SW-846)
B(304(h))
C(304(h))
2/0
0/1
0/0
6/30
3/33
3/33
Semivolatiles
A(SW-846)
B(304(h))
C(304(h))
0/1
1/1
0/0
3/27
0/34
0/25
PP - priority pollutants in Appendix VIII
App.VIII - other pollutants in Appendix VIII
*Maximum number is 72 (3 samples x 24 spiked compounds)possible false negative observations
for each laboratory (for voiatiles: 33-priority pollutants, 39-other Appendix Vlil compounds; for
semivolatiles: 24-priority pollutants, 48-other Appendix VIII compounds).
-------
CO
o
H
CD
00
CO
CD
O
£2-
CD
CD
o
CO,
O
m
C/3
•*—>
cd
H—'
JD
"o
DL
O
"k_
Q_
CD
O
o
o
CD
DC
•»— >
c
CD
O
O
Q_
CD
CD
CO
i__
CD
O
O
JD
CO
0)
i_
13
CD
-------
o
H
CD
CO
I
Q
CQ
O
CO,
CQ
CQ
O
CQ
CO
CO
o
I
Cfl
O
Q_
g
"k—
CL
o
1
CD
CO
CD
o
o
CD
DC
•+-'
CD
o
CD
Q_
CD
CD
2
CD
>,
O
O
JD
CO
CM
CD
O)
LJL
-------
10
o
H
CO
T3
C
D
O
CL
o
O
^
CD
CO
1
^^
CD
_J
^;
JZ
Xj"
O
CO^
CO
m
_j
^^3
^^
£,
^sh
o
co^
0
CQ
_J
|i§>i||
1
Q
O
CD
C
'E
03
"
c
CD
sz
CL
O
!c
o
CM
cu
.E
03
0
c/)
O
i_
•<— ;
|c
Q.
0
cL
cr
T3
CO
CD
r~
diphenylamine
hexachlorophel
^- LO
methoxychlor
CD
$
4^
CL*
methylmethac
"^
c
_o
i
o3
Q.
">^
•<— '
CD
CO
CD
CD
^
•0
i—
phenacetin
ther 7 compoui
o
CT>^
w
o
CD
o
Q.
Q.
CD
^1
•<— >
O
CD
O
1
CD
CO
CD
O
O
CD
DC
•+—•
c:
CD
O
CD
Q_
CD
O)
CO
CD
>~,
o
•g
o
JD
CO
CD
\—
13
D)
LL.
-------
106
INTRODUCTION
Follow-up to previous Chemical Manufacturers
Association (CMA) study
Objectives of present CMA study
• Compare performance of alternative GC/MS
methods (304(h)) to GC/MS methods of
SW-846 for analysis of organics in
Appendix VIII
• Compare performance of both groups of
methods for priority pollutants versus other
organic compounds in Appendix VIII
• Evaluate those compounds which are
amenable to GC/MS methods
-------
107
STUDY DESIGN
Ground water matrix spiked with organic compounds
in various concentrations and combinations
4 spiked samples
24 volatile compounds
• 1 1 priority pollutants (1 1 .9 - 78.4 jig/L)
• 1 3 other Appendix VIII compounds (1 1 .3 - 1 02.0 ja,g/L)
24 semivolatile compounds
• 8 priority pollutants (30.7 - 277.0 |ig/L)
• 1 6 other Appendix Vlll compounds (29.7 - 1 87.0
3 prominent laboratories
• 1 used SW-846 methods
• 2 used 304(h) methods
-------
108
STUDY DESIGN
304(h) GC/MS methods
• 624 (volatiles)
• 625 (semivolatiles)
SW-846 GC/MS methods
• 8240 (volatiles)
• 8270 (semivolatiles)
-------
109
PRECISION
Study not designed specifically for precision calculations
Qualitative comparison made based on precision equations
for methods 624 and 625
-------
110
PERFORMANCE COMPARISON OF LABORATORIES
BY INTERLABORATORY PRECISION FROM METHODS 624 AND 625 (304(H))
PRIORITY POLLUTANTS
Intel-laboratory Range*(ng/l) Number of Times Outside of Range
Sample 1
Lab A LabB LabC
(SW-846) (304(h)) (304(h))
Volatiles
acrolein
benzene
bromoform
chlorobenzene
chloroform
1,1-dichloroethane
1,2-dichloroethane
1,2-dichloropropane
ethylbenzene
toluene
1,1,1 -trichloroethane
Semivolatiles
2-chIorophenol
1,2-dichlorobenzene
1,3-dichlorobenzene
di-n-octylphthalate
naphthalene
4-nitrophenol
p-chloro-m-cresol
2,4,6-trichlorophenol
—• not given for Method 624 —
21-50 ** 9-17 32-83
8-25 21-50 47-101
17-37 ** 29-73 11-18
10-21
22-44
10-22
4-60
27-70
**
44-87
**
6-90
42-116
14-23
9-20
39-83
11-23
38-90
**
12-24
23-46
50-117
20-42
**
20-45
2-30
**
41-92
26-59
31-113
**
27-248
5-51
74-280
4-142
**
13-46
**
66-187
6-61
13-101
16-57
**
98-368
98-259
9-39
33-94
**
33-225
**
0-50
65-247
47-131
52-187
16-48
13-124
**
46-169
0-96
31-125
**
'Calculated from equations in Table 6,49 FR 43379
"Not spiked in sample
1
0
0
0
1
0
0
3
0
1
Total 6
0
0
0
0
0
1
0
0
Total 1
Grand
Total 7
0
2
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
2
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
16
-------
Ill
ACCURACY - VQLATILES
Volatile priority pollutants
• No statistical difference between labs at 5% level
• Average recoveries (80.3, 76.8, 78.5 %)
Other Appendix VIII volatile compounds
• Mixed recovery results on 3 compounds
• Other 10 compounds were never detected by any lab
-------
112
ACCURACY • SEMIVQLATILES
Semivolatile priority pollutants
• No statistical difference between labs at 5% level
• Average recoveries (78.8, 67.0, 70.2 %)
Other Appendix VIII semivolatile compounds
• No statistical difference between labs at 5% level
• 7 compounds were never detected by any lab
-------
CD
CO
I
£2.
CD
O
£0-
CQ
CD
O
CO,
O
CO
« .E
C CD
CO CD
O
O O O O O O
03
•*—'
JD
"5
CL
g
"»_
Q_
O
>,
o5
O
O
CD
cc
(D
O
CD
Q_
CD
O)
03
i_ •
CD
o
JD
Oi
CD
i_
13
D)
LL
-------
H
H
I
03
CD
00
C/D,
<
m
o
CO,
m
m
o
£2-
O
m
O
o
a.
.£•
v_
o
ol
1
o
1
CD
CO
CD
O
O
CD
DC
•*—•
c
CD
O
i_
CD
Q_
CD
CD
C3
CD
>>
O
O
JD
03
CvJ
CD
i_
ZJ
D)
LL
-------
LO
H
CO
•a
c
D
O
Q.
E
o
O
CD
00
CD
O
d
CD
CD
O
CO,
o
CD
. E R-S ,
CX Q. cd 2 O CD Ctf
O O tr- O >^ P Q.
n^»- »- >^-T= x£
^_g Q. § •§ o^>
9^ o .
CD >^ CO
^ s ®
co
o
r-t«
alE
c
E 815
8SS
c5
o3
r5 ^ ^
*-' T3 J=
Q) CD CD
E E E
CD
_
CD
cMco^LncDh-oocn+r
O
CD
O
E
CD
CO
Q)
>
O
O
CD
DC
•*— '
c
CD
o
CD
CL
CD
D)
CD
O
o
_a
CD
u_
-------
J.16
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
VOLATILES
Only 3 false positive observations reported
• 2 - 304(h); 1 - SW-846
Large number of false negative observations by all
Slabs
• 3 of 11 priority pollutants
• 11 of 13 other Appendix VIII compounds
-------
J.17
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
SEMIVOLATILES
Only 3 false positive observations reported
• 2 - 304(h); 1 - SW-846
Large number of false negative observations by all
3 labs, but fewer than for volatiles
• 1 of 8 priority pollutants
• 13 of 16 other Appendix VIII compounds
-------
118
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
SUMMARY
Laboratory
Volatiles
A(SW-846)
B(304(h))
C(304(h))
Number of Observations
False
Positives
(PP/App.VIII)
False
Negatives*
(PP/App.VIII)
2/0
0/1
0/0
6/30
3/33
3/33
Semivolatiles
A(SW-846)
B(304(h))
C(304(h))
0/1
1/1
0/0
3/27
0/34
0/25
PP - priority pollutants in Appendix VIII
App.VIlI - other pollutants in Appendix VIII
21
-------
JJ.9
FALSE NEGATIVE OBSERVATIONS
SUMMARY
VOLATILE
Priority pollutants
• SW-846: 18%
• 304(h): 9%
Other Appendix VIII compounds
• SW-846: 77%
• 304(h): 85%
SEMIVOLATILE
Priority pollutants
• SW-846: 13%
• 304(h): 0%
Other Appendix VIII compounds
• SW-846: 56%
• 304(h): 52-71%
VOLATILE and SEMIVOLATILE
• Priority pollutants: 3% (15 out of 432 analyses)
• Other Appendix VIII: 42% (182 out of 432 analyses)
-------
120
FALSE POSITIVE AND NEGATIVE OBSERVATIONS
SUMMARY
No significant difference among laboratories/methods
More false negative observations than false positive
observations
Less false negative observations for priority pollutants
False negative observations generally occur at lower
concentration levels
-------
121
CONCLUSIONS AND RECOMMENDATIONS
GC/MS methods from SW-846 and 304(h) perform equally
well for the analysis of organic Appendix VIII compounds
in ground water
Cost of 304(h) GC/MS methods are 1/4 to 1/3 cost for
SW-846 GC/MS methods
List of compounds amenable to GC/MS methodology is
shorter than Appendix VIII list (45% of spikes were
reported as false negative observations)
GC/MS methodology is a more amenable to analysis of
Appendix VIII compounds which are priority pollutants
•. 17 of 48 compounds never reported by any lab
• All 17 compounds were not priority pollutants
-------
122
CONCLUSIONS AND RECOMMENDATIONS
False negative observations are more likely to occur at
lower concentrations
Data user should interpreted analytical with caution
• False negative observations
• False positive observations
Labs were aware of study scope and objectives
• Analytical results represent "best" performance
Additional studies suggested
• 304(h) GC/MS method validation for Appendix VIII list
• Reagent water and ground water
• GC/MS reference spectra from study should be
incorporated into library similar to that for priority
pollutants
-------
123
QUESTION AND ANSWER SESSION
MR. APRIL: Could you
explain what some of the more striking differences
are between the 304(h) methods and the SW-846? I was
especially interested by the large cost differential.
MS. KOCUREK: Do you mean
differences in the analytical procedures or...
MR. APRIL: Just the
differences in the procedures, especially the ones
that lead to the large cost
differential.
MS. KOCUREK: I can't
really address the differences in the analytical
procedures because I'm not an analytical chemist. I
merely looked at the data that resulted from using
these methods. So I can't address analytical
methodologies, the specifics of them.
MR. LIN: Denis Lin from
ETC. Well, maybe I can answer Bob's question, too.
But let me first point out that ethyl benzene is a
priority pollutant. It is not on the Appendix VIII
list, therefore I don't know what affect it will
have in your false positive/false negative or position
study because one lab did not report ethyl benzene.
-------
124
I kind of suspect that lab was not looking for it
because it is not an Appendix VIII compound.
MS. KOCUREK: Yes, let me
take a look.
MR. LIN: While Diane is
looking, Bob, the answer to your question is if it is
comparing apples to apples, in other words, if somebody
asks for 8240 as opposed to 624, I'm sure the price
differential is not as great as stated, or for that
matter, 625 as opposed to 8270. However, if the
request is for all the methodology involved in so-
called Appendix VIII analysis, which include all the
GC method, all HPLC method, I'm sure the price
differential will add up very quickly.
If you are looking for say the volatile
analysis, no, I'm sure there is some difference,
because Appendix VIII volatiles will go up to about
50, 55 compounds.
MR. APRIL: ...then there's
no significant cost difference.
MR. LIN: We can do a survey
here. There's enough labs around. The answer to
your question is if you are looking for the same
compounds, I don't think there should be substantial
difference.
-------
125
MS. KOCUREK: Denis, I've
got an answer to your question. The false negative
observations for ethyl benzene for one of the labs
was reported three out of the three spiked samples,
so, yes, it's possible that if it were not reported
because it was not an Appendix VIII compound that
could explain the three false negatives.
MR. LIN: Yes. It is very
obvious to me. I checked the list, so I figured out
that probably is the reason.
MS. KOCUREK: Yes, it
would make a slight difference. It wouldn't make a
difference in the comparison between the methods
because overall, the performance wouldn't change that
much.
MR. STANKO: George Stanko
from Shell. I'd like to explain a little bit of
difference between Method 624, 625 and the 8240 and
8270. The 624, 625, the 304(h) methods were designed
primarily to satisfy the requirements of the Clean
Water Act. The list of priority pollutants is the
only compounds listed in these methods although the
method is amenable to other compounds. In other
words, you could look for priority pollutants plus.
Laboratories who offer that service will charge you
-------
126
almost the same as the priority pollutant list. If
they don't find any, you don't get charged any more
for 624 or 625 than you would if they had been looking
for an effluent or priority pollutants alone.
Laboratories who are using or supposedly using
8240 and 8270 are supposed to be looking for all
Appendix VIII compounds. Now, one of the compounds
was just pointed out right now may be a priority
pollutant but it's not on the Appendix VIII list. It
l
is amenable to the GC/MS methodology.
When you request analysis for a groundwater sample
for Appendix VIII compounds, whoever those laboratories
are who are capable of doing that, and I have some
doubts at this point in time, you will be charged
more for that sample and analysis on that sample.
Our experience has shown, when we have submitted
samples and requested Appendix VIII analysis which
include 8240 and 8270 plus some additional methods
which also don't work, the data you get back costs
you four times as much for the laboratory doing SW-
846 methods as the laboratories...if you sent the
same sample and requested 304(h) plus.
The analytical methods here again, the procedures
that you follow are the same. However, 304(h) will
not allow you to run a solid through the purge and
-------
127
trap device. Now, I don't know how it's done in SW-
846, but in theory, you can use SW-846, Methods 8240
and 8270 for solids as well as groundwater samples.
Now, this paper dealt nothing beyond groundwater
samples. But that is another difference in the two
methods. There is more flexibility allowed in 8240
and 8270 with respect to pre-sample preparation,
cleanup techniques and all that, which are not allowed
in the 304(h) methods.
MR. TELLIARD: George, do
you think that the cost of the analysis was just the
fact that Shell was sending out the samples?
MR. STANKO: There was a
true positive...
MS. KOCUREK: Are there
any other questions?
MR. TROIANO: Jeff Troiano,
Ford Motor Company. Could it be that this cost
differential, the magnitude that you indicated, is
that based on just using the costs associated with
using these three labs or a much larger population?
MS. KOCUREK: I believe
it's just using the three labs. George, you may want
to confirm that.
MR. STANKO: There is a lot
-------
128
of cost information available for Method 624 and 625.
For this study, we found only one laboratory that
claimed to be able to run SW-846 for all Appendix
VIII. We have a second laboratory that said they did
and they backed out.
MR. TROIANO: Well, going
back to my question, is the difference in price
between 304(h) and SW-846 methods, is it based on
these three laboratories or a survey of many
laboratories?
MS. KOCUREK: Well, I
think what George was saying is that the SW-846 cost
was based on one lab, and then the other ones were
based on known, I think, analytical costs for 624 and
625.
MR. TROIANO: And the SW-846
costs were based on all Appendix VIII or just the
ones in this test?
MS. KOCUREK: I think it
was derived from the cost estimate for this test. Is
that correct, George?
MR. STANKO: Basically, yes.
MR. ARLAUSKAS: Joe
Arlauskas with Martin Marietta Environmental Systems.
I found something else out, and somebody can correct
-------
129
me on this/ but the published list of compounds under
the October Federal Register 624 method, which are
supposedly priority pollutants, actually includes
compounds; at least I think one or two that are not
priority pollutants. And one of those, I think, are
the trichlorofluoro...
MS. KOCUREK: Are you
saying October '85 or '84?
MR. ARLAUSKAS: '84.
MS. KOCUREK: '84? I
don't remember that.
MR. ARLAUSKAS: Well, if
you compare the priority pollutant list against those
compounds, six...and published in that...
MS. KOCUREK: You're
talking about what was published in the Federal
Register?
MR. ARLAUSKAS: As the
compounds to be analyzed under 624.
MS. KOCUREK: Oh, okay.
I'm thinking of the tables that were presented under
the method for the, like precision and accuracy
equations. I only recall the priority pollutants
being listed for that.
MR. ARLAUSKAS: Okay.
-------
130
MR. TELLIARD: But you have
to remember on that publication, it's rather spurious
and you really can't count on it.
MR. ARLAUSKAS: But a lot
of laboratories will give you all that analysis as
they are priority pollutants, and if you're not
careful as a client...
MS. KOCUREK: Right.
MR. ARLAUSKAS: ...you're
going to get parameters that you don't have to look
for.
MS. KOCUREK: That's
correct. Yes.
MR. STITES: Ron Stites
with Cenref Labs. I don't know if I can shed any
light on this at all, but I can tell you the practice
we have in our laboratory which may show a difference
in pricing between the 624, 625 and then true Appendix
VIII analysis, which I don't think anybody can really
do.
If we have a reguest to run Appendix VIII
compounds, we have to talk to the client, and the
first thing we ask is, okay, do you really want us to
standardize for all of the compounds that we can get
our hands on and actually see if we can get some
-------
131
recovery, which is a quasi-research project because
you really don't have methods you can really go to
and say for sure this is going to work, or do you
want us to just run basically 624, 625 and see if
anything else shows up on the TICs, Tentatively
Identified Compounds. And there's a big cost
differential in those two approaches.
Just the fact of purchasing all sorts of compounds
and injecting on your GC/MS and standardizing and
getting response factors, there's where the big cost
differential comes from. I say again, I don't think
anybody can run Appendix VIII compounds, not really,
unless they just do it by definition and say we run
it this way, therefore we can do it.
MS. KOCUREK: Thank you.
MR. TUROSKI: Victor Turoski,
James River. What were the three spiking levels that
you used?
MS. KOCUREK: They were
all different for the different compounds. That's
why I gave the range on that one overhead for the
different priority pollutants. They were different
for each compound.
MR. TUROSKI: I'm sorry, I
missed that. What were they?
-------
132
MS. KOCUREK: You want it
for every compound?
MR. TUROSKI: General range,
MS. KOCUREK: General
range? Okay, just a second. For the volatiles the
priority pollutants range was about 12 to...it looks
like 78 Mg/L. For the other Appendix VIII compounds,
the volatiles, it was about 11.3 to 102 Mg/L. For
the semivolatiles, priority pollutants, the spike
range was 30.7 to 277 Mg/L. For the other Appendix
VIII compounds, semivolatiles, it was 29.7 to 187
Mg/L.
MR. PRESCOTT: Dianna,
while I'm walking around here I have a
question.
MS. KOCUREK: Yes, Bill.
MR. PRESCOTT: Did all
three of the laboratories know what compounds to
look for, or were they told the compounds that have
been spiked are on the Appendix VIII list?
MS. KOCUREK: They were
told to look for Appendix VIII compounds. I don't
think they were given the list of compounds.
MR. STANKO: The two
laboratories using 304(h) were told to look for
-------
133
priority pollutants plus anything else they could
identify; primarily the Appendix VIII compounds.
MS. KOCUREK: Right. Now,
let me add, on any of the false positive observations,
they were only limited to the list of spiked compounds.
We did not pull out any data beyond the list of 48
that we had.
MR. STANKO: One other
comment on the spiking levels that were used. They
were multiples, and they were chosen for a purpose not
at even numbers. Previous studies we have done 50,
100 and 150. This time we tried not to bias the
data. We picked some numbers between 30 and 80 for
the 50, and 100 was between 80 and 120. Whatever it
weighed out, that's what got put in. The numbers are
the true values that were actually put in the samples
and they were odd intentionally.
MR. TELLIARD: That figures.
Thank you very much, Dianna.
MS. KOCUREK: Thank you.
-------
134
MR. TELLIARD: Our next
speaker is Dave. Do you want to come on up?
-------
135
DAVID N. SPEIS
ENVIRONMENTAL TESTING AND CERTIFICATION CORPORATION
METHOD PERFORMANCE CHARACTERISTICS FOR
SELECTED RCRA APPENDIX VIII ANALYTES
MR. SPEIS: Actually, I feel that
my presentation might better be entitled, What I Did On My Summer
Vacation in 1982, since it is the third presentation on Appendix VIII
analytes this morning.
I'm happy to have the advantage of following Sam and Dianne since
I will be showing a different perspective for the performance of RCRA
Appendix VIII analytes in water.
Our laboratory has been able to develop a quality assurance data
base, which has given us the capability of gathering precision and
accuracy data for some of the more unusual Appendix VIII analytes wbj.:h
we spiked into groundwater. From that information, we can illustrate
performance expectations for SW-846. I'll be sharing this information
with you later on this morning.
Before I get started, I'd like to take care of a bit of
housekeeping and introduce my colleagues who worked with me on this
presentation. Of course, we have Denis Lin who provided significant
input into the initial classification of many of the compounds on the
-------
136
Appendix VIII list. He has also provided significant imput to EPA on
the recent short term guidance that was issued by the agency. Today,
he's my chief slide flipper.
Jim Bower is our data systems specialist at ETC. Jim is the
person who designed and developed the QA data base that we use for all
our analytes. His work enabled me to gather the information that I'm
going to be showing you a little bit later on.
ETC was presented with the unique opportunity of being involved
with the development of the Appendix VIII analytical chemistry from the
ground up. Our involvement began in July '82, when the agency first
published the list of Appendix VIII analytes. We perceived a need for
the analysis of groundwater samples for these compounds. Based on the
perceived need, we attempted to classify the compounds into analytical
methodology that we could perform for our clients.
We began the large and difficult task of classifying the Append!::
VIII compounds by common physical and chemical properties so they ccvild
be lumped into what we call a survey method approach. This approach if>
designed to analyze a large number of compounds with similar properties
using the fewest number of methods as possible. We are all aware of
survey method approaches and the difficulties associated with
performing them. As the number of compounds targeted by a specific
technique increases, you decrease your chances for successful analysis
for all those compounds targeted by the method.
-------
JL37
The primary objective in using a survey method approach is to get
data quickly. If you took a more singular approach, an approach that
used many methods for many compounds, obtaining data would be extremely
difficult, this approach was designed to produce data.
SLIDE 2
In September '82, the agency facilitated the method development
process by incorporating SW-846 as the methods of choice for Appendix
VIII. What we did at that point was to merge our list of classified
compounds into the SW-846 analytical scheme. The results of that
merging process are shown in the second slide.
There are six general techniques used for Appendix VIII analytes
in the SW-846 approach. Volatile organics are performed by Method
8240, extractable organics using a modification of Method 8270. The
modification is initial extraction of the sample at neutral pH or
ambient pH followed by basic extraction which is combined with the
neutral extract prior to analysis before going to acidic extraction of
the sample. Pesticides and herbicides are analyzed by Methods 8040,
8140, and 8150; water soluble volatile organics by direct aqueous
injection; polar and thermally labile organics by high performance
liquid chromatography using either direct aqueous injection or
liquid/liquid extraction. Metals and metal complexes were performed by
atomic absorption and inductively coupled Argon Plasma.
-------
138
The original Appendix VIII compound list started out with
approximately 375 compounds. Included on the list were 51 compounds
that were metals or metallic compounds. Also included on the list were
27 compounds that were considered exotic. By exotic, I mean they would
require a unique or specific method to determine their presence.
During the categorization process, we identified an additional 44
compounds that we thought were also exotic or unstable in water. We
omitted those from our classification. This left 253 compounds from
the original list for categorization into the SW-846 based analytical
scheme we have described.
By far, the two largest compound classes were the volatile
organics, which had a total of 59 compounds including 28 priority
pollutants and thfe extractable compounds which consisted of 164
compounds. As you can see and as I mentioned earlier, executing survey
methods for a group of compounds which numbered 164 is extremely
difficult.
I'll be spending a bit more time on the extractable organics and
the volatile organics later on in the presentation, when I discuss th>a
information we have extracted from our data base.
SLIDE 3
For our analytical scheme, we reduced the metals list, which had
started out as 51 metals and metal compounds, to a total of 22 metals.
-------
139
I'm going to be moving through these slides quickly since much of this
ground has previously been covered this morning.
SLIDE 4
The extractable HPLC list in this classification numbers 20
compounds. .One of the more difficult tasks in Appendix VIII analysis
was obtaining these compounds. The HPLC portion of the Appendix VIII
scheme, as most of you already know, has been excluded from the program
by the short term guidance. This change was extremely pleasing to our
HPLC specialist.
SLIDE 5
We also have a number of compounds that we classified for direct
aquesous injection HPLC. There are 18 compounds on that list. These
compounds have some water solubility and our early investigations
indicated that it would be very difficult to recover these compounds
from water using liquid/liquid extractions.
SLIDE 6
We've classified 26 compounds on the list as pesticides which can
be analyzed using gas chromatography (GC) with an electron capture
detector. Fifteen of those compounds are priority pollutant pesticides
and seven are Arochlor PCBs. We classified an additional four
-------
140
pesticides into this group which we determined would be amenable to
analysis by electron capture GC.
SLICE 7
Eight compounds are phosphorus-containing pesticides that can be
determined using GC with a flame photometric detector.
SLIDE 8
The herbicides rated their own special classification. We followed
the traditional approach for these compounds using liquid/liquid
extraction followed by diazomethane methane derivatization and electron
capture GC analysis.
SLIDE 9
A large number of compounds are water soluble volatile organics
that we determined could be analyzed using direct injection GC/MS.
Last year we presented some information at this symposium that showed
that many of these compounds could be successfully analyzed using a
heated purge and trap technique. Using that technique, we were able to
achieve much lower detection limits. Since that time, however, we
haven't devoted sufficient attention to heated purge and trap
methodology. We are therefore continuing to perform the analysis by
direct aqueous injection GC/MS. A number of the compounds on DAI/VOA
-------
141
list were not amenable to analysis using heated purge and trap, which
supports our reasons for using direct aqueous injection instead.
The most difficult task that we faced after categorizing these
compounds into the different methods that we could use for their
analysis was to find standard reference materials. We were not under
the same restrictions that Sam was. We were not required to use the
EPA repository, and as a result we were free to go to any source that
we could find.
Nonetheless, it was still extremely difficult to obtain many of
the compounds on the Appendix VIII list. I'd estimate of the
approximately 250 compounds that we classified, we could not find
sources for about one third of them.
Our next task, after procuring the standards, was to determine
whether we could actually analyze the compounds using the scheme we
proposed. Our first step in this process was to perform instrument
detection limit experiments which was equivalent to determining whether
these compounds could be analyzed by GC/MS. This work was performed
using standards only. It was not performed using spikes, and it was
done only for the Appendix VIII compounds, which were not on the
priority pollutants list. We borrowed detection limit data that had
previously been developed using Methods 624 and 625 for those priority
pollutants that were also listed as Appendix VIII targets. For those
-------
142
specific priority pollutant an instrument detection limit study was not
performed.
Using the instrument detection limit data developed for the
Appendix VIII compounds, we began spiking experiments to determine if
we could actually recover the compounds which had been assigned to
various categories using the analytical scheme designed for those
catagories. These spiking experiments used the IDL values that we had
obtained in our initial studies as the starting point. If we were not
successful with the initial concentration we had used for spiking, we
increased the spiked concentration until we were either successful or
we decided further increase was not justified because of previously
unknown technical concerns.
SLIDE 10
After going through the process of obtaining standards and
performing our IDL and spiking experiments, there were 28 compounds
remaining that we had categorized into analytical schemes that we were
experiencing extreme difficulties analyzing. Whether these
difficulties were the result of chromatographic problems associated
with compound polarity or thermal lability has not been investigated.
However, I'm certain that some of these problems were extraction
problems as well, and for lack of a better term, I've classified these
28 compounds as the Appendix VIII mystery organics.
-------
143
After performing the spiking experiments, we determined we were
ready to offer an Appendix VIII analytical scheme to clients. Our
system at ETC is somewhat different than most laboratories. When we
analyze batches of samples, the batches can range in size anywhere from
one sample to 16 samples. In that analysis we include a spiked blank
and a spiked matrix sample, which gives us two valuable pieces of QA
information. We can get recovery data for both spiked blanks and
spiked matrices.
About six months ago, our systems specialist, Jim Bower, designed
a QA data base which would allow us to automatically transfer processed
GC/MS data files from the GC/MS data system directly to an HP3000
computer. In the 3000 we can manipulate the information in a wide
variety of modes to obtain statistical information on any facet of
GC/MS analysis.
The construction of that data base gave us the opportunity to
compile some very valuable information on Appendix VIII analysis. T^.at
information consists of precision and recovery data for Appendix VII'I
compounds that previously had not been accumulated on a large scale.
I'll be focusing on this data for the next few minutes.
SLIDE 11
What I did initially was to quickly compare precision and accuracy
data for the priority pollutants which were common to the Appendix VIII
-------
144
list that had been analyzed using Method 8240 and Method 624. There's
some important pieces of information I'd like to point out to you on
this slide. During the study period for the Method 624 analytes we
were able to obtain all our comparison data within a relatively short
period of time. The reason for this is that our laboratory performs
many more priority pollutant analysis than we do Appendix VIII
analysis.
The length of the study period needed to obtain 15 data points
from samples which had been analyzed for Appendix VIII compound was
about three months. Excuse me, Denis is pointing out an error on my
slide here. This date is not 2/24/85, it 2/24/86, so the study period
is actually one week instead of one year and one week.
The spiking concentrations that we used for the priority
pollutants on Method 624 and Method 8240 were based on the detection
limits listed for the 600 methods. Another item that, I also would
like to point out to you is that it appears as though we've got better
precision for Method 624. The reason for that is when we intially
started collecting data on Method 8240, we were using a single internal
standard. All method 624 analysis is performed using multiple internal
standards which yeilds better precision.
If you notice, there's a few compounds on the list that show
especially good precision, in fact, better precision than Method 624,
these compounds happen to elute very close to the internal standard,
-------
J.45
which accounts for the better precision. As far as accuracy is
concerned, it's apparently equivalent but statistical equivalency has
not been determined for this data.
SLIDE 13
The data that I think you're most interested in seeing is for the
Appendix VIII compounds that are not common to the priority pollutant
list. Ordinarily when using an internal standard purge and trap
method, I'd expect to get good accuracy, or accuracy in the range of
about 80 to 120 percent recovery for all compounds. For a few of these
compounds, I'm getting much lower accuracy than I expected. I don't
really have a good explaination for that at this time. The focus of
this presentation is not to discuss the specifics of why any one
compound failed within this analytical scheme. It is geared to display
the type of data obtainable for compounds on the Appendix VIII list
using the analytical scheme I described to you earlier.
SLIDE 14
We also did the same comparisons for the extractable priority
pollutants. Again, I did a quick comparison of spiked blank data from
Method 625 with the data from the modified Appendix VIII scheme. The
compounds on this slide are common to the priority pollutant and
Appendix VIII list.
-------
146
You will notice that the number of data points chosen for the
Appendix VIII analytical scheme is smaller than those for the priority
polllutants. While preparing for this presentation, I found that
recovering some of the older data from our data base was going to be
more difficult than I had originally thought. The reason for this is
that the individual data points would have had to have been entered
manually into data base to be eligible for this study. As a result I
have a smaller data base to deal with for the data from the
modification of Method 8270 I think, however, that we've got enough
information from the reduced size data base to illustrate trends.
The precision of Method 625 and Modified 8270 seems to be almost
identical. In some cases, the precision for the Appendix VIII scheme
comes out better for some compounds. I'm trying to avoid making
statements on statistical significance since I have not applied any
statistical significance tests to the precision data, but in a
qualitative sense we do find that a few compounds exhibit a little bit
better precision. In some cases, we have also noted higher recovery
for the modified Appendix VIII extraction scheme.
SLIDE 15
Again, we have applied the same comparisions to the matrix spikes.
This time, we see that better precision favors the Method 625 scheme,
which seems to make a statement on whether the precision differences we
are observing are actually significant or not. I think from the
-------
147
information I have displayed we can conclude that the SW-846 schemes
for both volatiles and extractables perform satisfactorily for priority
pollutant compounds.
We also did a comparison of spiked blank data and matrix spike
data for the Appendix VIII compounds which are not on the priority
pollutants list. I've selected a cross section of compounds that are
exclusively on the Appendix VIII list to demonstrate the types of
recovery and precision that we are seeing using these techniques.
We see some unusual behavior that I think is important for each one
of you to consider. In many cases, we're seeing large standard
deviations. As a general rule, I'd say the non-polar compounds tend to
perform a little better within the scheme. The polar compounds have a
tendency towards lower recoveries and higher precisions. If you've
been forced into the Appendix VIII mode of monitoring for one reason or
another, I think it's extremely important that you know the value of
the data you're getting and whether you're going to be satisfied wibh
results for a specific compound where the percent relative standard
deviation for its concentration could be as much as 100 percent.
Where does this information I've presented mean for the regulatory
community? Many of the "exotic" compounds and poor performers have
been excluded by the short term guidance that EPA has recently issued.
This program is in its early adolescence. I'm sure that most of you
recall the evolutionary period that the 304(h) methods went through
-------
148
before they were accepted by all of us. I think that the analytical
scheme used for the Appendix VIII program is going to go through the
same type of evolutionary period.
I've displayed precision and recovery data for Appendix VIII
compounds that perform well within the scheme and we've also seen data
for compounds that perform poorly within the scheme and probably would
not be acceptable to any of you. I'm going to leave you with this
final thought and that is, before we can use these methods in a
regulatory or monitoring setting we must be aware of the type of data
that we get from these methods and what type of performance in terms of
precision and accuracy we can expect for any particular analyte. Thank
you.
-------
149
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
MR. FOSTER: Russ Foster from RAI. What was your
rationale for separating the neutral extraction, the basic extraction,
and did it meet the objectives?
MR. SPEIS: Well, we thought that for some compounds
there was a chance that we might hydrolyze them by making the solution
basic before extraction. So, to avoid that, we inserted a neutral
extraction followed by a basic extraction and an acid extraction. As
it turns out, doing a basic and neutral extraction usually recovers all
the compounds that we had spiked. There's very few compounds that end
up in the acidic fraction. I think it's less than five, and it may be
as low as two when you employ this extraction procedure.
AUDIENCE PATICIPANT: In doing this, we realize that
there's going to be problems with these new methods, and you've shovw
this statistically. Well, that's fine and good, but what if we're
analyzing for these analytes for a client, okay? Five years from now
they decide that the methodology we used was substandard in their
finding, you know.
-------
150
MR. SPEIS: This is exactly the point.
MR. TELLIARD: You're in deep trouble is what it is>
MR. SPEIS: How can you make a regulatory decision using
a method that gives you poor precision and accuracy in terms of 100
percent relative standard deviation?
MR. TELLIARD: No, it never stopped the agency in the
past and I don't think it will in the future. I think, going back to
the question, that we've gone through...! mean, it's taken me almost
nine years to convince STanko that GC/MS would work on priority
pollutants, so you're not going to do this overnight.
MR. SPEIS: It's only been three and a half years for
this program.
MR. RICE: Jim Rice. It is interesting to note though,
that the data you have put up is essentially a single lab's QC data.
MR. SPEIS: That's correct.
MR. RICE: All of which goes to say that if you were to
ever try at this stage of the game, an interlaboratory comparison would
be totally off the wall.
MR. TELLIARD: Thanks so much, Dave.
-------
J.5J.
We're running a little late, as usual. For those folks who are
new to this facility, for lunch there's a couple of restaurants in the
hotel. There's a number of restaurants next door at the Waterfront,
some fast food places. You can go make yourself a chocolate sundae,
put a lot of weight on, whatever you want to do, but let's get back
here about 1:30, quarter to., two^ at .'the, latest. . . ,.
: •--'•< .'••'•••..v'" "..- ' ,•-. - '•':"' "'''; • > ""V • ' ' •
Thank you so much for your attention.
(WHEREUPON, a lunch recess was taken.)
-------
152
METHOD PERFORMANCE CHARACTERISTICS
FOR
SELECTED APPENDIX VIII ANALYTES
DAVID N. SPEIS
DENIS C.K. UN
JAMES N. BOWER
Environmental Testing
and Certification Corporation
Edison, New Jersey
-------
153
APPENDIX VIII ANALYTICAL SCHEME
o VOLATILE ORGANICS:
Method 8240, GC/MS
o EXTRACTABLE ORGANICS:
Modified Method 8270, GC/MS
o PESTICIDES & HERBICIDES:
Method 8080, 8140, and 8150, GC/EC, GC/FPD
O WATER SOLUBLE VOLATILE:
Direct Aqueous Injection, GC/MS
o POLAR & THERMALLY LABILE ORGANICS:
HPLC Direct Aqueous Injection
o METALS & METAL COMPLEXES:
Representative Metals by AA & ICAP
-------
154
METALS (22)
ALUMINUM
ANTIMONY
ARSENIC
BARIUM
BERYLLIUM
CADMIUM
CALCIUM
CHROMIUM
COPPER
IRON
LEAD
MERCURY
NICKEL
OSMIUM
POTASSIUM
SELENIUM
SILVER
SODIUM
STRONTIUM
THALLIUM
VANADIUM
ZINC
-------
155
POLAR & THERMALLY LABILE ORGANIC COMPOUNDS
EXTRACTABLE HPLC (20)
WARFARIN
MITOMYCIN C
AZASERINE
BENZIDINE
CHLORAMBUCIL
CITRUS RED NO 2
DAUNOMYCIN
3,3-DJCHLOROBENZIDINE
3,4-Dl H YDROXY-ALPH-(M ETH YLAMINO)-
METHYL BENZYL ALCOHOL
3,3-DIMETHOXYBENZIDlNE
3,3-DIMETHYLBENZIDINE
1-N APHTH YL-2-THIOURE A
m-PHENYLENED«AMINE
o-PHENYLENEDIAMINE
p-PHENYLENDIAMINE
STREPTOZOTOCIN
THIOACETAMSDE
TOLUENE-2.4-DIAMINE
TRYPAN BLUE
ALDICARB
-------
156
POLAR AND THERMALLY LABILE
ORGANIC COMPOUNDS
Direct Aqueous Injection HPLC (18)
1-Acetyl-2-thiourea
Acrylamide
1-(o-Chiorophenyl) thiourea
Diethylstilbesterol/Ethyl carbamate
Ethyleneirnine
Ethylenethiourea
Maleic hydrazide
Malononitridine
Methomyl
2-Methylaziridine
Nicotinic acid
Nitroglycerine
N-Nitroso-N-ethylurea
N-Nitroso-N-methylurea
N-Phenythiourea
Reserpine
Thiourea
-------
157
PESTICIDES & POLYCHLORINATED BIPHENYLS
by
GAS CHROMATOGRAPHY
ELECTRON CAPTURE DETECTOR (26)
15 PRIORITY POLLUTANT PESTICIDES
7 POLYCHLORINATED BIPHENYLS (Arochlors)
CHLOROBENZILATE
DIMETHOATE
KEPONE
METHOXYCHLOR
-------
JL58
PESTICIDES
by
GAS CHROMATOGRAPHY
FLAME PHOTOMETRIC DETECTOR (8)
CARBOPHENOTH1ON
THIONAZIN
DISULFOTON
METHYL PARATHION
PARATH1ON
PHORATE
FAMPHUR
TETRAETHYLPYROPHOSPHATE
-------
159
HERBICIDES
by
GAS CHROMATOGRAPHY
ELECTRON CAPTURE DETECTOR
2,4-DICHLOROPHENOXYACETIC ACID
2,4,5-TRICHLOROPHENOXYACETIC ACID
2,4,5-TP (SILVEX ACID)
-------
160
WATER SOLUBLE COMPOUNDS
Direct Aqueous Injection GC/MS (15)
ALLYL ALCOHOL
CHLORAL
CHLOROACETALDEHYDE
2,3-DICHLOROPROPANOL
1,4-DIOXANE
ETHYL CYANIDE
ETHYLENE OXIDE
FLOUROACETIC ACID *
GLYCIDYLALDEHYDE
ISOBUTYL ALCOHOL
METHACRYLONITRILE
METHANETHIOL
N-NITROSOPYRROUDINE
2-PROPYN-1-OL
PYRIDINE
-------
161
APPENDIX VIII MYSTERY ORGANICS
Purgeable Organics (6)
Acetontrile
1-Chloro-2,3-epoxypropene
1,2:3,4-Diepoxy butane
Paraldehyde
Crotonaldehyde
Methyl hydrazine
Extractable Organics (22)
5-(AminomethyI)-3-isoxazoloI
Auramine
Dibenzo (aj) pyrene
DlisopropylfSuorophosphate
Thiofanox
a-a-Dimethylphenethylamine
Dimethyl phthalate
Dimethyl sulfate
Formic acid
Hydrazine
Maleic anhydride
Benzenethioi
Melphaian
2-Methyllactonitrile
N-Methyl-N-nitroso-
quandine
Methylthiouracil
N-Nitrosodiethanolamine
Endothal
n-Propylamine
Saccharin
Thiuram
tris-(2,3-Dibromoproplyl)
phosphate
io
-------
162
PRIORITY POLLUTANT VOLATILE ORGANICS
SPIKED BLANK
METHOD 624 METHOD 8240
METHYL CHLORIDE
VINYL CHLORIDE
ACRYLONITRILE
DICHLOROBROMOETHANE
CHLOROD1BROMETHANE
TR1CHLOROETHYLENE
1,1,2-TRICHLOROETHANE
BROMOFORM
106
95
94
95
100
87
1,1,2,2-TETRACHLOROETHANE1Q6
N = 19
STUDY PERIOD 2/24/86 - 3/2/86
SD
SD
16
8.7
9.9
8.5
9.3
9.2
9.3
12
19
97 28
107 19
93 22
95 6.1
94 9.7
97 5.9
99 10
89 10
108 16
N = 15
11/14/85 - 2/17/86
Spiked Concentrations
18 - 50 ug/I
-------
163
PRIORITY POLLUTANT VOLATILE ORGANICS
SPIKED MATRIX
METHOD 624
METHYL CHLORIDE
VINYL CHLORIDE
ACRYLONITRILE
DICHLOROBROMOMETHANE
CHLORODIBROMETHANE
TRICHLOROETHYLENE
1,1,2-TRICHLOROETHANE
BROMOFORM
1,1,2,2-TETRACHLOROETHANE
X
103
101
99
101
96
95
102
87
115
SD
18
16
15
16
15
13
12
19
24
N = 19
METHOD 8240
~x SD
106 22
106 18
105 57
99 13
99 22
114 35
111 16
92 27
105 31
N = 12
Spiked Concentration 18-50 ug/I
-------
164
APPENDIX VIII VOLATILE ORGANICS
METHOD 8240
1.2-DIBROMO-3
-CHLOROPROPANE
METHLY METHACRYLATE
1,2-DIBROMOMETHANE
1,2-DICHLOROPROPANE
2,3-DICHLOROPROPENE
METHYL ETHYL KETONE
DIBROMOMETHANE
FREON TF
BLANK SPIKE MATRIX SPIKE
x SD x SD
40 24 62 38
99
105
99
97
87
93
68
17
8.8
5.0
4.8
31
24
35
108
110
106
98
92
99
54
22
25
25
22
47
13
19
= 15
= 12
SPIKED CONCENTRATION
50 ug/l
-------
165
PRIORITY POLLUTANT EXTRACTABLES
SPIKED BLANK
Hexachioroethane
Hexachlorobutadiene
2-Chloronapthaiene
N-Nitrosodi-n-propylamine
Butyl benzyl phthalate
2,4-Dinltrotoluene
Bis (2-Chloroethoxy) methane
Pyrene
Benzo (a) anthracene
Chrysene
Phenol
2,4-Dinitrophenol
P-Chloro~M-Cresol
Method 625
X
55
59
83
91
77
89
86
85
85
87
47
78
SD
17
17
18
20
17
17
15
18
14
15
19
25
Appendix VIII
X
57
68
121
90
94
110
100
85
96
95
62
110
75
SD
8.8
9.1
41
18
19
20
20
15
6.8
12
22
21
38
n=18
n=6
Study Period
1/29/86 - 2/24/86 7/12/85 - 9/15/85
Spiked Concentration 100 ug/l
-------
166
PRIORITY POLLUTANT EXTRACTABLES
SPIKED MATRIX
Hexachloroethane
Hexachlorobutadiene
2-Chloronapthalene
N-Nitrosodi-n-propylamine
Butyl benzyl phthalate
2,4-Dinitrotoluene
Bis (2-Chloroethoxy) Methane
Pyrene
Benzo (a) anthrocene
Chrysene
Phenol
2,4-Dinitrophenol
P-Chloro-M-Cresol
Method 62S Appendix VIII
X SD X SD
62
65
84
94
78
89
89
89
84
86
50
73
19
15
13
21
14
23
12
24
11
15
21
37
48
59
86
83
99
99
83
77
87
88
99
31
27
54
21
17
32
27
32
18
19
32
n=18
n=6
Spiked Concentration 100 ug/l
-------
167
APPENDIX VIII EXTRACTABLE ORGANICS
1,2,3,4-Tetracfolorobenzene
2,3,5,6-TetrachIorophenol
2,4-Dithiobiuret
2-AcetyIaminofluorene
2-sec-Butyl-4,6-dinitrophenol
3-Chloropropionitrile
Benzotrichloride
Benzyl chloride
Diallate
Dibenzo(a,j)acridine
Dichloromethylbenzene
Hexachloropropene
N-Nitroso-N-methylurethane
Pentachioronitrobenzene
Pentachlorophenol
o-Cresol
SPIKED BLANK
If
82
59
84
21
84
>l 67
38
4
66
111
140
36
40
30
114
65
84
SD
46
42
78
28
53
42
22
6
4
18
89
33
28
31
25
54
16
SPIKED MATRIX
~x SD
92 66
19 32
79 55
32 19
107
87
37
30
28
103
64
73
48
79
36
25
33
65
48
20
N = 5
N = 6
Spiked Concentration*
100-300 ug/l
-------
168
MR. TELLIARD: Starting.off
our afternoon session, we have Tina Engel to talk
about pesticides determination in groundwater.
-------
169
TINA ENGEL
BATTELLE MEMORIAL INSTITUTE
DEVELOPMENT OF COMPREHENSIVE ANALYTICAL METHODS
FOR THE DETERMINATION OF PESTICIDES IN GROUNDWATER
MRS. ENGEL: Considerable
attention has been given the problem of contamination
of groundwater resources. In lieu of a National
Groundwater Survey to be implemented jointly by the
Office of Drinking Water and the Office of Pesticides
Programs, Battelle is conducting a study for the
development of comprehensive analytical methods for
pesticides in groundwater.
The groundwater analysis methods development
study is funded jointly by the Office of Drinking
Water and the Office of Pesticides Programs through a
contract with the Environmental Measurement Support
Laboratory in Cincinnati, Ohio.
The purpose of this study is to develop and
validate screening methods for pesticides and pesti-
cides metabolites in groundwater samples. These
validated methods would be made available for use in
the analysis of groundwater samples collected for the
National Groundwater Survey.
-------
170
Specific goals of the methods development study
include the ability to detect and quantify targeted
pesticides and pesticide metabolites at sub part per
billion levels in groundwater. Many of the pesticides
were included in the scope of the study because of
their proven or suspected toxic or carcinogenic
properties. In many cases, pesticide toxicity
information was suspected, so low detection limits in
groundwater were desirable as a precautionary measure.
During methods development, efforts were con-
tinuously made to consolidate and simplify analysis
methods. The potential scope of a groundwater analysis
survey is enormous. Methods are designed to incor-
porate as many pesticides as possible, yet are stream-
lined to simplify and ruggedize the methods.
Survey ground water samples will be analyzed by
different laboratories. For this reason, efforts
were made to ruggedize methods. The goal was to
simplify or ruggedize the method in order to minimize
variability of analysis results between laboratories.
After development and preliminary evaluation of
the analysis methods, the methods will be thoroughly
validated by determining analyte method detection
limits and method ranges and by evaluating potential
matrix effects on method results. .Validation of the
-------
171
groundwater analysis methods are now underway.
The scope of the groundwater analysis methods
development effort was defined by the Office of
Drinking Water and Office of Pesticides Programs
to include the determination of approximately 90
priority pesticides as chosen by an EPA Analytes
Selection Task Group. Additionally, inclusion of
approximately 80 non-priority pesticides to the
methods development effort. Third, development of
methods to look for these approximately 170 pesticides
in groundwater matrix.
Groundwater samples were not normally expected
to contain large levels or high levels of interfering
organics. Although a confirmation technique or
techniques were included in methods development
scope, methods development efforts were not designed
to include a cleanup method. Methods development
studies were conducted using reagent water matrices.
Validation efforts will include the use of groundwater
matrices.
A set of selected criteria were used by the
Analyte Selection Group to formulate the list of
priority pesticides for the methods development
effort. First, a list was compiled of known active
ingredients of pesticides currently or recently in
-------
172
use. This master list was then limited to those
compounds demonstrating physical properties of interest
to the groundwater survey. Properties evaluated in-
cluded adequate solubility of the pesticide In water;
adequate stability of the pesticide in water (hydro-
lysis data were considered when they were available);
volatility of the pesticide; distribution of the
pesticide between soil and water; mobility of the
pesticides through soil; and speciation of the pesti-
cide. Identified pesticide metabolites and decompo-
sition products, especially those displaying un-
desirable toxilogical properties, were included as
priority analytes.
The list of methods development analytes was
expanded to include approximately 80 non-priority
pesticides. Criteria used for selection of non-
priority pesticides included any analyte included in
any EPA 600 series method referencing a priority
pesticide. For instance, aldicarb was designated by
the analyte selection group as a priority pesticide.
Aldicarb is included as an analyte of EPA Method 531,
which is entitled, Measurement of N-Methyl Carbamoyl-
oximes and N-methyl Carbamates in Drinking Water by
Direct Aqueous Injection HPLC with Post Column
Derivatization. All other Method 531 analytes were
-------
173
included in the scope of the groundwater survey as
non-priority analytes.
The non-priority analytes were included in the
scope of the groundwater survey only if initial
methods development studies indicated no difficulties
in analysis or detection of the analyte. Non-priority
analytes were included in analysis methods designed
for the determination of priority analytes in
groundwater. Extraordinary efforts were not made to
include non-priority analytes in the survey. However,
since the non-priority analytes were already included
in one of the EPA 600 series methods, few difficulties
were encountered with non-priority analytes during
the methods development study.
Groundwater analysis methods were based on the
EPA 600 series methods. Methods incorporated extraction
and analysis techniques used in the 600 series methods
including extraction of analytes from groundwater
using separatory funnels; concentration of resultant
organic extracts using Kuderna-Danish equipment and
techniques; and finally, analysis of concentrated
sample extracts by packed column gas chromatography
using electron capture or nitrogen-phosphorus detectors,
or by high performance liquid chromatography using
ultraviolet or post-column derivatization detectors.
-------
174
EMSL 600 series methods used as a basis for
groundwater analysis methods development studies
included Methods 531, 608, 614, 615, 619, 622, 632,
633, 634, 643 and 645.
Five analysis methods were developed for the
determination of the 170 priority and non-priority
analytes in groundwater. The first method incorporates
separatory funnel extraction of the neutralized water
sample with triplicate 60 mL aliquots of methylene
chloride. The organic extract is dried and concentrated
using Kuderna-Danish techniques. The resultant
concentrated extract is analyzed by packed column gas
chromatography using a nitrogen-phosphorus detector.
Method 1 can be used to determine approximately
75 nitrogen- or phosphorus-containing pesticides and
pesticide metabolites in groundwater. Method 2 uses
the same extraction and concentration techniques as
Method 1. The resultant extract is analyzed by packed
column gas chromatography using an electron capture
detector. Method 2 can be used to determine
approximately 28 halogen-containing pesticides and
pesticide metabolites in groundwater.
The extraction and analysis procedures used in
Analysis Method 3 are based on EPA Method 615. Acids
and acid esters are extracted from the acidified water
-------
175
sample using separatory funnel extraction with 360 mL
aliquots of ethyl ether. Acid esters are converted
to the parent acid by base hydrolysis. The acids are
reextracted into ethyl ether and the extract is
concentrated using Kuderna-Danish techniques.
After methylation with diazomethane, the
resultant extract is analyzed by packed column gas
chromatography using an electron capture detector.
Method 3 has been used to determine approximately 26
halogen-containing acidic pesticides and pesticide
metabolites in groundwater.
In order to determine the analytes covered by
these first three GC methods it is necessary to use
three different packed columns; a non-polar column
packed with 1.5 percent OV-17/1.95 percent QF-1; a
moderately polar column packed with 3 percent SP-
2250; and a polar column packed with 5 percent Carbo-
wax 20M-TPA.
The fourth method incorporates separatory funnel
extraction of the neutralized water sample with
triplicate 60 mL aliquots of methylene chloride. The
organic extract is dried and concentrated using a
rotating evaporator. The resultant extract is analyzed
by reverse phase high performance liquid chromatography
using an ultraviolet detector. Method 4 can be used
-------
176
to determine approximately 21 pesticides and pesticide
metabolites in groundwater.
Method 5 is a direct injection HPLC procedure
based on EPA Method 531. The aqueous sample is
injected directly onto a reverse phase HPLC column.
Analytes are hydrolyzed with 0.5 N sodium hydroxide
at an elevated temperature after elution from the
column. The methylamine formed during hydrolysis is
reacted with 2-mercaptoethanol and ortho-phthalaldehyde
to form a highly fluorescent derivative which is
detected using a fluorescence detector. Method 5 can
be used to determine approximately 12 N-methyl
carbamoyloximes and N-methyl carbamate pesticides and
pesticide metabolites in groundwater.
After initial methods development studies, two
major modifications were made in the analysis methods.
The first modification was the substitution of a
tumbling liquid/liguid extraction procedure for the
sequential separatory funnel partitioning used in
most of the water analysis methods. The second
modification was the use of capillary columns instead
of packed columns in methods using gas chromatography
for analyte identification and quantification.
The tumbling extraction technique involves
tumbling the one liter aqueous sample with 300 mL of
-------
177
organic solvent end over end for one hour. The sample
and extraction solvent are placed in a sealed 1.7 L
culture bottle and a mechanical device is used to
turn the bottle reproducibly for a predetermined
length of time.
Substitution of the tumbling extraction technique
for the separatory funnel extraction technique was
based on mathematical calculations and later demon-
strated with real water samples. The partition
t
coefficient of an analyte between water and organic
solvent can be easily calculated. The aqueous sample
is equilibrated with solvent and the concentration of
the analyte in the organic phase can be determined.
The analyte coefficient, Kdf can be calculated
from the original amount of the analyte in the aqueous
sample, Ao, the amount of the analyte in the solvent
after partitioning, As, and the volumes of the two
phases, Vw and Vs. Calculation of Kd is simplified
when the equation is put in terms of analyte recovery
into the organic phase, R. The partition coefficient
can then be calculated in terms of percent recovery
and phase volumes. The assumption is made throughout
these calculations that analyte equilibrium between
the aqueous and organic phases is reached prior to
any recovery measurements.
-------
178
The expected analyte percent recovery from a
single partitioning can be calculated from the
partition coefficient and the phase volume ratio, Vr,
which is the ratio of the volume of the organic phase
to the volume of the aqueous phase.
The expected analyte percent recovery from
triplicate equilibrations with equal volumes of
organic solvent can just as easily be calculated from
the analyte partition coefficient and the volume ratio
from each equilibration.
Using the prior recovery equations, expected or
ideal recoveries can be calculated for partition
coefficient values representing a wide range of
analytes. This table contains calculated percent
recoveries for a fictitious group of analytes with
partition coefficient values varying from 1 to 80.
Calculations were made for a one liter aqueous sample,
assuming either triplicate separatory funnel extrac-
tions with 60 mL each of an organic solvent, or a
single tumbling partitioning with 300 mL of the
identical organic solvent.
Lower partition coefficient values indicate an
analyte which is more water soluble and thus harder
to extract from the water into the organic solvent.
As expected, calculated partition recoveries are
-------
179
lower for analytes having lower partition coefficients.
However, in no case does the calculated recovery
using a single tumbling partition vary from the
separatory funnel results by more than seven percent.
The experimental comparison of the extraction
techniques was conducted with many pesticides. Some
results are demonstrated in this table. These data
were generated from reagent water samples spiked with
the listed pesticides at the low parts per billion
level. Methylene chloride was used as the organic
phase. In most cases, comparable or superior recoveries
were observed for most compounds when the tumbling
extraction procedure was used. In most cases,
reproducibility of the measurements, expressed as the
percent relative standard deviation, was greatly
reduced. In other words, the procedure was more
repeatable.
The use of tumbling extraction provides several
advantages over the use of separatory funnels.
Tumbling is less time and labor consuming. Sample
processing is limited only by the equipment avail-
ability, specifically the number of available tumblers.
Use of separatory funnels requires a person to
physically shake the apparatus for the entire
equilibration period.
-------
180
Tumbling results in increased reproducibility.
The bottle motion, and thus the mixing of the two
phases, is mechanically controlled. Partitioning in
a separatory funnel is inherently dependent on the
operator. Last, tumbling results in an increased
method ruggedness evidenced by lower variability of
method results.
The second major methods modification was the
substitution of capillary column GC for the packed
column techniques originally evaluated. A 30 meter
by 25 millimeter ID SPB-5 fused silica capillary
column was used instead of the three packed columns
originally needed for the three gas chromatography
analysis methods.
Several advantages were obtained from using the
capillary GC column. The modified methods were more
sensitive for the method analytes. Use of capillary
columns resulted in sharper peaks leading to lower
analyte detection limits.
All analytes could be determined using only one
column. Using packed column would require analysis
of all sample extracts on all three packed columns.
Alternatively, one injection on a single column is
required when the capillary columns are used.
Finally, capillary columns demonstrate superior
-------
181
resolution properties. Lower analyte coelutions
allow for a more effective screening of the sample
extracts.
The advantages of using capillary columns are
demonstrated in the next two slides. This first
chromatogram was generated during the analysis of 13
nitrogen- and phosphorus-containing pesticides on a
non-polar packed column using a nitrogen-phosphorus
detector. Addition of other pesticides to the mixture
would most likely cause coelution problems.
This second chromatogram was generated during
the analysis of 34 pesticides on an SPB-5 fused silica
column. Although this mixture contains approximately
three times as many pesticides as that shown in the
previous chromatogram, the capillary column gas
chromatogram is much less complicated and crowded
than the packed column gas chromatogram.
In summary, five analysis methods have been
developed for the determination of approximately 170
pesticides in groundwater. These methods are suitable
for the screening and quantification of trace
levels of these analytes. The methods, originally
based on EPA 600 series methods, have been modified
to include tumbling extraction techniques and capillary
column gas chromatography where applicable. These
-------
182
modifications have resulted in improved method
repeatability and ruggedness. Further, method
ruggedization and methods validation studies are
ongoing.
Thank you.
-------
183
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
MR. TUROSKI: Victor
Turoski, James River Corporation. Did you notice any
difference in limited detection when you went from
packed columns to capillary columns, since you can
obviously inject 20 microliters on a packed column
and maybe one on capillary?
MRS. ENGEL: With the
detectors that we were looking at as far as the ECDf
no, because the detector really defines the linear
range that you have, and we found that you can only
really do about one and a half orders of magnitude
anyway with an BCD, which is what most of our work
was done with.
MR. TUROSKI: What level
spikes were these?
MRS. ENGEL: The work that
we did was usually 1:10 part per billion in the water
sample. We are actually going to be validating
the method at sub part per billion.
MR. TUROSKI: My concern is
that you can see a lot less. Your limited detection
should theoretically be less or your limited
-------
184
detection should be theoretically more on the packed
column. If I were going for limited detection, I'd
certainly use a packed column.
MRS. ENGEL: But these are
screening methods and they're trying to look for
everything in one run. They're trying to really
limit the costs of the analyses/ and if that's the
case, I don't see how you can use three different
packed columns and save any money.
MR. TUROSKI: I understand.
MR. TELLIARD: We just
don't want these labs to get fat on us, you know what
I mean? They're already just rolling.
MR. CALDWELL: I'm Chan
Caldwell, International Technology. Have you tried a
wider bore column, and do you expect that you'd get the
same type of results? Certainly, in regard to this
gentleman, you would be able to maintain higher
capacities in such a wider bore, say a .3 or .5 milli-
meter ID column.
MRS. ENGEL: The wide bore
column certainly would give you some kind of inter-
mediate results between using the packed and the
narrower bore capillary. No, we really haven't
looked into it. I suppose it depends on what's more
-------
185
important, the higher resolution capabilities or
being able to inject more sample and have higher
capacities.
MR. MOSESMAN: Neil
Mosesman, Supelco Incorporated. Did you do
any optimization of the amount of solvent in
the extraction technique? I see you're using
almost twice as much solvent in the tumbling technique
versus liquid, you know, separatory funnel extraction.
MRS. ENGEL: We found that
when you do the mathematical calculations, that the
less solvent you use for the single partitioning or
equilibration, obviously the lower your recoveries are
going to be; 300 mL seemed to pretty closely mimic
what you could do with triplicate extractions with
smaller volumes of solvent.
MR. MOSESMAN: But that
also means you've got twice as much solvent to
concentrate down.
MRS. ENGEL: Yes. That
didn't seem to pose any problems as far as using
Kd concentration techniques. If we used more than
300 mL we really didn't get any added recovery
advantages, and I suppose it would be harder to
handle, Kd concentration-wise.
-------
186
MR. APRIL: You had two
HPLC methods at the end. Are there any general
observations you can make about the HPLC methods as
contrasted with the GC methods?
MRS. ENGEL: The HPLC
methods were limited to those compounds that we could
not do by GC, the compounds that could be done by
Method 531, which is the post-column derivatization
detection technique, we left in that method because
it is highly specific for those compounds and it gave
good detection limits.
All compounds in Method 4 were screened by GC
and could not be done by GC. We were instructed in
this study that gas chromatography was the preferred
method simply because the detectors that were available
for use with the GC were more specific.
MR. APRIL: Can you comment
on the results you got from HPLC as contrasted with
the results for GC in terms of detection limits and
interferences?
MRS. ENGEL: Detection
limits for the PCD procedure were generally in the
low part per billion level. They did not get into
the sub part per billion level in water. The UV
methods we could go very low. In fact, in some cases
-------
187
you can go lower than the GC methods. But it probably
will be more likely to have interference problems
when we actually start looking at real samples.
MR. WHITLOCK: Stu Whitlock
from ESE. Did you try any cleanup techniques on
these things?
MRS. ENGEL: No, we didn't
try any cleanup techniques at all. That was not part
of the study. The general tack that's been taken
with these, or will be taken with these, samples is
that interferences are not really expected in most of
the samples, and there is going to be a backup analysis
technique or a confirmation technique. In other
words, either using mass spectrometry for confirmation
purposes or another column.
DR. GAIND: Arun Gaind from
Nanco. Liquid solid extraction procedures are up and
coming thing in preparatory chemistry. Are you
planning to try that in the future or no?
MRS. ENGEL: Would you like
to fund the study?
DR. GAIND: No, I am not
U.S. EPA, thank God.
MRS. ENGEL: Anybody? I
would love to, I would love to. Definitely. It has
-------
188
a lot of potential advantages if it can be demonstrated.
MR. SLOAN: You said you
used methylene chloride as extracting solvent. From
the BCD methods I know you had to change to something.
Was it hexane or iso-octane, and did you see any
difference between the two of them?
MRS. ENGEL: We used methyl
di-butyl ether, and we found generally that
gives you can do a solvent substitution with it
and it gives better recoveries because you don't have
to go to as high a temperature in the Kd, to use the
Kd concentration. And that's usually the solvent we
wind up in before we do any mass procedure.
MR. SLOAN: What about the
effects on the people who work with it?
MRS. ENGEL: We use hoods.
MR. TELLIARD: Thank you,
Tina. Thank you so much.
MRS. ENGEL: Thank you.
-------
I8y
DEVELOPMENT OF COMPREHENSIVE
ANALYTICAL METHOD FOR
PESTICIDES IN GROUND WATER
BY
T.M, ENGEL AND J.S. WARNER
-------
190
STUDY SUPPORTED
BY
THE OFFICE OF DRINKING WATER
AND
THE OFFICE OF PESTICIDES PROGRAMS
-------
191
STUDY PURPOSE
DEVELOP AND VALIDATE SCREENING METHODS FOR PESTICIDES
IN GROUND WATER IN SUPPORT OF ODW'S NATIONAL GROUND
WATER SURVEY
-------
192
PROGRAM GOALS
0 DETECT AND QUANTIFY PESTICIDES AT SUB PARTS-
PER-BILLION LEVELS IN GROUND WATER
0 CONSOLIDATE AND SIMPLIFY METHODS
0 RUGGEDIZE METHODS
0 VALIDATE METHODS
-------
193
STUDY SCOPE
0 DETERMINATION OF APPROXIMATELY 90 "PRIORITY"
PESTICIDES
0 INCLUSION OF APPROXIMATELY 80 "NONPRIORITY"
PESTICIDES
0 GROUND MATER MATRIX
-------
194
PRIORITY PESTICIDE
SELECTION CRITERIA
0 KNOWN ACTIVE INGREDIENTS
0 PHYSICAL PROPERTIES
WATER SOLUBILITY
STABILITY (HYDROLYSIS)
VOLATILITY
SOIL/WATER DISTRIBUTION
MOBILITY
SPECIATION
-------
195
NONPRIORITY PESTICIDE
SELECTION CRITERIA
0 ANY ANALYTE INCLUDED IN ANY EMSL-EPA 600-METHOD
REFERENCING A PRIORITY PESTICIDE
0 EASY INCLUSION IN A DEVELOPED METHOD
-------
196
METHODS DESCRIPTION
0 BASED ON EMSL/EPA 600-METHOD SERIES
SEPARATORY FUNNEL EXTRACTIONS
CONCENTRATION USING KUDERNA-DANISH
EQUIPMENT
ANALYSIS BY PACKED COLUMN GC (ECD OR NPD)
OR BY REVERSE PHASE HPLC (UV OR PCD)
-------
197
EMSL600-SERIES METHODS USED
FOR PRIORITY ANALYTES
531,608, (608,1,608,2), 614, (614,1), 615,619,622
(622,1), 632 (632,1), 633 (633,1), 634,643,645
-------
198
GROUND WATER ANALYSIS METHODS
METHOD
NO.
1
2
3
4
5
SAMPLE PREP CONDITIONS
SEPARATORY FUNNEL/NEUTRAL PH
SEPARATORY FUNNEL/NEUTRAL PH
SEPARATORY FUNNEL/ACID PH/
HYDROLYSIS/METHYLATION
(METHOD 615)
SEPARATORY FUNNEI7NEUTRAL PH
NONE
ANALYSIS CONDITIONS
PACKED COLUMN GC-NPD*
PACKED COLUMN GC-ECD*
PACKED COLUMN GC-ECD*
HPLC-UV
DIRECT INJECTION HPLC-PCD
•3 PACKED COLUMNS USED
-------
199
METHODS MODIFICATIONS
0 TUMBLING EXTRACTION
0 CAPILLARY COLUMN GC
-------
200
TUMBLING EXTRACTION
TUMBLE 1-L SAMPLE AND 300 ML SOLVENT END-OVER-END
FOR 1 HOUR
-------
201
CALCULATION OF PARTITION
COEFFICIENT (Ko)
KD - Cs - As x Vw
Cw Aw x Vs
As x Vw
(Ao - As) x Vs
R - 100_x As
flo
KD =
VM
Vs x [(100/R) - 1]
-------
CJJLAT.
MRIES
R = 100 x
(VR x KD)
TVRXKD) + 1
202
-------
203
R = 100 x
1- (fv 1 . . V
I [KD X VRj + I )
& «
-------
MATHEMATICAL COMPARISON
204
% RECOVERY
KD
1
5
10
20
40
80
SEPARATORY FUNNEL
16
54
76
91
97
99
TUMBLE
23
60
75
86
92
96
-------
205
EXPERIMENTAL COMPARISON
ANALYTE
AMETRYN
DISULFOTON SULFONE
METHYL PARATHION
PRONAMIDE
ALPHA-CHLORDANE
DIELDRIN
METHOXYCHLOR
% RECOVERY
StKAKAIUKY hUNNtL
88 ± 17
54 ±25
110 ±14
89 ± 15
68 ± 1
84 ±2
111 ± 11
(±2 RSD)
I UMttLtk
97 ± 5
99 ± 1
99 ±8
82 ±6
80 ± 4
82 ±3
97 ±4
-------
206
ADVANTAGES OF USING
TUMBLING EXTRACTION
0 LESS TIME-CONSUMING
0 INCREASED REPRODUCIBILITY
0 INCREASED RUGGEDNESS
-------
207
CAPILLARY COLUMN GC
SUBSTITUTION OF A 30 M x 25 MM ID SPB-5 FUSED SILICA
COLUMN FOR PACKED COLUMNS
-------
208
ADVANTAGES OF USING
CAPILLARY GC COLUMNS
0 MORE SENSITIVE
0 ONLY ONE COLUMN REQUIRED FOR ALL ANALYTES
0 SUPERIOR RESOLUTION CAPABILITIES
-------
209
-------
210
-------
209
-------
210
-------
211
SUMMARY
0 FIVE ANALYSIS METHODS DEVELOPED FOR THE
DETERMINATION OF 170 PESTICIDES IN GROUND WATER
0 TUMBLING EXTRACTION AND USE OF CAPILLARY COLUMNS
FOUND TO GIVE SUPERIOR RESULTS
0 FURTHER METHOD RUGGEDIZATION AND METHODS
VALIDATION ONGOING
-------
212
MR. TELLIARD: Before our next
speaker I have a couple of cleanup things that I'm
supposed to do. Is Jim King here? This morning's
speakers, those folks who have transparencies, if
you'll get them to Dr. King, he'll take them
downstairs and run a copy of them and give them
back to you before you leave today, Jim is in
the back corner, so that we have them for the
transcript. Did I do right? Good. Thank you.
Our next speaker is Ted Handel from Centec.
He's going to talk about some very, very expensive
data. I paid for it. I've never seen a platinum
ICP before but they said that's what it costs to
run them." Ted's going to talk about some metals
analysis.
-------
213
EDWARD HANDEL
CENTEC ANALYTICAL SERVICES
RAPID MULTI-ELEMENT SCREENING USING SEQUENTIAL ICP
DR. HANDEL: Thanks, Bill.
Basically, what I'd like to talk to you about today is
some preliminary work we've done with our new ICP.
Basically, the question that we're going to try and
answer today is what is SuperScan. If you take a
look in all your manuals for your various ICPs, you
won't find the term SuperScan, so don't bother. It's
something that we've just coined for lack of a three
letter word to call it by.
Basically, SuperScan is a way of screening an
environmental sample for the presence of up to 73
different elements. Since we chemists are fairly
curious types, we're always looking to see what's
actually in those samples. Take Bill, for example.
The first thing he said when he heard about SuperScan
is, a little smile came to his face, "You can measure
all that, huh?" The second thing, he says, "What's
this? Why can't we get that?" And he says, "If we
can get it, let's find a way." Well, we didn't, but
we might still work at it. That element, by the way,
-------
214
is technitium.
SLIDE 1
The 73 elements shown on this periodic table can
all be analyzed by an ICP and can be run on a sample
in a SuperScan analysis mode. As you can see, the
ICP analysis is able to identify the presence of most
of the transition metals and most of the alkali
metals. The rubidium and cesium, however, it should
be noted, suffer fairly high detection limits and may
not be useful in terms of a SuperScan.
As a laboratory, we often find people that come
to our facility and they're bringing us a sample,
maybe something like this Tina had, and said, can you
tell me what's in it. Well, when we had AA and we
were running our metals analysis by AA, we went and
counted our lamps, told them we could run this and it
was going to cost that.
Now, we take a look at our chart and we say,
well, we can run up to 73 different things using our
new ICP. If we were to run those four in quantitation,
it could cost guite a bit of money. Yet, on the
other hand, we can run this in a scanning mode or a
screening mode and save the client an awful lot of
money and still get some useful information as to
what the composition or the contaminates in that
-------
215
sample might be.
The SuperScan screening procedure is based on
the ability of high resolution sequential ICP
instruments to run such a wide range of elements
quickly and with relatively high selectivity. I
don't want to spend a lot of time comparing the
relative benefits of sequential versus simultaneous
ICP analysis, for we all know that they both offer
their own benefits. What I would like to do is point
out that the screening function offered by sequential
ICP is something that we can use in a valuable way.
SLIDE 2
We purchased a high resolution duo-monochrometer
sequential ICP earlier this year. We use .this
instrument heavily for the analysis of 19 of the 24
elements required for the inorganic Contract Lab
program. Our chemists found the power and the speed
of their new toy to be fairly irresistible and they
experimented on the prospect of running all 73 elements
in that periodic table shown previously.
In many cases, as they looked for standards and
stuff to compare, or to run on the machines, they
went through our chemical stockroom and they went
through the stockrooms of the various universities
around, and they were fairly unsuccessful in finding
-------
216
a lot of those elements. We've since sent out orders
and we are receiving standards for the various elements
so that we can quantitate the detection limits on our
particular instrument and see how useful the information
is going to actually turn out to be.
Our initial idea was to run a standard solution
or two in order to obtain a semi-quantitative result.
If we had continued this thinking we would have
probably never gotten anywhere. It would probably
take a dozen or more standards containing compatible
groupings of compounds of the various 73 elements in
order to even begin to calibrate a run. It could
take years just to figure out the chemistry and
probably several more years to convince anybody that
we had figured out the chemistry.
SLIDE 3
The worst part is that when we look at the
periodic table here and look at those elements that
we can screen in the SuperScan mode, the chances of
finding some of the odder ones are pretty slim. I
know I always am asking for trouble when I say you'll
never find some of those things in a sample, but the
odds are pretty much against you. So if you're going
to spend a lot of time and money trying to quantitate
a result before you even have a suspicion that the
-------
217
element is in your sample, you're defeating your
benefit of screening. So, we'd like to promote this
as a means to pre-screen a sample or to screen a
sample that's being run for other elements as well,
but to screen it to find some information about what
other things might be in it, and I'd like to show you
some data later on to point out how that might be of
benefit to us.
The bottom line question that one would have to
ask himself is how would you like to be able to screen
a water sample for the presence of 73 elements in
less than five minutes and for less than, let's say,
50, 60 cents an element. That's exactly what we're
talking about here.
I'd like to briefly now discuss how the SuperScan
works and then end up with a few quick examples of
SuperScan runs that we've made.
The elements shown in the periodic table emit
intense characteristic light wave lengths in the Uv
and visible part of the spectrum when they're subjected
to the high temperatures of plasma. The intensity of
the emission is proportional to the amount of the
specific element in the sample. In essence, all one
has to do is to carefully select 73 wave lengths which,
on the one hand have sufficient sensitivity to yield
-------
218
a useful detection limit, and on the other hand, are
not subject to significant interferences from nearby
emission lines of other elements.
Our ICP has a very powerful software application
package. It contains a data base of several thousands
of analytical emission wave lengths from which we can
choose our method file for the SuperScan run. In
addition to that, we can set our windows, our viewing
windows, for each particular emission line quite
narrow, and this excludes the possibility of mis-
identifying peaks of nearby emission lines.
Our instrument is a dual monochrometer system,
as I mentioned before. One high resolution mono-
chrometer has a spectral resolution of less than
100th of a nanometer, and the other monochrometer
has a little lower resolution of less than 200ths of
a nanometer. In creating a method file, one can use
the low resolution monochrometer for the easy elements
and set the high resolution monochrometer to measure
the more difficult ones. The other nice feature is
that when you're running these analyses you want to
make sure that you've got sufficient speed because
speed is where you're providing yourself with the
profit.
So each monochrometer, when you're running a dual
-------
219
system, is controlled independently of each other so
that the second monochrometer is getting in position
while the first one is making the measurement on the
previous element.
Another factor one has to control is the viewing
height. There's an optimal region in the plasma for
each element where the signal to noise ratio is
maximized. This factor can be preset for each element
when you're creating your method file, and during a
run then, the computer, under the control of your
method file for the SuperScan, rotates the grading to
a specific wavelength at a speed of 50 or so nanometers
per second, stops within a hundredth of a nanometer
on the wave length, with a very narrow window, makes
a measurement, and during that measurement time the
second monochrometer would be adjusting for the second
element out of the 73, and the process would continue.
Over a period of four or five minutes you'd have a
SuperScan run.
The fee of SuperScan analysis is one of the
critical factors to keep in mind. The ICP measurement
for a laboratory can make a laboratory a lot of money,
but not if it's tied up for half an hour or so making
a SuperScan measurement. You can't afford to run a
screening test that costs hundreds of dollars. So the
-------
220
screening test would have to be considerably less to
be of value to the client.
SLIDE 4
Right now, I'd like to take a look at some of
our data that we've generated on our SuperScan runs.
This particular slide here is shown in graphical
forms in the following slides so you don't have to
worry about the numbers here. What I'd like to point
out though, is that these elements here are some of
the 73 that were measured in a SuperScan run. The
other ones were unimportant because they didn't show
up in the particular samples at all. If you looked
at the periodic table, all of the elements on the
periodic table that were indicated in the first slide
were run.
This sample here is a groundwater sample from
around a hazardous waste site, or is from a batch
of groundwater samples from around a hazardous waste
site. We picked that as one of our samples and we
picked this from a different batch that's also a
groundwater sample from a hazardous waste site. As
it turned out, we pulled this off the shelf as I was
leaving to get ready to go to this meeting here. We
had these run and we had three samples run; this one,
this one and this one, in the SuperScan mode within
-------
221
20 minutes.
These values here are actual ICP quantitative
results in this column here. Some of the measurements,
like for arsenic and so forth, are run by furnace AA
for comparison. Mercury, of course, is run by cold
vapor. In this particular sample, we inadvertently
pulled it out and it turned out to probably be a
field blank, so it's kind of interesting to pick this
sample because as we look at the SuperScan result,
which is a ratio of the signal of the sample to a
digested blank...so a ratio of 1 says that there's
essentially no detection in the SuperScan mode...we
see that all these values here are essentially 1,
indicating that that is acting like a blank sample.
As we take a look at another sample that is an
actual sample that has something in it, we see that
that's not the case.
SLIDE 5
Now, I'd like to go to the next slide, which is
graphical, and we'll show this a little clearer.
Now, this is the second sample that we SuperScanned
and this is a signal over background of about 1,100,
and that's for calcium. I'll point out what elements
these are since they didn't come out very clearly.
Calcium came out very high. Sodium was about 400 times
-------
222
background. Strontium was about 700 times background
in the signal.
Now, I'll point out also that these elements
here, all the way up to zinc, which is right here,
are normal elements that we run in the Contract Lab
Program. The rest of them were additional elements
that showed something in this particular sample.
SLIDE 6
This is the same slide where we've just clipped
the top off and expanded from 0 to 100 instead of 0
to 1,000. It's really unimportant. This is still
1,100, this is still 700, and so forth. Values around
1 indicate that the sample showed no evidence of
that element being run in the SuperScan mode, however,
when we see aluminum here, we've got about two or
three times background. The same for barium. And in
the case of iron, we see we've got about 30 times the
background signal.
When we get down to the lower end of this group
of metals here, or elements, we find that sulfur
shows up, erbium shows up at about three or four
times background. Cerium, strontium, boron, cesium
and silicon and lithium show up. Iodine and phos-
phorus didn't show up, but they'll show up in the last
sample, so I left them all on for comparison purposes.
-------
223
Now, if we would have run this sample, as we did
run it for a client...when we ran this particular
sample, we only measured the first 24 elements, and
we used normal ICP and furnace methods for that and
we have quantitative results. In each case, the
quantitative result indicates that there was either
a high amount of material or, in a case where none
was detected, there was a good relationship between
the SuperScan being an indicator for these 24 elements
in this particular case.
However, the client didn't know anything about
the last group of materials that we identified. Some
of them may not be interesting. But if the client
can see this picture, then the client can make a
decision. He may say, what the heck is strontium
doing in there. Strontium tends to show up in more
and more lists these days. Strontium is not a normal
element that we'd run for the Contract Lab Program,
for example.
Osmium might be another one that we would iden-
tify in terms of a SuperScan mode where it wouldn't
be normally asked for. But a client can come back
now and take that same sample and request a quanti-
tative result for that particular element that showed
up in a SuperScan.
-------
224
SLIDE 7
This slide is of a sample...I like to do things
in threes, so instead of running another hazardous
waste site sample we ran a cup of coffee. Same 24
elements here. We see that we get the normal hits;
we get a hit on calcium, we get a hit on magnesium
and manganese. We also get a hit on potassium and
sodium, and a little bit of barium in there. This
is not a quantitative result, it's just two times the
background level of a digested blank, or a blank
water sample in this particular case.
We see there's sulfur in there at a level about
three times the background level of sulfur. And we
see strontium again, so I can't tell you what kind of
coffee it is. We've got silicon. I can tell you if
you put cream in your coffee, you'll probably find
more of this stuff in there, but besides that, we
also found the iodine and the phosphorus in this.
All the other 73 elements were not found in that
particular sample.
So, the question really becomes what kind of
usefulness can the SuperScan technique provide to us,
either the user or the laboratory. And it has benefits
for both of us. In the case of the laboratory, we've
got a number of different benefits that we can achieve
-------
225
by running a SuperScan. One of the problems of
running emission spectroscopy is the fact that you
can have interferences from high amounts of different
elements in that sample. Those interferences need to
be corrected for because you'll get a high value, for
example, for antimony and tin, if you've got a lot of
iron and calcium in there.
Well, if you're running a hazardous waste sample
for antimony and tin, chances are you're not going to
normally run iron and calcium, and so the only way you
can determine that you've got an interference from
high amounts of iron and calcium is by doing spiking.
So that would be one thing.
The second usefulness is when you get a client
that comes in your door, and every lab's got these.
Once a month or so we get somebody coming in and
saying, what can you tell me about this water, I
think I'm dying. You don't want to sit there and
tell the poor person that you're not going to measure
it because they'll die when they get the bill, you
know. So, it's nice to be able to have a little
method that you can say, for $50 or $40 or whatever
you're going to charge for a SuperScan, you can sit
there...
AUDIENCE PARTICIPANT: You're
-------
226
undercharging.
MR. HANDEL: Oh, okay, $100
or $200, we'll tell Bill that his water's all right,
or whoever.
The other advantages are your end clients. If
you've got a client that wants to know a certain
number of things for sure but is also out on a little
bit of a fishing expedition...maybe it's his own
backyard and he wants to see what he's got back
there...you can run a SuperScan in addition to the
things that he suspects to be back there that you
would normally run quantitative anyhow, as we did in
the first two examples.
We're proceeding. We haven't gotten real far on
this project yet. We're proceeding on an as we can
basis. It is interesting for us to do it, and we'd
be glad to talk to anybody about it if they're
interested in doing it on their own machines.
Thank you.
-------
227
QUESTION AND ANSWER SESSION
MR. BIRRI: John Birri,
EPA. Have you thought about using this as a screening
technigue, possibly for TOC and sulfate?
MR. HANDEL: Basically...
yes, I guess we could. We have in fact done that a
couple of times where we had a high carbon answer.
The detection limit on carbon is not the greatest in
the world, but then when you're running a TOC anyhow,
you're talking about part per million level. We have
correlated with our TOC runs that we've run for the
client after...for example, oil in his groundwater or
gasoline in his groundwater, we've actually picked
it up running this way.
MR. BIRRI: At what level
do you think you could pick something up?
MR. HANDEL: I think we
were talking in that particular case about 10 ppm.
MR. BIRRI: That's not too
bad. What about your detection levels in general?
What were you coming up with?
MR. HANDEL: Okay, the
detection limits are going to be element-specific
on this. They're going to also depend on what
-------
228
spectral lines you choose. You're going to have
different response factors for those lines. But in
general, a good ballpark range would be 50 to maybe
100 ppb.
You can improve on that if you want to, but we
want to make sure that we're keeping in the spirit of
a screening technique where we're not going to make
it cost a fortune to run it by increasing the amount
of run times or running two or three element lines
instead of just a single one. There's an awful lot
of things you can do to make it fancy, but it may
lose its value along the way.
MR. TELLIARD: As he pointed
out, it isn't very difficult, it's just expensive.
MR. BIRRI: My last question
is was this all done in vacuum?
MR. HANDEL: Yes.
MR. BIRRI: Thank you.
MR. TELLIARD: Thanks, Ted.
-------
229
cu
cd
o
CO
S-H
cu
O)
o
O)
-------
230
CO
CO
ex:
-------
231
oo CD
o t» o i-i o
•*»< CNI »-i o ^^ oo
m CD
c- co
O
•*-»
CO
O O> i-< __ »-< ..^ »-H
f> « ^0 ~~
t CO CO 00
I »-H ID
—.. O3 _,. _ rH £• CO
^^ CO CO r*
C O
i-( Ift CO CO CO i-H CD
o
fe«
-It
gill
= £0-5
J<
^ in 1-1
~c-'?1
OS W '" !
Cu Cu
cv
o
CD
E
t-
i-(
•D
C
CO
3
o
u
E
-I
•*-* .c
"O U
s "§
CO
E
5
; S>
o
53
a.
O
-a •-
-oo
< *•'
< CQ O Q W b O
O
I
o
f—»
CU
Q> w
£ +->
^S
^§
2 Q,
i E
.5 «>
Q«
^
eij
o>
43>
b.
-------
232
Element
Al
Sb
As
Ba
Be
Cd
Ca
Cr
Co
Cu
Fe
Pb
Mg
Mn
Hg
Ni
K
Se
Ag
Na
Tl
Sn
U
Zn
S
Th
Er
Pr
Ce
Sr
B
Cs
Si
Li
I
P
50301
ICP SuperScan
<7B
<58
<4
<21
<2
<4
<825
<8
<18
<19
<3B
<5
<351
<5
<0.2
<27
<2460
<3
<9
<523
<7
<40
<18
<11
0.96
1.01
1.01
1
0.99
1.0B
0
0.99
0.97
0.99
1.11
0.9B
0.3
0.99
1.01
0.97
1.03
1.04
0.97
0.78
0.98
1.01
1
1
0.92
0.96
0.99
1.01
0.95
0.35
1.54
0.96
4.09
1.05
1.11
1.08
50238
Coffee
ICP SuperScan SuperScan
15700
196
140
107
<2
12
2300000
59
<18
<19
8590
13
483000
83
<1
<27
32600
5
<9
6830000
<10
42
<18
79
3.76
1.11
1.1
3.93
1.02
1.11
1156
1.01
1.04
0.96
31.83
0.96
76.84
3.2
1.08
1.05
3.11
1.11
0.99
387.97
1.11
1.02
1.05
1.27
47.98
1.33
3.79
1.8
18.27
675.19
25.94
3.57
91.64
3.01
1.01
0.98
1.03
0.94
1.06
1.99
1
1
11.97
1.02
0.98
0.99
1.41
0.98
9.45
6.47
0.98
0.98
32.5
1
0.99
8.63
1.06
0.99
1.03
1.2
2.89
1.03
1.04
1.01
1.18
2.13
0.57
0.98
5.48
1.02
2.71
4.42
-------
233
O n.
Q |
(/] »
L. -a
°-
- 17)
1 \ \ \ \ \ \ \ \ \ \ \ \ \ \
i\\\.\\\\\
- a.
N
-ft
- F
-I
i\\\\\\\\\\\\\\\\\\\\\\\\
I i I i I i I i i i i
- Z
I
£
£
3 .
8
h 5
— c>
-------
234
D B.
o §
00 »
L- TJ
13 ""•
Cfl S
I
I \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \\ \ \\
K
k\W\\ V
.\\\\\\\\\\\\\\\\\\ \ \\ \\\\
^"X""\\ \'
C
[
K\\\ \\x\\S\\
[
. \ \ v \ \ \ \ \ \ \ \ \ \ \. \ \. \ \ \ v x \. \. \ v
I
E
k\\\\\\\\\\\\\\\\\\\
[
I^K- \\\\\\
!
\\\\\ \\x\\\\\\\\\\\\\\\\\
I
I
f\
i i t i i i i i i
- D.
- "Jj
I I I I I I I I I I I I I 1 I 1 1 1 I 1 I I 1 I 1 1 1 1 1 1 1 1 1
fil SbA^BtaBrsCdCaCrCoCuFc PbMglvfriHgr-4i K SeAgNaTI 5n V Zn 5 Th CrPrOsSr B CsSi
-------
235
l\\\- CL
r\ v\\f- is
8
ID
E
[
- |fl
8
>£
E-ij
E-f
E-
E-
K-
fi
w
F
D
l^
^
z
£
D»
!
£
3
\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ X
KXXXX
IX \VV\\\
E-
E-
K\\X\\\\\
o
(g
D
(D
^
ft
3
i i i
i i i
I I 1 I I I I
-------
236
MR. TELLIARD: Last year we
broke with tradition and had some of the critter
people come in. Our next speaker is a critter person.
Last year we had a meeting in San Diego on a public
hearing on the oil and gas regs, and Dr. Tom couldn't
make it because of no travel money, which I guess is
similar this year, so I had to give his talk for him.
Of course, being a real critter person like myself, I
talked about it, and in the process, one of the
questions was, why were we using mysids instead
of the California avocado or something. And I said,
well, it's a national standard. We have a national
method to compare national things. And from the
front row I hear a voice go, "Oh, shit, a National
Critter."
So, Tom is now the keeper of the National Critter.
He works down there in Gulf Breeze, Florida at our
laboratory, R&D lab, and they have bus loads of little
kids coming in to see the National Critter. Of
course, it's a lot easier than using bald eagles on
these tests.
But we thought about how we were going to really
bring this home, this fact that it's a National
Critter. We have the bald eagle. We thought of a
flag, a National Critter flag, but that really isn't
-------
237
going to sell. So, what we did is we arranged with
the Postmaster General to develop a stamp with the
National Critter on it. It will be 22 cents and it
will bring it out. So, when you go back to the lab,
Tom, you can give that to Hank and say, just send
it in to the Postmaster General and you're all set to
go.
DR. DUKE: I'll tell him
that, Bill.
MR. TELLIARD: Dr. Tom.
-------
238
-------
240
in the fluids), biocides, hydrocarbons for lubricity, caustic, and other
material. The exact formulation of drilling fluids discharged depends
upon the substrate through which the drilling is taking place, depth of
the well, and particular functions of the drilling fluid required at any
time. Therefore, there is no "typical" drilling fluid.
Best Available Technology .
It may be helpful at this point to review some of the criteria for a
toxicity test for BAT determinations because the guidelines are somewhat
unique. The test must be "generic" in the sense that BAT guidelines are
for general application as opposed to Section 403-c guidelines that pertain
to specific regional areas and can include requirements for indigenous
organisms and more extensive testing procedure. The test organism for
BAT must be reasonably sensitive to drilling fluids yet strong enough to
be transported to remote laboratory testing facilities if necessary.
Test procedures should be straight-forward yet rigorous enough to yield
scientifically acceptable.data. Also, a substantial data base involving
the test animals and drilling fluids should exist.
Methods
With these criteria in mind, the mysid, Mysidopsis bahia, was selected
as the test animal. A rather extensive data base is available on the
effects of laboratory-prepared as well as "discharged" drilling fluids
on mysids (Duke and Parrish, 1984). Methods for handling, acclimating
and sizing bioassay organisms given by Borthwick (1978) and Nimmo et al.
(1977) are followed in the toxicity tests.
Our laboratory's responsibilities in the comparative study conducted
-------
239
DRILLING FLUID TEST PROCEDURES:
PARTICIPATION, DATA COMPARISON AND IMPLEMENTATION
By
T.W. Duke and P.R. Parrish
Introduction
Our laboratory (Environmental Research Laboratory, Gulf Breeze) has
been involved in the development and implementation of toxicity.testing
methods for Best Available Technology (BAT) guidelines for discharges
from off-shore oil and gas platforms. We have contributed to the
test method described by Petrazzuolo (1983), tested various drilling
fluids (Duke and Parrish, 1984) and participated in a comparative
laboratory study where sub-samples of a laboratory-prepared drilling
fluid were tested by 11 laboratories.
The purpose of this talk is to provide background information on the
toxicity tests that serve as a basis for the statistical evaluation. A
statistical analysis of the comparative laboratory study will be presented
next by Drs. Eynon and Bailey. I apologize to those of you that attended
last year's Symposium because some of this material necessarily will be
repetitive.
Drilling Fluids
Drilling fluids are used in the rotary drilling process for several
purposes, including transporting cuttings produced by the bit, lubricating
the bit, coating the bore to prevent fluid loss, and reducing corrosion.
These fluids are a complex mixture of chemicals and clays and, in order
to perform their functions, can contain barite, bentonite, lignite,
lignosulfonate (these components comprise about 90% of the materials used
1
-------
241
by the Office of Water Programs included storing the drilling fluid
prepared by a commercial company, sub-sampling the original stock (a
commercially-prepared generic mud (number 8) with 3% mineral oil), sending
sub-samples to participating laboratories, establishing test concentrations
through range-finding tests, conducting definitive toxicity tests,
and supplying to the test method to the other laboratories. Our results
were included in th* comparative study.
A 96-hour concentration lethal to 50 percent of the test population
(LC50) was used to express the toxicity of the suspended particulate
phase (SPP) of drilling fluid sample used in the comparative study.
Details of the method were supplied to each participant and followed
the procedure published in the Federal Register (1985). In general, the
sub-sample to be tested is thoroughly mixed to a volumetric mud-to-seawater
ratio of 1 to 9. This slurry is mixed and the pH is adjusted, if necessary,
to within 0.2 units of seawater. Then, the slurry is allowed to settle
for 1 hour.
At the end of the settling period, the SPP is decanted (not siphoned)
into an appropriate container. Decanting the SPP is one continuous action.
The decanted solution is defined to be 100 percent SPP. The SPP is mixed
for 5 minutes (it must not be preserved or stored) and a sample taken
so that the filterable and unfilterable residue of the SPP can be measured.
Also, dissolved oxygen (should be at least 4.9 ppm or 60% saturation)
and pH of SPP are measured and adjusted if necessary. Appropriate volumes
-------
242
of 100 percent SPP are mixed with appropriate volumes of seawater to
obtain the desired SPP concentrations. The control is seawater only.
The animals are randomly selected and placed in dishes to begin the
96-hour toxicity tests.
Definitive test concentrations are based on results of the range-
finding test (in this instance, ERL-GB supplied range-funding values).
A minimum of five concentrations plus a negative and positive (reference
control) is required for the definitive test. To estimate the LC50, two
concentrations are be chosen that give (other than zero and 100 percent)
mortality above and below 50 percent. Cups constructed of nylon-mesh
screen are inserted into every test dish prior to adding animals. Twenty
organisms are exposed in each test dish and three replicates (total of
60 animals) are used in each test concentration. Individual animals are
randomly assigned to treatments. Throughout the test period, mysids are
fed daily with apporximately 50 Artemia (brine shrimp) nauplii per mysid.
Dishes are covered and incubated in a appropriate test chamber; test
mixtures are gently aerated throughout the test. The test medium is
not replaced during the 96-hour test.
At the end of 96 hours, all live animals are counted. Death is the
end point, so the number of living animals is recorded. Death is determined
by the lack of spontaneous movement.
Data are analyzed according to Finney (1971) to obtain the probit
model estimate of the LC50 and the 95 percent fiducial (confidence) limits
-------
243
for the LC50. These estimates are obtained by using the logrithmic
transform of the concentration.
Results of Comparative Toxicity Test
The results of the comparative toxicity test and a statistical
analyses will be presented in the next paper by Drs. Eynon and Bailey.
From a toxicologicaV point of view, the variability of acceptable results
from the various laboratories was within a reasonable range, i.e., the
range was within those limits reported for single chemical toxicity
tests conducted by several laboratories in similar studies. In our
opinion, the manner in which the SPP was prepared by each laboratory was
probably the largest contributing factor to variation of results among
the test laboratories. This conclusion is based on experience of our
staff as well as discussions with personnel from other laboratories.
There is no doubt that this BAT toxicity test can be improved, but it is our
opinion that it conforms to state-of-the-art testing methodology and is
appropriate to determine the acute effects of drilling fluids on mysids.
-------
244
Literature Cited
Borthwick, Patrick W. 1978. Methods for Acute Static Toxicity Tests with
Mysid Shrimp (Mysidopsis bahia). In: Bioassay Procedures for the Ocean
Disposal Permit Program, EPA-600/9-78-010: ERL-GB.
Duke, T.W. and P.R. Parrish. 1984. Results of the drilling fluids research
program sponsored by the Gulf Breeze Environmental Research Laboratory,
1976-1984, and their application to hazard assessment. EPA-600/4-
84-055, Environmental Research Laboratory, Gulf Breeze, FL. 94 pp
plus appendices.
Federal Register Aug. 26, 1985. 40 CFR Part 435. Oil and Gas Extraction
Point Source Category, Offshore Subcategory; Effluent Limitations
Guidelines and New Source Performance Standards; Proposed Rule.
Part II. pp 34631-34635
Finney, D.J. 1971. Probit Analyses. 3pd Edition. Cambridge University
Press.
Nimmo, D.R., L.H. Bahner, R.A. Rigby, J.M. Sheppard and A.J. Wilson, Jr.
1977. Mysidopsis bahia: an estuarine species suitable for life-
cycle toxicity tests to determine the effects of a pollutant. In:
Aquatic Toxicology and Hazard Evaluation, F.L. Mayer and J.L. Hamelink,
Eds., pp. 109-116. ASTM STP 634, American Society for Testing and
Materials, Philadelphia, PA.
Petrazzuolo, G. 1983. Proposed methodology: Drilling fluids toxicity test
for offshore subcategory; oil and gas extraction industry. Unpublished
Report. May 19, 1983. 45pp.
-------
245
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
AUDIENCE PARTICIPANT: Where
do you get a copy of this...
DR. DUKE: This has been
sent out by Bill's group.
MR. TELLIARD: It was in
the Federal Register notice on the proposed regulations
for the oil and gas industry, which date was August
28thf 1985.
AUDIENCE PARTICIPANT: The
second question concerning the method. You talk
about DO requirements. What is a DO requirement? I
mean, I can look it up in a method, but what is a DO
requirement for this particular method, when you're
setting it up?
DR. DUKE: We keep it at 60
percent...at least 60 percent of saturation.
DR. COWGILL: Michela
Cowgill, Dow. I have sort of a technical question.
You had an LC50 for selenium with something like 35
ppb? Did I misread it from back here?
DR. DUKE: No, the only
LCSO's I showed were drilling fluid LCSO's.
-------
246
DR. COWGILL: Excuse me?
DR. DUKE: Those were not
drilling fluid LCSO's. The table with selenium, I
think, was an analytical analysis.
DR. COWGILL: Of the?
DR. DUKE: Of a series of
elements that were analyzed by various laboratories.
DR. COWGILL: It wasn't an
LC50, I see.
50.
you.
Tom.
DR. DUKE: It wasn't an LC-
DR. COWGILL: Okay, thank
DR. TELLIARD: Thanks, Dr,
-------
247
MR. TELLIARD: Our next
speaker is Barry Eynon from SRI. He's going to talk
about the statistics and the round-robin that was run.
We ran 10 laboratories, as Tom pointed out. This
was with a drilling fluid that was a simulated
drilling fluid. It was hot-rolled and the labora-
tories ran these check samples. The Gulf Breeze
laboratory ran the range finders so it gave them the
window, and then they ran the samples for the
drilling fluid.
The drilling fluid was Number 8 MUCL with two
percent mineral oil added.
-------
248
-------
249
-------
250
-------
251
PRnfFTIURF;
WATER QUALITY
|fe*M>
siLvi ' .r"?!1"-,.1'. -
INCUBATE FOR 96 HOURS
COUNT LIVING WSIDS
CALCULATE 96-HOUR LC50
•'.-'. -.-'j -^m
.^t&isj&t
•••'*••'";•"«*?»<'
-------
252
-------
253
-------
254
-------
255
Csi CN1
CSI OJ
csi
," - ..-i." •;',','". ---,;-'1-' -.'"• ..^;'-, "•'-.'' -f'.^^&ri
- - •- • -. - ./.;"--L.. -, • A, , _ j ;..,'-•''-•",-'V'.^. j.'"^.-^---^^';^-^
-------
256
V ,-, -^:,•''..' :..•'•• .1"'"f;"|j"'•*•••','•' ••'>'• •••"f\'lj'i-v:,;>-f ^'•••'fffS1'''^
•".-•'•"' •;'••. ••.'•".•'• w'-'V^,':'--:" • ••'••:•., .'.'.'.j'A '>"'i/"'''v:!:;'||i:&?*?
""• •- •'.,- ". ,"'-i-^ "'' •«'-/', ••,':i'.!,f A- '"-'iiliwJtt
'.; , •: ". ••- ',•.'•'-,'"fe'v'r--,1'-'•A i,'",- 'l';'^i:: '","4&E%lf
SOURCES
FIUIDS
TEST ANIMALS
tEST MATERIAL PREPARATION— Phase se|
flEST MATERIAL COMPOSITION
CONDITIONS . ; ' . •: •'
V
-------
257
FLUIDS ACUTE TOXICITY
— Source
~ Age
— Condition
-------
258
i^"t^''vvi:"O«'•'-'" *"^"t?'.**:'.'„'•'' '•> ' '.".. • '•'
TK"W?\. •J^/^'.'JSi'rit*"-.*':,-^'. ':.•_',: ..•,.""'••:•' '•.•'
S^..;1..:'^/..-1."-"
tn*"W",. * ,;.
'
Pl^Ifi |pf E TOXlCilTY
CcMnpositipn,^ ^
Ass
Preparation
1. Phase separation
2. pH adjustment
3. Aeration
-------
259
^tosgfew*^;^iV' ''V^'v^x -r; '•'•, ,'... .'• v ''
^.W^:
5H*<.V"' ^"T = - "-'-' '- ':•' . - •'
%8&.?':y'^-X'•'
feSg^IiM::,:-':.,
SOORCES OF VARIATION
H*-^^i-*'' .*,J •-••'• '; '- :•-
r^^S^'v/" --
fplilNG FLUIDS ACUTE TOXICITY TfST
i •'-.''.
TEST CONDITIONS ~ Aeration
— Containers
— Other
-------
260
SOURCES OF VARIATION
DRILLING FLUIDS ACUTE TOXICITY TEST
! ' '" :'.-*' •> - ,V i ;^ ^";:
TEST FACILITY —Experience
— Skill
-------
261
BARRETT P. EYNON
SRI INTERNATIONAL
STATISTICAL EVALUATION AND VALIDATION
OF THE EPA DRILLING FLUID TOXICITY TEST PROCEDURE
MR. EYNON: Thanks, Bill.
Yes, I'm glad Tom was able to talk and show you
something about the shrimp and how that's all done.
That's something he's far more familiar with than I
am. What I would like to talk about is the results
of this round-robin study that was run.
SLIDE 1
MR. EYNON: The results
of the round-robin study that was performed about a
year ago. That study had two objectives. One was to
evaluate and analyze the performance of the method in
support of the effluent guideline preparations, and
also to aid in the selection of contract laboratories
for performances of these analyses, which that's been
done and analyses are ongoing.
This is part of the development process of the
method, and the kinds of analyses we're looking at
today are being fed back to Dr. Duke to help him
develop the method further and provide extra informa-
-------
262
tion. I guess we can go to the next slide.
SLIDE 2
I've got a couple background slides here. These
were drilling fluids. Pretty much this is what we
said before. These are discharged into the water and
that's why we mix the SPP down with water to determine
toxicity. Let's flip on here.
MR. TELLIARD: Barry, what's
the SPP?
SLIDE 3 & 4
MR. EYNON: Oh, I'm sorry.
The particulate...suspended particulate phase of
the material. Here's the reference to the method in
the Federal Register, Monday, 1985, and this method
was distributed to 10 laboratories who were participat-
ing in the study, along with a...we can go to the
next slide...along with a well mixed preparation of
drilling fluid Number 8. They all got the same
sample. Then they were asked to perform the method
using their procedures. Where not defined by the
method, the labs were given the procedure and told to
perform the analysis.
They also performed a reference toxicant. They
took three to five day old mysid shrimp, and as Tom
said, we took 96 hour exposure at five dilutions of
-------
263
the particulate phase and a control at 0 percent.
The three dishes of 20 shrimp of the drilling fluid
at each concentration. I will note that, as Tom
showed, the randomization procedure was used to make
sure that the shrimp were randomly distributed and
fairly distributed in the various dishes, and I think
that's an important competent of the procedure.
SLIDE 5
Then we have our obligatory National Critter
slide here with shrimp on it* Flip on.
SLIDE 6
In conjunction with the study, there were also
six replicate runs done at Gulf Breeze of the same
drilling fluid by the same method so we can compare
inter- and intralaboratory variation. We also
collected information on possible factors that could
be used to control the method or aid in controlling
for variability.
Some of those were the amount of total suspended
solids...each laboratory was instructed to measure
that...the amount of acid used in the sample, the
initial pH of the sample, and some other factors as
possible to measure.
SLIDE 7
The analytical form of the dose response function
-------
264
that we used in this analysis is the traditional
Probit function with an adjustment for spontaneous
response rate. Sof just to refresh us on that, the
form is a S-shaped dose response curve. Here we
applied it to the logarithms of the concentrations
and then the C+l-C is the correction for the spontaneous
response rate. So the intercept at O...as dose goes
off to 0, is C. These parameters were fitted by...
no, let's hold with that. Yes, let's hold with the
slide for a second there.
These parameters were fitted to the data for
each laboratory by maximal likelihood fitting techniques
to estimate the three parameters, C, Mu and Sigma.
The LC50 is then the exponential...because we're
working with log doses, the LC50 is the exponential of
the parameter Mu. Mu is the center, the log of the
50 percent point of the curve; Sigma determines the
rate at which the curve increases with increasing
dose, the rate of mortality increase; and then C, as
we said, is the spontaneous response rate.
The procedure that we wrote was written in the
statistical analysis system package as a macro program
using the procedures NLIN and MATRIX. And if
anybody's interested in the analytical techniques we
used, I'd be happy to talk to them later.
-------
265
We found that there were some limitations in
some of the published software. For instance, SAS
Proc Probit had some convergence problems with some
of the data, not that the data was bad but I think
the program has a few limitations. So we developed
our own analytical package to do this.
We also looked at, and I won't say a whole lot,
but we also did look at some other methods of
calculating LC50, including the moving average method
and also an extension to the Probit called the Burrit
model, which has some extra parameters for shape of
the response curve. Let's go to the next slide.
SLIDE 8
The next three slides are just for your edification
to see some examples of the actual data and the fitted
dose response curves in several situations. This is
the Gulf Breeze Laboratory data, and we have mud
concentration in percentage across the bottom. This
is on a log scale, and then the proportion of shrimp
which died at each concentration as the vertical
scale.
So, the data are the pluses at each concentration;
1, 2, 4, 8, 16 percent. And then there's a control
dose, which is way over on the left hand side. The
two plots are just...one plot is the raw response and
-------
266
the other plot is the adjusted response. Here, the
control response was 0, so the two plots are actually
the same.
SLIDE 9
The other two slides are just two of the labs.
This is Lab 3. A similar curve, but notice the slope
is a little bit shallower.
SLIDE 10
For Lab 4, we get a steeper slope response.
This kind of variability is characteristic of the
variability in the response curves on the same material
at different labs. So, that's the kind of variability
we're looking at. We'll come back to that a little
bit more.
SLIDE 11
The first objective of the study that we looked
at was to evaluate the laboratories for contract
performance. Several criteria were used. We compared
the results for each lab with the reference lab, with
the Gulf Breeze lab. Each calculation of the LC50
also obtains a confidence bound within the procedure,
and the width of that confidence bound is an indication
of the accuracy of the lab. Then we also performed a
comparison of each lab with the group of all other
labs so that we were sure...if the reference lab was
-------
267
distinctly different, then we would capture that
comparison.
Each of these factors was used to rank the labs,
and a combined ranking was put together which allowed
an overall ranking of the labs that was then given
back to EGD to aid them in the contract selection.
Okay, we've got the criteria and the rankings.
So that was the first step. Now we want to go look
at variability and possible explanations.
SLIDE 12
Here are the parameters of the fitted curves for
each of the 10 labs and for each of the six replicates.
The top plot is...one comment on the laboratory
ranking. We found that 3 of the 10 labs did not
perform what we thought were acceptable runs. The
results were too far away from the central lab. One
of the labs consistently got over 80 percent control
mortality and so was rejected on that basis, and the
other two labs; one was very high mortality and one
was very low.
For purposes of this study, because we felt that
with proper training those labs could be improved or
were not considered within the range of acceptable
variation, they were removed from basically the rest
of the study.
-------
268
So, here we have the parameters plotted for the
labs that we were considering. The Mu is the log of
the LC50, Sigma is the slope response, and you can
see that there is some variation between labs on the
top plot, and within labs on the bottom plot, and
that there's more, as we expect, there's more variation
between labs in the top plot. Go to the next slide.
SLIDE 13
We can lay out the actual results. The parameters
for each of the six trials, trials labeled A through
P, are given in the table for the drilling fluid and
for the reference toxicant. This table also shows
the achieved log likelihood, which is a measure of
fit, and the criterion by which we fit the parameters,
and that's always a negative number; 0 would be
perfect, perfect fit. Differences between the log
likelihood can be used to measure and test the
effectiveness of different models.
So, the table is designed to do three things;
first, to show the parameters; second, to test. At
the bottom, we can test for whether there are signi-
ficant differences between the samples within labs,
and the Chi-Sq test, listed as Chi-Sg of intralab,
does show significant...there are significant varia-
tions within labs, even within the reference lab, on
-------
269
the different trials. And that simply means there is
intralab variability above and beyond the natural
variability of the parameter estimates themselves.
The second Chi-Sq is the same test where we
allowed a separate spontaneous response rate to be
estimated for each trial and gets a similar result.
And the bottom Chi-Sq is a test of whether the
spontaneous responses are separate for each trial.
This is-also significant.
So, it appears that one feature to be considered
is that separate spontaneous response needs to be
considered for each trial. this may be due to
different batches of shrimp or different other time
factors. So, all runs should be allowed a separate
spontaneous response rate and then the LC50 is the
corrected value.
SLIDE 14
The next table is...that was intralab. This table
looks at interlab variations. This doesn't, unfor-
tunately, have the parameter estimates on them, but
there were two objectives on this table. One was to
test within each lab a goodness of fit result. We
can also test versus an arbitrary response curve
that's not the Probit curve, each lab individually,
and we can see that on each lab there is a not
-------
270
significant result for the test of goodness of fit,
which means the Probit model is doing a good job of
fitting the data.
Then, when we get down to the bottom, we can
also test whether there were interlaboratory differ-
ences. Given that we saw intra-, it would be surpris-
ing if there weren't. We do confirm that there are
interlaboratory differences. So, all of those sources
of variability are present in much the same way as
they are with an ordinary analytical function. So we
have that.
SLIDE 15
The next slide lays out...we can do almost an
analog to the analysis of variants table using the
likelihood functions, and we can split up the likelihood
into inter, intra, and several other factors, basically
which are all goodness of fit factors. The inter- and
intralab are highly significant. There are slightly
significant variations between the trials. That's, I
think, all I want to say about that slide.
SLIDE 16
To try and summarize the information that we can
get from the inter- and intralaboratory comparison, I
put together this table, which takes the LCSO's and
looks at what the average...geometric average LC50
-------
271
would be for both the inter- and intra-, and also
what a 95 percent variability multiplier, since these
are again on the log scale, what a 95 percent
variability multiplier would be for the amount of
variability we saw among these labs.
For the drilling fluid, six concentrations times
60 shrimp. Intralab...well, the minimum that this
could ever achieve would be the inter-experiment
variability that is just due to randomization and the
repetition of the process. That's the best you're
ever going to be able to get, and that's the number
I've listed under experiment. The best you could get
with six times 60 and the design series, about 1.2,
or about + or -20 percent.
The actual that we see; intralab is more like a
multiplier of 1.9, and interlab we see a multiple of
about 2.6. Tom says that in his experience this is
actually pretty good for biological testing and that
he's very pleased with the fact that the labs do a
pretty comparable job. This is larger than one sees
in analytical testing situations and one has to keep
that in mind in thinking about using these types of
procedures.
Similarly, on the reference toxicant we also see
that as you go from the intra-experiment to intralab
-------
272
to interlab there's also an increasing variability,
so this is an indication that differing lab standards
and procedures can have an effect on the result/ and
that one thing this leads us to is to ask the question
can we identify those qualities and propose better
details to be provided to the laboratories to help
them control these features. So, this leads us to
ways to measure other information that will help us
evaluate the test. If we can go to the next slide.
SLIDE 17
We have proposed some covariate models that take
the Mu and Sigma, which in the first round are simply
parameters to be estimated, and now we express
those, we attempt to express those, as a functional
model dependent on other covariate information. So
that, for instance, the LC50 or the log of the LC50
may be dependent upon some other factor we can measure,
and we can measure...by keeping track of that over
several trials, we can estimate that much like a
regression model.
This is imbedded in the maximal likelihood fitting
scheme. I don't think I'll say too much about that
except that it follows as a sort of a natural analog
to a regression type of analysis. We get the same
kinds of results, we get confidence intervals on the
-------
273
parameters, and we can test the significance of the
covariates. Next slide.
SLIDE 18
The reasons for looking at this are threefold,
depending on the particular nature of the covariate.
In each of these cases we may be able to do something
to improve methodologies.
If there are factors that can be both measured
and fixed, we can fix them if they're important. And
that's a cost versus benefit type tradeoff. If there
are procedures that are out there but we can't control
them and we can't measure them, we can randomize over
possible sources of variability to reduce their
effect. If we can measure but not control certain
parameters, we can consider the possibility of numer-
ically adjusting the results of the analysis to
compensate for the variations of those between runs.
All three of those procedures can show...have some
possibility of improving the results and that's
what we're working on now.
Just for instance...let's show the next slide.
SLIDE 19
One way of thinking of this is to look at the
MU, which is the log of the LC50, as a function of
one or more covariates. Here's a plot where we show
-------
274
the actual Mu's for the various labs plotted as a
function of the initial pH of the SPP of the pre-
pared dishes. Now, notice, we're looking at a very
restricted range here. The pH is controlled anyway,
and in fact, in this particular case, it's not clear
that it can be controlled more. It's one of those
that has been controlled, and the question is, is
there residual effect of that variable.
We can see a small, small trend here, possibly,
in the upper picture. That's of Mu versus pH. Then,
similarly, there may be a small negative trend in the
lower picture where the steepness of the slope also
depends on the pH. These are possible potential
relationships we're still exploring.
SLIDE 20
When we do this analysis, we can actually produce
tests of hypotheses and layouts similar again to the
regression or analysis of various types of approaches
where we test for the effects of factors. Here, this
is a table where we have the log of the original TSS,
the amount of acid, and Mu Ref, which is the log LC50
from the reference toxicant, all three of which we
thought might be predicters for toxicity or might be
influences on toxicity or correlated with toxicity in
the sample. Here we do find some significant results.
-------
275
Singly and in combination we get parameter estimates
for the various coefficients, and likelihood value
improves, and it improves it by a significant amount
in each case. So, this says that these parameters are
having some effect and we can think about controlling
them.
We've looked into them. I think that one thing
we've decided is that TSS needs to be more precisely
measured in the various samples because it's got a
high possibility of being important, and that some of
the other factors are also potential.
SLIDE 21
Let me skip these and show just the last slide,
which is just to come back to the Burrit model. To
compare some of the models, again for goodness of fit,
we fitted the Burrit function as opposed to the
Probit. The Probit uses the cumulative distribution
from the normal. The Burrit uses a different F
function which has two other parameters which control
its shape. We've made the two sets of parameters
comparable in that Mu and Sigma mean the same thing
in both models. We can see that under the Probit,
the parameters are very similar under the Burrit.
CB and KB are the shaped parameters. The KB likes
to be very large, which is maybe a Wibel model, which
-------
276
is different particular distribution, but way over on
the right hand side...well, I guess it's not. The
point is that these improvements are not significant
in any of the particular cases, and so we are reassured
again that the Probit is probably a good model to
use.
Let me stop there and entertain questions.
-------
277
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
MR. PLOST: Charles Plost
with EPA. As chemists, we're traditionally accustomed
to biological variability being the analyst, and so
we characterize our methods by running round-robins,
where typically we'll have two dozen laboratories as
opposed to 12. We'll go to great pains to throw out
one of the laboratories, and really give ourselves gas
when we try to throw out two, and at the end of that
we'll feel that we know something about the precision
and accuracy of the methodology when applied by any
laboratory. It seems that you've done something
that's somewhat akin to that.
May I translate this one?
MR. PLOST: Well, one, why
did you throw out three laboratories? Two, can you
extrapolate from this as we've extrapolated with...to
how would the next laboratory perform? Is this an
alternative way of running a round-robin?
MR. EYNON: The judgment to
throw out; one laboratory was just clearly not in
compliance with the procedures. It had very, very
bad response. The other two laboratories, one was
-------
278
the highest and one was the very lowest, and in a
best judgment situation, given that this is a
development study and not a validation study, the
best judgment was that those laboratories had failed
to perform the procedure as specified.
If we were going to try and actually state the
precision of the method rather than evaluate it, then
I would worry about that because you'd have to
say you have to either have some check or some formal
test for outliers. But the feeling was, and I think
Tom said to me, that what he wanted to do was get the
labs in and train them. They were given the procedure
and told to perform it, and he feels that they can
make improvements, and I think have made improvements
since then. He's been able to see improvements when
they learn the method and improve.
So, I could only express a personal desire to
have another round-robin at some point in the future.
I have no idea whether such a thing is in the works.
MR. PLOST: But would this
technique substitute for a round-robin?
MR. EYNON: Well, this is
a round-robin.
MR. PLOST: Sort of.
MR. TELLIARD: Well, Charlie,
-------
279
I think that the fact that we used 10 laboratories...!
mean we validated some of the 600 series on that same
number of labs, or less in some instances. On some of
the methods we only had, we won't bring that up. In
this case we did use 10.
I think also the universe of available laboratories
compared to...remember when we started back with 1624
and 1625 and all that, we didn't have...George had
the lab in his garage that he was generating API
data and we had a couple of others. Now we have this
magnitude of laboratories. I think the same thing is
true of the biological environment. There aren't
that many labs doing this type of work.
I was very surprised that We had 13 or 14 even
submit. We were thinking more like six when we orig-
inally started to do it. So, we were pretty pleased
with the study. Of course, I don't know what's a
good critter test as...but I think the folks that do
critter things are fairly pleased with this and we
are going to try to move on with it. If we're going
to do another validation, that costs money. I prefer
not to spend my own. I prefer to spend R&D's. So,
that's where we're coming down the road on it.
MR. RICE: Jim Rice. I'd
like to make a little comment on that. You know, one
-------
280
of the problems that we have...we've debated the
issue of casting out data in every round I've ever
heard.
MR. RICE: Unfortunately
what happens, once you cast it out...we hear your
reasons heref but they get lost. The conclusions
that you draw on the basis of only 70 percent, in
effect, of the laboratories participating is what
tends to stay for a long time, and it's hard, people
make decisions on the basis of that sort of repro-
ducibility. It would be well if you, in whatever
you publish out of this, if you perform both analyses,
but a much more rigorous interpretation on what you
cast out.
I understand the point when you're trying to do
it for methods development. That's perfectly
legitimate. You really want to find out what the
core of the problem is and get some of the really bad
actors out, but when the data sits around, as this
does, and it becomes very important to others in
making decisions on whether to use it or what's
possible with it or not, I think you owe it to
everybody to put both analyses in.
MR. EYNON: Sure. I think
if you noticed on a couple of the slides, I had double
-------
281
line entries for the...well, I probably went right by
it, but some of the slides do have double entries
with all labs and with 5, 9 and 10 excluded. I felt
the most interesting results are when you get rid of
the labs because the bad labs are going to be the
ones with the most influence, too, and if that's due
to reasons that you have not been able to look at or
the lab failing to follow the protocol at all, you
can totally mess up your results. Because of the
high influence and stuff, I think that a much stronger
piece of information comes from looking at those labs
that appeared good.
Certainly we have all the data and all the
results for all the labs in our reports, and we felt
that this was the appropriate subset to be looking at
to gain the most information. I agree with you, as
you start to use it, things like this do sit around,
and it's a good point.
MR. WHITLOCK: Stu Whitlock
from ESE. Did you go visit the laboratories before
you sent them the samples, or did you just do it on
publication only?
MR. TELLIARD: We didn't
visit the labs at all, Stu. On the five lowest
bidders, like we do anytime we issue...this was an
-------
282
IFB that we're issuing...we visited those five
laboratories. I don't know if those were any of the
ones we threw out. Do we know?
MR. WHITLOCK: Well, let me
ask a second question first and then you can answer
that. When you looked at the types of animals used,
did you look at whether or not they purchased the
animals they used, whether they cultured their own
animals or what the situation was there? That could be
a very large variability that probably should be
considered.
MR. EYNON: That's one of
the factors we thought about.
MR. TELLIARD: After we put
out that IFB, the price of mysids went up. You
could buy lobster for that price.
MRS. DENAGY: Susan DeNagy,
EPA. I guess I can comment on your question. The
laboratories just were not arbitrarily thrown out.
In the procedure, the whole testing was done because
of a contract, and we have to underline that. It was
IFB, the laboratories, everything was predesignated
in the contract bid. In the procedure, if you have
greater than 10 percent mortality, you're thrown out.
It's just...that's it. So it's not arbitrarily thrown
-------
283
out, it's part of the procedure.
There are other stated requirements for that IFB
that if you...if the data did not reach that peak of
quality, you were told ahead of time you're invalidated.
So, it's not as if you're good, you're bad. Everything
was predesignated. The testing was controlled in the
respect that Gulf Breeze purposely tested and gave
out the SPP concentrations ahead of time, which would
not normally be done. So this was not a round-robin
in the true sense, but we were gaining a lot of
information and learning where variability would be
coming in from that point on.
MR. EYNON: One way to
characterize the labs that were thrown out, and I
guess a point I didn't really mention, is that Gulf
Breeze...the procedure is actually a two step procedure
and contains a range finding step. The laboratories
were told basically the results of a range finding
which had been done, because Gulf Breeze was familiar
with the mud. They were told what concentrations to
use so that that source of variability would be
controlled. One lab had a very bad control response,
and no question, they failed to meet the stated
specifications. The other two labs, one was well
above and the other was well below the stated range
-------
284
of the concentrations, and for that reason alone, the
results at those labs are very unreliable, if for no
other reason.
So, that's actually...we didn't say we're going
to throw out all labs who come in above and below the
range. We looked at the data and we said this guy's
way high, this guy's way low and we think it's best
for our purposes that we remove those from the study.
MR. TELLIARD: Dr. Tom.
DR. DUKE: I just want to
comment real quickly about the animal question. It's
a good question. But for the purpose of this
experiment...! won't say round-robin, I'm just going
to say project...we wanted to let people use the
animals as they would if they got a sample. Some
would culture them, some would send off for them, what-
ever. That was just one of the variables that we saw.
The other was seawater. Some people used natural
seawater, filtered. Some used sea salts that they put
together themselves, put their own seawater together.
There was a requirement of salinity, temperature, pH,
so forth and so on, but all of these things...and to
me it's...I still, as a toxicologist, I've seen
studies with single pure compounds, and about the best
we can do when they're sent nationwide like this with
-------
285
10 to 20 labs...I know of two studies where there's a
factor of four between the lows and the highs. It's
sort of amazing to me that we came this close with a
complex drilling fluid under the conditions that we
did. It's encouraging. It leads us to say, well,
now we can look forward and do some other things with
it.
MR. TELLIARD: Thanks, Barry.
It's time for a break. Is Mr. Brown here from
Battelle? Would he go to the registration desk at
break time?
All right, 10 minutes or 15 minutes we'll get
back in here. Thank you.
(WHEREUPON, a brief recess was taken.)
-------
286
Statistical Evaluation and Analysis of the EP|
Drilling Fluid Toxicity Test Procedure
Barrett P. Eynon, SRI International
R. Clifton Bailey, EPA
Objectives:
Evaluate and statistically analyze the
performance of the Drilling Fluid Toxicity
Test in support of the NSPS/BAT Effluent
Guideline for the Offshore Segment of thi
Oil and Gas Extraction Industry.
Aid in selection of contract laboratories.
Participants1
Industrial Technology Division and Analys
and Evaluation Division of the EPA Office
of Water Regulations and Standards, and
EPA Gulf Breeze Laboratory.
-------
Background
While developing I1SPS/BHT Effluent Limitations
Guidelines for the Offshore segment of the oil and gas
w1 »
extraction industry point source category,
287
are
Fluids mere identified as toxic materials rohich
discharged into the aquatic environment surrounding
offshore drilling operations.
Drilling fluids provide lubrication and assist in the
extraction of cuttings.
II, ISSi
-------
288
To regulate the discharge of drilling fluids
EPfl proposed a
Drilling Fluids Toxicitg Test
to be used as a compliance tool.
The Industrial Technology Division (OWRS)
in cooperation urith the
Office of Research and Development Laboratorg
in
Gulf Breeze, Florida
and
Rnalgsis and Evaluation Division (OWRS)
refined
a laboratory procedure
for
evaluation of the Drilling Fluid Toxicitg
in
Support of the proposed DSPS/BflT Effluent
Guideline for the Offshore Oil and Gas Industrg
and
Published the test in the
Federal Register. Vol. 50. Do. IBS, Dlondag, Rugust 28,
1965, Oil and Gas Extraction Point Source Categorg,
Offshore Subcategorg; Effluent Limitations Guidelines and
Dem Source Performance Standards; Proposed Rule.
flppendix 3-Drilling Fluids Toxicitg Test. Pages
34631-34635 Hfllso contained in the Development
Document).
I, Uli
-------
289
Hethod:
Toxicity analysis of drilling fluid by
dilution of the suspended particulate phase
(SPP) of prepared samples into dishes
containing 3-5 day old mysid shrimp
(Hysidopsi bahia).
Shrimp mortality was evaluated after 96
hours of exposure, at 5 dilutions (\%, 2%,
4%, Q%, and }Q% SPP) plus a control.
Mortality on a reference toxicant (soldium
laurel sulfate) was also measured.
Each test concentration was applied to
three dishes of 20 shrimp each. Shrimp
were assigned to dishes according to a
randomization procedure.
-------
290
antennule
antenna
aatcnnal ink
donalproco
natocyn
8th thoncic limb . pfeopods
abdominal Mtmenis
•leison
•eodoood
thoracic K^ntcnis donaj process
xopod
Figure 17. Lateral and dorsal view of a typical mysid.(From Stuck et al.,
1979).
B
Figure 18. Morphological characteristics used in mysid identification
(Mysidopsis bahia). A, antenna 1, ventral male; B, antenna 2;
C, telson; D, right uropod, dorsal. Scale lines 0.5 mm in
length. (From Molenock, 1969).
_
105
-------
2Q1
Data:
10 test laboratories applied the protocol to
an EPA provided sample of the same well-
mixed, pretested batch of driling fluid.
Dose levels were determined based on
preliminary analyses at reference lab
Results were compared with six replicate
trials at the EPA laboratory.
Additional information was collected on
possible covariate factors, including the
total suspended solids (TSS) of the sample,
the amount of acid used in the
neutralization of the sample, the pH and
dissolved oxygen (DO) of dishes each day.
-------
Calculation of mortality:
Probit model :
292
P(d) = C + (1-C) * ( (log(d) - ji ) /
d : dose
C : spontaneous response rate
0 : probit function
jj, : dose-response parameters
LC50 = exp( JJL ) : toxicity (concentration
with 50% excess mortality over control)
Fitted by maximum likelihood (SAS macro;
NLIN and MATRIX)
Other models evaluated:
Moving Average
Burrit
-------
293
DRILLING MUD TOXICITY RESULTS
RESPONSE
1.0
0.8
0.6
0.4
0.2
0.0
CTRL
WITH FITTED PROBIT CURVE
IAB=0
12 4
MUD CONCENTRATION {
32
DRILLING MUD TOXICITY RESULTS
ADJRESP
1.0
0.8
0.6
0.4
0.2
0.0
ADJUSTED FOR NATURAL RESPON9IVITV
LAB=0
CTRL
1 2 4
MUD CONCENTRATION («J
8
32
Figure IV-1 DRILLING MUD TOXICITY RESULTS: REFERENCE LAB
19
-------
294
DRILLING MUD TOXICITY RESULTS
WITH FITTED PROSIT CURVE
LAB=3
RESPONSE
1.0
0.8
0.6
0.4
0.2
0.0
CTRL
1 i 4
MUD CONCENTRATIONS)
16
32
DRILLING MUD TOXICITT RESULTS
ADJUSTED FOR NATURAL RESPONSIVITV
LAB=?
ACURESP
1.0
0.8
0.6
0.4
0.2
0.0
CTRL
32
MUD CONCENTRATION (f ]
Figure IV-4 DRILLING MUD TOXICITY RESULTS: LAB NO. 3
22
-------
DRILLING MUD TOXICITY RESULTS
205
RESPONSE
1.0
0.8
0.6
0.4
0.2
0.0
CTRL
WITH FITTED PROBIT CURVE
LAB=4
124
MUD CONCENTRATION <
16
DRILLING MUD TOXIC ITT RESULTS
ADJRESP
1.0
0£
0.6
0.4
0.2
0.0
CTRL
ADJUSTED FOR NATURAL RE9PONSIVITV
LAB=4
1 2 4
MUD CONCENTRATION^)
16
32
Figure IV-5 DRILLING MUD TOXICITY RESULTS: LAB NO. 4
23
-------
Toble
v-e
summflRY OF BBBKS
FOB
TEST LHBOBRTOBIE5
BY
CBITEBIR
296
Criterion
Lab
*
1
2
3
4
5
6
7
8
9
10
Lab
vs.
Ref . Lab
3
1
5
7
*
2
4
8
6
9
Lab
vs.
Other
Labs
5
2
1
6
*
4
3
7
8
9
Width
Confidence
Interval
In-Mftl
3
1
7
7
6
2
4
8
9
10
Iff-Ogl
1
5
7
8
9
2
3.5
3.5
10
6
ji
3
6
7
1
10
5
4
2
9
8
0*
4
6
7
1
10
5
3
2
9
8
Not ranked
-------
I
m
1JO-
0.8-
OJ6-
0.4-
02-
Ttsl Labs 1,2,3,4,6.7.6
297
OJO
i
1JO-
OJB
OA-
02-
OJO
Ml
EPA Gulf Breeze Lob
MU
-------
Tests of Intro-Laboratory Variability IProbit Model, Reference Lab]
Drilling Fluid
Trial C Hu Sigma LogLik
A 0.00 0.87 0.74 -110.55
B 0.08 1.51 0.42 -113.92
C 0.07 0.79 0.78 -133.16
D 0.15 1.47 0.48 -140.87
E 0.01 0.94 0.59 -99.67
F 0.02 0.88 0.82 -128.31
Total -726.48
All* 0.04 1.01 0.76 -764.58
Chi-Sq llntra-Lab] 76.21
Of 15
All** various 1.04 0.73 -757.68
Chi-Sq ilntra-lab] 62.40
df 10
Chi-Sq [common spont. resp.l 13.81
df 5
Reference Toxicant
C Hu Sipa LogLik
0.04 2.01 0.34 -29.50
0.00 2.08 0.09 -13.87
0.10 1.78 0.43 -43.34
0.00 1.80 0.68 -35.47
0.05 1.98 0.10 -24.34
0.04 1.97 0.40 -31.84
-178.36
0.04 1.92 0.41 -197.19
37.65
15
various 1.94 0.39 -189.98
23.23
10
14.42
C
«
* with common spontaneous response rate
** with separate spontaneous response rate
-------
299
Pisbit Model and Summed Replicate Likelihoods
Diilling Fluid
Lab
Total - All
Tstal -75,9,
_ogLik(Model Loglik(Rep)
0
1
2
3
4
5
6
7
8
9
10
10
-110.55
- 1 08.02
-154.31
-174.60
-83.49
-52.53
-131:29
-126.80
-99.6 1
-87.52
- 1 1 0.50
-1239.22
-988.67
-100.57
-97.37
-144.72
-163.65
-73.59
-43.38
-125.96
-117.12
-88.14
-81.74
-100.96
. -1137.20
-911.12
Chi-Sq
19.95
21.31
19.19
21.90
19.79
1 8.30
1 0.66
1 9.36
22.94
11.56
19.08
204.04
155.10
Df
15
15
15
15
15
15
15
15
15
15
15
165
120
Signif
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
#
*
Notes
"fets of Interlaboratory Differences
CMC-All -1558.17 vs Total Model 637.90
Fised O Al 1 -1713.35 vs Total Model 948.26
Opt C/5,9,10
FisedC/5,9JO
-1122.19 vs Total Model 267.04
-1161.00 vs Total Model 344.66
20 **
20 ** (Approx)
14 **
14 ** (Approx)
-------
likelihoodTflblefor Inter/I ntrc Laboratory Effe cts
MaterialDrillingFluiil
L8bs:0(A-F)l234678
300
Model
Probit*
Probit/Lab*
Probit/Trial
PeiDose/Trial
PerRep/Trial
Complete
Parameters
15
29
39
78
234
4680
Up
-1790.51
-1635.80
-160460
-154L4Z
-1441.52
0.00
Effect
Inter-Lab
Intra-Lab
6-of-fit
Hetero-Rep
Residual
df
14
10
39
156
4446
Chi-Sq
309.42
62.40
112.35
213.81
2883.04
Z
55.83
11.72
8.31
3.27
-16.57
* Spon, Resp final
-------
301
Inter and Intra-Laboratory Variation
A/
Jbunu-Cu?
Co SH&mf
s
*>•««
Sff)
2?
£.0
9.9
/£?*-
-------
302
Covariate models :
Jl = JiO + jll Xj + J12 X2
log(tf) =
X2
Maximum likelihood fitting allows
estimation of parameters, calculation of
confidence intervals, and testing of
significance of covariates.
-------
303
Ue measure the variation to know what
can be expected. Ue attempt to explain
the variation so that we can improve
the test method. Ue can improve the
test method by
a) fixing or specifying influential
factors in the protornlr
e.g. specify materials used,
temperature and other
condidtions,
b] introducing procedures that
compensate for sources nf
variation, e.g. mixing,
randomization, blinding
c) numerically adjusting the
results to remove the
influence, e.g. the correction for
spontaneous response.
-------
2-2
3
0-
-1 -
7.80
7.85
LEGEND. LAP
SIGMA
1.50 -d
1.25 -
1.00 -
0.75 -
0.50 -i
0.25 -i
7.80
LEGEND. LAB
7.85
304
x
7.90
7.95
4-4-4 0
<:> c o A
2226
8.00
PH
y. x x
A £ A
8.05
1
5
9
8.10
* * * 2
« * * 6
A * * 10
8.15
.620
D D D
V Y V
x
7.92
4-4-4- 0
C O O 4
222 8
7.98 8.04
PH
XXX 1
• • i
8.10
8.16
8.22
* * * 2
s* » * 6
* * * 10
ODD
Y Y Y
7
Figure VI-4 PLOTS OF TOXICITY PARAMETERS VERSUS PH
45
-------
305
COVARIATES: InTSS.ACID.HUrcf UBS: 0123469
C:
Mil NPM1 InTM* IPIIS
nu sibnA miss ACID
L
P
InTSS
ACID
HUM
MSS,ACID
0.92
•1.19
1.20
0.07
•0.95
•1.19
0.57
•0.39
1.02
0.83
0.99
0.98
0.82
0.83
0.98
0.83
0.23
0.22 -0.04
0.23
-0.06
0.23 -0.09
-1057.3
-906.3
-10315
0.40 -1030.9
-902.6
0.01 -906.3
0.24 -1028.0
-0.27 -897.8
ACID | InTSS
ACIDlnilref
MSSIACID
ISSlhUref
HUrefllnTSS
InTSSlACID.HUref
ACIDIMSS,
302.0 1
51.5 1
52.8 1
309.3 2
302,0 2
58.5 2
318.9 3
7.3 1
5.7 1
257.8 1
249.2 1
7.0 1
0.0 I
260.4 I
9.6 I
16.9
).017
1002
-------
306
IIUHiTB: MS.ro.HM UK: 0123465
fc FIXED
HU
P
\m
m
m\m,
i)
•1057.3
•897.7
0.92 1.02
-1.82 IJ5 029
1.14 1.11
022 0.66 0.33 0.18 -1028.4
•1.54 1.78 027 -0.08 -0.02 -0.02 -895.9
•2.10 1.32 0.38 -0.16 -0.27 0.48 -877.8
0.58 0.94 -0.04 -0.04 021 -0.04 -1025.9
-1.37 0.49 0.34 -0.16 -0.05 0.11 -0.36 0.79
319.02
56.32
57.82
4
4
62.84
375.06
3.8 2 0.151
5.02
266.62
331.02
6.52
39.82
31232
5222
1622
March 8, 1986
-------
307
COVARIATES: iVflDO.pH
LABS: 01254578 C: Fie
none
avgDO 0.73 0.86
avgpH 30.25 0.87
avgDO.avgpH 29.84 0.87
avgDOIavgpH
avgpHlavgDO
CDVARIATES: avgDOJ
-1159.3
HYPOTHESIS HI SIGHAavgDOaflipH L CHI-SQ DF P
3.4 I 0.064
145.9 I 0.000
147.8 2 0.000
1.9 I 0.170
144.3 I 0.000
-3.64
0.06 -3.64
-1087.1
LABS: 01234678 C: OPTIflli
HYPOTHESIS HU SIGHAiVflDOavopH L CHI-SQ DF P
11.8 I 0.001
none 1.50 0.66 -1122.2
avgDO 0.42 0.64 0.16 -1116.3
avgpH 29.64 0.78 -3.55 -1068.4 107.7 I 0.000
avgDO.avgpH 28.15 076 0.13 -3.47 -1064.8 114.7 2 0.
avgDOIai/gpH
avgpHlavgDO
7.1 I
103.0 I 0.000
Morch 8, 1986
-------
308
TI
-^•H
i
n
IV
-t»
£
9-
"v^
)(
II
fi
-I-
CN
'— i
fc=» a
^N
?
*
z n
WB^iX
JP ^1 1
--, V
' l P «x:
»•-- ^
r
»
w
O
>
p5
k>
r\>
ro
P p p p p p p p p p p
O O O O — Co O O O O O
cy» o —* iQ —» 10 04 10 cn -ES. o
i
ro o ro —* —"O —• —»ooc
— — o4O4O-urocnvQ->j
p — o — — —
i ii
—» i i — — i
Co
ppppppppp p p
ooooocoooooo
sj —
O — — — O —
ro — — — *
Co
en — \o oo o to
— rocn — roro
O O O O O — * .— o O — —
ii ii
i i — — i • —* —
K>
Osi — ro— '
— Co OJ ro
ro ro ro ro ro ro ro ro ro ro ro
m
n
o
-n
m
co
O
GT>
i
ro
-------
309
MR. TELLIARD: Our next
speaker this afternoon is John Brown from Battelle.
He's going to describe some work that has been going
on with the Offshore Operators Committee, which is a
group within API.
We have proposed regulations for offshore oil
and gas, and one of the things the industry has really
appreciated is, of course, we've always been nice to
them, the question of prohibition in the regula-
tions on the use of diesel as a lubricity agent in
the drilling muds.
One of the problems that we have looked at in
this is how do you analyze drilling muds for number
two diesel fuel for the purposes, as you might want
to say, of enforcement...that's what we would want to
say...since it will be a condition in their permit,
which we've all grown to love and desire.
These general permits are due out shortly in the
Gulf and there are two proposed in Alaska and one in
California. So this work is rather important, not
to us, because after all, we can enforce without a
method, but to the industry at hand. They have taken
somewhat of a different view. John.
-------
310
JOHN BROWN
BATTELLE N.E.M.R.L
ORGANIC CHEMICAL CHARACTERIZATION OP DIESEL
AND MINERAL OILS USED AS DRILLING MUD ADDITIVES
MR. BROWN: In recent
years, concern over the impact of toxic diesel oil
components on benthic and pelagic communities sur-
rounding offshore drilling platforms has promoted
government agencies to issue a ban on the ocean
discharge of diesel-containing muds. In response,
the offshore oil industry has resorted to the use of
alternative lubricants, such as mineral oil, in
drilling muds destined for ocean disposal. The
selection of mineral oil was based on a presumed
decreased toxicity due to its lower aromatic hydro-
carbon content relative to diesel oil.
In anticipation of new federal regulations
banning the use of diesel oil in drilling muds,
Battelle New England was contracted by the Offshore
Operators Committee to initiate a multiphase study
aimed at developing an analytical method capable of
measuring the diesel content of drilling muds, as well
as distinguishing between individual diesel oils and
-------
311
mineral oils in mud formulations based on differences
in their organic composition.
One method proposed by EPA to analytically
measure diesel oil in drilling muds, what we call the
Top 10 method, involves GC/FID analysis of a drilling
mud extract and subsequent quantification based on
the concentration of the 10 major peaks in the sample
chromatogram relative to the same 10 peaks in a
chromatogram of a reference diesel oil.
This method appears to be analytically sound and
capable of yielding accurate quantitative results,
as is evidenced by the following slides of GC/MS
reconstructed ion chromatograms, or RIC's, of selected
mineral and diesel oils. These RIC's are virtually
identical to chromatograms that would be obtained
from GC/FID.
SLIDE 1
The first slide is an RIC of Mineral Oil A,
ortho-turphenyl is the added internal standard.
SLIDE 2
This figure represents another mineral oil,
Mineral Oil C.
SLIDE 3
This slide is representative of a California
diesel oil.
-------
312
SLIDE 4
This one is representative of a low sulfur
diesel.
As evidenced by the last four slides, there
appears to be compositional differences between
mineral and diesel oils which should be easily
distinguished by the Top 10 method.
SLIDE 5
However, in certain cases, a potential ambiguity
exists, as shown by this slide of Mineral Oil B, on
top, and Alaskan diesel on bottom. A comparison of
the chromatographic pattern exhibited by each oil
reveals that they are nearly identical, with the
exception of the relative abundance of the added
internal standards ortho-terphenyl. In fact, our
results show that if the EPA Top 10 method is applied
in this instance, Mineral Oil B is mistaken for diesel
oil.
These potential ambiguities emphasize the need
to develop tracers of specific additive types beyond
chromatographic pattern matching, which can serve as
definitive1 quantitative indicators of the presence of
mineral and diesel oils.
SLIDE 6
Nine oils representative of those used in
-------
313
different offshore regions were analyzed by the best
available methods for the following targeted compound
classes: organic sulfur, total sulfur and total
nitrogen; sulfur-, nitrogen- and oxygen-containing
PAH, or PAC's, carboxylid acids, phenolic acids,
aldehydes and ketones, phenol and its alkyl homologs,
up to C4, individual aromatic hydrocarbons, and total
aromatic content.
SLIDE 7
PAH analyses were performed using a modified
version of Standard Method D3239. The results
presented here show the percent total aromatic
content of the nine oil samples. The mineral oils
are found to have significantly lower aromatic con-
tent, with Mineral Oil A exhibiting the highest value
of 10.2 percent.
Among the diesel oils, EPA Number 2 fuel oil had
the highest value, 35.6 percent, while the remainder
ranged from 11.7 to 29 percent.
The total concentration of individual PAH and
their alkylated homologs, through C5, in general
mirror the total aromatic content of each respective
oil. However, distinct differences were found among
the distributions within the individual oils. The
mineral oils are found to have signficantly lower
-------
314
concentrations of benzene, naphthalene and their
alkyl homologs than the diesel oils.
The differences both in total and individual
aromatic contents of the oils indicate the potential
to use these parameters in future analytical programs
designed to distinguish between mineral and diesel
oils in actual mud samples.
SLIDE 8
Organic sulfur and total dibenzothiophenes were
determined by GC Hall BCD with dibenzothiophene peak
identifications being made by GC/MS. The results
show the organic sulfur, total dibenzothiophene and
percent dibenzothiophene concentrations of the diesel
and mineral oils analyzed. It is evident that the
mineral oils generally exhibit lower sulfur and
dibenzothiophene levels than the diesel oils, with
the exception of the low sulfur diesel.
SLIDE 9
This figure represents Hall BCD chromatograms of
four of the oil samples analyzed. Figures A and C,
corresponding to Mineral Oil A and California diesel,
are representative of the two extremes of organic
sulfur content, with Mineral Oil A essentially devoid
of peaks and the California diesel containing many
peaks, including a broad, unresolved hump. Thianthrene
-------
315
is the added internal standard.
The majority of the samples exhibited a distribu-
tion intermediate between these two extremes,
which are represented by Mineral Oil B and Gulf of
Mexico diesel. Mineral Oil B. Gulf of Mexico.
SLIDE 10
Total sulfur was determined by the oxygen bomb
method, Standard Method D129 Modified, and total
nitrogen was determined by the chemi-luminescence
method.
The sulfur contents of the three mineral oils is
low and differs by an order of magnitude from those
found for the diesels. The mineral oils also exhibit
a lower nitrogen content than the diesels, although
the differences here are not as great.
As with the aromatic hydrocarbon compositions,
the large differences in the sulfur contents of the
mineral and diesel oils make this a desirable para-
meter to use in future analytical programs, while
the use of total nitrogen would not be recommended.
SLIDE 11
The phenol and alkylphenol concentrations of the
nine oil samples were determined by GC/MS analyses.
The important feature of this data is the presence of
phenolic compounds exclusively in the diesel oils.
-------
316
No phenolic compounds were detected in any of the
mineral oils analyzed. This is a highly significant
finding which suggests that the alkylated phenols are
another suitable compound class for distinguishing
between diesel and mineral oils in actual drilling
mud formulations.
SLIDE 12
This figure shows a mass chromatogram map of the
alkylphenols in two of the diesels analyzed. The
distribution is analogous to those found in the
aromatic hydrocarbons in which the alkylated homologs
are prevalent over the parent compound in any one
series. You can see, in this case, phenol is absent.
Cl, C2, C3 and C4 phenols occur in greater concentra-
tions, and the same pattern here.
To sum up the results of Phase I, it was found
that quantitative differences between diesel and
mineral oils are evident in total aromatic, total
sulfur, organic sulfur contents, as well as in the
concentrations of individual PAH. In addition, phenol
and its alkyl homologs have been identified as a
compound class which has the potential to, by its
presence or absence, denote the occurrence of diesel
oil in drilling mud formulations.
The phenolic acids, aldehydes and ketones were
not found in measurable quantities in any of the oils,
while the carboxylic acids, the sulfur, nitrogen and
-------
317
not found in measurable quantities in any of the oils,
while the carboxylic acids, the sulfur, nitrogen and
oxygen PAC's, were detected in most of the oil samples,
but their compositional differences did not allow
differentiation between diesel and mineral oils.
The purpose of Phase II of this study was to
evaluate the efficiency of two extraction techniques;
retort distillation and solvent extraction, in
isolating the diesel oil organic tracers from actual
drilling mud formulations. In addition, we hope to
validate the analytical techniques from Phase I when
applied to drilling mud samples.
The compound classes from Phase I which were
considered to have the greatest potential to
differentiate between diesel and mineral oils were
the alkylphenols and the individual PAH. Total sulfur
and organic sulfur determinations were not chosen for
Phase II due to possible matrix interferences when
analyzing lignosulfonate muds.
SLIDE 15
You probably can't see this in back. I'll try
and walk everyone through it. This figure represents
the analytical approach used in Phase II. Briefly,
drilling mud samples were acidified to pH 1, mixed
with sodium sulfate after which the internal standards,
-------
318
orthoterphenyl and d-8-p-cresol were added. The mud
formulation was then extracted at ambient temperature
on a shaker table, two times with methanol and two
times with a 9:1 methylene chloride methanol mixture.
The combined organic extracts were partitioned
versus one normal hydrochloric acid, the organic
phase isolated, and the aqueous phase extracted again
with methylene chloride and ethyl ether.
At this point, the methylene chloride ether
extract was dried over sodium sulfate and a five
percent aliquot removed for extractable weight
determination and subsequent aromatic hydrocarbon
analysis.
The remaining extract was partitioned versus
base which was then acidified and extracted with
ethyl ether to isolate the alkylphenols. Both the
phenols and the aromatic hydrocarbons were analyzed
by the same electron impact GC/MS procedure used in
Phase I.
The retort distillates, containing the same
internal standards used in the solvent extractions,
were introduced into this analytical scheme just
prior to the acid water partitioning step and were
carried out through the remainder of the procedure in
a manner identical to the solvent extracts.
-------
319
The retorts were introduced here and were carried
out identically through the rest of the scheme.
The results of Phase II indicate that phenol
analysis by high temperature retort is flawed due to
artifact formation and should not be considered. It
was found that considerable concentrations of phenols
were produced in the drilling mud matrix, even with a
modified low temperature retort distillation.
It was determined that the solvent extraction
method be adopted for isolating phenols and hydrocarbons
from drilling muds since our preliminary analyses
have shown that this method yields accurate and
internally consistent data which compares favorably
with the composition determined previously in Phase I
for the neat Alaskan diesel.
In the future, we plan to conduct further
development work to purify the phenol isolates from
the solvent extracts, which would permit direct GC/MS
analysis. We've recommended that a derivatization
and subsequent absorption column chromatography be
explored, as initial results seem promising.
In addition, a column chromatography step to
clean up aromatic hydrocarbon extracts prior to the
GC/MS analysis would be recommended since this would
permit better chromatography and subsequently improve
-------
320
our accuracy and precision. We also plan to analyze
drilling muds with a mixture of additives, including
crude oil, and analyze drilling muds which have been
hot rolled at high temperature, and ultimately analyze
wild muds from offshore drilling platforms.
We hope that the final product of this research
will be an analytically validated method for determin-
ing the presence of diesel oil in drilling muds.
-------
321
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
DR. FRIEDMAN: Paul Friedman,
EPA. Have you looked into IR as a method for
characterizing these waste oils or oils in drilling
muds?
MR. BROWN: No, we haven't.
At the time of the study we didn't have IR capabilities.
DR. FRIEDMAN: Thank you.
MR. TELLIARD: No one else?
You're going to let him off this easy?
-------
. 322
ORGANIC CHEMICAL CHARACTERIZATION
OF DIESEL AND MINERAL OILS USED
AS DRILLING MUD ADDITIVES
A.G, REQUEJO
J.S. BROUN
P,D. BOEHM
-------
323
OBJECTIVES OF PHASE I
TO DEVELP AN ANALYTICAL METHOD CAPABLE OF
DIFFERENTIATING BETWEEN INDIVIDUAL DIESEL AND MINERAL
OILS, BASED ON DIFFERENCES IN THEIR ORGANIC COMPOSITION,
-------
-------
325
'
fev-
-------
-------
-------
MO-B-6-84-7
Alaska Diesel
GC/MS/OS Ur:.CONSTRIJCTFin ION CHROMATOC.RAMS (RIC) OF THE
MtNF.RAL OIL MO-B-f.-g';-? ANH THi:. M.A.SKAN nif-SIiL. NOTE; THf!
SIMILARITY IN THI- CHF^OMATOC.R APHIO I'ATTF-KN HXHIIMTfin BY
EACH. o-TF.RPHFlNYL IS THE AHOI-in INTI'KNAL STANOARO (IS).
-------
-------
330
PERCENT AROMATIC CONTENT OF DIESEL AND MINERAL OIL ADDITIVES
SAMPLE
PERCENT
MINERAL OIL
MQ-A-6-84-3
MO-B-6-84-7
MO-C-6-84-19A
10,2 ±0,7
2,1
3.2
DIESEL OIL
LOW S DIESEL
HIGH S DIESEL
GULF OF MEXICO DIESEL
ALASKA DIESEL
CALIFORNIA DIESEL
EPA-API NO. 2 FUEL OILA
16,1
29,0
23.8
11,7
15.9
35,6 + 3,9
ATRIPLICATE DETERMINATIONS
-------
. 331
ESTIMATES OF ORGANIC SULFUR CONTENT AND TOTAL
DIBENZOTHIOPHENES CONCENTRATIONS (PARENT COMPOUND TO
C3) USING GC/HECD,
SAMPLE
DIBENZOTHIOPHENES £
MIN
ERAL OIL
(UG/6 OIL)
i
[II
0 DIESEL
-APT RO.~2~FUEL OIL
rfBrra
-------
-------
cr\ oo "—*
O ox O
CM ^ CM
I I
< c6 u
666
Jfi
-------
334
s
|T
PC
U U U
11 ii i \' uucor"
t——tO I ' IU7
5 KT
m
*^
25
mF"
P2
•n:
:C3rp
m
-o
m
o
m
CO
70
O
CO
-------
ON OO ^
rf\ CM fS
oQ CO 6O
< ob u
666
-------
334
LJLJLJLJOC
UOLJLJOC
en
oo
Cd
en
oo
mac
en men
cr>
r—oo
CO
ro
-------
335
Gulf of Mexico Diesel
p—cresol | | m-*-p—cresol
'&£ phenols
.phenols
California Diesel
o- cresol
C-2 phenols
"1288 SCAN
28:08 TINE'.
C4 phenols
C^ phenols. /'
phenol (absent)-:
I:288-.-SCflN:-
28:88"TIME
GG/MS/DS MASS CHROMATOGRAM MAPS CORRESPONDING TO THE
ALKYLATED PHENOLS FOUND IN THE GULF OF MEXICO DIESEL AND
THE CALIFORNIA DIESEL. •;, /V'"x: O- • —" : :^--V.,-V-?.;';^
-------
. 336
CONCLUSIONS OF PHASE I
QUANTITATIVE DIFFERENCES BETWEEN DIESEL AND MINERAL
OILS ARE EVIDENT IN TOTAL AROMATIC, TOTAL SULFUR, AND
ORGANIC SULFUR CONTENTS, AS WELL AS IN THE
CONCENTRATIONS OF INDIVIDUAL PAH (BENZENE, NAPHTHALENE,
BIPHENYL, FLUORENE AND PHENANTHRENE ALKYL HOMOLOGUE
SERIES).
THE IDENTIFICATION OF ALKYL PHENOLS AS A COMPOUND CLASS
WHICH HAS THE POTENTIAL TO BY (ITS PRESENCE OR
ABSENCE), DENOTE THE OCCURRENCE OF DIESEL OIL IN MUD
FORMULATIONS.
-------
. 331
OBJECTIVES OF PHASE II
TO EVALUATE THE EFFICIENCY OF TWO EXTRACTION TECHNIQUES
(RETORT DISTILLATION AND SOLVENT EXTRACTION) IN
ISOLATING THE ORGANIC TRACERS (IDENTIFIED IN PHASE I)
OF DIESEL OILS FROM DRILLING MUDS.
-------
-1 30g Drilling
Mud
| I. Acidify with !.*ml6NHCI
2. Mix with 50g Na2SO$ (3:1 wet weight:Na2SO<,)
3. Add internal standards
Drilling Mud
I. Extract with methanol (2x, 100 ml each)
2. Extract wtth 9:1 methylene chloride:methanol (2x, 100 nil each)
3. Centrifuge after each extraction (3000 rpm, 10 min)
Combined
Methanol/Methylcn
Chloride Extracts
1. Partition v. 100 ml IN HC1
2. Isolate organic phase
3. Extract aqueous phase with methylene chloride
(50 ml) and diethyl ether (50 ml)
Combined
Methylene
Chloride/Ether
Extracts
II. Remove 3% aliquot
2. Dry over Na2SO$
Ielectrobalance
*. GC/MS
Aromatic
Hydrocarbons
1. Partition v. 100 ml
IN NaOH
1. Acidify with 6N HC1
2. Extract 3x with diethyl ether (50 mi each)
I Combined
Ether
I. Dry over Na
2. Concentrate
3. CC/MS
Alkyl Phenols
-------
339
CONCLUSIONS OF PHASE II
1. THE ANALYSIS OF INDIVIDUAL PHENOLS BY HIGH TEMPERATURE
RETORT DISTILLATION IS FLAWED DUE TO ARTIFACT FORMATION
AND SHOULD NOT BE CONSIDERED, PRELIMINARY EVALUATIONS
OF THE FEASIBILITY OF A LOWER TEMPERATURE RETORT
APPARATUS INDICATE THAT THESE ARE SIMILARLY SUBJECT TO
ARTIFACT FORMATION,
2, SOLVENT EXTRACTION SHOULD BE ADOPTED AS THE METHOD FOR
ISOLATING PHENOLS AND HYDROCARBONS FROM DRILLING MUD
SAMPLES, PRELIMINARY ANALYSES HAVE SHOWN THAT THIS
METHOD YIELDS ACCURATE, INTERNALLY CONSISTENT DATA
WHICH COMPARES FAVORABLY WITH THE COMPOSITION
DETERMINED PREVIOUSLY FOR THE NEAT ALASKAN DIESEL,
-------
340
FUTURE WORK
1, CONDUCT FURTHER DEVELOPMENT WORK TO SUFFICIENTLY PURIFY
THE PHENOL ISOLATES OBTAINED FROM SOLVENT EXTRACTS TO
PERMIT DIRECT GC/MS ANALYSIS. WE RECOMMENDED THAT
DERIVATIZATION AND SUBSEQUENT ADSORPTION COLUMN
CHROMATOGRAPHY OF THE PHENOL FRACTION BE EXPLORED AS
INITIAL RESULTS SEEM QUITE PROMISING,
2, INCORPORATE AN ADSORPTION COLUMN CHROMATOGRAPHY STEP TO
CLEAN-UP SOLVENT EXTRACTS PRIOR TO GC/MS ANALYSIS OF
AROMATIC HYDROCARBONS. THIS WOULD PERMIT BETTER
CHROMATOGRAPHY OF THE HYDROCABRON ISOLATES BY REMOVING
INTERFERING POLAR MATERIAL. WHICH IN TURN WOULD IMPROVE
ACCURACY AND PRECISION.
3. ANALYZE DRILLING MUD FORMULATIONS WITH A MIXTURE OF
ADDITIVES (INCLUDING CRUDE OIL). ANALYZE DRILLING MUDS
WHICH HAVE BEEN "HOT ROLLED" AT HIGH TERMPERATURES, AND
ULTIMATELY ANALYZE "WILD" DRILLING MUDS.
-------
341
MR. TELLIARD: Our next
speaker is going to talk on critters and the joys
thereof, and how you can use critters to find out
what's really bad around the house.
-------
342
ROBERT C. BARRICK
TETRA TECH, INC.
CORRELATION OF BIOLOGICAL INDICATORS
WITH CHEMICAL ANALYSIS DATA
MR. BARRICK: Thanks, Bill.
Actually, about a year ago at the last conference
Peggy Knight of Weyerhaeuser Technology Center
gave a summary of some results that they had
been working on using an isotope dilution technique
for very low level organic analyses of marine
sediments.
Dale Rushneck and I thought I might give
you some presentation on what happened to those data,
which are part of a larger data set for the Commence-
ment Bay Superfund investigation.
In that study, there were a number of chemical
biological measurements made. We threw these into a
blender and tried to come out with some fingers
pointed in one particular direction or another. What
I'm going to talk on today is a little bit about how
that came about, exactly what the requirements were,
and where we ended up. I'd like to have the slides,
please.
-------
343
SLIDE 1
The subject of my talk is specifically
correlation of biological indicators with
chemical analysis data. These correlations are
not proof of cause/effect relationships, but they're
suggestive of associations that you might see with
chemical and biological data.
I want to give credit to the Contract Laboratory
Program which was instrumental in providing the neces-
sary- funds to do the chemistry. The developmental
work for the procedures that we used, for example,
the isotope dilution technique, was a joint effort
between Tetra Tech and California Analytical Labora-
tories. Additional analyses for metals were done by
Rocky Mountain Analytical Laboratory. Weyerhaeuser
Technology Center provided additional organic analyses
and other people in Tetra Tech assembled the biological
data.
Basically, you've got a problem out in the
environment that can be measured. You've got industry
and other sources putting out chemicals into the
environment. We know that that's happening. We've
measured the waste streams, and we can measure chemi-
cals in the environment.
We also know, for example, that there are a
-------
344
lot of people that use the marine environment, as in
this slide, fishermen. These people can't go down
and see the chemicals. What they do are see the
effects.
Some of these effects can be measured. For example,
in this slide are livers of English sole. The liver
on the left has visible lesions. These are actual
malignancies. The liver on the right is a healthy
liver.
These are the kind of things that the public
sees. These are the kind of things that they react
to.
The next slide is a characterization that came in
an editorial cartoon. If you can't read in the back,
it talks about how the defendant is admitting that,
yes, she fed her husband Commencement Bay bottom fish
three nights in a row, which led to his demise,
presumably.
So, this is what the public sees, this is what
they react to, and this is what they want to have
explained.
Besides the health risk questions, we have other
kinds of problems in the environment; ecological
questions. These organisms in the next slide are
benthic infauna, or to analytical chemists like me,
-------
345
these are clams, worms and other slimy things. But
they react to chemicals, too, and we can measure the
effects, we can get some indices on effects, and we
can relate them to the chemistry.
So, basically, in the Commencement Bay Superfund
study, we focused on the sediments. In the sediments,
as seen in the next slide, we can measure the
chemistry on identical homogenates that are also
tested using bioassays. We can also collect
samples at the same station, same time, for the
benthic infauna (or worms).
We also looked at fish. In fish we can look at
bioaccumulation of substances in muscle tissue
and livers, and also look at exactly what's going on
with the pathology of the livers.
These are all different kinds of indicators.
They are relatively independent, even though they
might be measuring the same kind of thing, and we
need a way to link them together.
Some of the things we're going to be talking
about later on have to do specifically with the
results of bioassays and infauna. These bioassays
shown in this slide don't involve the National
Critter. Instead, we've got the national cousin,
specifically the Rhepoxynius abronius amphipod, which
-------
346
is used to measure toxicity via mortality.
We also used oyster larvae embryos as a check
on abnormality of organism development. That's
considered more of a chronic test. The abundance of
infauna is another biological indicator; the numbers
per square meter in a particular sample. And
chemistry/ you guys know"all about that.
This slide is an example of some kinds of relation-
ships we can see. On the vertical axis is plotted the
abundance of Praxillella gracilis; that's a worm.
On the horizonal axis is the concentration of pyrene,
a polynuclear aromatic hydrocarbon.
For the statisticians in the audience, I have
arbitrarily and capriciously drawn a line on this
slide. I have arbitrarily and capriciously indicated
a point which appears to be a transition from a wide
ranging abundance to relatively low abundance at a
particular concentration of pyrene. This kind of
relationship is suggestive of an association between
pyrene concentrations and the abundance of this
organism. It doesn't prove it. There may be other
chemicals out there producing the effect. Typically,
in the environment, we're dealing with a chemical
soup and pyrene is just one of the constituents
of that soup. We need a mechanism to unravel what
-------
347
things might be problems in a problem sediment, and
a means to actually identify a problem sediment.
Then, once we do that, we can start getting back to
the source and try and do some remedial action.
The kind of responses you may see in a bioassay
is shown in this slide. On the vertical axis we have
the response of the oyster larvae bioassay, the circles,
and the amphipod bioassay, the squares. These
bioassays tend to show the same behavior. Increasing
mortality and increasing abnormality appear to
occur with increasing concentration of 4-methyl-
phenol.
There's a very good relationship here...and in
fact, the samples that make up this data set go off in a
gradient from an outfall that is located at the sample
point in the upper right hand corner of the slide.
There's a decreasing gradient away from the outfall.
So, we see decreasing biological effects, decreasing
chemical concentrations, and an apparent relationship
between the two with distance from the outfall.
What we're left with is coming up with some idea
of how we sort all this information out. We had
the requirements shown in the next slide. First,
we needed to develop our relationships from field
data. Laboratory cause/effect data are sparse or
-------
348
nonexistent in a lot of cases, and their
applicability to the field has not really been
proved.
Second, we needed to provide chemical specific
values. We couldn't just go take a bioassay and say
this is a problem, this industry caused it. We
needed to have some way to demonstrate a problem;
these are the chemicals we suspect, these are the
chemicals that are associated with this particular
source.
Third, we also had to integrate several bio-
logical indicators. We wanted the system to be
driven by statistically significant effects. In
other words, the effect is statistically different
from what we see in a control area or reference
area which is presumably not contaminated, or contami-
nated very little.
Finally, we wanted to have something that was
supported by fairly strong evidence, non-contradictory
evidence. We didn't want to have to stand up in
court and say, this is contaminated and it caused
this effect while at another site the same
concentration was observed and there was no effect.
In this instance, how can you say there's a problem.
So, we wanted it to be fairly noncontradictory so we
-------
349
would have strong evidence of a problem.
The next slide is a schematic of what we worked
out that fits these different criteria. The
upper axis displays the elevation above reference
for lead. You don't need to pay too much attention
to that. All it means is that as you go along the
axis, concentrations of lead at the study area are
getting higher and higher relative to concentrations
at our reference area. On the bottom axis concen-
trations of lead are plotted.
The chemical data are divided into two general
groups on this slide. One group of stations had
statistically significant biological effects and the
remaining stations had no significant effects.
Stations indicated by the top bar on the slide had
no significant benthic depressions. In other words,
the abundances of infauna were similar at these
stations to what we found in the reference area.
There are about 32 of these sites, and what is plotted
is the range in lead concentrations for those
samples.
This center bar on the slide represents lead
concentrations at all the stations where we
found no significant toxicity by either of our
bioassays. Over this range of lead concentrations
-------
350
there was, apparently, for these samples, not a problem
with toxicity.
The bottom bar on the slide represents lead
concentrations at all of our stations where one kind
of effect or another was observed; either toxicity
or benthic depressions. There was a range of concen-
trations where we had a fair amount of contradictory
evidence. For example, we had a sample at 100 parts
per million lead that was toxic, but the lead
concentration in another sample was identical and we
didn't have an effect. Using these data, we defined
an apparent benthic effect threshold and an apparent
toxicity threshold. That's the data point, at 300
parts per million lead, for benthos, and 660 parts
per million lead for toxicity, above which we always
observed an effect, a statistically significant
biological effect.
This is really a pretty simple concept, but
it allows you to say we've got a problem up in
this region. By itself, lead doesn't explain an
awful lot of data points. What we're presuming is in
the region below the apparent effect threshold for
lead, we've got stations that are contaminated by
other chemicals perhaps, and that those chemicals
might be causing the observed biological effect.
-------
351
Likewise, for stations above the apparent effect
threshold, we're not necessarily saying that it's
lead, there's simply an association.
So what this technique will let you do is
look at all of your chemicals, plotting them
similarly to this slide, and see how many of your
impacted stations can be potentially explained.
In other words, you have some chemicals that are
above their apparent effect threshold at each
of your stations where you have effects. That's
the real proof.
What I want you to keep track of in this
slide for lead is the position of the impacted
stations indicated by arrows along the bottom bar.
The stations that are shown above the apparent
effects threshold for lead will jump to below the
apparent effects threshold for 4-methyl phenol in the
next slide. The data points indicated below the
apparent effects threshold for lead will jump up to
above the apparent effect threshold for 4-methyl
phenol.
The only difference between the two slides is
the chemical plotted. Stations indicated by arrows
above the apparent effect threshold for lead had
very high lead concentrations and fairly low or
-------
352
relatively low 4-methyl phenol concentrations. The
stations indicated by arrows above the apparent
effect threshold for 4-methyl phenol had low lead
concentrations and fairly high 4-methyl phenol
concentrations.
When we do this exercise for all the different
chemicals, we come up with an apparent effect
threshold for each chemical, based on the given
data set.
The next slide presents some of these differ-
ent threshold values for low molecular weight PAH
(these are basically the two and three ring compounds)
high molecular weight PAH (these are the four, five
and six ring compounds); PCBs; and 4-methyl phenol.
Apparent effect thresholds have been determined
for many substances, I just put up a few for an
example.
Using the amphipod bioassay, oyster larvae
bioassay, Microtox bioassy, (this is a bacterial
luminescence bioassay), and benthic infaunal
depressions, with the scheme that I just showed in
the last two slides, we come up with approximately
five to six parts per million as our threshold for
low molecular weight PAH. There is an error in
the slide, the columns labelled Microtox and
_
-------
353
benthic infauna should be reversed.
Basically, around the part per million level
for most of these organic chemicals, most of the
biological effects are explained. For arsenic, lead
and mercury...to put up some examples of metals...
the apparent effect threshold is around in the part
per million range for mercury and quite a bit higher
for arsenic and lead.
Now, we can come up with these different values,
but do they really tell us anything? Can we explain
our data? Does it say anything about the biology?
An evaluation of the approach is shown in the
next slide. This is actually taking not just
Commencement Bay data, but we've recently expanded
this for seven data sets from all over Puget Sound,
approximately 200 to 250 samples. For the amphipod
bioassay, for example, there were 150 samples that
went into this evaluation. Using the different
apparent effect thresholds and the different
impacted stations that I showed you in the previous
slides, we were able to account for approximately 54
percent of the significant amphipod bioassays.
In other words, at 54 percent of the stations where
the amphipod bioassay was significant, we had a
chemical that was above its apparent effect threshold.
-------
354
When we go down to the oyster larvae bioassay,
100 percent of the impacted stations were accounted
for/ Microtox 86 percent/ and benthic infauna 82 .
percent. So, in other words/ we had fairly high
predictive power. We had chemicals that were above
their apparent effect threshold at most of our
stations where we had these kinds of biological
effects. When we looked at only stations with
severe impacts...this is, for example, in bioassays
where we had greater than 50 percent mortality...
we accounted for 92 to 100 percent of the impacted
stations. The 92 percent shown in the slide, in
essence, is one station missed by the technique.
What this suggests is that by using this tech-
nique, we have fairly high predictive power for
identifying potential problem chemicals at biologically
impacted stations. Now, whether these chemicals are
the actual problem or not we don't know. .But we do
know that they're associated with biological effects,
and it suggests that they are the ones that should be
examined further.
As in any technique there are several sources
of uncertainty. The next slide summarizes these
sources. Primarily, the one we found to be most
important is the classification of bioeffects.
-------
355
There's a statistical error involved, which in our
data set was set experiment-wise at P = .05. In other
words, we had a five percent probability of saying
that a station was impacted when it actually wasn't.
The statistical power of this analysis, which is the
percent of stations that you say are not impacted
but actually were, is another story. It's more
complicated and I won't go into that now although
it was factored into our calculations.
A second source of uncertainty is the range
between the value that sets the apparent effect
threshold, (that's the non-impacted station with
the highest concentration of the chemical) and the
next higher value, which by definition is always an
impacted station. The AET may lie anywhere along
the concentration range between these two data values.
A third source of uncertainty is the reguire-
ment for representative sampling. If you haven't
sampled your environment very well you're not going
to be able to get enough data points to help sort out
the different possible combinations of effects
and chemical compositions. If you don't have
representative sampling you're going to have a problem.
Finally, there's uncertainty associated with
chemical analysis Variability, which we tried to
-------
356
control, for example, in the organic analyses with
the isotope dilution technique. Generally, in our
most recent analyses, the first two sources of error
I discussed are the predominant sources.
The next slide gives an example of the uncer-
tainty, for example, in benthic AET. The concen-
tration range between the AET and the next station up
with biological effects is the upper confidence
range. The lower confidence range is determined
as the 95 percent confidence interval based on
potential classification errors for nonimpacted
stations. What that means, for example, is that if
we statistically misclassify the two nonimpacted
stations with the highest concentrations, the AET all
of a sudden drops to the value of the nonimpacted
station with the third highest concentration.
In our application we simply estimated the
probability of misclassifying nonimpacted samples,
cranked that through, and derived the concentration
above which the AET lay with 95 percent probability.
Together, these uncertainty estimates give us a
confidence range for the AET.
Last, then, I want to show you what you can
actually do with AET. These values have several
potential .uses. Problem chemical identification
-------
357
is one use. In other words, identification of
what chemicals lie above their apparent effect
thresholds in areas that are impacted so you
can start allocating your resources for source
evaluation. Problem area identification is another
use. A lot of times we have chemical data and no
biological data. With these chemical data we can
start making predictions on where effects may occur.
Problem areas can be defined by how far out you have
to go to get everything below their apparent effect
thresholds.
Next, AET can be used as a screening tool for
biological sampling. Sediment concentrations that
are above, for example, the lower limit of an AET,
may indicate the need for additional bioassay testing.
You might want to go out and do some other biological
tests as well.
Also an important use of AET, and for analytical
chemists I think it's very important, as well as
for biologists, is helping to prioritize cause/effect
studies. There's potentially decades and decades and
decades of laboratory work that might have to be done
to sort out all the potential effect relationships
between each chemical and environmental indicators.
With this empirical evidence we can start prioritizing
-------
358
chemicals/ to focus on at least the chemicals
with a high probability of being associated with
significant biological effects in the field.
Finally, what we're having to do, at least in
the Puget Sound region, is to derive interim
regulatory and permitting guidance for sediments.
For example, in the Superfund cleanup program
in Commencement Bay, industries are likely going to
have to get sediment levels below apparent effect
thresholds. AET also have potential application in
dredge material disposal. Materials that are below a
certain level can be disposed of in Puget Sound
and materials that are above can't.
A specific example of the use of AET in the
Commencement Bay area is shown in this last slide.
Using the apparent effect thresholds and biological
data we're able to prioritize our worst problem areas
shown in red. Sediments in these areas contained
chemicals that exceeded-apparent effect thresholds,
and also exhibited multiple biological effects.
We used the apparent effect thresholds to define the
spatial extent of these problem areas, because we
didn't have biological data everywhere. The
lowest priority problem areas contained sediments
above AET but had no biological data available
-------
359
for confirmation of the predicted impacts.
In essence, we're able to use AET to rank and
prioritize all our stations and provide focus for
the continuing investigations on the Superfund
feasibility study.
Thank you.
-------
360
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
MR. EYNON: Barry Eynon,
SRI. Just, I thought it was a very interesting study
a'nd I'll be very interested to see it. Is it out yet?
MR. BARRICK: The final
report for the Commencement Bay was issued in October
or November...October of last year. It's available
from the Washington State Department of Ecology.
MR. EYNON: Some areas that
I, as a statistician, think that you might even be
able to go further with the data analysis. Let me
just mention them and see if you thought about them.
One is, what kind of correlation structure did the
presence or absence of the compounds that you were
looking at have? Were they correlated and did you
account for that?
MR. BARRICK: I want to
make sure I understand your...what correlation struc-
ture are we talking about?
MR. EYNON: The correlation
between the presence of the analytes in different
samples. In other words, if you had high lead, did
you also have high copper?
-------
361
MR. BARRICK: Oh, okay.
Typically, when we were looking at these, when we
looked at...let's take a particular station, Station
X. Only rarely when we had biological effects there
did we find a single chemical that was above its appar-
ent effect threshold. That did sometimes happen, but
in all the areas where we had real significant effects,
I mean major, major things like almost...azoic
sediments for example, or 100 percent mortality...we
had a number of chemicals that were up there. And
that's to be expected because chemicals aren't issued
out in a single little packet. They go out with lots
of other analytes. That's why we're very careful to
say this doesn't prove cause/effect. It's simply an
association.
MR. EYNON: Right.
MR. BARRICK: What we are
able to see though, is that chemicals that are above
their apparent effect thresholds in several of these
areas coincidentally are ones that you would expect,
given the local source discharging right into that
area. We've then got biological effects, we've got
chemistry, the chemistry suggests you're a potential
source, now you go ahead and figure it out. In
deciding how to handle the identified problem
-------
362
chemicals, they may take care of whatever's going on.
It could be that the real causative agent is not
being measured here. But supposedly, if this technique
is working, it's correlating with this, and supposedly,
if you control for these you will hopefully control
for the causative agent.
MR. EYNON: Okay. Yes, let
me suggest that I hope you can think a little fur-
ther about ways to cross-adjust for the various
compounds. I think that might make a...
MR. BARRICK: Some of the
industries think we've thought too far.
MR. EYNON: The second one
is along the same lines, that the different compounds
may have synergistic effects and stuff, too, obviously,
You've probably thought about that.
MR. BARRICK: Yes. One
thing I guess I would like to point out on that is
synergism is something that was brought up. There's
no...none of the sediment quality values or sediment
criteria can really take that into account. It's
very difficult to in a laboratory. My personal
hypothesis on this is that if we had...if synergism
and other kinds of chemical/chemical interactions were
really driving this in different areas, and weren't
-------
363
being accounted for by this technique, then we would
expect to have very low success in picking out the
stations. The fact that we do, suggests that maybe
synergism is going on but it is taken into account in
the setting of these values.
MR. EYNON: Okay. And my
third point was simply to note that your...it appeared
to me that when you compute your success rate in your
classification, that's taking the same data that you
used to define your classifications. Am I correct?
MR. BARRICK: Yes.
MR. EYNON:' So, is there
some thought to maybe cross-validating by subdividing
your data or going and getting new data to check
those success rates?
MR. BARRICK: We've done
that...yes, we've tried to do that two ways. One of
them is that there's no guarantee in the way the
method is set up that you will have things fall out.
Simply by using those data, the significant effects
data, to help define these groups doesn't guarantee in
itself that you will have chemicals falling above an
apparent effect threshold.
So the fact that we're going back and checking our
success isn't really predestined. In other words, we
-------
364
aren't going to have real good success simply because
we used those indicators in defining our two categories,
But we have taken this on. There's been some
samples, nine samples, collected in the San Francisco
Bay, and using these apparent effect thresholds, it
predicted fairly well. In the majority of the cases
where we said given that chemical concentration we'd
expect an effect, there was.
Right now, the status of this is it's being
expanded into a lot of areas in Puget Sound and we
will be adding later this fall a lot of samples to...
another whole data set with synoptic data, and we are
suggesting that we'll take these data and predict
effects in the other data set and then see how it
pans out.
MR. EYNON: Great. Thank
you.
MR. BARRICK: Thank you.
-------
365
-------
366
-------
367
-------
368
-------
369
-------
370
-------
371
-------
372
-------
373
-------
St-dS
91-dS
Zl-dS
U-dS
374
Ainviaon a
AinvwaoNav
-------
375
-------
376
-------
..377,
-------
378
-------
379
-------
380
-------
381
-------
382
-------
383
-------
384
MR. TELLIARD: Our last
speaker for today is going to talk about the
proposed...are you going to talk about the proposed
method? . . . ,
MR. KIMMELL: Sort of.'
MR. TELLIARD: Oh. Almost
proposed? Leaching test.
-------
385
TESTING CONSIDERATIONS FOR THE
TOXICITY CHARACTERISTIC LEACHING PROCEDURE
(TCLP)
NINTH ANNUAL ANALYTICAL SYMPOSIUM ON THE
ANALYSIS OF POLLUTANTS IN THE ENVIRONMENT
by
TODD A. KIMMELL
ENVIRONMENTAL SCIENTIST
OFFICE OF SOLID WASTE (WH-562B)
U.S. ENVIRONMENTAL PROTECTION AGENCY
MARCH 19, 1986
-------
386
ABSTRACT
TESTING CONSIDERATIONS FOR THE
TOXICITY CHARACTERISTIC LEACHING PROCEDURE (TCLP)
As a result of EPA's efforts to expand the capability of
the Extraction Procedure (EP) leaching test to address organic
components, including volatiles, and also to address.some of
its operational problems, a new leaching test, known as the
Toxicity Characteristic Leaching Procedure (TCLP), has been
developed. This test has been proposed for use in the Land
Disposal Restrictions Rule and in the expansion of the Extraction
Procedure (EP) Toxicity Characteristic. Both actions were
mandated by the Hazardous and Solid Waste Amendments of 1984.
This paper describes the new leaching test with respect to
its development, evaluation and use. Data are presented
regarding precision and ruggedness, and several operational
aspects of the procedure are described in detail. Emphasis
is placed on a comparison of the existing EP to the TCLP, in
terms of additional operational and analytical requirements.
Finally, several aspects of the test are discussed which are
being examined for possible improvement.
-------
387
Table of Contents
I. INTRODUCTION
A. Objectives
II. DEVELOPMENT OF THE TCLP
A. Disposal Environment and Model
C. Experimental Design
D. Results
III. OPERATIONAL ASPECTS
A. pH Adjustment
B. Liquid/Solid Separation
C. Use of Extraction Devices
D. Other Improvements
IV. OTHER ASPECTS
A. Particle Size Reduction
B. Treatment of Alkaline Wastes
C. Use of a Pre-Screen Test
D. Quality Assurance Requirements
V. OVERVIEW OF THE TCLP
A. EP Comparison
B. Metals and Semi-Volatiles
C. Volatiles
V. EVALUATION OF THE TCLP
A. Precision
B. Ruggedness
C. Collaborative Study
VI. CONCLUSIONS
-------
388
TESTING CONSIDERATIONS FOR THE
TOXICITY CHARACTERISTIC LEACHING PROCEDURE (TCLP)
Introduction
In carrying out Section 3001 of the Resource Conservation
and Recovery Act (RCRA), EPA identified a number of hazardous
waste characteristics. One of these characteristics, the
Extraction Procedure Toxicity Characteristic (40 CFR 261.24),
involved a leaching test, known as the Extraction Procedure,
that is used in identifying wastes that pose a hazard due to
their potential to leach significant concentrations of toxic
* ' .
compounds. When the EP was promulgated, however, the Agency
recognized that, while it is being used to address the leaching
of several organic pesticides, it was designed mainly to
model the leaching of metals.1 In addition, EPA has also
become "aware that the EP suffers a number of operational
problems, that, among other things, adversely affect its
precision.
As a result of these shortcomings, a new leaching test,
known as the Toxicity Characteristic Leaching Procedure (TCLP),
has been developed. This test addresses the leaching of inorganics
and organics, including volatiles, solves the operational problems
of the EP, and is also more precise than the EP. On January 14,
1986, EPA proposed this test for use in the Land Disposal
Restrictions Rule (51 PR 1602), and the test will soon be proposed
in an expanded Toxicity Characteristic under RCRA. Both of these
-1-
-------
389
actions were mandated by the Hazardous and Solid Waste Amendments
of 1984.
This paper describes the developmentr evaluation and use of
the TCLP. A Background Document detailing the various aspects of
the TCLP is available in EPA's docket.2
Development of the TCLP
In 1981, through an interagency agreement with the U.S.
Department of Energy's Oak Ridge National Laboratory (ORNL), EPA
began a research program to develop the TCLP. The experiments
used to develop the TCLP were set up to conform as closely as
possible with the sanitary waste codisposal model used to develop
the Ep.3'4
Briefly, industrial solid wastes containing a variety of
organic and inorganic species (target compounds), were loaded
into large glass columns and leached for approximately 3
months with municipal waste leachate generated from adjacent
lysimeters. The concentration of these target compounds in the
leachate was then measured over time and used to develop
lysimeter leachate target concentrations. These target
concentrations were established based on both practical
considerations, and the need to represent a mid-to-long term
leaching interval. This was important, as the purpose of the
leaching test is primarily to evaluate the leaching potential
of chronically toxic contaminants.^
The same industrial solid wastes were then subjected to
several laboratory leaching test procedures and leaching*
media, and the concentrations of the target compounds was
-2-
-------
390
determined. A variety of statistical tests was then used to
compare the various media as to their ability to reproduce
the target concentrationsf and the leaching medium which best
simulated the lysimeter/column was selected for the TCLP.
This medium is a 0.1 N acetate buffer with a pH of approximately
5.0.
Operational Aspects
As indicated previously, in moving from the EP to the TCLP,
EPA hoped to solve several operational problems that have been
associated with the EP. First, the EP involves continual pH
adjustment, which is tedious and is also probably the single most
important element in the EP protocol contributing to variability.
A
Using a pre-defined leaching medium, such as the acetate buffer,
eliminates this problem.
Second, the EP involves liquid/solid separation using 0.45 urn
pressure filtration. This separation has proved difficult for
some materials, such as certain types of oily wastes, which have
a tendancy to clog the filter. This problem is serious, since
materials which do not pass through the filter are operationally
defined as solids, even if they physically appear to be a liquid.
This problem is particularly serious for oily wastes, since
oils have been known to frequently migrate to ground waters. It
is important for the liquid/solid separation to treat, as liquids,
those materials which can behave as liquids in the environment.
•*
It is also important to recognize, however, that some materials,
such as certain paint wastes, while they have liquid properties,
they will generally behave as solids in the environment (i.e.,
-3-
-------
391
will not migrate).
After investigating several options, EPA opted for the
continued use of pressure filtration, but has changed the filter
medium to a 0.6-0.8 um glass fiber filter (nominal pore size).
Use of these filters decreases filtration time,6 improves the
precision of the method, and is believed to also provide a more
adequate differentiation between those materials which behave
as liquids in the environment, and those which behave as solids.
The third problem involves the use of extraction equipment.
The need to more precisely describe what constitutes acceptable
agitation and to adequately prevent volatilization of volatile
compounds during extraction was critical.
The EP protocol provides a descriptive definition of what
constitutes acceptable agitation. Two types of extraction
equipment are identified which were determined to meet this
definition? blade/stirrer, and rotary (end-over-end) agitation.
EPA has learned that this lack of specificity in agitation
conditions is also a major source of variability,7 and has centered
on the use of rotary agitation, and further specified an
agitation rate of 30 Hh 2 rpm. Rotary agitation is recognized
as a reproducible means of agitation, and has been incorporated
into several leaching and similar test methods.8
Loss of volatile contaminants can occur during liquid/solid
separation, and during extraction. With the assistance of
laboratory equipment manufacturers, EPA has addressed this
problem through development of the Zero-Headspace Extractor
(ZHE). This device, pictured in Figure 1, is designed to
-4-
-------
392
prevent volatilization when conducting the procedure. The
operation of this device is discussed in a subsequent section
of this paper.
Due to the need to have the ZHE compatable with common
laboratory equipment, such as off-gassing ovens and laboratory
sinks, and also the need to have a device that is easily handled
by analysts, a device smaller than the 2-liter internal volume
device EPA originally had in mind was necessary. Balancing the
need to also accomodate as large a sample size as possible, a device
with a 500 ml internal volume is specified. Due to the 500 ml
internal volume, however, the ZHE can only accommodate a maximum
sample size of 25 grams for a 100 percent solids sample. For a
waste containing less than 100 percent solids, the sample size
used in the ZHE is a function of the percent solids of the waste.
Although considerably more expensive than the bottles used in
the current EP, these devices are only required for volatile
components. Bottles are used for assessing non-volatile components.
While EPA had originally intended the ZHE to be capable of addressing
all waste components, the volume limitations and other constraints
have led EPA to only allow its use when dealing with volatiles.
Other improvements designed to reduce complexity and enhance
precision have also been introduced into the TCLP. For example,
agitation is conducted over 18 hours rather than the EP's 24 hours.
As another example, in transferring samples from container to
filtration apparatus to estractor etc., the TCLP procedure calls for
determining the weight of any residual sample material left behind,
and subtracting this from the total sample size.
-5-
-------
393
Other Aspects
There are also several other aspects of the TCLP which warrant
discussion. These are in the areas of particle size reduction,
=.
treatment of alkaline wastes, the use of a pre-screen test, and
the quality assurance requirements.
Whereas the EP allowed using what is known as the Structural
Integrity Procedure (SIP) for reduction of monolithic wastes, the
TCLP requires particle size reduction to 9.5 mm for all wastes,
if necessary. The SIP amounts to pounding monolithic wastes with
hammer-like blows, and was designed to simulate the action of
heavy landfill equipment, which can act to reduce monolithic
blocks into smaller pieces. In addition to the action of heavy
landfill equipment, however, wastes are also reduced in size by
natural weathering forces, such as wet/dry and freeze/thaw cycles,
which can also act to reduce monolithic wastes.9 By not allowing
use of the SIP, the Agency assures that wastes that are solidified
into monolithic blocks will not leach even if broken down.
Although the TCLP development work involved 11 wastes and
95 target compounds which leached from these wastes, alkaline
wastes were not adequately represented in the leaching experiments.
EPA believes that an increase in the leaching of inorganic and
some organic species may be observed as the ability of alkaline
wastes to resist the acidity encountered during leaching becomes
exhausted. Data from the TCLP development program (on a moderately
»-_
alkaline waste), and from subsequent studies on waste of moderate
•
to high alkalinity, demonstrated that the leaching rate of metals
was relatively constant or increasing slightly over liquid to
-6-
-------
394
solid ratios as high as 30 to I.10 The TCLP is carried out at a
20 to 1 liquid to solid ratio, during which most components from
non-alkaline wastes experience a decrease in leaching rate. This
data demonstrated that the acetate buffer system may not adequately
account for the leaching of some metals from wastes of moderate to
high alkalinity.
To address this problem, the TCLP specifies a second, stronger
leaching medium for the extraction of wastes of moderate to high
alkalinity. To define this second medium, the basis behind the
EP's maximum amount of acetic acid, 2 milliequivalents of acid
per gram of waste, is used.1 The acetate buffer supplies only
0.7 milliequivalents of acid per gram of waste.
i
A simple test of waste alkalinity is proposed as a means of
determining the appropriate leaching fluid to use. This test
involves mixing the solid portion of the waste with water and
determining the pH. If the pH is £ 5.0, the buffer is used. If
the pH- is > 5.0, a known amount of acid is added to the slurry
(i.e., 0.7 acid milliequivalents per gram of waste), the mixture
is heated, and after cooling, the pH is again measured. If the
pH is <_ 5.0, the acetate buffer is used, and if the pH is > 5.0,
the stronger leaching medium is used.11
While the TCLP is expected to be similar to the EP in cost,
due to the TCLP's increased number of analytes, the overall cost
of analysis of the TCLP extract will be considerably more than
the EP extract. To reduce these costs, EPA is proposing to
establish an optional prescreen test in lieu of the TCLP* This
prescreen will consist of a total analysis of the waste using
-7-
-------
395
SW-846 methods,12 to determine if the waste contains sufficient
amounts of individual contaminants for the appropriate -regulatory
threshold to be exceeded, assuming all the compound leaches from
the waste. If based on such an analysis, one can be certain that
the appropriate regulatory threshold could not be exceeded, then
the TCLP will not have to be performed.
This option will be especially useful for those generators
who wish to demonstrate that their waste is devoid of certain
contaminants. For example, since fly or bottom ashes resulting
from combustion processes are unlikely to contain volatiles,
the prescreen might prove a less costly option to performing
the TCLP. •• . '
The quality assurance (QA) requirements of the EP require one
blank per sample batch, and the Method of Standard Addition (MSA)
to be run for all samples. The Agency has been hearing for some
time now that the requirements with respect to the use of blanks
is unclear, and has changed the requirement to specify that one
blank should be run for every 10 extractions, and that the leaching
media should also be subjected to a blank.
EPA has also received comment that requiring MSA, which is
very expensive, is unnecessary for all situations. The Agency
agrees with this. Accordingly, MSA is only required when the
measured concentration of a contaminant is close enough to the
threshold, that matrix interferences could yield a wrong ^decision
regarding the determination of hazard, or that severe matrix
interferences are preventing the analytical method from accurately
determining the concentration of the analyte.
-8-
-------
396
EPA is also adding QA requirements dealing with acceptable
sample holding times. While specifying holding times for the EP
metals was not critical, since the TCLP involve,? quite a few more
organic analytes, sample holding times begin to take on more
importance. The sample holding times specified are consistent
with those used in the Agency's Contract Laboratory Program.
Overview of the TCLP
The major differences between the EP and the TCLP are
summarized in Table 1. The flow, diagrams of the two methods,
presented in Figures 2 and 3, respectively, indicate that the
TCLP is a batch leaching test similar to the EP in overall
conduct. It is obvious that the TCLP relies heavily on many
procedural aspects of the EP. The main differences are in the
areas of leaching fluid, filter type, particle size reduction,
extraction vessels, agitation, extraction time, and in the QA
requirements. I have discussed these differences in some detail.
Precedurally, then, the two protocols are very much alike.
For wastes containing less than 0.5 % solids, the waste, after
filtering through the glass fiber filter, is defined as the TCLP
extract. For wastes containing greater than 0.5 % solids, the
liquid phase, if any, is separated from the solid phase by glass
fiber filter filtration, and stored for later analysis. The
particle size of the solid phase is reduced (if necessary),
weighed, and then extracted with an amount extraction fluid equal
to 20 times the weight of the solid phase. The extraction fluid
employed is a function of the alkalinity of the waste,and "a special
extractor vessel, the ZHE, is used when testing for volatiles.
-9-
-------
397
Following an 18-hour agitation period, the liquid extract is
separated from the solid phase, again by filtration through a
glass fiber filter. If compatible (e.g., precipitate or multiple
phases will not form on combination), the initial liquid phase
of the waste (if any), is added to the liquid extract, and these
liquids are analyzed together. If incompatible, the liquids
are analyzed separately, and the results are combined to yield
a volume-weighted average concentration.
Evaluation of the TCLP
The TCLP has been subjected to a single laboratory precision
evaluation for both the bottle extractor (metals and semi-volatiles),
and for the ZHE (volatiles). Two wastes containing a variety of
contaminants were used (i.e., an oily waste and an alkaline waste),
and these wastes were also spiked with various volatile compounds.
As shown in Table 2, earlier research indicated that the TCLP
was more precise than the EP.4 Although the single laboratory
precision evaluation did not include a side-by-side evaluation of
the EP, the results for the bottle extractor show the TCLP to be
of acceptable precision.13 Tables 3 and 4 present some of the
results for metals and organics, respectively. For the most part,
the percent coefficient of variation (CV) between triplicate
extractions was less than 30 percent. This includes the variability
contributed by sampling variability and analytical variability.
Although sampling variability was minimized to the extent possible,
it is reasonable to expect a sampling variability contribution of
between 2 and 5 percent. Analytical variability was in many cases
comparable to, and in some cases exceeded the total variability.
-10-
-------
398
This observation is significant as the analytical methods used
•~ ,
are well-accepted and in widespread use.
The results of the precision evaluation for the ZHE,14 are
not as clearly interpreted. There are several reasons for this.
Firstr experience with the ZHE, especially by laboratory technicians
who were familiar with the work, was limited. Second, the evaluation
was conducted using several draft protocols, which differed in their
treatment of the ZHE. Substantial changes were made between the
draft protocols due to experience gained with the device. Third,
inadvertant errors were apparently made in following the protocol.
For example, whereas the protocol placed a maximum of 25 grams on
the amount of solid material the ZHE could accommodate, considerably
* .
more solid material was extracted in the device, providing for a
variable liquid to solid ratio. Fourth, due to extenuating
circumstances, two laboratories conducted the analytical work,
rather than the intended single laboratory. It is apparent that
higher concentrations were obtained by one of the laboratories.
As indicated above, these factors make the precision data
for the volatiles difficult to interpret. The percent CV's for
the alkaline waste were mostly less than 60 or 70 percent, which is
fair given the nature of volatiles. It is unrealistic to expect
any leaching procedure to provide the same variability for volatiles.
as it does for metals and semi-volatile organics. The precision
data for the oily waste- indicated more variability. Some of this
9\,
can be attributed to severe laboratory contamination problems, as
well as to the oily character of the waste, which seems t'o have
dominated the extraction. Due to the inconclusive nature of
-11-
-------
339
these results, EPA is in the process of another, more extensive
precision evaluation for the volatiles. ~
EPA has also conducted a ruggedness evaluation for the
TCLP. As with the precision evaluation, ruggedness was evaluated
for both the bottle extractors and the ZHE, and the same wastes
were used. While the evaluation of the volatiles is currently
ongoing, the evaluation of the bottle extractors has been completed.1
Table 5 presents the parameters which were evaluated for ruggedness
using both types of equipment.
The ruggedness evaluation for the bottle extractors demonstrated
that for the most part, the TCLP is fairly rugged. This is
especially true for the semivolatile organics, which for the most
part, were unaffected by the parameters investigated. For the
metals, the results suggest that two parameters may be critical.
As expected, the acidity of the extraction fluid directly affects
the extraction of metals. Recognizing this, the TCLP emphasizes
accuracy in the preparation of the extraction fluids, by specifying
the exact recipes for these media, and also indicating that the pH
of these media should be accurate to within +_ 0.05 pH units.
Bottle type (i.e., borosilicate vs flint glass) also appears
to affect extract concentrations for the metals. It appears that
using flint glass can result in significantly higher metal
concentrations. While acid washing or an expanded use of blanks
may help to resolve this problem, specifying borosilicate over
other types of glass would resolve the problem entirely. Due
to the substantially higher cost of borosilicate versus other types
of glass (i.e., from 3 to 5 times higher), this is one of the
-12-
-------
400
areas where further work may be warranted.
In addition to single laboratory ruggedness and precision
evaluations, two independent collaborative evaluations are being
conducted, one by EPA and the other by the Electric Power Research
Institute (EPRI). EPA's evaluation, in which a number of business
associations and individual companies are participating, involves
over 20 laboratories, five different wastes, and both types of
extraction e'quipment. This study is currently ongoing.
The EPRI study, which is nearing completion, is very similar
to an evaluation EPRI conducted on the EP.16 This study is limited
to the determination of inorganic species and deals with the bottle
extractors only. As with the 1979 study, EPRI is investigating
the contribution of both variability in sampling, and variability
introduced through the analytical methods. Unlike EPA's evaluation,
EPRI is also conducting EP extractions, and will be comparing the
variability of the two methods.
Conclusions
The development of test methods, whether they be physical
test methods like the TCLP, or analytical methods, is an evolving
process. As advances are made in technology and as additional data
is gathered and evaluated, test methods become more accurate,
precise, sensitive, rugged, and less costly. The evolution
from the EP to the TCLP is a prime example of this.
I have presented a -discussion of the development, evaluation
and use of the TCLP. Work is continuing to help resolve remaining
problems, and I am sure that future advancements will be made
that will lead to further improvement of the procedure.
-13-
-------
401
Acknowledgements ' "
I would like to express my appreciation to C. Francis*-
M. Maskarinec, N. Rothman, T. Varouxis and the rest of those
many people who had a part in the development and evaluation
of the TCLP. Special appreciation is in order for L. Williams
and D. Friedman. Through the dedication and expertise of all
these people, the TCLP has become a reality.
-14-
-------
402
Tables and Figures
Tables
Table 1: Comparison of EP and TCLP
Table 2: ORNL Precision Data Comparing EP to TCLP
Table 3: Precision Data For Metals
Table 4: Precision data For Semi-volatiles
Table 5: Parameters Investigated During TCLP Ruggedness Evaluation
Figures
Figure 1: ZHE Device
Figure 2: EP Flow Diagram
Figure 3: TCLP Flow Diagram
-15-
-------
403
Table ±
Conparison of the Extraction Procedure (EP) and the
Tbxicitv Characteristic Leaching Procedure (TCLP)*
Item
1) leaching Media
2) Liquid/solid
separation
3 f Monolithic mat-
erial/particle
size reduction
4) Extraction
Vessels
5) Agitation
6) Extraction time
7) Quality Control
requirements
EP
0.5 N Acetic acid added
to distilled deionized
water to a pH of 5 with
400 ml maximum addition.
Continual pH adjustment.
0.45 urn Filtration to
75 psi in 10 psi
increments.
Unspecified filter, type.
Use of Structural
Integrity Procedure
or grinding and milling.
Unspecified design.
• Blade/stirrer vessel
acceptable.
- Prose definition of
acceptable agitation.
- 24 hours.
- Standard additions
required.
- One blank per sample
batch.
TCLP
0.1 N pH 2.9 acetic
acid solution for moderate
to high alkaline wastes and
0.1 N pH 4.9 acetate buffer
for other wastes.
0.6-0.8 urn Glass fiber filter
filtration to 50 psi.
- Grinding or milling only.
- Structural Integrity Procedure
not used.
- Zero-headspace vessel
reguired for volatiles.
- Bottles used for non-volatiles.
-Blade stirrer vessel not used.
- Rotary agitation only in
an end-over-end fashion at
30 ±2 rpm.
- 18 hours.
- Standard additions required
in some cases.
- One blank per 10 extractions
and every new batch of extract.
- Analysis specific to analyte.
* All other attributes between the two tests are generally the same, although
there are seme minor differences. Note also that while the EP only addresses
those species for which National Interim Primary Drinking Water Standards
(NIPDHS) exist, the TCLP can be applied to other toxicants.
-------
404
TABLE V-. RANKING OF LABORATORY METHODS BY HEAN COEFFICIENT OF
VARIATION (PHASES I and II COMBINED, 95 TARGETS)*
Chemical
class
Organic and
Inorganic
Inorganic
Organic
Ranklna of media
Acetate
27.9
12.6
35.8
bv mean coefficient of
Carbonic add
39.6
21.6
50.8
variation 1%)
Extraction
procedure
39.1
20.4
47.8
No significant differences were observed (P < 0.05) within any
single chemical class.
-------
TABLE 3:
METALS IN API SEPARATOR SLUDGE, BUFFER A
METALS ANALYSIS
UNITS {UB/KL)
405
1
METAL
— »L
SB
AS
8ft
BE
B
CD
CA
CR
CO
cu
FE
1C
PB
KG
KN
HO
NI
SE
AB
NA
SR
TL
V
ZN
ZR
1
i
APIE BUFFER A
i : i
: SMPL J SHPL s
t SSAI : SSA2 :
: i !
: :
1 17.3 i
! 1.22 !
: ND :
s 1.01 ;
s ND :
S' 9.22 !
ND :
: - IBB !
S 18.3 i
: 0.103 i
: ND :
I 276 1
J 3.78 !
I NO S
S 39.2 !
: 6.91 :
i 0.076 i
S 90.3 1
! ND !
i ND i
i 125 :
: 1.49 :
: 0.620 :
i ND i
! 0.060 !
'. 360 !
i ND :
^ ! _
S
ts.9 :
0.117 I
NO :
1.02 :
ND !
10.0 i
. ND :
175 i
11.2 :
0.076 !
ND ;
233 1
4.01 I
ND I
38.7 !
6.95 !
0.087 I
76.3 !
ND ;
ND ;
136 :
1.54 S
ND i
ND 1
0.024 !
350 !
ND i
SMFL ! !
SSA3 '. MEAN I
: i
•
15.9 I
0.143 i
ND :
1.02 :
ND :
9.37 I
ND !
179 :
13.7 :
0.059 i
ND :
268 !
3.64 I
ND !
38.3 i
6.86 1
0.07 1
73.4 !
ND :
ND !
122 i
1.59 1
0.660 1
0.013 i
0.042 !
332 t
ND !
... J...
5
16.4 i
0.493 :
NA !
1.02 :
NA !
9.53 !
NA !
iBi :
14.4 ;
0.079 !
NA :
259 I
3.81 I
NA :
38.7 :
4.11 :
•.•78 :
».• :
NA :
NA :
128 :
1.54 :
0.64 1
NA I
0.042 i
347 !
NA :
t
SD t
i
•.808 :
f.429 i
M i
0.006 :
NA :
0.414 !
NA J
6.66 '
3.60 t
0.022 :
NA :
22.9 !
0.187 !
NA :
•.45i :
•,«5 :
•.009 i!
f.<04 !
MA I
NA 3
7.37 :
o.oso :
NA :
NA :
O.OIB :
14.2 ;
NA :
i
CV I
•
•
«
i
•.049
1.28
NA
0.006
NA
0.043 !
NA :
•.037 :
i.250 8
•.280 :
•A:
•.ess:
•.049 :
MA:
C.<012 S
Cu'Oo? :
•o.nm :
tO.115 3
INA :
INA :
C.05B 3
0.032 !
DA 6
NA :
0.429 !
0,041 :
NA t
1
1 i
'.METHOD !
1CV 1 BLANK A!
• i
i i
!
4.94 !
12B 1
NA !
0.568 !
NA I
4.34 S
MA i
3.69 •
25.0 t
28.0 S
•A 1
1.83 1
4.90 S
NA S
1.16 i
0.653 !
U.I J
11.3 1
NA !
NA i
5.77 I
3.25 !
HA 1
HA i
42.9 i
. 4.09 1
HA i
J
1
0.227 !
ND :
ND :
0.313 !
ND :
0.347 i
ND ;
1.40 :
no :
ND :
•.018 S
•.191 t
•.soo :
HO t
o.ue s
MD 1
ND :
u> :
ND :
ND :
3.25 !
ND :
•.530 !
ND ;
, ND :
0.560 !
ND :
' 1
1
1
1
1
1
i !
LAB ! DET. !
BLANK i LIMITS!
; i
i
O.OBO ;
ND :
NO :
ND !
ND :
ND !
0.030 :
0.050 i
ND i
ND :
ND i
o.no :
0.510 :
M) S
0.016 !
. WB 4
liD i
tffi :
KD :
MO i
0.168 !
liD I
0.230 :
NO :
0.036 :
0.167 i
ND :
. {
s
•
0.070 :
o.oso ;
0.004 :
0.040 :
o.oio :
o.oio ;
s
0.040 :
o.oto :
o.oos :
•
O.OBO :
o.ooi ;
o.oso ;
0.002 :
o.oso :
0.040 :
0.070 t
O.OiO 1
0.00? !
0.010 i
o.oso :
I
1
ND (NOT DETECTED)
NA (NOT APPLICABLE)
BUFFER A * PH 2.9
BUFFER S * PH 4.9
-------
TABLE
406
SEMI-VOLATILE ORGANIC COMPOUNDS IN STILL LIKE BOTTOM, BUFFER B
UNITS (UB/L)
ACID/EASE NEUTRAL
COMPOUNDS
PHENOL
ANILINE
52-HETHYLPHENDL
•M-HETHYLPHENOL
la^BIHETHYLPHENQL
{NAPHTHALENE
12-METHYLNAPHTHALENE
SDIBENZOFURAN .
JACENAPHIHYLENE
,'FLUORENE •
5PHENANTHRENE
{ANTHRACENE
5FLUDRANTHRENE
SPYRENE
JBISt2-ETHYLHE)WJPHTHALATE
IBEN2QlA)ANTHJy£ENE
JCHRYSENE * -
;DI-N-BUTYLPHTHALATE
j
1
SHPL i
SLBB1 I
j
•
17500 5
318 5
1820 5
7040 5
292 !
3480 i
254 !
170 i
635 5
141 !
234 i
32.4 J
24.9 5
15.2*5
ND :
ND :
ND 5
16.7*5
!
SHPL
SLBB2
21800
154
2340
9530
375
4300
340
213
804
171
266
39.8
27.3
14.5*
ND
ND
ND
14.5*
J
SHPL i
SLBB3 5
5
i
18600 5
230 5
1830 !
7250 5
296 5
3970 I
275 5
179 5
670 5
140 5
222 !
27.5 5
23.8 !
14.3*5
ND 5
ND t
ND !
23.8*!
:
STILL LIME
J
5
HEAN 1
i
•
,19300 5
234 1
2000 S
7940 i
321 i
3920 t
290 5
187 5
703 i
151 5
241 i
33.2 I
25.3 5
14.7*1
NA i
NA 5
NA 5
18.3*5
J
BOTTOM BUFFER B
S
i
SD !
1
. {
2230 5
82.1 !
297 5
1380 5
46.8 5
413 5
44.8 1
22.7 1
89.2 i
17.6 i
22.7 5
6.19 5
1.79 !
0.473 !
NA !
NA 5
NA :
4.86 !
J
I
CV
0.116
0.351
0.149
0.174
0.146
0.105
0.155
0.121
.0.127
0.117
0.095
0.186
0.071
0.032
NA
NA
NA
0.265
ID
11
35
14
17
14
10
15
12
12
11
9.
18
7.<
3.4
1
1
1
26
1
ND (NOT DETECTED)
HA (NOT APPLICABLE)
BUFFER A * FH 2.9
BUFFER B * PH 4.9
* INSTRUMENT DETECTION LIMITS ARE DETERMINED BY HL'LTIPLYINB BY THREE THE
STAE'ARD DEVIATION OF THREE SUCCESSIVE INJECTIONS OF A LOW-LEVEL STANDARD
SOLUTION. DETECTION LIMITS IN THE TABLE HAVE BEEN CORRECTED FOR THE
APPROPRIATE DILUTIGH FACTO* (10:1). QUANTlTATION LIMITS IN MANY CASES
CAN BE CONSIDERABLY HI&HER OR LOWER THAN THESE INSTRUMENT DETECTION LIMITS,
DEFENDING GN THE NATURE OF THE SAMPLE.
ICV
11.6
35.1
14.9
17.4
14.6
10.5
15.5
12.1
12.7
11.7
9.45
18.6
7.06
3.22
NA
NA
NA
26.5
DET.
LIKITS
9
2
9
15
3
13
5
17
5
26
12
4
9
17
14
9
2
30
-------
407
Table S
Parameters Investigated During TCLP Ruggedness Evaluation
Parameter
1) Liquid/Solid ratio:
2) Extraction time:
3) Headspace: ZHE:
Bottles:
4) Medium #1 acidity:
(milliequivalents acid)
5) Medium #2 acidity:
(milliequivalents acid)
6) Aliquots:
(taking of aliquots
directly form ZHE
for analysis)
7) Extractor vessel:
8) Acid wash filters:
9) Filter Type:
10) Pressurization of
ZHE during agitation
(psi)
11) ZHE extract collection
devices
TCLP
Specification
20
18
zero
variable
70
200
Allowed for ZHE
in sane cases
(See TCLP protocol
in Section VCI)
(See TCLP
protocol in
Section VEI)
Required
for metals
0.6-0.8 urn
glass fiber
5-10
Tedlar bag
or Syringe
ZHE
Device
19 - 21
0-5%
60 - 80
Common Equipment
(Bottles)
19-21
16 - 20
20 - 60%
190 - 210
Yes - No
Associated ZHE
Millipore ZHE
Borosilicate
Flint glass
Yes - No
Glass fiber -
Polycarbonate
0 - 20
Tedlar bag -
Syringe
-------
Extract Outlet
408
•Filter-
Waste/Extract ion Fluid
Piston
Top
Plate
Body
VI TON
o-rings
Bottom
Plate
Pressure Inlet
Figure ± : Zero-headspaoe Extraction Vessel
-------
"409
Wet Waste Sample Representative Wet Wast
Contains < 0.5% ^ Waste Sample ^ Contain*
Nonfilu
Solids
(rafale * > 1UU brams r Nonfiltw
1 ^
^. Dry Waste Sample ^
Liquid Solid ^_ ...
_ . •"•••» Solid
Separation r^"«
Lie
Discard
^
uid Partic
> 9.5mm < 9.1
4
Sample Size
Reduction
i
e Sample
> 0.5%
able
L
I *0,id „ Liquid Solid
^ - Separation
r
e Size Lie
>mm Monolithic
4
Structural
Integrity
Procedure
r *
^ Store
4
Solid 4-- Liquid Solid Separation
I ,
+ J
Discard
Liq
EPE
J
w •
\r
uid
,
u
Ktract
L
Methods
uid
r
it 4°C
-2
Figure 2: Extraction Procedure Flowchart.
-------
.410
FIGURE 3 : TCLP Flowchart
WET WASTE SAMPLE
CONTAINS < 0.5 %
NCN-FIUTERABIE
SOLIDS
REPRESENTATIVE WASTE
SAMPLE
ERY WASTE
WET WASTE SAMPLE
CONTAINS > 0.5 %
NON-FILTERABLE
SOLICS
I
LIQUID/SOLID
SEPARATION
0.6-0.8 urn
GIASS FIBRE
FIUTERS
DISCARD
SOLID
SOLID
SOLID
LIQUID/SOLID
. SEPARATION
0.6-0.8 urn
GIASS FIBRE
FILTERS
LIQUID
~~~i
STORE AT
4°C
REDUCE PARTICLE SIZE IF >9.5 mm
OR SURFACE AREA <3.1 crni2
TCLP EXTRACTICN*
OF SOLID
0-HEAD3PACE EXTRACTOR
REQUIRED FOR VOIATIIES
1
LIQUID/SOLID
SEPARATION
0.6-0.8 urn GIASS
FIBRE FILTERS
DISCARD
SOLID
LIQUID
TCLP EXTRACT
TCLP EXTRACT
ANALYTICAL
METHODS
TCLP EXTRACT
* The extraction fluid alloyed is a function of the alfcalinity of the
phase of the waste.
-------
411
References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(ID
(12)
U.S. EPA. Background Document, Section 261.24, Character-
istic of Extraction Procedure Toxicity. National Technical
Information Service (NTIS) PB 81 185-027. May, 1980.
U.S. EPA. Background Document, Section 261.24, Toxicity
Characteristic Leaching Procedure. March 10, 1986.
Francis, C.W. et al. Mobility of Toxic Compounds Prom
Hazardous Wastes. National Technical Information Service
(NTIS) PB 85 117-034. August, 1984.
Francis, C.W. and M. Maskarinec. Field and Laboratory Studies
in Support of a Hazardous Waste Extraction Test. Oak Ridge
National Laboratory Report No. 6247. February, 1986.
Kimmell, T.A. and D. Friedman. Model Assumptions and Rationale
Behind the Development of EP-III. Fourth ASTM Hazardous and
Industrial Solid Waste Testing Symposium. Proceedings. J.K.
Petros, W.J. Lacy, and R.A. Conway, Eds. ASTM Publication
Code Number (PCN) 04-886000-16. 1985.
Energy Resources Co. Filtration of Various Wastes Using
Various Filter Media. U.S. EPA Contract 68-01-7075. April,
1985.
Brown, O.K. et al. Mobility of Organic Compounds From
Hazardous Wastes. National Technical Information Service
(NTIS) PB 83 163-956
American Society For Testing and Materials (ASTM). Committee
D-34 Draft Method. Method for 24-Hour Batch Type Distribution
Ratio (DR) For Contaminant Sorption on Soils and Sediments.
ASTM D34.02-022RO. Philadelphia, PA. 1985.
Hannack, P. Personal Communication From P. Hannack, Canada
Ministry of the Environment, Alberta Research Centre, to T.A.
Kimmell, U.S. EPA Re: TCLP. October 28, 1985.
Francis, C.W. and M. Maskarinec. Leaching of Metals From
Alkaline Wastes by Municipal Waste Leachate. Oak Ridge
National Laboratory Report. January, 1986.
Energy Resources Co. Extraction Fluid Study and Development
of an Alkalinity Test For The Toxicity Characteristic Leaching
Procedure. U.S. EPA Contract 68-01-7075. February, 1986.
U.S. EPA. Test Methods For Evaluating Solid Waste - Physical/
Chemical Methods. Second ed. Government Printing Office
(GPO) 055-002-81001-2. EPA SW-846. Washington D.C. 1982.
-16-
-------
412
(13) S-Cubed. Precision Evaluation of the TCLP Protocol For
Non-Volatile Components. U.S. EPA Contract 68-03-1958.
January, 1986.
(14) Francis, C.W. and M. Maskarinec. Precision Analysis For
the Zero-Head Extractor. Oak Ridge National Laboratory.
January, 1986.
(15) Energy Resources Co. Evaluation of Bottle TCLP Draft
Protocol. U.S. EPA Contract 68-01-7075. February, 1986.
(16) Electric Power Research Institute (EPRI). Proposed RCRA
Extraction Procedure: Reproducibility and Sensitivity.
Palo Alto, CA. November, 1979.
-17-
-------
413
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
AUDIENCE PARTICIPANT: You
mentioned you got lower results for the borosilicate
glass. That was for metallic ions, I presume?
MR. KIMMELL: That's correct.
AUDIENCE PARTICIPANT: So
isn't that probably due to the ion exchange capacity
of the glass?
MR. KIMMELL: I don't think
we can conclude yet what it's due to.
MR. BIRRI: John Birri,
EPA. What's the cost of one of these units?
MR. KIMMELL: The price
ranges, depending on which company you buy it from,
I think from $1,200 or $1,300 to $1,500 or $1,600.
MR. BIRRI: I see.
MR. KIMMELL: The primary
differences in the cost of running the method will
be in the additional analytes that we're going to be
asking for.
MR. IMBUR: Bill Imbur, Law
Environmental Services. I've just gone through this
process. The Rotary extractor is about $2,100 and the
-------
414
Zero Headspace extractor is $1,500 on the current
market.
I have a question for you, Todd. How soon do you
expect this to become official, the methodology?
MR. KIMMELL: The method is
being used in two actions under RCRA. First of all,
the method has been proposed for use in the Land
Disposal Restrictions Rule, and in addition, we intend
to propose use of the method in an expanded toxicity
characteristic under RCRA within a month. It could
be later than that, but it should be within a month.
I've been wrong on my predictions before.
DR. LICHTENSTEIN: Harris
Lichtenstein, Spectrix, Houston. I'm trying to
understand the concept of the volatile organic addition
to the EP toxicity test. Is that how it should be
understood? You have a new method because we want to
understand more about the VGA's?
MR. KIMMELL: We want to be
able to...first of all, the characteristic of EP
toxicity is used in seeing if a waste poses as a
hazard due to its potential to leach primarily
inorganic compounds. We've developed the zero
headspace extractor specifically to address
the leaching of volatile organic compounds.
-------
415
DR. LIECHTENSTEIN: Do you
have any data that would allow one to design a mini-
extractor in a VGA bottle...let me just think out
loud...to where you basically...
MR. KIMMELL: We considered
that option. Specifically, we considered taking
large VGA vials and putting the waste in there with
an amount of extraction fluid. I think what it
amounted to was two grams of the waste.
I think most of you are aware of the problems
that everybody has with respect to representative
sampling. I think everybody agrees that 100 gram
sample size that the EP and the TCLP involved is...it
still doesn't get you to what you might call a repre-
sentative sample. Several samples are generally
evaluated. But going down to a two gram sample
might pose additional representative sampling
problems. That's why we opted to develop the zero
headspace extractor.
MR. TROIANO: Jeff Troiano,
Ford Motor Company. A strict reading of the proposed
procedure would seem to indicate that you are not to
use the ZHE extract for metals analysis.
MR. KIMMELL: Yes, and
there's a good reason for that. Due to the 25 gram
-------
416
maximum sample size that the device can accommodate,
what you've got is an extract that's less than 500
milliliters. EPA's analytical methods in SW-846
require a one liter sample size for just the analysis
of semivolatiles.
MR. TROIANO: What if you
know from a total analysis you have no semivolatiles
of any consequence in your sample?
MR. KIMMELL: I'm sorry,
can you repeat that?
MR. TROIANO: You spoke
earlier of a screening procedure whereby you analyze
a sample on a total basis to determine whether or not
you even have to consider certain compounds. Once
you've eliminated semivolatiles as being of concern,
you no longer have to worry about a volume constraint
for those parameters.
MR. KIMMELL: That's true.
Again, the method is written, as was the EP, to apply
to all wastes, so again, if you've got a better
idea...we've exhausted a lot of ideas though the
development of this method, but if you've got a
better idea on how to deal with it, we'd be glad to
hear it.
MR. TROIANO: Let me clarify
-------
417
what I've said. You've got your sample. You perform
a total analysis for metals, semivolatiles, volatiles,
and...
MR. KIMMELL: I think I
understand what you're saying. You're saying you
don't want to take a look at semivolatiles...
MR. TROIANO: Anymore.
MR. KIMMELL: You don't
want to take a look at them anymore, you just want
to evaluate metals. Why can't you use the ZHE?
MR. TROIANO: Metals and
volatiles.
MR. KIMMELL: Yes, and
again I invite you to comment on that in comments...
MR. TROIANO: We did.
MR. KIMMELL: ...on the
proposed rule. Then I'll be reading them. But
there's also another reason why it's not a good
idea to use different devices. We have found that
when you use different devices you get different
results. To reduce the variability of the method, it
makes sense to have one device, one set of operating
conditions. So that's another reason for going to
just the use of one device over another.
MR. TROIANO: Would it be
-------
418
that much more difficult to validate the ZHE extracts
versus the non-ZHE extract to show that the results
are basically equivalent, if they are?
MR. KIMMELL: In my opinion,
I think that you would find that the results are
different, for various reasons. One thing, the TCLP,
as applied to metals and semivolatiles, uses conven-
tional gas pressure filtration to separate the liquid
from the solid phase of the waste. The volatiles
procedure with the ZHE uses piston pressure. Those
two methods of liquid/solid separation can produce
different amounts of liquid; in other words, a different
result for a percent solids for the same waste.
So, I think there's a lot of give and take that
went along with the development of this method. For
example, we originally wanted to use the ZHE for
everything, that's why we developed the first devices
with a two liter sample size. But then we ran into
the problem that the device was so large that many
analysts, unless they're 6'6" and 240 pounds couldn't
deal with it. You couldn'.t fit it in laboratory
sinks and you couldn't fit it in an off-gasing oven,
so we had to go down to a smaller size.
MR. TROIANO: I just see a
tremendous amount of duplicative efforts, because I
-------
419
can see a lot of people might have wastes and...
MR. KIMMELL: Yes. In fact,
as part of our collaborative study...again, as I
said, we had a lot of input from various labs who
gave their time and their expertise to us, and one of
their comments from the very beginning was, you've
tried to do too much with the same procedure. By
developing the zero headspace extractor to apply to
all components, you're not taking a look at what it's
really good for, and that's volatiles.
So, basically, their comment that EPA is trying
to put too much into one device led us to separate
the volatiles from the metals and semivolatiles in
this procedure.
MR. TROIANO: Thank you.
MR. TELLIARD: We're running,
as usual, late, and I'm going to have to try to bring
this to closure because we only have an hour between
getting everybody out of here and getting back for
the murder mystery dofungus.
Todd, thanks so very much for your efforts.
Nine o'clock tomorrow morning. Nine.
(WHEREUPON, the proceedings were continued to March 20,
1986.)
-------
420
MR. TELLIARD: Could I
have this morning's speakers please come up and take
a seat and spread the fire, so to speak?
A number of folks have asked about a list of
analytes as it relates to the Groundwater Monitoring
Strategy. A copy of the new Appendix VIII guidance
can be had...had is a good term...through the RCRA
Hotline, which has a number which is 800-424-9346.
They will deliver it in a small brown paper bag.
800-424-9346. There's also a notice of availability
in the February 14th Federal Register, but as we pointed
out earlier today, or yesterday, you really can't trust
anything printed in the Federal Register.
There's one small change on the program. We're
talking about...instead of most of those things we
have listed, we're going to talk about animal breeding
and habits. Our last speaker from Duluth didn't make
it. I don't know if he even knows where Norfolk is.
He's up there counting snowflakes. Ray Maddalone
will be giving a talk on some metals analysis that
EPRI has been doing, which we heard a little bit
about last year.
Next year, folks, is the Big Ten. God willing
that I still be here...I would like someone to jot
some notes down. Any ideas you have for speakers or
-------
421
any ideas you have for what we can do for our Tenth
Anniversary...I have a few...I would certainly
appreciate. I would really like to have a nice show
next year...not that this isn't a nice show and the
last one wasn't a nice show...but I want to blow out
the stops. So any ideas on speakers, papers or
volunteers for papers, I certainly would be very
receptive, and remember, it's only money, particularly
the folks in the Contract Lab Program, it's only
money.
Our first speaker this morning is a man who
doesn't do metals. And there's a reason. Bob Beimer
is with S-Cubed. He couldn't hold his job with TRW
so he's now with S-Cubed. He is now going to talk a
little bit about one of our favorite subjects, isotope
dilution, as it relates to pesticides. Now, he does
do pesticides even though he doesn't do metals.
-------
422
ROBERT BEIMER
S-CUBED
DETERMINATION OF PRIORITY POLLUTANT PESTICIDES
BY ISOTOPE DILUTION GC/MS
DR. BEIMER: I'm really
happy to see such a nice turnout here this morning
after what most of you probably did to yourselves
last night. The guy I've got turning my slides really
hurt himself last night, so if he stumbles over them
a little bit we'll take that into consideration.
I left some reports in the back. I notice
they've all been picked up. They were supposedly on
sale. So those of you who picked them up, just leave
me $5 each and we'll take care of it. I will sign
them if you'd like.
A lot of what I'm going to present here today
is...
MR. TELLIARD: Benign.
DR. BEIMER: That too, but
there's a lot of data, and the report really goes
into more detail on listing a lot of this stuff. I
don't like statistics and there's a lot of that in
the report as well. So, if any of you would like a
-------
423
copy of the report, if you'd leave me one of your
business cards during the break I'll get it for you.
It is all right if I do that?
MR. TELLIARD: Oh, yes.
DR. BEIMER: Okay. To give
you a little historical background of what we're
doing, the isotope dilution method was developed, oh...
Bruce, how long ago? Five years? Nine years? Six
years ago. One of the movers and shakers of that
development was Bruce Colby.
What we're doing here is really just an extension
of that method, which has come to be known as Method
1624 and 1625. Specifically, the 1625 portion of
that is the extractable part, the extractable priority
pollutant determination by isotope dilution. We've
pretty well copied that for this determination of
pesticides.
Going into this I think we all knew that the
procedure was going to work. That wasn't really a
question. It was how well and what kind of interfer-
ences might one expect from chlorinated priority
pollutant pesticides which, generally speaking, have
somewhat messy mass spectra.
The procedures that we followed for this method
validation were provided by Dale Rushneck and the
-------
424
money to do this work was provided by Bill Telliard.
Hence, that's why I'm here today talking about this.
They're not going to stand and clap.
I would like to summarize just a little bit, if
I could, what we observed during the method, and then
we'll get into some of the slides and I'll show you
some of the numbers.
Certainly, the use of a labeled pesticide in
calculating the concentration of an unlabeled pesticide
provides for a significant reduction in the relative
standard deviation of the measurement itself. We
know that isotope dilution corrects for recovery, and
certainly we observed that in the data.
Interferences, although there are some, are
rather minimal. I'll get into a little more detail
on that. The detection limits that we were able to
achieve by the method approach 200 to 400 parts per
trillion in relatively clean water samples. There
are some exceptions to that, which I will also talk
about.
The method itself involves the extraction of two
liter water sample with the pH held in the range of
six to eight. The sample is extracted with methylene
chloride in a continuous liquid/liquid extractor
concentrated to one milliliter by Kd. Pass that
_
-------
425
extract then through an anhydrous sodium sulfate
column to dry it, reduce the volume to less than 200
microliters, add internal standard, bring the volume
back up to 200 microliters and analyze the sample by
GC/MS.
If you could put the first slide up, Bruce.
SLIDE 1
The analytical conditions for the determinations
that we're making. The GC, we're using a DBS capillary
column, helium carrier gas, of course...mass spectro-
meters like helium...30 centimeters a second linear
velocity, using a splitless injector and a silanized
quartz liner net injector.
For those of you who have done a lot of pesticide
determination, you know the importance of clean
injectors and properly cared for injectors, because a
lot of these pesticides get lost right there at the
front end, and you don't analyze them at the tail end
if you can't get them through the injector.
This was operated in a splitless mode. Column
temperature was 100 to 280 at eight degrees a minute
with a 10 minute hold at the column maximum. We
found that this temperature program provided for ade-
quate separation of the various pesticide components
and it kept the analysis time to a reasonable minimum.
-------
426
The mass spectrometer system we used was a
Finnigan 1020; typical scan range, one second per scan,
and the instrument was tuned to the DFTPP specifica-
tions of Method 1625B.
We used difluorobiphenyl as the internal standard.
We initially were going to use D-10-phenanthrene, like
everyone else, but there was a lot of interference
problems with D-10-phenanthrene with the various
pesticide materials that we were analyzing for, and
the 2,2-prime difluorobiphenyl eluted in a relatively
clean portion of the spectrum and showed no inter-
ference. The internal standard, of course, was used
for the calculation of recoveries of the labeled
compounds. The unlabeled compounds were calculated
relative to their labeled analogs.
What we established during the course of this
method development was the calibration range of the
instrument, calibration linearity, instrument detection
limit, analytical range of the instrument, analytical
detection limit, method detection limit, calibration
reproducibility, and the precision and recovery of
the method. We also have a limited amount of data on
interference problems.
The first step in doing an isotope dilution method
evaluation is to determine the masses that one is going
-------
427
to use for quantitation. We chose a set of primary
masses for quantitation which exhibited the minimum
amount of interference between the labeled and
unlabeled analogs, which generally coelute, and with
the other pesticide materials that would be in the
standard mixtures.
SLIDE
In addition to giving those various masses on
this slide, it also lists the priority pollutant
pesticides and the labeled analog which were used in
this determination. Not all of the labeled pesticides,
obviously, were available. Most of these were
deuterated compounds, although there were some carbon
13 labeled compounds as well. We only did the quanti-
tative determination on these primary masses.
I'm going to step back on my soapbox here a
little bit and make some suggestions. When one is
dealing with a limited number of analytes like we're
dealing with here, it would probably not be a bad
idea to choose a secondary set of quantitation masses,
repeat the quantitation, and determine if there's any
significant difference between quantitation at two
different mass numbers in a given GC/MS run. As far
as I could tell, this would probably only double the
cost of the analysis, but I think it's a wonderful
-------
428
idea.
Basically, we've got some pretty sophisticated
computer systems out there tied to these mass
spectrometers and when we're dealing with a limited
set of analytes it would make sense to make this
secondary measurement to determine if there was
interference problems occurring at your primary
quantitation mass.
Computers could make stupid decisions on these
as well as a human can. Determine if the difference
between the two numbers is significant and if so,
perhaps a tertiary mass or even a fourth mass could
be used in that quantitation, driving the price up a
factor of four. Then we could proceed to let the
computer make the decision as to whether or not there
was an interference problem, choose the two masses
which were interfered with the least...in other words,
they gave both the same number...and then that number
presented.
We've gone to a lot of trouble in determining
the accuracy of the fit of the library and the quality
of which that library is fitted to the analyte spectrum
in a run, but we haven't gone to a lot of trouble in
determining whether or not the quantitation mass is
interfered with for quantitation purposes.
-------
429
SLIDE
In these various samples, we also spiked chloro-
dane toxophene and PCB 1254. The purpose of these
initially was to try to do the analysis of these
materials also by isotope dilution. As it turned
out, it was more of an exercise in determining what
kind of interferences these mixture pesticides served
to be on the single component pesticides.
For those of you who have analyzed, especially
toxiphene and chlorodane, you know that the chromatogram
looks like a forest and the mass spectrum is certainly
the trees, because these isotope patterns are
throughout, and they certainly cause interferences.
SLIDE
It was our determination that these interferences
were not significant as long as these concentrations
of the mixture pesticides were in the neighborhood of
a factor of 10, at least not more than a factor of 10
greater than the single component pesticides.
SLIDE
These are the observed interferences that I'm
talking about. The top one there says that hepta-
clor is a component of chlorodane. That's obviously
an interference we're not going to be able to overcome.
If the chlorodane concentration is high enough, you
-------
430
will get, obviously, a positive interference with
heptaclor.
The chlorodane interferes also with the labeled
and unlabeled forms of alpha-endosulfan and with the
labeled alterant. Toxophene interferes with D4-endo-
sulfan as well as the unlabeled endosulfan, and PCB
1254 interferes with DDE and alpha-endosulfan...excuse
me, the labeled DDE alpha-endosulfan and labeled
alpha BHC.
Again, these concentrations have to be significant
for these interferences to be a real problem, but if
you do detect significant quantities of the mixture
pesticides you're not going to be able to get good
quantitative determination using isotope dilution on
single component pesticides in that sample.
SLIDE
The next step in the method validation was the
determination of the calibration range and the
estimated instrument detection limit. To do this, we
have analyzed standards covering...well, including
concentrations of .1, .3, 1, 3, 10, 30, 100, 300 and
1,000 micrograms per milliliter. In these mixtures,
the labeled pesticide concentrations were maintained
constant at 50 micrograms per milliliter and the
internal standard at 100 micrograms per milliliter.
-------
431
At each of these concentrations where the pesticide
was detected, we calculated the relative response
factors, the relative standard deviation for the
relative response factors. Again, all of this
data is presented in the report. It covers a number
of pages. Suffices to say that the estimated detection
limits were reasonable, as illustrated, a portion of
which on this slide.
Observation-wise, we were unable to detect the
unlabeled pesticides at the concentration levels of
.3 and .1 micrograms per milliliter...not terribly
surprising...and that we had some real linearity
problems with a number of the pesticides at the 1 and
3 micrograms per milliliter range. This is a range
in which a lot of absorption is occurring and we get
fall off in the relative response factor as the con-
centration approaches these levels.
The estimated detection limit for the instrument
was determined at that point in which the area for
the extracted ion current profile for the quantitation
mass was 1000; 1000 is a fairly arbitrary number
based on the use of an Incos data system. If one is
using a different type of mass spectrometer and data
system they would have to establish this number for
themselves. But on the Incos, that level of 1000 for
-------
432
a peak area is fairly reproducible and you're not
getting a lot of statistical variation in that
measurement itself. And we needed some arbitrary
limit to calculate the detection limit. That's what
Dale chose and that's what we used.
At the upper end, DDE, DDT and ODD all saturated
the mass spectrometer at the 1000 microgram per milli-
liter. The rest of the components did not saturate
the mass spectrometer, but they certainly did overload
the GC column. All of the peaks at that level showed
a decrease in resolution and began looking something
like bananas, and interference problems would certainly
increase dramatically because of the lack of separation
that one would achieve. May I have the next slide,
Bruce.
SLIDE
This is just a continuation of the estimated
instrument detection limits. I left the other one up
there longer because the numbers were smaller. These
numbers are somewhat larger. Beta endosulfan there,
you see, is a 30 microgram per milliliter and endo-
sulfan sulphate at 24. Obviously these materials are
not detected at the lower concentrations as well.
SLIDE
The next determination that was done was the
-------
433
calculation of the method analytical range and the
estimated detection limit for the method itself. To
do this we spiked seven 2-liter water samples with
pesticides of the concentrations of lf 3, 10, 30,
100, 300 and 1000. Again, we dropped off the .1 and
.3 because we were unable to detect the pesticides at
those levels and spiked the rest of the levels. The
labels were maintained at 50 and the internal standard
at 100. These calculations are based on the con-
centration in the final extract after it had been
concentrated down to 200 microliters.
Again, the calculation of estimated detection
limit is based on an extrapolation to an extracted
ion current profile of 1000 on the Incos data system.
SLIDE
Basically, the estimated method detection limits
paralleled the estimated instrument detection limits
reasonably well. We expected that the recovery would
be above 50 percent, and most of these numbers reflect
that.
Observation-wise, again at the high concentration,
the 1000 level, we're saturating the GC column, or
overloading the GC column, and in many cases saturating
the mass spectrometer. At the one microgram per
milliliter level, we're generally not detecting the
-------
434
pesticides. So, again, we're trimming a little bit
on both ends in order to come up with a range over
which calculations can actually be made.
SLIDE
The method detection limit was determined by
spiking another seven 2-liter water samples at a
concentration near that of the estimated detection
limit, which we have just determined in the previous
set of slides. The labels at this point are spiked
into the water at a five microgram per liter con-
centration, and the detection limit, as presented
on this and the next slide for the various pesticides,
is a calculation of three times the standard deviation
of the measurements. We're looking here at levels of
200 to 400 parts per trillion.
SLIDE
The worst case for the endosulfan sulfate is
about two and a half parts per billion.
SLIDE
The next step was a determination of a series of
five point calibration curves to determine calibration
linearity and calibration reproducibility. We ran...I
don't have any slides on this. We ran a series of
curves at concentrations of 25, 50, 100, 250 and 500
micrograms per milliliter, calculated the response
-------
435
factors of the various concentrations for individual
curves, relative standard deviation, and the between
curve relative standard deviations to determine that
indeed, the calibration was linear or reasonably linear,
very linear over this concentration range, and the
reproducibility of that calibration was within reason.
SLIDE
The last set of slides that I have here provide
the precision and recovery data. To achieve the
measure of precision in recovery we analyzed two sets
of four replicates each at- a concentration 20 times
the minimum detection limit for the unlabeled
pesticides. Again, the labels were maintained at a
five microgram per liter concentration.
Generally speaking, as you look down this slide,
the recoveries were reasonable. Again, the labeled
compounds were calculated, their recoveries based on
internal standardization. The unlabeled compounds,
which have labeled analogs, are calculated on the
basis of using the label isotope dilution method.
The column to the far right is the relative standard
deviation of the detected amount, and as you can see,
the relative standard deviation is always...is most
often less when you're dealing with an unlabeled
compound which has a labeled analog.
-------
436
The top one there, Alpha BHC, the labeled compound
has an RSD of 18, and the unlabeled compound an RSD of
8, and this held pretty true throughout the study.
We got a significant decrease in the relative standard
deviation of the measurement and the recoveries were
in the neighborhood of 100 percent or greater.
SLIDE
This just follows through the trend showing the
result. And I think the last slide is the next one,
which continues on.
SLIDE
The recovery of the labeled compunds in those
materials which did not have labeled analogs is also
reasonably good. We're talking in the 60 to 80
percent recovery range. At these levels that's not
too awfully bad. ,
I noticed that there were only three of us that
didn't submit an abstract as I was reading that thing.
They've got little stars by your names. Paul Friedman,
myself and some guy named BREAK, and so I did bring
some of the report. And as I mentioned at the start
of the meeting, I'd be happy to send you one if you
just want to leave me a card during that break.
Now, I'd be happy to answer any questions that you
might have.
-------
437
QUESTION AND ANSWER SESSION
MR. TELLIARD: Trying to
make up for the fact that you didn't submit an
abstract. Mailing some of those silly reports purely
isn't adequate. Dr. George has a question.
' MR. STANKO: George Stanko
from Shell Development. Bob, on one of your slides
you showed that the recovery of the carbon 13 analog
of aldrin was 39 percent with a relative standard
deviation of 29, and that the analyte itself, aldrin,
the recovery was 117 percent with a standard deviation
of 15. How can the isotope dilution procedure be
applied when you're not recovering your deuterated or
carbon 13 analyte at the same percentage as the
analyte itself?
MR. TELLIARD: I think
that's a knit-picky question.
DR. BEIMER: I tried to go
over that portion quickly, George. You must have
picked up one of the reports, right?
MR. STANKO: I didn't go
out last night, I stayed home.
DR. BEIMER: What we're
dealing with here, George, is the...where are we?
-------
438
MR. STANKO: The bottom
two. That is the whole basis for isotope dilution.
DR. BEIMER: Yes. The
labeled compound there, George, the recovery of the
labeled compound is calculated relative to an internal
standard. The unlabeled recovery is calculated
relative to the labeled compound, so it will naturally
be higher. It's corrected for that recovery. We did
not go through and calculate the unlabeled recovery
as a function of internal standard.
Did I make that clear?
MR. STANKO: It was clear,
but that would mean that the slide would be confusing
to someone who tried to interpret it the way I had
done. What you're really trying to tell me is that
the recovery of the aldrin itself was approximately
39 percent in this case.
DR. BEIMER: Yes, that's
correct.
MR. STANKO: And you're
showing me how the recovery corrected data would
appear.
DR. BEIMER: That's correct.
MR. STANKO: I think I would
recommend that you alter this slide to point this out.
-------
439
I think some people are going to be confused later
on.
DR. BEIMER: The idea here
was to point out that isotope dilution does indeed
correct for recovery of the materials, and I apologize
for the confusion that that might have presented.
The relative standard deviation in the final column,
though, is a measure of the absolute concentration as
determined, and I was looking more here at the
precision of the measurement than the actual recovery.
MR. MADDALONE: Ray Maddalone,
TRW. Bob, two questions. One, when you did your MDL,
did you use a distilled water sample or did you use
a real water sample with low pesticide concentrations?
DR. BEIMER: All of this was
done on as pure a laboratory water as we possibly could
obtain. We ran blanks alongside the...lab blanks
alongside of the analyses., We have yet to test this
method on real world samples, and I think there should
be some legitimate concern about concentrating real
world samples to 200 microliters. I'll put both hands
up in the air on that. We're undoubtedly going to
have to use some GPC cleanup in order to achieve
these kind of detection limits, and our recoveries
will probably drop off accordingly.
-------
440
MR. MADDALONE: The second
question. On the RSD's that you have for the labeled
compounds versus the unlabeled compounds, it looks
generally that the labeled compounds have a higher
relative standard deviation. Is there any particular
reason for that?
DR. BEIMER: Yes. The
unlabeled compounds...the scatter in the data is
reduced because they are calculated using the labeled
compounds as an internal standard. You're probably
confused the same way George is, and I apologize for
that. I probably should have put this on two separate
slides so that you could separate the two effects.
MR. MADDALONE: So in
effect, the deuterated compounds are telling us what
the actual recovery and precision is of the method,
and then both of those data bits were used to correct
the recovery and the relative standard deviation or...
DR. BEIMER: Well, you
don't actually correct the standard deviation.
MR. MADDALONE: Not correct
it, I'm sorry.
DR. BEIMER: The standard
deviation goes down because you're correcting for the
variability of the method because the labeled compound
-------
441
mimics the unlabeled compound in the extraction and
in the chromatography.
MR. LUCAS: Sam Lucas from
Battelle. You had a number of compounds you were
clearly disappointed in your detection limits on. I
wonder if you had a chance to try to determine whether
the fault was extraction of those compounds, losses
in your evaporator cavity injector, losses on
chromatography, poor mass spec characteristics or
whatever might cause the high detection limits.
DR. BEIMER: As you may
have noticed in the early slides, the detection limits
that we observed...the high detection limit compounds
were also those compounds that we got poor detection
with the mass spectrometer, so that we're not dealing
with a significant extraction problem, we're dealing
with a mass spectrometer problem. And, unfortunately,
in some of those cases, we're having to choose masses
because of potential interference problems which are
not the largest in the spectrum. If you throw away,
let's say, you look at a 40 or 30 percent abundance
mass peak, you're throwing away that fraction of the
total sensitivity to that compound in your instrument,
and we did that in some of these cases. We're down
with a couple of those compounds at mass abundances,
-------
442
mass peaks, that are in the 10 to 20 percent range in
order to not have interference problems.
MR. LUCAS: Did you have an
opportunity to try cold long column injection to see if
any of your thermile labile compounds could be improved
that way?
DR. BEIMER: No, we didn't
try that.
MR. LUCAS: You mentioned
that a proper deactivation of your injector liner was
extremely critical, and that's what made me think
that perhaps your method could be substantially
improved with cold long column injection.
DR. BEIMER: It's a very
good possibility.
MR. BARRICK: Bob Barrick,
Tetra Tech. Do you have any...for any of these
labeled analogs, do you have sufficient chromatographic
resolution such that they might possibly be done by
GCCD?
DR. BEIMER: I'm sorry, by
what?
MR. BARRICK: Do you get
enough chromatographic resolution on any of the
labeled analogs such that you might be able to take
-------
443
any of these and use just a few of those as recovery
standards using GCCD detection?
MR. TELLIARD: Electron
capture.
DR. BEIMER: Electron
capture. Okay, I'm sorry.
MR. BARRICK: Yes, for you
modern chemists, electron capture, yes.
DR. BEIMER: Okay, I'm
sorry. Generally speaking, the resolution on the
column is not sufficient to do that. The peaks would
at least be overlapped. I'm going to embarrass'
myself right here to tell you that I've not seen one
of the chromatograms. I'm not sure what the degree
of separation is.
MR. BARRICK: At least
you're honest there.
DR. BEIMER: The other
fellow whose name's on the report actually did
the work. He's the one who really deserves the credit
for what was done here. His name is Lee Helms.
MR. TELLIARD: Thank you,
Robert.
-------
444
ANALYTICAL CONDITIONS
GO - 30M, 0.25mm i.d. - DB-5
He carrier at 30 cm/sec
Splittless injector
100°C - 280°C @ 8°C/min (10 min hold)
MS - Finnigan OWA 1020
Scanned 35-450 amu @ 1 sec
Tuned for Method 1625B DFTPP
(2j2'-difluorobiphenyl used as internal standard)
-------
445
CHARACTERISTIC PESTICIDE MASSES
Compound
Aldrin
a-BHC
/3-BHC
-y-BHC
(5-BHC
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Analog
13,
'C
16
16
d
d
8
8
13
C
Primary
m/z's
263/269
219/224
219/224
235/243
246/254
235/243
263/269
-------
446
Characteristic Pesticide Masses (Continued)
Compound
a-Endosulfan
/3-Endosulfan
Endosulfan sulfate
Endrin
Endrin aldehyde
Heptachlor
Heptachlor epoxide
2,2c-Difluorobiphenyl (I.S.)
Analog
13
C
Primary
m/z?s
170/164
241
.'; , !
272
81
67
160/164
81
190
-------
447
INTERFERENCES
• Heptachlor is a component of chlordane
• Chlordane interferes with (d,) a-Endosulfan,
13
a-Endosulfan and ( C.) Aldrin
• Toxaphene interferes with (d.) a-Endosulfan and
a-Endosulfan
PCB 1254 interferes with (dg) 4,4'-DDE,
a-Endosulfan and (dfi) a-BHC
-------
448
CONCENTRATIONS AND STANDARD
DEVIATIONS FOR
PESTICIDE COMPOUNDS
Relative
Average Standard
Recovery (%) Deviation
402 (d6)a-BHC
102 a-BHC
103 /2-BHC
404 (d6h-BHC
104 7-BHC
105 &-BHC
400 (13C4) Heptachlor
100 Heptachlor
289 (13C4) Aldrin
089 Aldrin
75
18
80
65
65
99
59
103
75
39
117
8
22
23
8
23
26
21
29
15
-------
449
CONCENTRATIONS AND STANDARD
DEVIATIONS FOR
PESTICIDE COMPOUNDS (Continued)
101 Heptachlor epoxide
295 (d4) a-Endosulfan
095 a-Endosulfan
293 (d8)4,4<-DDE
093 4,4C-DDE
294 (d8)4,4'-DDD
094 4,4<-DDD
292 (d8)4,4'-DDT
092 4,4'-DDT
Relative
Average Standard
Recovery (%) Deviation
75
20
72
134
56
120
61
119
162
139
23
6
32
5
37
8
46
10
-------
CONCENTRATIONS AND STANDARD
DEVIATIONS FOR
PESTICIDE COMPOUNDS (Continued)
450
Relative
Average Standard
Recovery (%) Deviation
290 (13C4)Dieldrin
090 Dieldrin
098 Endrin
096 /3-Endosulfan
099 Endrin Aldehyde
097 Endosulfan sulfate
186 (d5)2c-Chlorobiphenyl
187 (d5)3c,4',56-Trichlorobiphenyl
188 (d6)3,3',4,4'-Tetrachloro-
biphenyl
58
35
113
59
83
53
100
69
66
62
22
30
31
32
36
11
24
37
-------
451
ESTIMATED INSTRUMENT
DETECTION LIMITS
FOR PESTICIDE COMPOUNDS
Compound
089 Aldrin
102 a-BHC
103 /3-BHC
104 7-BHC
105 &-BHC
094 4,4C-DDD
093 4,4'-DDE
092 4,4'-DDT
Estimated Instrument
Detection Limit (#g/mL)
5
5
5
5
5
3
3
3
-------
452
ESTIMATED INSTRUMENT
DETECTION LIMITS
FOR PESTICIDE COMPOUNDS (Continued)
Compound
090 Dieldrin
095 a-Endosulfan
096 /?-Endosulfan
097 Endosulfan Sulfate
098 Endrin
099 Endrin Aldehyde
100 Heptachlor
101 Heptachlor Epoxide
Estimated Instrument
Detection Limit f/zg/mL)
10
10
30
24
5
3
10
3
-------
453
ESTIMATED EXTRACTION
DETECTION LIMITS FOR
PESTICIDE COMPOUNDS
Compound
089 Aldrin
102 a-BHC
103 /3-BHC
104 7-BHC
105 £-BHC
094 4,4'-DDD
093 4,4'-DDE
092 4,4;-DDT
Estimated Extraction
Detection Limit
(2L Sample)
0.5
0.5
0.5
0.5
0.5
0.3
0.3
0.3
-------
454
ESTIMATED EXTRACTION
DETECTION LIMITS FOR
PESTICIDE COMPOUNDS (Continued)
Compound
090 Dieldrin
095 a-Endosulfan
096 /?-Endosulfan
097 Endosulfan sulfate
098 Endrin
099 Endrin aldehyde
100 Heptachlor
101 Heptachlor epoxide
Estimated Extraction
Detection Limit
(2L Sample)
1
1
3
2.4
1
0.5
2
0.3
-------
455
METHOD DETECTION LIMIT
FOR PESTICIDE COMPOUNDS
BASED ON INITIAL VOLUME
OF 2 L FOR WATER SAMPLE
Compound
102 a-BHC
103 /3-BHC
104 7-BHC
105 £-BHC
100 Heptachlor
089 Aldrin
101 Heptachlor epoxide
095 a-Endosulfan
Method Detection
Limit
0.21
0.34
0.19
0.26
0.47
0.44
0.16
0.80
-------
456
METHOD DETECTION LIMIT
FOR PESTICIDE COMPOUNDS
BASED ON INITIAL VOLUME
OF 2 L FOR WATER SAMPLE (Continued)
Compound
093 4,4C-DDE
094 4,4C-DDD
092 4,4'-DDT
090 Dieldrin
098 Endrin
096 /?-Endosulfan
099 Endrin aldehyde
097 Endosulfan sulfate
Method Detection
Limit //g/L
0.39
0.18
0.20
0.24
0.62
2.14
0.54
2.45
-------
457
MR. TELLIARD: Our next
speaker is John McGuire from our R&D lab in Athens.
Most of you know John as the father of Walt Shackel-
ford. He's not that old; I just made that up. John.
-------
458
JOHN MCGUIRE, PH.D.
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ATHENS E.R.L.
PLANS FOR AN IMPROVED ALGORITHM FOR FINDING
GC PEAKS IN GC/MS DATA
SLIDE 1
DR. MCQUIRE: This is
what it's all about. I don't know how legible that
slide really is, it's rather an old slide. In fact,
it dates back to the early days when EPA was getting
set up. Those are fish, all belly up, and had to do
with why EPA was established in the first place. The
Clean Water Act helped to focus on insults like this
to the environment, and since then, year by year,
there has been a general improvement.
SLIDE 2
This is a question that we were asking at the very
beginning of EPA...actually two questions...and, you
know, those questions are still being asked today.
So, I think they were very pertinent then, they are
very pertinent now, and have an awful lot to do with
analyzing samples connected with the RCRA program in
particular, and with all of the EPA programs in
general.
-------
459
In the early 70's, the infant EPA began to use
an extremely expensive detector for a gas chromato-
graph, but that extremely expensive detector did give
a fingerprint, a mass spectrum, of each GC peak as it
eluted from the column.
Our Athens lab, together with research grantees
in various parts of the country, led the way in
making a spectra matching program that had been
developed by Klaus Bieman and his associates at MIT
available, first of all, to all of the other EPA
laboratories to make tentative identifications
much more quickly than could be done by manual
interpretation; from them, it was picked up and
spread to other users of System Industries equipment.
This approach had the unfortunate drawback, of
course, that sometimes people who didn't know what
they were doing used the computer's interpretation as
being gospel.
We also realized that the available spectral
collections were biased strongly towards petroleum
spectra and biological spectra; and we began a project
to collect spectra of pesticides and other environmental
pollutants, which later was picked up by Steve
Heller and Bill Milne in Washington, and has now
become the EPA-NIH data base that is the standard for
-------
460
all GC/MS users.
In '76, EPA entered into a consent decree with
NRDC and other environmentalist groups concerned
about establishment of effluent guidelines for
industry. This gave visibility to the Effluent
Guidelines Division which, as you all know, is now
ITD.
Our Athens Research Lab, together with the
Cincinnati Lab, helped the Effluent Guidelines
Division in setting up the priority pollutant list in
its final, analyzable form. I'm sure you all are
aware of the difference between the original list and
the final one. The original Consent Decree has been
modified several times.
When the survey phase was well underway, parties
to the original consent decree agreed to a significant
amendment: yesterday Tom Fielding made reference
to Paragraph 4(c) of the consent decree, which agreed
that EPA would massage data collected by contract
labs as part of the EGD Effluent Study for non-
priority pollutants. Walter Shackelford, who at
the time was part of my group in Athens, has reported
to this meeting in the past on plans, progress and
some of the results of that study. Since his work
was what Tom referred to as the "first bite of
-------
461
Paragraph 4(c)" and this is the second, I want to
refresh your memory on what was involved.
By the way, yesterday Telliard questioned the
repute of the Federal Register as a scientific journal,
so much of what I have said to date can be found in
Environmental Science and Technology. I'm sure Bill
would approve of that reference since he is one of
the authors.
Data from the contract labs' priority pollutant
runs were stored on mag tape and sent, ultimately at
least, to our Athens lab for processing. Walter and
his contractors had set up a set of programs based on
the CLEANUP and HISLIB, Historical Library, programs
developed at Stanford, and the PBM program, Probabi- ,
lity Based Matching, developed at Cornell.
SLIDE 3
The first stage of that set of programs,
RETRIEVE, was a simple one designed to read 800 bits
per inch tape from Finnigan, HP, or System Industries
data systems, produce a chromatogram of the data, and
store the data in a fixed format for future processing.
The second, or BUILD stage, was an interactive
one wherein the chemists made sure that all pertinent
parameters, such as scan speed, internal standards,
mass range and chromatographic column, were entered
-------
462
into a data base for the particular GC/MS run.
The next three programs located the internal
standards and assessed the GC peak profiles for all
standards. These profiles were considered model
profiles for compounds eluting nearby. I'll get
into that more in a moment.
SLIDE 4
MODELS was followed by CLEANUP, a program that
extracted a typical mass spectrum reasonably free of
background for each GC-separated compound. This
spectrum was then put through McLafferty's data
compression program to emphasize unique mass numbers
and processed through what we then felt was an extremely
large, 50,000 plus, collection of spectra. It was,
in fact, the super set of the EPA/NIH library at that
t ime.
Walter and his colleagues found that reliability
of the matches could be improved greatly...and that
means less "false positives"...by requiring fairly tight
retention time windows. This relative retention time
window was calculated for the PBM hits in the same
part of the program that calculated concentration,
based on the known concentration of internal standards.
If any of you remember Walter talking about
problems with the initial tape study several years
-------
463
ago, finding these internal standards at known
concentrations was one of the biggest problems we had
initially. The d^o anthracene that was supposed to
be there as an internal standard simply wasn't there.
Finally, HISLIB checked to see if the tentative
identifications fit the historical relative retention
time for that particular compound. If it did, a HIT
file was updated to show a new identification of that
particular compound. If it did not, a MISS list was
consulted to see if that same unidentified spectrum
had been found in that same relative retention time
window for other samples before. If it had been, the
MISS file was updated. If it had not, or if either
the PBM score or the relative retention time window
was a 'close but no cigar1 type, the spectrum was
output for an analyst's decision as to which file it
would go into. This, briefly, was the program Athens
and EGD had in place during the "first bite" phase.
SLIDE 5
A summary of what was found is given in the next
slide. I don't want to take time to dwell on any of
the points here except the number of spectra not
identified. That 2,500 represents spectra that
occurred more than once and were felt by the chemists
who did the final decision making to be reasonable
-------
464
appearing spectra, but were not identified. There
were many other spectra that were not identified that
were judged by those same chemists to be trash and
they were not included in the MISS list.
As I have already indicated, we superimposed the
relative retention time windows on the PBM searches,
and we tightened those windows up to eliminate false
positives. Well, any analyst knows if you tighten
things up to eliminate false positives you create
false negatives. Going through a portion of our MISS
list by hand, it seems quite evident that a substan-
tial portion of that list, perhaps as high as a quarter
of it, is indeed false negatives.
As well as the results and the summary, certain
other conclusions came out of this "first bite". These
included the obvious facts that with some caution,
capillary column data was more apt to give
chromatographically resolved peaks than packed columns.
And that more than one or two internal standards were
needed.
As a result, contractors for the data collection
of the "second bite" were instructed to use experimental
improvements. Now those improvements, basically, so '
far as my particular concern at this point, come down
to these. Fused silica capillary columns were used
-------
465
for all but the VGA's, and multiple internal standards
were used. It was also felt, based on our results in
that first study, that significant changes in the
algorithm would be needed to do justice to capillary
column data.
SLIDE 6
Yesterday, Bill Telliard alluded to his having
more money. Well, the implication when one says he
has more money is that one had less money in the
past. We, down at Athens, have had that problem, too.
So, at the end of the "first bite" tape study, we put
our production tape program in mothballs for lack of
money and saw one after another of our contract
programmers and contract chemists who had gotten the
Athens system operating in the first place, leave for
bigger and better things.
SLIDE 7
Accordingly, our first priority on setting up
for the "second bite" has been to restaff for the study.
SLIDE 8
A contract programmer with numerous phone
consultations with Walt Shackelford, has brought the
old programs back on line on the PDF 11/70 used in the
earlier work. He has made the programs portable by
redoing all machine language I/O calls into FORTRAN 77,
-------
466
and is now in the process of transferring the system
to the VAX 785 to give both the faster speed and the
double precision arithmetic that we found we needed
in the first study.
We're now having the contractor recruit a data
chemist to work on the project; if anybody knows
one, let me know. We have placed an order for the
May, 1986, release of the Wiley Library. This will
contain over 100,000 spectra. We plan to merge that
with our present collection, which is the current EPA/
NIH collection plus our original collection, by
eliminating all identical spectra.
Incidentally, along with the tape program, we
also have been working with ITD in other ways. We've
been analyzing selected samples for the Domestic
Sewage Study, and as a result of that we recommended
that 30 organics from Appendix VIII be added to 1625B,
and I believe from one of the booklets that Bill has
handed out, they have been.
Back on the main theme though, we've identified...
and that was no mean task if any of you have tried
improving someone else's software...those areas that
need to be updated for multiple internal standards,
and are currently beginning to address them.
SLIDE 9
-------
467
When one remembers that the early priority
pollutant days for semivolatiles had only a single
internal standard, and compares that to the abundance
of, for example, the cocktails that are added
for isotope dilution, it seems obvious that program
modifications must be made to best find and utilize
the information that will be available in the newer
data. This slide shows a single spike set, but all
told, there are around 75 spikes that have been used
at one point or another in this phase of the program.
In work evaluating a different program, the
Master Analytical Scheme that was developed by the
Athens lab and a contractor, we found that in capillary
column GC/MS with internal standards it was much
more important to compare the analytes to the standard
eluting closest to them than it was to compare them
to similar chemical compounds. And that's fortunate
since the spiked sets...again using the same spiked set
that's here...tend to be reasonably well spaced over
the entire chromatogram. This gives convenient
references for almost all the analytes that we're apt
to find.
SLIDE 10
I said earlier we had felt, based on the original
tape study, that significant modifications of the
-------
468
algorithms would be needed to apply the programs to
cap column data. I'd intended to go into these
changes in depth today; however, as we have dug deeper
and deeper into the old programs in the past three
months, we found that that may not be necessary. I
believe right now our job will be much simpler. Rather
than constructing new peak recognition algorithms, I
now feel that tuning existing system parameters will
do, and this work has just started.
SLIDE 11
The need for the tuning is shown in the next
slide. Here we have real time traces of a continuous
monitor of GC peaks eluting: first, the early part
of a capillary column run; next, a little bit later;
then, a little bit later; and, finally up towards
the point where one is starting to get into an
isothermal portion of the run. The peak width at the
base of this peak, which is at the beginning of the
run, is eight and a half seconds. That is,
incidentally, a single compound. As RT increases,
peak width decreases: four and a half seconds on
the second peak, three seconds on the next, and two
and a quarter seconds on the last.
Now, consider what happens if your mass spectro-
meter is scanning at one second per scan. If you do
-------
469
your scans on the first peak, you're going to get
eight and a half scans across it. If the scans
sync with the eluting GC peak, your mass spectrometer
is going to come up with a very nice RIC that looks
like a single peak.
If the sync is slightly different, the mass
spectrometer is going to think that it sees two
peaks. Similarly, further in the chromatogram
an improper sync will cause the system to miss the
top part of the peak.
For that reason we feel we're going to have to
do a little bit of massaging. The GC peak shown
here is indeed a valid peak shape for the early part
of the run. We are going to have to tune our
parameters so that the system can sense that.
During the programmed portion of the GC run, we've
no problems. At the second isothermal stage, we're
going to have to come up with a means of extrapolating.
SLIDE 12
The confusion of the tape processing programs is
illustrated by the next few slides. This one is a
reconstructed chromatogram as it conventionally
appears. (For those who use Finnigan terminology, a
RIC.)
SLIDE 13
-------
470
This slide is of the actual data points that
went into that pretty scan of Slide 12. Note the
scans on the ascending (front) edge of the "939"
peak, followed by the large gap on the descending
side where there were no data being taken. At the
"1021" peak, we've a nice set of data points going
up the peak and then nothing at all until we get down
to scan 1022, where the chromatogram comes back up
again. Well, again, we have to tune our parameters
so that the system can recognize that.
I'd like you to take a look at the two GC peaks
near scans 902 and 1143. The leftmost arrow is
pointing to the front edge of the small GC peak near
scan 902; the next arrow points to the trailing edge,
but the top of the particular GC peak is actually
between them. In other words, I do not have an arrow
pointing to it. Similarly, near scan 1142, I've an
arrow pointing to the front and an arrow pointing
to the rear, but none pointing to the top. The
probability of finding scans on the sides of peaks
rather than the top is real and these are possibili-
ties that we could run into in actual data scans.
Now, I'm going to show you the four spectra that
I have "arrowed". The four consist of two pairs of
spectra, and the important thing here is to notice in
-------
471
the first pair (Slides 14 and 15) the relative
abundance of 131 to 232. It's around 30 percent in
Slide 14, nearly twice that in Slide 15. The next
one (Slide 16)...look at 198 to 442...it's around
two to one in Slide 16 and about six to one in
Slide 17. In other words, and very much to be
expected, an instrument that scans from low mass to
high mass will give more high mass bias on the
ascending side as the concentration in the ion source
is building up, and more low mass bias as it's
dropping off.
The best spectrum for the GC-peak, the one that
should most closely correspond to that in the reference
spectrum collection, should be at the top of the peak.
For example, our friend DFTPP gives a spectrum at the
top of the...gives this spectrum (Slide 18) at the
top of the peak, and a library match (Slide 19) shows
that agreement with the reference is very good. You
see the disagreement is plotted here, and if we look
at a library output (Slide 20), it has a purity of
846 and a fit...almost a perfect fit. Now, that is
again taken at the top of the peak.
So, the point is we have a bias towards the high
mass side on the ascending side of the GC peak and
the low mass side on the descending side: it is
-------
472
this bias in the spectra that we plan to investigate
as a means of estimating where the top of a GC peak
would have been if our mass scan happened to miss it.
We will estimate linear correction factors to
permit extrapolating the data points and compare
quantitative results with those obtained by the
existing programs. We haven't done any of it yet.
I hope to include a discussion of this approach in
the overall report of the "second bite" at next
year's meeting.
In concluding, I want to thank Dr. Walter
Shackelford and David Cline, who are now of RTF but
who were originally at the Athens lab, who set up the
original programs with help from Computer Science
Corporation, Al Thruston of our group in Athens who
was the one responsible for analyzing the domestic
sewage samples; Bruce Bartell of Computer Science
Corporation, who succeeded in getting the computer
system operational again and making substantial
improvements in the present coding. I must also
recognize the Stanford and Cornell researchers who
started the two main portions of the system.
Thank you. Are there any questions? Bill, take
over.
MR. TELLIARD: Any questions?
-------
473
You're going to let him off? People must really be
hurting. Thank you, John.
-------
474
-------
475
-------
476
-------
477
CD O
-------
478
-------
479
-------
480
-------
481
-------
482
oo
CD
<_5 CD
os:
-------
483
-------
484
-------
485
-------
486
S
-------
487
O 31= UU
LU 10
Cl_ TO •• «•
1-0 V UJ .
CSJ i.O o C3
-------
488
_
LU
-------
489
.
.. IX< UJ
XSi C-J O
O *= UJ
LU U~«
Q_ CO — »• CU
ui x LU • .sr
10 _J ".O LU
Lfl -'- •""• '
'.'"I , .
a: u~j •
-------
490
-------
491
-------
492
-------
_j in
LU.< ID
493
-" t N ro -o oo
u. DO>t- ru n ru
DC
=n-C O CO 03-0
' ~ O O
•>-< CO CM HI 01 CM
f-
3 ,
co n co N co
•p ru o co o co
V} dj ID
O .0 -
LU I _j
a. i >
ea LU a. lu LU
Q - 2 z:
Q -, T
_i _! DL
O Z UJ Z LU
c t> ru o m ru
Q.-Z ru tn
o -o . ru z
i o a a .
. . - . in
i« . O O 03 O
r-< in ru
t- CD co ru
-------
494
MR. TELLIARD: Our next
speaker is Lee Wolfe from the Athens Laboratory and
he's going to talk a little bit about the joys of
hydrolysis, or the lack thereof.
-------
495
N. LEE WOLFE
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ENVIRONMENTAL RESEARCH LAB
ATHENS, GEORGIA
HYDROLYTIC TRANSFORMATION OF
RCRA APPENDIX VIII COMPOUNDS
MR. WOLFE: The objective
of this talk is to acquaint you with a report that
assesses the hydrolysis of many of the compounds that
have been discussed here yesterday and today. The
hydrolysis data were prepared by EPA1s Office of
Research and Development for the Office of Solid
Waste (Slide 1). The tabulation of data provides
the bfest available data base, I think, on hydrolysis
for these compounds. The kinetic data contained in
this report should be useful in evaluating, designing
sampling procedures, sample storage and sample
analyses.
Specifically, I want to address parameters that
are in the report: hydrolysis rate constants, half-
lives, key to the data set, partial summary of the
reactivity, and some data-generating efforts that
are underway at the Athens Environmental Research
Laboratory.
A while back, at an ACS meeting I bumped
-------
496
into an old professor. I said, "Well, Dick, what
are you doing now; what systems are you working on?"
He said, "I just happen to have a slide in my
briefcase." He pulled out the slide (Slide 2) and
said, "I've got this exo-bicylco-2, 2-hexyl-2-yl
tosylate system. These are the products we're
getting." I said, "Well, Dick, you haven't changed
much, you're still looking for complex solutions to
simple problems."
He said, "Well, what are you doing?" I said,
"I've got this really neat system. We've been working
with it for quite a while now and I just happen to
have a slide (Slide 3). This is a pretty exciting
system." "Well*," he said, "it doesn't look to me like
you've changed much either; you're still looking for
simple solutions to complex problems."
I use this example to make the point, that in the
last several years or so it has been shown, that with
a great deal of confidence one can extrapolate
laboratory hydrolysis data to the field, and that's
pretty much the basis for this report and the work
that we have done for the Office of Solid Waste.
The report is titled, Screening of Hydrolytic
Reactivity of OSW Chemicals. The report, as I said,
was prepared for the Office of Solid Waste to support
-------
497
their efforts in modeling, fate and transport in
groundwaters, and certainly the data that were tabulated
is applicable to the work that's being done in
environmental analytical chemistry.
The list departs, I guess, from the Appendix VIII
list in that there are only 362 compounds. I'm not
really sure of the origin. It has things like Fast
Track, California List, and then thirds. My
understanding was that originally it was the
Appendix VIII.
We've only looked at data on hydrolysis for the
organic compounds. There are some organo-metallic
compounds on the list, and what data was available, we
have tabulated and evaluated. We have not made
any attempt to do anything with the inorganics such
as hydrolysis of organic ligands.
What can this report do for us? We know that
hydrolysis reactions or hydrolytic reactions can
result in products that are unigue to the list. For
example, benzyl chloride, one of the compounds on the
list, undergoes very rapid hydrolysis (Slide 4).
It has a half-life of a couple hours in water and
gives benzyl alcohol as a product, which would say
that you're wasting your time if you look for benzyl
chloride in a water sample, but if you want to know
-------
498
if it was there, the presence of benzyl chloride
might be an indication.
Also, products that are already on the list, for
example, 1,1,1-trichloroethane, readily undergoes
reaction in water, in the presence of base, to give
1,1-dichloroethene. 1,1,1-trichloroethane is on the
OSW list, and also 1,1-dichloroethene is on the list.
One doesn't have to look very far to find examples
of compounds that are not on the list but will undergo
hydrolysis reactions to produce compounds that are on
the list.
Well, let's be a purist for just a second and go
back and define hydrolysis (Slide 5). The definitions
do become impdrtant sometimes. I define hydrolysis
as a reaction in which a bond between two atoms of a
molecule is cleaved and a new bond is formed with the
oxygen atom of a water molecule. These reactions are
often, though not necessarily, mediated by acid or
base. Certainly a good example of this is the
hydrolysis of 2,4-D acid esters. The acid esters
undergo alkaline hydrolysis to give the acid, and in
many cases, the hydrolysis of compounds result in
more polar compounds, that are difficult to extract
from water and often more difficult to analyze for.
Another term that's thrown around a lot, a
-------
499
much broader term, is hydrolytic degradation, a term
that's used to denote a chemical reaction that occurs
in water. For example, the reaction you just saw,
1,1,1-trichloroethane going to 1,1-dichloroethene
is really an elimination reaction, but it's generally
just grouped as hydrolytic degradation.
Many hydrolysis reactions are affected by pH,
and I put this slide up (Slide 6) a plot of PH
versus half-life to demonstrate some of the effects
that you might run into. You see three types of
effects. Looking at D, an acid catalyzed hydrolysis
reaction, you see that as the pH increases, the
half-life for this reaction increases. The thing to
note Is that for each pH unit increase, the half-life
for this reaction decreases by one order of magnitude.
The other type of reaction you have is a base
catalyzed reaction shown by E, whereas the pH
increases, the half-life decreases.
The third type of reaction, which is a neutral
hydrolysis reaction is indicated by B. We see
that this reaction is pH independent, that a
change in pH of, as shown here, six orders of
magnitude, does not change the hydrolysis half-life.
The report addresses this in the following way
(Slide 7). What you're interested in, in assessing
-------
500
whether or not your compound is going to hydrolyze,
is the observed rate constant. The observed rate
constant is made upf in many cases, of three
contributions: acid hydrolysis, neutral hydrolysis
and alkaline hydrolysis. The report contains the
second-order acid catalyzed hydrolysis rate constants,
the first-order neutral hydrolysis rate constants,
"and the second-order alkaline hydrolysis rate
constants.
What you need to calculate K-observed is the
hydrogen ion concentration or the hydroxide ion
concentration, and those numbers are available from
the pH of the solution.
K-observe'd is a fairly difficult number to live
with. It's a fairly abstract number, and to most
people it doesn't have much meaning. A much better
way to get a feel for these numbers is to convert
them to half-lives (Slide 8). Half-life is generally
defined as the time reguired for the initial
concentration of the reactant to be reduced to one-
half the initial concentration. In this expression,
remember that one-half life, T one-half is egual to
the constant .693 divided by K-observed, recalling
that K-observed is really the summation of the three
hydrolysis processes.
-------
501
Because of the large amount of data and diversity
of reactivity of the compounds, varying from a few
seconds to hundreds, or even thousands of years, we
divided the report up into 15 fields (Slide 9).
The first two fields are the constituent code and
the chemical abstract register numbers. The constitu-
ent code is a code that was provided to us by OSW.
The chemical abstract's register number is important
to us because that's an unambiguous way of identifying
all the compounds.
The rank, field three, is really a scheduling
priority. That's the order in which OSW is going to
be making decisions on whether or not these compounds
can bte dumped into landfills, and this was the order
in which they wanted us to do our work.
Compounds, field four, were named as received.
When you get down to fields five and six, this is
where we start to get into the meat of the report.
What we established was whether or not the compounds
hydrolyze or not, and if they do hydrolyze, will
their hydrolysis half-lives be less than a year or
greater than a year.
Field number five indicates that there is no
hydrolyzable functional group on the compounds. For
example, a compound like chrysene, an aromatic hydro-
-------
502
carbon, has no functional groups that can hydrolyze.
Field number six, non-labile functional groups,
is something like chlorobenzene. Chlorobenzene does
have a carbon-chlorine bond that can hydrolyze, but
it does not hydrolyze at a significant rate in water
even under extreme reaction conditions, so we indicated
those compounds as non-labile functional groups.
Fields seven and eight indicate whether the
compounds have a half-life of less than a year or a
half-life greater than a year. Fields 9, 10 and 11
indicate where the data came from. Field 9 is
experimental values. These were experimental values
that came either out of the Athens laboratory or
literature. 'Sen indicates estimated values. These
are values that can be estimated through various
free energy relationships. In some cases you can
estimate the values almost as well as you can measure
them. Then we have expert judgment, field 11. For
some of the compounds there was not any hydrolysis
data available, so we used a panel of four experts
and their best estimates.
The numbers in fields 12, 13 and 14 are the
acid hydrolysis rate constant, base hydrolysis rate
constant and the neutral hydrolysis rate constant
respectively. We also set up an additional field,
-------
503
a kind of catch-all field that included compounds that
are not stable, or compounds which we felt there was
not much known about their reactivity and we couldn't
even make good expert judgments, and gave these high
priority for measurement.
This slide (Slide 10) is page six out of the
report, and I know it's difficult or impossible to
read. I've put i.t up just to show you the format of
the report. What you see are the 15 fields across
the top with Code, CAS number, rank, compound and
then the data.
For example, dimethylsulfate has a half-life of
less than a year, it is a measured value and it is a
neutrtal hydrolysis. The hydrolysis rate constant of
0.6 translates into hydrolysis half-life of about one
hour.
Some of the compounds are so unstable in water •
that measured values for the rate constants are not
available. They're just so reactive that you can't
measure them by conventional means (Slide 11). This
is a list of some of these compounds. There are
eight compounds that undergo very fast hydrolysis
reactions; half-lives on the orders of seconds or
less.
There are other compounds that do not undergo
-------
504
hydrolysis reactions but do undergo oxidation-
reduction reactions, and we identified these as such.
The two peroxides are very strong oxidizing agents
and very reactive. The three hydrazines are very
strong reducing agents, very reactive, and will
not exist in most natural waters.
Also, we identified seven aldehydes that will
likely exist in water as the hydrates. The aldehydes
react with water to form the hydrate. It's a
reversible reaction and by adjusting the pH, you can
sometimes force a reaction back to the aldehyde. But
it does present problems in extracting these compounds
out of water.
I've pull«ed out some of the data to give you an
idea of how reactive some of the compounds are
(Slide 12). These are neutral hydrolysis half-lives
and are pH independent. These compounds will
hydrolyze with these half-lives at pH 4, the same
as they will at pH 10. We see that the half-lives
vary anywhere from the order of seconds down to about
90 days. This does indicate, I think, the reactivity
of some of these compounds.
Now, many of the other compounds react very
fast, or may not react very fast, depending on the
pH of the water. Some of the compounds will be very
-------
505
stable at pH 5, but at pH 9 their half-lives might
well be on the order of days, and vice versa.
Compounds such as the epoxides, which are very stable
at alkaline pH's, react very fast at acidic pH's.
Does the report include data on hydrolysis for
all the compounds on the OSW list? The answer is I'm
afraid not. But we have a new branch at the Athens
Laboratory, the Measurements Branch, that's being
headed up by Mr. Bill Donaldson, that is in the
business of measuring hydrolysis rate constants
(Slide 13). Not only are they going to provide
hydrolysis rate constants, they're going to address
such things as precision and accuracy and cost-
effectiveness in attaining these numbers.
In some cases, they'll be doing product iden-
tification. They have the capability to do that.
Also, they are going to publish some standard
measurement procedures, and along with these standard
measurement procedures they will include some standard
reference compounds. I think they're going to have a
standard reference compounds that can be used to test
your measurements procedures for an acid catalyzed
reaction, a neutral hydrolysis reaction and an alkaline
hydrolysis reaction. They will then tabulate the data
and keep statistical data on accuracy and precision.
-------
506
Last but not least, they are also going to be
developing a data base. This data base will contain
what's called process reactivity constants. It will
include hydrolysis rate constants, octanol-water
partition coefficients, biodegradation rate
constants and other data that's deemed necessary.
One additional area that I wanted to bring in
is that OSW is concerned about fate and transport of
organics in groundwater. We had done some work, and
they are supporting us to do additional work on soil
mediated hydrolysis reactions. I just want to point
out the state of the art at this point (Slide 14).
In a sediment water system or a soil water system you
have two phase's, the solid phase and the aqueous
phase. We can say with a great degree of confidence
that how fast the compounds react in the aqueous
phase of this two-phase system, depends primarily on
the structure of the compound and the pH of the
water. The solid particles do not affect the
hydrolysis rate constant in the aqueous phase.
In the same two-phase system, part of the
compound is sorbed to the solid phase, and we know
now that alkaline hydrolysis is retarded in the solid
associated phase, relative to reaction in the water
phase. Neutral hydrolysis is unaltered in the solid
-------
507
phase, and acid mediated hydrolysis is actually
accelerated in the solid phase; again, when everything
is referenced back to reactivity in water. Some of
our ongoing efforts in the Chemistry Branch in the
Athens Laboratory will be taking this hypothesis and
expanding the data base and the theoretical basis for
these reactions.
Well, I'd like to summarize with this slide
(Slide 15). This is really just a summary of all the
fields in the report. As I said, there are 362
compounds on this list of which we've looked only at
the organics. There are quite a few metals. The
thing I think is important is that there are 49
compounds that have no hydrolyzable functional group,
there are 79 compounds that have no labile functional
group. So, these two groups are not going to
hydrolyze, so they present no problem in sampling, in
sample storage or in sample analysis.
In field seven, we see 85 compounds that have
hydrolysis half-lives of less than a year, and for
these compounds, their half-lives will be dependent
on pH. We see in field eight, that 85 of the
compounds have half-lives of greater than a year and
should not present any problem in sample analysis and
sample storage.
-------
508
That pretty much sums up what I have to say
about hydrolysis. I'd be glad to answer any
questions.
-------
509
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
MR. LUCAS: Sam Lucas of
Battelle. I think your first or second slide showed a
report to EPA in 1985. Can you tell us something
about the availability of that report?
MR. WOLFE: Yes. That
report is available through the Athens Laboratory at
the present time.
MR. LUCAS: From you
directly?
MR. WOLFE: Yes.
« MR. LUCAS: Thank you.
MR. FOSTER: Russ Foster
from RAI. What sort of laboratory technique was used
to prepare the samples for hydrolysis study?
MR. WOLFE: Well, maybe I
didn't make that point quite clear. For 20 or 25 of
these compounds, the values came out of the Athens
Laboratory, and some of them are literature values.
The methods of measuring are quite varied. There is
really no standard technique. All you really need to
do is be able to measure concentration as a function
of time. I think most of these compounds are done
-------
510
easier with HPLC or GLC. Hopefully, the Measurements
Branch, as it gets set up, is going to come up with
some standard procedures, some protocols that can be
used to obtain these numbers.
MR. WISE: Hugh Wise, ITD.
The appearance of your bis (chloromethyl) ether
struck me odd when I first came with the agency and
saw the priority pollutant list for the first time;
that a compound that hydrolyzes at that rate would
ever appear on a list of compounds we were to look
for in wastewater samples.
Have these other half-lives of things that
hydrolyze very rapidly been factored in to eliminating
compounds front the List of Lists and so on?
MR. TELLIARD: The List of
Lists, yes. When we put together the ITD List and
the List of Lists, we used a much smaller window. We
used 12 to 18 hours as our window, realizing that you
can't run back to the lab that fast and analyze it.
So if we aren't going to see it in 12 to 18 hours...
we care, but we don't care as much as we would.
Thank you very much, Lee. Ladies and gentlemen,
we have a break for coffee and the tinkeltorium, and
then get back in here, please. Thank you.
(WHEREUPON, a brief recess was taken.)
-------
511
Possible Mechanistic Routes for the Solvo;lysis
of ^-Bicyclo [ 2.2.0] Hex-2yl-Tosylate
'OTs
.OTs
OAc
OAc
AcO "-
7a 9a
-------
512
-------
CO
iH
m
O
<
O
oo
>-
o
zo
Ld
LJ ^r
UJ
c£
o
o_
LJ
C£
00
-------
o:
o
LJ
I—
Ld
Q_
UJ
o
UJ
a:
o
o
Q
O
-
M
Z
Ld
m
o
Ld
CM
o
Q
_
O
A
Ld
§ ">
CM
< IT
o o
Q
O
a.
Ld
Ld
-T1 o
O =!
O
Q
Ld
Ld
O i
hO o
_L_ jv;
O i-
-------
515
DEFINITIONS
HYDROLYSIS -- A REACTION IN WHICH A BOND BETWEEN TWO ATOMS OF A
MOLECULE IS CLEAVED AND A NEW BOND IS FORMED WITH THE OXYGEN ATOM
OF A WATER MOLECULE. THESE REACTIONS ARE OFTEN,, ALTHOUGH NOT
NECESSARILY,, MEDIATED BY ACID OR BASE.
HYDROLYTIC DEGRADATION -- A BROADER TERM USED TO DENOTE ANY
CHEMICAL REACTION THAT OCCURS IN WATER.
-------
516
Possible pH Rate Profiles for the
Reaction of Organic Compounds in Water
o
CD
(f)
cn
o
~5
8x10
8xlO"3
IJLJ
8x10
-1 7
•=3:
8x10
1
-------
CO
I
o
C_J
LU
h-
DC
CO
>-
O
DC
a
n:
X
o
X
O
II
_Q
O
_^:
0
0
JZ
c
o
•4—
to
c
o
o
0
TJ
0
N
D
V)
o
C
D
CO
C
O
o
0
o
V-
.to
*to
D
o
TJ
I I I
I I I
c
0
CO
c
o
o
CD
CD
TJ
0
31
O
c
o
o
c
o
to
c
0
0
-t-
0
TJ
0
£
0
CO
jQ
O
D
O
TJ
*O
D
^_
0
TJ
O
1
TJ
C
0
o
0
to
-4—
c
0
O
C
o
o
.—
c
0
0
TJ
^~
-E-
_jr-
-t—
0
C
0
TJ
0
JL
CO
H-
i_
0
to
D
JQ
5^
0
TJ
i^.
|
I
TJ
C
O
o
0
to
0
cone
c
o
• —
0
TJ
*X
O
~o
>
-------
518
Half-life is defined as the time required for the initial
concentraiton of a reactant to be reduced to one-half the
initial concentration.
The general half-life expression for a acid or base
catalyzed hydrolysis assuming pseudo-first-order reaction
kinetics is:
0.693
kH[H+]
kOR[-OH]
-------
519
KEY TO HYDROLYTIC DATA
FIELD
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
ABREVIATION
CODE
CAS NO.
RANK
COMPOUNDS
NHYF
NLFG
LESS
MORE
MES
EST
EXP
ACID
BASE
NEU
MIS
REPRESENTS
CONSTITUENT CODE
CHEMICAL ABSTRACTS REGISTRY NUMBER
SCHEDULING PRIORITY
NAMES AS RECEIVED
NO HYDROLYZABLE FUNCTIONAL GROUP
NON-LABILE FUNCTIONAL GROUP
HALF-LIFE LESS THAN A YEAR
HALF-LIFE GREATER THAN A YEAR
EXPERIMENTAL VALUES
ESTIMATED VALUES
EXPERT JUDGEMENT
ACID HYDROLYSIS RATE CONSTANT
BASE HYDROLYSIS RATE CONSTANT
NEUTRAL HYDROLYSIS RATE CONSTANT
OTHER REACTIONS
-------
Page
No. 6
520
05/22/85
'CODE
U053
U061
U063
U064
U066
IW67
U074
IM86
U089
U103
Ul'}8
UII5
111:4
U133
U13?
U151
U154
U155
U1S7
U153
CAS NO
123-73-9
50-29-3
53-70-3
189-55-9
96-12-8
106-93-4
764-41-0
1615-80-1
56-53-1
77-78-1
123-91-1
75-21-8
110-00-9
302-01-2
193-39-5
7439-97-6
67-56-1
91-90-5
56-49-5
101-14-4
RANK
1ST THIRD
1ST THIRD
1ST THIRD
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
1ST
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
THIRD
1ST THIRD
1ST THIRD
OS*
CDHP« NHYF NLFS LESS HQRE HEAS EST EXP ACID BASE NEU HIS
CROTONALDEHYDE YES 2
DDT YES YES .35 6.8E-6
DIBEHZO (A, H) ANTHRACENE YES
1,2,7,8DIBENZOPYRENE YES
l,2-DIBRO«Q-3-CHLOROPROPA YES YES
NE
ETHYLENE DIBRDHIDE YES YES 3.7E-5
l,4-DICHLORO-2-BUTENE YES ' YES 1.7E-3
N,N!-DIETHYLHYDRAZINE YES 3
DIETHYLSTILBESTRQL YES
DLIETHYL SULFATE YES YES .6
1,4-DIQXANE YES
ETHYLENE OXIDE YES YES 33.5 2.4E-3
FURAN YES
HYDRAZINE- YES 3
INDENO(1,2,3-CD)PYRENE YES
HERCURY 5
HETHANOL YES
HETHAPYRILENE YES
3-HETHYLCHOLANTHRENE YES
4,4-HETHYLENE-BIS-(2-CHLQ YES YES lE->i
ROANILINE5
U171 79-46-9 1ST THIRD 2-NITROPROPANE
YES
-------
521
COMPOUNDS UNSTABLE IN WATER
HYDROLYSIS
CHLORIDE CYANIDE
PHOSGENE
PHOSPHINE
ACETYL CHLORIDE
CARBONYL FLUORIDE
CYANOGEN BROMIDE
CHLOROMETHYL METHYL ETHER
BIS-(CHLOROMETHYL) ETHER
REDOX
METHYL ETHYL KETONE PEROXIDE
N,N'-DIETHYLHYDRAZINE
1,1-DIMETHYLHYDRAZINE
1,2-DIMETHYLHYDRAZINE
61 ,a-DIMETHYLBENZHYDROPEROXIDE
HYDRATES
7 ALDEHYDES WILL EXIST AS THE
HYDRATE IN MOST NATURAL WATER
SAMPLES.
-------
522
HALF-LIVES FOR THE NEUTRAL HYDROLYSIS
OF SELECTED COMPOUNDS
COMPOUND
BROMO ACETONE
BENZENE SULFONYL CHLORIDE
CARBONYL FLUORIDE
METHYL CHLOROCARBONATE
_3 -CHLOROMETHYL ETHER
METHYL BROMIDE
DIMETHYL SULFATE
ETHYLENE OXIDE
DIMETHYL CARBONYL CHLORIDE
BENZOTRICHLORIDE
ETHYL METHANE SULFONATE
BENZYL CHLORIDE
METHYL ISOCYANATE
BENZAL CHLORIDE
1,3-DICHLOROPRENE
2>2'-BIOXIRANE
MALEIC ANHYDRIDE
PHTHALIC ANHYDRIDE
HALF-LIFE
4 MIN
35 MIN
24 SEC
20 DAYS
1 HR
12 DAYS
4 MIN
3 MIN
19 HR
9 HR
8 MIN
7 MIN
11 DAYS
29 DAYS
26 SEC
3 MIN
-------
CO
(M
CO
CO
_J
o<
Q -o
in
o
DL. JH
O uj
~" — * ~X^
UJ —
""*"*>»> \/
Ld
Q
m
o
z:
DD
(/)
h-
z:
Ld
Ld
ID
CO
Ld
CO
h-
<
h~
CO
O
o
Ld
H-
<£
CE
Ld
Q
^>
0
Q_
Z
O
h-
o
Lu
h-
z
Ld
Q
h-
O
ID
Q
O
(y
Q_
Ul
Ld
Q
• *
Ld
O
O
Q_
H-
z:
Ld
Ld
C£
CO
-------
524
SOIL-WATER MEDIATED HYDROLYSIS REACTIONS
IN WATER
HOW FAST THE COMPOUNDS REACT DEPENDS ON THE
STRUCTURE OF THE COMPOUND AND THE pH OF THE WATER
IN SOILS
ALKALINE HYDROLYSIS IS RETARDED IN THE SOLID
ASSOCIATED PHASE
NEUTRAL HYDROLYSIS IS UNALTERED IN THE SOLID
ASSOCIATED PHASE
ACID MEDIATED HYDROLYSIS IS ACCELERATED IN THE
SOLID ASSOCIATED PHASE
WHEN REFERENCED BACK TO REACTIVITY IN WATER
KINETIC STUDIES FOR 10 ADDITIONAL COMPOUNDS WILL
CARRIED OUT TO FURTHER SUPPORT THIS MODEL
THEORETICAL BASIS
CONSISTENT WITH SIMILAR SYSTEMS
TRANSITION STATE CHEMISTRY
-------
525
Field
1 CONST CODE
2 CAS NO
3 RANK
4 COMPOUND
5 NHYF
6 NLFG
7 LESS
8 MORE
9 MEASURED
10 ESTIMATED
11 EXPERT
12 ACID HYDRO
13 BASE HYDRO
14 NEUT HYDRO.
15 MIS
1
2
3
4
25
8
10
9
Count
362
359
362
362
49
79
85
85
44
9
115
15
58
64
50
-------
526
MR. TELLIARD: We'd like to
get the session started.
Our next speaker is Ted Martin from our Cincinnati
lab/ who is going to talk a little about some metals
analysis, and then followed by Ray Maddalone from
TRW, who is also going to be talking about the EPRI
metals study that has been going on for the last
couple of years on the round-robin for the Edison
Electric Institute. Ted.
-------
527
THEODORE D. MARTIN
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
EVALUATION OF METHOD 200.1 DETERMINATION OF
ACID SOLUBLE METALS
(ABRIDGED REPORT)
INTRODUCTION
This report describes a practical and controlled evaluation of the
"acid-soluble" metal methodology, Method 200.1 DETERMINATION OF ACID SOLUBLE
METALS. The method is basically a sample extraction/preparation procedure
with reference to the determinative step being the 200 series methods,
Method for Metals - Methods for Chemical Analysis of Hater and Wastes,
EPA-600/4-79-020, March 1983.For the purposes of this report the
discussion of Method 200.1 is limited almost entirely to the different
aspects of sample preservation and extraction.
The term "acid soluble" metal, along with the concentration requirements
as related to water quality criteria as a possible enforceable standard
pursuant to Section 304 (a) (1) of the Clean Water Act [33 U.S.C.
1314 (a) (1)] is given in the Federal Register (FR), July 29, 1985 [50 FR
30784]. Included in this FR reference is the National Technical Information
Service (NTIS) listing of publication order numbers for the criteria
documents for each of the six metals of concern: arsenic (As), cadmium
(Cd), chromium (Cr), copper (Cu), lead (Pb), and mercury (Hg). Four
additional metals: aluminum (Al), nickel (Ni), silver (Ag), and zinc (Zn)
were evaluated in this report since regulations concerning them are being
considered. ,
This FR states that for ambient water the Agency believes that the
measurement of "acid-soluble" metal will provide a more scientifically
correct basis upon which to establish criteria. The latest criteria [50 FR
30784] has been developed on this basis. The concept of "acid-soluble"
metal is a measurement believed to be less rigorous than "total" or
"total-recoverable" metal analyses. Although it is only a partial
determination of the metal content in a sample, it is intended to provide an
estimate of the available portion associated with suspended and particulate
material in addition to the dissolved metal fraction. It should be noted
that for those metals with stable oxidation states of reported different
toxicity, the Agency has included required measurement of "acid-soluble"
metal speciation as a part of the water quality criteria. At present only
arsenic and chromium are affected, but other metals may be added at a later
date. Also, the FR states that this methodology should be applicable to
analysis of saltwater matrix as well as freshwater. The present version
-------
528
(1.4) of Method 200.1 does not attempt to provide for these two specific
needs, but is a generalized approach and first offering methodology to the
analysis of "acid-soluble" metals. The purpose of this report is to discuss
the ruggedness of Method 200.1 and provide a standard method that may be
used to evaluate the toxicity of "acid-soluble" metals to aquatic life. A
final draft copy of the proposed method with acknowledgment given to
reviewers and others who have provided significant contributions is
available from the Environmental Monitoring and Support Laboratory -
Cincinnati, Ohio 45268..
CONCLUSIONS
The findings of this evaluation support the opinion that the measurement
of "acid-soluble" metal is less rigorous than other established methods used
to measure metal concentration in multi-phase samples. Since the
measurement includes both the dissolved metal fraction as well as an
extracted portion of metal from the particulate material, it is by
definition a new parameter and is affected by the operational technique
used. The concentration of metal extracted will vary with pH and is not
uniform between metals. An example is an 81 percent increase in lead
concentration at pH 1.5 over that extracted at pH 2.0, while for cadmium
there is less than a 7 percent increase for the same change in pH. The
technique is precise when the sample is acidified to a specific pH range of
narrow limits (1.75 ± 0.1) and remains at that pH during the extraction
period. The pH selected for extraction of lead and chromium is most
critical. The concentration of metal extracted also is affected by the
holding period after pH adjustment. However, with the exception of chromium
the evaluation data indicates the rate of extraction at pH 1.75 and 22°C
decreases considerably beyond 96 hours (4 days) and there is usually less
than a 20 percent increase in concentration between 16 and 96 hours of
extraction. Although the amount of particulate material in the sample, up
to 2.5 grams/liter, did not appear to affect extraction efficiency, other
factors that do affect the concentration measured are acidification at the
time of sampleocollection to preserve the dissolved metal fraction, and
transport at 4°C to reduce chemical activity. Of the ten metals evaluated
the data indicate that the solution chemistry of silver and mercury are not
suited to "acid-soluble" metal measurement.
RECOMMENDATIONS
From the compiled data and observations made during this evaluation of
Method 200.1, certain changes are suggested and further work recommended
before Method 200.1 is adopted as acceptable methodology for freshwater
"acid-soluble" metals analyses. These recommendations are as follows:
1. Continue to analyze silver and mercury as "total" or "total recoverable"
metals and set limits on that basis. The vapor pressure of
organic-mercury compounds will most probably result in partial losses
during filtration under vacuum. Also, silver chloride and other
compounds of silver and mercury are relatively insoluble in the
-------
529
specified acid conditions, subject to the dynamics of sample matrix and
in some cases either lost or not recovered as a result of using the
"acid-soluble" criteria.
2. Develop an improved method for mercury analysis with a lower detection
limit. The water quality criteria limit set for "acid-soluble" mercury
is more than an order of magnitude below the detection limit of the
present cold vapor mercury methodology.
3. Evaluate and verify that interlaboratory precision and percent recovery
data for "acid-soluble" graphite atomization atomic absorption analyses
will meet the criteria limits listed in Table 1.
4. Verify that the determinations for "acid-soluble" metals analyses as
described in Method 200.1, Version 1.4 can be experimentally correlated
to known aquatic toxicity.
5. Develop separation methods for the analysis of stable oxidation states
of arsenic, chromium and selenium and the organometallic species of
arsenic, mercury, selenium and tin as required.
EXPERIMENTAL DESIGN
In planning the experimental work for the evaluation of Method 200.1,
the allowable concentration given in 50 FR 30784 was considered. For most
metals the concentration limit for both freshwater and saltwater is very low
and a level most suited to analysis by graphite atomization atomic
absorption. Also, the allowable "acid-soluble" concentration of cadmium
(Cd), trivalent chromium (Cr+3), copper (Cu), and lead (Pb) in freshwater
vary with the determined concentrations of hardness as calcium carbonate.
Listed in Table 1 are examples of the expected limits for different types of
water. For the testing and evaluation of the procedure, an assumption was
made that the extraction characteristics of these metals would be similar at
higher concentrations as well as at the expected limits. This assumption
permitted the use of simultaneous inductively coupled-plasma atomic emission
spectrometry (ICP) in the determinative step.
The sample used for the evaluation was a composited sample of Ohio river
water spiked with particulate material and an acidfied-aqueous spike of
silver (Ag), arsenic (As) and mercury (Hg). The particulate material was
uniform and homogeneously mixed with a moisture content of approximately 50
percent. The particulate material, taken from an industrial sludge stock,
was added to the aliquot of river water to a concentration of 5
grams/liter. The addition of 5 grams/liter yielded a test solution
containing 2.5 grams per liter of particulate material. All test solutions
were prepared in this manner, except those used for the generation of
precision and percent recovery data. In this case the sample solutions were
prepared at various levels of particulate concentration.
-------
530
TABLE 1. EXAMPLES OF ACID SOLUBLE METAL CRITERIA LIMITS*
Average Concentration, pg/L
Metal
Arsenic"1"3
Arsenic"1"5
Cadmi urn
50 mg CaCOs/L
100 mg CaC03/L
200 mg CaCOa/L
Chromium"1"3
50 mg CaC03/L
100 mg CaCOs/L
200 mg CaCOs/L
Chromium*6
Copper
50 mg CaC03/L
100 mg CaC03/L
200 mg CaCOs/L
Lead
50 mg CaCo3/L
100 mg CaCOs/L
200 mg CaC03/L
Mercury
Freshwater
4 Day
190
••^
0.66
1.1
2.0
120
210
370
11
•_»
6.5
12
21
wm —
1.3
3.2
7.7
0.012
1 Hr
360
__
1.8
3.9
8.6
980
1700
3100
16
__
9.2
18
34
vimrM
34
83
200
2.4
Saltwater
4 Day 1 Hr
36 69
9.3 43
— __
_ _ __
— —
__ __
__ «._
— —
50 1100
2.9
_« __
— —
5.6 140
M_
_— __
0.025 2.1
* FR, Volume 50, Number 145, July 29, 1985
-------
531
Before using the sludge particulate material in this study, it was
analyzed by ICP in the EMSL-Cincinnati laboratory and by the USEPA Region 4,
Analytical Support Branch. In the Cincinnati Laboratory, a 2 gram aliquot
of wet sludge was prepared for analysis using a combination nitric-
hydrochloric acid reflux, while the Region 4 laboratory used a 1 gram
aliquot of dried sludge and the nitric acid-hydrogen peroxide digestion
(Method 3050) given in SW-846, Test Methods for Evaluating Solid Waste. The
results, reported on a dry weight basis of those determinations constitute a
"total" analysis and the comparative data are given in Table 2. It should
be noted that over the period of time needed to complete the evaluation, the
stock sludge with exposure to the atmosphere from repeated weighings lost
moisture causing a gradual increase in metal concentration on a wet weight
basis. To validate the precision and percent recovery data at the time of
analysis, four additional replicate aliquots were analyzed just prior tothat
aspect of the evaluation. These concentrations are reported as "total"
values in Table 15 under "total acid reflux" and expressed in mg/L for
comparison to the "total recoverable" data.
Since "acid-soluble" metal analysis includes both the dissolved metal
fraction as well as an acid extracted portion of metal from the particulate
material, it is by definition a new parameter and affected by the
operational technique used. Although it is the intent that the measurement
be less rigorous than "total recoverable" analysis, it must be rugged for
the determination to be meaningful. The following list of variables were
considered and tests were conducted to determine the significance of each:
1. The conditions of filtration including type of filter material, cleaning
procedure, prefiltration, and filter prewash requirements;
2. The effect of varying pH and whether allowable range limits would be
required around the selected pH;
3. The effort of varying the time period of extraction following
acidification;
4. The need for acid preservation at the time of sample collection and
whether specified transport conditions to the laboratory would be
required;
5. The effect varying amounts of particulate material would have on
extraction efficiency.
The test solutions for each experimental phase were prepared in acid
cleaned cubitainers. The amount of particulate material needed was weighed
into the cubitainers and the appropriate volume of river water added. If a
soluble spike was to be added to the solution, it was done immediately prior
to acidification. The volume of (1+1) nitric acid added to each type of
solution for pH adjustment was predetermined and verified after addition.
Following extraction and filtration the pH was again measured to verify that
the pH was maintained and constant throughout the extraction period.
-------
532
TABLE 2. INDUSTRIAL SLUDGE DATA - ACID REFLUX TOTAL ANALYSES
Metal
Ag
Al
As
Cd
Cr
Cu
Hg
Ni
Pb
Zn
Concentration, wg/gram*
U.S. EPA Region 4 EMSL - Cincinnati
Anal. Support Branch Phys. & Chem. Methods Branch
6
3800
19(a)
58
7000
840
3(b)
360
450(c)
3400
10
3300
21
67
7000
920
2
370
530
3600
(a) Determined by graphite furnace atomic absorption.
(b) Determined by cold vapor atomic absorption.
(c) Determined by flame atomic absorption.
*Reported on a dry weight basis.
-------
533
The concentrations of all spiking solutions and laboratory control
standards used in this evaluation were verified with a quality control check
sample obtainted from the EMSL-Cincinnati, Quality Assurance Branch. In
addition to control standards, method blanks, and river water control
samples were prepared and processed with the test solutions.
In the final phase of the evaluation, precision, and percent recovery
data were generated by varying the amount of particulate material in the
test solution. In addition to the river water control, seven replicate
solutions were prepared for each of three different levels of particulate
material concentration. To calculate percent recovery, the average river
water control value was subtracted from the average test solution
concentration and compared to the expected concentration as determined from
the "total recoverable" analyses of the same samples.
RESULTS AND DISCUSSION
Conditions of Filtration
In the first phase of the evaluation various types of filters were
examined. This examination included both the 0.45 vm filters and the
prefilters that were considered necessary to the procedure. It was
important to determine that the specified filtering material could withstand
the mild acid (0.2% v/v HNOs) used for extraction and not contaminate the
samp]le with trace impurities. To determine if filter contamination would
be a problem, each filter was evaluated separately. Also, to check the
resistance to acid, a more concentrated acid blank (1..535 v/v HNOs) was
used to extract the filter. A 10 ml aliquot of the acid blank was recycled
through the filter three times while allowing one minute of contact with the
filter each time before vacuum was applied. This aliquot was then analyzed
by ICP to determine if any metals of interest were extracted. The types of
filters tested were the following:
Course Prefilters
1. Gelman glass fiber filter: Type AE
2. Gelman 5 ym PVC membrance: Type VN-1
Fine Prefilter
1. Gelman PVC arcylic copolymer: Type DM-800
Fine Filters
1. Mi 11ipore mixed esters of cellulose: Type HAWP-047
2. Gelman PVC acrylic copolymer: Type DM-450
3. Gelman PP/PTFE (Teflon): Type TF-450
The filtering apparatus used throughout this evaluation was a
polysulfone Gelman 47 mm Magnetic Filter Funnel #4201 and 500 ml suction
flask. The magnetic lock between the funnel housing and base was most
-------
534
convenient and easy to use. The manufacturer notes that polysulfone is
resistant to nitric and hydrochloric acids, but not to chromic nor sulfuric
acids. Cleaning of the funnel, as with the suction flask, was effectively
accomplished using a detergent wash, rinsing with water followed by a dilute
nitric acid rince and copious amounts of deionized distilled water. Also,
rinsing the apparatus with copious amounts of deonized distilled water
between samples appears to be sufficient cleansing to avoid cross
contamination as long as a previous sample does not contain an oily phase.
A key factor in keeping contamination to a minimum was to never allow the
filtering apparatus or receiving labware to dry without being properly
rinsed or cleaned before reuse. Since the rubber stopped on the funnel stem
comes in contact with the suction flask over which the filtrate is poured,
to eliminate contamination, it was wrapped 1 inch PTFE laboratory tape.
Of the fine filters (0.45 ym) tested only the Gelman PP/PTFE type proved
unacceptable. This filter would not wet with sample contact alone, but
required initial wetting with methanol before aqueous filtration would
occur. Since adding an extra step to the procedure seemed impractical, the
use of the teflon filter is not recommended.
The other two types of fine filters, the PVC acrylic copolymer and the
mixed esters of cellulose, both proved acceptable. The cellulose filter was
able to withstand the acid because of the relatively dilute acid used and
limited contact time. The extract analyses from both type of filters gave
values near the acid blank and below the instrumental detection limits.
Although the PVC acrylic copolymer fine prefilter (0.8 ym) also proved
acceptable giving similar analytical results to that of a 0.45 pm filters,
the same was not true for the coarse prefilter. The extract analysis of the
PVC type VM-1 prefilter indicated that significant contamination of aluminum
(Al) and zinc (Zn) may occur. Concentrations of approximately 0.5 mg/L of
both elements were found in the extract. Also, over 0.3 mg/L of Al was
detected in the extract of the glass fiber type AE filter. Since Al and Zn
may eventually be added to the list of "acid-soluble" metals, the need for a
coarse prefilter was reconsidered.
In the preliminary work of this evaluation the three filter system was
observed to be very bulky and in some cases did not allow a good seal
between the funnel housing and base. This resulted in leaks and affected
the rate of filtration. The actual need for coarse prefiltration to
eliminate clogging of the 0.45 ym filter is most probably minimal. On this
basis and because possible contamination and problems of leaking during
filtration, the use of the coarse prefilter was removed from the procedure.
To check the possibility that metals may be leached from the cubitainer
during transport, seven cleaned cubitainers were subjected to a five percent
nitric acid leach for 28 days. At the end of the period, the acidified
deionized distilled water extract was concentrated by evaporation and
analyzed. These results verified that any metals that may be leached from
the cubitainer were at a concentration insignificant to ICP analyses. The
determined concentration of all metals in the leaching solution were below
the ICP instrumental detection limits.
-------
535
Although the combination 0.8 ym prefilter and 0.45 pin fine filter
revealed no contamination from the filtering material, a 50 ml sample
prewash of the filtering apparatus was used to remove traces of the
deionized water rinse and to condition the membrane filters. The
combination of the 0.8 wm prefilter with either of the 0.45 pm filters was
equally effective in filtering the prepared samples used in this
evaluation. The time of filtration for an acidified sample following sample
prewash was less than 2 minutes for a 200 ml aliquot decanted from a sample
that had settled no longer than one-half hour.
Effect of Varying pH
Once the various aspects of the filtering process were determined, the
effect of varying pH was investigated. For determining the effect of pH on
the extraction of "acid-soluble" metals, six levels of hydrogen ion
concentration (pH 1.0, 1.5, 1.75, 2.0, 3.3 and 7.0) were selected. The wide
pH range and points on either side of 1.5 and 2.0 were selected in response
to concerns for investigating the least rigorous extraction while providing
a practical and rugged method. The intent of those researchers who
suggested the concept of "acid-soluble" metals was to provide a measurement
where the sample would be acidified to a low enough pH for the sufficient
time to dissolve the carbonates, hydroxides and metal precipitates without
leaching or dissolving substantial quantities of the metals occluded in
minerals, clays and sand or strongly sorbed to particulate material.
Originally pH 4 was selected for extraction of "acid-soluble" metals,
however, it was changed to pH range 1.5 to 2.0 because of the buffering
capacity of the carbonate system of ambient waters. The adjustment of
samples to the higher pH of 4, in practice, is more time consuming and
difficult to accomplish since this is the lower boundary of the
carbonate-bicarbonate buffering system..
The concentrations of "acid-soluble" metals extracted at each selected
pH level are given in Table 3. Four replicate samples were prepared for
each pH. Each set of samples was prepared in the same manner by spiking
river water with particulate material (5 grams/liter) and an acidified-
aqueous spike of Ag, As and Hg to total concentrations of 0.050, 0.15 and
0.1 mg/L, respectively. All samples were mixed and allowed to extract for
16 hours at 22°C. In reviewing the data, it is apparent that precision
between replicates is very good with the highest relative standard deviation
at concentrations greater than 5x the instrumental detection limit being 6.5
percent for Al at pH 1.5. Also, it is obvious from the data that varying
the pH produces trends in the measured concentration of all metals with some
being more drastic than others. This effect is best appreciated from the
graphic illustration of Table 6 data shown in Figures 1 through 4.
-------
in
CO
tvj
CM
CM
If) in
10 O
• •
CM O
•H
co o
• •
CM O
in
co oo
CM O
•H
Ch
CO O
• *
i-H O
•H
LO Cn
If) O
O O
o
0
CM
CM
Cn LD
to i-H
r-. o
CM CO
r^ CM
LO O
•f
Cn CO
LO i-H
LO CM
i-l CM
CO O
i-H CO
in o
o o
0
O I
re
00
LO
LO CO
«d- O
en o
• •
o o
i-H CM
en o
o o
CM LO
O i-l
CO O
• •
O O
-H
in co
CM CM
CO O
• •
o o
r-. in
o o
LD o
• •
o o
CO «3-
r*- o
o o
o o
LO «d-
CO O
o o
• •
o o
•H
CM CO
CO O
O O
• •
o o
•H
cn co
CO O
o o
• •
o o
-H
LO CM
co o
o o
• •
o o
-H
(Ft CO
LO O
o o
• •
o o
•H
in in
CO O
o o
• •
o o
o
UD
CJ1
O t-H
co o
• •
CM O
•H
CD
•* 1^
CM O
• •
CM O
•H
o o
• •
CM o
•H
CM
cn 10
O O
• •
CM O
-H
CO
i-H CO
LO O
CM CM
CM O
O O
• •
O O
•H
Q.
in
CO t-l
•»
CO
o t->.
O)
UJ
CO
00
in
. to ^j-
r- in o
°
LO
«* CM
cn CM
• •
CO O
LO cn
LO o
co
0
in o
co
LO ^~
o o
CM
co
UJ
cr> 1-4
o o
o o
o o
o o
LO i-l
o o
o o
CO t-H
o o
O O
CO i-H
o o
O O
• •
o o
-o
at
4-»
cu
(O
00
in
(O
a>
00
00
(O
3: a>
00
-------
537
ui
CJ
1.75
SAMPLE EXTRACTION, pH
FIGURE 1. fletal Extraction as a Function of Saraole oH -
Aluminum (Al), Zinc (Zn) and Copper '
-------
538
i
UJ
o
I 0.4
7.0
FIGURE 2.
5.0
3.3
2.0 I 1.5 1.0
1.75
SAMPLE EXTRACTION, pH
^a Function of Sample oH -
(Nv), Lead (Pb) and Cadmium (Cd) '
-------
539
o
20.0
18.0.
16.0
14.0
12.0
10.0
LU
O
5
o
8.0
6.0
4.0
2.0
7.0
5.0
—i »—i—i 1
3.3 2.0 I 1.5 1.0
1.75
SAMPLE EXTRACTION, pH
FIGURE 3. Metal Extraction as a Function of Sample pH
Chromium (Cr)
-------
540
UJ
1
o
0.20
0.18
0.16-
0.14
0.12
0.10
0.08
0.06
0.04
0.02-
7.0
5.0
: M'.•.,!
3.3
SAMPLE EXTRACTION, pH
o As
a
2.0 I 1.5 1.0
1.75
FIGURE 4. Stability of Soluble Spike as a Function of Sample oH-
Arsenic (As), Mercury (Hg) and Silver (Ag)
-------
541
In Figure 1 the extraction trend exhibited for the Cu and Zn are very
similar. Concentration increases as the pH is lowered, but for pH 2 and
below there is little change in the amount of metal extracted. While for
Al, the other metal in Figure 1, the concentration extracted continues to
increase as the pH is lowered, but still only amounts to a 5 percent
increase in concentration between pH 2.0 and 1.5.
In Figure 2, Cd displays the least effect from pH change while nickel
(Ni) shows a 10 percent increase in concentration between pH 2.0 and 1.5.
However, when the change in Ni concentration is compared to that of Pb, it
is relatively minor. The increase in Pb concentration extracted between pH
2.0 and 1.5 is 81 percent. Of all the metals investigated the extraction of
Pb is most affected by slight changes in acid below pH 3.
In Figure 3, the effect on Cr extraction is illustrated by itself
because of the high Cr concentration in the prepared sample. As can be
seen, the effect of varying pH on Cr is also extreme being second only to
Pb. The change in Cr concentration extracted between pH 2.0 arid 1.5 is 41
percent.
In Figure 4 the three metals As, Hg and Ag are displayed. For As there
is a slight increase in concentration, but since the As concentration in the
particulate material is very low, a conclusive statement on extraction can
not be made. However, it is apparent that the As soluble spike is not lost
as is that of Hg. For Hg, as the pH is lowered, its solubility decreases.
Speculation on this occurrence is that the lowering of pH solubilized a
matrix constituent of the particulate material which reacted with the
soluble Hg spike forming insoluble mercury compound. The loss of the Ag
spike is attributed to the precipitation of silver chloride. For silver,
there is a slight increase in concentration as pH is lowered, but less than
20% of the possible 0.05 mg/L is solubilized at pH'1.0. It should be noted
from the data given in Table 15 that both the Hg and Ag spikes are recovered
using the "total recoverable" analyses.
Since both Pb and Cr demonstrate extreme changes in the concentration of
metal extracted for slight changes in pH, more exacting pH limits were
incorporated into version 1.3 of Method 200.1. The midpoint (pH 1.75) of
the "acid-soluble" pH range (1.5 to 2.0) was selected with a required
tolerance limit of ±0.1 pH units. The midpoint was selected because the
dissolved metals can be acid preserved by eliminating the carbonate
buffering capacity of ambient waters with little likelihood that the sample
pH will actually reach or go below 1.75. Also, this pH provided less
rigorous data while also being rugged.
Effect of Extraction Time
Having decided on pH 1.75 ±0.1 for the extraction of "acid-soluble"
metals, the length of contact or extraction time was investigated. Samples
were prepared as described earlier, adjusted to pH 1.75, mixed, and held at
22°C for various periods of time (1, 4, 16, 96, 168 and 600 hours). The
-------
542
concentrations of "acid-soluble" metals extracted for each time period are
given in Table 4 and illustrated in Figures 5 through 8.
With the exception of Ag and Hg, the data in Table 4 indicate a gradual
increase in concentration with increased extraction time. For example there
is for all metals, except Cr, less than a 10 percent increase in
concentration between the 4 and 16 hour extraction time. Also, only Al, Cr
and Zn show an approximate 20 percent increase in concentration between 16
and 96 hour (4 day) extractions, and except for Cr and Zn, there is little
increase in concentration beyond four days. From these findings and for the
convenience of receiving samples and processing them in the laboratory, a 16
hour extraction has been included in Method 200.1.
Since extended holding of the sample after collection and acid
preservation may increase the concentration of "acid-soluble" metals, a
limit of 3 days for sample collection and transport at 4°C, has been
incorporated into the method. It is assumed that once the sample is
received into the laboratory and equilibrated to room temperature,
processing, including pH adjustment and the 16 hour extraction period will
begin. The data in Table 4 should be evidence that a time period of two to
three days needed for sample collection and transport should not greatly
affect an "acid-soluble" metal determination using Method 200.1.
Sample Preservation and Transport
In the present USEPA metal methods, except for Cr+6, all samples are
acid preserved at the time of collection and shipped to the laboratory in
the most convenient manner. Actual conditions of transport and shipment
time are not considered critical since the metals are either already in
solution or will be subjected to an acid digestion or solubilization prior
to analysis. However, in determining "acid-soluble" metals sample
preservation and transport conditions should be of prime concern, since the
dissolved fraction must be preserved while the extraction of metals from the
particulate material must be minimized during transport. These two
procedural aspects of the method are the least controllable variables.
To determine the significance of both acid preservation and sample
transport conditions on the determination of "acid-soluble" metals, a series
of tests were devised in an attempt to reflect the various types of
environmental conditions that might occur. To simulate transport, the
samples were divided into three groups and held in storage for three days,
each group at a different level of temperature (4°C, 22°C and 49°C). To
simulate acid preservation, half of the samples were acidified to pH 1.75
before being placed in storage, while the other half were maintained at a
neutral pH until the start of sample processing. In these tests, following
mock transport, the extraction period used was 16 hours at pH 1.75 and
22 C. (Note: It should be realized when reviewing the data from these
tests that the acidified samples actually experienced the equivalent of a 4
day or 96 hour extraction).
-------
3
^ LO
3 O LO
O
o
CM
CM
| —
LO
[•-.
rH
n:
D-
1
UJ
i — i
I—
Z.
O
1 —
o
I—
X
UJ
CD
z.
1— )
=£
>
1—
HH
3
oo
i
=i
UJ
UJ
i
CO
oo
f — \
H-l
o
•
«=!•
1 1 1
1
CO
1 —
CM O
c • •
NJ CM O
O CM
CM rH
t~\ ^- CO
Q_ • .
O O
O LO
LO CM
y • •
O O
•H
rH CM
CO O
Crt O O
O O
•H
en rH
E CM r^
3 en o
« o • •
I "°
•p
re
S- CM
.p «* LO
C S- LO rH
co O • •
o •* o
!= -H
o
o
r^ LO
CM O
-O rH O
o • •
t-x LO
rH O
tO rH O
°'°*
en
«3- CM
1 — CM rH
^c * •
oo o
-H
o o
rH O
Oil O O
^f 1 * *
O O
-H
c •
o >
•i- CO
O CO S- C •
re E 3 re -o
S- •!- O CO -P
-PI — n: s: oo
X
UJ rH
CO rH O
«3- LO r^ LO LOLO r^ LO OLO
OOO OOO COO COO rHrH
CMO CMO CMO CMO "3- O
-H -H •« £
to
CO
cnvo cnco COCM «* co LOCO re
CMLO LOrH ^- LO o-* cnoo o
^- o «* o «* o LOO . ^- o, . -r~
OO OO OO OOOO Q-
•« •" •" -i
CO
rHrH CMLO OCn l^f-s COOO CO
CnOO OrH CTlCM Cr>rH OOOO S
i^ o coo coo coo en o -t->
. . • . • . • . •• CO
OO OO OO OO OO -Q
•H -H -H <„
3
re
Looocnoo • • • >
LO O OO O Q O Q
oooo • • • . S>
.. • • z. \ z. \ z. \ en
oooo z
•H -H c? .
OO » ^- -stCM LOCO El
OOO OOO ^O «*O LOO -i— LO
rHO rHO rHO rHO rHO _JCM
OO OO OO OO OO C~O
•H -H -H ° *=
•r- re •
-l-> C
o >> o
co re -i—
LD LO CM «* - — • LO OO rH rH -P "O -(->
OO rHO CM OO rHO COlre
rH O rH O *— ' rH O rH O C2 *3- -t-
• • •• | •• •• >
OO OO OO OO i — S-CO
•fl -H -H « 0 T3
c -a o>
CO "OS- r-
CT QJ fQ f~}
en rH LO 3 s- TS re
CMcn Locn OCPI LOCTI oorH s_rec'—
LO i— i LOO «*o LOO OrH -pa.re-r-
• • •• • • • • •• toco-pre
OOO OOO ^l"O «!l" O LOO Ci-tO>
•H -tt • •« i— i o- re
M—
V to O -P
CO 0
OrH LOrH .'• « T3" — COC
rHO OO Q, Q O CUQ-O
oo oo • • "•• • -pErere
.. • . z' I 2:1 2:1 o re i— 4->
oooo cotocxre
m •« -P _ -o
co o c:
*-. *—. Q 3 ••- 0
. — . • -— » --^ • rH tO +J •!-
• +J > i— 1 . — . tO > >, +-> C C
> jz & — • vt to>>cotoreco o>>coco
co tocnQ to>) co s-re Q s-Qccn z i — > to
»_-co o Q •— » --^
LOO LO LO O «rHCM
^J. rH^^ CTl rH LO Z. •
-------
544
UJ
M.
O
7.0.
6.5-
6.0 -
5.5-
5.0
4.5
4.0
3.54
3.0
•
2.5
2.0
1.5.
1.0 •
0.5
0
1116
4
96
168
600
EXTRACTION HOLDING TIME, HOURS
FIGURE 5. Metal Extraction as a Function of Sample Holding Time
Aluminum (Al), Zinc (Zn) and Copper (Cu)
-------
545
O
i—i
§
UJ
o
O
o
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Pb
-©
l] 16 96 168~
600
EXTRACTION HOLDING TIME, HOURS
FIGURE 6. Metal Extraction as a Function of Sample Holding Time
Nickel (Ni), Lead (Pb) and Cadmium (Cd)
-------
546
UJ
o
6QO
EXTRACTION HOLDING TIME, HOURS
FIGURE 7. Hetal Extraction as a Function of Saroole Holding Time
Chromium (Cr) •
-------
547
^
^
o
»— 1
t:
CONCENT*/
0.26
0.24
0.22
0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0
0
© ©
Hg o
\
\
'A'-^V, T
96
168
600
EXTRACTION HOLDING TIHE, HOURS
FIGURE 8. Stability of Soluble Spike as a Function of Saronle Holding Time
Arsenic (As), Mercury (Hg) and Silver (Ag) '
-------
548
To test the stability of the dissolved metal fraction under the
described conditions, one set of samples was prepared by spiking Ohio river
water with a soluble spike containing all ten metals. To test the
extraction from particulate material alone, a second set of samples was
prepared where river water was spiked with weighed aliquots of the sludge.
To test the two fractions together, a third set of samples was prepared
where the river water was spiked to contain both the soluble spike and
sludge particulate material. In each set, duplicate samples were prepared
for each condition described, and as in previous experiments, the
particulate material was spiked into river water to a concentration of 5
grams/liter.
Because the data from these tests are extensive and are best reviewed by
comparing the results from each of the selected conditions, the data for
each metal are given in individual tables. Included in the "acid-soluble"
data of the prepared samples are the analyses of a river water control and
laboratory control standard. Also included in the tables for comparison are
the results of direct analysis of the spiking solution diluted in two
different acid matrices, as well as the "total recoverable" analyses of the
laboratory control standard and the river water containing the soluble
spike. The mean value listed for each analysis is the average of the
duplicates with the range value being their difference. The data for all
ten metals are contained in Tables 5 through 14. Each metal will be
discussed in the same order as the tables.
Aluminum - Table 5.
The analyses data of the laboratory control standard show very good
agreement between the "acid-soluble" and "total recoverable" analyses. The
data also agree with the spiking solution which gave the same analyzed value
in both acid matrices.
As might be expected, the "acid-soluble" data for the acid preserved
samples gave higher extracted concentrations than the non-preserved
samples. In both groups, the range values appear to be less than 10 percent
of the mean indicating relatively good precision with the acid preserved
samples showing somewhat better agreement.
The acid preserved samples stored at 49°C yielded considerably higher
extracted concentrations than those stored at 4°C. Increases in
concentration ranged from 20 to 50 percent. Since there is a similar
increase in all three types of sample-spike combinations, the increase
appears to be due to Al extracted from the river water. A comparison of the
river water control to total recoverable analyses of the river water plus
soluble spike support this statement, since only 20 percent of the Al in
river water was made available using Method 200.1. In the non-preserved
samples, the Al concentration showed aoslight decrease for those samples
stored at the elevated temperature (49°C).
For Al the need for acid preservation is not confirmed, but sample
transport and storage below 22°C is required until the time of sample
processing. This is most practically accomplished by cooling with ice to
4°C.
-------
549
TABLE 5. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - ALUMINUM
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1
Diluted in 1.0* (v/v) HN03 + 5%
mg/L
Acid Preserved Not Acid Preserved
Mean Range Mean Range
0.406
2.08
2.17
2.55
4.23
6.23
6.60
8.25
6.70
7.02
8.45
-0.410
10.2
10.8
.5)
(v/v) HC1
0.008
0.02
0.04
0.01
0.19
0.03
0.09
0.03
0.08
0.24
0.14
0.003
0.05
0.81
Value
0.400
0.404
0.403
_ —
1.58
1.89
1.87
1.64
5.42
5.32
5.21
5.96
5.71
5.38
8.51
11.6
0.02
0.04
0.04
0.02
0.16
0.16
0.44
0.19
0.17
0.23
._
0.06
0.03
-------
550
Arsenic - Table 6
From prior acid-reflux analyses the determined As concentration in the
sludge is very low. In fact, if all were extracted by the "acid-soluble"
procedure, the concentration would be only 0.056 mg/L. From the data given
in Table 6 it would appear that 50 percent of the As is extracted, but the
poor precision between duplicates and low concentration of the analysis does
not permit a conclusive statement.
The highest recovery of the soluble As spike is only 87 percent. Acid
preservation does not appear to be necessary because it does not stabilize
the dissolved fraction, since storage at the elevated temperature (49°C) in
both situations indicates additional losses. This consistently low recovery
cannot be explained, but the "total recoverable" analyses indicate the loss
to be as a precipitate and not a loss to the walls of the cubitainer.
For As the need for acid preservation is not confirmed, but sample
transport and storage below 22°C is required to reduce chemical activity and
loss of the dissolved fraction.
An important note to the As analysis is the 75 percent recovery of the
laboratory control standard. During evaluation the As analysis of the
control standard was erratic. This behavior was found to be a phenomenon of
the ICP analysis and was attributed to the difference in the concentration
of the acid matrix between the extracting acid solution and the calibration
standard. (See the direct analysis of the spike solution.) It is assumed
this problem did not occur in prepared river samples because of the
naturally occurring chloride and higher dissolved solids concentration.
Cadmium - Table 7
The data for Cd given in Table 7 indicate that there is little
difference between the acid preserved and non-preserved samples and that
temperature control during transport is not critical. Although there is a
slight increase (7%) in the Cd extracted from the sludge material during
acid preservation, Cd was the least affected by simulated acid preservation
and the mock transport conditions.
Chromium - Table 8
The data given in Table 8 for Cr indicate that both acid preservation
and temperature control are critical. Without'acid preservation, the
dissolved fraction is partially lost (13%) as is shown in the non-preserved
river water plus spike data. Also, this loss (37%) is accelerated as the
temperature is increased to 49°C. Although there is an increase in the
extracted "acid-soluble" concentration because of acid preservation, the
data indicate the increase is limited to less than 16 percent when the
samples are stored .at 4°C over the three day holding period.
The conclusion for Cr "acid-soluble" analysis is that for valid
determinations both acid preservation and storage below 22°C are recommended.
-------
551
TABLE 6. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - ARSENIC
Concentration, mg/L
Acid Soluble Analyses
Lab. Control Spike Std.
Ohio River
Ohio River + Spike
Ohio River + Spike
Ohio River + Spike
Ohio River + Sludge
Ohio River + Sludge
Ohio River + Sludge
River + Sludge + Spike
River + Sludge + Spike
River + Sludge + Spike
Acid Preserved
Mean Range
(22°C)
(22°C)
( 4°C)
(22 C)
(49°C)
( 4"0C)
(22 C)
(49UC)
( 4°C)
(22 C)
(49°C)
0.296
0.016
0.360
0.350
0.315
0.020
0.032
0.026
0.366
0.362
0.339
0.021
0.023
0.004
0.004
0.009
0.011
0.011
0.000
0.013
0.031
0.004
Not Acid Preserved
Mean Range
N.D.
0.353
0.342
0.318
N.D.
0.016
0.022
0.356
0.334
0.283
0.007
0.007
0.004
0.003
0.004
0.027
0.012
0.002
Total Recoverable Analyses
Lab.
Ohio
Ohio
Control
River +
River +
Spike Std.
Spike
Spike
(22°
( 4°
(22
C)
C)
C)
0
0
0
.421
.435
.433
0
0
0
.013
.006
.004
0
0
.428
.431
0.000
0.002
Direct Analysis of Spike Solutions
Value
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HNOs + 5% (v/v) HC1
0.400
0.302
0.409
N.D. - Not Detected < 0.016 mg/L
-------
552
TABLE 7. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - CADMIUM
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22"C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HMOs (pH 1.5)
Diluted in 1.0% (v/v) HNOa + 5% (v/v)
mg/L
Acid Preserved Not Acid Preserved
Mean Range Mean Range
0.163
0.001
0.165
0.162
0.164
0.150
0.150
0.159
0.310
0.301
0.307
0.160
0.160
0.158
HC1
0.001
0.000
0.002
0.000
0.000
0.005
0.004
0.002
0.004
0.006
0.003
0.001
0.001
0.001
Value
0.160
0.166
0.162
___
0.001
0.160
0.162
0.156
0.140
0.142
0.143
0.296
0.291
0.286
0.155
0.158
___
0.001
0.004
0.001
0.001
0.001
0.001
0.001
0.013
0.009
0.009
0.000
0.000
-------
TABLE 8. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - CHROMIUM
553
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HN03 + 5% (v/v
mg/L
Acid Preserved Not Acid Preserved
Mean Range Mean Range
0.167
0.007
0.173
0.173
0.177
6.81
6.97
11.1
7.01
7.85
10.1
0.164
0.181
0.179
r) HC1
0.001
0.003
0.009
0.006
0.006
0.25
0.55
0.24
0.45
0.25
0.16
0.002
0.001
0.001
Value
0.160
0.166
0.166
0.005
0.149
0.130
0.108
5.88
5.34
5.28
6.46
5.84
5.29
...
0.173
0.182
0.001
0.007
0.008
0.001
0.26
0.26
0.38
0.03
0.09
0.10
0.001
0.002
-------
554
Copper - Table 9
The data for Cu given in Table 9 show good agreement between
duplicates. The data indicate there is little difference, less than 10
percent, between the acid preserved and non-preserved samples, when
temperature is controlled to 4°C.
In the non-preserved river water plus spike samples stored at 49°C there
is a loss (13%) in the dissolved fraction of the samples. However,, when all
the non-preserved sample data are reviewed, a conclusion of loss because of
elevated temperature is not supported. Also, a comparison of the
"acid-soluble" data to the "total-recoverable" data for the non-preserved
river water plus spike samples indicate not all Cu is made available for
analysis using Method 200.1, as is the case in acid preserved samples. The
complete extraction of Cu is attributed to the longer acid contact time
(96 hours) of acid preservation (see Table 4).
These data indicate that for Cu neither acid preservation nor controlled
temperature during transport and storage are required.
Lead - Table 10
The data for Pb given in Table 10 show good agreement between the
laboratory control standard and the other standard solutions analyzed.
Comparison of the "acid-soluble" concentration to the "total recoverable"
analyses of the river water plus spike indicate no loss of the soluble spike
or dissolved fraction.
The Pb concentration in the acid preserved river water plus sludge is
very similar to that in the non-preserved sample with acceptable precision
between duplicates. As expected, if the temperature is increased to 49°C
during storage, the Pb extracted from the particulate material in acid
preserved samples increases. However, this is not true for the
non-preserved sample. In fact, although not conclusive, there appears to be
some indicationoof loss in the non-preserved samples with elevated storage
temperature (49°C).
The data in Table 10 indicate for consistent and valid "acid-soluble" Pb
analyses, sample transport and storage at 4°C is recommended. With the
temperature controlled at 4°C the increase in "acid-soluble" Pb
concentration from acid preservation over the three day holding period is
limited to less than 5 percent.
Mercury - Table 11
The data for Hg given in Table 11 are evidence that Hg as an
"acid-soluble" metal can not easily be analyzed. Even the acid-preserved
soluble spike at 4°C was not stable for the three day holding period giving
a recovery of less than 80 percent. Also, when the soluble spike was
combined in the river water with the sludge particulate material, recovery
was further reduced to less than 50 percent. In this evaluation only the
"total recoverable" analyses of Hg appeared to be acceptable.
-------
TABLE 9. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - COPPER
555
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HNOa (pH 1.5)
Diluted in 1.0% (v/v) HMOs + 5% (v/v
mg/L
Acid Preserved Not Acid
Mean Range Mean
0.396
0.010
0.414
0.420
0.415
2.28
2.30
2.34
2.73
2.70
2.72
0.384
0.420
0.415
) HC1
0.001
0.001
0.002
0.006
0.004
0.03
0.08
0.03
0.05
0.10
0.01
0.006
0.002
0.001
Value
0.400
0.392
0.393
___
0.008
0.394
0.393
0.341
2.11
2.16
2.08
2.57
2.54
2.40
0.418
0.432
Preserved
Range
0.001
0.005
0.001
0.001
0.02
0.04
0.05
0.08
0.11
0.09
0.020
0.017
-------
556
TABLE 10. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - LEAD
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted' in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HN03 + 5% (v/v!
mg/L
Acid Preserved Not Acid Preserved
Mean Range Mean Range
0.405
0.020
0.431
0.422
0.422
0.441
0.434
0.520
0.785
0.776
0.841
0.389
0.439
0.434
) HC1
0.002
0.016
0.017
0.027
0.033
0.044
0.021
0.038
0.020
0.018
0.005
0.012
0.021
0.013
Value
0.400
0.413
0.406
____
0.025
0.424
0.427
0.393
0.440
0.393
0.405
0.751
0.759
0.646
0.420
0.434
.MM,,^
0.019
0.033
0.022
0.002
0.042
0.013
0.005
0.054
0.013
0.017
0.000
0.012
-------
557
TABLE 11. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - MERCURY
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HN03 + 5% (v/v)
Acid
Mean
0.163
N.D.
0.126
0.116
0.083
N.D.
N.D.
N.D.
0.067
0.038
N.D.
0.163
0.169
0.165
HC1
Concentration,
mg/L
Preserved Not Acid
Range Mean
0.005
—
0.007
0.003
0.002
—
0.020
0.018
0.004
0.002
0.004
Value
0.160
0.272
0.164
u^ulll ,
N.D.
0.070
0.044
0.015
N.D.
N.D.
N.D.
0.036
0.030
N.D.
0.157
0.145
Preserved
Range
t JMVU
—
0.001
0.001
0.003
—
0.003
0.002
0.006
0.004
•
N.D. - Not Detected < 0.015 mg/L
-------
558
An additional observation on the analysis of Hg from this evaluation is
that there is a significantly different response in the ICP analysis of the
spiking solution diluted in the two different acid matrices. In the dilute
nitric there is a 60 percent increase in the mercury signal. Over a period
of 2 to 3 days reanalysis showed the increase to Tessen, and if then mixed
with hydrochloric acid, the response equaled that of the calibration
standard. Unfortunately, if Hg spiking solutions are freshly prepared in
0.2 percent nitric acid and then immediately mixed with hydrochloric acid
the response remains elevated. This situation appears to have a chemistry
and reaction rate not understood and merits additional study.
Nickel - Table 12
The data for Ni given in Table 12 are straightforward, reliable and are
similar to that of Cu. Only non-acidopreserved river water with a soluble
spike at the elevated temperature (49°C) showed a decrease in
concentration. The effect of acid preservation on samples stored at 4°C was
less than a seven percent increase over non-preserved samples stored at the
same temperature. Therefore, for the determination of "acid-soluble" nickel
acid preservation is recommended.
Silver - Table 13
As in the case of Hg, the data given in Table 13 for Ag indicate it also
is not suited to "acid-soluble'J analysis. When spiked into river water,
acid preserved and stored at 4°C, Ag recovery approximated only 80 percent.
When combined with the sludge particulate material recovery dropped to 30
percent. Only the "total recoverable" analyses of Ag gave consistent and
reliable data.
Zinc - Table 14
The data given in Table 14 for Zn exhibit similar characteristics to
data of other metals. There is good agreement between the laboratory
control standard and the direct analysis of the spiking solutions. The
difference between "acid-soluble" analyses of acid preserved samples and
non-preserved samples stored 4°C is less than 10 percent. The soluble spike
in the non-preserved samples stored at 49°C is partially lost, while acid
preserved samples stored at the same temperature show an increase.
Therefore, the data given in Table 14 for "acid-soluble" Zn analyses
indicate samples should be acid preserved and stored at 4°C until time of
processing.
The following is a summation regarding sample preservation for
"acid-soluble" analyses:
1. Ag and Hg are not suited to "acid-soluble" analysis. Acid
preservation does not prevent loss from the dissolved fraction.
Additionally, these metals are not extracted with dilute acid and
recoveries are poor unless analysis is completed as "total" or
"total recoverable".
2. Although dissolved As will remain in solution without acid
preservation, losses will occur even with acid preservation if
-------
559
TABLE 12. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - NICKEL
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HNOs + 5% (v/v
mg/L
Acid Preserved Not Acid
Mean Range Mean
0.409
0.019
0.434
0.431
0.438
0.932
0.929
0.964
1.35
1.32
1.35
0.404
0.454
0.440
) HC1
0.000
0.003
0.001
0.002
0.003
0.020
0.029
0.014
0.03
0.03
0.02
0.000
0.021
0.002
Value
0.400
0.414
0.406
—
0.016
0.409
0.388
0.315
0.872
0.865
0.857
1.27
1.26
1.18
0.428
0.490
Preserved
Range
___
0.000
0.016
0.011
0.004
0.008
0.020
0.020
0.06
0.00
0.01
0.001
0.019
-------
560
TABLE 13. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - SILVER
Concentration, mg/L
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22eC)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2% (v/v) HN03 (pH 1.5)
Diluted in 1.0% (v/v) HNOs + 5% (v/v)
Acid Preserved
Mean Range
0.042
N.D.
0.035
0.033
0.030
N.D.
0.003
0.005
0.014
0.012
0.005
0.042
0.046
0.046
HC1
0.000
0.004
0.002
0.001
0.001
0.000
0.000
0.003
0.001
0.002
0.000
0.000
Not Acid Preserved
Mean Range
«_.
0.003
0.023
0.020
0.011
N.D.
0.003
N.D.
0.005
0.005
0.004
0.044
0.044
Value
0.043
0.043
0.043
___
0.000
0.003
0.000
0.002
0.001
0.001
0.002
0.002
0.000
0.001
N.D. - Not Detected < 0.003 mg/L
-------
561
TABLE 14. THREE DAY SAMPLE HOLDING PERIOD AT
VARIOUS TEMPERATURES - ZINC
Concentration,
Acid Soluble Analyses
Lab. Control Spike Std. (22°C)
Ohio River (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Ohio River + Spike (49°C)
Ohio River + Sludge ( 4°C)
Ohio River + Sludge (22°C)
Ohio River + Sludge (49°C)
River + Sludge + Spike ( 4°C)
River + Sludge + Spike (22°C)
River + Sludge + Spike (49°C)
Total Recoverable Analyses
Lab. Control Spike Std. (22°C)
Ohio River + Spike ( 4°C)
Ohio River + Spike (22°C)
Direct Analysis of Spike Solutions
Theoretical Spike Concentration
Analyzed Spike Concentration
Diluted in 0.2$ (v/v) HNOs (pH 1.5)
Diluted in 1.0% (v/v) HNOs + 5% (v/v
mg/L
Acid Preserved Not Acid Preserved
Mean Range Mean Range
0.161
0.099
0.261
0.263
0.283
2.86
2.95
3.62
2.99
3.03
3.82
0.153
0.316
0.314
) HC1
0.002
0.003
0.002
0.000
0.004
0.07
0.05
0.06
0.07
0.05
0.10
0.002
0.007
0.001
i
Value
0.160
0.163
0.159
•
0.088
0.243
0.238
0.205
2.65
2.66
2.64
2.81
2.80
2.69
0.301
0.355
0.001
0.007
0.001
0.001
0.02
0.01
0.03
0.05
0.04
0.09
0.005
0.034
-------
562
samples are transported or stored at elevated temperatures (49°C).
3. Of the elements investigated Cd does not appear to be adversely
affected by varying storage temperatures or lack of acid
preservation.
4. The remaining six metals (Al, Cr, Cu, Ni, Pb and Zn) all had losses
from the dissolved fraction, when samples are not acid preserved
and the storage temperature is increased (49°C). Only Cr showed a
loss (13%) for a non-preserved sample stored at 4°C for three
days. For samples that were acid preserved and stored at 4°C, the
increase in concentration of Cu, Ni, Pb and Zn was less than 10
percent while for Al and Cr, it was near 15 percent.
Since the dissolved metal fraction is most available to aquatic life and
should have the greatest toxicological impact, acid preservation is
recommended, especially for Cr. To reduce extractability and chemical
reactivity during transport, samples should be stored at 4°C until
processing can begin. These two provisions have been incorporated into
Method 200.1 Acid preservation is accomplished by adding 2 ml of (1+1)
nitric acid to approximately 800 ml of sample at the time of collection. It
has been determined that this amount of acid will not lower deionized
distilled water below pH 1.75. Samples are to be held at 4°C and processing
must begin within three days after sample collection.
Effect of Particulate Material
In the final phase of the evaluation the effect that varying amounts of
particulate material have on extraction efficiency was investigated. To
accomplish this, four different levels of the sludge particulate material
were added to river water and preserved as instructed in version 1.4 of
Method 200.1. To some samples, soluble spikes of Hg, Ag and As were added
because the concentration in the sludge material was too low for analysis.
Seven replicates were prepared for each level as well as for the river water
control. To calculate percent recovery of the "acid-soluble" analyses,
average concentration found in the river water control was subtracted from
the average value determined for each level. These net values were then
compared to the expected net concentration as determined from the "total
recoverable" analyses of the same samples. The analytical data from these
determinations are given in Tables 15 through 17. Also included are "total"
or estimated "total" analyses data as calculated from the acid-reflux
digestions.
A review of the standard deviation data for "acid-soluble" analyses
given in all three tables shows the precision between the replicates to be
very good. Although the extraction efficiency is not the same for each
metal, comparison of the mean values for each metal across the various spike
levels indicates extraction is uniform for varying amounts of particulate
material. In fact, the "acid-soluble" metal extraction was more uniform
across the prepared samples tested than the "total recoverable" analyses.
-------
563
TABLE 15. ACID SOLUBLE METAL PRECISION AND PERCENT
RECOVERY DATA -2.5 grams/L
Concentration,
mg/L
Acid Soluble*
Metal
Ag
AT
As
Cd
Cr
Cu
Hg
Ni
Pb
Zn
Total
Acid Reflux
0.015
5.13
0.028
0.098
10.1
1.26
N.D.(b)
0.525
0.810
5.46
Total
Recoverable
0.012
4.83
0.031
0.085
7.48
1.17
0.114(c)
0.466
0.601
3.43
Mean
0.003
2.18
0.007U)
0.074
3.18
1.11
0.029
0.457
0.311
1.35
Std. % Total
Dev. Recoverable
±0.002
±0.047
±0.001
±0.002
±0.085
±0.025
±0.004
±0.008
±0.010
±0.018
25%
45%
23%
87%
43%
95%
25%
98%
52%
39%
* Data are from 7 replicate samples
(a) Arsenic filtrate concentrated 4x before analysis
instrumental detection limit = 0.003 mg/L.
(b) N.D. - Not detected < 0.015 mg/L.
(c) A soluble mercury spike of 0.125 mg/L was added to the sample
before processing.
-------
564
TABLE 16. ACID SOLUBLE METAL PRECISION AND PERCENT
RECOVERY DATA - 1.0 grams/L
Concentration, mg/L
Metal
Ag
AT
As
Cd
Cr
Cu
Hg
Ni
Pb
Zn
Est. Total
Acid Reflux
0.031U)
2.05
0.323(b)
0.039
4.04
0.505
0.063(c)
0.211
0.324
2.19
Acid Soluble*
Total
Recoverable
0.026
1.74
0.328
0.036
3.28
0.502
0.065
0.194
0.268
1.50
Mean
0.014
0.870
0.313
0.030
1.21
0.442
0.021
0.191
0.137
0.548
Std. % Total
Dev. Recoverable
±0.001
±0.045
±0.007
±0.002
±0.070
±0.019
±0.002
±0.008
±0.008
±0.017
54%
50%
95%
83%
37%
88%
32%
98%
51%
37%
* Data are from 7 replicate samples
(a) Silver value includes a soluble spike of 0.025 mg/L.
(b) Arsenic value includes a soluble spike of 0.313 mg/L
(c) Mercury value includes a soluble spike of 0.063 mg/L.
-------
565
TABLE 17. ACID SOLUBLE METAL PRECISION AND PERCENT
RECOVERY DATA - 0.25 grams/L
Concentration, mg/L
Acid Soluble*
Metal
Ag
Al
As
Cd
Cr
Cu
Hg
Ni
Pb
Zn
Est. Total
Acid Reflux
0.052(a)
0.513
0.063(b)
0.010
1.01
0.126
0.025(c)
0.053
0.081
0.546
Total
Recoverable
0.050
0.421-
0.061
0.010
0.669
0.130
0.024
0.051
0.067
0.298
Mean
0.024
0.261
0.065
0.009
0.326
0.115
N.D.(d)
0.062
0.042
0.150
Std.
Dev.
±0.001
±0.019
±0.003
±0.001
±0.032
±0.006
±0.006
±0.007
±0.010
% Total
Recoverable
48%
62%
103%
90%
49%
88%
132%
63%
50%
* Data are from 7 replicate samples.
(a) Silver value includes a soluble spike of 0.050 mg/L
(b) Arsenic value includes a soluble spike of 0.063 mg/L
(c) Mercury value includes a soluble spike of 0.025 mg/L
(d) N.D. - Not Detected < 0.015 mg/L
_
-------
566
Only Pb showed a gradual increase (10%) in extraction efficiency as the
amount of particulate material was decreased.
As expected, recovery of the soluble As spike was near 100 percent with
good precision. However, recoveries of the soluble Hg and Ag spikes were
not readily improved because of the lower concentration of particulate
material. In each case where the concentration was measurable, precision
was good but recovery of Hg and Ag was limited to near 30 and 50 percent,
respectively.
Of the metals investigated that are suitable to the "acid-soluble"
analysis, seven can be grouped into two categories. One group (Cd, Cu and
Ni) are easily extracted and give recoveries near 90 percent. For the other
group (Al, Cr, Pb and Zn) extraction is more affected by pH, temperature and
extraction time, and give recoveries nearer 50 percent. When experimental
conditions are well controlled the "acid-soluble" extraction of all of these
metals'is precise, uniformly consistent, but different. .These differences
cannot be attributed to the amount of particulate material present, but
rather to the chemical nature of these metals. Factors that affect metal
solubility using Method 200.1 are the ease that insoluble compounds are
dissociated with the lowering of pH or the stability of a metal complex
under acidic conditions. These factors are affected by the sample matrix,
anion constituents, and when "acid-soluble" metal concentrations are
compared to the "total-recoverable" analyses, the comparative data may vary
from sample to sample. Although these differences are expected, the
solubilized metal concentration in each case when using Method 200.1 is by
definition the "acid-soluble" metal concentration.
-------
567
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
DR. GAIND: Arun Gaind from
Nanco Labs. The silver and mercury may have gotten
out of solution because the river water had high level
of chloride.
MR. MARTIN: Yes.
DR. GAIND: So, the procedure
now is going to institutionalize Ohio River water
use for this procedure? No? I mean...
MR. MARTIN: I'm sorry, I
didn't understand you.
DR. GAIND: You used...to
verify the method you used Ohio River water.
MR. MARTIN: Yes.
DR. GAIND: Right. Okay,
but the procedure now when it becomes used for other
conditions, it will use DI water?
MR. MARTIN: No, I don't
believe so. It's going to be an ambient water
measurement.
DR. GAIND: Okay.
MR. MADDALONE: As part
of this EPRI program during this current round,
-------
568
we are doing the analysis of water for mercury, and
we talked with Gary McKee, Chief of the Inorganic
Analyses Section at EMSL-Cincinnati, and originally,
in the Federal Register a couple years ago, they
talked about using dichromate to stabilize mercury
samples. It turned out that use of dichromate and
gold and all those other substances to stablilize
mercury, were really related to stabilized mercury in
distilled water, and we confirmed this in our labs.
We spiked some distilled water and our ash point
overflow water. Ash pond overflow samples were fine.
They didn't show any drop in the mercury concentration
with the normal acidification scheme, but the distilled
water showed a very significant drop in the amount of
mercury present. So for our distilled water samples
in our program we had to put dichromate in them to
stabilize them, and that did stabilize them over a one
month period.
-------
569
-------
570
-------
571
RAYMOND F. MADDALONE
RESULTS OF THE TRW EPRI ANALYTICAL METHODS
QUALIFICATION PROGRAM, PART I
Dr. MADDALONE: I'm going to
discuss today the work that TRW is doing for the
Electric Power Research Institute. The program
that we have from the Electric Power Research
Institute is divided into three distinct areas.
The first area, literature studies,
identified the problems and scope the future
work. As part of that effort, we collated and
performed an analysis of all the available data on
the concentrations of parameters and elements in
the aqueous discharge streams from power plants.
A report summarizing that data was written. The
report contains data from open literature studies,
and a summary of 100 of the 1980 2C NPDES
permits. All of this discharge data was placed in
a computerized data base.
The next activity conducted was to acquire
precision and bias data from a number of sources,
including ASTM, EPA/Effluent Guidelines Division,
and UWAG. Those data were then collated and
reviewed to help us determine what were the
literature values for limits of detection and
limits of quantitation. A second report
summarized those findings.
-------
572
A third report was written summarizing the
results of a literature review of sampling and
analysis methods for the 13 priority pollutant
metals. This report assessed the state of the
art, provided background information on the
interferences associated with current methods, and
looked at new methods or procedures.
SLIDE 1
This slide summarizes the reports that we
have already published or to publish under the
program. As you can see, this first set here
above the dotted line is primarily the literature
work that we have done. Below the line we have
the reports that will come out at the end of each
one of the field studies. Through the good
offices of EPA, we obtained copies of the DMR/QA
data sets from the first four years. We are
currently analyzing those results and will publish
a report
The QA/QC guideline document is directed
towards the utility industry to provide the
utility chemist information on how to go about
setting up a quality assurance program. We think
that's probably one of the more important things
to do, because in the utility industry, you have
laboratories that range from the central lab, with
-------
573
the highly qualified individuals, to basically a
boiler water chemistry lab that's being used to do
some environmental analyses at the smaller plants.
The real focus of the program, and what I'm
going to be talking about today, is the method
validation studies. The literature studies led
the way and told us what trace metals we ought to
study. We then set-up a validation program
consisting of four parts called Analytical Methods
Qualification (AMQ). Let me show you how that is
configured.
SLIDE 2
This is the design of the field study
program. To date, we have completed Parts I and
II. Part I consisted of two rounds. During the
first round we did arsenic and selenium by
Graphite Furnace AAS, and in round 2 chromium,
copper, nickel and lead were determined by
Graphite Furnace AAS.
We performed these validation efforts in a
number of different matrices, including a river
water; an ash pond overflow stream, which
consisted of the river water source plus leachate
from the ash pond. We had a seawater intake and
seawater discharge. We used a reagent grade water
sample to provide a sample with no matrix
-------
574
interferences while a sample of treated chemical
metal cleaning wastes provided a very challenging
sample for analysis. EPA/EMSL provided us with a
number of their QA/QC ampules that we used to have
an absolute measure of the bias of each
laboratory.
What I will be presenting today are the
results from the first part, AMQ-I. We had 42
laboratories participate in the round-robin
study. Approximately 30 of them were what we
called our freshwater laboratories. Those are the
labs that analyzed the river water, ash pond
overflow, reagent grade water, the QA/QC ampule
and the treated chemical metal cleaning waste.
The seawater labs substituted for the two
freshwater streams, a seawater intake and a
seawater discharge. The data that I'll present
today will be from the freshwater laboratories.
The data that was used to calculate the
limits of detection was processed using standard
ASTM precision and bias statistics. That is, we
used D-2777-85. In that process you're allowed to
rank the labs and then go through a standard t-
test (1% double tail). You are permitted to use
two iterations of the t-test which in this case is
a one percent double-tailed t-test. The precision
data obtained fram data analysis using D-2777 was
-------
575
regressed versus the mean test concentration. The
resulting y-intercept (standard deviation at zero
concentration) was used to compute an LOD.
EPA/EMSL has published a procedure for the
Method Detection Limit in ES&T a number of years
ago for organics. It has now been promulgated in
the Federal Register as part of the GC organic
methods. The EPA/EMSL MDL procedure first
determines that the analyte concentration that you
are using is between one and five times the
estimated limit of detection of the method. If
so, you can then use that sample directly to
calculate a method detection limit. If it's
greater than five times the method detection
limit, but less than 10 times the detection limit,
you can still use the sample, though they would
probably prefer for you to find a sample with a
lower concentration.
The MDL calculation is based on a standard t-
value. The sample is analyzed seven times by a
single operator, and the t-value for a one percent
single tail with six degrees of freedom is
found. The standard deviation of the mean of the
replicate analyses is multiplied by that t-value
to calculate the method detection limit. The way
that we did the MDL calculation under the EPRI
program utilized the number of labs and data
-------
576
points obtained for each matrix and element. We
used the number of laboratories or the number of
data points, depending on whether we were dealing
with overall or the single operator MDLs to
provide the degrees of freedom.
SLIDE
As I mentioned, we also calculated an LOD
using the precision data from the D-2777
analysis. Each one of the samples that we sent to
our participants consisted of four concentration
levels, the background plus three spikes. Using
that data, we calculated a precision regression
equation and then extrapolated it to the zero
concentration, which would give you the precision
at zero concentration (i.e., the standard
deviation of a blank). By using that extrapolated
value, you can then apply the standard ACS
definition of three times the standard deviation
of the blank to determine the limit of detection.
All the data presented today was calculated
using method detection limit. The limits of
detection based on the regression equation data
are fairly similar to these data, and in fact they
agreed quite wel1 .
-------
577
SLIDE
This table here presents the results from the
freshwater streams (i.e., freshwater, ash pond
overflow, and reagent grade water samples). The
data were very close and there is no reason not to
collate the LOD estimates from those data. On the
other hand, I'll be showing you the MDL's
calculated from the treated chemical metal
cleaning waste and those obviously were quite
different from the freshwater1 s. In all cases
they showed much higher limits of detection,
perhaps indicating a matrix problem.
The data summarized in this slide are for the
six elements from the AMQ-I; this is the single
operator precision data. We have the overall
precision based limits of detection also. The EPA
quoted values are the values quoted by the EPA in
the "Methods of Chemical Analysis for Water and
Waste". All the values on this table are in parts
per bi 1 1 i on .
The EPRI survey LOD value is the simple mean
of the LOD data that we collated under our
literature survey. This LOD includes, again,
values from ASTM, US6S, EPA from the MCAW, and
EPA/EGD "Metals Methods Validation" study. The
plus or minus gives you some indication of what
the range was of the values reported.
8
-------
578
The final column here, EPRI field study, is
the results that we obtained from the EPRI AMQ-I
study. Those results were obtained using utility
laboratories and real world samples that were
spiked, split, acidified, and sent out to the
participating laboratories. We tried to simulate
as close as possible the situation that occurs in
the normal NPDES sampling exercise.
For the single operator data, the arsenic and
copper, nickel, lead and selenium MDLs are
slightly higher than the EPA quoted value and are
\
pretty much at the same level as we found in the
literature survey. Now remember, this is single
operator data.
SLIDE
Let's graphically take a look at this single
operator precision data for graphite furnace.
Remember all these LODs are based on the EPA MDL
calculation. For the freshwater matrices, you can
see here that the green are the quoted EPA values,
and the orange and the red bars, are the EPRI
field study. Again, there's fairly good agreement
with the survey data, but the calculated MDLs
generally are higher than the EPA quoted LODs. In
the 1974 version of the MCAW, the EPA quoted
values for graphite furnace were on the order of
-------
579
0.3 ppb. In the latest revision (1983) of the
MCAW, they are now 1.5, which is a little bit
closer to what we're seeing in terms of the real
world data.
SLIDE
This slide shows the overall precision data
for the freshwater streams. Now you can see that
we're starting to get, both in the EPRI survey
data and in the EPRI field study data, a slightly
higher and, in come cases, much .higher
calculations of the limit of detection. Also, you
can see that there is some range associated with
the values that we have. Arsenic and selenium are
two of the more difficult elements to do by
graphite furnace. In the case of some of our more
difficult matrices, such as the seawater and the
treated chemical metal cleaning waste, all the
participants complained about the difficulties
that they were having.
The people performing this work were very
dedicated. Even though we told them to perform
the analysis according to exact procedures that
are provided by the MCAW, several of participants
said, "we were having a great deal of difficulty
getting this element to recover properly or
getting the instrument to work, so we did it by
10
-------
580
Flame Atomic Absorption or we did it by Gaseous
Hydride AAS." It's an indication that these
people really want to do well, but in the case of
our study, it really didn't help us at all because
we only wanted to evaluate Graphite Furnace AAS.
SLIDE
Let's show what the overall precision MDLs
look like graphically. We start to see a great
deal of divergence in terms of what's quoted by
the EPA as an LOD (which is probably really a
single operator/single laboratory value) compared
to the overall precision-based limit of detection
that we found, both from our field study and from
our EPRI survey. An overall precision-based limit
of detection is really a better measure of what
goes on in the real world, because what you're
talking about is one laboratory comparing its
results versus another laboratory. In the case of
NPDES sampling, it could be the utility versus the
EPA or EPA contractor laboratory that might have
taken a sample. So you really want to use the
overall precision data, in that case, to evaluate
the results of the two laboratories.
SLIDE
Let's go to the treated metal chemical
11
-------
581
cleaning waste data. Here are the data from the
treated chemical metal cleaning waste. Treatment
started by adding various polymers and floculating
agents to the waterside boiler wash. It was
pumped into a receiving pond, allowed to stand
there for several days, and then it was pumped
through a filter. Analysis samples were taken
until the typical one ppm copper and iron values
were obtained, and then that solution was then
directly discharged into the receiving waters.
The sample analyzed in the EPRI program was the
sample that was treated to meet NPDES permit
di scharge limits.
You can see here from the field study data
for this matrix that we have much higher single
operator values than the EPA quoted or the EPRI
survey LODs. Furthermore, they are much higher
than the freshwater values that I had previously
di spl ayed.
Participants had a great deal of difficulty
with copper and nickel in the TCMCW samples. The
MDL for selenium, 3.5 ppb, is low, but I think it
was low because we had a stability problem with
the selenium in the treated chemical metal
cleaning waste. We didn't find much of the
selenium, and the precision then was fairly good
because they were dealing with low concentrations.
12
-------
582
SLIDE
Let's take a look at these single operator
data graphically. Note that there's a break in
the scale. The yellow correspond to the copper
and nickel results, which you can see have a much
higher MDLs in this matrix. This was a very
difficult matrix. Even though it meets discharge
requirements, it still probably has some portion
of the floculating agents, it has a high calcium
value, and a very high chloride concentration.
So, it's a very nasty matrix, but it's a matrix
that the utility industry has to monitor under the
permit laws.
SLIDE
Here's the table for the LODs based on the
overall precision from the treated chemical metal
cleaning wastes. Again, the scale break
correspond to the bar colors. You'll note that
the computed MDLs for copper and nickel show-up
very high and definitely higher for selenium,
arsenic, chromium, .and lead based on the data that
we obtained from the field validation effort. You
can see that the literature survey data fits in
about the same level as the field study data.
13
-------
583
Let me summarize now just where we stand in
this program. We have, as I mentioned, completed
the two rounds associated with AMQ-I. We expect
to finish the report in the summer 1986 and
probably have a report out, published by EPRI, by
the end of the year. AMQ-II has been completed
and we've just received the last bit of data two
weeks, ago. We have loaded the data onto our
computer system, but we haven't started to analyze
it yet. The AMQ-II data report will be out a bit
1ater in the year.
For the two remaining parts of the AMQ
project, we want to do a study using ICP for 10 to
15 elements. We also want to do total suspended
solids (non-filterable residue) and oil &
grease. Those two parameters, which are in
everybody's permit, have little precision data
associated with them. Bias data is very hard to
come by because the ability to make up a stable
sample for oil & grease and total suspended solid
is very difficult.
So, the EPRI AMQ program is continuing. We
are publishing reports on our validation effort
and reports are available from our survey
studies. If you have any questions, I'll be happy
to answer them at this time.
MR. TELLIARD: Any questions?
14
-------
584
Thank you very much.
15
-------
585
-------
586
-------
587
-------
588
-------
589
-------
590
-------
591
-------
592
CD ~
_ w CD
Ere
O O
-------
591
-------
-------
593
-------
594
-------
595
-------
oo u.
C3 C3 ^ C3
in O Lf) C3
ro co cvj esi
O o
-------
595
-------
«AU. g
CD ^
»*- o
— 03
E 2
O o
J0 C3 O O
LO C3 ITS ^
n ro csi cvi
-------
597
-------
598
-------
599
MR. TELLIARD: Thank you,
ladies and gentlemen. I'd like to thank the County
Court Reporters. They're down to one. I'd like to
thank you for your attention, your attendance, your
patience for all the slides that you couldn't read,
and hope to see you all back here next year.
We'll try to get the proceedings out before the
next meeting. Thank you very much for coming.
(WHEREUPON, the proceedings were concluded at 11:45 a.m.)
-------
600
ROSTER OF ATTENDEES
JAMES R. ALEXANDER
SUPERVISORY CHEMIST
NORFOLK NAVAL SHIPYARD
CODE 134.2 BLDG. 184
PORTSMOUTH VA 23709
804-396-3373
B. A. ALLEN
MANAGER, ENVIRONMENTAL SERV.
DOW CHEMICAL COMPANY
P. 0. BOX BB, ENVIRONMENTAL SERVK
FREEPORT TX 77541
409-238-2415
MIKE ALSOP
DEVELOPMENT SCIENTIST
UNION CARBIDE CORP.
P.O. BOX 8361
S. CHARLESTON WV 25303
304-747-5467
BOB APRIL
ENVIRONMENTAL SCIENTIST
USEPA-ITD
401 M STREET, SW, (WH-565E)
WASHINGTON DC 20460
202-382-4654
JOE ARLAUSKAS
MARTIN MARIETTA ENVR. SYSTEM
9200 RUMSEY ROAD
COLUMBIA MD 21045
301-964-9200
JOHN J. AUSTIN
CHEMIST
USEPA-REGION III
839 BESTGATE ROAD
ANNAPOLIS MD 21401
301-224-2740
CLIFTON BAILEY
OURS
USEPA-ITD
401 M STREET, SW, (WH-552)
WASHINGTON DC 20460
202-382-5411
ROBERT C. BARRICK
PRINCIPAL CHEMIST
TETRA TECH, INC.
11820 NORTHUP WAY, SUITE 100
BELLEVUE WA 98005
206-822-9596
ROBERT BEIMER
MANAGER, CHEMISTRY PROGRAM
S-CUBED
P.O. BOX 1620
LA JOLLA CA 92038
619-453-0060
JOHN BIRRI
QA CHEMIST
USEPA-REGION II
WOODBRIDGE AVE.
EDISON NJ 08837
201-321-6722
DAVE BOTTRELL
USEPA-EMSC-LV QAD
P.O. BOX 15027
LAS VEGAS NV 89104
702-798-2142
C. DON BOWYER
CHEMICAL LAB CERT. OFFICER
NJDEP-OSR-QA (CN 409)
436 E. STATE ST.
TRENTON NJ 08625
609-292-3950
-------
601
STEVE BRADY
CHEMIST
EASTMAN KODAK COMPANY
KODAK PARK. BLDG. 34
ROCHESTER NY 14650
716-722-3430
BILL BRAMSTEDT, Ph.D
SDS BIOTECH
BOX 348
PAINESVILLE OH 44077
216-357-3350
MICHAEL BRENNAN
SALES ADMINISTRATOR
ENVIRITE ANALYTICAL SERVICES
OLD WATERBURY ROAD
THOMASTON CT 06787
203-283-8235
WADE BROUGHTON
CHEMIST II
CITY OP ATLANTA
2640 JONESBORO ROAD, SE
ATLANTA GA 30316
404-627-1222
JOHN BROWN
MARINE CHEMISTRY RESEARCHER
BATTELLE N.E.M.R.L.
397 WASHINGTON ST.f BOX AH
DUXBURY MA 02332
617-934-5682
EARNESTINE BUTTONS
CHEMIST
CITY OF ATLANTA
2640 JONESBORO RD. SE
ATLANTA GA 30315
404-627-1222
CHAN CALDWELL
SUPV. OF GAS CHROMATOGRAPHY
PRINCETON AQUA SCIENCE/IT
165 FIELDCREST AVE., CN7809
EDISON NJ 08818
201-225-2000
GENE CHIU
LABORATORY DIRECTOR
TP ANALYTICAL LABS
P.O. BOX 919
CONYERS GA 30207
404-922-8000
WINSTON CHOW
PROJECT MANAGER
ELECTRIC POWER RESEARCH INST
3412 HILLVIEW AVE.
PALO ALTO CA 94304
415-855-2868
GORDON CHU
CAL LAB EAST, INC.
2240 DABNEY ROAD
RICHMOND VA 23230
804-359-1900
YEUH CHUANG
MARTIN MARIETTA ENVR. SYSTEM
9200 RUMSEY ROAD
COLUMBIA MD 21045
301-964-9200
BRUCE N. COLBY
PRESIDENT
PACIFIC ANALYTICAL, INC.
1989 B PALOMAR OAKS WAY
CARLSBAD CA 92008
619-931-1766
-------
602
DR. U. M. COWGILL
ASSOC. ENVR. CONSULTANT
DOW CHEMICAL
2030 WHDOW CENTER
MIDLAND MI 48674
517-636-1735
MARK J. CROCHET
ANALYTICAL CHEMIST
BASF CORP.
P.O. BOX 457
GEISMAR LA 70734
504-387-0631 EXT.2424
MICHAEL D. CROUCH
PRESIDENT
TOXICON LABS, INC.
3213 MONTERREY BLVD.
BATON ROUGE LA 70814
504-925-5012
JOHANNA M. CULVER
CHEMIST
NORFOLK NAVAL SHIPYARD
CODE 134.2 BLDG. 184
PORTSMOUTH VA 23709
804-396-3373
SEYED DASTGHEYB
GE/MS-SUPERVISOR
UNITED STATES TESTING CO.
1415 PARK AVE.
HOBOKEN NJ 07030
201-792-2400
WILLIAM DAVIS
MARKETING MANAGER
FINNIGAN CORP.
355 RIVER OAKS PARKWAY
SAN JOSE CA 95134
408-433-4800
SUSAN deNAGY
PROJECT OFFICER
USEPA-ITD
401 M STREET, SW (WH-552)
WASHINGTON DC 20460
202-382-7141
KATHY J. DIEN HILLIG
RESEARCH STAFF
BASF CORP.
1419 BIDDLE AVE.
WYANDETTE MI 48192
313-246-6334
DR. THOMAS W. DUKE
SR. RESEARCH SCIENTIST
USEPA-REGION IV
ENVR. RESEARCH LAB, SABINE ISLAND
GULF BREEZE FL 32561
904-932-5311
TINA ENGEL
RESEARCH SCIENTIST
BATTELLE MEMORIAL INSTITUTE
505 KING AVENUE
COLUMBUS OH 43201
614-424-4149
BARRETT P. EYNON
STATISTICIAN
SRI INTERNATIONAL
333 RAVENSWOOD AVE.
MENLO PARK CA 94025
415-859-5239
PAUL FARROW
VG MASSLAB
TUDOR ROAD, ALTRINGHAM
CHESIRE CD WAI4 5RZ
061-941-3552
-------
603
THOMAS FIELDING
USEPA-ITD
401 M STREET, SW, (WH-552)
WASHINGTON DC 20460
202-382-7156
RUSSELL D. FOSTER, JR.
TECHNICAL DIRECTOR
RESOURCE ANALYSTS, INC.
P.O. BOX 778
HAMPTON NH 03842
603-926-7777
DR. PAUL FRIEDMAN
USEPA OFFICE OF SOLID WASTE
401 M STREET, SW, (WH-562-B)
WASHINGTON DC 20460
202-382-7700
DR. ARUN GAIND
EXECUTIVE VICE PRESIDENT
NANCO LABS INC.
P.O. BOX 10
HOPEWELL JCT. NY 12533
914-221-2485
DANIEL A. GARTIEZ
VESTEC CORPORATION
THOMAS J. GEYER
ENVIRONMENTAL LAB SPVR.
FORD MOTOR COMPANY
15201 CENTURY DRIVE
DEARBORN MI 48047
313-845-1648
WILLIAM E. GORE
AMERICAN CYANAMID
P.O. BOX 60
STAMFORD CT 06904
203-655-8873
KATHY GROSS
CAL LAB EAST, INC.
2240 DABNEY ROAD
RICHMOND VA 23230
804-359-1900
ROBERT E. HACKLEY
CHEMIST
BURLINGTON NORTHERN R.R.
1612 N. LEXINGTON
SPRINGFIELD MO 65802
417-864-3177
DR. EDWARD HANDEL
CHAIRMAN OF THE BOARD
CENTEC ANALYTICAL SERVICES
2160 INDUSTRIAL DRIVE
SALEM VA 24153
703-387-3995
WILLIAM C. HARRIS
LABORATORY SUPERVISOR
UNION CAMP CORP.
P.O. BOX 178
FRANKLIN VA 23851
804-569-4263
JOHN C. HENDRICKS
CHEMIST
AMERICAN ELEC. POWER &
1 RIVERSIDE PLAZA
COLUMBUS OH 43215
614-223-1238
SERV.
-------
604
DIANA HOLLEY
SR. ENVIRONMENTAL ENGINEER
ONION CARBIDE-INSTITUTE PLT.
P.O. BOX 2831
CHARLESTON WV 25330
304-747-1885
MIN 6. BU
GROUP LEADER, IND. ANALYSIS
ROLLINS ENVR. SERVICES
2027 BATTLEGROUND
DEER PARK TX 77536
713-479-6001
WILLIAM P. HUFF
TECHNICAL DIRECTOR
ANDREW S. McCREATH & SON
301 WEST FREEMASON ST.
NORFOLK VA 23510
804-622-6136
WILLIAM IMBUR
MANAGER, TESTING & MEASURE.
LOW ENVIRONMENTAL SERVICES
2749 DECK ROAD, SE
MARIETTA GA 30067
404-952-9005
PETER ISSACSON
CHEMIST
VIAR AND COMPANY
300 N. LEE ST., SUITE 200
ALEXANDRIA VA 22314
703-683-0885
RICHARD A. JAVICK
SR. RESEARCH CHEMIST
FMC CORPORATION
P.O. BOX 8
PRINCETON NJ 08540
CHARLOTTE W. K1MBROUGH
CHEMIST
MARTIN MARIETTA ENERGY SYST.
P.O. BOX Y
OAK RIDGE TN 37831
615-574-3715
TODD KIMMELL
USEPA-OFFICE OF SOLID WASTE
401 M STREET, SW, (WH-562-B)
WASHINGTON DC 20460
202-382-4795
JAMES R. KING
SAMPLE CONTROL CTR. COORD.
VIAR AND COMPANY
300 NORTH LEE ST., SUITE 200
ALEXANDRIA VA 22314
703-683-0885
JOHN A. KIRCHSTEIN
TECHNICAL SERVICES ENGINEER
E.I. DuPONT
ENGINEERING TEST CENTER
WILMINGTON DE 19898-7104
302-366-3608
DIANNA KOCUREK
PARTNER
TISCHELER-KOCUREK
116 E. MAIN ST.
ROUND ROCK TX 78664
512-244-9058
MARCIA KUEHL
DONOHUE & ASSOCIATES, INC.
4738 N. 40TH
SHEBOYGAN WI 53081
414-458-8711
-------
605
SUJITH KUMAR
TECHNICAL DIRECTOR
INTERNATIONAL TECH. CORP.
5103 OLD WILLIAM BERN HWY.
EXPORT PA 15632
412-731-8806
ROBERT LANNAN
ATTORNEY
ROBINSON & MCELWEE
P.O. BOX 1791
CHARLESTON WV 25326
304-344-5800
PATIRCIA LASKA
CHEMIST
USEPA-EMSL-LAS VEGAS
944 E. HARMON AVE.
LAS VEGAS NV 89109
702-798-2671
LINDA LAUGHLIN
LAB COORDINATOR
MARTIN MARIETTA ENVR SYSTEMS
9200 RUMSEY ROAD
COLUMBIA MD 21045
301-964-9200
EDWARD LAWLER
SENIOR CHEMIST
CAMBRIDGE ANALYTICAL ASSOC.
1106 COMMONWEALTH AVE.
BOSTON MA 02215
617-232-2207
SIM D. LESSLEY
ASSOCIATE DIRECTOR
UBTL, INC.
520 WAKARA WAY
SALT LAKE CITY UT 84108
801-584-3232
CONNIE LESTER
CHEMIST
CITY OF ATLANTA
2440 BOLTON RD. NW
ATLANTA GA 30318
404-351-6120 EXT.284
HAROLD LESTER
MARKETING DIRECTOR
POLICY PLANNING & EVAL., INC
8521 LESSBURG PIKE, SUITE 310
VIENNA VA 22180
703-893-0315
NATHAN LEVY
PRESIDENT
A&E TESTING, INC.
1717 SEABORD, SUITE 3,03
BATON ROUGE LA 70810
504-769-1930
JAMES W. LEWIS
LABORATORY MANAGER
BIONETICS CORP.
20 RESEARCH DRIVE
HAMPTON VA 23666
804-864-0680
HARRIS A. LICHTENSTEIN, Ph.D
PRESIDENT
SPECTRIX CORPORATION
3911 FONDREN, SUITE 100
HOUSTON TX 77063-5821
713-266-6800
DENIS LIN
ETC CORP.
P.O. BOX 7808
EDISON NJ 08818-7808
201-225-6759
-------
606
TOM A. LOWE
MGR. ANALY. & SURFACE CHEM,
KAISER ALUMINUM
P.O. BOX 877
PLEASANTON CA 94566
415-847-4626
JEFFREY C. LOWRY
IND. CS/MS SUPERVISOR
ROCKY MT. ANALYTICAL LAB
5530 MARSHALL ST.
ARVADA CO 80002
303-421-6611
SAMUEL V. LUCAS
BATTELLE COLUMBUS LABS
505 KING AVE.
COLUMBUS OH 43201
614-424-4178
LARRY LaFLEUR
ORGANIC PROGRAMS MANAGER
NCASI
720 S.W. 4TH STREET
CORVALLIS OR 97333
503-754-2015
CHRISTINE MACKO
ENVIRONEMTNAL ENGINEER
SAIC
8400 WESTPARK DIRVE
MCLEAN VA 22102
703-734-3106
RAYMOND F. MADDALONE, Ph.D
PROGRAM MANAGER
TRW SPACE & TECHNOLOGY GRP.
ONE SPACE PARK
REDONDO BEACH CA 90278
213-536-2451
GORDON DEAN MARBURY
SENIOR CHEMIST
TRIANGLE LABORATORIES, INC.
4915-F PROSPECTUS DRIVE
DURHAM NC 27713
919-544-5729
DR. MARK F. MARCUS
DIRECTOR, ANALYTICAL CHEM.
CHEMICAL WASTE MGMT.
150 W. 137TH STREET
RIVERDALE IL 60627
312-841-8360
THEODORE D. MARTIN
CHEMIST
USEPA/EMSL-CINTI
26 W. ST. CLAIR
CINCINNATI OH 45208
513-569-7312
ROBERT METTER
LABORATORY SERVICES MANAGER
BECKMAN INDUSTRIAL CORP.
4141 PALM ST.
FULLERTON CA 92635
714-447-2326
DR. DEBORAH S. MILLER
UNION CARBIDE CORP.
P.O. BOX 8361, B770-318
S. CHARLESTON WV 25314
304-747-4463
RAYMOND F. MINDRUP
MKT. DEVELOPMENT SPECIALIST
SUPELCO, INC.
SUPELCO PARK
BELLEFONTE PA 16823-0048
814-359-3441
-------
607
SANDY MOSER
COUNTY COURT REPORTERS, INC.
30 SOUTH CAMERON ST.
WINCHESTER VA 22601
703-667-0600
NEIL H. MOSESMAN
RESEARCH CHEMIST
SUPELCO INC.
SUPELCO PARK
BELLEFONTE PA 16823
814-359-3441
DR. JOHN J. MOUSA
MANAGER OF CHEMISTRY DIV.
ENVR. SCIENCE & ENG., INC.
P.O. BOX ESE
GAINESVILLE FL 32602-3053
904-332-3318
JOHN S. MURRAK
EMS LABS
7901 W. MORRIS ST.
INDIANAPOLIS IN 46231
317-298-4146
ALEXANDER C. McBRIDE
CHIEF,WATER QUALITY ANALYSIS
USEPA-ITD
401 M ST., SW., (WH-553)
WASHINGTON DC 20460
202-382-7046
KEVIN McGONNAGHY
COMPUCHEM LABORATORIES
P.O. BOX 12652
RESEARCH TRIANGLE PARK NC 27709
919-549-8263
DONNA L. McGOVERN
CHEMIST
NORFOLK NAVAL SHIPYARD
CODE 134.2 BLDG. 184
PORTSMOUTH VA 23709
804-396-3373
DR. JOHN McGUIRE
USEPA-ATHENS E.R.L.
COLLEGE STATION RD.
ATHENS GA 30613
404-546-3185
GEORGE ODELL
MANAGER/ORGANICS LAB
NANCO LABS INC.
P.O. BOX 10
HOPEWILL JCT. NY 12533
914-221-2485
VIJAYAMOHAN K. PALAT
LAB MANAGER
CHEM CLEAR INC.
992 OLD EAGLE SCHOOL RD.
WAYNE PA 19087
215-687-8990
JERRY L. PARR
DIRECTOR OF TECH. SERVICES
ROCKY MT. ANALYTICAL LAB
5530 MARSHALL ST.
ARVADA CO 80002
303-421-6611
GARY PETRAZZUOLO
TECHNICAL RESOURCES, INC.
3202 MONROE ST. SUITE 300
ROCKVILLE MD 20852
301-231-5250
-------
608
CHARLES PLEST
SENIOR CHEMIST
USEPA-OFFICE OF ACID RAIN
401 M.. STREET, SW
WASHINGTON DC 20406
202-382-5796
RICHARD POSNER
VICE PRESIDENT
UNITED STATES TESTING CO.
1415 PARK AVE.
HOBOKEN NJ 07030
201-792-2400
DAVID H. POWELL, Ph.D
MANAGER, GC/MS
ENVR. SCIENCE & ENG., INC.
P.O. BOX ESE
GAINESVILLE FL 32602-3053
904-332-3318
WILLIAM B. PRESCOTT
CONSULTANT
724 HAWTHORNE AVENUE
BOUND BROOK NJ 08805
201-469-1198
JAMES K. RICE
JAMES K. RICE CHARTERED
17415 BATCHELLORS FOREST RD
OLNEY MD 20832
301-774-2210
ROSS K. ROBESON
DIRECTOR, ENVR. MARKETING
COMPUCHEM LABORATORIES
P.O. BOX 12652
RESEARCH TRIANGLE PARK NC 2770!
919-549-8263
ANN ROSECRANCE
ORGANIC LABORATORY DIRECTOR
JTC ENVR. CONSULTANTS, INC.
4 RESEARCH PLACE
ROCKVILLE MD 20850
301-921-9790
DARCEY ROSENBLATT
TECHNICAL RESOURCES, INC.
3202 MONROE ST. SUITE 300
ROCKVILLE MD 20852
301-231-5250
CURTIS ROSS
CENTRAL REGIONAL LAB DIR.
USEPA-REGION V
536 SO. CLARK ST.
CHICAGO IL 60604
312-353-8370
DALE R. RUSHNECK
INTERFACE INC.
P.O. BOX 297
FT. COLLINS CO 80522-0297
303-223-2013
EDWARD H. SANDERS, Ph.D
TECHNICAL MANAGER
UBTL, INC
520 WAKARA WAY
SALT LAKE CITY UT 84108
801-584-3232
ROBERT B. SCHAFFER
VICE PRESIDENT
CENTEC CORP.
11260 ROGER BACON DR.
RESTON VA 22090
703-471-6300
-------
609
ANDREW SCHKUTA
MANAGER, MASS SPECT. SERVICE
CAMBRIDGE ANALYTICAL ASSOC.
1106 COMMONWEALTH AVE.
BOSTON MA 02215
617-232-2207
RITA MARIE SCHMON-STASIK
SUPV. OF GAS CHROMATOGRAPHY
PRINCETON AQUA SCIENCE/IT
165 FIELDCREST AVE., CN7809
EDISON NJ 08818
201-225-2000
JUDITH W. SCOTT
CHEMISTRY DEPARTMENT
TRW
ONE SPACE PARK, 01/2161
REDONDO BEACH CA 90278
213-535-6886
JANICE L. SEARS
PROJECTS ADMINISTRATOR
CENTEC CORPORATION
11260 ROGER BACON DRIVE
RESTON VA 22090
703-471-6300 EXT.260
ANNETTE SIMON
CHEMIST
OCCIDENTAL CHEMICALS
LONG ROAD
GRAND ISLAND NY 14072
716-773-8655
MARGARET S. SLEEVI
CAL LAB EAST, INC.
2240 DABNEY ROAD
RICHMOND VA 23230
804-359-1900
TIMOTHY O. SLOAN
SR. ORGANIC CHEMIST
ROGERS & CALLCOTT ENG.
718 LOWNDES HILL RD.
GREENVILLE SC 29607
803-232-1556
JAMES S. SMITH, Ph.D
CHEMIST
WALTER B SATTERTHWAITE ASSOC
720 N. FIVE POINTS ROAD
WEST CHESTER PA 19380
215-692-5770
STEPHEN B. SMITH
VICE PRESIDENT, R&D
ENVIRITE ANALYTICAL SERVICES
OLD WATERBURY ROAD
THOMASTON CT 06787
203-283-8235
DAVID N. SPEIS
MANAGER, CHROMATOGRAPHY
ETC CORP.
P.O. BOX 7808
EDISON NJ 08818-7808
201-225-6759
DAVID E. SPLICHAL
CHEMIST
WILSON LABORATORIES
525 NORTH 8TH ST.
SALINA KS 67401
913-325-7186
GEORGE H. STANKO
STAFF RES. CHEMIST
SHELL DEVELOPMENT CO.
P.O. BOX 1380
HOUSTON TX 72251-1380
713-493-7702
-------
610
WILLIAM STEINHAUER
PRINCIPAL RESEARCH SCIENTIST
BATTELLE MEMORIAL LABORATORY
397 WASHINGTON STREET
DUXBURY MA 02332
617-934-5682
JAMES STEPHENSON
CAL LAB EAST, INC.
2240 DABNEY ROAD
RICHMOND VA 23230
804-359-1900
RON C. STITES
GENERAL MANAGER
CENREF LABS
P.O. BOX 68
BRIGHTON CO 80601
303-659-0497
BRUCE O. STRASSER
RESEARCH SCIENTIST, R&D
UNION CAMP CORP., R&D
P.O. BOX 3301
PRINCETON NJ 08543-3301
609-891-1200
DONALD SUMLIN
LABORTORY DIRECTORY
CITY OF ATLANTA
2640 JONESBORO RD., SE
ATLANTA GA 30316
404-627-1222
ROBERT SWAIN
MARKETING ENGINEER
FINNIGAN CORP.
355 RIVER OAKS PARKWAY
SAN JOSE CA 95134
408-433-4800
WILLIAM A. TELLIARD
CHIEF, ENERGY & MINING BRANC
USEPA-ITD
401 M STREET, SW, (WH-552)
WASHINGTON DC 20460
202-382-7131
DR. P. MICHAEL TERLECKY
VICE PRESIDENT
FRONTIER TECHNICAL ASSOC.
8675 SHERIDAN DRIVE
BUFFALO NY 14221
716-634-2293
KURT THEURER
ALLIED CORP.
BOX 1021R-CRL
MORRISTOWN NJ 07960
201-455-2141
KATHLEEN E. THRUN
ARTHUR D. LITTLE, INC.
ACORN PARK
CAMBRIDGE MA 02140
617-864-5770 EXT.2311
LIAL F. TISCHLER
PARTNER
TISCHLER/KOCUREK
116 EAST MAIN ST.
ROUND ROCK TX 78664
512-244-9058
SAMUEL TO
USEPA-ITD
401 M STRETT, SW, (WH-552)
WASHINGTON DC 20460
201-475-8322
-------
611
DAVID TOMPKINS
CENTEC ANALYTICAL SERVICES
2160 INDUSTRIAL DRIVE
SALEM VA 24153
703-387-3995
ALLAN TORDINI
ASST. VICE PRESIDENT
UNITED STATES TESTING CO.
1415 PARK AVE.
HOBOKEN NJ 07030
201-792-2400
DONALD P. TREES
SAMPLE CONTROL CTR. MANAGER
VIAR AND COMPANY
300 N. LEE ST., SUITE 200
ALEXANDRIA VA 22314
703-683-0885
JEFFEREY TROIANO
SR. ENVIRONMENTAL ENGINEER
FORD MOTOR COMPANY
48229 MALLARD
DEARBORN MI 48047
313-322-3890
VICTOR TUROSKI
JAMES RIVER CORPORATION
1915 MARATHON AVENUE
NEENAH WI 54956
414-729-8168
DR. JACK TUSCHALL
SR. PROJECT SCIENTIST
NORTHROP SERVICES INC.
P.O. BOX 12313, 2 TRIANGLE DR.
RESEARCH TRIANGLE PARK NJ 27709
919-549-0611
JOSEPH VIAR, JR.
PRESIDENT
VIAR AND COMPANY
300 N. LEE ST., SUITE 200
ALEXANDRIA VA 22314
703-683-0885
FRANK VINCENT
JAMES RIVER CORPORATION
1915 MARATHON AVENUE
NEENAH WI 54956
414-729-8168
DAVID L. VINCI
SR. ENVR. CHEMIST
WESTCHESTER CTY. LAB DEPT.
HAMMOND HOUSE ROAD
VALHALLA NY 10595
914-347-3155
JOSEPH VITALIS
CHEMICAL ENGINEER
USEPA-ITD
401 M ST., SW (WH-552)
WASHINGTON DC 20460
202-382-7172
DR. DALLAS WAIT
VICE PRESIDENT
ERCO/A DIVISION OF ENSECO
205 ALEWIFE BROOK PARKWAY
CAMBRIDGE MA 02138
617-661-3111
TONIE WALLACE
PRESIDENT
COUNTY COURT REPORTERS, INC
30 SOUTH CAMERON ST.
WINCHESTER VA 22601
703-667-0600
-------
612
BRUCE K. WALLIN, Ph.D
TECHNICAL DIRECTOR
E.G. JORDON CO.
562 CONGRESS ST., P.O. BOX 7050
PORTLAND ME 04112
207-775-5401
STAN WEST
CHIEF CHEMIST
SHERRY LABS
2203 S. MADISON ST.,P.O. BOX 2i
MONCIE IN 47302
317-747-9000
STUART A. WHITLOCK
ASSOCIATE VICE PRESIDENT
ENVR. SCIENCE & ENG., INC.
P.O. BOX ESE
GAINESVILLE FL 32609
904-376-0056
BRUCE E. WILKES
PRESIDENT
ENVR. ANALYTICAL CONSULTING
5176 CYRSTAL DRIVE
CROSS LANES WV 25313
304-776-2730
TOM WILSON
SENIOR CHEMIST
IT CORPORATION
5815 MIDDLESROOK PIKE
KNOXVILLE TN 37921
615-588-6401
HUGH WISE
ENVIRONMENTAL SCIENTIST
USEPA-ITD
401 M. ST., SW, (WH-552)
WASHINGTON DC 20460
202-382-7177
N. LEE WOLFE
USEPA-ATHENS E.R.L.
COLLEGE STATION ROAD
ATHENS GA 30613
404-546-3185
MARK W. WOOD
SR. LABORATORY SUPERVISOR
PPG INDUSTRIES, INC.
P. 0. BOX 1000
LAKE CHARLES LA 70602
318-491-4450
ROBERT WOODS
GC/MS SUPERVISOR
ANALYTICAL TECH., INC.
225 WEST 30TH ST.
NATIONAL CITY CA 92050
619-477-4173
LAUREN M. YELLE
MASS SPECTRONMETRIST
ARTHUR D. LITTLE, INC.
15 ACORN PARK
CAMBRIDGE MA 02140
617-864-5770 EXT.2586
-------