Degradation of 2,5-dihydroxy-1,4-benzoquinone by hydrogen peroxide under moderately alkaline conditions resembling pulp bleaching: A combined kinetic and computational study

Article abstract of DOI:10.1021/jo401486d

2,5-Dihydroxy-1,4-benzoquinone (DHBQ) is one of the key chromophores occurring in all types of aged cellulosics. This study investigates the mechanism of H₂O₂ degradation of DHBQ under conditions relevant to pulp bleaching (3.0% H₂O₂, NaOH, pH 10), to obtain insights useful for improved pulp processing. DHBQ is degraded quantitatively into malonic acid with an activation energy (E_a) of 16.1 kcal/mol and activation entropy (Δ^aS) of ～28 cal/mol·K. Higher concentrations of sodium cations increase the reaction rate. Theoretical computations indicate the formation of an intermediate I_O having an O-O bridge between C-2 and C-5 of the 1,4-cyclohexadione structure. I_O undergoes O-O homolysis to form a biradical Bt, which is fragmented into malonate anions. The calculated E_a (17.8 kcal/mol) agrees well with the experimental one. Coordination of Na⁺ to I_O and Bt decreases their energies and enhances the O-O homolysis rate, which is consistent with the acceleration by sodium cation and the negative Δ^aS. The homolysis of I_O is much favored over that of the neutral counterpart, with the unpaired electrons of Bt being stabilized by the geminal anionic oxygen. This difference in the stability of the intermediates translates into significant variations in the reaction rate and the product distribution between pH 10 and neutral/acidic conditions.

Full text of DOI:10.1021/jo401486d

Article
pubs.acs.org/joc
Degradation of 2,5-Dihydroxy-1,4-benzoquinone by Hydrogen
Peroxide under Moderately Alkaline Conditions Resembling Pulp
Bleaching: A Combined Kinetic and Computational Study
Takashi Hosoya and Thomas Rosenau*
Department of Chemistry, University of Natural Resources and Life Sciences, Muthgasse 18, A-1190 Vienna, Austria
*
S Supporting Information
ABSTRACT: 2,5-Dihydroxy-1,4-benzoquinone (DHBQ) is one of the key chromo-
phores occurring in all types of aged cellulosics. This study investigates the mechanism of
H O degradation of DHBQ under conditions relevant to pulp bleaching (3.0% H O ,
2
2
2
2
NaOH, pH 10), to obtain insights useful for improved pulp processing. DHBQ is
degraded quantitatively into malonic acid with an activation energy (E ) of 16.1 kcal/mol
a
⧧
and activation entropy (Δ S°) of ∼28 cal/mol·K. Higher concentrations of sodium
cations increase the reaction rate. Theoretical computations indicate the formation of an intermediate I having an O−O bridge
O
between C-2 and C-5 of the 1,4-cyclohexadione structure. IO undergoes O−O homolysis to form a biradical Bt, which is
+
fragmented into malonate anions. The calculated E (17.8 kcal/mol) agrees well with the experimental one. Coordination of Na
a
to I and Bt decreases their energies and enhances the O−O homolysis rate, which is consistent with the acceleration by sodium
O
⧧
cation and the negative Δ S°. The homolysis of I is much favored over that of the neutral counterpart, with the unpaired
electrons of Bt being stabilized by the geminal anionic oxygen. This difference in the stability of the intermediates translates into
significant variations in the reaction rate and the product distribution between pH 10 and neutral/acidic conditions.
O
INTRODUCTION
matrices are looking for mild and compatible ways for DHBQ
destruction. The chemistry of DHBQ has thus become a point
of great interest from several viewpoints in cellulose science.
The structure of DHBQ and its resonance-stabilized dianion
CB is given in Scheme 1.
■
2
,5-Dihydroxy-1,4-benzoquinone (DHBQ) has been described
as one of the key compounds contributing to discoloration
effects of cellulosic materials as a consequence of aging. This
overall process is commonly denoted as ″yellowing″ of the
pulp. This comprises natural aging as well as all kinds of stresses
by thermal treatment, irradiation, oxidation, or impact by
extreme pH conditions. DHBQ is a nearly ubiquitous cellulosic
chromophore as it is both a prime survivor of bleaching
treatments and a regeneration product from low-molecular
Scheme 1. Structures of 2,5-Dihydroxy-1,4-benzoquinone
(DHBQ) and Its Resonance-Stabilized Dianion CB
1
weight fragments of cellulose aging. Its exceptional resonance
stabilization in alkaline medium is responsible for its resistance
toward bleaching chemicals. Most of them, e.g., ozone,
hydrogen peroxide, and chlorine dioxide, destroy chromophoric
structures mainly by acting on (localized) double bonds.
However, the charges, and thus also the double bonds, in the
DHBQ-derived dianion are highly delocalized and are thus
much more slowly attacked than those in most other quinoid
chromophores that lack such resonance stabilization. This
makes DHBQ a much better ″survivor″ of industrial pulp
bleaching sequences than other aromatic or quinoid chromo-
Several previous accounts have addressed the chemistry of
2
DHBQ, which has recently been reviewed. Its main reaction
3
a−h
paths are electrophilic substitution,
nucleophilic substitu-
3i
4
tion, and condensation with amines. However, these works
have employed DHBQ as a starting material in organic
synthesis or as a target in analytical endeavors. With the
1
phores in pulps. Condensation of carbonyl-containing and
″
revival″ of the compound, which is connected to its pivotal
carboxyl containing C2−C4 fragments, such as those formed
from carbohydrates upon aging, condense under formation of
DHBQ as a main product, since it represents a global minimum
on the C−H−O thermodynamic formation energy map.
The pulp and paper industries are interested in better, i.e.,
faster and cheaper, removal of the compound from cellulosic
pulps; cellulose scientists search for good methods of DHBQ
detection and quantification of its effects on cellulose
properties; and conservationists working with historic cellulosic
role in cellulose yellowing, the degradation chemistry of
DHBQ, i.e., its ″discoloration″, became a matter of much
interest.
We have previously attempted to contribute to the
mechanism of H O degradation of DHBQ under nonbuffered
2
2
Received: July 11, 2013
Published: October 25, 2013
©
2013 American Chemical Society
11194
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
5
conditions. Under these conditions, the reaction solution is
done to check whether the pseudo-first-order approximation
would hold also true under pH 10 degradation conditions.
From Figure 1 it is evident that there is a nice, linear
slightly acidic (pH ≈ 5.5) from the beginning due to the acidity
2
of DHBQ. As shown in Scheme 2, DHBQ is eventually
Scheme 2. Summary of Detailed Reaction Steps of DHBQ
5
Degradation by H O under Non-buffered Conditions
2
2
Figure 1. Relationship between ln[DHBQ] and reaction time t at
2
3
78.15 K (■), 288.15 K (▼), 298.15 K (●), 313.15 K (▲), and
23.15 K ( ).
degraded into malonic acid, acetic acid, and carbon dioxide.
According to the detailed mechanism, DHBQ forms a primary
intermediate IOH through nucleophilic attack of H O . Next
◆
2
2
relationship between ln[DHBQ] and t, indicating correctness
of the pseudo-first-order approximation. The rate constants k at
each reaction temperature follow from the slope of the lines.
reaction step, homolysis of the O−O bond of IOH, is rate-
determining. β-Fragmentation of the resulting biradical
produces oxaloacetic acid OA and ketene; see Scheme 2. In
this process, the electronic structure remains the most stable
singlet (S0) state. OA is then further attacked by H O and
−4
The determined k values ranged from 6.14 × 10 to 2.61 ×
−2 −1
1
0
s , as summarized in Table 1.
2
2
Figure 2 shows the corresponding Arrhenius plot of the
releases carbon dioxide and water to form malonic acid, while
ketene is converted into acetic acid by addition of water. This
mechanism has been well supported by kinetic analysis, product
identification, and computational considerations. However, the
conditions in that study were chosen in a way to be as simple as
possible and were thus avoiding any additives or adjustment of
pH. Although interesting from the scientific point of view, the
setting was unfortunately not relevant to cellulose processing,
as pulp bleaching with H O , the so-called bleaching ″P-stage″,
logarithm of the pseudo-first-order rate constants k (ln k)
versus the reciprocal of the reaction temperature (K) to have a
linear relationship that allows determination of the activation
parameters. The activation energy (E ) of the degradation was
a
⧧
1
6.1 kcal/mol, and the activation enthalpy (Δ H°) was similar
⧧
at 15.5 kcal/mol; see Table 1. The activation entropies (Δ S°)
were considerably negative (−27.7 to −22.6 cal/mol·K). It is
interesting to compare these activation data to those of the
reaction under the nonbuffered conditions: under the pH 10
2
2
is commonly performed at alkaline pH around 10 and above.
Our attempt to verify that the degradation chemistry of DHBQ
under such alkaline conditions is similar to that in neutral
media failed rather quickly and led us into a different chemistry
that, by parallel computational treatment, provided some
interesting aspects of DHBQ degradation in particular and
bleaching chemistry in general. This is the topic of the present
combined experimental and computational study, with which
we would like to address the DHBQ degradation under
industrially relevant peroxide bleaching conditions, and we
hope to provide a solid basis for considerations in cellulose
bleaching chemistry and optimization of pulp bleaching
sequences.
⧧
conditions, the activation enthalpies Δ H° are lower and the
⧧
activation entropies Δ S° are more negative; see Table 1. In
other words, the alkaline degradation is enthalpically favored,
⧧
but disfavored from the viewpoint of entropy. The Δ G° values
(
21.8−24.1 kcal/mol) are thus moderately smaller for the
reaction under the pH 10 conditions, meaning that the reaction
proceeds somewhat faster than in neutral medium.
1
Figure 3A presents the H NMR spectrum of the reaction
mixture (sample A, reaction time 240 s) in DMSO-d obtained
6
at 298.15 K; see the Experimental Section for details of sample
preparation. The spectrum shows singlets corresponding to
nonreacted DHBQ, its diester with boronic acid, and malonic
acid. Here and in the following, all assignments were confirmed
by spiking with authentic samples. The boronic ester originated
from the borate buffer and was formed under the acidic
RESULTS AND DISCUSSION
■
Kinetic Analysis and Product Characterization. In our
13
conditions used to stop the reaction (see also the C NMR
5
previous study, we reported that H O degradation of DHBQ
11
2
2
and B NMR spectra of sample A in Figure S1 of the
under nonbuffered conditions was following pseudo-first-order
kinetics if an excess amount of H O (257 molar equiv of H O
Supporting Information for detailed information on the borate
6
2
2
2
2
ester). Since the signal of malonic acid is overlapping with the
relative to DHBQ) was used. In the pseudo-first-order
approximation, the concentration of DHBQ [DHBQ] at
reaction time t is presented by following equation:
1
water signal in DMSO-d , it was thus quantified from the H
6
NMR spectrum of sample B in D O, where the singlet of
2
malonate dianion is neatly separated. Note that sample B
contains minute traces of acetic acid (see Experimental Section
for the difference between samples A and B). The yields of
malonic acid were found to be nearly quantitative (95.4−
100%), and thus 2 equiv of malonic acid were formed per
−
kt
[
DHBQ] = [DHBQ] e
(1)
0
where [DHBQ] is the initial concentration of DHBQ and k is
0
the kinetic rate constant. A plot of ln[DHBQ] against t was
1
1195
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Table 1. Experimental Pseudo-first-order Rate Constants (k) and Activation Parameters in the Degradation of DHBQ by H O
2
2
under pH 10 Conditions
a
without sodium sulfate addition
b
d
activation parameters
with sodium sulfate addition
⧧
⧧
⧧
k (s−1
)
−3
temp (K)
k (s−1)
Δ H° (kcal/mol)
Δ S° (cal/mol·K)
Δ G° (kcal/mol)
4
278.15
288.15
298.15
313.15
323.15
6.14 × 10−
8.20 × 10
2.22 × 10−
15.5
−22.6
−24.2
−24.1
−27.7 (−16.7)
−23.6 (−14.7)
21.8
1.60 × 10
3.70 × 10
−
4
3
3
−3
−3
15.5
22.5
4.10 × 10 (3.54 × 10−3
)
e
15.5
22.7
1.44 × 10− (4.66× 10−4
)
c
c
15.5 (19.9)
15.5 (19.8)
24.1 (24.9)
23.1 (24.5)
2.13 × 10
2.44× 10
−2
2.61 × 10− (3.01× 10−3)
2
−2
a
b
The reaction solution without sodium sulfate addition also contains some amount of sodium cation since a NaOH-borate buffer is employed. The
activation energies (E ) were determined to be 16.1 kcal/mol under pH 10 conditions and 20.4 kcal/mol under nonbuffered conditions. Values for
the nonbuffered conditions from previous study for comparison. Under these conditions, no metal cation is involved in the reaction solution. The
c
a
5
d
concentration of sodium ion is enhanced by the addition of sodium sulfate. The amount of sodium sulfate added was 150 equiv (mole) against
e
DHBQ. Rate constant (k) in the presence of 300 equiv (mole) of sodium chloride.
Figure 2. Arrhenius plot of the degradation reaction in the
temperature range from 278.15 to 323.15 K.
equivalent of degraded DHBQ under the employed pH 10
7
reaction conditions. Malonic acid was also one of the major
5
products under nonbuffered conditions. However, under those
nonbuffered conditions malonic acid, acetic acid and carbon
dioxide were formed in equimolar amounts, and only one
molecule of malonic acid was produced from one molecule of
DHBQ (see Scheme 2). Under pH 10 conditions, by contrast,
DHBQ produced almost exclusively malonic acid, and only tiny
1
Figure 3. H NMR spectra of the degradation reaction at 298.15 K
and 240 s, sample A in DMSO-d (A) and sample B in D O (B); see
6
2
the Experimental Section section for the preparation methods for
samples A and B. In spectrum A, nonreacted DHBQ is found mainly
as the diester of boronic acid (from the buffer system, see also Figure
S1 in Supporting Information). In the spectrum B, the signal of
DHBQ is not detected due to H-D exchange of the protons at C-3 and
C-6. The small signal at 4.2 ppm in spectrum B probably correspond
to glycolic acid or 2-hydroxymalonic acid, which were detected in
GC−MS analysis (see Figure S2 in Supporting Information).
1
amounts of acetic acid were seen as byproduct: the H NMR
spectrum in Figure 3B shows an almost negligible signal of the
acetate anion at 1.9 ppm. These results suggested that the
degradation mechanism under the moderately alkaline (pH 10)
conditions is considerably different from that under the
nonbuffered conditions.
The degradation experiment was also carried out in the
presence of 150 molar equiv of sodium sulfate (relative to
DHBQ). In this case, the concentration of sodium cation is
additionally enhanced by the addition of sodium sulfate
compared to the original buffer solution. The salt addition
significantly increased the reaction rate, as shown in Table 1
chloride, indicating the effect of the cation, but not the anion,
to be decisive. This change in the reaction rate by the addition
of the salts cannot be explained by changes in the ionic strength
of the reaction solution as this would be rather different for the
+
(
see also Figure S3 in the Supporting Information for the
two salts used. It was therefore likely that Na affects some
relationship between ln[DHBQ] and t). A similar effect was
observed when we substituted sodium sulfate for 300 molar
equiv of sodium chloride, maintaining the same amount of the
step(s) of the degradation mechanism directly and this way+
enhances the degradation rate. We will discuss the effect of Na
in combination with computational results in the next section.
A detailed kinetic analysis of the effect of different alkali, earth
alkali, and aluminum salts is the topic of an upcoming report.
+
−3 −1
Na relative to DHBQ (Table 1): k was 4.10 × 10
s
in the
−
3 −1
presence of sodium sulfate and 3.54 × 10 s with sodium
1
1196
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Unfortunately, we were unable to determine the activation
parameters in the presence of sodium sulfate since the
Arrhenius plot was not linear for the most common rate law
cases (see Figure S4 in Supporting Information). The reason
for this nonlinear relationship is unknown at the moment.
Degradation Mechanism; Quantum Chemical In-
sights. Quantum chemical calculations were carried out to
get some deeper insight into the reaction mechanism. Malonic
acid was a major product in the nonbuffered system, formed via
The detailed mechanism including the transition states is
shown in Figure 4, and schematic description of the mechanism
is shown in Scheme 3. The first step is a proton transfer from
H O to CB followed by nucleophilic attack of the resulting
2
2
−
HOO to the C-2 carbonyl group, to form intermediate Ih.
This reaction proceeds via transition state TSh . In the second
step, Ih is transformed to I through an intramolecular reaction
via transition state TSh . The mechanism of the formation of I
1
O
2
O
from Ih is similar to that of the first step (CB to Ih): a proton
transfer from the second hydrogen peroxide oxygen (O-7′) to
C-6 and a subsequent nucleophilic attack of the anionic O-7′ at
C-5 (see Figure 4). The rate-determining step in the formation
5
intermediate IOH (see Scheme 2). It appeared reasonable to
assume the intermediacy of I under the pH 10 conditions (see
O
Figure 4 and Scheme 3 for the structure), which is the dianionic
⧧
of I is the first H O attack, with the Δ G° value (32.5 kcal/
O
2
2
mol) of TSh being higher than that of TSh (23.4 kcal/mol).
1
2
I was calculated to be 23.5 and 13.1 kcal/mol less stable than
O
10
CB based on Gibbs energy and potential energy, respectively.
The next step, also by analogy to the reaction in neutral
medium, should be the O-7−O-7′ homolysis of I . Hence, the
O
singlet biradical Bs , as the most plausible product, was
O
11
considered at the DFT(UB3LYP) level (see Figure 4).
However, the obtained potential energy of Bs (26.7 kcal/mol)
O
was so high that it far exceeded the experimental E value (see
a
Figure 4 and Table 1). This led us to assume coordination of
+
−
Na to I , with the electrostatic interaction between O and
O
+
Na stabilizing I . As expected, the ΔE value of I was
O
O
substantially decreased to 2.7 kcal/mol (coordination of two
+
Na ions; see I in Figure 4). Because this coordination is
Na
entropically unfavored, the Gibbs energy of INa becomes
comparable to that of I , suggesting that IO and INa are in
O
equilibrium in water solution. We also calculated the
corresponding singlet biradical Bs formed by O−O homolysis,
+
Bs thus being the Na -coordinated Bs . The potential energy
O
of Bs (15.8 kcal/mol) is much lower than that of Bs . From
O
these lines of results, we focused our further investigation on
+
the Na -coordinated species, such as I and Bs, rather than on
Na
the anionic, noncoordinated ones. It should be also noted that
+
our assumption of the Na coordination is well supported by
the above experimental facts: the addition of the sodium salts
significantly increasing the reaction rate and the reaction under
the moderately alkaline conditions being entropically unfavored
(
see previous section). The enhanced sodium cation concen-
Figure 4. Proposed detailed mechanism for the degradation of DHBQ
under moderately alkaline (pH 10) conditions, computed at the MP2/
BS-II//DFT(B3LYP)/BS-I level of theory. Notes: (a) Bond lengths
are given in Å. (b) Energy values are presented in kcal/mol. (c)
tration will shift the equilibrium between I and I further to
O
Na
I , resulting in an enhanced homolysis rate of the O−O bond.
Na
Since the geometry of Bs was optimized as a stable species at
the DFT(UB3LYP) level, there must be a transition state that
connects INa and Bs. The geometry optimization of such
transition state, however, was neither successful with the
restricted nor the unrestricted DFT(B3LYP) methods. Instead,
the potential energy curve of the path from INa to Bs was
investigated in regard to the dependence of the O-7−O-7′
bond length, this way approximating the structure of the
transition state. As shown in Figure 5, the potential energy
increases up to a O-7−O-7′ distance of 1.889 Å and then
decreases again from a distance of 2.263 Å, indicating that the
transition state is located between these O-7−O-7′distances.
The failure to localize any transition state structure is due to the
fact that exactly in this distance range the scission of the O−O
bond occurs, which requires treatment by different computa-
tional methods for nonradical and radical species, i.e., RB3LYP
and UB3LYP, making it impossible to run the geometry
optimization constantly. We eventually decided to employ the
geometry optimized at a O-7−O-7′ distance of 2.263 Å (TS1 in
Figures 4 and 5) for the determination of the activation energy,
Energies of TS1, Bs , Bs, Bt, CI2, and CI3 are calculated according to
O
the DFT(UB3LYP)/BS-II method relative to the DFT(RB3LYP)/BS-
II-calculated energy of INa because the MP2 method did not provide
reliable energy data due to an instability of the Hartree−Fock wave
functions for these species. (d) The structures of TS1, CI2, and CI3
are approximated ones, and the potential energies do not include the
zero-point energy; see Figures 5 and 8 and the pertinent discussion in
the text.
conjugate base of IOH. Under the pH 10 conditions, DHBQ is
completely converted into its dianionic conjugate base CB
(
4
Scheme 1) since the pK and pK of DHBQ are 2.95 and
al a2
8
.87, respectively. Although H O is also a very weak acid,
2 2
−
formation of deprotonated H O such as HOO should not be
predominant at pH 10 from its high pK of 11.7. We thus
2
2
9
a
investigated a reaction in which CB is attacked by H O to
2
2
form I , using computations at the MP2//DFT(B3LYP) level
O
of theory.
1
1197
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Scheme 3. Schematic Description of the Detailed Mechanism for the Degradation of DHBQ under Moderately Alkaline (pH
1
0) Conditions
have similar orbital energies with strong σ-like interaction. This
interaction destabilizes one electron and stabilizes two
electrons, leading to overall stabilization of the biradical: ESO
=
2ΔE1 − ΔE > 0, see Scheme 4A. Since the biradical
O
2O
molecule is symmetric, the same interaction exists also between
the unpaired electron of O-7′ and the lone pair electrons of O-
5
. The atomic charges and the spin densities of Bt, calculated at
the DFT(UB3LYP)/BS-II level, showed that O-2, O-7, O-7′,
and O-5 have almost the same atomic charges around −0.68
and spin densities around +0.55, indicating that the electrons
are evenly delocalized between those atoms (Figure 7A). This
result is consistent with the orbital interaction in Scheme 4A.
The electronic structure of Bs was essentially the same as that
of Bt; see Figure S5 in the Supporting Information for the
important orbitals of Bs.
Figure 5. Potential energy curve regarding the O7−O7′ length from
to Bs, computed at the DFT(UB3LYP)/BS-I level of theory. We
obtained the potential energy curve by carrying out the geometry
optimization with the O7−O7′ length set to the respective values.
I
Na
Then in the triplet state biradical Bt undergoes β-
fragmentation to form an acyl radical anion A1 (see Figure
4
). Interestingly, this reaction needs to pass through a conical
intersection CI2, which connects the most stable triplet state
T1) of Bt to the next stable triplet state (T2), see Figure 6A
as its structure is considered to be the closest to that of the
actual transition state. As frequency calculation is meaningless
(
⧧
and B. To clarify the reason for the necessity of this conical
intersection, we studied the electronic structure of Bt in the T2
state (T2-Bt). The important orbitals of T2-Bt show that one
of the SOMOs (HOMO-1) contains the p-orbital of O-7
orthogonal to the SOMO of T1-Bt. In addition, the p-orbital of
O-7 interacts with a p-orbital of C-2 in a π-like antibonding way
with the bonding counterpart of this interaction in the HOMO-
8. As shown in Scheme 5, T1-Bt cannot undergo the β-
fragmentation because the p-orbital of C-2 cannot interact with
the unpaired electron. In T2-Bt, on the other hand, the
unpaired electron of O-7 does interact with the p-orbital of C-2,
rendering β-fragmentation possible. This notion is also
supported by the potential energy curves in Figure 8, where
the T2 curve leads to A1 while that of T1 monotonously
increases. The energy of the T2 state of Bt is only 3.2 kcal/mol
higher than that of the T1 state, calculated at the DFT-
(UB3LYP)/BS-II level. Also, the T2 state becomes more stable
than the T1 state when the C-1−C-2 bond is elongated by 0.05
Å from the equilibrium structure of T1-Bt (see CI2 in the
in this case, the Δ G° value was not determined; the obtained
E value for TS1 was 17.8 kcal/mol relative to CB.
a
We also optimized the structure of a triplet biradical Bt, the
structure and energy of which were quite similar to those of Bs;
see Figure 4 and Scheme 3. These results suggest that Bs and
Bt are able to interconvert through a conical intersection CI1
after O-7−O-7′ homolysis.
The detailed electronic structures of the biradical Bs and Bt
were worth investigating in detail for a better understanding of
the degradation mechanism. In this paragraph, we thus wish to
discuss the electronic structure of Bt on the basis of its
molecular orbitals calculated at the ROHF/BS-II level of
theory. As shown in Figure 6A, the p-orbitals of O-7 and O-2 as
well as those of O-7′ and O-5 interact with each other by σ-like
antibonding and evenly contribute to the SOMOs (singly
occupied molecular orbitals) of Bt; see Scheme 4A. The
HOMO-5 and HOMO-6 of Bt involve the bonding counter-
parts of this interaction. The p-orbital of O-7 (the unpaired
electron) and the p-orbital of O-2 (the lone-pair electrons)
1
1198
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Figure 6. Important molecular orbitals of Bt in the T1 and T2 states and of the protonated Bt for comparison, calculated according to the ROHF/
BS-I method.
Scheme 4. Difference in Orbital Interactions between Bt (A) and Protonated Bt (B)
Figure 7. NBO atomic charges and Mulliken spin densities of important atoms of Bt and protonated Bt in the T1 state, calculated at the
DFT(UB3LYP)/BS-II level of theory, where α spin is taken to be positive. The geometry of protonated Bt was obtained by geometry optimization,
starting from that of Bt.
1
1199
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Scheme 5. Difference in Orbital Interactions between the T1
and T2 States of Bt
Scheme 6. Suggested Mechanism of the Formation of
Malonic Acid from Radical A2
based on the Gibbs energy and −14.3 kcal/mol based on the
potential energy. The rate-determining step in the overall
process is the O-7−O-7′ homolysis of I to form Bs, with TS1
Na
having the considerably high potential energy of 17.8 kcal/mol.
When the entropy calculated for Bs was adapted to TS1, the
⧧
Δ G° value for TS1 was estimated to be 36.8 kcal/mol, which
was much higher than that in the I formation (32.5 kcal/mol,
O
see above). The E value for TS1 is in good agreement with the
a
⧧
experimental E and Δ H° values in Table 1, indicating that the
a
proposed mechanism in Figure 4 was likely to be correct.
The formation of malonic acid could be theoretically
explained also by ionic reactions as shown in Scheme 7.
Figure 8. Potential energy curves of the S0 (
◆
), T1(■), and T2 (▲)
states of the biradicals in regard to the C1−C2 distance, calculated
according to the DFT(UB3LYP)/BS-I method. The potential energy
curve was obtained by geometry optimization with the C-1−C-2
lengths fixed at the specified values.
Scheme 7. Hypothetic Ionic Degradation of the Intermediate
Ih
potential energy curve in Figure 8). Although determination of
the exact E value for this T1-to-T2 transition is not trivial, we
a
estimated it to be 16.9 kcal/mol relative to CB (Figure 4) at the
DFT(B3LYP)/BS-II level by assuming the geometry of CI2 in
Figure 8 as a reference for the real conical intersection
structure. Bs in the S0 state can also change its spin state to the
T2 state through another conical intersection CI3 (see Figure
According to this hypothetic ionic mechanism, one of the keto
groups of the intermediate Ih is first hydrolyzed to form IhH,
followed by ionic β-fragmentation of IhH into dicarboxylic acid
DC and water. The diketo group of DC reacts analogously to
afford two molecules of malonic acid. To check whether this
ionic mechanism is competitive, we simulated the reaction from
IhH to DC. IhH was 34.3 kcal/mol less stable than the reactant
CB based on Gibbs energy. We then optimized the transition
8
). We estimated the E value for CI3 to be 16.1 kcal/mol,
a
assuming the geometry of CI3 for the real conical intersection.
After the formation of A1, another type of β-fragmentation
occurs to yield two molecules of a malonyl radical anion A2 via
TS2 with a small activation energy of 14.4 kcal/mol (see Figure
4
and Scheme 3). A2 readily produces malonic acid through
12
further attack of H O , according to the literature. The most
2
2
plausible mechanism for the formation of malonic acid from A2
involves several reaction steps, as shown in Scheme 6: H-
abstraction by A2 from H O to form the corresponding
state TSh connecting IhH to DC; see Figure S7 in Supporting
3
Information for detailed geometry change in this reaction. The
⧧
Δ G° and E values of TSh were calculated to be 51.7 and 41.5
2
2
a
3
⧧
aldehyde and an OOH radical, disproportionation of OOH
radical to gaseous O and H O , and further oxidation of the
kcal/mol, respectively. Since these Δ G° and E values for TSh
were much larger than those of TS1 (see above), it was evident
a
3
2
2
2
aldehyde into malonic acid. Though we did not investigate
these reactions in detail, the above mechanism is supported by
an experimental observation of gaseous oxygen formed during
the degradation (see Figure S6 in Supporting Information).
The overall computed reaction pathway in Figure 4 and
Scheme 3 accounts for the overall conversion of DHBQ into
malonic acid in a plausible way agreeing with the experimental
fact that malonic acid is formed from DHBQ in quantitative
yield (see the previous section). The reaction energy of the
whole degradation is considerably negative: −19.8 kcal/mol
that the ionic reaction via TSh was much less favorable than
that via TS1. From these lines of computational results, we
ruled out the hypothetical ionic mechanism.
3
Degradation of DHBQ under Moderately Alkaline
(Buffered, pH 10) and Nonbuffered, Neutral Conditions:
A Comparison. The mechanism for the degradation of
DHBQ under pH 10 conditions is considerably different from
5
that under the nonbuffered conditions. In the upcoming
section, we would like to address those differences and
elaborate the reasons for them.
1
1200
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
One important difference is the stability of the biradicals
Scheme 8. Mechanism of the β-Fragmentation from Bt to A2
formed by the O−O bridge homolysis of the intermediates I
via A1
O
(
alkaline) and I (neutral, nonbuffered). While Bs and Bt are
OH
stable species as discussed in the previous section (Figure 4 and
Scheme 3), the conjugate acid of Bs and Bt, formed by the O−
O bond homolysis of I in the nonbuffered system (Scheme
OH
5
1
), is not a stable species, as reported in our previous study. In
other words, deprotonation of the two hydroxyl groups of IOH
stabilizes the biracial and renders it stable as intermediate
species. In fact, the E value calculated for the degradation
a
under the pH 10 conditions (17.8 kcal/mol, Figure 4) is much
lower than that calculated for the degradation under the
nonbuffered conditions (23.3 kcal/mol) reported in our
5
previous study, and so are the experimental E values (16.1
a
kcal/mol vs 20.4 kcal/mol, see Table 1). It was thus worth
clarifying what changes in the electronic structure of Bt are
caused by the protonation/deprotonation.
As described in the previous section, Bt is stabilized by the
orbital interaction shown in Figure 6A and Scheme 4A. For
comparison, Figure 6C shows the important molecular orbitals
of the protonated Bt, also calculated at the ROHF/BS-I level of
theory. Contrary to the SOMOs of Bt, the SOMOs of the
protonated Bt are almost localized at p-orbitals of O-7 and O-
CONCLUSIONS
We have investigated the degradation of DHBQ by H O in
alkaline medium (pH 10, buffer). Kinetic and product analysis
showed DHBQ to be degraded into malonic acid in
■
2
2
quantitative yield with an E value of 16.1 kcal/mol. The
a
mechanism of the degradation under the pH 10 conditions was
thus significantly different from that under the nonbuffered
conditions. This clearly disqualifies the conventional assump-
tion in pulp bleaching with H O (P-stage) that a change in pH
influences only the reaction rate for chromophore removal, but
not the actual mechanism.
The clarified mechanism under alkaline conditions of about
pH 10 is as follows: (1) the reaction of DHBQ with H O
results in the formation of an addition product, intermediate IO.
DHBQ is reacting in the form of its dianion CB, so that also IO
7
′; compare the orbitals in Figures 6A to those in 6C for that
particular feature. The spin density of the protonated Bt is
much higher at O-7 and O-7′ (+0.695) than at O-2 and O-5
2
2
(
(
+0.025), while the negative charge is higher at O-2 and O-5
−0.794) than at O-7 and O-7′ (−0.454). These results
indicate that the delocalization of the spin and charge between
2
2
O-7, O-7′, O-2, and O-5, which considerably stabilized Bt
(
Figure 6B), scarcely exist in the protonated Bt. The reason for
this change is displayed in Scheme 4B: the orbital energies of
the O-2 and the O-5 lone pairs are significantly lowered by
protonation, and the energy difference between the O-7 and O-
is a dianion. (2) I suffers homolytic O−O bond cleavage to
O
form biradical Bs in the S0 state and Bt in the T1 state. This
+
step is much facilitated by addition of Na . (3) Bs and Bt
7
′ single electrons and the O-2 and O-5 lone-pair electrons
change the electronic structure to the T2 state through the
conical intersections and then undergo β-fragmentation to form
the acyl radical anion A1. (4) A1 is further degraded into the
malonyl radical anion A2, which finally is neatly converted to
malonic acid by the attack of excess H O .
becomes significantly larger. The stabilizing interaction between
these orbitals is thus substantially weakened: ESOH ≪ E . The
SO
changes in the orbital interaction caused by the protonation of
Bt will lead to a much higher potential energy of the protonated
Bt compared to that of the original Bt, and this fact is very
likely to be a major reason why Bt can exist as a stable species,
but the conjugate acid (the protonated Bt) cannot.
2
2
+
The beneficial effect of Na , which is significantly stabilizing
IO and in this way increasing the homolysis rate, is supported
both by the experimental data and the theoretical calculation.
The addition of salts, such as sodium sulfate that is a byproduct
in many pulp production units anyway, during the P-stage
would be an interesting conclusion to be tested in real-world
bleaching of cellulosic pulps. The kinetic analyses and the
computation indicated DHBQ to be degraded more easily
under the pH 10 conditions than under the neutral conditions.
In a P-stage, alkaline media are used anyway in order to
minimize damage to the carbohydrate that significantly
increases with pH decreasing. The use of an alkaline pH
would now get additional support as it is needed not only from
the viewpoint of carbohydrate preservation but also with regard
to a faster degradation of DHBQ-based chromophores.
Another important difference in the degradation reaction
under the pH 10 and nonbuffered conditions is the product
distribution: DHBQ produces malonic acid in quantitative yield
in alkaline medium, while a maximum of 50% yield was
5
obtained under neutral conditions. This difference in the
products is well explained from the difference in the spine state
from which the β-fragmentation of the biradicals starts. In the
T2 state of Bt (pH 10 conditions), the two unpaired electrons
have the same spin so that they cannot form a covalent bond.
Thus, β-fragmentation does not proceed further to form
nonradical species after the formation of A1, and two molecules
of a malonyl radical anion are formed (see a schematic
description in Scheme 8). Under nonbuffered conditions, on
the other hand, such a triplet state is only involved in a minor
Another conclusion is that a very high pH (14 and above) is
negative from the viewpoint of DHBQ degradation. The first
step of the degradation sequence, attack of neutral H O at the
5
pathway, as previously reported, while the major degradation
2
2
path starts from the singlet (S0) state in which the unpaired
electrons can readily form a covalent bond after the β-
fragmentation due to their different spin. Hence, ketene and
oxaloacetic acid OA are produced as nonradical products. Since
only OA is a precursor of malonic acid, its yield comes down to
dianionic form of DHBQ, would be strongly impeded if H O
2 2
is deprotonated, due to repulsion between the two negatively
charged reaction partners. Thus, a very high pH is not
beneficial with regard to DHBQ discoloration; a pH between
10 and 12 is most appropriate in this regard and in addition
saves alkali, which is one of the main cost factors in P-stages.
50% under nonbuffered conditions (cf. Scheme 2).
1
1201
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
The studied mechanisms of DHBQ degradation by hydrogen
peroxide suggest that both reaction rate and reaction
mechanism (and thus product distribution) are controlled by
the pH since the underlying (de)protonation processes change
the stability and the reactivity of the biradical formed by the
O−O bond homolysis of the respective intermediates. We hope
that the insights will be helpful with regard to improving the
efficiency of pulp bleaching processes.
ASSOCIATED CONTENT
Supporting Information
■
*
S
Complete ref 13. C NMR and 11B NMR spectra of the
13
degradation reaction at 298.15 K and 240 s, sample A in
DMSO-d . Total-ion chromatogram in GC−MS analysis (after
6
trimethylsilylation) of degradation reaction at 298.15 K and 240
s, sample A. Relationships between ln[DHBQ] and t in the
degradation of DHBQ under the presence of sodium sulfate
and sodium chloride. Arrhenius plot in the degradation of
DHBQ under the presence of sodium sulfate. Important
orbitals of Bs. Picture of the reaction solution under the pH 10
conditions. Geometry changes and energy profile in the ionic
degradation of DHBQ. Cartesian coordinates of the optimized
geometries. This material is available free of charge via the
Internet at http://pubs.acs.org.
EXPERIMENTAL AND COMPUTATIONAL DETAILS
■
Experimental Section. All chemicals were purchased from
commercial providers. They were of highest purity available (p.a.
grade) and were used without further purification. Proton nuclear
1
magnetic resonance ( H NMR) spectra were recorded at 400.13 MHz
proton resonance frequency.
The degradation reaction was started by adding a 30% aqueous
solution of hydrogen peroxide (2.2 mL, 19.6 mM of hydrogen
peroxide) to 20 mL of an pH 10 buffer solution (NaOH-borax) of
DHBQ (10.7 mg, 0.0763 mM) in a 50 mL round-bottom flask. The
buffer solution of DHBQ contained 4.0 mg of 1,3,5-tricarboxylbenzene
as internal standard. The solution of DHBQ was preheated at
temperatures between 278.15 and 323.15 K before addition of the
hydrogen peroxide solution. After the reaction started, sampling of the
reaction mixture was done by adding 2.0 mL of the reaction solution
separately into 10.0 mL of a 0.005 M HCl solution and 10.0 mL of
deionized water, to obtain two types of the sample: sample A is
prepared from the 0.005 M HCl solution, and sample B from the pure
water. These samples were then immediately cooled to 0 °C in an ice
AUTHOR INFORMATION
Corresponding Author
E-mail: thomas.resenau@boku.ac.at.
Notes
■
*
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We performed quantum chemical calculations with the
workstation in the Sakaki group, Fukui institute for
fundamental chemistry at Kyoto University, Japan and are
thankful for the access. The financial support of the Austrian
Christian Doppler Research Society (CDG) through the CD-
lab “Advanced cellulose chemistry and analytics” and of the
Austrian Research Promotion Agency (FFG, project 829443) is
gratefully acknowledged.
bath and frozen at 193.15 K. After freeze-drying of the samples at this
1
temperature, sample A was analyzed by H NMR in DMSO-d to
6
quantify the amount of nonreacted DHBQ. Sample B was monitored
1
with H NMR in D O. Since sample B is not neutralized with HCl
2
during its preparation, all acids produced from DHBQ exist as their
corresponding sodium salts, which are not removed in the freeze-
drying process.
REFERENCES
The degradation of DHBQ was also carried out in the presence of
■
1
.63 g (11.4 mM) of sodium sulfate. In this case, sodium sulfate was
(1) (a) Rosenau, T.; Potthast, A.; Milacher, W.; Hofinger, A.; Kosma,
P. Polymer 2004, 45, 6437. (b) Rosenau, T.; Potthast, A.; Milacher, W.;
Adorjan, I.; Hofinger, A.; Kosma, P. Cellulose 2005, 12, 197.
(c) Rosenau, T.; Potthast, A.; Kosma, P.; Suess, U.; Nimmerfroh, N.
Holzforschung 2007, 61, 656. (d) Krainz, K.; Potthast, A.; Suess, U.;
Dietz, T.; Nimmerfroh, N.; Rosenau, T. Holzforschung 2009, 63, 647.
(e) Rosenau, T.; Potthast, A.; Krainz, K.; Yoneda, Y.; Dietz, T.;
Shields, Z. P.; French, A. D. Cellulose 2011, 18, 1623.
added into the buffer solution of DHBQ in advance, and the
degradation was started in the same manner as described above.
Sampling was carried out according to the procedure described above
for sample A. The freeze-dried mixture was extracted with dimethyl
1
ether, and the soluble part was subject to H NMR analysis in DMSO-
d6 after removal of the solvent in vacuo.
Computations. We employed the GAUSSIAN 09 program
13
packages for the calculations. The geometry optimization in water
(2) Hosoya, T.; French, A. D.; Rosenau, T. Mini-Rev. Org. Chem.
2013, 10, 309.
1
4
was carried out by the DFT method with the B3LYP functional and
the PCM method. The 6-31G(d) basis sets were employed for H, C,
O, Na where a diffuse function was added to each of C and O and a p-
polarization function was added to H. We name these basis sets as BS-
I. For single point calculations, we employed MP2 or DFT(UB3LYP)
method, the latter was used only for the calculations of I, Bs, and Bt.
The DFT(UB3LYP) functional is adequate to evaluate the property of
(3) (a) Brassard, P.; L’Ecuyer, P. Can. J. Chem. 1958, 36, 1346.
(b) Nicolaides, D. N.; Gautam, D. R.; Litinas, K. E.; Papamehael, T. J.
Chem. Soc., Perkin Trans. 1 2002, 1455. (c) Lang, M.; Muhlbauer, A.;
Jagers, E.; Steglich, W. Eur. J. Org. Chem. 2008, 20, 3544. (d) Shaabani,
A.; Ghadari, R.; Sarvary, A.; Rezayan, A. J. Org. Chem. 2009, 74, 4372.
(e) Shaabani, A.; Ghadari, R.; Ghasemi, S.; Pedarpour, M.; Rezayan,
A.; Sarvary, A.; Ng, S. W. J. Comb. 2009, 11, 956. (f) Misiolek, A.;
Ichimura, A. S.; Gentner, R. A.; Huang, R. H.; McCaffrey, V. P.;
Jackson, J. E. Inorg. Chem. 2009, 48, 9005. (g) Jimenez-Alonso, S.;
Estevez-Braun, A.; Ravelo, A. G.; Zarate, R.; Lopez, M. Tetrahedron
2007, 63, 3066. (h) Jimenez-Alonso, S.; Perez-Lomas, A. L.; Estevez-
Braun, A.; Martinez, F. M.; Orellana, H. C.; Ravelo, A. G.; Gamarro,
F.; Castanys, S.; Lopez, M. J. Med. Chem. 2008, 51, 7132. (i) Koulouri,
S.; Malamidou-Xenikaki, E.; Spyroudis, S. Tetrahedron 2005, 61,
10894.
(4) (a) Manthey, M. K.; Pyne, S. G.; Truscott, R. J. W. Aust. J. Chem.
1989, 42, 365. (b) Ikeda, M.; Kitahara, K.; Nishi, H. J. Heterocycl.
Chem. 1992, 29, 289. (c) Zhang, D.; Jin, G. X. Organometallics 2003,
22, 2851. (d) Gellerman, G.; Rudi, A.; Kashman, Y. Tetrahedron 1994,
50, 12959. (e) Reuben, G.; Shonle, H. A. J. Am. Chem. Soc. 1946, 68,
2246. (f) Placin, F.; Clavier, G.; Najera, F.; Desvergne, J. P.; Pozzo, J.
L. Polycyclic Aromat. Compd. 2000, 19, 107. (g) Lehaire, M. L.;
5
O−O bonds as we reported before, while the MP2 method did not
work due to the instability of the wave function that arises from the
homolysis of the O−O bond. In the single point calculations, cc-pVDZ
basis stets were employed for H and Na, and aug-cc-pVDZ basis sets
were employed for C and O. These basis sets are named as BS-II.
In all calculations, the solvation energy was evaluated with the PCM
method. For the determination of the cavity size in the PCM
calculation, the UFF parameters were used along with the united atom
topological model optimized for the HF/6-31G(d) level of theory for
geometry optimization and energy evaluation, respectively. It was
ascertained that each equilibrium geometry exhibited no imaginary
frequency and each transition state exhibited one imaginary frequency.
Enthalpy, entropy, and Gibbs energy changes were evaluated at 298.15
K. Zero-point energy, thermal energy, and entropy change were
evaluated with the DFT(B3LYP) method. The translational entropy in
15
water was evaluated according to the literature method.
1
1202
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203

The Journal of Organic Chemistry
Article
Scopelliti, R.; Herdeis, L.; Polborn, K.; Mayer, P.; Severin, K. Inorg.
Chem. 2004, 43, 1609. (h) Seillan, C.; Brisset, H.; Siri, O. Org. Lett.
2
008, 10, 4013. (i) Tang, Q.; Liu, J.; Chan, H. S.; Miao, Q. Chem.
Eur. J. 2009, 15, 3965. (j) Lee, D. C.; Cao, B.; Jang, K.; Forster, P. M. J.
Mater. Chem. 2010, 20, 867. (k) Tang, Q.; Liang, Z.; Liu, J.; Xu, J.;
Miao, Q. Chem. Commun. 2010, 46, 2977. (l) Di, C.; Li, J.; Yu, G.;
Xiao, Y.; Guo, Y.; Liu, Y.; Qian, X.; Zhu, D. Org. Lett. 2008, 10, 3025.
(
5) Hosoya, T.; Rosenau, T. J. Org. Chem. 2013, 78, 3176.
(
6) Since the borate ester is water-sensitive and alkali-sensitive, the
ester is not formed during the degradation and thus has no effect in
the kinetic analysis.
(
7) The degradation of DHBQ is much faster than that of malonic
acid under the employed conditions. Hence, malonic acid was
quantitatively produced from DHBQ.
(
(
(
8) Mostafa, S. I. Trans. Met. Chem. 1999, 24, 306.
9) Lee, H.; Park, A. H.; Oloman, C. TAPPI J. 2000, 83, 94.
10) The ΔG° value is much higher than the E value due to decrease
a
in entropy by the covalent bond formation between H O and CB.
2
2
(
11) The DFT(UB3LYP) calculations for BsO and Bs resulted in a
considerable degree of spin contamination, but we have described that
the DFT(UB3LYP) functional can reproduce a correct energy profile
in the O−O bond cleavage of H O (see ref 5).
2
2
(
12) (a) Dagaut, P.; Wallington, T. J.; Kurylo, M. J. J. Phy. Chem.
1
988, 92, 3836. (b) Vaghjiani, G. L.; Ravishankara, A. R. J. Phys. Chem.
1
989, 93, 1948. (c) Lightfoot, P. D.; Veyret, B.; Lesclaux, R. J. Phy.
Chem. 1990, 94, 708. (d) Lightfoot, P. D.; Lesclaux, R.; Veyret, B. J.
Phy. Chem. 1990, 94, 700. (e) Wallington, T. J.; Dagaut, P.; Kurylo, M.
J. Chem. Rev. 1992, 92, 667.
(
13) Frisch, M. J., et al. Gaussian 09, Revision A.1; Gaussian, Inc.:
Wallingford, CT, 2009. See Supporting Information for the complete
description.
(
14) (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098. (b) Becke, A. D. J.
Chem. Phys. 1983, 98, 5648.
15) Mammen, M.; Shakhnovich, E. I.; Deutch, J. M.; Whitesides, G.
M. J. Org. Chem. 1998, 63, 3821.
(
1
1203
dx.doi.org/10.1021/jo401486d | J. Org. Chem. 2013, 78, 11194−11203