Markus
Mickel
,
Hee-Cheol
Kim
and
Norbert
Hampp
*
University of Marburg, Institute of Physical Chemistry, Hans-Meerwein-Str. Geb. H, D-35032, Marburg, Germany. E-mail: hampp@mailer.uni-marburg.de
First published on 16th January 2003
The use of mediator-based processes for the delignification of wood pulps is of great interest as they promise to be environmentally as well as economically interesting alternatives to the currently dominating chlorine-based processes. Mediator-based processes suffer from the fact that substantial amounts of the mediator molecules are irreversibly lost during the reactions. We have now analyzed in detail the primary and in addition the secondary reactions of the oxime type mediators violuric acid and dimethylvioluric acid with lignin model compounds and softwood pulp in order to find the reasons for the repeatedly reported loss of mediator molecules during enzymatic and electrochemical delignification processes. Using a set of lignin model compounds, representing the various structural subunits in the lignin network, we were able to demonstrate that the loss of mediator is not due to the primary reaction of the mediator radicals with the lignin but occurs in a secondary reaction where the activated lignin subunits react with the mediators in the solution. The primary reaction of the mediator radicals is hydrogen abstraction from phenols and activated aromatic rings which leads to the formation of phenoxyl and phenyl radicals. The secondary reaction is the formation of colored semi-stable covalent adduct molecules from an activated lignin subunit with a mediator molecule in the solution leading to polycyclic N-hydroxy compounds. Only in this secondary reaction mediator molecules are removed from the solution due to covalent attachment to the lignin network. No reaction with the cellulose content in the pulp is observed. The findings reported in this paper point the way to an improved mediator design and appropriately modified processes, where the secondary mediator reactions are suppressed, and removes a major road block towards a technical application of the mediator-based delignification procedures.
Green ContextThe development of more environmentally benign alternatives to chlorine in pulp bleaching processes is an important large scale industrial challenge—Redox mediators and more recently N-hydroxy compounds have been proposed for this purpose. They generate radicals in the presence of enzymes, and these in turn cause pulp delignification. Unfortunately, significant quantities of these ‘mediators’ are lost in the delignification process. This paper analyses this loss, explains its cause and thus opens the door to improved mediator design. Thus electrochemical mediator-based delignification is shown to be an attractive alternative to existing chlorine-based processes.JHC |
To overcome these restrictions we introduced an electrochemical process where the enzymatic mediator oxidation is replaced by an electrochemical oxidation at the anode of an electrolyzer cell.5,9 The use of an electrolyzer cell removes the restriction to physiological conditions (pH, temperature) implied by the use of enzymes.
For the electrochemical process described the cyclic oximes violuric acid (Vio) and dimethylvioluric acid (dmVio) are suitable mediators. The molecules are electrochemically as well as enzymatically oxidizable to the corresponding iminoxy radicals5 which are extraordinarily stable in the pH range 2–10 and show a half-life of more than 30 min. It has been shown that the radicals undergo a fast reaction with the lignin portion in the pulp leading to a delignification of up to 50% in a single step.9,10 The redox systems of Vio as well as dmVio show a single electron oxidation peak at 1.02 V vs. NHE. The redox system is completely reversible in water as derived from cyclovoltammetric experiments.9,10
However, like in the enzymatic systems a loss of mediator during the electrochemical delignification is observed with Vio and dmVio. This loss limits the efficiency of the total process. We have now analyzed the primary molecular mechanisms of the radical catalyzed depolymerization of lignin in more detail.
For the investigations the model compounds shown in Table 1 were used. The compounds tested comprise classical lignin model compounds as well as monofunctional aromatic molecules. A focus is on guaiacol (Gua) and veratryl alcohol (VA) as representatives for phenolic and for non-phenolic structural lignin subunits. The dimeric lignin models are representatives of the O-4 linkage type, which frequently occurs in lignin, and show similar reactions but due to the multi-functionality of these compounds several chemical reactions occur in parallel and lead to a more complex analysis.
R1 | R2 | R3 | Compound |
---|---|---|---|
OH | OCH3 | H | Guaiacol |
OH | OCH3 | CH2OH | Vanillyl alcohol |
OH | OCH3 | GE | Guaiacolglyceryl-β-guaiacyl ether |
OCH3 | OCH3 | H | Veratrole |
OCH3 | H | H | Anisole |
OCH3 | H | CH2OH | Anisyl alcohol |
OCH3 | OCH3 | CH2OH | Veratryl alcohol |
OCH3 | OCH3 | GE | Veratrolglyceryl-β-guaiacyl ether |
OCH3 | OCH(OH)CH2OH | H | Guaiacolglyceryl ether |
CH2OH | H | H | Benzyl alcohol |
Violuric acid (Vio) as well as N,N′-dimethylvioluric acid (dmVio) form almost identical radicals upon electrochemical oxidation (Fig. 1). From the EPR spectra their radical structure is derived with the unpaired electron localized on the NO group.12 The lifetime of these radicals in water at room temperature is on the order of several tens of minutes.9
Fig. 1 Structure of violuric acid (Vio) and dimethyl violuric acid (dmVio) and EPR-spectra of the corresponding iminoxy radicals. The radicals are the reactive species in the so-called delignification process. |
The reaction of Vio˙ with a number of test compounds is shown in Fig. 2. EPR was employed to monitor the time-dependence of the radical concentration in the reactions with different lignin model compounds. 50 mL of 100 mmol L−1 aqueous solutions (or suspensions where required) of the model compounds were added to 1 L of 0.5 mmol L−1 Vio solution (containing about 40% as Vio˙). The at least 10-fold excess of the model compound guarantees that the kinetics observed is corresponding to the different reactivities of the model compounds. No reaction is observed upon addition of cellobiose (1) but phenolic lignin model compounds (8–10) cause an immediate radical consumption. With non-phenolic aromatic model compounds (2–7) a slower reaction kinetics is observed.
Fig. 2 Time dependent decrease of the EPR radical signal of Vio˙ after the addition of different lignin model compounds. Before t = 0 the thermal decay of the Vio˙ radical in water is seen. At t = 0 the model compounds are added and a small change in the signal intensity due to dilution of the radical is observed for all compounds. |
The electrochemical formation of the mediator radicals Vio˙ and dmVio˙ is equivalent to the abstraction of a hydrogen atom from the oxime group. The fast reaction of Vio˙ with phenolic groups is due to hydrogen abstraction from the phenolic group which leads to the formation of a phenoxyl radical and recovery of Vio. The slower reaction of Vio˙ with the different aromatic model compounds is due to an oxidation of the aromatic ring. The rate constants correspond to the activation of the ring by its substituents and its thermodynamic oxidation potentials (Uox). This is seen from the initially same reaction rates of veratryl alcohol (Uox = 1.35 V vs. SCE) 7, a benzyl alcoholic compound, and veratrole (Uox = 1.27 V vs. SCE) 6, an activated aromatic compound. The comparison of the reaction rates of veratrole 6 and anise alcohol (Uox = 1.45 V vs. SCE) 5 show a clearly higher reaction rate with veratrol which proves that abstraction of benzylic hydrogen atoms is of minor importance. The structural unit where lignin is attacked with Vio˙/dmVio˙ is an aromatic system activated by two methoxy groups, a benzyl alcohol function or a phenolic group (Uox = 0.7 V vs. SCE). The spontaneous radical decrease observed with EPR after addition of pulp is related to the reaction with the phenolic subunits9 because they show the fastest reaction due to the acidic hydrogen atom.
In order to prove the selectivity of the radical the reaction with cellobiose 1 was tested which is a soluble model compound for the carbohydrate content in pulp. By adding cellobiose to a Vio˙ solution no reduction of the radical signal is observed. Thus no reaction takes place with the cellulose part of the pulp, which guarantees a high quality of the pulp produced. Viscosity investigations at electrochemically delignified pulp confirm the very good selectivity of the reaction.
We have analyzed the recovery rate of Vio˙ for the reaction with equimolar solutions of the model compounds guaiacol and veratryl alcohol as well as for softwood Kraft pulp. RDE-voltammetry is a suitable tool to measure simultaneously the amount of Vio and Vio˙ with acceptable time resolution. The decrease of radical as well as the recovery of Vio from Vio˙ after the addition of lignin model compounds or pulp is quantitatively determined. The initial amount of mediator radical (Vio˙) should be theoretically 100% be converted back to mediator (Vio). The amount of mediator recovered in the experiment is given by the difference of mediator at the time where no more radical is detected (Vioend) minus the initial amount of mediator (Viostart). The recovery rate RR is then calculated from RR = ([Vio]end − [Vio]start)/[Vio˙]start. Recovery rates of Vio˙ to Vio with the lignin model compounds veratryl alcohol and guaiacol (Fig. 3) of 52% and 61%, respectively, are obtained. For softwood pulp a recovery rate of 66% is found. The reactions are completed within several minutes.
Fig. 3 Recovery of violuric acid from its radical upon reaction with the model compounds guaiacol and veratryl alcohol as well as softwood kraft pulp. The diagrams show the RDE-analyses of the reaction of Vio˙ with a = guaiacol, b = veratryl alcohol and c = pulp, respectively. The numbering corresponds to measurements at different times: (1) before (2) 30 s; (3) 3 min; (4) 7 min; (5) 11 min after the adding of the model compounds. The recovery rate RR is calculated from RR(t) = ([Vio]end − [Vio]start)/[Vio˙]start. The end of the reaction is reached as soon as no remaining Vio˙ is detected. |
A more detailed analysis is required to investigate which processes occur in parallel. The turnover of each mediator molecule of Vio and dmVio may be calculated to range between two and three cycles in a real experiment. These values are in contradiction to the earlier finding that the mediators show an excellent electrochemical reversibility in aqueous solution. If the primary mediator reaction is not responsible for the apparently low mediator cyclicity then secondary reactions must occur which cause that mediator is removed from the solution and chemically attached to the model compounds or the pulp.
Fig. 4 Analysis of the products obtained from the reactions of mediator radicals Vio˙ and dmVio˙ with the model compounds guaiacol and veratryl alcohol. Shown is the total ion count (positive mode) chromatogram after 2 min and 24 h reaction time and the mass spectra of the formed adduct molecules after 2 min reaction time. In the last column the time dependent UV/Vis (solid line: 2 min reaction time, broken line 24 h reaction time) spectra are shown. |
However, also a striking difference is observed in that while the Vio adduct disappears after 24 h reaction time, the corresponding dmVio product is stable.
The formation of a covalent adduct is not only observed with the phenolic model compound guaiacol, but also with benzyl alcoholic model compounds like veratryl alcohol and also with activated aromatics like veratrole (not shown). In the case of veratryl alcohol the expected mass of the adduct molecule is 324 u with Vio and 352 u with dmVio. Both signals are observed, but the most intense ion signals correspond to an ion after an internal condensation reaction (308 u and 334 u).
After 24 h reaction time with both mediators (Vio and dmVio) no residual adduct molecules are detected which is in contrast to the phenolic lignin model compounds. In the Experimental section a more detailed summery of the reactant and product peaks and their identification is given (Table 2).
Retention time/min | Mass/u | Substance | |
---|---|---|---|
a Most abundant ion. | |||
Reactants | 9.2 | 156 | Vio (not shown) |
25.5 | 184 | dmVio | |
20.9 (UV) | No ion | Gua | |
29.8 | 167, 151a | VA | |
Gua + Vio | 18.9 | 280,a 138 | Adduct |
33.3 (24 h) | 276,a 260 | Quinoid dimer | |
Gua + dmVio | 25.7 | 184 | dmVio |
27.1 | 308,a 292, 138 | Adduct | |
32.6 924 h) | 276,a 260 | Quinoid dimer | |
VA + Vio | 25.9 | 324, 306,a 182 | Adduct |
29.8 | 168, 151a | VA | |
32.5 | 167,a 151 | Veratric aldehyde | |
VA + dmVio | 25.7 | 184 | dmVio |
29.4 | 352, 334a | Adduct | |
30.1 | 168, 151a | VA | |
32.8 | 167,a 151 | Veratric aldehyde |
The formation of a semi-stable adduct also can be monitored by UV/Vis spectroscopy. From reaction of the radicals with the model compounds intensely redish colored solutions result, with an absorption maximum found at around 500 nm (Fig. 4). A continuous reduction of the absorption is observed which conforms to the semi-stability of the adducts found by LC-MS measurements.
From these experiments it is concluded that the reaction of mediators and model compounds lead to 1∶1 covalent adduct molecules. In most cases this adduct molecule is not stable and the adduct complex hydrolyses (type a), but in some cases, here e.g. dmVio and guaiacol, a stable adduct is formed (type b). The first mechanism (type a) could explain a transient loss of mediator, but the latter mechanism (type b) also may hold for an irreversible loss of mediator due to chemical attachment to the pulp.
The mass spectra indicate the formation of dimeric quinoid compounds from the lignin model compounds (Fig. 5). The same product is formed from reactions involving the different mediators Vio and dmVio. This corresponds with the reaction scheme that in a first step the model compound reacts with the Vio˙ under formation of an aromatic radical and in a second step a dimerisation with another model compound molecule occurs. Oxygen-rich quinoid systems are formed via reactions with the oxygen content of the aqueous solution.14–16 Performing MS/MS experiments with the product ion of mass 276 u leads to a major fragment with 244 u which correlates to the loss of molecular oxygen (Fig. 5) which proves the above mentioned mechanism.
Fig. 5 Formation of mediator type independent guaiacol dimers. The MS and MS/MS spectra of the molecule and a proposed reaction scheme is shown. |
The reaction mechanism found here is only triggered by the reaction of the mediator radical with the lignin model compound but the mediator does not take part in the second reaction step. Such type of reactions will cause undesired products but it would not cause any loss of mediator. Firthermore the dimerisation reaction is of little importance for the lignin in the pulp, as there is no free mobility of the aromatic subunits.
Hydrogen abstraction of the mediator-radical from lignin or the model compounds leads to the formation of secondary radicals which should be detectable by EPR spectroscopy as long as they have a sufficient lifetime. In non-protic organic solvents an extended lifetime of the secondary radicals is expected. For the detection of these secondary model compound radicals lyophilized Vio˙ and dmVio˙ was prepared and dissolved in cold acetonitrile (−40 °C). As a suitable lignin model compound the dimeric phenolic guaiacolglyceryl-β-guaiacyl ether was chosen. A small amount of the ether dissolved in cold acetonitrile was added to the mediator radical solution in acetonitrile. When Vio˙ was used no secondary radical signal could be detected after the addition of the model compound. Using dmVio˙ led to the formation a new radical signal, which clearly indicates a disappearance of the oxime structure (Fig. 6). The EPR signal of iminoxy radicals is split into three groups with a hyperfine constant ranging from 28 to 32 G due to an interaction with the oxime nitrogen atom (Fig. 6, upper curve).17 After the addition of guaiacolglyceryl-β-guaiacyl ether a radical signal is detected, which shows splitting into three groups of lines of same intensity. This indicates a coupling with a spin I = 1 atom like nitrogen. However, now the hyperfine constant is only approx. 10 G typical of N-hydroxy radicals.18 The further splitting of each group of lines into a doublet with a hyperfine constant of 2 G is caused by a proton at the α-carbon or a proton from a hydroxy group at the nitrogen atom. The detected radical signal belongs to the primary radical adduct between dmVio and the model substance guaiacolglyceryl-β-guaiacyl ether. With the formation of a covalent bond the oxime structure is lost.
Fig. 6 Radical reaction monitored by EPR in acetonitrile (−40 °C). (Upper curve) The EPR-signal of dmVio˙ with its large coupling constant is typical for an oxime radical. (Lower curve) After reaction of dmVio˙ with guaiacolglyceryl-β-guaiacyl ether the dmVio˙ signal disappears and a new radical signal with a much smaller coupling constant appears. |
In Fig. 7 a model for the structure of the adducts as well as their development is proposed on the basis of veratrol and guaiacol. Primarily a radical adduct is formed from the model compound radical obtained in the first step with a mediator molecule which leads to the formation of a N-hydroxy radical structure as seen in EPR. Due to the radical structure it is assumed that further internal reactions will occur, in particular ring-closure, before the radical character is lost via an inhibitor.
Fig. 7 Proposed mechanism for the mediator–substrate adduct formation. From the LC-MS measurements a one-to-one stoichiometry between radical and lignin model compound is derived. |
The reactions cause the formation of semi-stable molecules. Using the proposed reaction model (Fig. 7) the fragmentation pattern of the APCI mass spectrum of the veratryl alcohol–Vio adduct may be interpreted (Fig. 8). The formed adduct releases H2O during ionization and a break down of the mediator part of the adduct is induced. Similar reactions occur in the reaction solution. They are responsible for the disappearance of the adduct signal within 24 h reaction time.
Fig. 8 Shown is the APCI mass spectrum of the adduct of Vio˙ and veratryl alcohol. The fragmentation pattern observed fully corresponds with the ring formation theory. |
It was found that the mediator radicals in aqueous solution react very rapidly with phenolic groups, and more slowly with activated aromatic rings under abstraction of hydrogen atoms. Hydrogen abstraction from the benzyl alcoholic carbon is of minor importance. No reaction with the cellulose content of pulp is expected as experiments with cellobiose have shown.
The reaction stoichiometry between the mediator radicals and the model compounds was found to be one to one.
The molecular cause for the loss of mediator is due to the formation of covalent adducts from mediator-activated model compound radicals (and their counterparts in the lignin network, respectively) with mediator molecules. The adduct molecules predicted by this model have been identified by LC-MS spectrometry. The adducts are colored semi-stable compounds. Primary radical N-hydroxy type adducts were identified by EPR measurements. As a consequence of the radical character further internal reactions are assumed which stabilize the adduct.
The findings reported in this paper point the way to an optimization of mediator-based enzymatic as well as electrochemical processes for pulp delignification. The interception of the induced secondary radicals by additives will increase the cyclicity of the mediator.
For rotating disk electrode (RDE) measurements a BM-EDI101 electrode and CTV 101 speed control module from Radiometer Analytical were used. As a working electrode a glassy carbon disk with 3 mm diameter, as a counter electrode a glassy carbon rod with 5 mm diameter and a saturated calomel electrode (SCE) as a reference electrode were used. The electrode was rotated at 1000 rpm and the scan rate was 20 mV s−1 in the potential range from 0.1 to 1.1 V vs. SCE. All experiments were performed in 0.1 M acetate buffer (pH 4.5). For the determination of the mediator recovery rate for the reaction with guaiacol (UOx = 0.7 V vs. SCE) it is important to add only a small excess of guaiacol to minimize the error due to direct electrochemical oxidation of guaiacol. The RDE-measurements with pulp were done at a pulp consistency of 3% (1.3 g dried pulp in 40 ml 0.1 M acetate buffer pH 4.5). A direct oxidation of the pulp is not observable due to the fact that a direct contact of electrode and pulp scarcely occurs in the experiment.
LC-MS/MS measurements were done on a LCQduo from Thermoquest Finnigan equipped with an LC-system comprising of a P4000 gradient pump, a UV2000 UV/Vis detector and a RP18 Nucleosil column from Bischoff (Germany) (length 250 mm, 2 mm diameter, particle size 5μm). The mobile phase was a linear gradient from water containing 0.5% formic acid (A) and methanol, containing 10% water and 0.5% formic acid (B) (0–10 min: 100% A; 10–30 min: linear gradient 0% to 100% B; 30–60 min: 100% B). The concentration of the samples was 1–5 mmol L−1. APCI ionization (positive mode) with nitrogen as shooting gas was used. The radical solution in 0.1 M acetate buffer pH 4.5 was prepared electrochemically, and the radical concentration was determined by RDE-voltammetry. An excess of the model compounds in 0.1 M acetate buffer (pH 4.5) was added. Samples were taken from the reaction solution at different times and immediately frozen in liquid nitrogen. Then the samples were successively measured by LC-MS. In Table 2 the retention times and masses of the different peaks are given. The main peaks were identified by their mass spectra apart from the peak with the retention time 31.5 min when reacting Vio and dmVio with VA. The peak consists of a major ion with 182 u and a fragment with 151 u. The signal does not fit in its retention time and fragmentation pattern of veratric acid (m/z = 200, 183 (100%), 139, 117).
UV/Vis spectra were measured on a UV900 from Kontron. The probe was a diluted sample (1 to 1000) from the LC-MS investigations.
This journal is © The Royal Society of Chemistry 2003 |