DOI: 10.1002/cssc.201300509 Methanol Fractionation of Softwood Kraft Lignin: Impact on the Lignin Properties Tomonori Saito,*[a] Joshua H. Perkins,[b] Frederic Vautard,[b] Harry M. Meyer,[b] Jamie M. Messman,[c] Balazs Tolnai,[d] and Amit K. Naskar*[b] Introduction Lignin, the most abundant natural aromatic polymer, exists in the cell wall of plants. For instance, it represents 18–35% of wood by weight.[1] The annual natural production of lignin on Earth is estimated as 5–36108 tons.[2] Lignin is currently pro- duced on an industrial scale in paper and pulping factories as well as in biorefineries. The paper and pulping industry alone produces lignin in quantities that exceed 50 million tons annu- ally,[3] but the majority of lignin produced is currently utilized as a low-cost fuel to balance energy needs. Only approximately 1 million tons per year of lignin is currently used for commer- cial applications, which include concrete additives, dyestuff dis- persants, binders or surfactants for animal feed, dust control, and pesticides.[4,5] However, the market demand for all of these nonfuel lignin usages represents a negligible fraction of cur- rent lignin production. One of the biggest hindrances for the commercialization of high-performance lignin products is that lignin has heterogeneous properties, which include molecular weight, functionality, and thermal properties, from different sources and processing methods. Therefore, the isolation of high-purity lignin for high-performance lignin-derived materials should be urgently pursued for the development of new ap- proaches for lignin in various applications. It is critical to devel- op a wide utilization of the renewable and abundant lignin re- source for a sustainable society. Some of the high-performance lignin derivatives utilize lignin as a macromonomer in a component of lignin copoly- mers as well as carbon fiber precursors.[6] We have recently re- ported the successful synthesis of new thermoplastic copoly- mers from lignin, which are rubbery polyester copolymers[7] and robust polyurethane copolymers.[8] These thermoplastic copolymers from lignin exhibited a two-phase behavior in dy- namic mechanical analysis, characteristic of multiphase ther- moplastic copolymers.[9] Our previous work demonstrated that the lignin content and glass transition temperature (Tg) can be tuned to achieve the desired properties. The capability to tailor the Tg of lignin is extremely important for its processing in ap- plications such as carbon fiber precursors[10] because it directly correlates to the melt-processing temperature or the stabiliza- tion conditions prior to carbonization. Key to the successful synthesis of lignin-based thermoplastic copolymers was the utilization of a high-molecular-weight (HMW) lignin fraction to produce an efficient connection between the hard segment (i.e., lignin) and the soft segment. The HMW lignin fraction was prepared by formaldehyde crosslinking, whereas as-received lignin contained a significant amount of a low-molecular- weight (LMW) fraction. In both studies, the crosslinked HMW The development of technologies to tune lignin properties for high-performance lignin-based materials is crucial for the uti- lization of lignin in various applications. Here, the effect of methanol (MeOH) fractionation on the molecular weight, mo- lecular weight distribution, glass transition temperature (Tg), thermal decomposition, and chemical structure of lignin were investigated. Repeated MeOH fractionation of softwood Kraft lignin successfully removed the low-molecular-weight fraction. The separated high-molecular-weight lignin showed a Tg of 2118C and a char yield of 47%, much higher than those of as- received lignin (Tg 1538C, char yield 41%). The MeOH-soluble fraction of lignin showed an increased low-molecular-weight fraction and a lower Tg (1178C) and char yield (32%). The amount of low-molecular-weight fraction showed a quantitative correlation with both 1/Tg and char yield in a linear regression. This study demonstrated the efficient purification or fractiona- tion technology for lignin; it also established a theoretical and empirical correlation between the physical characteristics of fractionated lignins. [a] Dr. T. Saito Chemical Sciences Division Oak Ridge National Laboratory Oak Ridge TN 37831-6210 (USA) E-mail: saitot@ornl.gov [b] J. H. Perkins, Dr. F. Vautard, Dr. H. M. Meyer, Dr. A. K. Naskar Materials Science and Technology Division Oak Ridge National Laboratory Oak Ridge TN 37831-6053 (USA) E-mail: naskarak@ornl.gov [c] Dr. J. M. Messman Center for Nanophase Materials Sciences Oak Ridge National Laboratory Oak Ridge TN 37831-6494 (USA) [d] Dr. B. Tolnai Industrial Products Division Kruger Inc. 3285 Bedford Road, Montral, Qubec H3S 1G5 (Canada) Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cssc.201300509.  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 9 &1& These are not the final page numbers! ÞÞ CHEMSUSCHEM FULL PAPERS fraction of lignin showed a higher Tg than as-received lignin. The removal of the LMW fraction also reduced its polydispersity index (PDI), which enabled the suc- cessful generation of two distinctive phases in the thermoplastic copolymers. Our previous work[7] not only reported the formal- dehyde crosslinking but also the increase of the lignin molecular weight and Tg by methanol (MeOH) fractionation. Solvent fractionation of lignin has been used in industry and investigated for several different sources of lignin and different solvents.[11–20] Solvent fractionation to increase the molecular weight and Tg is a more facile approach than performing a formalde- hyde crosslinking reaction; however, our work showed that the isolated yield from MeOH fractionation of lignin was very small (17%) and that its molecular weight distribution was broad. The previous work investigated hardwood-based solvent-ex- tracted lignin, and the as-received lignin had a low molecular weight (Mn =1840 gmol1 , PDI=122) with a low Tg (1088C). In this study, we have investigated MeOH fractionation of soft- wood Kraft lignin, which has a higher molecular weight and the Tg of as-received lignin was higher than hardwood lignin in our previous work. The effect of MeOH fractionation process on the molecular weight, molecular weight distribution, Tg, thermal decomposition, and chemical structure of lignin was investigated. Furthermore, these properties were empirically correlated to fit into a physical theory. The elucidation of these lignin properties through a lignin purification processes is crucial to foster technology for advanced lignin-based materials. Results and Discussion Effect of MeOH fractionation of lignin on its molecular weight Medium ash grade lignin (as-re- ceived lignin) from Kraft-pro- cessed softwood biomass was put into MeOH, stirred, shaken, and centrifuged to separate MeOH-insoluble lignin and MeOH-soluble lignin (Figure 1). As-received lignin gave Mn =10000 gmol1 and PDI=110 (Table 1) from DMF size-exclusion chroma- tography (SEC) without LiBr. The very broad molecu- lar weight distribution is consistent with that ob- tained from a hardwood lignin in our previous stud- ies.[7,8] The SEC curve of as-received lignin revealed a HMW peak and a LMW peak (Figures 2a and 3a). The presence of HMW and LMW peaks is consistent with other work.[7,8,21,22] The HMW peak is assigned as the peak area located before a retention time of 22 min, whereas the LMW peak is assigned as the peak located after a retention time of 22 min. The area of the HMW peak corresponds to 93.6% of the Figure 1. Schematic diagram of the MeOH fractionation of lignin. Table 1. Molecular weight of lignin from MeOH fractionation (without LiBr). Lignin Number of washes Mn [a] [gmol1 ] Mw [a] [gmol1 ] PDI[a] HMW area [%] LMW area [%] as-received 0 10000 1110000 110 93.6 6.4 MeOH-insoluble 1st 28200 1310000 46.3 97.2 2.8 2nd 60600 1400000 23.1 98.8 1.2 3rd 63800 1530000 24.0 98.8 1.2 4th (last) 443000 1550000 3.5 100 0 MeOH-soluble 3800 492000 129 84.7 15.3 [a] Determined by RI detection. Figure 2. (a) SEC curves of as-received lignin (top), MeOH-washed lignin [MeOH-insoluble lignin; first, second, and fourth wash (bottom)] and (b) area of LMW fraction from the SEC curve at higher than 22 min retention time and PDI as a function of the number of MeOH washes. Figure 3. SEC curves of as-received lignin (c), MeOH-insoluble lignin (b), and MeOH-soluble lignin (a) (a) without and (b) with LiBr.  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 9 &2& These are not the final page numbers! ÞÞ CHEMSUSCHEM FULL PAPERS www.chemsuschem.org total area, whereas the area of the LMW peak corresponds to 6.4% of the total area. The presence of a LMW fraction with an area of only 6.4% contributed to its very large PDI. Sequential MeOH washing removed the LMW fraction of lignin. As the number of MeOH washes increased, the LMW peak area of the MeOH-insoluble lignin decreased significantly (Figure 2a and b, Table 1), so that MeOH-insoluble lignin contained a minimal LMW peak after the fourth wash. As a result of the removal of the LMW fraction, the PDI value decreased dramatically (Fig- ure 2b, Table 1). After the fourth wash, the DMF SEC analysis of MeOH-insoluble lignin gave Mn = 443000 gmol1 and PDI=3.5. The isolated yield after the fourth wash was 50%, which is much greater than that of hardwood lignin as reported in our pre- vious work.[7] In addition, MeOH fractionation for the hardwood lignin never achieved a single-digit PDI, possibly because of the presence of a huge LMW fraction and the lack of using a centrifuge and a Sil- verson mixer. This Kraft-processed softwood-based as-received lignin appears to contain a much higher content of the HMW fraction than solvent-extracted hardwood-based lignin. As the successful removal of the LMW fraction was achieved after the fourth wash, the rest of this study will focus on this MeOH-in- soluble lignin residue obtained after four washes with MeOH. The term MeOH-insoluble lignin will refer to this lignin that was washed four times. MeOH-soluble lignin was characterized by SEC, and the results were compared with those of as-re- ceived lignin and MeOH-insoluble lignin. As shown by the chromatograms in Figure 3a and the data in Table 1, the LMW fraction in MeOH-soluble lignin increased compared to that in as-received lignin. This indicates that MeOH extraction enriches the LMW fraction of lignin by dissolving it in the extraction medium. As a result of an increase of the LMW area (15.3%), the mo- lecular weight of the MeOH-soluble lignin decreased (Mn = 3800 gmol1 ), and it showed a broader PDI (PDI=129) than as-received lignin. This principle of the removal of a LMW frac- tion of lignin with MeOH should be applicable for all sources of lignin or at least the majority of them. Notably, MeOH-solu- ble lignin still contains a large quantity of the HMW fraction, al- though the HMW peak shifted toward a slightly higher reten- tion time in SEC. The HMW fraction of lignin might be partially soluble in MeOH, or the colloids of lignin in MeOH solution might stay in the solution rather than precipitating as a solid chunk. It was impossible to distinguish tiny colloid particles and soluble lignin as the appearance of the MeOH solution was dark brown. This MeOH washing strategy is applicable to separate the HMW fraction of lignin as discussed above, but it might not be possible to obtain only the LMW fraction be- cause of its solubility or its tendency to form a colloidal sus- pension. If one aims to separate only the LMW fraction from the HMW fraction, a membrane filtration type of separation might be recommended. Lignin is known to aggregate in solution, so the lignin peak from neat DMF in the SEC most likely represents both single lignin molecules and lignin aggregates in DMF.[22,23] The HMW peak in the SEC might represent aggregates of lignin, whereas the LMW peak might correspond to non-aggregate-forming LMW lignin molecules. LiCl and LiBr have been reported to dis- rupt lignin aggregation.[22,24] Therefore, for comparison, DMF SEC with LiBr (0.05m) was performed, and the data are sum- marized in Table 2 and Figure 3b. All the SEC curves in Fig- ure 3b show a monomodal peak, rather than the bimodal peaks seen in SEC curves obtained with DMF SEC without LiBr (Figure 3a). As shown by their narrower monomodal SEC peaks, the PDI of these samples was between 1.90 and 3.05, which is much smaller than that obtained from DMF SEC with- out LiBr. As seen in the narrower SEC curves in Figure 3b, the PDI of both MeOH-insoluble lignin (2.28) and MeOH-soluble lignin (1.90) were smaller than the as-received lignin (3.05), an indication of achieving effective fractionation by MeOH. The obtained molecular weights (MW) followed the same trend as those obtained with DMF SEC without LiBr, in which MeOH-in- soluble lignin fractionated HMW components and MeOH-solu- ble lignin fractionated LMW components. However, the MW values are significantly smaller. The Mw values corresponding to MeOH-insoluble lignin and MeOH-soluble lignin were 14900 and 3000 gmol1 , respectively, whereas the Mw of as-received lignin was 7190 gmol1 according to refractive index (RI) de- tection. If a light scattering detector was used, the Mw of MeOH-insoluble lignin, MeOH-soluble lignin, and as-received lignin was 188000, 86600, and 96100 gmol1 , respectively. Because the MW measured by the RI detector is relative to the poly(2-vinyl pyridine) standard, the absolute MW measured by the light scattering detector most likely represents a more ac- curate MW. If the actual MW distribution of these lignins is a monomodal distribution as shown in the SEC data obtained from DMF SEC with LiBr (Figure 3b), the multimodal peaks ob- tained from DMF SEC without LiBr represent aggregation and heterogeneity of lignin molecules in neat DMF. The correlation of the LMW fraction obtained from DMF SEC without LiBr with various properties will be discussed in the following sections. Effect of MeOH fractionation of lignin on its thermal re- sponse properties MeOH fractionation altered the Tg of as-received lignin (Figure 4, Table 3). MeOH-insoluble lignin had a Tg of 2118C, which is 508C higher than the Tg of as-received lignin (1538C). The increased Tg after MeOH washing is consistent Table 2. Molecular weight of lignin with the addition of LiBr to the SEC solvent. Lignin Mn [a] [gmol1 ] Mw [a] [gmol1 ] PDI[a] Mw [b] [gmol1 ] dn/dc[c] [mLg1 ] as-received 2360 7190 3.05 96100 0.18670.0011 MeOH-insoluble 6520 14900 2.28 188000 0.18370.0092 MeOH-soluble 1590 3000 1.90 86600 0.16360.0028 [a] Determined by RI detection. [b] Determined by light scattering detection. [c] Deter- mined off-line.  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 9 &3& These are not the final page numbers! ÞÞ CHEMSUSCHEM FULL PAPERS www.chemsuschem.org with the trend in solvent-extracted hardwood lignin reported in our previous work.[7] However, the Tg of MeOH-soluble lignin was 1178C, much lower than that of as-received lignin. The trend of Tg alteration can be explained by the content of LMW lignin measured by DMF SEC without LiBr. The LMW fraction should provide lignin with a plasticizing effect. In DMF SEC without LiBr, the LMW fraction with a retention time higher than 22 min corresponds to a peak MW (Mp) of 700– 1100 gmol1 , although Mn and Mw could not be determined because of its molecular weight range, which was outside of the calibration curve. The Mp value of 700–1100 gmol1 sug- gests that the LMW fraction consists of lignin oligomers. The presence of a LMW fraction is represented as LMW area [%] in Table 1. As the LMW area increases, the Tg of the correspond- ing lignin decreases; this trend follows a linear regression based on the Fox equation[25] [Eq. (1)] with a correlation coeffi- cient R2 of 0.974 (Figure 5): 1 Tg ¼ wA TgA þ wB TgB ð1Þ where wA and wB correspond to the weight fractions of A and B and TgA and TgB correspond to the Tg of components A and B, respectively. In this study, TgA represents the Tg of the LMW fraction and TgB corresponds to the Tg of MeOH-insoluble lignin (2118C). The area from the RI peaks in SEC represents the weight fraction of the sample in the solution, and thus the LMW area [%] in Table 1 is converted to wA in Figure 5. The linear regression follows the equation y=0.00336x+0.00207. Equation (1) derives Equation (2), and TgA can be determined from the slope of Equation (2) according to Equa- tion (3). 1 Tg ¼ wA TgA þ 1  wA TgB ¼ 1 TgA  1 TgB   wA þ 1 TgB ð2Þ 1 TgA  1 TgB ¼ 0:00336 ð3Þ 1 TgA ¼ 0:00336 þ 1 TgB ¼ 0:00336 þ 0:00207 ¼ 0:0543 TgA ¼ 184:3 K The calculated TgA results in 898C. Although the LMW fraction could not be isolated in pure form (as the extracted mass contained some agglomerated HMW frac- tion), the Fox equation suggests that the Tg of LMW oligomeric lignin is 898C (Table 3). The Tg value derived from the Fox equation for oligomeric lignin is so low that it justifies the plas- ticization effect of the LMW lignin fraction. As Tg is one of the most important lignin processing parameters, the ability to alter Tg is critical for the development of lignin-derived prod- ucts. MeOH fractionation as depicted here provides the ability to tune the Tg of lignin from existing sources. MeOH fractionation also alters the thermal stability of lignin (Figure 6, Table 3) under N2. MeOH-soluble lignin decomposed the quickest, as seen in the 10 wt% weight loss at 2608C and in the maximum of the decomposition derivative peak located at 3548C. The initial degradation of MeOH-insoluble lignin (10 wt% weight loss at 2938C, the maximum of the decompo- sition derivative peak located at 3628C) was slightly faster than Figure 4. DSC curves of as-received lignin (middle), MeOH-insoluble lignin (top), and MeOH-soluble lignin (bottom). Table 3. Tg obtained from DSC and thermal degradation properties obtained from TGA. Lignin Tg [8C] 10 wt% weight loss [8C] Derivative weight peak [8C] Residual weight at 10008C [%] as-received 153 301 384 41 MeOH-insoluble 211 293 362 47 MeOH-soluble 117 260 354 32 LMW fraction 89[a] – – – [a] Estimated from the Fox equation. Figure 5. Tg as a function of LMW content in the lignin. Figure 6. TGA curves of as-received lignin (c), MeOH-insoluble lignin (b), and MeOH-soluble lignin (a).  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 9 &4& These are not the final page numbers! ÞÞ CHEMSUSCHEM FULL PAPERS www.chemsuschem.org that of as-received lignin (10 wt% weight loss at 3018C, maximum of the decomposition derivative peak located at 3848C). The alteration of the degra- dation temperature indicates that not only the mo- lecular weight affects the thermal stability but also the chemical composition as some functional groups have been altered in comparison to as-received lignin, which is addressed in the next section. More importantly, the char yield (residual weight) at 10008C increased for MeOH-insoluble lignin (47 wt%) and decreased for MeOH-soluble lignin (32 wt%) compared to that of as-received lignin (41 wt%). If lignin is used as a carbon precursor, such as in the case of carbon fibers or activated carbons, the increase of char yield is highly beneficial for economic and environmental reasons. Thus, MeOH-insoluble lignin, or HMW lignin, could be more valuable as a carbon precursor than LMW lignin. Interestingly, the char yield followed a linear regression with the area of LMW fraction obtained from SEC without LiBr (Figure 7). The coefficient of correlation R2 was 0.997. If we consider that the area of the LMW fraction in SEC by RI intensity corresponds to the mass ratio, such a good fit with a linear regression indicates that the majority of the LMW fraction does not form char. In fact, an increase of ...