Vol. 7, No. I , January-February 1974 be used as an approximate measure of the extent of an otherwise difficult otm characterize compositional heterogeneity. For example, copolymers A, B, and C of Table I probably contain some relatively longer sequences of acrylonitrile since the estimate of the average acrylonitrile sequence length based on Rz is much larger than those based either on RI or the final composition. Naturally,
Simulation of Reactions with Lignin by Computer
17
this kind of measure is more accurate for that range of copolymer composition in which very long sequences are not likely (which is the case for the copolymers of Table I). This is true since the actual nmr measurement is in terms of triads. Thus, the measure of deviations from theoretical predictions of average sequence lengths is most sensitive for those lengths which are on the order of about 2-4, or approximately the length of a triad.
Simulation of Reactions with Lignin by Computer (SIMREL). I. Polymerization of Coniferyl Alcohol Monomers1 Wolfgang G. Glasser*2aand Heidemarie R. G l a s s e r z b Department of Forestry and Forest Products, Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Received June 6, 1973 ABSTRACT: A computer program was written to simulate the formation of a dehydrogenation polymer of coniferyl alcohol starting with a monomer of the type CgHgOzOCHs with another monomer of the same type, forming a polymer of up t o 100 units (aglomerated monomers). Coupling reactions are selected according to a statistical distribution of predetermined values found in the literature. Coupling modes include combinations of any of four radical species leading to previously reported interunit bonds. Tables are printed out giving the reaction sequence and the percentage figures of the reaction types at the time of simulation. A summary with the final configuration of the polymer is printed at the end of the simulation run, also listing a modified Rydholm diagram and other tables. The simulation mechanism is based on code numbers of molecule positions in Ca units that identify chemical elements and reactivity coefficients and that decide upon the possibility of a molecule position to perform a reaction. Three states of reactivity are possible: reactive, not reactive, and “latent.” A “latent reactivity” classifies a reaction as possible with the exception that a statistical percentage figure does not permit further reaction with this particular position because a prescribed analytical value is lower or equal to one being computed. The SIMREL program searches for a reactive position in the polymer under formation and performs a reaction with a new monomer unless it is statistically impossible; then, the reactive position is converted into a latent one. A limit on the number of units in the polymer or exhaustion of reactive positions terminates the simulation. 1. Introduction
Lignin is known to be a three-dimensionally branched polyphenol, synthesizled by some plants from phenylpropenoid precursors (for review, see ref 3-6). Enzymatically triggered phenolic coupling has been recognized as the motor of polymerization. Although this mechanism is widely believed to proceed in a random manner, some schemes have been proposed to the c ~ n t r a r y . ’ . However, ~ one way or the other, there seems to be no doubt as to the initial enzymatic control of its formation from coniferyl alcohol and related phenylpropanes. If there is a control mechanism that would cause lignin to form in “repeating units,” it would be unique to the biochemistry of plants, as other natural polyphenols such as tannins and phlobaphenes are considered to be disordered polymers. Also, the existence of such a control mechanism would probably have little effect on the overall chemical makeup of lignin and its reactions. The concept of lignin as a completely ordered molecule might be as erroneous as that of lignin (1) Part of the results have been presented in a paper given at the 4th (2) (3) (4) (5) (6) (7) (8)
Canadian Wood Chemistry Symposium, held in Quebec, Canada, July 4-6, 1973. (a) Assistant Professor ‘of Wood Chemistry; (b) M. S., formerly a Research Associate, College of Forest Resources, University of Washington, Seattle, Wash. K. V. Sarkanen and C . H . Ludwig, Ed., “Lignins-Occurrence, Formation, Structure and :Reactions,” Wiley Interscience, New York, N . Y., 1971. K. Freudenberg and A. C. Neish, “Constitution and Biosynthesis of Lignin,” Springer-Verlag, New York, N . Y., 1968. J. M. Harkin in “Oxidative Coupling of Phenols,” W. I. Taylor and A. R. Battersby, Ed., Marcel Dekker Inc., New York, N . Y., pp 256-263. T. Higuchi, Aduan. Enzymol., 34,207 (1971). K. Forss and K . E. Fremer, Pap. Puu, 47,443 (1965). K. Forss, K . E. Fremer, and B. Stenlund, Pap. Puu, 48, 565, 669 ( 1965).
as an entirely homogeneous substance; recent indications obtained by several investigators are that lignin configuration varies with its location in the cell ~ t r u c t u r e . ~The -l~ assumption that a statistical polymer formed essentially randomly through a limited number of oxidative coupling reactions of one, two, or three p-hydroxycinnamyl alcoholtype phenols makes this substance well suited for mathematical simulation by computer. Such mathematical models have previously been used in the field of pulp and paper technology to simulate pulping processes in terms of such process parameters as liquor: wood ratio, chemical charge, and reaction r a t e ~ . l These ~ - ~ ~ models allowed to make reasonable predictions as to yields and lignin contents of resulting pulps. However, no attempts have been (9) B. J. Fergus and D.A . I,. Goring, Holzforschung, 24, 113,118 (1970). (10) R. Draganova and R. C. Simionescu, Cellulose Chem. Technol., 2, 359 (1968). (11) R. Draganova, Tseluloza Khartiya, 2 (16), 17 (1971); cited in Abstr. Bull. Zmt. Pap. Chem., 43,2790 (1973). (12) R. Draganova, Cellulose Chem. Technol., 5,451,457,463 (1971). (13) 0. P . Grushnikov and N . N . Shorygina, Usp. Khim., 40, 1394 (1971); citedinAbstr. Bull. Inst. Pap. Chem., 42,11138 (1972). (14) M . Matsukura and A. Sakakibara, Mokurai Gakkaishi, 15,35 (1969). (15) T . J. Williams, Ind. Eng. Chem., 59,53 (1967). (16) J . Schmied, Papier (Darmstadti, 21 (10A). 672 (1967). (17) A. J. Kerr, APPITA, 24,180 (1970). (18) A . H.Morris, Pap. TradeJ., 152 (39), 42 (1968). (19) Yu. N . Nepenin, I. S. Shul’man, V. P. Denisenko, M. I. Ternitskii,
and I. A. Faradkhev, Mater. Nauch.-Tekh. Konf. Lesotekh. Akad., I , 10 (1967), cited in Abstr. Bull. Inst. Pap. Chem., 41,4347 (1971). (20) Yu. A. Makhov, Sb. Tr. TNIZ Bumagi, No. 4, 158 (1969); cited in Abstr. Bull. Inst. Pap. Chem., 41,4345 (1971). (21) I. E. V’Yukov, Sb. Tr. VNIZ Tsellyd-Bumarh. Prom. No. 57, 204 (1970): cited inAbstr. Bull. Inst. Pap. Chem., 41.10268 (1971). (22) I. E. V’Yukov, V. N . Smorodin, and I. V. Gornostaeva, Sb. Tr. VNZZ Tsellyd-Bumath. Born., No. 57, 245 (1970); cited in Abstr. Bull. Inst. Pap. Chem., 41,10269 (1971).
Macromolecules
18 Glasser, Glasser Coniferyl alcohol monomer dehydrogenation
I
steering mechanism
reaction
R,,
Rb,
Table I Identilecation of Molecule Positions
Analytical literature data
species
R,, R,
“kinetics”
-
Dehydrogenation-polymer
I
5 6 7
I
7 CH I
+
OCH
100
/
Figure 2. Codes used in phenylpropane units.
made to simulate the chemical reactions that are involved in the formation and the reactions of lignin and to make predictions as to the chemical structure of this natural product. The objective of this investigation, therefore, is to present a simulation of the formation of lignin following presently accepted concepts of mechanisms operative in plants. This simulation is to result in a structural model of lignin that satisfies analytical requirements obtained with various methods by several investigators and described in the literature. The success of this simulation will be tested by simulating reactiops of lignin under a variety of conditions and comparing observed with predicted results. Since most studies with lignin have been carried out using fragments of lignin such as model compounds or mildly isolated lignin preparations comprising only small portions of the total material present in wood, results cannot be applied to lignin in general. How well certain fragments represent lignin can probably best be judged after applying results obtained with fragments to the polymer in a simulation model. By superimposing the results of different reactions, degradations, and isolations in a simulated lignin model, one might be able to discover discrepancies, interpret them, and even solve them. Simple addition of results obtained with fragments might not be representative for the macromolecule at all. Possibly, changes will have to be made in the concept of lignin, as much of the information on this polymer is based on the behavior of fragments. In addition to gaining information about the structure of lignin, the simulation model is expected to become a useful tool in assessing new reactions, determining theoretical yields and giving insight into reaction kinetics. 2. Principles of Simulation The mathematical simulation of the polymerization of coniferyl alcohol is schematically presented in Figure 1. Monomeric phenylpropane units ( C9 units) are converted
C H D (double bond) CHO CHz CHO-Sugu C-OCHs
12 13
linkages
8 CH
co
8 9 10 11
Content of functional groups and types of
9 CH,OH
C CH CH20H OH CH-OH 0 CHZOR
4
steering mechanism
Figure 1. Schematic outline of the principles of polymerization used in the SIMREL model.
Characterization
1 2 3
Relative frequency of forming and coupling radical species
polymerization
1
ID
into radical species by “dehydrogenation.” The polymer is subsequently formed by coupling reactions of radicals. The distribution of such reactions in the SIMREL model has been selected in accordance with analytical data published in the literature (cf. ref 3-5). The structural makeup of the polymer is monitored continuously, and the obtained analytical data are compared with the desired distribution of interunit bonds. Deviations of the two data sets are used to influence the frequencies of formation and types of coupling reactions of radicals in such a way that the deviations are eliminated. Monomeric phenylpropane units are coded with ten molecule positions, as depicted in Figure 2. Each molecule position is given an “identification” (ID) which characterizes the type of substitution in this position. Identifications are listed in Table I. Thus, a C g unit is described by molecule positions and identifications. 2.1. Basic Types of Coupling Reactions. The simulation of phenolic coupling reactions proceeds in accordance with established pathways. In Figure 3, the four presently recognized mesomeric forms of the coniferyl alcohol radical are shown which may lead to the formation of dimers by coupling. The theoretical number of combinations is 4z = 16. However, this number is reduced to only five different coupling patterns if all dimers resulting from radical type &, all “mirror images” (e.g., 8-10 us. 10-8), and the theoretical peroxide resulting from the coupling of two Ra radical species are disregarded. These five inter-unit bonds are as follows SARKANEN’S Code3 (3-0-4
5-0-4
0-0 P-5 5-5
SIMREL Code 8-10 5-10 8-8 5-8 5-5
Synonym 0-aryl-ether biphenyl ether pinoresinol
phenylcoumaran biphenyl
In the beginning stages of simulating lignin formation, radical & was neglected, as its coupling behavior is largely dependent upon side-chain substituents. It is suspected that the stabilization of a radical in position 1 of the phenylpropane monomer is suppressed in favor of other positions, particularly in 8 and 10. In comparison to Ra, Rb, and Rc it is expected that F& has a low rate of formation. Freudenberg expressed the opinion that coupling with position 1 will probably occur only with those units that have already been incorporated in oligomers or polym e r ~ Therefore, . ~ ~ & radicals were excluded from reacting as monomers in the simulation model, and its COUpling reactions were introduced only in the higher mers. Initially, polymerization was simulated by adding reactive sites of monomers to reactive sites of the growing (23) K . Freudenberg, K. Penzien, and H. Renner, Chem. Ber., 102, 1320 (1969).
Simulation of Reactions with Lignin by Computer
Vol. 7, No. 1, JanuaryFebruary 1974 CH,OH
I
CH
19
C
CH,OH
I
HC.
I
C
C
I c.. . .. .
C
I
(1
CH20H
,
CH-OH
I
I
\
0
R,.
OH
0
R‘.
Figure 3. Mesomeric radicals of coniferyl alcohol.
polymer through one of the five different coupling patterns resulting from Ra, &, and Rc. Sites were thereby defined as “reactive” if (1) there was a free phenolic hydroxyl group present, meaning position ten had identification four, and (2) position five and/or eight was still available for coupling. In order to prevent complete exhaustion of phenolic hydroxyl groups in the simulation model, through constantly reacting all reactive sites, three different levels of REACT M T I E S were introduced for molecule positions five, eight, and ten. These three were: REAC 0: not reactive, because the requirements described above are not satisfied; REAC 4: reactive; the requirements for reactive sites are satisfied, and coupling may occur; REAC 3: latent; the requirements for reactive sites are satisfied, but coupling cannot occur because the statistical distribution of coupling types or substituents in certain positions in the already formed polymer does not permit a reaction. Consequently each Cs uni.t in the simulation model was completely identified only with the codes MOLECULE POSITION (1-lo),IDENTIFICATION (1-13, expandable), and REACTIVITY (0,3, or 4). 2.2. Basic Coupling Reactions with Rd-Type Radicals (Side-Chain Displacement Reactions). The participation of Rd-type radicals in the formation of lignin has first been eluted from the observation of 1,2-diarylpropane der i v a t i v e ~ which , ~ ~ are apparently formed through coupling of & and Rb with simultaneous displacement of the side chain of &. Larsson and Miksche25 have since discovered that the displacement of side chains is restricted to molecules with benzyl alcohol groups. In addition, side-chain displacements had also been observed with compounds containing carbonyl groups in a: position.26 Other substituents in a: position seem to favor coupling in position 6 rather than 1 with no elimination of the side chain, as shown in Figure 4.25 ]Benzyl alcohol groups, however, are results of coupling reactions in p position where the intermediate quinone methide has added water. Such coupling reactions in p position may be p-0-4, (3-1, or in some instances also p-5 and p-p. consequently, displaced side chains are actually not lost, but they remain attached to (24) H.Nimz. Chem. Eer., 98,3160 (1965); ibid.,99,469, 2638 (1966). (25) S. Larsson and G. E. Miltsche, Acta Chem. Scand., 23,917 (1969) (26) J . C. Pew and W. J. Connors, J . Org. Chem., 34,585 (1969).
OCH,
Figure 4. Coupling of &-type radicals in dependence of substituents in the cy position (after Larsson and Miksche, ref 25).
other Cg units. This mechanism also implies that sidechain displacement reactions occur only after the &forming C9 unit has been incorporated in an oligomer or polymer; coniferyl alcohol with a radical in position 1 would apparently couple with its position 6 (see Figure 4), as was proposed by Larsson and M i k ~ c h e . ~ ~ The computer simulation of reactions of &-type radicals was, for the above reasons, limited to nonmonomeric units. A further restraint was introduced in the SIMREL program by requiring &-type radicals to be free of substituents in position 5 in order to guarantee a better mobility of the displaced aromatic rings. This restraint, in all probability, does not comply with lignin formation in viuo. Theoretically, Rd-type radicals can lead to nine different coupling modes as illustrated in Table 11. Only four of them have been o b ~ e r v e d ; ~ these ~ . ~ ’ are the 1-5, 1 - 0 - 4 , p-1, and 5-6 linkage. The 6-0-4 linkage has been searched for, but was not detected;27 it appears to be absent. Three other coupling modes, a 1-1link, a 6-6 link, and a 1-6 link are improbable, as they involve either double displacement or double rearrangement reactions of Rd-type radicals. One other linkage (p-6) must be considered likely, although it has not yet been detected. Thus, five different types of linkages were included in the simulation of coupling reactions with &-type radicals. These are the following SARKANEN’S Code 1-5 P-1
1-0-4 6-5 P-6
SIMREL Code 1-5 1-8 1-10 6-5 6-8
Synonym biphenyl 1,2-bis-guaiacylpropane biphenyl ether biphenyl 1,2-guaiacyl propane
2.3. Distribution of Basic Coupling Reactions. The structure of lignin as a statistical polymer built through random coupling reactions of four different mesomeric forms of one radical species depends obviously on the relative frequency with which each mesomer occurs. This frequency in t u m is a result of (a) how often a specific mesomer is formed and (b) how long it will survive under the conditions of lignification in vivo. Freudenberg described the average lifetime of a radical with 45 sec.28 The actual (27) S. Larsson and G. E. Miksche, Acta Chem. Scand., 23,3337 (1969). (28) K. Freudenberg, in ref 4, p 86.
20 Glasser, Glasser
Macromolecules
Table I1 Theoretically Possible Coupling Reactions of Rd-Type Radicals
6
OCH
0 OH
OH
I
R
=
-CH-C-C
It
I
4s
-(’H-C-C
R HO H4
O 0
C
H
OCH
OCH,
CH $0
R I H O CH 0
P O CH 0
@OCH 0
F CH 0
1-0.4
’\
6H 6.04
I R
CH-OH
f
I
on
OH
OH
16
6-6
OCH,
0 R
-I
CH-OH
CH,O
OCH
OCH,
OH
1-1
1-G
C C.
i
q
I
I
c
C CH,O
O
H
OCH,
J=+
OCH,
C CH,O J + O H
OCH;
0
OH 8-1
OH 8.6
structure of lignin suggests that the p position of a Cg unit is a favored position for radical formation since almosk all Cb atoms are involved in inter-unit bonds. The assignment of numerical values to express the relative abundance of each mesomeric species would be a prerequisite for a kinetically correct simulation model. However, some doubts would remain even if such values could be developed because of differences in the coupling rate of coniferyl alcohol dehydrogenated in the Zulauf us. the Zutropf procedure, as pointed out in detail by Sarkanen.Zs Therefore, the simulation of lignin formation has to be based upon some other mechanism in order to determine the relative rate and consequently distribution of basic cou. pling reactions. In the SIMREL model this problem was solved by prescribing “distributions of coupling reactions” and allowing “reaction margins.” The “distribution of coupling reactions’’ consisted thereby of a weighted mixture of all possible inter-unit bonds, and “reaction margins” were percentages of the total number of units by which fixed distributions were allowed to deviate if the propagation of simulated lignin formation depended on it. This method ( 2 9 ) K. V . Sarkanen, in ref 3, pp 146, 150
which emphasizes the result of lignin formation rather than the kinetic forces involved, seemed to guarantee best an even distribution of coupling types throughout the development of the polymer. This concept is based on the idea that lignin-particularly that from softwood species-shows only limited variation in its structure. The distribution of basic coupling reactions was determined in accordance with data obtained from the literature, and compiled in Table 111. Thus, 20% of all coupling reactions was based on &-type radicals with three-fourths of them occurring with benzyl alcohol or a-carbonyl units (C,-R, R = OH, R = 0),and one-fourth with units having other than benzyl or carbonyl constituents. The majority of coupling reactions in position one occurs with Rb-type radicals, so that about 10% of all linkages in the lignin model will be of the 6-1 type. Biphenyl ether linkages were increased over most previously reported data to 10% on the basis of recent results by Pew, Connors, and Kunishi30 and Larsson and M i k ~ c h e Most . ~ ~ other data were kept in close agreement with data obtained from the literature. The simulation of lignin formation was carried out by adding monomeric coniferyl alcohol to some reactive site in the polymer. For this purpose, the polymer was searched by the program for a REAC 4 position of the unit with the lowest possible unit number. If a unit was found with one REAC 4 position-for example, in molecule position 5-the addition of the monomer to this position was attempted with any coupling reaction involving position 5 . This could be either a 5-5, a 5-8, or a 5-10 reaction. The random choice between these different possibilities was restricted by imposing an analytical templet on the model in form of the “distribution-of-coupling-reactions” value. Thus, for example, a 5-5 reaction was allowed if the total number of 5-5 links which would be present after the reaction was less than a preset limit. If this limit were exceeded the reaction was not permitted, and another reaction with position 5 , namely, 5-8 or 5-10, would be tested. If none of these reactions was feasible, the REAC 4 of the unit in the polymer was converted to a REAC 3 and the simulation was continued with the renewed search for a REAC 4 position. In the event a unit with two or three REAC 4 positions was found, one of the reactive sites was chosen randomly over the others. By this method molecule positions 5, 8, and 10 were equivalent and had the same chance of reacting. In essence, the simulation of the lignin formation was based on the result of the unknown selection mechanism prevailing in nature, rather than its driving kinetic force. The result was decided to be the analytical distribution of linkages listed in Table 111, plus whatever margin was required to permit the simulation of a lignin molecule with 75 or more units corresponding to a molecular weight of about 14,000. Side-chain displacement reactions which were defined to be all reactions resulting from Rd-type radicals, independently of whether the sidechain is actually displaced or not, were simulated immediately following the attachment of monomeric Cg units. Such a reaction was initiated, if, at a certain point of polymerization, it was found that less than 20% of all reactions could be classified as side-chain displacement reactions. Any of the five possible reactions was chosen randomly as long as the analytical requirements of Table I11 were satisfied. A reaction of molecule position 1 required the presence of a hydroxyl or carbonyl group in a position (ID 7 or 8 in position 7 ) . If such a reaction was required by the analytical templet, but was not possible because the desired substituent was (30)J. C. Pew, W. J. Connors, and A . Kunishi, Chim. Biochim. Lignine Cellul. Hemicellul. Actes Symp. Int., 229 (1964). (31) S. Larsson and G. E. Miksche, Acta Chem. S c a d . , 21, 1970 (1967).
Simulation of Reactions with Lignin by Computer 21
Vol. 7,No.1, January-February 1974
I CH
I
1
CH
OCH,
0
CH.OH
CH
I
1I
0 1 H FI’ ‘CHI
I I
I
CH
CH
bo”‘
“-4 I
3 4
OH
I CH.
CH=CH
CH.OH
0
CH,OH
CH,OH
HC-CH
.CH CH ,O
CH
CH ,OH
n
I
-O .C H ,[
I
CH=CH CH,OH
\ \
OH
OH
R-CH
CH 0
4 I
CH
’’
CH-R
I
I I
I O
C H { O Y 4 OH
\/
OCH 1 4
1
H C L ,CH?
OH
OH I1
R = any benzyl substitution
Figure 5.6-5 coupling.
I
CH -CH
I
HC
-CH
HC,
/C-0
I
/o
1
I O
Hx
not available in the t~ position, the reaction was deferred until such a substitution at molecule position 7 appeared. In addition to directly limiting the types of reactions that could occur in the formation of lignin, some other restrictions were imposed that indirectly limited certain types of coupling reactions. These were reactions that involved the formation of such end groups as coniferyl alcohols, aldehydes, and phenolic hydroxyls. Considering that all coupling reactions involving monomeric Ra radicals, as well as most couplings of Rb with Rc, would automatically lead to the irreversible formation of a coniferyl alcohol or addehyde end group, it is surprising that softwood lignins contain only 9% of such groups.32 In the SIMREL model all reactions producing such end groups by depriving a unit of its reactive phenolic hydroxyl groups were restricted to less than 9%. On the other hand, the survival of about 25% of phenolic hydroxyl groups is guaranteed through assigning “latent reactivities” (REAC 3) to phenolic positions. Obviously, nature possesses a n extremely sensitive mechanism leading to a (random) polyphenol that exhibits little structural variations. As already recognized by S a r k a n e ~ ~exclusive ,~~ 8-0-4 coupling will form chain polymers without phenolic and coniferyl endgroups (except for one at each end), and exclusive 8-8 and 13-5 reactions will form polymers with plenty of such end groups. And yet lignins seem to exhibit little variation in their overall distribution of types of linkages and end groups. The simulation of formation of lignin described so far has been based on reactions of monomeric radicals of coniferyl alcohol, exclusively. This simplification is unrealistic in view of experiments by Freudenberg and his group that have also been discussed by SarkanemZgThese experiments showed that the in vitro formation of oligomers only follows the conversion of 80% of coniferyl alcohol monomers to dimers, suggesting that polymers are formed only after dimerization is virtually complete and monomers are depleted. The prevalence of 80% dimers a t some stage of dehydrogenation makes the participation of monomers in the polymerization questionable. However, the addition of oligomers instead of monomers to the growing
polymer will require refinement of the present simulation model. 2.4. Simulation Mechanics. After the decision has been made how to attach a Cg unit to the polymer, some molecular changes are carried out on both reaction partners. These changes concern “identifications” as well as “reactivities” of certain molecule positions in the units. These consequences of coupling reactions are discussed in the following and are listed in Table IV for the example of a 6-5 coupling reaction. 5-5 Coupling. Biphenyl structures, resulting from the coupling of two R, radicals, are subject to the smallest number of constraints, as they do not contribute to the irreversible formation of coniferyl end groups or to the depletion of phenolic hydroxyls. The formation of 5-5 and 4 - 0 - 5 linkages is reported to occur predominantly with higher oligomers rather than with monomers.33 However, no such restriction was added in the SIMREL program. 0-5 Coupling. The coupling of Rb- with &-type radicals results in most instances (-75%) with internal ring closure through addition of the phenolic hydroxyl group to the quinone methide intermediate (I, Figure 5 ) . However, in some cases the quinone methide stabilizes through addition of other hydroxyl-containing compounds, as indicated in the case of I1 (Figure 5 ) . The relative proportion of one or the other stabilization pathway was derived from data available in the literature and reviewed by Sarkanen.32 In the instance of open @-5derivatives, Sarkanen’s Rydholm diagram was referred to in ref 32. p-6 Coupling. Pinoresinol (111) is the major result of coupling of two Rb radicals as shown in Figure 6. Additional minor dimerization products are IV and V (Figure 6) which result from Rb-type coupling of ferulic acid in
(32) Y Z . Lai and K . V. Sarkanen. in ref 3, pp 195,227
(33) K. Freudenberg and K. C. Renner, Chem. Ber., 98,1879 (1965).
CH30
v
OH IV
R
= any benzyl substitution
Figure 6. p-/3 coupling.
22 Glasser, Glasser
Macromolecules CH,OH
CH.OH
II
CH.
4 I
CH ,O
0 Ri,
CH.O sugar O/CH,
CH-OH
I
I
4-
CH-CH,OH
I
CH
@
II
FH
OCH
OH
-
CH20H
CH-CH,OH
I
sugar
\
OCH
0.
CH-CHPH
1
CH-0
CH -CH
OH
CH ,O
OH
CH,OH
C-0
CH,O
I
(x
CH-CH,OH
OCH ,
OH
Figure 7. p - 0 - 4 coupling.
the case of IV, and C,-Cs condensation in the case of V. The remaining quinone methide intermediate of V stabilizes again through addition of a hydroxyl containing compound like water, sugar (or carbohydrate), or phenol. The relative distribution of the three optional derivatives was again derived from Sarkanen’s Rydholm diagram.32 5-0-4 Coupling. The long neglected coupling of Ra and Rb radicals leading to biphenyl ether structures is subject to end-group constraints, as it will establish coniferyl endgroups if the Ra-type radical couples with a REAC 4 in position 8. Recent experiments by Pew and coworkers have shown that 5-0-4-type coupling patterns are almost as frequent as 5-5 couplings.30 The same is indicated by oxidation studies of Larsson and M i k ~ c h e . ~ ’ p-0-4 Coupling. The coupling of R, and Rb radicals leads to @-arylether derivatives which have been recognized as the most representative dimeric structures in lignin. The quinone methide intermediate which is formed initially becomes stabilized through the addition of some hydroxyl containing moiety as water, sugar, or phenol, as is indicated in Figure 7. A number of additional constraints are imposed on 0-0-4 coupling reactions as these reactions may establish end groups and consume phenolic hydroxyl groups. As a consequence, coupling of Ra-type radicals must be favored only after some side-chain substitution has previously occurred. Therefore, most pethers will result from the addition of Rb-type monomeric radicals to Ra-type oligo- or polymeric radicals; otherwise, a much greater number of coniferyl end groups will be formed. Coupling Reactions Based on Rd-Type Radicals. Reactions resulting from &-type radicals involve approximately 20% of all unitsS2 with three-fourths of them causing the displacement of side chains, and the rest of the
coupling reactions taking place in position 6 of the aromatic ring. Displacement reactions initially require the presence of benzyl alcohol or carbonyl groups, leaving the a position after the displacement reaction as aldehyde group or unconjugated carbonyl group detached from the aromatic ring. However, the displaced side chain will not be lost, but remains attached to whatever coupling counterpart in the p position that had led to the secondary formation of a benzyl alcohol or carbonyl function. Generally, coupling reactions in position 1 will take place with positions 5 , 8, and 10 (corresponding to Rc-, Rb-, and Ratype radicals), whereas position 6 will probably become attached to position 5 or 8. p-1 Coupling. The most abundant and longest known result of coupling with Rd-type radicals is the 1,2-diarylpropane derivative which is reported to be present in lignin to the extent of about 0.10-0.15/c9 unit.32 Upon coupling, the quinone methide intermediate will stabilize much in the same fashion as the 0-0-4 dimer by addition of some hydroxyl-containing compound. 1-5 and 1-0-4 Couplings. Coupling reactions of position 1 with Ra- and Rc-type radicals to form 1-0-4 and 1-5 inter-unit bonds seem much less pronounced than p-1type linkages. Only about one unit each in a polymer of 50 units participates in these types of reactions (cf. ref 27). 6-6 and 6-5 Reactions. The only recently discovered coupling reactions of Rd-type radicals with units lagging ketonic or hydroxylic benzyl carbon atoms lead to condensations at the CS atoms of the aromatic rings. In the SIMREL model, only 2-3% of each are assumed to occur. As in all other coupling mechanisms of the /3 position, the p-6 linkage may be responsible for the formation of a benzyl alcohol and, following oxidation, an a-carbonyl group or a lignin-carbohydrate bond or a noncyclic benzyl-aryl ether linkage. Carbonyl groups have received much attention in lignin ~ h e m i s t r y . ~They ~-~~ can be formed through either disproportionation reactions occurring in cell structures38 or through aut~xidation.~S Three types of carbonyl groups are presently assumed to exist in lignins (cf. ref 3 ) . These are coniferyl aldehyde end groups, conjugated ( a ) ,and isolated carbonyl groups. The presence of other aldehyde groups, such as in glyceraldehyde @-arylether structures40 is questionable. Also, the possibility of cinnamic acid end groups cannot be ruled o ~ t . However, ~ ~ , ~such ~ groups must be considered rarities and can probably be neglected in structural schemes such as the one presented here. Coniferyl aldehyde groups are simulated through simple conversion of identification 3 into ID 10 in position 9 of coniferyl end groups. A similar process generates conjugated carbonyl groups from benzyl alcohol groups in units with ID 5 in position 7 . Isolated carbonyl groups are assumed to be the result of displaced side chains in the form of aldehyde end groups.32 These groups are now given responsibility for the observation that borohydride-reduced milled wood lignin is more difficult to sulfonate than genuine lignin.42 Presumably, the presence of the aldehyde group is necessary to activate the primary hydroxyl group of the C, atom of the displaced side chain to form a car(34) E. Adler and J . Marton, Acta Chem. Scand., 13,75,357 (1961). (35) E. Adler, J . Marton, and K. I. Person, Acta Chem. Scand., 15, 384 (1961). (36) J. Marton and E. Adler, Tappi, 46,92 (1963). (37) E. Adler, Pap. Puu, 43,634 (1961). (38) K. Freudenberg and N . Schluter, Chem. Ber., 88,617 (1955). (39) K. Freudenbergand B. Lehmann, Chem. Ber., 93,1354 (1960). (40) K. Lundquist, G. E. Miksche, L. Ericsson, and L. Berndtson, TetrahedronLett., 46,4587 (1967). (41) T. M. Vasil’eva, R. K . Boyarskaya, and G . P. Grigor’ev, Zh. Prikl. Khim. (Leningrad), 41, 603 (1968); cited in Abstr. Bull. Inst. Pap. Chem., 40,4693 (1970). (42) D. W . Glennie, in ref 3, p 627.
Vol. 7, No. I , January-February 1974
Simulation of Reactions with Lignin by Computer 23 Table 111 Distribution of Basic Coupling Reactions
Sarkanen (Ref 3)
Freudenberg (Ref 4)
P-0-4
0 . 4Eia
P-P P-5 5-5 4-0-5
0. 17u 0 . 14a 0.25b 0 . 05bic
0.44' 0.22f 0.090 0.25b 0.06e
Side-chain "displacements" Total
0.20d
Guidelines
Harkin (Ref 5)