Development of Pooled Method Detection Limits (MDLs) and 

Minimum Levels of Quantitation (MLs) for EPA Method 1668C.   

May 18th, 2010 

Introduction

This document describes how EPA used MDL data from three (3)
laboratories to produce pooled MDLs and MLs for EPA Method 1668C.

Method detection limit (MDL) data were received from three laboratories.
 LAB 1’s MDLs and LAB 2’s MDLs were from spikes of each congener
into water at a concentration of 10 pg/L; Lab 3’s MDLs were from
spikes of various congeners into water at concentrations of 50 to 150
pg/L.  LAB 1’s MDLs were from analyses run in 2009; LAB 2’s MDLs
were from analyses run in 2005; Lab 3's MDLs were from analyses run in
2006. 

Table 2 of Method 1668 (Revisions A and B) list retention times for
individual congeners and for coeluted congener groups (coelutions). 
Coelutions exist because the gas chromatographic (GC) column does not
resolve all 209 congeners; in fact, and in spite of repeated attempts,
no GC column resolves all 209 congeners.  Because of slight differences
in GC columns, LAB 2’s and Lab 3's coelutions are different than LAB
1’s coelutions and those in Table 2 of Method 1668 (LAB 1’s
coelutions and those in Method 1668 agree).  The difference in LAB 2’s
and Lab 3's coelutions necessitated deconvolution of LAB 2’s and Lab
3's MDLs in coelutions that differed from the coelutions in Table 2 of
Method 1668.

The process of producing pooled MDLs and MLs involved three steps:

1)  Deconvoluting LAB 2’s and Lab 3's MDLs in coelutions that did not
match the coelutions in Table 2 of Method 1668.

2)  Combining the individual congener MDLs deconvoluted from LAB 2’s
and Lab 3's coelutions into MDLs for the coelutions in Table 2 of Method
1668, and

3)  Combining LAB 1, LAB 2, and Lab 3 individual MDLs and MDLs for the
coelutions in Table 2 of Method 1668 to produce pooled MDLs and MLs for
individual congeners and coelutions.

The MDLs and MLs calculated by these three steps are in Table 2 of
Method 1668C.

Determination of Laboratory MDLs for individual Congeners

For MDLs determined for an individual congener (i.e., a congener not
involved in a coelution), the laboratory-reported MDLs were not
adjusted; however, all applicable calculations were reviewed to confirm
that the MDL procedure was followed correctly.

Deconvolution of LAB 2’s and Lab 3's coelutions

LAB 1’s 2009 MDLs are listed in Table 1; LAB 2’s 2005 MDLs are
listed in Table 2; and Lab 3's 2006 MDLs are listed in Table 3.  For all
three labs, MDLs were reported for at least one group of two or more
coeluting congeners.  A spike of each of the 209 congeners at a constant
concentration (e.g., 10 pg/L) results in combined spikes 2 to 6 times
the concentration of the individual congeners (e.g., 20 - 60 pg/L for a
spike of 10 pg/L for an individual congener) because 2 to 6 congeners
elute from the GC as a single peak (see Table 2 of Method 1668 for the
coelutions).  MDLs for these coelutions would be expected to be 2 to 6
times higher than for an individual congener that elutes as a single
peak.  However, in reviewing MDLs from the three labs, it was noticed
that the MDL for a coelution was not consistently higher by the multiple
of the number of congeners in the coelution.  Therefore, to establish
individual MDLs for the LAB 2 and Lab 3 coeluting congeners that were
inconsistent with Table 2 coelutions, and to not underestimate the MDL
for an individual congener in a coelution, we did the following.  The
MDL for an individual congener in a coelution was estimated as the
greater of the MDL for the coelution divided by the number of congeners
in the coelution, or the mean of the MDLs for four individual congeners,
two of which eluted most closely before, and two of which eluted most
closely after, the coelution.

For example, for coeluted congeners 100/93, LAB 2’s MDL was 7.6 pg/L. 
Dividing by 2 to produce an estimate for an individual MDL for each
congener yields 3.8 pg/L.  However, the mean of the MDLs for congeners
94, 95, 84, and 89 (the 2 individual congeners that elute most closely
before and the 2 individual congeners that elute most closely after the
coelution) was 5.7 pg/L.  Therefore, we set the MDL for each congener
(100 and 93) to 5.7 pg/L.

LAB 2 and Lab 3 coelutions that required deconvolution into individual
congeners, and the choice of deconvolution (divide by the number of
congeners or the average of the 4 most closely eluted individual
congeners) are listed in Tables 4 and 5, respectively.

Determination of Laboratory MDLs for Coelutions

Once MDLs were determined for each individual congener for LAB 2’s and
Lab 3's coelutions inconsistent with the coelutions in Table 2 of Method
1668, the individual MDLs were added to form MDLs for the coelutions in
Table 2 of Method 1668.  LAB 1’s original coelutions already
corresponded to the coelutions in Method 1668; therefore, LAB 1’s MDLs
for the coelutions in Table 2 of Method 1668 were used as received.

Calculation of pooled MDLs and MLs

Pooled MDLs were calculated for each individual congener and each
coelution by first calculating a pooled standard deviation.  If
possible, this pooled standard deviation was calculated from the
standard deviations in data from the three labs.  In cases where an MDL
was determined by deconvolution, the standard deviation was
“back-calculated” by dividing the MDL by the 99th percentile
t-statistic with the appropriate degrees of freedom (i.e., the number of
replicates analyzed by LAB 2 minus 1 (7 - 1 = 6)).  The pooled standard
deviation was then calculated by weighting the labs' standard deviations
by the number of replicates minus 1 (so LAB 1's standard deviation was
weighted more because LAB 1 analyzed 9 replicates whereas LAB 2 and Lab
3 analyzed 7).  This additional weighting resulted in a slight increase
in the pooled MDL above lab MDLs that were identical.

To form minimum levels of quantitation (MLs), the pooled MDLs were
multiplied by 3.18 and rounded to the number nearest to 1, 2, or 5 x 10
n, where n is 0 or an integer, per the ML procedure (see 68 FR 11790). 
MDLs and MLs for other media and MLs in the final extract were
calculated by taking into account the sample and extract volumes.  The
resulting MDLs and MLs were then inserted in Table 2 of Method 1668C.

MDLs for Future Coelutions Different from Those in Table 2 of Method
1668C

If, in the future, a laboratory has coelutions different from those in
Table 2 of Method 1668C, the laboratory can use the information in this
document to deconvolute its coelutions to obtain MDLs for individual
congeners in a coelution, and reconstitute an MDL for a Method 1668
Table 2 coelution by adding the pertinent individual MDLs.

Table 1   LAB 1's 2009 MDLs (pg/L)*

Congener	MDL

Congener	MDL

Congener	MDL

Congener	MDL

1	3.9

48	3.3

108/124	4.0

164	3.4

2	2.5

69/49	8.1

107	2.1

165	2.4

3	3.8

50/53	5.0

110/115	13.2

167	2.7

4	2.8

52	5.6

111	4.4

169	2.9

5	3.0

54	3.1

112	5.1

170	4.7

6	1.5

55	2.4

114	2.2

171/173	6.3

7	1.9

56	3.1

118	6.0

172	3.8

8	3.6

57	1.8

120	3.1

174	5.7

9	2.1

58	2.5

121	4.7

175	4.1

10	2.1

59/62/75	8.4

122	1.8

176	3.6

11	36.4

60	2.4

123	2.4

177	4.7

12/13	5.3

61/70/74/76	9.6

126	2.1

178	5.0

14	2.2

63	2.8

127	2.0

179	4.8

15	4.0

64	3.5

128/166	3.3

180/193	12.7

16	3.2

66	4.9

138/163/129/160	16.1

181	4.3

17	3.7

67	3.1

130	2.1

182	4.5

30/18	7.5

68	3.0

131	2.7

183/185	8.4

19	3.3

72	2.2

132	3.5

184	2.8

28/20	5.8

73	3.1

133	3.0

186	5.3

21/33	4.0

77	2.0

134/143	3.5

187	11.7

22	3.1

78	2.2

151/135/154	15.8

188	2.2

23	3.2

79	2.4

136	3.3

189	2.0

24	2.8

80	2.3

137	2.9

190	2.4

25	1.8

81	2.0

139/140	3.9

191	3.4

26/29	2.7

82	4.3

141	5.2

192	2.5

27	3.5

83/99	10.6

142	2.1

194	2.3

31	4.5

84	3.8

144	3.6

195	2.7

32	1.5

117/116/85	8.7

145	3.8

196	4.0

34	2.5

109/119/86/97/125/87	22.8

146	3.8

197/200	7.7

35	3.6

88/91	6.6

147/149	20.8

198/199	5.1

36	2.2

89	4.3

148	3.2

201	2.8

37	2.4

113/90/101	19.6

150	3.2

202	3.5

38	1.5

92	4.6

152	3.5

203	2.5

39	2.4

95/100/93/102/98	19.2

153/168	19.9

204	3.1

41/40/71	8.5

94	3.7

155	3.0

205	2.2

42	4.7

96	3.6

156/157	4.1

206	1.8

43	5.0

103	2.7

158	1.7

207	1.6

44/47/65	10.4

104	3.2

159	1.7

208	3.3

45/51	4.6

105	3.4

161	1.9

209	2.6

46	2.6

106	4.1

162	2.9



	*From 9 replicate analyses of a spike of each of 209 congeners into
water at a concentration of 10 pg/L.

Table 2   LAB 2's 2005 MDLs (pg/L)*

Congener	MDL

Congener	MDL

Congener	MDL

Congener	MDL

1	6.7

49/69	8.5

107	3.7

165	4.1

2	4.2

50/53	9.2

108/124	22.9

167	5.0

3	6.4

52	7.5

110/115	7.2

169	3.6

4	10.4

54	4.7

111	3.4

170	2.9

5	4.6

55	6.1

114	4.7

171/173	7.8

6	6.6

56	5.0

118	6.3

172	3.4

7	3.9

57	4.6

120	3.4

174	6.5

8	8.6

58	2.8

121	3.4

175	5.6

9	4.6

59/62/75	11.6

122	4.6

176	2.2

10	7.3

60	4.8

123	5.0

177	2.2

11	36.4

61/70/74/76	23.8

126	2.2

178	2.9

13/12	20.4

63	4.8

127	6.6

179	2.5

14	5.8

64	5.0

128/166	10.8

180/193	7.8

15	10.8

66	12.0

129/138/163	8.9

181	5.4

16	8.6

67	5.7

130	8.7

182	3.6

17	10.9

68	3.9

131	1.3

183/185	4.3

18/30	11.5

72	3.7

132	4.6

184	3.3

19	9.7

77	4.4

133	2.7

186	4.2

20/28	16.6

78	4.4

134/143	10.4

187	4.5

21/33	12.6

79	3.2

135/151	6.3

188	4.3

22	9.9

80	3.6

136	3.4

189	2.8

23	3.2

81	3.4

137/164	3.5

190	2.5

24	11.2

82	8.3

139/140	4.4

191	3.1

25	7.7

83	9.3

141	3.8

192	3.7

26/29	9.0

84	6.0

142	4.4

194	5.0

27	5.6

85/116/117	8.4

144	5.5

195	6.2

31	10.1

86/87/97/109/119/125	10.5

145	3.1

196	6.2

32	5.7

88/91	7.4

146	4.9

197/200	5.6

34	3.4

89	5.6

147/149	4.5

198/199	13.0

35	9.6

90/101/113	4.7

148	5.0

201	4.3

36	7.5

92	3.7

150	3.4

202	3.9

37	9.0

93/100	7.6

152	2.3

203	4.9

38	4.7

94	4.5

153/168	7.1

204	3.1

39	7.3

95	6.8

154	4.9

205	5.5

40/41/71	6.5

96	2.6

155	3.2

206	3.2

42	4.0

98/102	12.1

156/157	4.5

207	2.7

43/73	9.3

99/112	17.7

158	2.5

208	3.5

44/47/65	10.7

103	2.5

159	3.4

209	2.5

45/51	12.1

104	5.7

160	7.2



	46	2.6

105	4.4

161	2.6



	48	2.6

106	5.8

162	4.1



	*From 7 replicate analyses of a spike of each of 209 congeners into
water at a concentration of 10 pg/L.

Table 3   Lab 3's 2006 MDLs (pg/L) *

Congener	MDL

Congener	MDL

Congener	MDL

Congener	MDL

1	17.1

65/47/44	71.2	 	111	24.8	 	169	26.1

2	11.2

62/75/59	64.8	 	120	24.0	 	188	26.3

3	18.6

42	28.7	 	108/124	47.7	 	179	25.2

4	20.5

41/71/40	75.6	 	107/123/106	88.8	 	184	25.4

10	22.7

64	22.0	 	118	53.0	 	176	22.1

9	12.1

72	22.4	 	122	21.7	 	186	25.3

7	13.7

68	24.2	 	114	26.3	 	178	24.4

6	11.8

57	20.3	 	105	30.3	 	175	25.2

5	14.3

58	26.0	 	127	24.3	 	187	26.0

8	25.4

67	21.7	 	126	29.9	 	182	22.7

14	12.4

63	20.9	 	155	25.2	 	183/185/174	75.0

11	20.4

61/70/76/74	104.0	 	152	25.3	 	177	19.4

13/12	27.2

66	27.0	 	150	26.8	 	181	23.2

15	27.5

55	21.3	 	136	29.0	 	171/173	54.0

19	10.2

56	27.2	 	145	28.2	 	172	23.3

30/18	25.9

60	25.8	 	148	25.2	 	192	23.5

17	11.2

80	19.8	 	151/135/154	80.0	 	193/180	52.6

27/16/24	39.0

79	23.8	 	144	25.6	 	191	23.7

32	13.1

78	28.6	 	147/149	57.3	 	170	21.4

34	11.2

81	32.8	 	134/143	58.3	 	190	24.6

23	10.8

77	24.9	 	139/140	52.0	 	189	23.8

29/26	20.2

104	23.9	 	131	31.5	 	202	44.2

25	13.2

96	27.0	 	142	29.7	 	201	35.9

31	31.5

103	20.2	 	132	29.4	 	204	37.9

28/20/21/33	72.3

94	22.4	 	133	22.0	 	197/200	77.2

22	13.4

95	36.5	 	165	22.6	 	198/199	65.2

36	13.2

100/93/102/98	85.7	 	146/161	48.0	 	196	35.4

39	13.3

88/91	38.9	 	153/168	49.0	 	203	32.7

38	12.3

84	18.1	 	141	29.9	 	195	38.7

35	13.7

89	21.5	 	130	22.7	 	194	32.7

37	15.0

121	20.7	 	137/164	50.2	 	205	27.5

54	25.7

92	23.6	 	138/163/129/160	112.5	 	208	28.2

50/53	44.0

113/90/101	82.9	 	158	28.7	 	207	33.7

45/51	38.1

83/99/112	66.4	 	166/128	52.1	 	206	28.7

46	18.6

119/109/86/97/125/87	131.7	 	159	24.8	 	209	28.8

73/52/43	69.1

117/116/85	68.1	 	162	22.8



	69/49	45.0

110/115	68.4	 	167	23.2



	48	25.8

82	26.4	 	156/157	42.3



	*From 7 replicate analyses of spikes into water of mono- thru tri-
chloroCBs at 50 pg/L; tetra- thru hexa- chloroCBs at 100 pg/L and hepta-
thru deca- chloroCBs at 150 pg/L.

Table 4   Individual Test America-Knoxville MDLs deconvoluted from MDLs
for coelutions that do not match coelutions in Table 2 of Method 1668A

MDL for an individual congener was determined as the larger of the MDL
for the coelution divided by the number of congeners in the coelution,
or the mean of the MDLs for four individual congeners, two of which
eluted most closely before, and two of which eluted most closely after,
the coelution

Coelution	Congener numbers of congeners averaged

100,93	94,95,84,89 (means that MDLs for these 4 congeners were averaged)

102,98	/2 (means that the MDL for these congeners was divided by 2)

99,12	/2

151,135	145,148,154,144

137,164	141,130,160,158

138,163,129	141,130,160,158



Table 5   Individual Lab 3-Columbus MDLs deconvoluted from MDLs for
coelutions that do not match coelutions in Table 2 of Method 1668A

MDL for an individual congener was determined as the larger of the MDL
for the coelution divided by the number of congeners in the coelution,
or the mean of the MDLs for four individual congeners, two of which
eluted most closely before, and two of which eluted most closely after,
the coelution

Coelution	Congener numbers of congeners averaged

27,16,24	/3 (means that the MDL for these congeners was divided by 3)

28,20,21,33	/4

52,73,43	54,46,48,42 (means that MDLs for these 4 congeners were
averaged)

95,100,93,102,98	94,95,84,89

83,99,112	121+92+82+111

107,123,106	111,120,118,122

146,161	133,165,141,130

137,164	141,130,158,159

183,185,174	/3



Engineering and Analysis Division		Brian Englert

Office of Science and Technology		202-566-0754

05/18/2010

US EPA Engineering and Analysis Division		Brian Englert

Office of Science and Technology		May 18, 2010

 PAGE   2 

 PAGE   7 

