Products, rates, and mechanism of the gas-phase condensation of phenoxy radicals between 500-840 K

Article abstract of DOI:10.1002/(SICI)1099-0690(200003)2000:6<921::AID-EJOC921>3.0.CO;2-P

Phenols are demonstrated precursors of 'dioxins' - polychlorinated dibenzo-p-dioxins (DDs) and dibenzofurans (DFs) - in thermal processes, especially incineration. Heterogeneous catalysis, depending on conditions, can play an important role, but mere gas-phase combination of phenolic entities to ultimately DD and/or DF is always possible. The present paper addresses the fundamental role of phenol itself. Phenol has long been known to give DF upon pyrolysis and in similar thermal reactions. In the liquid phase under oxidative conditions it yields five condensation products (A-E); this clearly occurs through the dimerization of two phenoxy (PhO) radicals, followed by enolisation/rearomatisation. Our study shows that in the gas phase, at the lower T end, such dimers are also formed, but still with very little DF. That DF, indeed, is almost the only condensation product at elevated temperatures is substantiated by thermochemical-kinetic analysis (favouring the pathway of ortho-C/ortho-C combination of two PhO radicals), as well as by results obtained with two plausible intermediates, viz. 2,2'- dihydroxybiphenyl (A) and 2-phenoxyphenol (C). Mechanisms for the requisite enolisation and dehydration steps leading to DF are discussed.

Full text of DOI:10.1002/(SICI)1099-0690(200003)2000:6<921::AID-EJOC921>3.0.CO;2-P

FULL PAPER
Products, Rates, and Mechanism of the Gas-Phase Condensation of Phenoxy
Radicals between 500–840 K
Izabela Wiater,[a][‡] Jan G. P. Born,[a] and Robert Louw*[a]
Keywords: Gas-phase chemistry / Phenoxy radicals / Dioxins / 2-Phenoxyphenol / Reaction mechanisms
Phenols are demonstrated precursors of ‘‘dioxins’’ – polychlo-
rinated dibenzo-p-dioxins (DDs) and dibenzofurans (DFs) –
radicals, followed by enolisation/rearomatisation. Our study
shows that in the gas phase, at the lower T end, such dimers
in thermal processes, especially incineration. Heterogeneous are also formed, but still with very little DF. That DF, indeed,
catalysis, depending on conditions, can play an important
role, but mere gas-phase combination of phenolic entities to
ultimately DD and/or DF is always possible. The present pa-
per addresses the fundamental role of phenol itself. Phenol
has long been known to give DF upon pyrolysis and in sim-
is almost the only condensation product at elevated temper-
atures is substantiated by thermochemical–kinetic analysis
(favouring the pathway of ortho-C/ortho-C combination of
two PhO radicals), as well as by results obtained with two
plausible intermediates, viz. 2,2’-dihydroxybiphenyl (A) and
ilar thermal reactions. In the liquid phase under oxidative 2-phenoxyphenol (C). Mechanisms for the requisite enolis-
conditions it yields five condensation products (A–E); this cle-
ation and dehydration steps leading to DF are discussed.
arly occurs through the dimerization of two phenoxy (PhO)
Introduction
Thermolytic reactions of phenol have been investigated
for a long time and are still actively researched for, for ex-
ample, a better understanding of combustion mechanisms.
When subjected to heat, phenol yields dibenzofuran
(DF). A German patent from the beginning of the 20th
century, for example, outlines that phenol vapour, diluted
with steam, mixed with air, and heated with or without con-
tact substances resulted in dibenzofuran formation.[1] In an-
other early example, dibenzofuran is also produced upon
distillation of phenol over various substances, such as lead
oxide.[1] DF formation in the homogeneous gas-phase pyro-
lysis and slow combustion of phenol has been ascribed to
the dimerisation of two phenoxy radicals as the first
step.[2,3] More recent kinetic and product studies[4,5] showed
that DF formation from two phenoxy radicals can occur
with an overall rate constant of 108 –1 s–1 and above, not
so much less than k for the first step, which is approximately
Scheme 1. Formation of dibenzofuran through the dimerisation of
phenoxy radicals
At higher temperatures, phenoxy radicals decompose
unimolecularly into carbon monoxide and a cyclopenta-
dienyl radical[8] (Equation 1):
C6H5O• Ǟ c-C5H5• ϩ CO
(1)
1010 –1 s–1. A simple rationale is given in Scheme 1.[3,6]

This important reaction limits lifetimes – and concentra-
tions – of phenoxy-type radicals, and hence their signific-
ance in ‘‘dioxin’’ (DF, DD) formation in incinerators.[9]
The phenoxy radical is resonance stabilized by 16 kcal/
mol;[10] the electron spin is delocalized from oxygen to
ortho- and para-C, and the C–O bond is shortened.[11,12]
According to FMO theory, coupling of radicals occurs pref-
erentially at the sites of the highest spin density. SOMO
electron-spin densities calculated by AM1/UHF show the
highest value for the para position,[13] but ortho–ortho,
ortho–para, and para–para C–C coupling are all feasible.
While dimer I will revert to phenoxy radicals at elevated
temperatures, enolisation to A (2,2’-dihydroxybiphenyl) can
ultimately lead the way to DF. A has been found in trace
amounts (next to DF) in the pyrolysis of phenol[4] and from
the hydrogenolysis of anisole.[5] Also, at 500 °C authentic A
can be rapidly converted into DF with high yield,[3] presum-
ably by a radical (chain) process. The formation of dibenzo-
p-dioxin (DD) was mentioned in none of these cases; signi-
ficant proportions of (chlorinated) DDs arise, however,
when ortho-chlorine substituents are introduced.[3,4,6,7]
[a]
Center for Chemistry and the Environment, Leiden Institute of Coupling can also involve the oxygen, to give O–ortho-C,
Chemistry, Gorlaeus Laboratories, Leiden University,
O–para-C dimers. The probability for the formation of the
O–O dimer is low and, more importantly, it will be unstable
even at ambient temperature. Methyl radicals gave cresols
POB 9502, NL-2300 RA Leiden, The Netherlands
E-mail: R.Louw@chem.leidenuniv.nl
On leave (through TEMPUS grant IMG-97-Pl-2135) from
[‡]
´
Jagiellonian University, Faculty of Chemistry, Krakow, Poland in more than 90% (expectedly via the methylcyclo-
Eur. J. Org. Chem. 2000, 921Ϫ928 921
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
1434Ϫ193X/00/0306Ϫ0921 $ 17.50ϩ.50/0

I. Wiater, J. G. P. Born, R. Louw
FULL PAPER
hexadienones) and gave less than 10% anisole.[14] The ortho- It also implicitly means that these radicals are quite unre-
/para ratio of cresol is 1.2–1.5, essentially independent of active towards (air) oxygen.[20]
temperature. Armstrong et al.[15] investigated the products
Detailed insight into the thermal conversion of phenol/
of the phenoxy ϩ phenoxy combination upon the decom- phenoxy is a prerequisite for understanding the behaviour
position of di-tert-butyl peroxide (dTBP) in neat phenol of (poly)chlorinated analogues. The latter are demonstrated
at 140 °C over 24 h, and found all possible (enolised) precursors of the notorious family of polychlorinated di-
condensation products; they are, in decreasing order: benzo-p-dioxins and dibenzofurans (PCDD/Fs), (often
A Ͼ B Ͼ C ഠ D Ͼ E (Figure 1).
toxic) by-products of the combustion and other thermal
processes involving chlorine-containing materials.[21]
This paper addresses the chlorine-free basic system:
phenol. With quartz flow reactors, dilute vapours of phenol
were induced to form phenoxy radicals, and their fate was
monitored with emphasis on condensation products. At
temperatures from 500 K, free radical initiators, e.g., tert-
butyl hydroperoxide, served to form phenoxy in inert (N2)
atmospheres; at higher temperature (773 K), slow combus-
tion – with substantial degrees of conversion of phenol –
was also studied, the oxygen now enabling the generation
of phenoxy radical intermediates.
Results
Figure 1. Structure of compounds; DF ϭ dibenzofuran, DD ϭ di-
benzo-p-dioxin, A ϭ 2,2’-dihydroxybiphenyl, B ϭ 2,4’-dihydroxybi-
phenyl, C ϭ 2-phenoxyphenol, D ϭ 4-phenoxyphenol, E ϭ 4,4’-
dihydroxybiphenyl, F ϭ 4-dibenzofuranol, and G ϭ 2-dibenzofur-
anol
The reactivity of phenoxy radicals was studied in flow
reactors at 1 bar between 500–850 K, with emphasis on the
higher temperature range (above 700 K).
Three types of experiments were performed:
(1) Peroxide-initiated pyrolysis of phenol, between 500–
613 K, with residence times around 12 s.
(2) Nitromethane-initiated pyrolysis of phenol between
713 and 833 K, with residence times of ca. 8 s.
(3) Slow combustion of phenol together with an excess
Born repeated this experiment, and found analogous re-
sults.[4] Interestingly, in a comparable gas-phase flow experi-
ment at 340 °C, he found only A (major), C, and D with
little DF (5% based on A).
Phenolic compounds and the corresponding phenoxy of benzene at 773 K, with a residence time of approxim-
radicals play an important role in many biological and ately 45 s.
chemical processes. Lipids and other natural organic mat-
Data of representative examples of experiments 1, 2, and
erials are protected from oxidative degradation by 3 are collected in Table 1.
phenols.[16] The antioxidant action of sterically hindered
As seen in Table 1, the main product of phenoxyl com-
phenols is widely used to protect synthetic materials. bination in series 2 was dibenzofuran, with somewhat
Phenols can break (aut)oxidation chains by H-donation to, higher output levels at the lower end of the temperature
e.g., chain-carrying peroxyl radicals; the hindered, reson- range. As minor products, 2- and 4-phenoxyphenol (C and
ance-stabilised aryloxyl radical is in most systems too unre- D), and 2,2’-dihydroxybiphenyl (A) were also observed. The
active to continue the chain; they rather dimerise. In model compounds 2,4’- and 4,4’-dihydroxybiphenyl were not
oxidation studies of phenols with available unsubstituted found (detection limit 10–4, calculated on PhOH). The re-
ortho or para positions, dihydroxybiphenyl derivatives are sults differ markedly from those of series 1, where DF was
formed.[17] If all ortho and para positions are substituted (so close to the detection limit, and A is the most abundant
that tautomerisation to the biphenyl skeleton is impossible) ‘‘dimer’’ found. The exit gases of series 1 and 2 contained
dimerisation can still take place; the type of the product CO, C2H6, and also some C2H4 in series 2, as degradation
depends on structure and experimental conditions.[18] If products of dTBP. As expected, the products of methyl–
both ortho positions are occupied and the para position is phenoxy combination were also formed. The most abund-
free, p-diphenoquinones can be isolated. These arise from ant was o-cresol, from 45% in experiment 2.2 to 54% in
the para C–C coupled bisketo compound, through enolis- experiment 1.1; anisole was the least abundant product,
ation to the 4,4’-dihydroxybiphenyl and further oxida- from 8.7% in experiment 1.1 to 17% in experiment 2.3.
tion.[18b] Oxidative coupling of phenols, via phenoxy rad- Experiments 3.1 and 3.2 exemplify the slow combustion of
icals, is also of importance in biomimetic synthesis.[19] phenol diluted in benzene (1:30). Benzene was chosen as a
These examples of liquid-phase (aut)oxidation amply dem- diluent, because it is expected to be very stable under those
onstrate the capacity of phenoxy-type radicals to dimerize. conditions and, when reacting, to give similar products to
922
Eur. J. Org. Chem. 2000, 921Ϫ928

Gas-Phase Condensation of Phenoxy Radicals
FULL PAPER
Table 1. Gas-phase conversions of phenol: dimeric products
Exp.no
Temp.
[K]
Ratio of ‘‘dimers’’ output[a]
Sum, yield
[2 ϫ (mol/mol PhOH in)]
DF
:
A
:
C
:
D
1.1[b]
1.2
503
543
573
613
713
753
773
793
833
ND
0.009
0.011
0.05
20
:
:
:
:
:
:
:
:
:
1
1
1
1
1
1
1
1
1
:
2
:
:
:
:
:
:
:
:
:
1
0.0010
0.0010
0.0012
0.0010
0.0072
0.0063
0.0030
0.0028
0.0032
:
:
:
:
:
:
:
:
0.5
0.25
0.15
0.10
0.25
0.30
0.41
0.50
0.15
1.3
0.33
0.37
0.4
1.4
2.1[c]
2.2
27
0.4
2.3
20
0.45
0.40
0.33
2.4
16
2.5
22
DF
:
:
:
A
1
1
:
:
:
C
1.04
1.98
:
:
:
G
2.12
1.95
:
:
:
F
1.04
1.04
3.1[d]
3.2
773
773
773
843
5
0.01
0.01
51
3a[e]
DF, see text
DF, see text
3b[e]
[a] A, C, D, F, G, DF – see Figure 1; ND ϭ below detection limit. – [b] Series 1: Inputs (mol fraction): PhOH 0.02; tert-butyl hydroperoxide
(TBHP) from 0.004 in experiment 1.1 to 0.002 (exp. 1.4); dTBP: 10 mol-% on TBHP, carrier gas N2. Outputs of PhOH Ն 90%, TBHP
conversion ca. 85% in exp. 1.1 to complete (other exp’s); di-tert-butyl peroxide (dTBP) conversion ca. 20% in exp. 1.1, ca. 40% in exp.
1.2, ca. 90% (exp. 1.3), to completion in exp. 1.4. Other salient products: ethane; anisole/o-cresol/p-cresol with ratio 10:53:37, 2–3%,
[c]
based on PhOH. – Series 2: Inputs (mol fraction): PhOH 0.0042; MeNO2 0.0035; filler gas N2. Outputs of PhOH Ն 90% except exp.
2.4: 80%; conversion of MeNO2 Ͻ 10% (exp. 2.1), ca. 20% (exp. 2.2), 65% (exp. 2.3), 96% (exp. 2.4). Other salient products include
[d]
ethane, anisole, cresols, as above. For further details, see ref. [4] Series 3: Inputs (mol fraction): PhOH 0.00165 (exp. 3.1 ), 0.0014 (exp.
[e]
3.2); benzene 0.050; filler gas O2 and N2 (19:81). Outputs of PhOH Ն 55%. – Series 3a: Inputs (mol fraction): benzene 0.05; filler gas
O2 and N2 (19:81).
that of phenol. For comparison, the slow combustion of present, comprise less than 1% on DF. Compared with the
benzene only (exp. 3a), under the same conditions, was also output of experiment 3a (benzene only), the output of
conducted. The ‘‘major’’ product of this blank run was phenol was six times higher. The selectivities from A to
phenol, around 0.01%, based on the input of benzene. The phenol and DF were 20–25% each. The latter output is less
only other detectable product was DF (about 1% based on than half of that from neat A, but still quite significant.
phenol, which is only ca. 0.8% of the DF obtained in exp.
The thermolysis of A in the lower temperature range
3.1 and exp. 3.2).
(560–600 K), initiated by TBHP, was also studied briefly.
In experiments 3 the conversion of phenol was approxim- Diluted in benzene (1:50) with about 1% peroxide on ben-
ately 40%. The main product was again DF, with a yield of zene, and with residence times around 90 s, ca. 20% of A is
1–2% based on converted phenol. Further products found converted at 560 K and ca. 80% is converted at 600 K. The
were A, C, D, 4-dibenzofuranol (F), and 2-dibenzofuranol selectivity to DF is low, in the 1–2% range only. The forma-
(G), but in amounts nearly two orders of magnitude less tion of products such as biphenyl (from benzene) and the
than DF. After concentration of the obtained sample, a mono- and dimethyl ethers of A and of methylated cresol,
large number of other products, at trace levels (3–4 orders as well as derivatives such as 2,2’-dihydroxy-5-methylbi-
of magnitude below DF), could also be identified. These phenyl, at least partially explain this. The oxyradical which
include DD, 2,4’-dihydroxybiphenyl (B), isomers of benzo- can arise from A will be relatively long-lived at these mild
bisbenzofuran, phenoxyphenoxyphenol, and also methyl- temperatures, so it can undergo bimolecular reactions with
ated derivatives of A. The gaseous products were CO, CO2, other radicals, e.g., CH3.
CH4, and C2H2.
Slow combustion was focused on again, and several ex-
To better understand the chemistry of the condensation periments with 2-phenoxyphenol (C) diluted in benzene,
of phenoxy radicals, two possible intermediate products on 1:220–600, were conducted at 773 K. Conversion of C was
the way to DF, namely 2,2’-dihydroxybiphenyl (A) and 2- about 95% throughout. Remarkably, the major product next
phenoxyphenol (C), were chosen and their reactivity was to phenol (about 10% on C input) was 4-dibenzofuranol (F)
examined.
(about 4%). Two more products were detectable, viz. DF and
When dilute vapours of A are subjected to thermolysis DD, with reproducible yields of 14 and 9% on F, respectively.
(in N2 atmosphere) or to slow combustion at 773 K with Based on the slow combustion of benzene only (exp. 3a), it
residence times of around 100 s, conversion is quantitative, turns out that about 25% of phenol and less than 10% of the
with high yields of DF (80 and 55%, respectively).[3] Once DF can have been produced from benzene.
it was known that A is a precursor of DF, the slow combus-
tion of A in a large excess of benzene (1:200) was con-
ducted, as a step towards more realistic conditions. Under
Discussion
(a) Radical-Induced Lower-Temperature Approaches
the conditions of experiments 3, the conversion of A was
99.93% and the only important organic products were
Both TBHP and nitromethane generate methyl radicals
phenol and DF; other products with two benzene rings, if which will abstract hydrogen from phenol to form phenoxy
Eur. J. Org. Chem. 2000, 921Ϫ928
923

I. Wiater, J. G. P. Born, R. Louw
FULL PAPER
radicals (Equation 2). Judging from recent data reported by (NO2) Ͼ 10–5 ; the half-life of ArOH in this process will
Pedulli et. al.,[39] k2 should be around 106 –1 s–1 at 600 K. be less than 0.1 s.
This can be followed by combination of phenoxy with
methyl radicals to give anisole, plus o-/p-cresols, the latter
two compounds after tautomerisation of the respective keto
forms (Equation 3).[14]
(b) Slow-Combustion Series
Combustion reactions are mostly of a radical-chain na-
ture.[24] In the slow combustion of phenol, when free-radical
initiators are not deliberately used, measurable reaction oc-
curs above 700 K. O2, a triplet biradical, may abstract H
from phenol (Equation 9), but this reaction – endothermic
by ca. 40 kcal/mol, will be several orders of magnitude too
slow to account for observed degrees of reaction. Without
detailing the mechanism here, radical species (e.g., HO2•
CH3• ϩ C6H5OH Ǟ CH4 ϩ C6H5O•
(2)
CH3• ϩ C6H5O• Ǟ C6H5OCH3 ϩ o-CH3C6H4OH ϩ
p-CH3C6H4OH
(3)
The site selectivities of O/o-C/p-C are ca. 10:55:35 in the
peroxide series, and ca. 15:45:40 in the runs with MeNO2;
these results tally with those of Mulcahy and do not deviate
much from what expected on the basis of spin-density
data.[13] The strengths of the bonds made, 65 kcal/mol for
O–C in anisole, and similar strengths in the keto-cresols
(vide infra),[22] ensure that the CH3 additions to PhO• are
irreversible in the temperature range studied. Therefore,
tautomerisation to the ‘‘real’’ cresols is the most likely fate
of the latter. A simple intramolecular H-shift will be rather
slow due to orbital-symmetry restrictions,[23] but polar mo-
lecules and/or surfaces can be of great help here. We will
come back to this aspect in section (c).
•
and, more importantly, OH at higher T, Equation 10) will
be instrumental in the (depending on conditions, reversible)
formation of phenoxy radicals.[24,25]
PhOH ϩ O2 Ǟ PhO• ϩ HO2•
(9)
PhOH ϩ Z• Ǟ PhO• ϩ ZH (Z ϭ HO2•, •OH, •O•)
(10)
Phenoxy radicals can recombine as described in section
(a) or can undergo decomposition.[8]
Product patterns and levels strongly depend on temper-
ature. So the reaction of benzene/PhOH (20:1) at 843 K led
to larger outputs of phenol (due to its substantial formation
from benzene at this temperature), high levels of CO and
CO2 and concomitant higher outputs of DF. Also, some
products not seen in experiment 3a, such as benzonaphtho-
furan, are detected.
In experiments 3 at 773 K, about one-half of phenol is
smoothly converted, and by far the major condensation
product is again DF. Only a few percent (based on DF) of
A and B are formed, and products involving the para site of
phenoxy (e.g., D) are even much less abundant. Relatively
important are F and G, hydroxy derivatives of DF. Given
the extremely low yield of phenol upon slow combustion of
benzene (exp. 3a), F and G have not been formed by sec-
ondary oxidation of DF, but rather via some other ‘‘dimer’’
Analogously to methyl–phenoxy recombination, phenoxy
radicals can recombine through C–O coupling (requiring
one following keto–enol tautomerisation) and by C–C-
bonding at the ortho and/or para positions (needing two re-
arrangements).
2 C6H5O• Ǟ 2-C6H5OC6H4OH (C)
2 C6H5O• Ǟ 4-C6H5OC6H4OH (D)
2 C6H5O• Ǟ 2,2’-HOC6H4–C6H4OH (A)
2 C6H5O• Ǟ 2,4’-HOC6H4–C6H4OH (B)
2 C6H5O• Ǟ 4,4’-HOC6H4–C6H4OH (E)
(4)
(5)
(6)
(7)
(8)
Dimerisation to give C6H5O–OC6H5 will be highly re- precursors. In the formation of DF, o-C/o-C combination is
versible due to its very weak peroxide bond which, due to an important (but not necessarily the only) product chan-
the resonance energy in C6H5O•, will be ca. 15 kcal/mol nel, as the results of slow combustion with added A demon-
only.
strate. We will discuss other possible pathways to/via dimers
In both series 1 and 2 only three of the five possible prod- of phenoxy in more detail after the discussion of the therm-
ucts have been positively identified: A, C, and D; the p-C/ ochemical–kinetic analysis of section (c). Here we will bri-
p-C and o-C/p-C condensation products, if present, were efly comment on the results of the ‘‘blank’’ slow combus-
below 1% based on A. This contrasts with the results of tion of benzene only.
Armstrong et al.[15] on the peroxide-initiated liquid-phase
At 773 K, benzene is still almost inert. Traces of phenol
reaction (140 °C), wherein the overall site selectivities and DF (ratio 100:1) are the only aromatic products seen.
centred around O/o-C/p-C ഠ 10:60:30. This points at re- After evaporating nearly all of the benzene from the prod-
versibility in the gas phase, at least with respect to the p- uct, the MS analysis begins to show ‘‘noise’’ from ultratrace
site in phenoxy. Thermokinetic aspects are discussed in sec- compounds, which may also stem from contaminants in the
tion (c); it will suffice here to point out that bonds between benzene. To get more meaningful results, the temperature
two PhO radicals are much weaker than those between PhO must be well above 773 K. In the slow combustion of ben-
and methyl.
zene at 843 K (exp. 3b), yields of phenol and DF are in-
At T Ͻ 600 K, series 1, outputs of DF are very minimal, creased (PhOH 2% on benzene input, DF 1.5% on PhOH).
but in series 2, (T Ͼ 700 K) it is the major product. If A is Some other products are seen by GC-MS, especially after
accepted to be the intermediate, its conversion into DF is sample concentration: Biphenyl ഠ 50% on DF; also traces
probably facilitated by NO2; H-abstraction from ArO–H by of other PAH, for example, naphthalene, etc.; C and D each
NO2 will be mildly endothermic (ca. 10 kcal/mol), but with are present in a few percent, based on DF. Furthermore,
924
Eur. J. Org. Chem. 2000, 921Ϫ928

Gas-Phase Condensation of Phenoxy Radicals
FULL PAPER
some methylated derivatives of biphenyl and DF, as well as
B and benzobisbenzofuran, each with yields two to three
orders of magnitude less than DF; finally traces of DD,
three to four orders of magnitude less than DF, were de-
tected. The product pattern is not unlike that of the slow
combustion of phenol/benzene, but the levels are much
lower.
Figure 2. Structures of compounds IV, V, and VI
(c) Thermochemical–Kinetic Aspects
In slow combustion and thermolysis at elevated temper-
ature (Ͼ 700 K), net rates of formation of DF, the major
condensation product from two PhO• radicals imply rate
constants of 108–108.5 –1 s–1.[5,6] This also means substan-

tial reversibility (if not equilibration) in the process: 2 PhO
o dimers, as the forward rate constant will be close to
1010 –1 s–1.[26] Then, site selectivities for combination will

lose relevance at the expense of (differences in) bond
strengths, combined with rates for consecutive reactions,
e.g. tautomerization.
We will now discuss this for three of the five possible
PhO/PhO dimerizations, taking into account the best avail-
able thermochemical data; see Scheme 2.[10] Standard gas-
phase heats of formation of A, C, and E have, to our know-
ledge, not been directly measured, but can be approximated
well, by starting from biphenyl (43.5 kcal/mol) and diphenyl
ether (12.5 kcal/mol),[27] and by employing the increment
benzene Ǟ phenol of –43.[28] To arrive at the enthalpies of
I and III, twice the heats of ‘‘ketonisation’’ must be added.
Accepting that V is 4 kcal less stable than IV[10] (see Fig-
ure 2), we place the absolute value for IV at –8 kcal/mol,[29]
on various grounds, so 15 kcal/mol endothermic, compared
to phenol. Applying the respective increments to I and III
leads to the bracketed values in Scheme 2. The enthalpy of
II is based on that of IV, since replacement of an H at (sp3)
carbon, R–H Ǟ R–OPh, R ϭ e.g., Me or Et, has almost
no change on the enthalpy of formation. From Scheme 2
the enthalpy diagram of Figure 3 is obtained, which can
serve to rationalise rates and products, also as a function
of temperature.
Figure 3. Enthalpy diagram (cf. Scheme 2)
At lower temperatures, say up to 400 K, any of the five
possible combinations can be formed, obeying the kinetic
site selectivities in the phenoxy radical, and the dimers will
tautomerize to the final aromatic products, certainly in the
liquid phase. With increasing temperature, in the vapour
phase, redissociation comes into play, first for III, and last
for I. If we accept an Arrhenius log A/s–1 value of 15
throughout, calculated half-lives at 700 K will be ca. 0.02 s
(for I), 0.002 s for II, and ca. 1 µs for III. At 800 K, the k
value for dissociation of I is ca. 105 s–1, that for III is two
orders of magnitude larger. Recalling that, in slow combus-
tion, the (net) rate of condensation of two PhO• groups to
DF is 108ϩ –1 s–1, up to two powers of ten below k for

radical/radical combination, formation of I–III (and the
other two isomers, from O/p-C and o-C/o-C coupling) must
be reversible. Then, I will greatly predominate. In order to
go to the final product(s) – taking 800 K as a reference
again – the (pseudo-first-order) rate constant to describe
the consecutive enolisation, I Ǟ A, has to be 103–104 s–1 to
satisfy the thermokinetic model. (At the same time, this rate
constant should not be so very much larger. When 106 s–1
is exceeded, combination of two PhO• groups will no longer
be reversible, and overall rates (to DF) would approach the
radical/radical encounter rate limit).
Direct 1,3-H shifts from C to O will be (too) slow, being
‘‘forbidden’’ due to orbital symmetry mismatch.[23] Neither
this, nor ‘‘substitution of hydrogen’’ by some unprecedented
simultaneous H shift, as very recently advanced by Hagen-
maier[31] can be considered as a satisfactory mechanistic in-
terpretation. A clue to this is the behaviour of allyl phenyl
ether upon (attempted) gas-phase Claisen rearrange-
ment.[32] This reaction does not occur in dry, inert environ-
ments (this implies that the first step to VI is endothermic,
Scheme 2. Pathways for the condensation of phenoxy radicals
Eur. J. Org. Chem. 2000, 921Ϫ928
925

I. Wiater, J. G. P. Born, R. Louw
FULL PAPER
which puts a lower limit to its heat of enolisation[29]). With
proper (e.g., hydroxylic) agents the overall Claisen two-step
reaction to o-allylphenol proceeds with rates consistent with
extrapolated liquid-phase data. These phenomena can also
be explained by enolisation rates such as those mentioned
above.[33]
(d) The Fate of Intermediates
In slow combustion, 2-phenoxyphenol (C) is much less
stable than phenol itself (95 vs 45% conversion at 773 K,
respectively). While formation of the 2-phenoxyphenoxy
radical may be easier due to a somewhat lower O–H bond
strength, the increased reactivity of this radical is in our
view more important. One possibility is fragmentation, (a)
in Scheme 3; the o-quinone will be rapidly degraded, and
phenyl radicals can in part form phenol, via phenoxy rad-
icals formed upon reaction with O2.[34,35] The intramolecu-
lar reactions (b), (c), and (d) in Scheme 3, adequately ex-
plain the formation of F, the most prominent product, DF
and DD.[36]
Scheme 4. Slow combustion of 2,2’-dihydroxybiphenyl (A)
The slow combustion of 2,2’-dihydroxybiphenyl (A) gives
In addition to DF and the other, intermediate, products
essentially only one condensation product, DF, in large built from two phenoxy entities as discussed before, several
yields. Here, splitting of the corresponding aryloxy radical ‘‘tricyclic’’ condensation products were detected by GC-
is not possible. The overall dehydration A Ǟ DF can be MS, in particular, (isomers of) benzobisbenzofuran and of
rationalized again in terms of free radical steps [Scheme 4, phenoxy-phenoxyphenol. The simplest rationale is by con-
(a)], but it is not immediately clear why the alternative (b) densation of phenoxy with a higher homologue, such as α
with about the same thermokinetic characteristics, does not (Scheme 4) or γ (Scheme 3). These combination reactions
seem to occur. (The expected product, 1-hydroxydibenzofu- can indeed be competitive with unimolecular steps of, e.g.,
ran, could not be detected; if present it must be below 0.1% α to DF, due to the substantial barriers of around 35 kcal/
based on DF). The overall high rate of conversion A Ǟ DF mol for such reactions.[36] Scheme 5 illustrates how radical
suggests that the loss of H2O may occur at least in part by α can condense with phenoxy, by o-C/o-C coupling, either
nonradical reactions (at the wall, or promoted by H2O), by radical or nonradical pathways, to (one isomer of)
schematically depicted as an intramolecular reaction (a’).
benzobisbenzofuran. Whether or not still higher condens-
Scheme 3. Conversion of 2-phenoxyphenol (C)
926
Eur. J. Org. Chem. 2000, 921Ϫ928

Gas-Phase Condensation of Phenoxy Radicals
FULL PAPER
Scheme 5. Rationale for the formation of benzobisbenzofuran
N2, regulated with mass-flow controllers, was passed through an
impinger filled with the liquid substrate. For phenol itself, an oil-
operated thermostatic bath was used to maintain the molten com-
pound at a fixed temperature (74 °C) in a stainless steel impinger;
a stable vapour pressure at the required level was thus created. The
mixture of dTBP and TBHP (initially 20:80) was kept in a Pyrex
evaporator at 20 °C and was added to the phenol by passing
through a calibrated stream of N2. Nitromethane was evaporated
from a stainless steel impinger kept at 20 °C by a water-run thermo-
static bath.
ates were formed is unknown, as these products were not
volatile enough to be seen by our GC-MS analysis.
In summary, we conclude that in the gas phase at moder-
ately elevated temperatures, up to approximately 850 K,
phenoxy radicals condense to yield dibenzofuran, DF, with
only (very) small proportions of other products of similar
complexity, including dibenzo-p-dioxin, DD. This may be
understood on the basis of thermochemical kinetics, if the
best available consistent data is used; this indeed favours
the product channel by o-C/o-C coupling with subsequent
tautomerization and, overall, dehydration steps.
In experiment 3, gas flows of N2 and O2 were regulated with needle
valves and were measured with capillary flow-meters. Liquid react-
ants were introduced by means of a calibrated motorised syringe
pump (B.Braun Melsungen, Perfusor VI type 871222/0) through a
gas-tight rubber septum, and they were vaporised into the gas flow
before they entered the reactor.
Experimental Section
Purity of Chemicals: Tank N2 99.99% and O2 99.99% were supplied
by Hoekloos in standard cylinders. 20% Di-tert-butyl peroxide in
80% tert-butyl hydroperoxide (Fluka, pure), nitromethane (Janssen,
96%) were used without further purification, except for the experi-
ments with 2,2’-dihydroxybiphenyl, for which the tert-butyl hydro-
peroxide was distilled prior to use. Phenol (Merck, pro analysis),
benzene (Merck, for spectroscopy, distilled), toluene (J. T. Baker,
Ͼ99.5%), 2,2’-dihydroxybiphenyl (Janssen Chimica, 99%) were
checked by GC to be of adequate (Ͼ99%) purity and were used
as such. For 2-phenoxyphenol synthesis: 2-bromophenol (Janssen
Chimica, 98%), KOH (Boom B.V Meppel, 85%), diethyl ether (J.
T. Baker, Ͼ99.5%), petroleum ether 60–80 (Baker, pure) were used
without further purification. Reference compounds: bromoben-
zene, dibenzofuran, dibenzo-p-dioxin, 2-dibenzofuranol, and 4-di-
benzofuranol were all high-grade (Ͼ99%) products.
The entrance and exit tubes of the reactors were heated to over 200
°C by wrapped heating tapes to prevent condensation of the less
volatile organic substances.
For series 1 and 2, an on-line set-up was used. A two-way switching
valve led the gas mixture either directly or via the reactor towards
the on-line gas chromatograph for analysis before or after reaction.
The exit gas, after mixing with methane (internal standard) was
also analysed on-line by GC-FID with a packed carbowax (2 m ϫ
1/8’’ ID) column and a methaniser.[4]
In experiment 3, condensable products were collected in a cold trap.
The trap, containing toluene with an internal standard (bromoben-
zene), was cooled by acetone with liquid nitrogen. Exit noncon-
densable gases were analysed with a Packard series 428 GC
equipped with an FID detector, Carboplot 007 column and a me-
thaniser, calibrated by independent injections of a standard gas
mixture.
General: All experiments were conducted in quartz cylindrical flow
reactors of 6.2 mL (length 47 mm, 13 mm ID) for series 1 and 2,
and of 340 mL for experiments 3; these reactors were placed either
horizontally or vertically in an electrically heated oven. The tem-
perature was controlled by a proportional regulator and was mon-
itored with chromel-alumel thermocouples displayed on digital
thermometers (Therma 1, type ST-861-107). The upper and lower
ends of the oven were insulated by quartz wool. For further details
on these set-ups see Refs.[4,6]
Products in the toluene catch were quantified with a Hewlett Pack-
ard 5890A gas chromatograph with FID with a CP-SIL5-CB col-
umn (50 m ϫ 0.32 mm ID), operated by HP Chem Stations. Tolu-
ene solutions (2 µL portions) were injected by an autosampler
(Hewlett Packard 6890 Series) at a split ratio of 1:15. Absolute
concentrations were deduced from the peak areas, after calibration
by injecting standard mixtures of known composition. A Hewlett
Packard 5890 GC-MS was used to identify unknown products.
In experiments 1 and 2, substrates were continuously fed into the
reactor by evaporation of the liquid. For this a constant inflow of
Eur. J. Org. Chem. 2000, 921Ϫ928
927

I. Wiater, J. G. P. Born, R. Louw
FULL PAPER
[23]
[24]
R. B. Woodward, R. Hoffmann, The Conservation of Orbital
Symmetry, Acad. Press, 1970, chapter 7.
Physical and Chemical Aspects of Combustion (Eds.: F. L.
Dryer, R. F. Sawyer), Gordon and Breach, 1997, chapters 1
and 4.
Some samples were concentrated by solvent evaporation with a
gentle stream of N2, to allow better measurement of trace products.
Synthesis of 2-Phenoxyphenol: 2-Phenoxyphenol was prepared ac-
cording to the procedure described in ref.[38] A comparable amount
of DD was formed. The product mixture was liquefied in toluene,
treated with 10% KOH and extracted with water. This extract was
acidified with diluted HCl, extracted with toluene and checked by
GC-MS. Toluene was gently evaporated with a stream of N2. The
crystalline product was dissolved in n-pentane and was kept at –20
°C for a few days for recrystallisation. The purity of the sample
used in the slow combustion with benzene was checked by GC to
be 99% (1% of DD).
[25]
[26]
[27]
[28]
P. Mulder, R. Louw, J. Chem. Soc., Perkin Trans. 2 1986,
1541–1550.
Z. Tao, Z. Li, Int. J. Chem. Kinet. 1999, 31, 65–72 and refer-
ences cited therein.
Thermochemical Data of Organic Compounds (Eds.: J. B. Ped-
ley, R. D. Naylor, S. P. Kirby), Chapman and Hall, NY, 1986.
This approach means that the C–OH bond strengths in A and
C are taken to be identical to that in phenol. Likewise,
∆Hf0298 of hydroxy-biphenyl should be ca. 0 kcal/mol. Remark-
ably, the NIST compilation,[10] based on the original French
paper,[37] advocates ϩ8.5 for the p-isomer and 4 kcal/mol for
ortho-hydroxybiphenyl. The idea that these C–OH bonds are
much (4–8.5 kcal) weaker than in phenol and the idea that the
para form should be no less than 4 kcal destabilized relative to
ortho form must be in error.
[1]
Beilsteins Handbuch der Organischen Chemie 1933, XVII, 71.
[2]
R. Cypres, B. Bettens, Tetrahedron 1975, 31, 359–365.
[3]
J. G. P. Born, R. Louw, P. Mulder, Chemosphere 1989, 19,
401–406.
J. G. P. Born, Ph. D. Thesis 1992, Leiden University (in En-
glish).
I. W. C. E. Arends, R. Louw, P. Mulder, J. Phys. Chem. 1993,
97, 7914–7925.
Ref.[10] advances that ∆Hf0298 (IV) ϭ –17 ± 3 kcal/mol, so
roughly only 6 kcal greater than that of phenol. As will be ex-
plained in more detail elsewhere,[30] this cannot be correct. One
argument is that it would make VI about 5 kcal more stable
than allyl phenyl ether, with the consequence that, contrary to
observation,[32] the gas-phase Claisen rearrangement should
also proceed in an inert environment (enolisation of VI occur-
ring after cooling and condensing the reactant vapours). An-
other reason for rejecting this is that the value of –17 kcal/
mol for IV is very much lower than that calculated from group
additivity, from cyclohexadiene (25) and the increment for Cϭ
O replacing –CH2–, i.e., –25, hence 0 kcal/mol; it is unrealistic
to have an additional delocalization energy of more than
15 kcal/mol in this dienone system.
[29]
[4]
[5]
[6]
H. H. Grotheer, R. Louw, 26th Symp.(Int.) on Combustion
1996, 2405–2411.
[7]
S. S. Sidhu, L. Maqsud, B. Dellinger, G. Mascolo, Comb. and
Flame 1995, 100, 11–20.
[8]
C. Y. Lin, M. C. Lin, J. Phys. Chem.1986, 90, 425–431.
[9]
W. M. Shaub, W. Tsang, Environ.Sci. Technol. 1983, 17, 721–
730.
[10]
NIST Chemistry WebBook (Ed.: W. G. Mallard), NIST Stand-
[30]
[31]
[32]
[33]
R. Louw, D. Santoro, (to be) submitted to J. Phys. Chem.
R. Weber, H.-P. Hagenmaier, Chemosphere 1999, 38, 529–549.
R. Louw, Recl. Trav. Chim. Pays-Bas 1971, 90, 1119–1122.
ard Reference, Database 69 – November 1998 Release, Na-
tional Institute of Standards and Technology, Gaithersburg,
MD, 1998.
[11]
With ∆Hf0298 values of –1 kcal/mol for PhO–CH2–CHϭCH2,
and ϩ7 kcal/mol for VI, and accepting rate parameters for the
forward reaction: log A/s–1 ϭ 11, Ea ϭ 32, those for the reversal
of VI to starting compound will be log A ഠ 11 and E ϭ
24 kcal/mol. At 800 K this means k ഠ 104.4 s–1. The vapour of
added acetic acid could make the ‘‘rate constant’’ for competi-
tive enolisation equal to or larger than this.
D. M. Chipman, R. Liu, X. Zhou, P. Pulay, J. Chem. Phys.
1994, 100, 5023–5035.
[12]
Y. Qin, R. A. Wheeler, J. Chem. Phys. 1995, 102, 1689–1698.
[13]
K. Tuppurainen, I. Halonen, J. Tarhanen, J. Ruuskanen, J.
Mol. Struct. (Theochem). 1994, 312, 139–144.
[14]
M. F. R. Mulcahy, D. J. Williams, Aust. J. Chem. 1965, 18,
20–38.
[34]
[35]
[36]
[15]
P. Frank, J. Herzler, T. Just, C. Wahl, Symp. Int. Combust. Proc.
D. R. Armstrong, C. Cameron, D. C. Nonhebel, P. G. Perkins,
1994, 25, 833–840.
J. Chem. Soc., Perkin Trans. 2. 1983, 563–568.
[16]
M. Foti, K. U. Ingold, J. Lusztyk, J. Am. Chem. Soc. 1994,
P. M. Sommeling, P. Mulder, R. Louw, D. V. Avila, J. Lusztyk,
K. U. Ingold, J. Phys. Chem. 1993, 97, 8361–8364.
116, 9440–9447.
[17]
W. A. Waters, C. Wickham-Jones, J. Chem. Soc. 1951, 812–823.
Without detailing the thermochemistry, reaction (a) is ca.
47 kcal endothermic, while the reactions by (b)–(d) are associ-
ated with activation barriers of some 35 kcal/mol. While scis-
sion (a) will have an Arrhenius A value of ca. 1015 s–1, the other
are intramolecular additions with much lower A values, e.g.,
1010 s–1. As a result, (a) can effectively compete with (b)–(d),
the more so at higher temperatures.
G. Bertholon, M. Giray, R. Perrin, M.-F. Vincent-Falaquet-
Bernet, Bull. Soc. Chim. France 1971, 9, 3180.
E. Klarmann, L. W. Gates, V. A. Shternov, J. Am. Chem. Soc.
[18] [18a] R. F. Moore, W. A. Waters, J. Chem. Soc. 1954, 243–246. –
[18b]
E. C. Horswill, K. U. Ingold, Can. J. Chem. 1966, 44 (3),
263–268.
[19]
Comprehensive Organic Chemistry, Vol. 5 (Ed.: E. Haslam),
Pergamon Press, New York, 1979, Ch. 29, pp. 956–957.
[20]
D. Wang, I. György, K. Hildenbrand, C. von Sonntag, J. Chem.
[37]
[38]
[39]
Soc., Perkin Trans. 2 1994, 45–55.
[21]
R. Addink, K. Olie, Environ.Sci. Technol. 1995, 29, (6), 1425–
1435.
[22]
1932, 54, 1204–1211.
The standard heats of formation are –18 kcal/mol for anisole
and approximately –30 kcal/mol for o- and p-cresol. The
‘‘keto’’ structures of the latter make them some 15 kcal/mol less
stable;[33] their heats of formation and hence methyl–R bond
strengths are close to that of anisole.
P. Franchi, M. Lucarini, G. F. Pedulli, L. Valgimigli, B. Lunelli
J. Am. Chem. Soc. 1999, 121, 507–514.
Received June 18, 1999
[O99364]
928
Eur. J. Org. Chem. 2000, 921Ϫ928