2502
Anal. Chem. 1'903, 65, 2502-2509
Electron-Acceptor and Electron-Donor Chromatographic Stationary Phases for the Reversed-Phase Liquid Chromatographic Separation and Isomer Identification of Polychlorinated Dibenzo-pdioxins Kazuhiro KimataJ Ken Hosoya,t Takeo ArakiJ Nobuo Tanaka:*+ Elizabeth R. BarnharttJ Louis R. Alexander3 Sarath Sirimanne3 Patricia C. McClure3 James Grainger? and Donald G. Patterson, Jr.t Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606, Japan, and Centers for Disease Control and Prevention, Division of Environmental Health Laboratory Sciences, 4770 Buford Highway, NE, Atlanta, Georgia 30341 -3724
To assess the impact of environmental toxicants on human health, researchers routinely use GC/ MS, and the compounds they use for analytical reference standardsmust be synthesized,purified, and characterized. Reversed-phase HPLC separation and structure identification of polychlorinated dibenzo-pdioxin (PCDD) isomers (potentially hazardous environmentalcontaminants) in synthesis reaction mixtures were accomplishedby taking advantage of the different retention mechanismsat work on four different stationary phases bonded to silica gel. Hydrophobic, charge-transfer, and dipole-dipole interactions characterize the analyte elution orders of the positional isomers of PCDDs from columns packed with C I ~pyre, nylethyl, (nitrophenyl)ethyl,and (nitrophenoxy)propyl stationary phases.
INTRODUCTION
Table 1. Properties of Stationary Phases phase Cis PYE NPE NPO
elemental a n a l . O C (%) N (%)
19.17 18.45 8.81 10.48
0 0 1.02d 1.03
surface
separation factorb (k') (pmol/m2) CHa COOCHs (C&)C
3.22 2.99 2.56 2.89
1.96 1.83 1.61 1.57
0.80
1.86 1.42 1.39
(3.36) (1.86) (0.91) (1.06)
a Prior to trimethylsilylation. The k ' ratio between toluene and benzene, and between methyl benzoate and benzene, in methanolwater (6040). The k' value of benzene in methanol-water (60:40). d Corresponds to 0.99 NO2 group per phenyl group.
and 1,2,4,6,7,9-/1,2,4,6,8,9-HxCDDl. Further research on the separations and identification for these congeners and for dichlorodibenzo-p-dioxins(DCDDs) and trichlorodibenzop-dioxins (TrCDDs),about which little information has been published, seems warranted. Researchers have attempted to characterize PCDD mixtures by "-NMR and FTIR spectral interpretation,3*4but complications arose when quantitative ratios approached unity. This is the problem for mixtures of isomers with the highest degrees of similarity. Contradicting and ambiguous structure assignments for the elution order of the isomers have been reported even after determinations with the liquid crystal stationary phase.2 The isolation of individual congeners of the chlorodioxins is indispensible to provide pure analytical reference standard compounds for determinations of equivalency in teratogenicity, toxicity, carcinogenicity, and other human health effects.' When PCDD reference compounds are being prepared, each one of an isomer pair, coproduced by the Smiles rearrangement in synthesis,6*6must be purified. Until recently, isolation of pure isomers was not possible, even with the highly efficient polar GC capillary columns. For preparative-scale operations, HPLC with silica and, in reversedphase mode, with octadecyl- (CIS) bonded phase provided different selectivities and investigatorsused them frequently74
Separating polychlorinated dibenzo-p-dioxin(PCDD) congeners by GC and HPLC is critical to determining human health hazards in environmental science. Analysis of human samples relies mainly on GC separations with MS detection in the select-ion recording mode.1 However, unequivocal identification of all the GC peaks of PCDD congeners has not been feasible on the columns ordinarily used.2 From one comprehensive study there is a report of GC separations of all the congeners with four or more chlorine substituents on nine different GC stationary phases.2 In that study, selectivity was shown to increase with increasing polarities of the stationary phases. Use of a liquid crystalline stationary phase effected the separation of all the isomers, including those that are inseparable on more robust, conventional GC column preparations. However, an unequivocal assignment is yet needed for identities of individual isomers of six pairs of PCDDs [1,2,4,6-/1,2,4,9-tetrachlorodibenzo-p-dioxin (TCDD); 1,2,4,7-/1,2,4,8-TCDD; 1,2,4,6,7-/1,2,4,8,9-pentachlorodibenzo-p-dioxin (PnCDD); 1,2,4,6,8-/1,2,4,7,9-PnCDD; (3)Gelbaum, L. T.; Patterson, D. G., Jr.; Ashley, D.; Groce, D. F. 1,2,3,6,7,9-/1,2,3,6,8,9-hexachlorodibenzo-p-dioxin (HxCDD); Chemosphere 1988,17,551-558.
(4)Grainger, J.; Gelbaum, L. T. Appl. Spectrosc. 1987,41,-820. (5)Kende, A. S.;Decamp, M. R. Tetrahedron Lett. 1976,2875-2880. (6)Koester, C. J.; Hites, R. A. Chemosphere 1988,17,2355-2362. t Centers for Disease Control and Prevention. (7)Nestrick, T.J.; Lamparski, L. L.; Stehl, R. H. Anal. Chem. 1979, (1)Rappe, C., Buser, H. R., Dodet, B., ONeill, I. K.,Eds.Enuironmental Carcinogens Methods of Analysis and Exposure Measurement, Vol. 11, 51,2273-2281. (8)Taylor,M.L.;Tiernan,T.O.;Ramalingam,B.;Wagel,D.J.;Garrett, Polychlorinated Dioxins and Dibenzofurans; International Agency for J. H.; Solch, J. G.; et al. In Chlorinated Dioxins and Dibenzofurans in Research on Cancer: Lyon, France, 1991;pp 31-50. (2)Ryan, J. J.;Conacher,H.B.S.;Panopio,L.G.;Lau,B.B.Y.;Hardy, the Total Environment; Keith, L. H., Rappe, C., Choudhary, G., Eds.; Butterworth Boston, 1985;Vol. 11, Chapter 2,pp 17-35. J. A.; Masuda, Y. J. Chromatogr. 1991,541,131-183. t Kyoto Institute of Technology.
0003-2700/83/0365-2502$04.00/0
0 1983 American Chemlcal Society
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
2503
c N P E
5
0
3
d NPO 3
p y E 1 5
,
1
5
0
#in
10
10
15
20
25
5
Bin
Flgurr 1. Reversdphase chromatographic patterns showing differences in retentlon of stationary phases for naphthalene derivatives: peaks, (1) l,8dlnltro, (2) 1,5dinltro, (3) naphthalene, (4) 1-methyl, and (5) 1,5dimethyl; stationary phases, (a) Cle, (b) PYE, (c) NPE, and (d) NPO. 1.0
c
x
1,2.4
x
0 1.2,3
0.5
2 0
1,4 1,2
1,3 1.2
-0.2 0
0
I
3
2
1,4
2
4
Number o f C 1
3
4
Number o f C 1
P Y E
1.01
x
x
m
5
1
1.2.3
m O
r 1,3
2
N P E
m
0.5
0 -'
5 0
'
2
I
3
4
'
'E e 1,4
-0.2
1
2
3
4
Number o f C1 Number o f C 1 Fburr 2. OIaphS of log k'of polychlorobenrenesversus the number of chlorine substituents on four stationary phases: mobile phase, methanolwater (80:20, v/v); stationary phases, (a) Cle, (b) PYE, (c) NPE, and (d) NPO.
Currently, a fairly wide selection of bonded stationary phases for HPLC is available for use in purifying isomers. Planar hydrophobic aromatic hydrocarbons and substituenta present certain properties advantageous to differential partitioning into particular bonded stationary phases in HPLC.lOJ1 The superiorityof 2-(nitrophenyl)ethylsilyl (NPE)
and 2-(l-pyrenyl)ethylsilyl (PYE) bonded silica packings for TCDD isomer segregation has been previously deucribed.12 In this paper we report the application of electron donoracceptor stationary phases, PYE, NPE, and 3-@-nitrophenoxy)propylsilyl (NPO) bonded to silica, for the RPLC separation of all the synthesis isomer pairs of PCDDs.
(9) Barnhart, E. R.; Patterson, D. G., Jr.; Tanaka, N.; Araki, M. J. Chrornatogr. 1988,445,14&164. (10) Tanaka, N.; Tokuda, Y.; Iwaguchi, K.; Araki,M. J. Chromatogr. 1982,239,761-772.
(11) Tanaka, N.; Tanigawa, T.; Kimata, K.; Hosoya, K.; Araki, T. J. Chromatogr. 1991,549, 29-41. (12) Kimata, K.; Hoeoya, K.; Tanaka, N.; Araki, T.; Patterson, D. G. J. Chromatogr. 1992,595,77-88.
2504
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
Table 11. Y Values and Separation Factors for PCDDs on the Four Stationary Phases.
k’ PCDD 0 1 2 12/14 16/19
Cle (90% CHsOH) 1.25 1.68 2.13 2.60
17/18
2.45 (71) 2.25 (29) 11.091 2.85
27/28
3.23
13 23 1261129 1271128 1361139 1371138 123 124 146 147 178 237 123611239 123711238
3.66 3.58 3.77 (76) 3.49 (24) 11.081 4.16 (72) 4.33 (28) 11.041 5.16 (59) 4.67 (41) [1.103 3.46 5.28 4.99 3.47 4.24 4.77 5.19 7.29 (78) 6.66 (22) 11.091 7.58
124611249
6.60
124711248
7.63
126711289
PYE (CH30H)C 0.40 0.76 0.81 1.72 1.73(74) 1.57 (26) [1.101 1.74 1.76 1.91 2.02 4.41 (77) 3.87 (23) [1.14] 4.08 5.25 (59) 4.68 (41) 11.121 4.87 5.14 4.87 4.53 4.66 5.18 4.82 20.09 (80) 17.21 (20) [1.17] 15.98 (69) 15.39 (31) 11.041 17.55 16.47 (61) 15.77 (39) 11.051 13.48 (80) 11.27 (20) 11.201 16.21 (53) 14.40 (47) 11.131 21.63 (65) 17.81 (35) 11.211 18.34 19.00 19.54 20.89 15.51 14.70 21.15 17.10 2.56 (68) 2.27 (32) 11.131 3.03 (62) 2.80 (38) r1.081 2.51
12467112489
5.51 (80) 5.24 (20) 11.051 7.40 (53) 6.74 (47) L1.101 9.64 (64) 8.70 (36) [1.11] 8.21 8.55 4.98 6.94 5.25 6.80 6.88 7.71 10.02 (67) 9.36 (33) 11.071 12.89 (57) 11.72 (43) F1.101 9.36
12468/12479
12.10
2.98
12346
10.85 -. ..
2.78
126811279 1368/1379 1234 1378 1469 1369 1269 1278 1478 2378 12367/12389 12368112379
NPE (80% CH30H)d
NPO (80% CH30H)d
1.10 1.78 1.91 2.73 (80) 3.02 (20) L1.111 2.64 (69) 3.29 (31) [1.24] 2.82 (69) 3.24 (31) [1.151 2.96 (76) 3.11 (24) 11.051 3.17 3.28 4.25 (76) 5.68 (24) [1.271 4.52 (69) 5.33 (31) [1.18] 4.89 (59) 5.28 (41) [1.081 4.84 (49) 5.23 (51) 11.081 5.34 4.96 4.66 4.65 5.51 5.45 7.76 (79) 9.03 (21) 11.161 7.78 (65) 8.64 (35) L1.111 7.63 (65) 8.30 (35) 11.091 7.67 (56) 7.96 (44) 11.041 6.55 (80) 9.53 (20) 11.461 7.62 (53) 8.64 (47) [1.14] 8.19
1.36 2.20 2.37 3.40 (84) 3.90 (16) [1.15] 3.22 (75) 4.24 (25) [1.32] 3.49 (69) 4.11 (31) 11.181 3.66 (68) 3.95 (32) 11.081 3.95 4.19 5.31 (78) 7.89 (22) [1.49] 5.62 (71) 6.96 (29) [1.24] 6.11 (58) 6.83 (42) 11.121 5.96 (43) 6.46 (57) i1.091 7.09 6.36 5.81 5.62 6.40 6.20 10.02 (81) 12.75 (19) 11.273 9.93 (65) 11.29 (35) 11.141 9.88 (62) 11.35 (38) 11.151 9.55 (55) 10.17 (45) 11.071 8.08 (81) 14.05 (19) 11.741 9.55 (53) 11.51 (47) [1.21] 9.91 (53) 10.28 (47) [LO41 11.98 9.71 8.85 9.51 9.95 9.93 9.21 9.85 3.93 (71) 5.28 (29) [1.34] 4.11 (57) 4.60 (43) [1.12] 4.18 (68) 4.81 (32) 11.151 4.06 (41) 4.34 (59) 11.071 5.08
8.97 8.23 7.49 7.77 7.50 8.03 7.62 8.21 3.04 (68) 3.67 (32) 11.211 3.24 (67) 3.41 (33) [1.051 3.13 (68) 3.41 (32) 11.091 3.16 (46) 3.30 (54) 11.041 3.58
dipole momente (D)
3.095 4.121 1.710 2.668 2.178 2.922 0.623 1.607 0.023 4.220 1.489 2.724 0.023 1.221 3.727 1.323 0.024 1.478 2.414 2.467 2.480 0.021 1.698 3.557 1.409 2.211 1.458 2.687 0.218 1.229 3.545
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
2505
Table I1 (Continued) PCDD
Cis (90% CHaOH)
PYE (CH3OH)O
12347 12369 12378 12469 12478 123678/123789 123679/123689
11.74 9.71 11.08 9.08 11.59 16.89 (68) 15.66 (32) 11.081 16.23
2.54 3.25 2.69 3.05 2.92 6.54 (69) 6.01 (31) 11.091 7.21
124679/124689
15.45
123467 123468 123469 123478 1234678 1234679 12346789
15.21 18.34 14.72 17.08 24.00 24.04 35.70
7.61 (78) 7.30 (22) [1.041 5.84 7.22 8.39 6.42 16.16 19.12 45.12
k’ NPE (80%CH30H)d 3.43 3.13 3.27 3.11 3.14 4.43(65) 4.90 (35) 11.111 4.39 (35) 4.65 (65) 11.061 4.38 4.84 4.73 4.58 4.64 6.53 6.26 8.62
NPO (80% CH8OH)d
dipole momente (D)
4.60 4.32 4.23 4.23 3.98 5.75 (66) 7.04 (34) 11.221 5.92 (32) 6.50 (68) 11.101 5.99 (52) 6.17 (48) 11.031 6.98 6.50 6.59 6.15 9.16 9.06 12.72
1.999 2.875 1.248 1.685 1.093 0.019 2.383 1.183 2.059 0.030 1.220 2.557 1.528 2.796 0.335 1.136 1.337 0.016
0 Flow rate, 1.0 mL/min; 2.0 mL/min with PYE. * The numbers in parentheses indicate the isomer ratios in each pair coproduced during synthesis. The numbers in brackets are separation factors. Mobile phase for PnCCD-OcCDD with PYE, dichlorornethane-ethanol(50:50). Mobile phase for PnCDD-OcCDD with NPE and NPO, methanol-water (%lo). e Reference 6.
Recognizing the types of molecular interactions involved in the retention process, we assigned a tentative structural identification to each peak (including those for DCDDs and TrCDDs). In addition, all of the isomer pairs coproduced during preparation were separated using a simple elution of isomer mixtures on the two silica columns that contained bonded aromatic electron-acceptor or -donor moieties. Identification is based on the elution order for PYE and NPE or NPO silicas together with the peak size ratio of isomers coproduced in a process involving thermodynamic equilibrium. PYE behaves as an electron donor, preferentially retaining the PCDD isomer with the more symmetrically arranged chlorine substituents on the aromatic nucleus. On the other hand, the nitroaromatic moiety exhibited the reverse tendency, preferentially retaining the more dipolar compounds, which have more proximal chlorinesubstituents. NPO selectivity proved to be superior to that of NPE. EXPERIMENTAL SECTION Safety. On the basis of animal studies, researchers have determined that PCDDs are potentially hazardous to human health and should be handled and disposed of safely.13 Columns. NPE, PYE, and Cis columns (5-pm particle-size Cosmosil5NPE,5PyE, and 5 C 3 were commercialproducts from Nacalai-Tesque, Kyoto, Japan. The NPE stationary phase on silica possesses a mixed functionalityof about 70% p - and 30% o-nitro-substituted phenylethyl groups.12 The NPO and PYE stationary phases were 3-@-nitrophenoxy)propylsilyl and 2-(1pyreny1)ethylsilylgroups,bonded onto silicaparticles with surface areas of approximately 330 m2/g. We prepared the NPO phase using 3-@-nitrophenoxy)propyldimethylchlorosilane as a silylating agent by the method previously reported for bonding Cls.14 Chemicals. All mobile phases were mixed by volume measurements from HPLC-grade solvents. Using a standard procedure that has been reported previously? we prepared PCDDs with a chlorocatechol and a chloronitrobenzene in dimethyl sulfoxide in the presence of anhydrous potassium carbonate. HPLC Equipment and Measurement. We used three systems: a Beckman gradient bioseparation with System Gold workstation, Model 126 pump and Model 166 programmable detection module, a Waters M6000 pump and LDC Spec(13)Myers,G.L.;Patterson, D. G., Jr. Prof. Saf. 1987,32(6),30-37. (14)Jinno, K.; Shimura, S.; Tanaka, N.; Kimata, K.; Fetzer, J. C.; Biggs, W. R. Chromotographia 1989,27, 285-291.
troMonitor UV detector,and a Shimadzusystem of LC-6A pump, SIL-6A automatic injector, SPD-6A UV detector, and C-R3A data processor. Column temperature was maintained at 30 O C in a constant-temperature water bath, and the elution time of acetone was used as to for the calculation of k’values. Data were corrected for dead volume, and chromatographic measurements were performed in duplicate.
RESULTS AND DISCUSSION Characterization of Stationary Phases. We previously reported the use of NPE together with PYE for the separation and structural assignment of TCDD isomer pairs.12 NPO was designed to provide better selectivity, since the nitro substituent, unlike that in NPE, is exclusively para to the phenyl-silica linkage. Table I shows the elemental analysis of NPE, NPO, PYE, and CIS,together with the k’values and the separation factors for benzene derivatives, on these stationary phases. The results indicate that the hydrophobicities of NPO and NPE are less than those of PYE or Cis, as estimated from the contribution to the retention of analytes by one methylene group. The elemental analysis showed relatively low surface coverage of NPE due to the cleavage of bonded groups with highly acidic reaction conditions during the nitration of phenylethylsilylated stationary phase. NPO can be readily prepared with higher surface coverages. The two types of aromatic stationary phases, PYE phase and nitroaromatic phase, displayed opposite retention preferences regarding aromatic compounds with electron-withdrawing substituents. Figure 1 shows the chromatograms obtained for dinitronaphthalenes with the CIS, PYE, NPE, and NPO phases. The more dipolar 1,8-dinitronaphthalene was retained longer than 1,5-dinitronaphthaleneon NPE and NPO phases with the latter phase resulting in a greater separation factor for the compounds. As shown in Figure 2,PYE displayed longer retention for chlorobenzene molecules with chlorine atoms spaced as far apart as possible, and NPE and NPO showed preferential retention for chlorobenzenes with chlorine atoms positioned as close as possible. PYE preferentially retained isomers with the more symmetrical substitution or charge distribution (which results in the more electron-deficient aromatic ring
2106
ANALYTICAL CHEMISTRY, VOL. 85, NO. 18, SEWEMBER 15, 1993
I
I
a-1
30
b-1
d-1
c-1
0
. -
1
b-2
2
N
d-2
c-2
I
0
5
a-3
10
15
20
0
(mid
2
4
6 (nin)
0
2
6
4
c-3
b-3
8 (min)
d-3
1 r
O N c l
A
n
-
b-4
a-4
d-4
-
IC
L
0
L
5
15
IO
(min)
0
5
15
10
(mid
0
2
'
6
Gin)
L
I
I
0
2
4
I
,
6 8 (.in)
ANALYTICAL CHEMISTRY, VOL. 85, NO. 18, SEPTEMBER 15, 1993 Q
a-5
b-5
N c
c
(1
N
a-7
b-7
c-5
1
2507
d-5
.w N h
c-7
d-7
Flgwo 3. Chromatographic separation of seven different mlxtures (1-7) of PCDD isomer pairs (identified in the figure) showing elution order reversal between Clo or PYE and nltroarometic stationary phases: statlonary phases, (a) Cle, (b) PYE, (c) NPE, and (d) NPO.
leading to the more favorable charge-transfer interaction). In contrast, the nitroaromatic bonded phases preferentially retained those isomers with more crowded chlorine substituents (which also have the greater dipolar character leading to the more favorable dipole-dipole interaction). Using these characteristic retention mechanisms, it appeared possible to chromatographically identify the positional structure of substituents in isomer pairs of PCDDs. PCDD Isomer Separation. Table I1 lists the k' values and peak size ratios (235 nm) obtained using the four stationary phases for the PCDD isomer pairs that were coproduced. For the f i t time, separation factors and identity assignments for all isomer pairs were possible in a single mode of chromatography,reversed-phase. Table I1 and Figure 3, showing the separation of all 75 PCDDs, will serve as
references for those persons undertaking the identification of PCDD isomers. Isomers that are separable on CU include 1,2,3,6-/1,2,3,9-TCDD;1,2,6,7-/1,2,8,9-TCDD;1,2,6,&/1,2,7,9TCDD; 1,3,6,8-/1,3,7,9-TCDD; 1,2,3,6,7-/1,2,3,8,9-PnCDD; 1,2,3,6,8-/1,2,3,7,9-PnCDD; and 1,2,3,6,7,8-/1,2,3,7,8,9-HxCDD. Because of confidence in the identities of these isomers: their chromatographiccharacteristics are discussed first. As reported previously,'2 the CISstationary phase showed the longest retention time for the PCDD isomer molecule that had the least steric crowding at positions 1and 9 or for the isomer with the most hydrophobicity (Figure3a). Chlorine substituents on the aromatic ring increase retentive forces for hydrophobicity-based mechanisms.*6 Furthermore, a greater distance between positional substituents decreases
2508
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
19
1289
123789
123678
HxCDD
CI
126
129
CI 12367
12389
Flgure 4. Time scale for elution windows of each PCW homolog provlded by NPO stationary phase with 80% methanol mobile phase.
steric hindrance, thus intensifyingthe retention effect (Figure 3). The peak size difference in each pair partially reflects the relative thermodynamic stability of the isomers and can be used for tracking the retention order on the stationaryphases. PYE showed preferential retention for those isomerswhose chlorine atoms were farthest apart (Figure 3b). This finding can be explained in terms of charge-transfer interactions and the hydrophobic retention mechanisms of PYE. The retention order on PYE is usually the same as on Cl8, yet PYE is more selective and, therefore, allowsfor the resolution of more isomer peaks. In contrast to PYE and (218, NPE showed preferential affinity for those PCDDs whose chlorine atoms were closest together. NPO demonstrated the same tendency,with slightly greater separationfactorsfor isomersthan NPE. Presumably, this outcome stems from dipoledipole interactions. We observed no separation on NPE for the 1,3,6,&/1,3,7,9-TCDDs that could be separated with the Cl8 phase. Apparently the 1,3,7,9-TCDD,the more polar isomer with a stronger dipole dipole interaction potential, is retained by NPE and NPO to about the same degree as the 1,3,6,8-TCDD isomer, which has a more hydrophobic interaction, a factor that accounts for resolution on the C18 phase. However, both mechanisms are at work in reversed-phase retention on these nitroaromatic columns. It appears that,with these stationary phases, PCDD isomers with small dipole moments6 are less separable than those isomers with larger dipole moments even with similar differences between isomers. However,we achieved a partial separation of the 1,3,6,&/1,3,7,9-TCDDisomer pair with NPO. The PCDD isomers 1,2,4,6-/1,2,4,9-TCDD;1,2,4,7-/1,2,4,8TCDD 1,2,4,6,7-/1,2,4,8,9PnCDD; 1,2,4,6,&/1,2,4,7,9-PnCDD 1,2,3,6,7,9-/1,2,3,6,8,9-H~CDD;and 1,2,4,6,7,9-/1,2,4,6,8,9HxCDD; are hard to separate by GC2and cannot be separated by C18 HPLC. We did not observe the separation of 1,2,4,6-/1,2,4,9-TCDDon Cl8 although others have reported on the prospect of such separation.l6 Cla either partially separated or failed to separate the isomers in these pairs: (16)Rekker, R. F. Hydrophobic Fragmental Constant; Elsevier: Amsterdam, 1977;Chapter 3. (16)Swerev, M.; Ballschmiter,K. Chemosphere 1986,15(9-12), 11231126.
1,2,3,7-/1,2,3,8-TCDD; 1,7-/1,8-DCDD; 2,7-/2,8-DCDD; 1,2,7-/1,2,&TrCDD;and 1,3,7-/1,3,8-TrCDD. Of the pairs inseparable on C18, PYE separated 1,2,4,7-/ 1,2,4,8-TCDD; 1,2,3,7-/1,2,3,8-TCDD; and 1,2,4,6,7,9-/ 1,2,4,6,8,9-HxCDD. In some cases, however, the retention time of PCDDs on this stationary phase was too long to produce good peak shapes. Of t h e PCDDs tested, only two isomer pairs (1,2,4,6,7,9-/1,2,4,6,8,9-HxCDD9 and 1,3,6,8-/1,3,7,9-TCDD) were not separated by NPE when we used one of two aqueous mobile phases (80%and 90% methanol). This inability to separate the pairs was due to the weak dipolar character of these isomers. NPO showed improved selectivity compared with NPE and separated all the PCDD isomer pairs with 75 % and 90% methanol. The elution order corresponded to that on NPE and was the reverse of that on PYE. There was a clear correlation of peak size and elution order, especially for those isomers easily separable with (218. The solutions of synthesis products containing these easily separable isomer pairs generally possess components in widely differing amounts, (i.e., with large peak size ratios). The smaller peak elutes first on Cl8 and PYE, and the larger peak elutes first on NPE or NPO. These orders of elution are understandable given the relative thermodynamic stability, hydrophobicity, and dipolar character of the isomers that are determined by the positions of chlorine Substituents. Because of the contribution of the Smiles rearrangement, the more thermodynamically stable isomers tend to form preferentially during the preparation reaction. More sterically unhindered (i.e., widely separated) chlorine substitution results in greater stability of these isomers. In agreement with the heat of formation predictions? such an arrangement confers on the isomer the greater hydrophobicity of the pair17 and results in the longer retention time on (218. Those isomers whose substituents are more symmetrically positioned allow more charge-transfer interactions between the isomer and stationary phase than do isomers whose substituents are positioned less symmetrically. Therefore, isomers with the more symmetrically positioned substituents will have the (17)Sierra, A. R.; Lettinga, G. Appl. Microbiol. Biotechnol. 1991,34, 644-650.
ANALYTICAL CHEMISTRY, VOL. 05, NO. 18, SEPTEMBER 15, 1993
2509
calculations of dipole moments for the di- and trichlorinated greater retention on PYE. These mechanisms account for the similar retention order on the two columns, with the larger isomers to confirm our assignments. Greater differences in peak eluting last. dipole moments between isomers facilitate separation on NPE The more thermodynamically stable isomers possess the and NPO. As stated above, selectivity diminishes with the smaller dipole moments6 because their substituents are more size of the dipole moments. This diminishment is due to the symmetricallyarranged. This arrangement lessens attraction relatively weak interactions between the analyte and the to NPE or NPO phases. Thus the less stable isomer with the stationary phase. more congested (and asymmetric) substituent arrangement (i.e., the smaller peak) is the isomer that elutes later on these Although isomer elution order assignments differ for some two columns. This tendency of the smaller peak to elute isomersthat were reported for the GC liquid crystal stationary later was also observed on polar and liquid crystal GC phase in which arbitrariness existed, the retention order on stationary phases.2 NPE and NPO and the dipolar character of each isomer All members of each isomeric group of PCDD congeners suggest that the present assignment is correct. elute between the most crowded and least crowded chlorine substitution on NPO where dipole-dipole interaction mechGenerally, the production of the more thermodynamically anisms seem to be the most dominating forces in retention stable isomer correlates with the contribution of the Smiles (Figure 4). For example, all TCDDs elute between the two rearrangement. The correlation between peak size and peaks of 1,2,6,7-and 1,2,8,9-TCDD. Similarly,DCDDs appear retention order, however, may not hold strictly for every between 1,6-and 1B-DCDD;TrCDDs between 1,2,6-and 1,2,9isomer pair. Peak size ratios can vary as a result of workup TrCDD; PnCDDs between 1,2,3,6,7- and 1,2,3,8,9-PnCDD; of crude products. Therefore, the small deviations in and HxCDDs between 1,2,3,6,7,8- and 1,2,3,7,8,9-HxCDD. correlation of peak size ratios with retention order should not The observations from these nitroaromatic columns support be taken to indicate any ineffectiveness of the present scheme the interpretation of the present results based on the dipoleof isomer structure assignment for HPLC peaks. Even when dipole interactions. separation is not attained on the nitroaromatic phases, the Structure Identity Assignment. In most cases where relative retentions of the two isomers on PYE and NPE or separation was achieved with PYE, the elution order was NPO permit assignment, on the basis of either the spatial reversed on the nitroaromatic phases. Our results suggest symmetry of substituents and their effects on charge-transfer the possibility of isomer identification based on the elution interaction or the asymmetrical substitution resulting in order of PCDDs separated on either of these phases. Because dipole-dipole interaction. all the isomers could be separated on NPO, we were able to provide tentative structure assignments for all of the isomers The NPO phase seems to afford PCDD isomer distinction (later in this section), including those undetermined in the in HPLC similar to that seen with the liquid crystal stationary comprehensive GC analyses.2 As explained below, molecular phase in GC. The NPO column should be tested further for structures corresponded to identified peak elution orders on effective purification, analysis, and the isomer structure the phase. assignment of many other compounds by HPLC. Although Although not confirmed by other means, the earliest eluting there is little in the literature about GC separation of DCDDs (and larger) peaks in the separation of 1,6-/1,9-DCDD and and TrCDDs, such information would enhance clarification 1,2,6-/1,2,9-TrCDDon NPO and NPE, or the second eluting of HPLC data. peaks on &and PYE, should be 1,B-DCDDand 1,2,6-TrCDD, as predicted from consideringthe elution orders for the 1,2,3,6The NPO phase, which distinguishes dioxin isomers /1,2,3,9-TCDD isomer pair, for which isomer identification primarily by dipole-dipole interaction, can separate all of is known with some ~ertainty.~J?8 The substitution of chlorine the isomer pairs coproducedin preparation. In addition, when at the 2- and 3-positions of dibenzo-p-dioxin in 1,6-DCDDor it is reinforced with the results on PYE phase and with the 1,g-DCDD was not expected to alter the retention order results on CU phase, structure assignment is permitted for determined by hydrophobic, charge-transfer, and dipoledipole interactions. In fact, the extra chlorine substituents the peaks of those congeners omitted in previous studies.2 actually enhanced the separation (Table I1 and Figure 3). The chromatographic peak assignments, which are based on Similarly, substituting a chlorine atom at the 4-position of retention order, are in agreement with predictions of calcu1,2,6-/1,2,9-TrCDD should not reverse the retention order, lated dipole moments and heats of formation.6 although the substitution prevents separation on Clsand PYE. Therefore, predicting the elution order and peak size for this ACKNOWLEDGMENT pair on the chromatograms on NPE and NPO, we assigned 1,2,4,6-TCDDto the first and larger peak. The remainder of the previously unassigned peaks correspondingto the PnCDD We thank the Japanese Ministry of Education for partial and HxCDD isomers of pairs were then assigned (Figure 3). funding of the study through the Monbusho International In other words, because we confirmed identities and retention Scientific Research Program Projects (01044081,03044089), orders of 1,3,6,8-/1,3,7,9-TCDDand 1,2,6,8-/1,2,7,9-TCDD, which included suggestions and supportive contributions from we could extrapolate the retention orders and identities of Shigeru Terabe, Himeji Institute of Technology, and Koushi 1,2,3,6,8-/1,2,3,7,9-PnCDD. We could also extrapolate from Fukunishi, Kyoto Institute of Technology. For technical 1,2,6,8-/1,2,7,9-TCDD to 1,2,4,6,8-/1,2,4,7,9-PnCDD; from assistance, we thank Hironobu Miyanishi, James Gill, and 1,2,6,7-/1,2,8,9-TCDDto 1,2,4,6,7-/1,2,4,8,9-PnCDD;and from Chester Lapeza. We thank Emory Dixon for use of the 1,2,4,6,7-/1,2,4,8,9-PnCDD to 1,2,4,6,7,9-/1,2,4,6,8,9-HxCDD. Beckman HPLC system. Use of trade names is for identiStructural identities to the pairs 1,2,4,7-/1,2,4,8-TCDD; fication only and does not constitute endorsement by the 1,7-/1,8-DCDD; and 1,2,7-/1,2,8-TrCDD were similarly asPublic Health Service or the US.Department of Health and signed based on the peak elution order reversal between PYE Human services. and the nitroaromatic phases. The elution order agrees with the behavior of 1,2,4,6,7,9-/1,2,4,6,8,9-HxCDD. All assignments agree with the calculated dipole moments published RECEIVED for review February 8, 1993. Accepted June 5, for TCDDs, PnCDDs, and HxCDDs.12 All peaks for DCDDs 1993. and TrCDDs were also assigned. We could expect the
