A New Mechanism in the Ozone Reaction with Lignin

M. Ragnar et al.: Formation of Superoxide Radical in the Reaction of Ozone with Lignin

423

Holzforschung 53 (1999) 423–428

A New Mechanism in the Ozone Reaction with Lignin Like Structures By Martin Ragnar1, Tord Eriksson1, Torbjörn Reitberger2, 3 and Peter Brandt4 1

Kungliga Tekniska Högskolan, Institutionen för Pappers- och massateknik, Träkemi, Stockholm, Sweden Kungliga Tekniska Högskolan, Institutionen för Kemi, Kärnkemi, Stockholm, Sweden 3 Visiting scientist to Skogsindustrins Tekniska Forskningsinstitut (STFI), Stockholm, Sweden 4 Kungliga Tekniska Högskolan, Institutionen för Kemi, Organisk kemi, Stockholm, Sweden 2

Keywords

Summary

Ozone Hydroxyl radicals Superoxide Pulp bleaching Lignin model compounds Mechanisms Quantum chemical calculations

A new mechanism including a homolytic cleavage of a trioxide intermediate forming superoxide is suggested to be the main course of radical formation in reactions of ozone and lignin like structures. The suggested mechanism is supported by quantum chemical and thermochemical methods.

Introduction When ozone is used to bleach pulp, the desirable reaction involves degradation of unsaturated structures in the lignin via ionic mechanisms known as ozonolysis (Criegee 1957). This has been reported to be the major reaction pathway for lignin model compounds in non-aqueous media (Kaneko et al. 1981; Eriksson and Gierer 1985). However, in aqueous media the reactions become more complex (Pryor 1994). Electron transfer from phenolic lignin structures to ozone was proposed to give hydroxyl radicals directly (Eriksson and Reitberger 1995; Lind et al. 1997). Hydroxyl radicals rapidly attack pulp constituents unselectively and in the presence of oxygen a superoxide mediated chain reaction is established. This chain reaction is probably the main cause of fibre strength losses in ozone bleaching. In a recent paper (Ragnar et al. 1999), radical formation in reactions between ozone and lignin model compounds was investigated by a new method affording a complete blocking of the chain reaction. Using this method initial radical yields were measured for the first time. Surprisingly, it was found that superoxide rather than the hydroxyl radical is the initial radical species formed, i. e. hydroxyl radicals arise predominantly by way of the superoxidemediated chain reaction. In this paper, we discuss feasible mechanisms for the reactions of ozone with lignin in the light of our new findings. To corroborate our suggested mechanisms the stability of key intermediates was estimated by quantum chemical calculations. Computational Methods Bond Dissociation Energies (BDE) All calculations were performed using the Gaussian 94 program (Frisch et al. 1995). Geometry optimisations were initially performed at the HF level together with the 6–31G basis set in Holzforschung / Vol. 53 / 1999 / No. 4  Copyright 1999 Walter de Gruyter ⋅ Berlin ⋅ New York

Gaussian 94; as some of the species were anionic, a set of diffuse s and p functions was added to all oxygen atoms, i. e. 6–31+G. This basis set will be referred to as 6–31(O+)G. After this step, a second geometry optimisation was performed using the B3LYP hybrid functional (Lee et al. 1988; Becke 1993) with the same basis set. B3LYP is a density functional theory type method based on hybrid functionals. The final energy assessments were calculated with B3LYP using the large 6–311+G(2 d, p) basis set in the Gaussian 94 program. Solvation We have tried to quantify the effect of solvation in water by single point calculations using the polarisable continuum model (PCM/DIR) (Cossi et al. 1996). We have also performed calculations where the anionic oxygen is stabilised by partial solvation, adding three directly coordinating water molecules.

Results and Discussion Changing the oxidation state of the para-substituent on a (phenolic) guaiacyl ring results in rather different rate constants for the reaction of ozone with lignin model compounds (Ragnar et al. 1999). We have estimated the oxidation potentials for these compounds using a refined technique described by Jonsson (1995). When the logarithm of the rate constant was plotted as a function of the oxidation potential, a straight line was obtained (Fig. 1). Similar behaviour was observed when the same plot was drawn for some non-phenolic model compounds, i. e. veratric acid, 1,2-dimethoxybenzene, veratryl alcohol, 1,4dimethoxybenzene and 3,4-dimethoxytoluene. The linear relationship in Figure 1 suggests that the reaction of ozone with these compounds proceeds via a charge-transfer state. From the charge-transfer state two reaction pathways are possible. In pathway A (Fig. 2), a complete electron-transfer takes place, resulting in the formation of an aromatic cation radical and an ozonide radical, which after protonation decomposes to oxygen

424


mixture from ozone treatment of creosol, showing a small formation of phenoxyl radicals in the system. Ozone adds to the aromatic ring, preferentially to the oxygen-substituted carbons (Kaneko et al. 1983) as shown in Figure 2, pathway B. The resulting zwitterion subsequently reacts via different routes, which will be discussed separately for phenolic and non-phenolic structures.

Fig. 1. The rate constant for radical formation as a function of the estimated one-electron reduction potential for a series of phenolic lignin model compounds.

and a hydroxyl radical, i. e. pathway A describes a direct route to hydroxyl radical formation. Presumably, this reaction pathway is more important for phenolic compounds under alkaline conditions, since the aromatic cation radical immediately yields the corresponding phenoxyl radical. A small amount of the phenolic coupling product bis-creosol was indeed detected in the reaction

Phenolic structures If R = H (Fig. 2), the hydroxyl substituent of the zwitterion will immediately be deprotonated, forming a keto-function. Direct decomposition of the trioxide by demethoxylation to form an ortho-quinone (Fig. 3, pathway C), is less probable, since the oxygen formed would be in the singlet state. Pathway D (Fig. 3) shows a homolytic cleavage of the trioxide to yield superoxide and a quinol radical, which may eventually yield a quinone product. Indeed, the strongly coloured solutions observed after partial ozonation of model compounds indicate the formation of quinones. Pathway E (Fig. 3) is a heterolytic cleavage of the aromatic ring that yields the same reaction products as does the ozonolysis. In this reaction, hydrogen peroxide is also formed.

Fig. 2. Proposed mechanism of the initial reaction between ozone and an aromatic lignin structure, via a charge-transfer state.

Fig. 3. Proposed mechanisms of ozone reaction with creosol in aqueous media. Holzforschung / Vol. 53 / 1999 / No. 4


425

Fig. 4. Proposed mechanisms of ozone reaction with 3,4-dimethoxytoluene in aqueous media.

There is always a competition between route D and E. In acidic solution, route D becomes less favourable, since homolytic cleavage of the hydrotrioxide would lead to formation of the oxidising hydroperoxyl radical. Hence, one may expect an increase in radical formation at around the apparent pKa-value of the trioxide, which under our experimental conditions was estimated to be about 4 (Ragnar et al. 1999).

Fig. 5. Hydroxyl radical formation from initially formed superoxide.

Fig. 6. Different forms of the trioxide intermediate used in calculations.

Non-phenolic structures If R = CH3 (Fig. 2), deprotonation of the zwitterion cannot occur. Instead heterolytic ozonolysis, route F (Fig. 4), becomes favourable. This reaction route may be regarded as a classical Criegee type of reaction (Criegee 1957). Under alkaline conditions, a nucleophilic attack by hydroxide ion on the aromatic ring is another possibility, as shown in route G (Fig. 4). This leads to demethoxylation and superoxide formation, analogous to pathway D (Fig. 3). pH-dependence For all phenolic lignin model compounds investigated a rapid increase in the radical yield was observed at pH 3, whereas non-phenolic compounds monotonically yield more radicals at pH ≥ 6 (Ragnar et al. 1999). The latter behaviour may be explained from pathway G (Fig. 4). For phenolic compounds, the lower radical yield in acidic solution reflects both a lower fraction of homolysis according to pathway D and the protonation and elimination of superoxide (Fig. 5). Quantum chemical calculations In order to substantiate the proposed mechanisms quantum chemical calculations were performed to assess the O–O BDE of the trioxide intermediates (Fig. 6). The results of the calculations are summarised in Table 1 and will be discussed below.

Fig. 7. The hydrotrioxide (I) has a calculated O–O bond length of 1.53 Å.

Isolated molecular species Using the largest basis set we obtained a BDE of about 14 kcal/mol (Table 1) for the RO–OOH bond (I) (Fig. 7). The validity of the calculated BDE was investigated by performing a similar calculation on hydrogen peroxide. The calculated O–O BDEs in hydrogen peroxide are shown in Table 2. Holzforschung / Vol. 53 / 1999 / No. 4

426


Table 1. Homolytic O–O BDE (∆E[kcal/mol])a Computational level

RO–OOH (I)

RO–OO– (II)

B3LYP/6–31(O+)G//HF/6–31(O+)G B3LYP/6–31(O+)G B3LYP/6–31+G*//B3LYP/6–31(O+)G B3LYP/6–311+G(2 d,p)//B3LYP/6–31(O+)G

15.1 18.1 15.4 14.4

7.9 ——b ——b ——b

B3LYP/6–31(O +)G//HF/6–31(O +)G + ∆G(solvation)c

11.4

–6.1

RO–OO–(H2O)3 (III) 4.1 9.9 ——d ——d –10.8

a Zero point energy corrections (ZPC) not included. b Not a stationary point on the B3LYP/6–31(O+)G PES. c Solvation contributions calculated using PCM/DIR (HF/6–31G*//HF/6–31(O+)G) at 298 K. d Due to problems in the SCF convergens the B3LYP6–31+G* and B3LYP/6–311+G(2 d,p) energies could not be determined.

Table 2. O–O BDE (∆E[kcal/mol]) in H2O2 Computational level

∆E

B3LYP/6–31(O+)G B3LYP/6–31+G*//B3LYP/6–31(O+)G B3LYP/6–311+G(2 d,p)//B3LYP/6–31(O+)G ZPCa Thermal correction to enthalpya

42.2 48.6 47.9 –5.3 –3.0

a

ZPC and thermal B3LYP/6–31(O+)G.

correction

to

enthalpy

calculated

at

The BDE for the O-O bond was calculated to be 47.9 kcal/mol at the B3LYP/6–311+G(2d, p)//B3LYP/6–31 (O+)G level. Including a thermal correction to enthalpy calculated at the B3LYP/6–31(O+)G level, the bond enthalpy was calculated to 45 kcal/mol. The experimentally determined value of this dissociation enthalpy is 51 ± 1kcal/mol (Kerr 1966). The calculations thus seem to underestimate the O–O BDE by a few kcal/mol. The trioxide intermediate II was found to be a non-stationary point at the B3LYP/6–31(O+)G potential energy surface (PES), eventually optimising to singlet oxygen and a hemi-ketal anion. This reflects the anion stabilising effect of the adjacent methoxy group. The BDE of the II was calculated to be about 8 kcal/mol at B3LYP/6–31 (O+)G//HF/6–31(O+)G, 7 kcal/mol less in energy than I.

Energy of solvation Evaluation of relative BDEs in molecules of different total charge is a very delicate problem, especially in polar protic solvents where solvation could affect the relative energy difference. The reaction under investigation is carried out in water, a solvent that strongly solvates charged species. The effect of solvation on I was found to be about –3 kcal/mol. Since the neglect of ZPC will lead to an overestimation of about this size, the relevant BDE for RO–OOH in water can be estimated to be about 15 kcal/mol. The effect of solvatisation on II was investigated by adding a set of three water molecules coordinated to II. This species (III) was found to be stable also at the B3LYP PES, Holzforschung / Vol. 53 / 1999 / No. 4

although the RO–OO– bond length is strongly elongated, (1.75 Å) (Fig. 8). The solvation of III, lowered the BDE to 4 kcal/mol at B3LYP/6–31(O+)G//HF/6–31(O+)G, indicating that a real solvation in water of II will further decrease the stability. It should be noted that the decrease in BDE upon partial solvation was achieved although the dissociation of the O–O bond included breaking of hydrogen bonds. It is clear from the results from partial solvation that solvation plays an important role. In order to further evaluate the effects of solvation, a final estimation of solvent effects was performed using single point calculations on the Hartree-Fock (HF) geometries using PCM/DIR. These results strongly support the conclusion that II is a very labile compound, although the exact BDE is difficult to estimate. Thermochemical discussion The quantum chemical calculations yielded a value of the BDE of RO–OOH (I) of about 15 kcal/mol, whereas the

Fig. 8. The partially solvated anion III (B3LYP/6–31(O+)G) has a calculated O–O bond length of 1.75 Å.

Fig. 9. Correlation between relative acidities and BDEs.


BDE of RO–OO– (II) could only be assessed to be in the interval 0–10 kcal/mol. Provided that II is a real specie and not only a transient state the thermochemical scheme in Figure 9 should be valid. From Figure 9 it is possible to express energy relationships. By Hess’ law it follows that: ∆H + ∆HA⋅ = ∆HA + ∆H′

(1)

The difference in BDE between the trioxide (II) and the protonated trioxide (I) could thus be expressed as: ∆H – ∆H′ = ∆HA – ∆HA⋅

(2)

∆GA = RT In Ka, A = ∆HA – T∆SA

(3)

∆GA⋅ = –RT In Ka, A⋅ = ∆HA⋅ – T∆SA⋅

(4)

Substitution of equation (3) and equation (4) in equation (2) then leads to the expression  K a,A  ∆H – ∆H′ = T(∆SA – ∆SA⋅) – RT In    K a,A ⋅ 

(5)

∆H – ∆H′ = 2.3 RT [pKa,A – pKa,A⋅]

(6)

pKa,A⋅ = pKa (HO⋅2) = 4.88 (Behar et al. 1970)

(7)

pKa,A⋅ may be estimated by comparing the pKa-values for water (14.0) and hydrogen peroxide (11.6). This would suggest that: (8)

Cherkasov (1998) derived a correlation equation to predict any unknown pKa-value on the basis of calculated inductive and steric constants of substituents attached to the terminal oxygen in peroxide derivatives. Using his equation we obtain pKa = 9.7 for the hydrotrioxide. Using these pKa-values in equation (6) we obtain ∆H – ∆H′ ≥ 6 kcal/mol. This can also be regarded as the difference in BDE between RO–OOH and RO–OO–, i. e. ∆E – ∆E′. From the quantum chemical calculations we obtained ∆E ≈ 15 kcal/mol which gives ∆E′ ≤ 9 kcal/mol in accordance with the quantum chemical estimation. Using the Arrhenius equation: k1 = Ae/E/a/RT

(10)

where kH+ is the rate of the protonation of II. Using the obtained value for k1 in equation (10) gives a rate constant for the protonation 1010 M–1s–1 ≤ kH+ ≤ 1011 M–1s–1. This seems to be in good accordance with the findings of Eigen (1954). The observed pKa-value in the radical yield versus pH-curve may thus be kinetically rather than thermodynamically determined. Conclusions A new mechanism for the reaction of ozone with lignin is proposed. The stability of key intermediates in the reaction was investigated by quantum chemical and thermochemical methods. It is shown that trioxide intermediates are easily cleaved to yield superoxide. This is in accordance with our experimental finding that superoxide rather than the hydroxyl radical is the primary radical species formed when ozone reacts with lignin model compounds. Acknowledgements

Since the differences in entropy change must be small

pKa, A = pKa (ROOOH) ≥ 9

k1 = kH+ 10–pKa

427

(9)

it is now possible to calculate the rate constant for the homolysis. Taking the pre-exponential factor A ≥ 1013 s–1 we obtain k1 ≥ 2 × 106 s–1 which means that the lifetime of II should be less than 0.5 µs. Similarly we obtain a life time of the order of 10 ms for I. This shows that the hydrotrioxide (I) has a sufficient life time to participate in heterolytic reactions as indicated in pathway E (Fig. 3). The radical yield versus the pH-curve shows an apparent pKa-value between 4 and 5 (Ragnar et al. ). If this change in the radical yield merely reflects the competition between homolysis and protonation of II, the rate of the homolytic reaction, k1, may be estimated from equation (10).

The Swedish Research Council for Engineering Sciences (TFR), the Nordpap-programme of the Nordic Industrial Fund, the Ernst Johnson Foundation and the Trygger Foundation are thankfully acknowledged for financial support. Dr. Mats Jonsson for fruitful discussions on oxidation potentials. We are also grateful to Prof. Jacopo Tomasi and Dr. Maurizio Cossi for a copy of the PCM/DIR program.

References Becke, A.D. 1993. Density-functional thermochemistry III. The role of exact exchange. J. Chem. Phys. 98, 5648–5652. Behar, D., G. Czapski, J. Rabani, L.M. Dorfman and H.A. Schwarz. 1970. The acid dissociation constant and decay kinetics of the perhydroxyl radical. J. Phys. Chem. 74, (17), 3209–3213. PCM/DIR courtesy of Cossi, M., V. Barone, R. Cammi and J. Tomasi. 1996. Ab initio study of solvated molecules: A new implementation of the polarizable continuum model. Chem. Phys. Lett. 255, 327–335. Cherkasov, A.R. (1998). Private communication. Criegee, R. 1957. The course of ozonization of unsaturated compounds. Record of Chemical Progress. 18, (2), 111–120. Eigen, M. 1954. Ionic reactions in aqueous solutions with half-times as short as 10–9 second. Applications to neutralization and hydrolysis reactions. Discussions Faraday Soc. 17, 194–205. Eriksson, T. and J. Gierer. 1985. Studies on the ozonation of structural elements in residual kraft lignins. J. Wood Chem. Technol. 5, (5), 53–84. Eriksson, T. and T. Reitberger. 1995. Formation of hydroxyl radicals from direct ozone reactions with pulp constituents. 8th International Symposium on Wood and Pulping Chemistry. Helsinki, Finland. vol. II, 349–354. Frisch, M.J., G.W. Trucks, H.B. Schlegel, P.M.W. Gill, B.G. Johnson, M.A. Robb, J.R. Cheeseman, T. Keith, G.A. Petersson, J.A. Montgomery, K. Raghavachari, M.A. Al-Laham, V.G. Zakrzewski, J.V. Ortiz, J.B. Foresman, J. Cioslowski, B.B. Stefanov, A. Nanayakkara, M. Challacombe, C.Y. Peng, P.Y. Ayala, W. Chen, M.W. Wong, J.L. Andres, E.S. Replogle, R. Gomperts, R.L. Martin, D.J. Fox, J.S. Binkely, D.J. Defrees, J. Baker, J.P. Stewart, M. Head-Gordon, C. Gonzales and J.A. Pople. 1995. Gaussian 94. Revision B.2, Gaussian Inc. Holzforschung / Vol. 53 / 1999 / No. 4

428


Jonsson, M. 1995. One-electron oxidized substituted benzenes – thermochemistry and kinetics. Ph.D. Thesis. Kungl Tekniska Högskolan, Stockholm, Sweden Kaneko, H., S. Hosoya and J. Nakano. 1981. Degradation of lignin with ozone. Reactions of biphenyl and a-carbonyl type model compounds with ozone. Mok. Gakk. 27, 678–683. Kaneko, H., S. Hosoya, K. Iiyama and J. Nakano. 1983. Degradation of lignin with ozone – Reactivity of lignin model compounds toward ozone. J. Wood Chem. Technol. 3, (4), 399–411. Kerr, J.A. 1966. Bond dissociation energies by kinetic methods. Chem. Rev. 66, 465. Lee, C., W. Yang and R.G. Parr. 1988. Development of the Colle-Salvetti correlation-energy formula into functional of the electron density. Phys. Rev. B. 37, 785–789. Lind, J., G. Merényi and K. Wegner. 1997. The chemistry of ozone during pulp bleaching. J. Wood Chem. Technol. 17, (3), 297–326. Pryor, W.A. 1994. Mechanisms of radical formation from reactions of ozone with target molecules in the lung. Free Radical Biology and Medicine. 17, (5), 451–465.

Holzforschung / Vol. 53 / 1999 / No. 4

Ragnar, M., T. Eriksson and T. Reitberger. 1999. Radical formation in ozone reactions with lignin and carbohydrate model compounds. Holzforschung in press. Received March 12th 1998 Martin Ragnar Tord Eriksson Kungl Tekniska Högskolan Department of Pulp and Paper Chemistry and Technology Division of Wood Chemistry SE-100 44 Stockholm Sweden Torbjörn Reitberger Kungl Tekniska Högskolan Department of Chemistry Division of Nuclear Chemistry SE-100 44 Stockholm Sweden

Peter Brandt Kungl Tekniska Högskolan Department of Chemistry Division of Organic Chemistry SE-100 44 Stockholm Sweden