Structure of the HexA-derived chromophores
The CRI procedure was applied to Eucalyptus pulp with high glucuronoxylan content, which had been aged before at elevated temperature in dry environment. This protocol, applied to 2.8 kg of pulp, afforded a mixture of chromophores from which five different, well-defined compounds were isolated (structures shown in Scheme 2). All of the five compounds are strongly colored and thus potent chromophores. The five compounds account for 92% of the isolated chromophoric matter, the residual 8% consisting of an intractable black remainder. When the pulp was pre-aged before chromophore isolation in a humid environment, but under otherwise identical conditions, the same number of chromophores was found, and their sum amounted to 86% of the total chromophore mixture.
In a next set of experiments, the HexA model compound methyl 4-deoxy-β-L-threo-hex-4-enopyranosiduronic acid (1)Footnote 3 was used as the starting material and was subject to thermal and acidic stress—under conditions similar to industrial acidic washing stages. The black precipitate obtained was separated by column chromatography and provided a mixture of five well-defined compounds that, most notably, were identical to those obtained from the pulp. In the model compound case, the combined mass of the five chromophoric compounds amounted almost to the whole mass of aging products, nearly 95%. The similarity of the compounds to those isolated from pulp was readily proven by identical retention times in thin layer chromatography (TLC), high-performance chromatography (HPLC), using toluene with 10% ethanol as the eluent in both cases, and by HR-MS. We were thus able to conclude that, first, model compound 1 was indeed a quite fitting model for the HexA unit in pulp and, second, it produced the same chromophoric compounds as the real-world pulp.
However, determination of the chemical structures of the compounds was far from being trivial. While mass spectrometry allowed deriving a chemical formula that corresponded with that derived from microanalysis (see experimental), fragmentation patterns were largely inconclusive. Evidently, all compounds contained no other elements than carbon, hydrogen, and oxygen. The molecular formulae derived indicated structures containing 10 (4), 12 (5), 15 (6), 18 (7), and 22 (8) carbon atoms with rather low C/H ratios of 2.5 (4,6,7), 3 (5) and even 5.5 (8). This was clearly indicative of highly condensed structures which were in addition moderately oxygenated (C/O ratios between 2 and 2.5). While not much information was expected from the 1H NMR spectra because of the low H content, also most 13C spectra NMR spectra were disappointingly void of indicative resonances, although well resolved. Except for 6, which showed the expected number of 15 carbons, all other compounds displayed peaks only for half the number of contained carbons. Evidently, those compounds were symmetric with a two-fold symmetry element and exclusively contained pairs of magnetically equivalent carbons. Moreover, all peaks were rather sharp and well-resolved so that inclusion compounds or valence isomerism were excluded. The opposite case was observed for compounds contained as key chromophores in pulps, such as 2,5-dihydroxy-1,4-benzoquinones (DHBQ) (Hosoya et al. 2013a) or 5,8-dihydroxy-1,4-naphthoquinone (Hosoya et al. 2013b), which are highly resonance-stabilized moieties and hence give rather broad resonances in 13C NMR. To conclude, we failed to derive the chemical structure of the five isolated compounds by a combination of the conventional analytical techniques (NMR, MS, IR, microanalysis).
Fortunately, all five compounds exhibited rather low solubilities in all kinds of organic solvents, especially polar ones, and a high tendency to form crystals suitable for crystal structure determination. For crystallization, the compounds were dissolved in hot glacial acetic acid and slowly cooled down to form large and well-shaped crystals. Recrystallization from both chloroform and toluene was also possible, but provided smaller specimens along with amorphous precipitates. The crystal structures of the five compounds are shown in Scheme 3, their chemical structures thus being established. The overlay with the drawn chemical formulae (Scheme 3, lower two rows) illustrates in a simple way an interesting aspect of the structure of the compounds: they are all completely planar. Almost all atoms lie perfectly well in the molecular plane, which renders the deviations to the drawn formula—of which the atoms also all lie in one plane, the paper plane—rather small. The torsion angles of 180 °C between the anellated rings confirm this simple illustration in a more “scientific” way. Such “two-dimensional” molecular arrangements are known, for instance, for graphene (Geim and Novoselov 2007), or polycylic aromatic hydrocarbons (PAH) (Rieger and Muellen 2010) without substituents—even a substituent as small as a methyl group (or more precisely its hydrogens) would protrude from the molecular plane and add a “third dimension” to the molecule.
Properties, structure comparison, and reactivity of the chromophores
The degradation products derived from HexA are very potent chromophores. One molar solutions of all chromophores appear to be black; 10 µM solutions are sufficiently dilute so that color differences can be seen. Even at this relatively low concentration, compounds 7 and 8 are still so strongly absorbing that their solutions appear nearly black. A 10 µM equimolar mixture of the five compounds 4–8 appears black, while a 1 nM (!) solution has a distinctly yellow overall color (see Fig. 1). This concentration translates into about 0.2 ppb of each compound—this is roughly the realistic concentration range for commercial pulps. This simple experiment helps to illustrate why the identification of chromophores in pulps is such a tricky endeavor. The concentrations of the individual compounds are extremely low, but they are still very well perceptible to the human eye due to their high absorption coefficient, especially in the yellow range for which the human eye is particularly sensitive.
Compound 4 is a known compound—it has been reported as an intermediate in the synthesis of organic conductors (Takahashi and Kobayashi 2000). The other four compounds have not been described so far and represent novel structures. With the exception of 6, all compounds are symmetric with an inversion center I. This corresponds to the NMR data showing only “half the number” of carbon resonances for those four molecules, and to the mass spectra being poor in informative fragmentation mass peaks. The inversion center is evidently located in the geometric middle of the central 1,4-benzoquinone moiety in these molecules.
From the viewpoint of formal structure, compound 4 can be seen as a basic motif from which the other four compounds are derived: 5 has two carboxyl substituents in addition, 7 two anellated p-hydroxybenzoic acid moieties, and 8 two [2,3-f]-anellated benzofuran units. In all of these compounds the additional substituents (relative to 4) are introduced in a symmetric way, so that the inversion symmetry is maintained. Only in compound 6, just one p-hy-droxy-benzoic acid substituent is appended onto 4, breaking the symmetry; 6 is the only non-symmetric compound among the isolated chromophores, with each of its 15 carbons giving an individual resonance in 13C NMR since no magnetic equivalence due to symmetry reasons is prevailing.
The compounds are ladder-type oligomers of mixed furanoid/benzoid and mixed aromatic/quinoid nature. In all five chromophores, there are six-membered rings, either p-benzoquinone units or p-hydroxybenzoic acid units, anellated to (= fused with) furan units, i.e. the six-membered rings are amended with ethenyloxy units. By this anellation (lat. anellus = little ring), a C = C double bond is at the same time part of the furan unit and the p-benzoquinone unit (or p-hydroxybenzoic acid, respectively). The furan-fused benzoquinones can also be conceived as furan moieties linked through carbonyl bridges; alternatively, they can be seen as partially oxidized phenolic units that are at the same time linked through ortho-positioned aryl-O-aryl and aryl–aryl carbon bonds, or as 2,5-dihydroxy-1,4-benzoquinone units of which the hydroxyl groups are incorporated into fused heterocyclic aromatics (furan units) or etherified with phenol derivatives (2-hydroxy group of 2,4-dihy-droxy-benzoic acid). It is interesting that every second furan unit is flipped by 180° so that the two oxygens of the furans constitute at the same time two para-configured oxygen-substituents of the connecting six-membered structures. The compounds seem to have a pronounced Janus character with regard to contained structural elements, and these ambiguities in the structural description already reflect the fact that the electronic structure will be rather complex and cannot be described in simple terms of (anti)aromaticity or substituent effects. The UV/Vis and fluorescence properties of the chromophores are currently studied, along with a computational treatment of their electronic structures. Preliminary results indicate interesting semiconducting properties of the parent compound and predict the derived (radical)-ions to be very stable or even persistent, both being interesting features for organic conductors. This indeed would be a completely new and unexpected facet of the unwanted “black sludge” from HexA degradation. This matter will be discussed in an upcoming account.
It is important to note that the structure of the compounds as identified does not agree with linear furan polymers as postulated in the literature (Gopalakrishnan et al. 2014; Gandini 2008, 2011; Gandini et al. 2008, 2009). Hitherto it was assumed that furfural and other furan derivatives link in a way that furan units are bridged by C1 links and form relatively flexible systems with extended conjugation that accounts for their color.
The structure determination of the chromophores might have yet another important implication. In biorefinery scenarios, hexoses and pentoses are often converted into furan derivatives that are to be further used as chemical building blocks. Formation of furfural from xylose/xylan and 5-hydroxymethylfurfural (HMF) from glucose/cellulose are the most well-known examples. These conversions are always accompanied by formation of black condensation products that decrease the yield and cause problems in separation and purification. These dark by-products are usually denoted as “humins”—perhaps indicating similar color and complexity, but being structurally completely different from humic matter in soil. Such humins are formed by side reactions of mono-substituted or disubstituted furans, exactly as the HexA-derived chromophores were. Without having done in-depth structure determination of humins formed upon furfural or HMF production, we would like to point out the likelihood that those humins have very similar structures to the chromophores derived from HexA, thus being similar ladder-type oligomers of mixed aromatic and quinoid characters and mixed five-membered and six-membered structural building blocks.
The formation mechanism of the chromophores from HexA seemed quite confusing at first glance. Since 2,5-dihydroxy-[1,4]-benzoquinone (DHBQ) was found to be a fundamental structural unit of chromophores in cellulosics (Hosoya et al. 2013a), it was logical to assume it to be a key element also in the HexA-derived chromophores, especially as HexA and DHBQ have the same number of carbon atoms. However, relative to DHBQ, compound 4 is distinguished by two additional ethenyl groups, and compound 5 by two additional propenoic acid motifs (see Scheme 4). The build-up of chromophores from such small fragments had been observed in the case of cellulose-derived chromophores, and thus seemed likely also for the HexA case. In 6 and 7, even one and two, respectively, whole p-hydroxybenzoic acid units are “added” onto 4. The origin of such additional C2-, C3- or aromatic units remained puzzling for quite some time, especially the question why attachment of these small fragments to the central benzoquinone unit proceeded largely in a highly symmetric way. In the end, the formation mechanism could only be clarified by means of isotopically labeled HexA. For this purpose, six different 13C-isotopomers of HexA model compound 1 had been prepared (1a-1f) as described in the preceding part of this series.
The solution as to the formation mechanism was as simple as it was plausible. From a retrosynthetic point of view, the “trick” was to dissect the molecule not into p-benzoquinoid structures, but rather into furanoid fragments (Scheme 4). This way, compound 4 is composed of two 2-furanoyl (2-carbonylfuran) units that are linked in a symmetric way so that the carbonyl group of the first unit is linked to C-3 of the second, and the carbonyl group of the second linked to C-3 of the first furan. Compound 5 would be constructed in a similar way, with two 2-furanoyl-5-carboxylic acid units being symmetrically joined through carbonyl-C3 bonds. The two 2-furanoyl units making up 4 are two furan-2-carboxylic acid (3) units that are completely unchanged (apart from the two bonds linking them and a hydroxyl loss at the COOH group). The same applies to the 2-furanoyl-5-carboxylic acid units in 5: they represent almost unchanged 5-formylfuran-2-carboxylic acid (2) precursors.
How can now 13C-labeling help to support this formation mechanism of the chromophores? The 13C-isotopomers 1a-1f carry one carbon atom each with 99 + % 13C isotopic abundance. This atom remains like a marker throughout the chemical processes during chromophore formation. For instance, methyl 2-13C-4-deoxy-β-L-threo-hex-4-enopyranosid-uronic acid (1b) translates into the chromophore 4b—or methyl 4-13C-4-deoxy-β-L-threo-hex-4-enopyranosiduronic acid (1d) into 4d in an analogous way (Scheme 5). In each case, it was necessary, of course, to separate the respective chromophores (4, 4b-4f) from the other chromophores (5-8) that are formed alongside. The 13C NMR spectra of the 13C-labeled products show one dominant resonance each—the carbon position carrying the label—while the other carbons are more or less “blanked out” by their roughly 100-fold lower sensitivity. The fact that only one prominent resonance is seen proves that HexA was neatly converted into furanoid intermediates that combine in the presented way into the symmetrical compound 4. This symmetry renders the two corresponding 13C-labels magnetically equivalent so that they produce only one peak.
A final proof of the linking and its positioning was provided by the following experiment: a mixture of two differently isotopically labeled HexA model compounds, namely methyl 5-13C-4-deoxy-β-L-threo-hex-4-enopyranosiduronic acid (1e) and methyl 6-13C-4-deoxy-β-L-threo-hex-4-enopyranosiduronic acid (1f) was converted into the corresponding chromophores. In the starting mixture (1e/1f) as well as in the mixture of furanoid intermediates (2e/2f and 3e/3f) only singlets appear for the labeled position. In the chromophoric products, however, where two furanoyl building blocks combine, C-4 and C-5 (of the former HexA motif) form a C–C bond. This is reflected by the occurrence of homonuclear C–C coupling with 1
C,C coupling constants of about 65 Hz; thus the two labeled carbons do not appear as two singlets any longer, but as two doublets (see schematic drawing of the spectra in Scheme 5). In fact, the experiments, starting from a 1:1 mixture of 1d and 1f, provided a mixture of the isotopomers 4d, 4f and 4d/f close to the theoretical ratio 1:1:2, which was similarly true for 5d, 5f and 5d/f. In Schemes 5 and 6 the labeled carbon positions are indicated by red dots. These experiments confirmed the linkage patterns of the 2-furanoyl units in chromophores 4 (Scheme 5) and 5 (Scheme 6).
One interesting aspect of compound 5 was that the carboxyl groups of the former HexA units (C6) are engaged in the bonds making up the central p-benzoquinone unit (Scheme 6). The two “exo”-carboxyl groups, in turn, actually represent the former C1-carbons in HexA, i.e. formyl (aldehyde) groups. However, in all isolated chromophores (and also in spectra of chromophore mixtures) we have never detected any aldehyde functions, which are readily discernible in NMR spectra by proton resonances >9 ppm and carbon resonances >180 ppm. Evidently, the aldehyde groups have been oxidized to the corresponding carboxylic acids quite rapidly. An experiment in an inert atmosphere could verify that this oxidation is—at least in part—an autoxidation by oxygen: the chromophore mixtures initially contained aldehydes which had disappeared after 1 h in contact with air. But neither was it examined to which chromophores the unstable aldehydes belonged nor whether an additional redox process with an oxidizing intermediate was involved. Since the labeling experiments were able to unambiguously establish the position and linkages of the 2-furanoyl building blocks the minor issue of the oxidant in the C1-oxidation was not further followed.
In a similar way, the connections of building blocks were studied for chromophore 8 (see Scheme 7). HexA labeled in position 2 (methyl 2-13C-4-deoxy-β-L-threo-hex-4-enopyranosiduronic acid, 1b) produced two prominent singlets in the 13C product spectrum coming from 2 × 2 magnetically equivalent carbons, while mixtures of 1a/1d, 1c/1f and 1d/1f were used to confirm the three different inter-furanoyl linkages in the molecule (Scheme 7), again by the occurrence of two doublets in the NMR spectra caused by the homonuclear 1
C,C coupling of the directly linked 13C-labeled carbons.
Analysis of the formation of chromophores 6 and 7 was somewhat more complicated: in these compounds, benzene rings (p-hydroxybenzoic acid moieties) were present for which an analogous formation simply by condensation of furanoyl moieties could not be easily imagined. A similar approach as used for the linkage and formation analysis of the other chromophores—based on labeled HexA precursor and position analysis of the 13C-labeles in the products—provided the labels as shown in Scheme 8. Evidently, 6 consisted of an inner “core” of compound 4, to which one furanoyl (C5) unit was appended to eventually give the anellated p-hydroxybenzoate unit in 6. Compound 7 was its binary counterpart, formed by anellation of two p-hydroxybenzoic acid units to the central compound 4 in a symmetric way. This way, 6 would be an intermediate in the formation of 7, in which only one “side” of precursor 4 has reacted under p-hydroxybenzoate formation, while in 7 both “sides” had been involved in such an anellation. Compound 6 is thus made up of three furanoyl motifs, compound 7 of four. The linkage analysis for compound 6 is shown in Scheme 8, analysis for compound 7 was done analogously.
The assumption that 6 and 7 are derived from parent compound 4 offered at the same time a plausible assumption as to the mechanism. As furan units have limited aromaticity and can be regarded as dienes (McKillip and Sherman 1980; Hoydonckx et al. 2012; Takebayashi et al. 1983), they readily undergo Diels–Alder reactions, both as dienes and as dienophiles. In the most simple case, a furan derivative reacts “with itself” by assuming the role of ene and diene at the same time. That such a process is possible at all demonstrates that the energetic disfavoring by the loss of aromaticity is of little importance and readily overcompensated by entropic favoring (Gribble and Gilchrist 1998; Scaiano et al. 1988; Isidor et al. 1973). For compound 6, we can assume a mechanism shown in Scheme 9, with compound 4 undergoing a Diels–Alder-type process with 5-formyl-furan-2-carboxylic acid (2) reacting as the corresponding diene. Intermediate I
can be isolated when starting from a 1:1 mixture of 2 and 4. The regioselectivity of the diene approaching the dienophile in the [4 + 2]-cycloaddition is determined by the hydrogen bonding between the carboxyl group to the furan ring’s oxygen. Note that the intermediate has no endo-/exo-differences since 4 is planar and both approachable faces are equivalent. To form compound 7, such sequence would proceed on both “sides” of precursor 4, not just at one side only as it happens in the formation of 6.
The oxygen-bridged intermediate (I
in Scheme 9) would open up to eventually give a p-hydroxybenzoic acid motif, as indeed observed in 6 and 7. Detailed mechanistic investigations—which will be communicated in a separate account—have shown 5-formylfuran-2-carboxylic acid (2) to be the actual co-reactant of core chromophore 4 and thus precursor of 6 and 7. Furan-2-carboxylic acid (3) reacts in a similar Diels–Alder mode with 4, but the final products have anellated benzoic acid (not p-hydroxybenzoic acid) units and are thus missing the hydroxyl groups of 6 and 7 (they can be imagined as deoxy-6 and bisdeoxy-7). From the Diels–Alder intermediate I
, opening of the oxygen bridge is the first step that proceeds regioselectively, governed by the higher stability of the intermediate tertiary carbenium ion in α-position to the carboxyl function, followed by loss of the neighboring proton. The formyl group is lost in an unconventional process similar to the deformylation described in the preceding paper. Why the formyl substituent is eliminated—which would be expected to be reasonably stable—and not the hydroxyl group—which on the other hand would be reckoned to be rather elimination-prone—will be discussed in the above-mentioned treatise, also in connection with computational results. The final step is an oxidation of the cyclohexenone (hydroxycyclohexadiene) product driven by rearomatization, to establish the p-hydroxybenzoic acid moieties found in chromophores 6 and 7.
With the linkage analysis by means of isotopic labeling, all chromophores 4-8 were shown to consist of non-fragmented and largely unchanged furanoyl units (Schemes 5, 6, 7, 8, 9). It should be noted that this is in stark contrast to the chromophores from (over-)oxidized celluloses, such as 2,5-dihydroxy-1,4-benzoquinone or 5,6-dihydroxy-1,4-naphthoquinone, which are formed by complex degradation of oxidized units in cellulose to C2–C4 units and recondensation of such small fragments. In the HexA-derived chromophores, the furanoyl degradation intermediates 2-furancarboxylic acid (3) and 5-formylfuran-2-carboxylic acid (2) are linked together and undergo follow-up transformations, but they remain largely intact without further degradation. They build up the chromophores like whole bricks a wall.
One particular feature of the chemistry of all five chromophores is their ease of reduction, not only to the respective hydroquinones—this would be quite expectable for [1,4]-benzoquinones—but even further to the corresponding hydrocarbons, i.e. under loss of the oxygen functions, the resulting structures being shown in Scheme 10. The classical Clemmensen reduction (Clemmensen 1913), a method to reduce aldehydes and ketones to the corresponding alkanes in acidic media that usually uses (amalgamated) zinc and hydrochloric acid gives excellent yields above 90% in quite short reaction times. Conventionally, it gives good yields for dialkyl ketones and alkylaryl ketones, but it is rather problematic for diaryl ketones. The chromophores evidently represent a special case here where yields are unusually high. Interestingly, also reduction under alkaline conditions, for instance the standard Wolff–Kishner reduction (Szmant 1968; Wolff 1912), carried out with solid KOH and hydrazine in diethylene glycol, proceeds neatly with similarly high yields. It is rather unusual that a ketone can be reduced well to the alkane under both acidic and alkaline conditions; normally a clear preference for one type of medium is found. The reason for the particular ease and neatness of the reduction of the chromophores is evidently the high stability of reduction intermediates (both carbanions and carbenium ions) brought about by the extensive conjugation, and the extended aromaticity and exceptional stabilities of the resulting systems, which resemble those of polycyclic aromatic hydrocarbons (Mackay and Callcott 1998; Fetzer 2000; Harvey 1998). The redox behavior of the HexA-derived chromophores, along with the analytical data of the products 4R–8R (Scheme 10) will be discussed in detail in an upcoming account.