Investigating homogeneous Co/Br-/H2O2 catalysed oxidation of lignin model compounds in acetic acid

Article abstract of DOI:10.1039/c8cy01902a

Oxidation of α-O-4, β-O-4 and monomeric lignin model compounds by Co/Br^-/H₂O₂ in acetic acid at 70 °C was investigated. Co and Br^- were introduced as cobalt acetate tetrahydrate and KBr respectively. The degree of methoxylation of the substrate was found to affect its reactivity. For the α-O-4 model compounds, increased methoxylation of the benzyl moiety influenced product selectivity, while increased methoxylation of the phenolic moiety increased substrate conversion. The β-O-4 model compounds exhibited similar conversions to the α-O-4 models, but afforded a lesser amount of monomeric products. The formation of phenol and guaiacol from α-O-4 bond cleavage inhibited substrate conversion and sequestered oxidation products due to the formation of phenoxy radicals and polyguaiacols. Similar to the α-O-4 model compounds, increased methoxylation of the monomers changed the types of products formed, from polyphenols (phenol and guaiacol) to quinones (syringol). The behaviour of syringol was explored extensively, revealing that the corresponding 1,4-hydroquinone strongly inhibited syringol oxidation, and the syringol oxidation product, 4,4′-diphenoquinone, was susceptible to over-oxidation. The deleterious effects of phenols on oxidation of an α-O-4 model could be reduced by substitution of the Br^- co-catalyst with N-hydroxyphthalimide (NHPI), improving substrate conversion and product selectivity.

Full text of DOI:10.1039/c8cy01902a

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Investigating homogeneous Co/Br−/H2O2
catalysed oxidation of lignin model compounds in
Cite this: DOI: 10.1039/c8cy01902a
acetic acid†
Edwin B. Clatworthy,
Richard T. Maschmeyer, Anthony F. Masters
Julia L. Picone-Murray, Alexander K. L. Yuen,
and Thomas Maschmeyer
*
Oxidation of α-O-4, β-O-4 and monomeric lignin model compounds by Co/Br−/H2O2 in acetic acid at 70
°C was investigated. Co and Br− were introduced as cobalt acetate tetrahydrate and KBr respectively. The
degree of methoxylation of the substrate was found to affect its reactivity. For the α-O-4 model com-
pounds, increased methoxylation of the benzyl moiety influenced product selectivity, while increased
methoxylation of the phenolic moiety increased substrate conversion. The β-O-4 model compounds
exhibited similar conversions to the α-O-4 models, but afforded a lesser amount of monomeric products.
The formation of phenol and guaiacol from α-O-4 bond cleavage inhibited substrate conversion and se-
questered oxidation products due to the formation of phenoxy radicals and polyguaiacols. Similar to the
α-O-4 model compounds, increased methoxylation of the monomers changed the types of products
formed, from polyphenols (phenol and guaiacol) to quinones (syringol). The behaviour of syringol was ex-
plored extensively, revealing that the corresponding 1,4-hydroquinone strongly inhibited syringol oxidation,
and the syringol oxidation product, 4,4′-diphenoquinone, was susceptible to over-oxidation. The deleteri-
ous effects of phenols on oxidation of an α-O-4 model could be reduced by substitution of the Br− co-
catalyst with N-hydroxyphthalimide (NHPI), improving substrate conversion and product selectivity.
Received 11th September 2018,
Accepted 11th December 2018
DOI: 10.1039/c8cy01902a
rsc.li/catalysis
kyl aromatics to their respective aldehydes, alcohols and carbox-
ylic acids. The AMOCO Mid Century Process (Schemes 1 and 2),
Introduction
Lignin is a polymer composed of a variety of cross-linked aro-
matic sub-units and accounts for up to 30 wt% of woody bio-
mass. It represents the largest source of potentially renewable
aromatics.1 The major source of industrial lignin is derived
from pulp and paper manufacturing where, once separated
from the cellulose fibres and recovered from black liquour, it is
typically burnt as a low grade fuel. Despite its availability and
volume, lignin is currently an under-valorised resource, with
only ∼2% of the lignin produced annually being used for com-
mercial applications other than process heat in the pulping op-
erations.2,3 The established nature and scale of the petro-
chemical industry, combined with the technical challenges
presented by the structural complexity of lignin, have provided
limited incentive to valorise lignin to aromatic chemicals. How-
ever, the renewed push for renewable chemicals is changing
the landscape, and new opportunities are emerging.4
utilises cobalt and manganese acetates with hydrobromic acid
in acetic acid in the presence of pure oxygen.5 The success of
this system revolves around several factors including its high ac-
tivity over a broad range of temperatures (25–300 °C), substrate
flexibility, simple product recovery, high yields (>90%), and sim-
ple catalyst components.6,7 The fundamental system operates
with a cobalt salt, a source of bromide, and an oxidant (typically
O2 or H2O2). Cobalt (typically as the acetate) reacts rapidly with
peroxides and peracids, and generates benzylic radicals via a
CoIJIII) species. Bromide (HBr, NaBr or KBr) facilitates both rapid
generation of benzylic radicals and reduction of CoIJIII) to CoIJII).
Adding different metal acetates (Mn, Zr) to the cobalt system
can result in even higher activities and selectivities, e.g. replac-
ing 20% Co with Mn resulted in a 5-fold increase in the rate of
oxidation of the intermediate p-toluic acid.8–10 In addition,
Cobalt is one of the largest scale homogeneous oxidation cat-
alysts in industrial chemistry, primarily in the conversion of al-
School of Chemistry, The University of Sydney, Sydney, NSW 2006, Australia.
E-mail: thomas.maschmeyer@sydney.edu.au; Fax: (+61) 0293513329;
Tel: (+61) 0293512581
Scheme 1 AMOCO Mid-Century method for the oxidation of p-xylene
to terephthalic acid. Adapted from Partenheimer.6
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c8cy01902a
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
The propagation reaction is
kp


(4)
ROO  RH  ROOH  R
At steady-state, the rate of termination equals the rate of
initiation, and so
k2[ROOH][RH] + kcat[ROOH][CoII] = kt[ROO˙]2
(A)
So
Scheme 2 General mechanism for the AMOCO MC catalyst. Adapted
from Partenheimer.6

II 

ROOH
k
RH  k Co


2 



cat

2

(B)
ROO



kt
cyclohexane can be oxidised by a similar catalytic system, ulti-
mately to adipic acid in the DSM process.11 Both current large-
scale industrial oxidations operate at ca. 170 °C and 10–30 atmo-
spheres of air and use cobalt-based catalysts. They differ, how-
ever, in that the cyclohexane oxidation is performed selectively
to the alcohol and ketone at low conversions, whereas the arene
is oxidised to the carboxylic acid in high conversion and em-
ploys a bromide co-catalyst. The specific structure/speciation of
the Co/Br− catalyst is unknown, but investigations by a number
of different researchers have attempted to reveal fundamental
aspects. Solution studies of cobalt acetate and bromide in acetic
acid and water (5–15 wt%), comparable to industrial conditions,
report that the catalyst solution contains a mixture of different
active catalyst species. As the fraction of water in the solvent in-
creases less bromide is coordinated directly by cobalt centres,
but instead has been suggested to remain in a second coordina-
tion sphere held there by ion pairing.12 A number of published
kinetic studies of bromide speciation have suggested that the
main active species are metalIJII) dibromides. These species are
oxidised by metalIJIII) centres and subsequently undergo sponta-
neous reductive elimination to generate dibromide radical
anions.13–15 At least three different cobaltIJIII) species, including
dimers and a mixed valence CoIJII/III) trimer, have been proposed
while studies of oxidised acetic acid/water solutions of cobalt ac-
etate in the presence of pyridine have revealed dimeric, trimeric,
cubane and octo cobaltIJIII) structures.16
If the rate of propagation, rp, is given by
d RH



(C)
(D)
 r  kp ROO RH



p
dt
Then, substituting for [ROO˙]

II 

k2 ROOH RH  k ROOH Co
 cat 




r  kp RH


p
kt

II 

k2 ROOH RH  k ROOH Co
 cat 



d RH



(E)
 kp RH


dt
kt
Furthermore, Hermans has recently reviewed free radical
hydrocarbon oxidations18 and Adamian and Gong have
discussed the specific case of p-xylene oxidation catalysed by
a combination of Co/Mn/Br− derivatives, illustrating the com-
plexity of the process.15 Our own experience with cobalt-
catalysed hydrocarbon oxidation included the investigation of
the partial oxidation of 4-tert-butyltoluene to 4-tert-
butylbenzaldehyde (37% yield, 79% selectivity) using Co/Br−
at 70 °C with H2O2 as the oxidant.19 In addition, we have also
investigated the catalytic reaction of a model cobaltIJIII) acetate
complex with H2O2, observing the relatively rapid decomposi-
tion of H2O2 and the formation of O2.20 This is consistent
with reports describing the ability of cobaltIJIII) species to re-
act with peroxides, however, the exact mechanism is un-
known. Spier, Neuenschwander and Hermans have noted
that depending on the type of ligand coordinated to Co (acac
vs. 2-ethylhexanoate) and the Co–OOR bond strength, perox-
ides may react through different mechanistic pathways.21
This catalytic system has been applied to the oxidation of
other functional groups,22 including the curing of paints,23
and the subtle effects of ligand variation are noted.24 As
outlined by Partenheimer,25 and recognising the relative dis-
tributions of oxidisable linkages in lignin, investigating the
potential of this catalyst system to valorise lignin is attractive
It is instructive to note that Neuenschwander and
Hermans17 have described a simplified scheme of a thermal
and catalytic free radical hydrocarbon autoxidation (reactions
1–4), which can be described analytically by eqn (E):
The radical initiation occurs via a thermal step
k2


(1)
(2)
(3)
ROOH  RH  RO  H2O  R
And a catalytic step
kcat
III
ROOH  CoII  Co  OH  RO

The radical termination is given by
kt


ROO  ROO  ROH  Q  O  O2
for several reasons:26–28 firstly, as
a
successful and
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
established industrial catalyst many developmental and engi-
neering problems such as corrosion and catalyst/solvent
recycling have been solved. Secondly, lignin break-down
products or fragments contain a broad mixture of hydroxyl-
ated, methoxylated and alkylated aromatics. As lignin un-
dergoes extraction and chemical modification during
pulping, cleavage of sub-units increases the frequency of phe-
nolic hydroxyl groups from 7–13% to over 70%, based on
aminolysis.29,30 Some of these fragments, either received
from the process, or in pure form as model compounds, have
been already investigated with Co/Mn/Br− catalyst systems.31
However, these studies, performed at high temperature and
pressure, were not sufficiently selective to warrant further
commercially targeted development. It was found that the ar-
omatic hydroxyl group inhibited the rate of oxidation of the
benzylic group, but acetylation of the hydroxyl group im-
proved the rate significantly.25 In addition, application of the
catalyst system to commercial grade organosolv lignin has
also been performed under various conditions. Use of a Co/
Mn/Zr/Br− catalyst at 180 °C and 138 bar afforded 10.9 wt%
hydroxyaromatic benzaldehydes and acids.25 Another exam-
ple describes the use of a Co/Mn/Br− catalyst to oxidise four
different types of lignin under variable oxygen pressures (0.5–
1.5 MPa) at 210 °C to yield vanillin and vanillic acid in 4–4.5
wt% based on the initial amount of lignin.32 Thus, these ex-
amples suggest that the catalyst system might not be selective
enough to convert industrial grade lignin to economically at-
tractive yields of valuable products due to the complexities
associated with C–C and C–O bond cleavage and the re-
polymerisation of cleavage products. Alternatively, pre-
treatment procedures could be employed to afford less com-
plex mixtures of relatively smaller oligomers or related com-
pounds such as syringyl derivatives.33–35 We are currently un-
aware of any use of the Co/Br− or a derivative catalyst system
to oxidise exclusively phenolic substrates. However, the use
of either cobalt or manganese acetate with excess H2O2 has
been demonstrated for the oxidation of selected phenols to
their respective benzoquinones in 30–75% yield.36 This is im-
portant as phenol reactivity is distinct to that of alkyl or
mixed alkyl/hydroxyl substituted aromatics. As reported previ-
ously, phenols are found to be strong anti-oxidants and in-
hibitors of alkylaromatic oxidations via the formation of
long-lived phenoxy radicals that can terminate chain propa-
gation by means of a number of processes, including the gen-
eration of unwanted side products, such as quinones, which
may also act as inhibitors.37–43
Fig. 1 Lignin model compounds, from left to right: monomers, α-O-4
(red link) and β-O-4 (blue link) models.
might other agents such as NHPI, be a more selective radical
promoter in lignin oxidation chemistry than the more com-
monly used bromide system.44,45 Accordingly, in the present
study we report the use of Co/Br−/H2O2 in acetic acid at 70 °C
and 1 atm for the oxidation of a series of lignin model com-
pounds with varying degrees of methoxylation including phe-
nolic monomers, and α-O-4 and β-O-4 models (Fig. 1), and of
organosolv lignin. Cobalt and Br− were introduced as cobalt
acetate tetrahydrate and KBr respectively. A preliminary as-
sessment of the applicability of the analytical solution to the
Neuenschwander and Hermans rate equation was conducted
using syringol as a simple substrate.
Experimental
Chemicals
All chemicals were used as received unless otherwise specified.
Glacial acetic acid, potassium bromide, potassium carbonate,
and sodium hydroxide (Ajax), 3,4-dimethoxybenzaldehyde
(3,4DMB-CHO), 3,4-dimethoxybenzyl alcohol (3,4DMB-OH),
3,4-dimethoxybenzyl acetate (3,4DMB-OAc), 3,4-dimethoxy
acetophenone, 2,6-dimethoxyphenol (syringol), 2,6-dimethoxy-
1,4-benzoquinone (2,6DMBZQ), 2,6-dimethoxy-1,4-hydroqui-
none (2,6DMHQ), N-bromosuccinimide, benzoic acid, phthalic
anhydride, sodium hydride (60% w/w in mineral oil), vanillin,
(Sigma Aldrich), N-hydroxyphthalimide (Alfa Aesar), aqueous
hydrogen peroxide (30% w/v), 2-phenylethyl bromide, copperIJII)
bromide, phosphorus tribromide, potassium hydroxide, tolu-
ene (Merck), cobaltIJII) acetate tetrahydrate, benzaldehyde (Asia
Pacific), phenyl benzoate (British Drug Houses) were of reagent
grade. Organosolv lignin was purchased from GreenValue En-
terprises and was produced by the Alcell process. The feedstock
was a mixture of eastern Canadian hardwoods (∼50% maple,
30% birch and 20% poplar). Chloroform-d, acetic acid-d4
(Cambridge Isotope Laboratories) were diagnostic grade. All
water used was de-ionised with a Milli-Q (Millipore) Ultra Pure
Water System. Guaiacol (Sigma Aldrich) was vacuum distilled
and stored under nitrogen before use. Phenol (Sigma Aldrich)
was recrystallised from hexanes before use. Synthesis and
The foregoing discussion raises several questions, which
need to be answered to determine the in-principle viability of
low temperature and pressure oxidations with a Co/H2O2 sys-
tem: (i) what is the efficacy of catalysis for realistic model
compounds and for lignin itself under milder conditions (70
°C with H2O2, rather than O2 under pressure, as the oxidant);
(ii) is the simplified analysis of Neuenschwander and
Hermans, notwithstanding the complexity of the chemistry
described for p-xylene oxidation by Adamian and Gong, perti-
nent under these conditions and for these substrates;15,17 (iii)
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
purification of α-O-4 and β-O-4 model compounds, and
4-bromo-2,6-dimethoxyphenol (4Br2,6DMP) was performed
following published procedures (see ESI†).
(0.06 mmol), substrate (2.7 mmol) and KBr (0.5 mmol) in acetic
acid (7.5 mL) with H2O2 (10 × 113 μL) added every 12 min for 2
h, and samples (150 μL) taken every 20 min. Oxidation of
syringol involved reacting CoIJOAc)2·4H2O (0.24 mmol), syringol
(10.8 mmol) and KBr (2 mmol) in acetic acid (20 mL), H2O2
(375 μL aliquot) was added every 12 min for either 2.5 h, or 1 h
followed by 1 h of heating. Samples (200 μL) were taken 5 min
after H2O2 addition. Oxidation of 3,3′,5,5′-tetramethoxy-4,4′-
diphenoquinone (coerulignone) and 2,6DMBZQ as substrates
were conducted similarly to syringol. Samples were extracted
into ethyl acetate for analysis by a Shimadzu GC-2010 Plus GC-
FID installed with a Rtx-5MS capillary column, 30 mm × 0.25
mm × 0.25 μm. The GC oven ramp method employed for all
substrates, except B2, was 3 min isothermally at 50 °C followed
by a ramp of 15 °C min−1 to 300 °C which temperature was
held for 5 min, giving a total method time of 24.67 min. The
GC oven ramp method employed for analysis of B2 was 3 min
isothermally at 50 °C, followed by a ramp of 10 °C min−1 to 300
°C which temperature was held for 5 min, giving a total
method time of 33 min. For reactions with coerulignone, the
product was washed with methanol (20–30 mL) and dried in
air for 5–10 minutes followed by oven drying at 80 °C for up to
4 h. Repeat experiments of the catalytic oxidation of syringol
were performed to determine the experimental error of the cat-
alytic testing procedure. The standard deviation of the data
points of the substrate and products was <3%. All other sub-
strates were reacted in the catalyst system once unless other-
wise specified. Coerulignone was identified by melting point,
mass spectroscopy and elemental analysis (ESI). The residue
from reactions involving syringol was collected by removing the
acetic acid in vacuo to afford a dark red/brown amorphous ma-
terial. This was then washed with dichloromethane (5 × 50 mL)
to remove the remaining soluble organics and dried in an oven
at 80 °C for up to 4 h leaving a dry brown powder/residue. Anal-
ysis of the organic filtrate by GC-MS identified syringol,
2,6DMBZQ, 2,6DMHQ and 4Br2,6DMP. Oxidation of
organosolv lignin was conducted similarly to syringol for 2 h,
however a time-course analysis was not performed. After 2 h
the reaction was cooled to room temperature and diethyl ether
and water (40 mL each) were added to the reaction mixture to
precipitate the unreacted lignin. The mixture was filtered and
the lignin residue washed with diethyl ether and water (3 × 20
mL each) followed by drying at 60 °C overnight. The combined
aqueous fraction was extracted with diethyl ether until the or-
ganic fractions were colourless (5 × 100 mL). The ethereal frac-
tions were combined and concentrated in vacuo followed by re-
dissolving of the products in ethyl acetate in a volumetric flask
(10 mL) and analysis by GC-MS, and quantification by GC-FID
using chlorobenzene as an external standard.
Instrumentation
Gas chromatography-mass spectrometry (GC-MS) was
performed using a Shimadzu GCMS-QP2010 and gas chroma-
tography using an FID (GC-FID) was performed using a
Shimadzu GC 2010 Plus. Both gas chromatographs were
fitted with a Restek Rtx-5MS column (30 mm × 0.25 mm, I.D.;
0.25 μm thickness) and a Shimadzu AOC-20i auto injector.
Data were collected and analysed using Shimadzu
LabSolutions GCMSsolution version 2.72 and Shimadzu
LabSolutions GCsolution version 2.41 respectively. Atmo-
spheric pressure chemical ionisation mass spectroscopy
(APCI-MS) and electrospray ionisation mass spectroscopy
(ESI-MS) data were acquired using a Bruker amaZon SL mass
spectrometer. 1H NMR (300.13 MHz) and 1H NMR (200.13
MHz) spectra were recorded on a Bruker Avance DPX300 and
a Bruker Avance DPX200 spectrometer respectively, both at
300 K. Spectra were collected and processed using Bruker
TopSpin 3.0 software and referenced internally to residual
solvent signals or tetramethylsilane (TMS, δ 0.00 ppm). The
melting points of compounds were determined in air using a
Stanford Research Systems OptiMelt Automated Melting
Point System, and a Gallenkamp melting point apparatus,
and were reported uncorrected to a known standard.
Catalytic testing
The following procedures were derived from van de Water
et al.19 Oxidation of lignin model compounds was conducted
with a mole ratio of cobalt acetate tetrahydrate, to substrate, to
potassium bromide and excess hydrogen peroxide (per aliquot)
of 1 : 45: 8 : 15–18, heated at 70 °C in a 100 mL multi-necked
round bottom flask with a magnetic stirrer, and an un-capped
reflux condenser open to the atmosphere. Products were identi-
fied by GC-MS and comparing the EI-MS spectra with those in
the (NIST05) database. Substrates and products were quantified
by GC-FID using individual co-efficient factors determined
from calibration curves of authentic samples. Chlorobenzene
was used as an internal standard. Conversions are expressed as
a percentage defined as (moles of initial substrate – moles of fi-
nal substrate) × 100. Selectivities are expressed as a percentage
defined as (moles of product/moles of converted substrate) ×
100. Yields are expressed as a percentage defined as (conver-
sion percentage × selectivity percentage)/100. Mass balances
are calculated from the calibrated substrates and products
measured by GC-FID, or gravimetrically. Oxidation of α-O-4
model compounds, guaiacol, phenol, 3,4DMB-OH, and toluene
involved reacting CoIJOAc)2·4H2O (0.08 mmol), substrate (3.6
mmol) and KBr (0.67 mmol) in acetic acid (10 mL) with ali-
quots of H2O2 (11 × 150 μL) added every 12 min for 2 h, and
samples (150 μL) taken every 20 min (samples taken every 10
min for the first hour of 3,4DMB-OH oxidation). Oxidation of
β-O-4 model compounds involved reacting CoIJOAc)2·4H2O
Results and discussion
Catalytic oxidation of α-O-4 and β-O-4 model compounds
The efficacy of the Co/Br−/H2O2 catalyst system was evaluated
for lignin model compounds first. Results for the catalytic ox-
idation of α-O-4 and β-O-4 model compounds by Co/Br−/H2O2
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Table 1 Results for the Co/Br−/H2O2 catalysed oxidation of α-O-4 and β-O-4 model compounds, 2 h of H2O2 addition
Substrate
A1
Conversion (%) Products (yield mol%, selectivity%)
17
Mass balance (mol%)
Benzyl moiety 96
Phenolic moiety 84
(12, 71)
(1.1, 6.5)
14
Benzyl moiety 98
Phenolic moiety 86
A2
(7.6, 54)
(4.1, 29)
47
Benzyl moiety 77
Phenolic moiety 59
A3
B1
(6.1, 13)
(1.2, 2.6)
(4.6, 24)
(23, 49)
19
Benzyl moiety 86
Phenolic moiety 81
Not quantified
38
Benzylic moiety 84
Phenolic moiety 85
B2
(6.3, 16)
(4.7, 12)
(17, 45)
are summarised in Table 1. The unsubstituted α-O-4 model
compound, A1, exhibited a relatively slow rate of conversion
(17% overall) over 2 h with minimal change to the mass bal-
ance of the benzyl moiety (96%) (Fig. S1†). Benzaldehyde was
identified as the major product with a small amount of phe-
nyl benzoate (1.1%) also identified. Small amounts of phenol
and bromophenol isomers (<1%) were observed by GC-MS
(Fig. S2†). Due to the small amounts detected of the latter
species and the rapid oxidation of phenols to quinones and
CO2, the mass balance of the phenyl moiety could not be de-
termined accurately. Methoxylation at the para- and
ortho-positions of the benzyl moiety of the α-O-4 linkage
gives substrate A2. Under the standard oxidation conditions,
when comparing to A1, the product distribution is signifi-
cantly altered, yielding 3,4-dimethoxybenzyl acetate (3,4DMB-
OAc) and decreased selectivity for the corresponding benzal-
dehyde (from 12% to 8%, Fig. S3†). In contrast to the phenyl-
benzoate observed for A1, no corresponding phenyl-3,4-
dimethoxybenzoate was observed for A2. Phenol and bromo-
phenol products were barely detectable, attributed to their
rapid oxidation. Similarly to A1 however, A2 exhibited a rela-
tively slow rate of conversion (14% overall) over 2 h and com-
parable mass balance values. The formation of 3,4DMB-OAc
is attributed to the stabilising effect of the methoxy groups
on the intermediate benzylic radical.46 Assuming the benzyl
moiety has been cleaved, greater stability of the benzylic
Scheme 3 The equilibrium established between benzylic radicals and
O2. The electron-donating methoxy substituents impart stability to the
benzylic radical.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
radical reduces its rate of interaction with molecular oxygen.7
This would be consistent with a relatively higher concentra-
tion of the more stable benzylic radical due to a greater reso-
nance stabilisation than that present in its corresponding
benzylperoxy radical (Scheme 3). In addition to methoxy-
lation at the 3- and 4-positions of the benzyl moiety of A2,
the methoxylation of the phenolic moiety in the
ortho-position of A3 resulted in significant improvements in
the rate and overall conversion (47% over 2 h) of the sub-
strate A3 compared with A1 and A2 (Fig. S4†). In contrast to
A2, 3,4DMB-OAc was the major product of A3 oxidation,
while the yield of the corresponding benzaldehyde dropped
significantly. In addition a greater amount of the phenolic
moiety (guaiacol, 6%) was detected. Similarly to the observa-
tions with both A1 and A2, a small amount of the brominated
phenolic moiety (bromoguaiacol) was detected by GC-MS.
The improved substrate conversion of A3 (as compared to A1
and A2) is attributed to the increased stability of the phenoxy
radical generated from guaiacol, compared to that generated
from phenol, due to the electron donating effect of the
methoxy group in the ortho position.47 The improved conver-
sion can be further rationalised from the calculated C–O
bond dissociation energies (BDE) of A1, A2 and A3 (56.43,
56.24 and 51.32 kcal/mol respectively).48 Addition of methoxy
groups at the ortho-position on the phenolic moiety of the
α-O-4 model compounds led to the largest changes in the
C–O BDE and this change was attributed to greater steric hin-
drance of the methoxy group added.48 However, despite the
increased conversion of A3 compared to A1 and A2, the over-
all mass balance for the reaction was lower, indicating an al-
ternative reaction pathway, such as polymerisation.
The unsubstituted β-O-4 model compound (B1) yielded a
conversion (19% over 2 h) similar to the unsubstituted α-O-
4 model (A1), plateauing after 80 min (Fig. S5†). Benzalde-
hyde was the only major product quantified (4.6% yield),
however, no colour change, the observation of which can
indicate the formation of phenoxy-radicals or polyphenols
was observed. Based on the intensity of the peaks in both
the GC-MS and GC-FID chromatograms, it is clear that a
significant amount of the substrate underwent bromination
(Table 1, Fig. S6†). The exact isomer formed could not be
identified conclusively, the fragmentation pattern suggesting
bromination of the phenolic moiety (Fig. S7†). By assuming
the brominated ether to have a similar FID molar signal re-
sponse as to that of the substrate, the yield of the bromi-
nated product was approximated at 10% (Fig. S8†). This es-
timate is in line with expectations based on the previous
observations of low conversion and high mass balances of
the benzyl moieties of A1 and A2 over 2 h of reaction. In
addition, hydroxyphenylethyl phenyl ether and phenacyl
phenyl ether were also observed as minor products, pre-
sumably arising from an analogous oxidation process for
the α-carbon that results in the phenyl benzoate observed
in the oxidation of A1 (<1%, Table 1, Fig. S6†). Despite the
similar rates of conversion of B1 and A1, the yield of benz-
aldehyde from the conversion of B1 was 3 times less than
that from A1. This is likely a consequence of the higher
BDE values for C–C bonds compared with C–O bonds.48
Furthermore, reaction pathways such as bromination, and
hydroxylation of the α-carbon proceeded to the extent that
these species were just observable spectrometrically. From
the pattern of substrate reactivity observed with the α-O-4
model compounds it was expected that the more elaborately
functionalised (and thus more realistic) β-O-4 model com-
pound, B2, would exhibit greater conversion and cleavage
in the Co/Br−/H2O2 system than B1. After 2 h 38% of the
substrate was converted, then the rate of conversion
appeared to plateau after a further 40 min (Fig. S9(a)†).
The
dimeric
molecule
1-(3,4-dimethoxyphenyl)-2-(2-
methoxyphenoxy)ethanone was identified as the major prod-
uct, indicating oxidation at the α-carbon and elimination of
the primary alcohol of B2. Smaller amounts of 3,4DMB-
CHO and guaiacol were also observed. Analysis of the reac-
tion mixture by GC-MS revealed a mixture of aromatic
monomer species (Fig. S10(a and b)†). Apart from the
bromoguaiacol species, many of the low intensity ion peaks
could not be reliably identified by the NIST library, how-
ever, structures consistent with the m/z ratio of the ion
peak are proposed. The relatively low intensity ion peaks of
these products suggests that they are present in the reac-
tion at very low concentrations. In addition to aromatic
monomeric species, two ion peaks with an m/z ratio consis-
tent with the dimeric species 1-(3,4-dimethoxyphenyl)-2-(2-
methoxyphenoxy)-2-propen-1-one and acetylation of one of
the alcohol groups of B2 were also observed (Fig. S10(c)†).
In comparison to B1, bromination of the substrate was not
observed, but a similar yield of the benzaldehyde product
was obtained. The higher BDE values for C–C bonds com-
pared with C–O bonds, cited earlier in the discussion of
B1, may direct selectivity to dimeric oxidation products over
monomeric products. In contrast to the α-O-4 model com-
pounds A2 and A3, the presence of 3,4DMB-OAc was not
detected. The mass balance for the reaction of B2 was
slightly lower compared to B1 (when including the esti-
mated yield of brominated B1), consistent with the behav-
iour of the α-O-4 model compounds with increased
methoxylation of the phenolic moiety. The colour of the re-
action turned dark orange during H2O2 addition, suggesting
the formation of species with an extended π-system, such
as polyphenolics. In comparison to earlier results obtained
by DiCosimo et al., oxidation of B2 using a similar catalyst
system (CoIJII) acetate in 80% acetic acid at 170 °C and 3.45
MPa, 4% O2 in N2) reported 49% substrate conversion to
afford 3,4DMB-CHO, 3,4-dimethoxybenzoic acid, and
guaiacol as the major products. Acetylation of the substrate
at either the primary alcohol or both the primary and sec-
ondary alcohol was also observed, but no products consis-
tent with the elimination of the primary alcohol of the sub-
strate were reported.49 The authors also investigated the
stability of B2 in the absence of the CoIJII) catalyst (80%
acetic acid at 170 °C and 3.45 MPa, 4% O2 in N2) observing
77% conversion of the substrate and affording a higher
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
selectivity for acetylation of the substrate relative to the
monomeric products. In this work, heating B2 in acetic
acid at 70 °C for 2 h demonstrated a negligible change in
the amount of substrate, confirming that the Co/Br−/H2O2
catalyst is primarily responsible for the conversion of B2.
Analysis of the reaction mixture by GC-MS revealed the
presence of an ion peak consistent with the acetylation of
B2 at either the primary or secondary alcohol (Fig. S11†).
monomer moieties. Due to the radical scavenging ability of
phenol and guaiacol, the effect of these molecules on the oxida-
tion of a benzylic substrate was investigated. 3,4DMB-OH was
chosen as the test substrate due to its resemblance to the ben-
zyl moiety of the α-O-4 model compounds A2 and A3. Catalytic
oxidation of 3,4DMB-OH in the Co/Br−/H2O2 system resulted in
complete conversion of the substrate after 50 min to afford
3,4DMB-CHO (93% selectivity) and 3,4DMB-OAc (7% selectiv-
ity) (Fig. S18†). Addition of either phenol or guaiacol at 5 mol%
(with respect to substrate) reduced the conversion of 3,4DMB-
OH (68% and 61% respectively), and reduced the mass balance
to ∼80%, indicative of the formation of oligomeric/polymeric
phenols (Fig. S19 and S20†). In a similar trend to that observed
for the α-O-4 model compounds A2 and A3, which feature the
dimethoxybenzyl moiety, the yield and selectivity of 3,4DMB-
OAc increased at the expense of the yield and selectivity of
3,4DMB-CHO in the presence of phenols and to a greater extent
in the presence of guaiacol. At even higher concentrations of
phenol or guaiacol (20 mol%) the conversion decreased even
more (∼38%), while the mass balance remained relatively
unchanged (Fig. S21 and S22†). The selectivity for 3,4DMB-OAc
increased further with it becoming the major product, while
the yield of 3,4DMB-CHO dropped to less than 10%. It is note-
worthy that regardless of the type or amount of phenolic addi-
tive, the yields of 3,4DMB-OAc were all nearly identical (14, 13,
18, 14 mol%). Due to the reduced formation of benzylic radi-
cals in the presence of phenols, 3,4DMB-OH is more likely to
undergo acetylation in the presence of excess acetic acid. These
results demonstrate clearly that phenols both inhibit the con-
version of benzylic substrates, including α-O-4 model com-
pounds, and demonstrate that product selectivity can be di-
rected by the type of phenol employed.
Behaviour of phenol and guaiacol
The low conversion of A1 and A2, and the decrease in the
mass balance for catalytic oxidation of A3 and B2 are attrib-
uted to the formation of phenoxy radicals and polymeric spe-
cies. A colour change to red-brown after addition of H2O2 was
observed for all α-O-4 model compounds, indicative of the
formation of phenoxy radicals.50 These species are known
radical scavengers, reducing the availability of catalytically
useful radicals (benzylic) and terminating the propagating
radical species.51,52 As phenols undergo an increasing degree
of polymerisation the antioxidant activity increases due to
the greater number of hydroxyl groups and extended
π-system, affording greater stability of the polymer.53 Cata-
lytic oxidation of either phenol or guaiacol revealed the
monomers reacted more rapidly than the α- and β-O-4 com-
pounds. Similar to the α-O-4 model compounds, increased
methoxylation of the phenolic substrate resulted in a greater
rate of conversion (Fig. S12 and S14†). GC-MS analysis
detected only small amounts of the brominated substrate as
well as 2-methoxyhydroquinone (species identified by the
NIST library), but may also be methoxy resorcinol or methoxy
catechol (Fig. S13 and S15†). Large molecular weight prod-
ucts (larger than dimers), not detectable by GC, were detected
by APCI-MS analysis. Consistent with this, interrogation of
the guaiacol reaction mixture revealed the presence of multi-
ple ions with a m/z ratio comparable to structures of guaiacol
and hydroxylated guaiacol derived oligomers including
guaiacol dimer (m/z = 247.04), trimer (m/z = 369.16), tetramer
(m/z = 491.13) and pentamer (m/z = 613.15) species (Fig.
S16†), confirming that phenolic substrates undergo oxidative
radical polymerisation in the Co/Br−/H2O2 system. Quinones
and organic acids were not detected by GC-MS for the cata-
lytic oxidation of guaiacol and phenol, suggesting over-
oxidation is not a major pathway for these substrates under
the conditions employed.
Catalytic oxidation of syringol
Results for the catalytic oxidation of syringol by Co/Br−/H2O2
are summarised in Table 2. The substrate behaved similarly
to phenol and guaiacol in that after addition of H2O2 the re-
action turned dark red indicating the formation of phenoxy
radicals. However, unlike the cases of the other phenolic
monomers several products arising from distinct processes
were detected. Over a 2.5 h period, with peroxide addition
in aliquots, 96% of the syringol was consumed (Fig. S23†).
Immediately after initiating the reaction with H2O2, a dark
purple precipitate formed which was later identified as
coerulignone by ESI-MS. Analysis by GC-FID and GC-MS re-
vealed the formation of 2,6DMBZQ and 4Br2,6DMP as mi-
nor products, and a very small amount of 2,6DMHQ (<1%)
(Fig. S24†). However the products and unreacted substrate
accounted for only 63% of the mass balance. Removal of
the solvent and extraction of the volatile organic molecules
yielded a brown residue that is suspected to be the result of
polymerisation of phenoxy radicals, and accounted for 31%
of the mass balance (based on the initial mass of the sub-
strate). Analysis of the residue by APCI-MS revealed multiple
high molecular weight species (similar to those generated
For the catalytic oxidation of A3, similar to the oxidation of
guaiacol, large molecular weight products were detected by
APCI-MS: a complex spectrum was obtained displaying ions
with a m/z ratio consistent with a guaiacol trimer (m/z = 369.16)
and protonated guaiacol pentamer (m/z = 613.27) species. Spe-
cies containing fragments consistent with both benzyl (m/z =
150) and guaiacyl (m/z = 122) moieties less two hydrogens each
were also detected (Fig. S17†). These observations demonstrate
that the progressive decrease in mass balance for the oxidation
of A3 is due to the increasing formation of oligomeric and poly-
meric species derived from the radical coupling of phenolic
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 2 Results for the Co/Br−/H2O2 catalysed oxidation of syringol, 2.5 h H2O2 addition
Conversion
(mol%)
Mass balance
(mol%)
Substrate
Products (yield mol%, selectivity%)
(Coerulignone) (2,6DMBZQ)
96
(4Br2,6DMP)
(63)a
(6, 6)
(12, 13)
(41, 43)
a
A significant amount of the substrate was converted to a non-tractable residue (31%, based on the mass of starting material) which was recov-
ered by removing the solvent.
from guaiacol) consistent with structures composed of syringyl
moieties (less two hydrogens, m/z = 152) (Fig. S25†). Methoxy
groups in both ortho positions of syringol promote the
dimerisation of the phenoxy radical, the product of which is
then oxidised to form the highly insoluble coerulignone
(Scheme S1†). Formation of the benzoquinone is possible by
reaction of the phenoxy radical with a hydroxyl radical to form
the hydroquinone, followed by oxidation by a either a bromine
radical or hydroxyl radical. The benzoquinone could also be
formed by the reaction of the phenoxy radical with O2 in the
para position, forming a p-Ph-O2˙ adduct. Dimerisation of the
p-Ph-O2˙ adduct could lead to Russel termination, forming the
benzoquinone, hydroquinone and O2.54 Formation of the
bromophenol is likely a consequence of the phenoxy radical
reacting with a bromine radical (Scheme S2†).
4Br2,6DMP. Similar behaviour of deactivation has been ob-
served by others for the oxidation of alkyl aromatics by the
formation of catalytically inactive benzyl bromides and rapid
bromination of phenol by-products.57,58 It has also been
reported that in Co/Br− catalysed systems the bromide anion
may exist within the second coordination sphere of the metal
through ion pairing.12 With less Co present in the reaction to
coordinate bromide, the co-catalyst is more likely to be se-
questered by phenoxy radicals. Decreasing initial Co concen-
trations also resulted in decreasing yields of coerulignone
(from 42 to 16%), a small overall decrease of the yield of
2,6DMBZQ (from 9 to 4%), and significant decreases in the
residue recovered from solvent removal. Based on these re-
sults the yield of coerulignone appears dependent on the con-
centration of Co across the range of 0.6 to 12 mM. This inter-
pretation is supported by a control reaction in which Co was
absent, affording <1% of coerulignone (Fig. S26†). Previous
studies concerning oxidation of 2,6-disubstituted phenols by
Co–O2 complexes demonstrated higher ratios of benzoqui-
none (BZQ) to diphenoquinone (DPQ) with increasing Co
concentration in non-polar solvents.59 However, the ratio of
BZQ to DPQ significantly decreases with increased tempera-
ture in polar solvents, with DPQ (4,4 C–C bonding) being
favoured in acidic solvents.43,60
Variation of initial co and phenolic substrate concentration
To better understand the reactivity of phenolic substrates
with the Co/Br− catalyst, variation of the initial concentra-
tions of Co catalyst and substrate were performed. Syringol
was used as the test substrate due to its broader product dis-
tribution compared to phenol and guaiacol, and its preva-
lence in organosolv and hardwood lignin extracts.33,55,56 The
conversion of syringol began to decrease significantly from
82 to 71 mol% when the initial concentration of Co was less
than 6 mM (Table 3). Additionally, increased yields of
4Br2,6DMP, from 4 to 22%, were observed. This result indi-
cates that when the Co : Br− ratio decreases, more of the bro-
mide anion is sequestered as the catalytically inactive form
In contrast to decreasing the initial Co concentration, de-
creasing the initial substrate concentration increased conver-
sion of the substrate (Table 4). The yields of coerulignone
were reduced and the yields of 4Br2,6DMPs remained rela-
tively unchanged, however the yields of 2,6DMBZQ increased
significantly (22%, 42% selectivity). With less substrate
Table 3 Results for the catalytic oxidation of syringol using different initial concentrations of Co (as CoIJOAc)2·4H2O), 1 h of H2O2 addition
[Co]0
(mM)
Conversion
(mol%)
Coerulignone
(yield, mol%)
2,6DMBZQ
(yield, mol%)
4Br2,6DMP
(yield, mol%)
Entry
Residue (%)
Mass balancec (%)
1
2
3
4
5
24
12a
6
3
0.6
85
82
82
71
65
46
42
32
21
16
6
8
9
7
4
2
4
8
11
11
22
20
N/Ab
6
69
72
67
68
66
7
a
Same initial concentration as results in Table 2. b N/A: residue was not recovered. c Based on substrate and identified products.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Table 4 Results for the catalytic oxidation of syringol using different initial concentrations of substrate, 1 h of H2O2 addition
[Syringol]0
(M)
Conversion
(mol%)
Coerulignone
(yield, mol%)
2,6DMBZQ
(yield, mol%)
4Br2,6DMP
(yield, mol%)
Residue
(%)
Mass
Rate constantd
Entry
balancec (%)
(min−1
)
1
2
3
4
1.08
0.54a
0.27
0.14
59
82
98
33
42
31
27
4
8
16
22
3
4
5
3
N/Ab
20
N/A
57
81
72
54
52
0.0120
0.0213
0.0440
0.0970
100
a
b
c
d
Same initial concentration as results in Table 2. N/A: residue was not recovered. Based on substrate and identified products. As an
approximation to facilitate general comparisons of the relative rates, pseudo-first order rate constants were obtained by fitting the initial rates
as first-order.
present fewer phenoxy radicals are produced in situ, reducing
the probability of dimerisation. With relatively higher con-
centrations of Co, O2 and H2O2, phenoxy radicals are more
likely to follow reaction pathways to form benzoquinone. To
approximately compare the relative rates of the catalytic oxi-
dation of syringol for different initial syringol concentrations,
pseudo-first order rate constants were obtained from plotting
the natural logarithm of syringol concentration vs. time for
the initial part of the reaction. Plotting the pseudo-first order
rate constants against the initial syringol concentration re-
vealed that the initial pseudo-first order rate constant of
syringol conversion is inversely proportional to the initial
syringol concentration (Fig. S27 and S28†). Again, similar be-
haviour (increased substrate conversion, BZQ yield) has been
observed for oxidation of 2,6-disubstituted phenols by Co–O2
complexes when lowering the substrate concentration but
maintaining the Co concentration (0.017 M Co and 0.66, 0.33
M substrate59 vs. 0.012 M Co and 1.08, 0.54, 0.27, 0.17 M sub-
strate, this work).
system is affected by oxidation products such as hydroqui-
nones and quinones. Time-course experiments involving the
addition of an equimolar amount of 2,6DMBZQ or 2,6DMHQ
with syringol were performed with peroxide addition for 1 h
(Fig. 2 and 3, Table 5, Fig. S29†). For the addition of equimo-
lar 2,6DMBZQ the conversion of syringol was not dramati-
cally inhibited (Fig. 2). However, at higher concentrations
(∼0.25 M) 2,6DMBZQ is partially insoluble, and due to the
similar retention time/peak overlap of 4Br2,6DMP both mole-
cules could not be accurately monitored. In comparison to
the absence of added 2,6DMBZQ (Table 5, entry 1), the con-
version of syringol and the observed pseudo-first order rate
constant are slightly lower (82%, 0.0213 min−1 vs. 79%,
0.0193 min−1, Table 5, entry 2). The yield of coerulignone is
slightly less (32% vs. 42%), suggesting that the dimerisation
of syringyl radicals may have been inhibited.
In the presence of equimolar 2,6DMHQ, the catalytic oxi-
dation of syringol was strongly inhibited (Fig. 3), while
2,6DMHQ was rapidly oxidised to 2,6DMBZQ. Unlike
2,6DMBZQ, 2,6DMHQ remained soluble at higher concentra-
tion. However, the rapid formation of 2,6DMBZQ meant
that the product and 4Br2,6DMP could not be monitored
accurately. The data suggests that until the concentration of
2,6DMHQ drops below 10% of its initial value the conver-
sion of syringol is strongly inhibited. The pseudo-first order
rate constant for the conversion of 2,6DMHQ was 0.104
The effects of quinones and hydroquinones on the reaction
of syringol
It is well known that phenols and quinones are inhibitors of
the oxidation of benzylic carbons on aromatic molecules,
however it is unclear if phenol oxidation in the Co/Br−/H2O2
Fig. 2 Time-course profile of the catalytic oxidation of syringol with
Co/Br−/H2O2 showing only syringol consumption with and without the
presence of equimolar 2,6DMBZQ.
Fig. 3 Time-course profile of the catalytic oxidation of equimolar
syringol and 2,6DMHQ with Co/Br−/H2O2.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 5 Results for the catalytic oxidation of syringol with Co/Br−/H2O2 in the presence of equimolar 2,6DMBZQ or 2,6DMHQ
Syringol conversion
(mol%)
Coerulignone
(yield, mol%)
Rate constanta
(syringol, min−1
Rate constanta
Entry
SubstrateIJs)
)
(2,6DMHQ, min−1
)
1
2
3
4
5
Syringol (1 h H2O2)
82
79
36
96
80
42
32
6
41
35
0.0213
0.0193
0.0110
0.0254
0.0183
N/A
N/A
0.104
N/A
0.0944
Syringol, 2,6DMBZQ (1 h H2O2)
Syringol, 2,6DMHQ (1 h H2O2)
Syringol (2.5 h H2O2)
Syringol, 2,6DMHQ (2.5 h H2O2)
a
As an approximation to facilitate general comparisons of the relative rates, pseudo-first order rate constants were obtained by fitting the ini-
tial rates as first-order.
min−1 (Table 5, entry 3), approximately five times greater
than that for syringol (0.0213 min−1). Extending the H2O2
Modelling the catalytic oxidation of syringol
addition for 2.5
h
syringol conversion continued until
As discussed above, to approximately compare the differences
observed for the conversion of syringol under different condi-
tions, pseudo-first order rate constants were obtained from
the linear plots of the natural logarithm of the substrate con-
centration vs. time at the start of the reaction. While this
treatment provides a convenient way to compare the behav-
iour of the substrate during the initial stage of the reaction (t
< 65 min), it does not satisfactorily describe the behaviour of
the substrate over the entirety of the reaction. The model
based on pseudo-first order rate constants breaks down at
lower concentrations of syringol, e.g. ∼20% of the initial con-
centration (Fig. S31†), as its assumptions for its applicability
(high concentration of substrate) become invalid. Recently,
Neuenschwander and Hermans17 were able to successfully de-
scribe the thermal and catalytic initiation of the free radical
hydrocarbon autoxidation of α-pinene. To test whether the
rate equation of Neuenschwander and Hermans17 (eqn (E))
could be employed as a useful first approximation to model
the entire catalytic reaction of syringol, an analytical solution
for the concentration of substrate vs. time was developed con-
sistent with the scheme of Neuenschwander and Hermans,
i.e. there is a thermal and catalytic pathway, a steady-state in
radical concentration, and that the concentrations of the
peroxide (H2O2 in this case) and Co are constant. The time-
course of syringol oxidation was fitted using the analytical so-
lution of the rate equation and the results are illustrated in
Fig. 4. The derivation of the analytical solution of the rate
equation is described in the ESI,† pp. 7–10. Although individ-
ual rate constants cannot be extracted, the good fit of the the-
oretical expression to the data indicate that the simplified
mechanistic description of Neuenschwander and Hermans is
a useful first approximation to describe the chemistry over
the whole of the reaction and can be used as a basis for fur-
ther in-depth microkinetic analysis of this complex system. In
addition to Co, we also know that Br− is involved in the cata-
lytic step, however, it is added to speed up the reaction and
may not be involved in the rate-determining step as described
by Neuenschwander and Hermans, and discussed in detail by
Adamian and Gong.15,17 Despite some Br− being consumed in
the bromination of the substrate, the concentration of Br− re-
mains in excess throughout the reaction (>60% of the initial
Br− concentration based on the 4Br2,6DMP concentration).
plateauing at 80% after 2 h (Fig. S30†). Similar to the equi-
molar addition of 2,6DMBZQ (Table 5, entry 2), the pseudo-
first order rate constant and coerulignone yield were slightly
lower compared to the reaction of only syringol for 2.5 h
(0.0183 min−1, 35% vs. 0.0254 min−1, 41%, Table 5, entries
4 and 5). The relatively rapid reaction of 2,6DMHQ may ex-
plain its presence in only very low concentrations (<1%) ob-
served in samples from the catalytic oxidation of syringol.
Furthermore, the inhibition of syringol conversion by
2,6DMHQ may explain why relatively higher initial concen-
trations of syringol result in lower amounts of syringol con-
version. With
a higher syringol to cobalt ratio, more
2,6DMHQ is generated, which inhibits syringol conversion.
The mechanism of inhibition of syringol conversion by
2,6DMHQ would be consistent with a disproportionation
pathway: in the absence of radical scavengers, neutral
semiquinone radicals decay through bimolecular self-
reactions that afford the respective quinone and hydroqui-
none.43 The generation of both syringyl and 2,6-
dimethoxysemiquinone radicals could result in the transfer
of a hydrogen radical from a 2,6-dimethoxysemiquinone rad-
ical to a syringyl radical (Scheme 4). However, this assumes
the rate of H-atom transfer between the phenoxy radical
species is significantly greater than the rate of other reac-
tion pathways such as syringyl radical dimerisation to form
coerulignone.
Scheme
4
Proposed scheme of H˙ transfer from
a
2,6-
dimethoxysemiquinone radical to a syringyl radical.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
loss of the substrate, and the initial mass of 2,6DMBZQ was
recovered. This suggests that the molecule is stable under
the catalytic conditions and is not converted to organic
acids or COx. Samples were also analysed by GC-MS, reveal-
ing a species corresponding to bromination of 2,6DMBZQ
with a yield of <5% based on the co-efficient factor of
2,6DMBZQ (Fig. S32†). The reaction of coerulignone could
not be monitored by GC-FID due to its extremely poor solu-
bility, however, samples were still taken to observe any po-
tential decomposition products. Samples analysed by GC-MS
revealed tribromomethane and 2,2-dibromoacetic acid
methyl ester (Fig. S33a†). Comparison of the chromato-
grams for the oxidation of coerulignone and a blank reac-
tion indicate the brominated molecules are formed from
the decomposition of coerulignone (Fig. S33b†). Conversion
of coerulignone was determined gravimetrically by recover-
ing the remaining material. A brown residue similar in ap-
pearance to the residue observed for the catalytic oxidation
of syringol was also recovered. Like the syringol oxidation
residue, the coerulignone residue was only soluble in acetic
acid and was sparingly soluble in water. 1HNMR spectro-
scopy of the coerulignone residue in acetic acid-d4 revealed
a complex spectrum with methoxy and phenolic signals sim-
ilar to that for the syringol oxidation residue (Fig. S34–
S36†). By extrapolation, the mass of coerulignone unac-
counted for (23 and 57% of the mass balance, Table 6) cor-
responds to 5.7 and 5.6%, respectively, of the mass balance
of syringol (for 1 h of H2O2 addition) in the earlier experi-
ments. This suggests that excess H2O2 may directly affect
conversion of coerulignone. Overall these results suggest the
loss of coerulignone in the catalytic system could contribute
to the poor mass balances recorded for the catalytic oxida-
tion of syringol, including any contribution from CO/CO2,
which was not explicitly quantified.
Fig. 4 Plot of the experimental data points for the catalytic oxidation
of syringol over 2.5 h of peroxide addition, compared with the plot
arising from the Neuenschwander and Hermans model, eqn (E),
(including the standard error).
Over-oxidation of syringol-derived products
As reported above, a significant portion of syringol is recov-
ered as a residue that is suspected to be the result of
syringyl radical polymerisation. However, the literature pro-
vides scant examples of cobalt-catalysed formation of poly-
phenols.61,62 It is unknown whether the Co/Br−/H2O2 catalyst
system could yield polymeric products, and whether any
polymeric products would be resistant to over-oxidation. It
is known that under Co/Br− catalyst conditions both the
acetic acid solvent and the aromatic substrate are subject to
over-oxidation.58,63–66 The main by-products of solvent oxida-
tion include CO2 and CO, followed by smaller amounts of
methyl acetate, acetoxyacetic acid, formic acid and methane.
It has been demonstrated that solvent decomposition can
be initiated by thermal decomposition of a CoIJIII) acetate
species and formation of an acetoxy radical via homolysis of
a CoIJIII)–O bond of an acetate ligand.65 Products of substrate
decomposition are similar to those from solvent decomposi-
tion, predominantly COx species and smaller amounts of or-
ganic acids, esters and brominated molecules.66 Suspected
mechanisms of substrate decomposition include decarboxyl-
ation of aromatic acids to produce COx species, or genera-
tion of phenols which are oxidised to quinones and
CO2.6,42,43 Coerulignone and 2,6DMBZQ were both tested
for over-oxidation and the results are summarised in
Table 6. Monitoring of 2,6DMBZQ by GC-FID showed no
Substitution of bromide with N-hydroxyphthalimide for
catalytic oxidation of α-O-4 model compound (A3) and
organosolv lignin
The highly corrosive conditions of the catalyst system, due
to the presence of HBr, have encouraged the investigation
of alternative radical promoter co-catalysts such as NHPI,
which is converted to the phthalimide-N-oxyl (PINO) radi-
cal.45 Co/NHPI systems have been demonstrated to oxidise
toluene and other hydrocarbons effectively, but to the best
of our knowledge have not been investigated for oxidation
of lignin model compounds. Reaction of A3 with the Co/
Table 6 Results for the catalytic oxidation of 2,6DMBZQ and coerulignone in the Co/Br−/H2O2 system, 1 h of H2O2 addition
Run
Initial substrate, ga Recovered substrate, g, (%) Consumed substrate, g, (%) Recovered residue, g, (%) Mass balance (%)
2,6DMBZQ
Coerulignone1 0.412
Coerulignone2 0.165
0.455
0.455 (100)
0.27, (66)
0, (0)
0 (0)
0.142, (34)
0.165, (100)
0 (0)
0.047, (11)
0.071, (43)
100
77
43
a
The loadings of 2,6DMBZQ and coerulignone correspond to 25, and 25 and 10% of the loading for the catalytic oxidation of syringol (1.67 g,
10.8 mmol), 1 mole of coerulignone is equivalent to 2 moles of syringol.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 7 Results for the catalytic oxidation of organosolv lignin (1.67 g) with Co and either Br− or NHPI as co-catalysts, 2 h of H2O2 addition. Monomeric
aromatic and quinone products contained in the organic extract of the reaction mixture were analysed only
Run
Vanillin, mg (wt%)
Syringaldehyde, mg (wt%)
2,6DMBZQ, mg (wt%)
Total, mg (wt%)
Br− 1
2.8 (0.17)
3.6 (0.21)
2.8 (0.17)
2.5 (0.15)
7.2 (0.43)
9.1 (0.54)
3.3 (0.2)
18 (1.1)
22 (1.3)
30 (1.8)
28 (1.7)
28 (1.7)
35 (2.1)
36 (2.2)
33 (2.0)
Br− 2
NHPI 1
NHPI 2
2.0 (0.12)
The wt% product yields are calculated relative to the amount of initial lignin.
NHPI/H2O2 system, replacing Br− with an equimolar amount
of NHPI, resulted in improvements in substrate conversion
and product yields (Fig. S37, Table S1†). The conversion of
the substrate increased from 47 to 68%, while the yield of
3,4DMB-CHO increased from 1.2 to 8.2%, corresponding to
a 5-fold improvement in selectivity. The yield and selectivity
of guaiacol significantly improved (from 6 and 13%, to 29
and 43%), while the yield of 3,4DMB-OAc also improved
slightly. The mass balance of the benzyl moiety of A3
exhibited a similar profile to A3 reacted in the presence of
Co/Br−, while the mass balance of the phenolic moiety im-
proved relative to the improved yield of guaiacol. This im-
proved guaiacol selectivity suggests that substitution of Br−
with NHPI reduces the rate of polyguaiacol formation,
which reduces inhibition of substrate conversion and the
loss of oxidation products. Benzylic substrates undergo less
conversion when using NHPI as a co-catalyst than when
using Br− (e.g. toluene, Fig. S38 and S39†). This is reflected
by the less thermodynamically favoured hydrogen atom ab-
straction by PINO radicals compared to bromine radicals.67
In addition, benzylic radicals have a shorter lifetime relative
to phenoxy radicals.6 However, the greater stability of the
PINO radical appears advantageous in the presence of more
reactive molecules such as phenolics (relative to benzyls),
compared to highly active bromine radicals. Rate constants
reported for the reaction between dibromide radicals and
methoxylated phenols are ∼108 M−1 s−1, which may be con-
sidered close to diffusion control limited.68 This would sug-
gest that the higher reactivity of bromine radicals results in
lower selectivity, causing an increase in guaiacol radical
concentration and subsequent polyguaiacol formation.
vanillin was relatively unchanged. This is inconsistent with
the improved aldehyde yield from the reaction of A3 with
NHPI as a co-catalyst. The wt% yield of 2,6DMBZQ increased
by a factor of 1.3–1.5, indicating less syringol was lost to poly-
merisation and over-oxidation. This is consistent with the in-
creased guaiacol yield from the reaction of A3. No coerulign-
one was observed suggesting a relatively low concentration of
free syringol generated from the lignin polymer, inferred from
the syringol concentration variation experiments. In addition,
a significant amount of the NHPI co-catalyst was converted to
other species, including phthalic anhydride (∼20 wt%),
suggesting deactivation of the catalyst system (Fig. S41†).
These preliminary results demonstrate that NHPI could po-
tentially be useful for improving the overall product yield
from catalytic lignin oxidation, but further studies are neces-
sary to understand and optimise the system. In comparison to
the two literature examples (see ESI† for details, pp. 36–37),
the yields of aromatic aldehydes, calculated as the wt% of the
initial lignin, are ∼5 times less.25,32 The other examples use
more active catalyst compositions, such as Co/Mn/Br− and Co/
Mn/Zr/Br−, which are ∼4 and 9 times more active towards al-
kyl aromatic substrates than Co/Br− respectively.6 In addition,
the other literature examples employ either a higher reaction
temperature or pressure which has been shown to improve
the yield of aromatic aldehydes. However, the complexity of
the catalyst system and the variation of conditions between
the different literature reports makes comparison of the per-
formance of the different systems difficult.
Conclusions
Following these encouraging results, preliminary experi-
ments using NHPI as a co-catalyst for oxidation of organosolv
lignin were conducted. NHPI was compared against Br− based
on the yield of aromatic and quinone products extracted in
the organic phase. For the lignin oxidation experiments, the
molar ratio of the cobalt catalyst was doubled to reduce the
probability of forming brominated phenols and aromatics,
evidenced from variation of the cobalt concentration for
syringol oxidation. Oxidation of organosolv lignin with Co/
Br−/H2O2 and analysis of the diethyl ether extract revealed
three major products identified by GC-MS as vanillin,
2,6DMBZQ and syringaldehyde, and quantified by GC-FID
(Table 7, Fig. S40†). Substitution of Br− for an equivalent mo-
lar amount of NHPI resulted in the decrease of the wt% yield
of syringaldehyde by a factor of 2–4 while the wt% yield of
The catalytic oxidation of lignin model compounds by Co/
Br−/H2O2 in acetic acid at 70 °C has been investigated, dem-
onstrating the efficacy of the system to react with a variety of
substrates, the successful application of a simplified mecha-
nistic scheme, and the ability of NHPI to act as a more selec-
tive radical promoter for the oxidation of lignin model com-
pounds. For the oxidation of model compounds, the degree
of methoxylation of the substrate was observed to have a sig-
nificant effect on its reactivity and the product distribution
across all types of model compounds tested. For the α-O-4
model compounds the selectivity could be varied by methoxy-
lation of the benzyl moiety while substrate conversion could
be improved by methoxylation of the phenolic moiety. The
improved substrate conversion is attributed to the greater sta-
bility of the guaiacyl radical (vs. phenoxy radical) and lower
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
C–O BDE. Similar behaviour was observed for the β-O-4
model compounds, with the addition of methoxy and alcohol
functional groups increasing the complexity of the product
distribution. The catalytic oxidation of phenolic monomers
such as phenol and guaiacol resulted in the formation of
polyphenols/guaiacols. These species were found to inhibit
substrate conversion and reduce the mass balance of α-O-4
model compounds by incorporation of oxidation products
into polymeric species. In comparison to phenol and
guaiacol, catalytic oxidation of syringol yielded dipheno- and
benzoquinones, the selectivity for the formation of which was
dependent on the ratio of Co to substrate. However, the
diphenoquinone was susceptible to over-oxidation. Analysis
of the time-course data of syringol oxidation using the analyt-
ical solution to the rate equation of Neuenschwander and
Hermans17 demonstrated that the oxidation could usefully be
modelled to a first approximation by this simplified mecha-
nistic scheme. The successful application of this mechanistic
scheme will be explored for other aromatic substrates, such
as 4-tert-butyltoluene, and other lignin model compounds in
future investigations. The substitution of Br− for NHPI as the
radical promoting co-catalyst reduced the deleterious effects
of phenols for the conversion of a model α-O-4 compound,
improving the yield and selectivity of 3,4DMB-CHO and
guaiacol products. Preliminary application of NHPI to an ac-
tual lignin source afforded a modest improvement in the
yield of benzoquinone product, but, not unexpectedly, the to-
tal yield of aromatic aldehydes was lower. These results dem-
onstrate that NHPI can be a more selective radical promoter
for pure lignin model compounds than Br−, while further in-
vestigation is necessary to improve the oxidation of lignin
with Co/NHPI/H2O2.
edges the receipt of an Australian Postgraduate Award. This
work was supported by the Australian Research Council.
Notes and references
1 J. Zakzeski, P. C. Bruijnincx, A. L. Jongerius and B. M.
Weckhuysen, Chem. Rev., 2010, 110, 3552–3599.
2 R. Gosselink, E. De Jong, B. Guran and A. Abächerli, Ind.
Crops Prod., 2004, 20, 121–129.
3 W. Zhu, G. Westman and H. Theliander, J. Wood Chem.
Technol., 2014, 34, 77–97.
4 R. Rinaldi, R. Jastrzebski, M. T. Clough, J. Ralph, M.
Kennema, P. C. Bruijnincx and B. M. Weckhuysen, Angew.
Chem., Int. Ed., 2016, 55, 8164–8215.
5 R. Landau, Uncaging Animal Spirits: Essays on Engineering,
Entrepreneurship, and Economics, MIT Press, Cambridge,
1994.
6 W. Partenheimer, Catal. Today, 1995, 23, 69–158.
7 R. R. A. Tomás, J. O. C. Bordado and J. O. F. Gomes, Chem.
Rev., 2013, 113, 7421–7469.
8 W. Partenheimer, J. Mol. Catal. A: Chem., 2003, 206,
105–119.
9 W. Partenheimer and V. V. Grushin, Adv. Synth. Catal.,
2001, 343, 102–111.
10 W. Partenheimer, Catal. Today, 2003, 81, 117–135.
11 J. T. Tinge, in Liquid Phase Aerobic Oxidation Catalysis:
Industrial Applications and Academic Perspectives, ed. S. S.
Stahl and P. L. Alsters, Wiley-VCH, Weinheim, 2016, pp. 33–
39.
12 W. Partenheimer, J. Mol. Catal. A: Chem., 2001, 174, 29–33.
13 X.-D. Jiao and J. H. Espenson, Inorg. Chem., 2000, 39,
1549–1554.
Overall this study demonstrates the complexity of Co/Br−
catalysed oxidation of lignin substrates by revealing the effects
of discrete structural changes to the substrate on the conversion
and resulting product distribution. While the catalyst system is
versatile enough to react with a variety of substrates, poor prod-
uct selectivity from real lignin sources, and inhibition by
substrate-derived phenols remain as critical issues. Despite the
improved product yield when using NHPI with a model com-
pound, the inconsistency when using an actual lignin source
encourages further investigation of NHPI as the co-catalyst.
Nonetheless, the low temperature and pressure system
shows much promise and further optimisation, including ad-
ditives like Mn or Zr mirroring the commercial system, might
lead to acceptable conversion at much higher selectivities.
14 P. D. Metelski and J. H. Espenson, J. Phys. Chem. A,
2001, 105, 5881–5884.
15 V. A. Adamian and W. H. Gong, in Liquid Phase Aerobic
Oxidation Catalysis: Industrial Applications and Academic
Perspectives, ed. S. S. Stahl and P. L. Alsters, Wiley-VCH,
Weinheim, 2016, pp. 41–66.
16 J. K. Beattie, T. W. Hambley, J. A. Klepetko, A. F. Masters
and P. Turner, Polyhedron, 1998, 17, 1343–1354.
17 U. Neuenschwander and I. Hermans, J. Catal., 2012, 287,
1–4.
18 I. Hermans, in Liquid Phase Aerobic Oxidation Catalysis:
Industrial Applications and Academic Perspectives, ed. S. S.
Stahl and P. L. Alsters, Wiley-VCH, Weinheim, 2016, pp. 3–
14.
19 L. G. A. van de Water, A. Kaza, J. K. Beattie, A. F. Masters
and T. Maschemyer, Chem. – Eur. J., 2007, 13, 8037–8044.
20 E. B. Clatworthy, X. Li, A. F. Masters and T. Maschmeyer,
Chem. Commun., 2016, 52, 14412–14415.
21 E. Spier, U. Neuenschwander and I. Hermans, Angew. Chem.,
Int. Ed., 2013, 52, 1581–1585.
Conflicts of interest
The authors declare no conflict of interest.
22 S. Mahmood, T. Li, B. H. Xu, Y. F. Guo and S. J. Zhang,
Asian J. Org. Chem., 2017, 6, 768–774.
Acknowledgements
The authors wish to thank Dr Kaitlin Beare for help with the
synthesis of lignin model compound B2. E. B. C. acknowl-
23 L. H. Oakley, F. Casadio, P. K. R. Shull and P. L. J.
Broadbelt, Angew. Chem., Int. Ed., 2018, 57, 7413–7417.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
24 E. Spier and I. Hermans, ChemPhysChem, 2013, 14,
3384–3388.
25 W. Partenheimer, Adv. Synth. Catal., 2009, 351, 456–466.
26 M. Erickson, S. Larsson and G. E. Miksche, Acta Chem.
Scand., 1973, 27, 1–9.
27 H. Nimz, Angew. Chem., Int. Ed. Engl., 1974, 13, 313–321.
28 J. D. Gargulak, S. E. Lebo and T. J. McNally, Kirk-Othmer
Encyclopedia of Chemical Technology, 2000, pp. 1–26.
29 D. R. Robert, M. Bardet, G. Gellerstedt and E. L. Lindfors,
J. Wood Chem. Technol., 1984, 4, 239–263.
30 B. Biannic and J. J. Bozell, Org. Lett., 2013, 15,
2730–2733.
31 W. Partenheimer, Adv. Synth. Catal., 2004, 346, 1495–1500.
32 A. R. Gonçalves and U. Schuchardt, Appl. Biochem.
Biotechnol., 1999, 77, 127–132.
47 H. M. Ali, A. Abo-Shady, H. A. S. Eldeen, H. A. Soror, W. G.
Shousha, O. A. Abdel-Barry and A. M. Saleh, Chem. Cent. J.,
2013, 7, 1–9.
48 R. Parthasarathi, R. A. Romero, A. Redondo and S.
Gnanakaran, J. Phys. Chem. Lett., 2011, 2, 2660–2666.
49 R. DiCosimo and H. C. Szabo, J. Org. Chem., 1988, 53,
1673–1679.
50 P. C. Chrostowski, A. M. Dietrich and I. H. Suffet, Water
Res., 1983, 17, 1627–1633.
51 T. Mimura, K. Yazaki, K. Sawaki, T. Ozawa and M.
Kawaguchi, Biomed. Res., 2005, 26, 139–145.
52 K. Ingold, Chem. Rev., 1961, 61, 563–589.
53 E. Anouar, C. Calliste, P. Kosinova, F. Di Meo, J. Duroux, Y.
Champavier, K. Marakchi and P. Trouillas, J. Phys. Chem. A,
2009, 113, 13881–13891.
33 V. Roberts, V. Stein, T. Reiner, A. Lemonidou, X. Li and J. A.
Lercher, Chem. – Eur. J., 2011, 17, 5939–5948.
34 Q. Song, F. Wang, J. Cai, Y. Wang, J. Zhang, W. Yu and J. Xu,
Energy Environ. Sci., 2013, 6, 994–1007.
35 K. Barta, G. R. Warner, E. S. Beach and P. T. Anastas, Green
Chem., 2014, 16, 191–196.
54 G. A. Russell, J. Am. Chem. Soc., 1957, 79, 3871–3877.
55 C. Amen-Chen, H. Pakdel and C. Roy, Biomass Bioenergy,
1997, 13, 25–37.
56 J. N. Murwanashyaka, H. Pakdel and C. Roy, Sep. Purif.
Technol., 2001, 24, 155–165.
57 W. Partenheimer, Adv. Synth. Catal., 2004, 346, 297–306.
58 P. Roffia, P. Calini, L. Motta and S. Tonti, Ind. Eng. Chem.
Prod. Res. Dev., 1984, 23, 629–634.
36 R. Mostaghim and Y. Ahmadibeni, Acta Chim. Slov.,
2003, 50, 569–572.
37 R. Moore and W. Waters, J. Chem. Soc., 1952, 2432.
38 J. Dunn and W. A. Waters, J. Chem. Soc., 1953, 2993–2998.
39 J. Igarashi, J. Lusztyk and K. Ingold, J. Am. Chem. Soc.,
1992, 114, 7719–7726.
40 J. Igarashi, R. Jensen, J. Lusztyk, S. Korcek and K. Ingold,
J. Am. Chem. Soc., 1992, 114, 7727–7736.
41 A. Pokutsa, A. Zaborovskii, R. Sheparovich, V. Kopylets and
D. Maksim-Lutsik, Kinet. Catal., 2003, 44, 121–128.
42 W. Partenheimer, Adv. Synth. Catal., 2005, 347, 580–590.
43 E. T. Denisov and I. Khudyakov, Chem. Rev., 1987, 87,
1313–1357.
44 Y. Ishii, K. Nakayama, M. Takeno, S. Sakaguchi, T. Iwahama
and Y. Nishiyama, J. Org. Chem., 1995, 60, 3934–3935.
45 Y. Ishii, S. Sakaguchi and T. Iwahama, Adv. Synth. Catal.,
2001, 343, 393–427.
59 R. S. Drago, J. Gaul, A. Zombeck and D. K. Straub, J. Am.
Chem. Soc., 1980, 102, 1033–1038.
60 V. M. Kothari and J. J. Tazuma, J. Catal., 1976, 41, 180–189.
61 R. M. Strom, US Pat., US4438284A, The Dow Chemical
Company, 1984.
62 A. Puzari and J. B. Baruah, React. Funct. Polym., 2000, 46,
101–107.
63 P. Roffia, P. Calini and S. Tonti, Ind. Eng. Chem. Res.,
1988, 27, 765–770.
64 S. Tang and B. Liang, Ind. Eng. Chem. Res., 2007, 46, 7826–7829.
65 C. E. Sumner Jr, J. Little, A. S. Howard and W. C. Liang,
Inorg. Chem., 2010, 49, 4606–4610.
66 W. Partenheimer, Appl. Catal., A, 2011, 409, 48–54.
67 N. Koshino, Y. Cai and J. H. Espenson, J. Phys. Chem. A,
2003, 107, 4262–4267.
46 R. A. Jackson and M. Sharifi, J. Chem. Soc., Perkin Trans. 2,
1996, 775–778.
68 S. V. Jovanovic, M. Tosic and M. G. Simic, J. Phys. Chem.,
1991, 95, 10824–10827.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018