Temperature dependence of DCDD/F isomer distributions from chlorophenol precursors

Article abstract of DOI:10.1016/S0045-6535(00)00246-0

The temperature dependence of the gas-phase, rate-limited formation of dichlorodibenzo-p-dioxin (DCDD) and dichlorodibenzofuran (DCDF) isomers from 2,6-dichlorophenol and 3-chlorophenol, respectively, has been studied experimentally in an isothermal flow reactor over the range 300-900°C under pyrolytic, oxidative and catalytic conditions and computationally using semi-empirical molecular orbital methods. At high temperatures, distributions of sets of DCDD/F condensation products are consistent with the calculated thermodynamic distributions, indicating that the relative rates of formation are governed by differences in symmetry and steric hindrance present in the isomer product structures. At low temperatures, however, this is not the case. In the case of 1,6- and 1,9-DCDD formed from 2, 6-dichlorophenol via Smiles rearrangement, the 1,6 isomer is favored at low temperatures more than thermodynamically predicted. This result appears to be consistent with kinetic effects of either the expansion of the five-membered ring Smiles intermediate or a lower activation energy six-membered ring intermediate pathway that produces only the 1,6 isomer. For formation of 1,7-, 3,7- and 1,9-DCDF from 3-chlorophenol, the 1,7 isomer fraction increases at low temperatures whereas thermodynamics predicts a decrease. This result can be attributed to steric effects in alternative sandwich-type approach geometries of phenoxy radicals to form the o, o′-dihydroxybiphenyl (DOHB) intermediate via its keto-tautomers. Higher level molecular theory (ab initio) is needed to provide a more quantitative description of these kinetics.

Full text of DOI:10.1016/S0045-6535(00)00246-0

Chemosphere 42 (2001) 719±727
Temperature dependence of DCDD/F isomer distributions
q
from chlorophenol precursors
*
James A. Mulholland , Umesh Akki, Yun Yang, Jae-Yong Ryu
Environmental Engineering, Georgia Institute of Technology, 200 Bobby Dodd Way, Atlanta, GA 30332-0512, USA
Received 1 September 1999; received in revised form 1 November 1999
Abstract
The temperature dependence of the gas-phase, rate-limited formation of dichlorodibenzo-p-dioxin (DCDD) and
dichlorodibenzofuran (DCDF) isomers from 2,6-dichlorophenol and 3-chlorophenol, respectively, has been studied
experimentally in an isothermal ¯ow reactor over the range 300±900°C under pyrolytic, oxidative and catalytic con-
ditions and computationally using semi-empirical molecular orbital methods. At high temperatures, distributions of sets
of DCDD/F condensation products are consistent with the calculated thermodynamic distributions, indicating that the
relative rates of formation are governed by di€erences in symmetry and steric hindrance present in the isomer product
structures. At low temperatures, however, this is not the case. In the case of 1,6- and 1,9-DCDD formed from 2,
6
-dichlorophenol via Smiles rearrangement, the 1,6 isomer is favored at low temperatures more than thermodynami-
cally predicted. This result appears to be consistent with kinetic e€ects of either the expansion of the ®ve-membered ring
Smiles intermediate or a lower activation energy six-membered ring intermediate pathway that produces only the 1,6
isomer. For formation of 1,7-, 3,7- and 1,9-DCDF from 3-chlorophenol, the 1,7 isomer fraction increases at low
temperatures whereas thermodynamics predicts a decrease. This result can be attributed to steric e€ects in alternative
0
``sandwich-type'' approach geometries of phenoxy radicals to form the o; o -dihydroxybiphenyl (DOHB) intermediate
via its keto-tautomers. Higher level molecular theory (ab initio) is needed to provide a more quantitative description of
these kinetics. Ó 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Dioxins; Furans; Semi-empirical molecular orbital modeling; Toxic combustion byproducts
1
. Introduction
(1993) identi®ed the former as an important route in
combustion processes. Detailed mechanisms for PCDD/
F formation from chlorophenols were ®rst proposed by
Born et al. (1989, 1993). PCDD formation can be un-
derstood by phenoxy radical addition and Cl-atom-
elimination at an ortho-carbon site of phenol molecule
to form a o-phenoxyphenol (POP) intermediate. Alter-
natively, phenoxy radical±radical combination by oxy-
gen±carbon coupling produces the keto-tautomer of
POP. Both routes are shown for 2,6-dichlorophenol in
Fig. 1. The radical±radical route is likely favored in low
temperature, slow combustion due to the much larger
energy barrier (more than 20 kcal/mol) of the radical±
molecule route (Fleming, 1976; Wiater and Louw, 1999).
In either case, the resonance-stabilized phenoxy±phenoxy
Two major pathways for the formation of polychlo-
rinated dibenzo-p-dioxins (PCDDs) and dibenzofurans
PCDFs) have been identi®ed: precursor (e.g., chlor-
ophenols) mechanisms and de novo synthesis. Altwicker
(
q
Presented at the Sixth International Congress on Toxic
Combustion Byproducts, 27±30 June, 1999, University of
Karlsruhe.
*
Corresponding author. Tel.: +1-404-894-1695; fax: +1-404-
8
94-8266.
E-mail address: james.mulholland@ce.gatech.edu (J.A.
Mulholland).
0045-6535/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ( 0 0 ) 0 0 2 4 6 - 0

7
20
J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
Fig. 1. Reaction pathways for the formation of DCDDs from 2,6-dichlorophenol (2,6-DCP). Condensation of phenoxyphenoxy
radical is shown to occur via six-membered ring (path a) and ®ve-membered ring (path b ± analogous to Smiles rearrangement)
structures.
radical is formed, followed by ring condensation either
via six-member ring closure (path a) at another
Cl-substituted ortho-carbon or via ®ve-member ring
closure (path b) similar to Smiles rearrangement (Truce
et al., 1970); loss of a second Cl-atom completes the
formation of PCDD. The former route produces only
one PCDD product, whereas the latter route produces a
pair of PCDD isomers (Sidhu et al., 1995).
substituted phenoxy radicals at low temperatures,
Armstrong et al. (1983) found that phenoxy radicals
preferentially approach each other with ``sandwich-like''
geometries, with coupling occurring at sites of highest
spin density and for geometries of least steric hindrance.
There are four possible ortho±ortho couplings for pairs
of phenoxy radicals without ortho-carbon Cl substitu-
ents, with two of the route degenerate in the case of
identical phenoxy radicals as shown in Fig. 2.
Formation of PCDFs, on the other hand, can be
understood by ortho±ortho carbon coupling of phenoxy
Weber and Hagenmaier (1999) reported results of
gas-phase experiments on the temperature dependence
of total PCDD/F yield and the ratio of PCDD to PCDF
from chlorophenols; the temperature dependence of
PCDD/F isomer distributions, however, was not ad-
dressed. Such information is needed to predict isomer
0
radicals at unsubstituted sites to form a o; o -di-
hydroxybiphenyl (DOHB) intermediate (Weber and
Hagenmaier, 1997) via its keto-tautomer, followed by
H
2
O elimination to form PCDF. This is shown for 3-
chlorophenol in Fig. 2. In studies of the coupling of
Fig. 2. Reaction pathways for the formation of DCDFs from 3-chlorophenol (3-CP). The formation of DOHB intermediate that
condenses to form 1,7-DCDF (path b) has a reaction path degeneracy of two, whereas only one coupling yields 3,7- and 1,9-DCDFs
(
paths a and c, respectively).

J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
721
distributions and total toxic equivalency for PCDDs and
PCDFs formed at both high and low temperatures in
incinerators. In a comprehensive study by Wehrmeier et
al. (1998), patterns of PCDD isomers, obtained from
municipal waste incinerator samples, experiments on the
chlorination and hydrodechlorination of PCDDs, and
semi-empirical method AM1 calculations of thermody-
namic stability and reactivity, were used for the eluci-
dation of thermal formation mechanisms. They found
that, while the relative abundances of PCDD isomers
calculated from thermodynamic stability were similar to
those measured from combustion, distinct di€erences
were observed that could be attributed to kinetic e€ects.
They did not, however, address the formation of PCDD/
F isomers by the coupling of chlorophenol precursors.
Previous study has shown that the thermal coupling
of o-dichlorobenzenes results in the formation of a
thermodynamic distribution of polychlorinated biphenyl
signi®cantly di€erent than thermodynamic distributions.
Here, we assess these hypotheses experimentally and
computationally, with the underlying assumption that
PCDD/F formation from chlorinated phenol precursors,
while not the only pathway, is an important pathway in
incineration systems.
2. Experimental
Experiments were conducted in an electrically heated,
quartz tube ¯ow reactor, 40 cm in length and 1.7 cm in
diameter, at temperatures ranging from 300°C to 900°C
in 50°C increments. High purity chlorinated phenol re-
actant (40 mg, nominal) was placed in a glass vessel and
heated to a temperature above its melting point. Reac-
tant vapor was transported to the reactor by carrier gas
through a heated transfer line. Reactor residence times
ranged between 2 and 10 s in these experiments. An
experiment took approximately 7 min to complete, with
all reactant vaporized and delivered to the reactor. The
concentration of reactant in carrier gas delivered to the
reactor was measured to be approximately constant,
between 0.5 and 0.7 mol%, over 90% of the duration of
an experiment. For pyrolysis experiments, helium carrier
gas was used. For study in an oxidative environment, a
gas mixture of 92% nitrogen and 8% oxygen was used.
(
PCB) isomers, with isomer distributions governed by
statistical factors and steric e€ects associated with
chlorine substitution (Mulholland et al., 1993). Recent-
ly, we presented experimental results on the pyrolytic
formation of 1,6- and 1,9-dichlorodibenzo-p-dioxin
(
DCDD) isomers from 2,6-dichlorophenol (Akki and
Mulholland, 1997) and of 1,7-, 3,7- and 1,9-dic-
hlorodibenzofuran (DCDF) isomers from 3-chlorophe-
nol (Yang et al., 1998) suggesting that thermodynamic
factors govern these distributions produced at high
temperatures. In this paper, we present additional ex-
perimental results on the reactivity of these chlorophe-
nols under oxidative and catalytic conditions, thereby
extending the range of study to lower temperatures.
Semi-empirical molecular orbital modeling is used to
assess the extent to which PCDD/F thermodynamic
properties govern isomer distributions produced under
kinetic-control from single chlorinated phenol precur-
sors, and to explain deviations of the experimental
PCDD/F distributions from thermodynamic distribu-
tions at low temperatures.
For study under catalytic conditions, a 1 g SiO
2 2
/CuCl
particle bed containing 0.5±1% (mass) copper chloride
catalyst was loaded into the reactor. When an experi-
ment is begun with the ¯ow of chlorinated phenol vapor
in N
be produced by CuCl
2
/O
2
carrier gas to the reactor, chlorine gas (Cl
2
) can
2
decomposition or by Deacon
process chemistry. In addition to providing a source of
reactive chlorine gas, Cu species can catalyze biaryl
synthesis through heterogeneous reaction (Stieglitz
et al., 1989).
Unreacted chlorinated phenol reactant, aromatic
products and soot were collected by sampling the entire
product gas stream using a dual ice-cooled dichlorom-
ethane (DCM) trap. After each experiment, impinger
and rinse samples were combined and soot, de®ned as
the DCM-insoluble fraction, was extracted by ®ltration
and measured gravimetrically. Sample solutions were
then analyzed by GC/MS using DB-5 ms and DB-dioxin
columns for aromatic compounds ranging from benzene
through C20 compounds. In the experiments reported
here, 2,6-dichlorophenol and 3-chlorophenol were re-
acted to produce DCDD and DCDF isomer sets. For
identi®cation, samples containing ®ve of 10 DCDDs,
including the 1,6 and 1,9 isomers of interest, and 12 of
16 DCDFs were obtained from the centers for disease
control (CDC) and the US Environmental Protection
Agency (EPA). The elution order of DCDFs on DB-5
columns is known (Hale et al., 1985). Two standards
obtained commercially were used to estimate DCDD
The pathways shown in Figs. 1 and 2 illustrate an
important di€erence in the formation of PCDD/F iso-
mer sets from chlorinated phenol precursors. The
branching step in PCDD formation is the ®nal step of
®ve-member ring expansion to two product isomers.
Di€erences in the product isomer structures are likely to
be similar to di€erences in the transition states; thus,
thermodynamic factors are likely to play an important
role in the relative rate of formation of the PCDD iso-
mer pair. In the case of PCDF formation, however, the
initial ortho±ortho carbon coupling of phenoxy radicals
is the branching step that leads to di€erent PCDF iso-
mers. The proposed sandwich-like geometries of react-
ing phenoxy radicals are very dissimilar to the nearly
planar PCDF product structures; therefore, we expect
rate-limited formation of sets of PCDF isomers from
chlorinated phenols to result in distributions that are

7
22
J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
and DCDF response factors: 2,7-DCDD and 2,6-
DCDF, respectively. Based on the range of GC/MS re-
sponse factors measured for chlorinated phenol and
benzene isomers obtained commercially, the accuracy of
using equal response factors for isomer groups is esti-
mated to be Æ20%. Findings on the temperature de-
pendence of isomer distributions are, however, not
a€ected by this assumption.
been studied by molecular modeling. Huang et al. (1996)
compared group additivity and semi-empirical methods
for calculating Gibbs free energies of formation of
PCDDs to predict thermodynamically favored reductive
dechlorination pathways, whereas Lynam et al. (1998)
used semi-empirical and ab initio methods to obtain
molecular descriptors that govern PCDD/F reductive
dechlorination.
Here, semi-empirical molecular modeling methods
available in the molecular orbital program MOPAC
(Stewart, 1990) have been used to estimate thermody-
namic distributions of PCDD/F isomers. While less ac-
curate in predicting absolute thermodynamic properties
than ab initio methods employing large basis sets and
high levels of theory, semi-empirical methods can ac-
curately and eciently predict isomer di€erences if the
structural di€erences between isomers are adequately
represented in the parameterizations based on experi-
mental data. On the other hand, group additivity
methods allow for very close agreement between calcu-
lated and measured properties when experimental data
for deriving group interactions are available.
3
. Molecular modeling methods
Due to the lack of experimental data on PCDD/F
thermodynamic properties, a wide variety of molecular
modeling techniques have been used to identify ther-
modynamically favored pathways to speci®c PCDD/F
isomers. Bozzelli et al. (1991) modi®ed and expanded the
group additivity method of Shaub (1982) for estimating
PCDD/F thermodynamic properties to include non-
next-nearest neighbor interactions. Unsworth and Dor-
ans (1993) compared PCDD/F isomer distributions from
incinerators to thermodynamic distributions calculated
using a semi-empirical molecular orbital method. Tu-
ppurainen and Ruuskanen (1999) used semi-empirical
and ab initio methods to study copper-catalyzed oxi-
dative chlorophenol coupling reactions. Isomer-speci®c
PCDD/F reductive dechlorination pathways have also
Heats of formation for the 10 DCDD isomers and 16
DCDF isomers calculated by the semi-empirical meth-
ods modi®ed neglect of diatomic di€erential overlap
(MNDO), Austin model 1 (AM1) and Parametric
method 3 (PM3) are shown in Table 1. Group additivity
Table 1
DCDD/F heats of formation calculated using semi-empirical methods
DCDD
DHf;298° (kcal/mol)
DCDF
DHf;298 (kcal/mol)
a
Group
AM1
PM3
MNDO
AM1
PM3
MNDO
2,7
2,8
2,3
1,7
1,8
1,3
1,2
1,6
1,9
1,4
)50.84
)50.84
)48.75
)40.62
)40.62
)39.63
)38.53
)30.4
)10.09
)10.08
)8.69
)7.67
)7.64
)7.24
)6.23
)5.3
)22.19
)22.18
)21.1
)20.76
)20.74
)20.6
)19.78
)19.33
)19.27
)19.27
)36.2
2,8
2,7
3,7
1,8
1,7
1,3
2,3
2,6
3,6
1,2
23.07
23.17
23.29
23.49
23.53
23.99
24.65
24.68
24.81
24.91
25.06
25.06
25.29
26.1
12.42
12.38
12.36
13.17
13.12
13.31
13.51
13.94
13.92
14.22
14.09
14.68
14.76
14.89
15.51
15.73
)0.34
)0.11
0.15
0.85
1
)36.18
)33.64
)34.26
)34.22
)33.71
)31.66
)32.36
)32.01
)31.95
1.59
2.53
1.16
1.45
3.46
1.68
2.26
2.71
3.86
2.9
)30.4
)30.01
)5.1
)4.98
2
1
1
3
4
1
,4
,6
,4
,4
,6
,9
26.41
26.94
5.74
Group
C1/O ortho
Cl/Cl ortho
Cl/Cl bay
Cl/O ortho
Cl/Cl ortho
Cl/H bay
3.9
1.6
1.4
0.4
0.4
0.1
3.3
1.5
1.1
0.8
0.15
0
6.1
1.4
2.6
1.1
0.6
0.2
10.2
2.1
2.4
1.4
1.5
1.1
2
2.6
Cl/Cl meta
1
0.4
0.15
0.5
Cl/Cl meta
Cl/O meta
a
Group additivity method of Shaub (1982).

J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
estimates for DCDDs by Shaub (1982) are also shown.
723
The stability ranking of DCDD isomers using group
additivity methods is very similar to that based on the
AM1 and PM3 results, although the range in heat of
formation values calculated by group additivity is much
wider: 20.83 kcal/mol compared to 5.11 and 2.92 kcal/
mol for AM1 and PM3, respectively. The large range in
group additivity values is unrealistic, according to
Unsworth and Dorans (1993), due to the derivation of
Cl±O interactions in PCDDs from Cl±OH interactions
in chlorophenols. Correlation with group additivity
2
values is high for AM1 (R ˆ 0.963) and PM3
Fig. 3. Yields of unreacted 2,6-dichlorophenol (2,6-DCP) and
total DCDD products under pyrolytic, oxidative and catalytic
conditions. Data points are not shown for 2,6-DCP yields.
2
(
(
R ˆ 0.927) data, and much lower for MNDO data
2
R ˆ 0.725). MNDO tends to overestimate non-bonded
repulsions such as the Cl/Cl ortho group interaction, as
evidenced by the higher relative heats of formation of
DCDDs, were detected. Over the temperature range
studied, other pyrolysis products observed include 2-
chlorophenol, 1,3-dichlorobenzene, monochloroben-
zene, napthalenes with up to three chlorine substituents,
2
,3-DCDD and 2,3-DCDF predicted by MNDO than
by the other methods. Thus, it is likely that MNDO
underestimates the thermodynamic isomer fractions of
the most toxic PCDD/F congeners, which have chlorine
substituents at lateral ring sites (i.e., the 2,3,7,8 cong-
eners).
1
-monoCDD, dibenzofuran and benzonaphthofuran
products with up to two chlorine substituents, and soot;
these results are reported elsewhere (Akki and Mulhol-
land, 1997). 1,6- and 1,9-DCDDs are produced directly
from coupling of two 2,6-dichlorophenol molecules,
whereas 1-monoCDD and the furan compounds are
secondary products from reactions of 2,6-dichlorophe-
nol with 2-chlorophenol and naphthalene products. In
the presence of oxygen, 2,6-dichlorophenol decomposi-
tion began at 400°C and was 99% reacted at 700°C. The
same two DCDDs produced under pyrolytic conditions
were produced with a peak total yield of 1.2% observed
at 550°C. A similar set of secondary products was ob-
served as well. The higher yield at lower temperatures of
DCDDs under oxidative conditions is explained by en-
hanced phenoxy radical formation due to H-atom ab-
straction reactions by oxygen species. In the presence of
the copper chloride catalyst, a peak 1,6- and 1,9-DCDD
yield of 3% was observed at 360°C. Above 450°C, the
The semi-empirical method AM1 was chosen for re-
action pathway analysis in this study. We limit our
presentation of AM1 results here to relative heats of
formation among isomer groups. In previous work
(
Mulholland et al., 1993), AM1 was shown to most ac-
curately predict the measured torsional angle of biphe-
nyl (40°) and its internal rotational barriers through
planar and perpendicular conformations (2 and 1 kcal/
mol, respectively). These properties may be important in
the coupling reactions to form POP and DOHB inter-
mediates and the ring condensation reactions to form
PCDDs and PCDFs, respectively. AM1 has been found,
however, to underestimate PCB rotational barriers,
particularly for overcrowded congeners, when compared
to the ab initio method B3LYP/6-31G* (Biedermann
and Agranat, 1999). Therefore, the AM1 results re-
ported here should be viewed as qualitative estimates of
thermodynamic stability and transition state barriers.
2
CuCl catalyst is not stable.
4. Results and discussion
4.1. Total DCDD/F yields
Yields of unreacted 2,6-dichlorophenol and total
DCDD products formed under pyrolytic, oxidative and
catalytic conditions are shown in Fig. 3. In the absence
of oxygen, 2,6-dichlorophenol decomposition began at
550°C and was more than 99% reacted at 800°C. As
expected from the ®ve-member ring intermediate path-
way shown in Fig. 1, 1,6- and 1,9-DCDDs were the
primary dioxin products observed, with a peak total
yield of 0.4% (carbon conversion) found between 600°C
and 700°C. Above 900°C, no chloroaromatics, including
Fig. 4. Yields of unreacted 3-chlorophenol (3-CP) and total
DCDF products under pyrolytic, oxidative and catalytic
2
(CuCl ) conditions. Data points are not shown for 3-CP yields.

7
24
J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
Yields of unreacted 3-chlorophenol and total DCDF
products formed under pyrolytic and oxidative condi-
tions, as well as in a heterogeneous environment with
copper chloride catalyst, are shown in Fig. 4. As ex-
pected based on the mechanism shown in Fig. 2, the
primary furan products observed were 1,7-, 3,7- and 1,9-
DCDF isomers. In the absence of oxygen, 3-chlor-
ophenol decomposition began at 500°C and was more
than 99% reacted at 800°C. A peak total DCDF yield of
0
.3% (carbon conversion) was observed between 600°C
and 700°C. Other pyrolysis products observed include
phenol, monochlorobenzene, naphthalenes, cyclopen-
ta[def]phenanthrene, monoCDFs and dibenzofuran,
monochlorodibenzofuranol, and soot; these results are
reported elsewhere (Yang et al., 1998). In the presence of
oxygen, 3-chlorophenol decomposition began at 400°C
and was 99% reacted at 700°C, with the same three
DCDF products observed in the pyrolysis experiments
observed under oxidative conditions. A peak total
DCDF product yield of 12% was observed at 600°C. As
in the case of DCDD formation, the e€ect of oxygen on
total DCDF formation was to increase yields at lower
Fig. 5. DCDD isomer fractions from 2,6-dichlorophenol
(
resent a ®t to data such that the rate of formation of 1,9 relative
squares ˆ 1,6-DCDD; triangles ˆ 1,9-DCDD). Solid lines rep-
Ã
�
DE =RT
to that of 1,6 is given by Arel
Ã
e
, where Arel ˆ 1:6 and
DE ˆ 1:4 kcal/mol; for comparison, dashed lines representing
calculated thermodynamic estimates are also shown.
closure by the ®ve-member ring intermediate (path b,
Fig. 1). The AM1 reaction path heats of formation
shown in Fig. 6 demonstrate that there are two possible
explanations of the relative kinetics of formation of
these isomers. The di€erence in calculated heats of for-
mation for transition states from the ®ve-member ring
intermediate to 1,6- and 1,9-DCDD is 1.0 kcal/mol,
which is close to the observed temperature dependence.
Alternatively, the ®ve-member ring intermediate has a
heat of formation that is 1.2 kcal/mol greater than that
of the six-member ring intermediate, and a transition
state heat of formation that is greater by 3.9 kcal/mol.
Therefore, the observed favored formation of 1,6-
DCDD over 1,9-DCDD relative to thermodynamic
yields may be explained either by the kinetics of ®ve-
temperatures. In the presence of the SiO
bed, the same three DCDF products were observed at
00°C, and a peak total DCDF yield of 4% was ob-
served at 400°C. Thus, the presence of CuCl catalyst
2 2
/CuCl particle
3
2
lowered the temperature of DCDF formation still fur-
ther, which may be due to the formation of chlorine gas
and Cl atoms that enhance phenoxy radical formation
by H-atom abstraction or due to heterogeneous phenoxy
coupling reactions.
4
2
.2. 1,6- and 1,9-DCDD isomer distribution from
,6-dichlorophenol
Measured and calculated isomer fractions of 1,6- and
,9-DCDDs from 2,6-dichlorophenol are shown in
1
Fig. 5. Also shown is CDC data from synthesis experi-
ments at 175°C (Grainger et al., 1996). The experimental
data (points) approach the thermodynamic distribution
(
dashed lines) at high temperatures with 1,6-DCDD fa-
vored slightly over 1,9-DCDD. At low temperatures,
however, 1,6-DCDD is favored more than thermody-
namically predicted. This ®nding does not depend on the
semi-empirical method used since, as shown in Table 1,
1
,6-DCDD has only a slightly lower heat of formation
than 1,9-DCDD as calculated by any of the methods
0.35, 0.20 and 0.05 kcal/mol by the MNDO, AM1 and
(
PM3 methods, respectively). From the experimental re-
sults, a relative rate of formation of 1,9- to 1,6-DCDD
(
solid lines, Fig. 5) is found, with a temperature depen-
dence of 1.4 kcal/mol, greater than the thermodynamic
estimates by any of the semi-empirical or group addi-
tivity methods.
Fig. 6. Energy diagram for the conversion of phenoxy±phen-
oxy radical to DCDD products via six-member and ®ve-mem-
ber ring closure. AM1 heats of formation relative to that of
phenoxy±phenoxy radical are shown for stable products and
transition states.
Formation of the 1,6- and 1,9-DCDD isomers from
,6-dichlorophenol is evidence of the mechanism of ring
2

J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
725
member expansion to the 1,6 isomer, or by competition
at low temperatures of a second reaction pathway, that
of six-member ring closure which produces only 1,6-
DCDD. Further computational study using ab initio
methods to more accurately estimate entropy e€ects is
needed to better understand these kinetics.
imentally, the 1,7-DCDF isomer is formed in highest
yields, most signi®cantly at low temperatures, which
would not be expected thermodynamically. The relative
rates of formation of the three isomers found experi-
mentally under oxidative and pyrolytic conditions sug-
gest that the di€erences in activation energies between
3
,7-DCDF and 1,7- and 1,9-DCDFs are )1.0 and 2.5
kcal/mol, respectively. Under catalytic conditions, 1,7-
DCDF is kinetically favored by an even greater amount.
The di€erences between catalytic and oxidative results at
the same temperatures suggest that the heterogeneous
phenoxy radical coupling reactions were involved in the
catalytic experiments.
4
3
.3. 1,7-, 3,7- and 1,9-DCDF isomer distribution from
-chlorophenol
Measured and calculated isomer fractions of 1,7-,
,7- and 1,9-DCDF products from 3-chlorophenol are
3
shown in Fig. 7. As was the case for 1,6- and 1,9-DCDD
formation from 2,6-dichlorophenol, the experimental
data (points) approach a thermodynamic distribution
Preferential formation of 1,7-DCDF at low-temper-
atures is not explained by thermodynamic di€erences in
the reaction intermediates, that is, the DOHBs and their
keto-tautomers (Fig. 8). These intermediates have the
same Cl substitution pattern as the corresponding
DCDF products, and, therefore, similar relative stabil-
ities. Transition state analysis of the ortho±ortho carbon
coupling of phenoxy radicals is needed to understand
di€erences in the rates of formation of the DCDF iso-
mers. Qualitatively, we can explain the results by con-
sidering alternative sandwich-like geometries of reacting
(
dashed lines) at high temperatures (pyrolytic condi-
tions). Similar ®ndings are obtained by analysis of py-
rolysis data on the distribution of 1,2,7,8-, 2,3,7,8- and
1
,2,8,9-TetraCDFs from 3,4-dichlorophenol (Yang et
al., 1998). For the three DCDF products, di€erences in
chlorine substitution at bay sites have a large e€ect on
their thermodynamic distribution due to steric hin-
drance. As shown in Table 1, 3,7-DCDF, which has no
bay Cl substituents, has the lowest heat of formation of
the three isomers; 1,7-DCDF, which has one bay Cl
substituent, and 1,9-DCDF, which has two bay Cl
substituents, have AM1 heats of formation that are
higher by 0.24 and 3.65 kcal/mol, respectively. Another
factor a€ecting the thermodynamic distribution of these
three isomers is symmetry, which translates to a reaction
path degeneracy of two for 1,7-DCDF formation com-
pared to one for the formation of 3,7- and 1,9-DCDFs
with symmetry. Thus, 1,7-DCDF is thermodynamically
favored at high temperatures, whereas 3,7-DCDF is
thermodynamically favored at low temperatures, a result
found using any of the semi-empirical methods. Exper-
Fig. 8. Energy diagram for the formation of DCDF products
from 3-chlorophenol. In the top panel, AM1 heats of formation
relative to that of two chlorophenoxy radicals are shown for
intermediates and products. Transition state energies were not
calculated. In the bottom panel, alternative sandwich-like ge-
ometries of approach for ortho±ortho carbon coupling of 3-
chlorophenoxy radicals to form the DOHB keto-tautomers are
shown.
Fig. 7. DCDF isomer fractions from 3-chlorophenol (squar-
es ˆ 1,7-DCDF; triangles ˆ 3,7-DCDF; circles ˆ 1,9-DCDF).
Solid lines represent a ®t to oxidative and pyrolytic data, with
the rates of 1,7- and 1,9-DCDF formation relative to 3,7-
�
Ã
DE =RT
Ã
DCDF given by e
respectively; dashed lines representing AM1 thermodynamic
estimates are shown for comparison.
where DE ˆ )1.0 and 2.5 kcal/mol,

7
26
J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
phenoxy radicals. Six di€erent approach geometries re-
sulting in eight possible ortho±ortho carbon coupling
reactions of 3-chlorophenoxy radicals are shown in the
bottom panel of Fig. 8. The geometry that minimizes
steric hindrance is the staggered con®guration (upper
left geometry), with only one O±O interaction; therefore,
this approach, which results in the formation of 1,7-
DCDF, is expected to be favored kinetically. Geometries
depicted with one O±Cl interaction (other two geome-
tries in top row) also lead to 1,7-DCDF. Geometries
that eventually produce 3,7-DCDF (bottom left and
center), on the other hand, have Cl±Cl interactions. The
References
Akki, U., Mulholland, J., 1997. Gas-phase formation of dioxin
and other aromatic products from 2,6-dichlorophenol
pyrolysis. Organohalogen Compd. 31, 475±479.
Altwicker, E.R., 1993. Municipal waste combustion. Proc. Air
Waste Manage. Assoc. 32, 202±206.
Armstrong, D., Breckenridge, R., Cameron, C., Nonhebel, D.,
Pauson, P., Perkins, P., 1983. The role of stereoelectronic
factors in the oxidation of phenols. Tetrahedron Lett. 24,
1071±1074.
Biedermann, P.U., Agranat, I., 1999. The internal barriers of
rotation for the 209 polychlorinated biphenyls. Environ. Sci.
Pollut. Res. 6, 67±68.
®nal geometry shown has two O±Cl interactions and
Born, J.G.P., Louw, R., Mulder, P., 1989. Formation of
dibenzodioxins and dibenzofurans in homogenous gas-
phase reactions of phenols. Chemosphere 19, 401±406.
Born, J.G.P., Mulder, P., Louw, R., 1993. Fly ash mediated
reactions of phenol and monochlorophenols: oxychlorina-
tion, deep oxidation, and condensation. Environ. Sci.
Technol. 27, 1849±1863.
leads to 1,9-DCDF formation. Therefore, 1,7-DCDF is
expected to be favored kinetically, whereas 3,7-DCDF is
favored thermodynamically, as indicated by the crossing
of the energy pro®les in the radical combination reaction
(
®rst step shown in Fig. 8). Thus, the experimental re-
sults demonstrate the kinetic control of PCDF isomer
formation.
Bozzelli, J.W., Wu, Y.G., Ritter, E.R., 1991. Thermodynamic
insights on pathways to formation of chlorinated dioxins
and dibenzofurans. Chemosphere 23, 1221±1232.
Fleming, I. (Ed.), 1976. Frontier Orbitals and Organic Reac-
tions. Wiley, New York, pp. 195±196.
Further computational study using ab initio methods
is needed to explore the kinetics of the ortho±ortho
carbon coupling of phenoxy radicals. The sandwich-like
geometry is preferred due to enhanced stabilization by
overlap of the aromatic ring systems during reaction.
Therefore, molecular orbital modeling used to study this
reaction must properly account for this e€ect.
Grainger, J., McClure, P.C., Liu, Z.Y., Botero, B., Sirimanne,
S., Patterson, D.G., Sewer, M., Gillyard, C., Kimata, K.,
Hosoya, K., Araki, T., Tanaka, N., Terabe, S., 1996. Isomer
identi®cation of chlorinated dibenzo-p-dioxins by orthogo-
nal spectroscopic and chromatographic techniques. Chemo-
sphere 32, 13±23.
5
. Conclusions
Hale, M.D., Hileman, F.D., Mazer, T., Shell, T.L., Noble,
R.W., Brooks, J.J., 1985. Mathematical modeling of tem-
perature programmed capillary gas chromatographic reten-
tion indexes for polychlorinated dibenzofurans. Anal.
Chem. 57, 640±648.
The formation of PCDD/F isomer sets from chlor-
ophenols was studied in gas-phase, ¯ow reactor experi-
ments under pyrolytic, oxidative and catalytic
conditions. At high temperatures, the distributions of
Huang, C.-L., Harrison, B.K., Madura, J., Dol®ng, J., 1996.
Gibbs free energies of formation of PCDDs: evaluation
of estimation methods and application for predicting
dehalogenation pathways. Environ. Toxicol. Chem. 15,
1
3
,6- and 1,9-DCDDs from 2,6-dichlorophenol and 1,7-,
,7- and 1,9-DCDFs from 3-chlorophenol are similar to
calculated thermodynamic distributions. At low tem-
peratures, however, the preferential formation of 1,6-
DCDD and 1,7-DCDF demonstrate kinetic control. In
the case of PCDD formation, the results at low tem-
perature are consistent with lower transition state bar-
riers for 1,6-DCDD formation via either a ®ve-member
ring closure pathway alone that produces both isomers
or a competing six-member ring closure pathway that
produces only 1,6-DCDD. In the case of PCDF for-
mation, the results suggest that steric factors associated
with the geometries of phenoxy radical ortho±ortho
carbon coupling play an important role in determining
the distribution of PCDF product isomers.
8
24±836.
Lynam, M.M., Kuty, M., Damborsky, J., Koca, J., Adriaens,
P., 1998. Molecular orbital calculations to describe micro-
bial reductive dechlorination of polychlorinated dioxins.
Environ. Toxicol. Chem. 17, 988±997.
Mulholland, J.A., Saro®m, A.F., Rutledge, G.C., 1993. Semi-
empirical molecular orbital estimation of the relative
stability of polychlorinated biphenyl isomers produced by
o-dichlorobenzene pyrolysis. J. Phys. Chem. 97, 6890±6896.
Shaub, W.M., 1982. Procedure for estimating the heats of
formation of aromatic compounds: chlorinated benzenes,
phenols and dioxins. Thermochim. Acta 55, 59±73.
Sidhu, S.S., Maqsud, L., Dellinger, B., Mascolo, G., 1995. The
homogeneous, gas-phase formation of chlorinated and
brominated dibenzo-p-dioxin from 2,4,6-trichloro- and
2
,4,6-tribromophenols. Combust. Flame 100, 11±20.
Acknowledgements
Stewart, J.J.P., 1990. MOPAC: A semi-empirical molecular
orbital program. J. Comput.-Aided Mol. Des. 4, 1±105.
Stieglitz, L., Zwick, G., Beck, J., Bautz, H., Roth, W., 1989.
Carbonaceous particles in ¯y ash ± a source for the de-novo-
Support of the National Science Foundation, Grant
No. CTS-9457028, is gratefully acknowledged.

J.A. Mulholland et al. / Chemosphere 42 (2001) 719±727
727
synthesis of organochlorocompounds. Chemosphere 19,
83±290.
Weber, R., Hagenmaier, H., 1999. Mechanism of the formation
of polychlorinated dibenzo-p-dioxins and dibenzofurans
from chlorophenols in gas-phase reactions. Chemosphere
38, 529±549.
2
Truce, W.E., Krieder, E.M., Brand, W.W., 1970. The Smiles
and related rearrangements of aromatic systems. Org.
React. 18, 99±215.
Wehrmeier, A., Lenoir, D., Schramm, K.-W., Zimmermann,
R., Hahn, K., Henkelmann, B., Kettrup, A., 1998. Patterns
of isomers of chlorinated dibenzo-p-dioxins as tool for
elucidation of thermal formation mechanisms. Chemo-
sphere 36, 2775±2801.
Tuppurainen, K., Ruuskanen, J., 1999. A plausible mechanism
for copper-catalyzed oxidative chlorophenol coupling reac-
tions. Semi-empirical AM1 and ab initio 3-21G molecular
orbital studies. Chemosphere 38, 1825±1832.
Unsworth, J.F., Dorans, H., 1993. Thermodynamic data for
dioxins from molecular modelling computations: prediction
of equilibrium isomer composition. Chemosphere 27, 351±
Wiater, I., Louw, R., 1999. Reactions of diphenyl ether with
chlorine and bromine atoms around )750 K ± relavance for
gas-phase dioxin formation. Eur. J. Org. Chem., 261±265.
Yang, Y., Mulholland, J.A., Akki, U., 1998. Formation of
furans by gas-phase reactions of chlorophenols. In: 27th
International Symposium on Combustion. The Combustion
Institute, Pittsburgh, PA, pp. 1761±1768.
3
58.
Weber, R., Hagenmaier, H., 1997. On the mechanism of the
formation of polychlorinated dibenzofurans from chlor-
ophenols. Organohalogen Compd. 31, 480±485.