Introduction

Lignin constitutes the structural support of vascular plants’ tissues and algae contributing up to 30% of biomass [1,2,3,4]. As a result, lignin is one of the most abundant sources of renewable carbon, a potential feedstock for renewable liquid fuel and valuable phenolic chemicals [5]. Various approaches to transform lignin into commercially beneficial substances are being investigated including extraction as well as thermal, catalytic, and biological degradation [1, 3, 6,7,8,9].

The determination of produced low molecular weight (MW) species (< 400 Da) present in the mixture of decomposition products is usually accomplished by one- or two-dimensional gas chromatography (GC or GC×GC) with either flame ionization (FID) or mass spectrometry (MS) detection, or by liquid chromatography with electrospray ionization (ESI) MS (LC-MS) [9]. This analysis allows for a comprehensive characterization of up to 20% of the product mass extractable by organic solvents, so-called bio-oil [9]. However, a significant non-GC-able portion of degradation products, which consists of phenolic oligomers and a “coke” solid soot-like residue originating from native lignin, is only tentatively characterized [9]. Furthermore, for a comprehensive assessment of the degradation efficiency and understanding of decomposition pathways, a thorough characterization of lignin in the form of a native polymer prior to the chemical modification and degradation is essential, yet remains extremely challenging.

The most common method employed for intact lignin characterization is size exclusion chromatography (SEC). Recent studies suggest Agilent PLgel columns to be the most suitable for biomass and lignin analysis in tetrahydrofuran (THF)-based systems [10, 11]. While this method is an excellent tool for regular polymer MW determination, its application to lignin is limited because of the lack of pertinent calibration standards. As a result, calibration is usually conducted with polystyrene standards, which do not match the structure of lignin [12]. Additionally, lignin’s heteropolymeric nature contributes to secondary interactions of various functional groups with the column stationary phase [12,13,14,15].

Steric effects typical for rigid phenolic polymers also tend to skew SEC results [12]. To improve the accuracy of MW determination by SEC, detectors such as low-angle light scattering and multi-angle laser light scattering are frequently used [13, 16,17,18,19,20,21]. However, the determined MW values may still be skewed by the fluorescence and UV absorptivity of lignin constituents along with possible lignin macromolecule association [22].

While mass spectrometry is often the method of choice, Table 1 demonstrates the strengths and the limitations of various MS approaches and unresolved issues reported for intact lignin characterization alone [14], or in combination with pyrolysis (Pyr) GC [46,47,48,49,50,51,52,53,54,55,56]. Pyr-GC is traditionally used to determine the type of lignin or lignocellulose; however, it neither provides any MW information nor allows for a differentiation of small and large MW species [9]. The application of Pyr-GC-MS typically targets the speciation of syringol and guaiacol within lignin structural units [55].

Table 1 Comprehensive Overview of the MS Approaches Employed for the Analysis of Intact Lignin and High MW Standards
Full size table

Matrix-assisted laser desorption/ionization (MALDI) MS is currently considered the only suitable MS approach [40] for determination of MW distribution in synthetic polymers, proteins, and polysaccharides [40]. However, MALDI provides only limited structural information [37, 41, 44] due to partial fragmentation, which makes it impossible to distinguish small and large molecules [37, 41, 44]. This issue could be addressed by coupling MS to LC; however, MS detection with MALDI as an ion source cannot be used for an online analysis of the LC eluent due to the required sample preparation procedure, which includes the analyte co-crystallization in a mixture with a matrix. Moreover, sample co-crystallization leads to undesirable association effects [30]. MALDI ionization also results in the predominant formation of single-charged species, thus suppressing the ionization of high MW species and promoting the formation of ions featuring low m/z ratios [41]. Additionally, the analyte MW is typically limited to the mass analyzer operation range [38]. These limitations of MALDI ionization result in a noisy lignin spectrum lacking resolved peaks, particularly on the higher end of the m/z scale [38, 40, 42, 44].

Intact lignin has also been investigated using MS with ESI and atmospheric pressure chemical ionization (APCI). However, these studies focused only on a limited m/z range (up to 1000 m/z [23,24,25,26,27, 34, 57] and 1500 m/z [25, 28, 29, 31, 32]) aiming to evaluate lignin degradation products (Table 1) [14, 22]. The significance of these studies was in the successful elucidation of the structural patterns and common linkages in various types of lignin, yet the MW was left undetermined [23,24,25,26,27,28,29, 31, 32, 34, 57], as well as a potential formation of multiply charged species characteristic for ESI, which is widely used in proteomics [58], yet has not been reported to our knowledge for lignin. Several research groups utilized ESI with an extended m/z range (up to 3500 m/z [33] and 7000 m/z [30]); however, the reported spectra showed a low signal-to-noise ratio and low intensity above 1000 m/z, possibly due to the use of ionization conditions optimized for low MW lignin model compounds. To the best of our knowledge, a thorough optimization of ionization conditions and mass analyzer parameters within an extended m/z range aiming for a greater peak resolution and higher signal-to-noise ratio has not been reported.

The intermediate products of lignin degradation, i.e., lignin mono- to oligomeric standards, appear to be suitable model compounds for understanding the ionization mechanism of intact and degraded lignin [23, 25, 34, 57]. An extensive optimization of ESI and APCI conditions toward the analysis of mono- and several diarene lignin model compounds was reported, to our best knowledge, only in the study by Haupert et al. while utilizing a linear quadrupole ion trap MS [59]. Kosyakov et al. assessed ESI, APCI, and atmospheric pressure photoionization (APPI) as ionization techniques focusing on specific additives for analysis of lignin and its model compounds using a quadrupole-orbitrap MS [23].

The limitation of APCI and APPI is their applicability only to a relatively low MW range, below 1800 Da, and the predominant formation of singly charged ions. In addition, APCI was shown to suffer from excessive fragmentation of lignin model compounds, thus making this method inappropriate for both analysis of complex degradation mixtures [26, 59] and determination of intact lignin MW [27]. Thus, APPI and APCI may be preferred for studying oligomeric species and structural patterns in lignin. However, these ionization techniques cannot be applied for detection of high MW species in lignin.

Sodium hydroxide and sodium chloride have been claimed to be the most effective electrolytes while utilizing ESI, as minimal molecular ion fragmentation was observed for acidic lignin model compounds in the negative mode and for nonacidic compounds in the positive mode [59]. The ESI MS with NaOH method was successfully applied in combination with reversed phase LC by Owen et al., enabling a structural elucidation of some low MW components in a mixture of lignin degradation products [60]. By contrast, Kiyota et al. showed that oligomers with a MW up to 852 Da could be successfully ionized in the negative mode in the presence of 20 mmol L−1 formic acid and created a do-it-yourself oligomer database, which included mono- to tetrameric phenolic species synthesized in-house [34]. Proposed in 2010, the MS-based sequencing of lignin oligomers as an identification tool allowed for coining the term “lignomics” [25, 57]. Oligomeric model compounds and isolated oligolignols (up to 650 m/z) were efficiently ionized with APCI and ESI in the negative mode; furthermore, the fragmentation pathways of the major bonding structures in the gas phase were elucidated [25, 57]. All of these approaches laid the foundation for the development of an MS-based sequencing algorithm, yet they varied in ionization approaches and were limited by their focus on relatively low m/z values (< 900). Thus, detection of high MW species, which are essential components of lignin, remains challenging and prevents lignin MW elucidation with the currently available MS toolkit.

In this study, we have developed an MS protocol extending the lignomics toolkit through the detection of high MW lignin components as well as low MW phenolics in kraft alkali lignin, which belongs to the largest lignin class produced as a by-product of the paper industry [61, 62]. Building on previous studies [25, 31, 32, 34, 59], we investigated the effect of a broader suite of electrolytes on ESI of two representative lignin-like dimers. This was followed by narrowing the range of electrolytes while assessing the ionization of 11 mono-, di-, and triarene lignin model compounds featuring different oxygenated functional groups and linkages typical for lignin and then expanded the method to native lignin. This investigation led to the development of a protocol for successful ionization of high MW lignin species with a focus on the multiply charged ion formation. Due to lignin’s highly heteropolymeric origin, the species within a broad range of MW and of a number of possible monomeric unit combinations may occur in the sample. This heteropolymeric nature, the lack of the regular repeating unit along with their somewhat random crosslinking, complicates identification. Thus, our focus was not on identification alone but primarily on characterizing the distribution of the MW, which is essential when working with polymers. Thus, we postulated and confirmed that higher MS resolution led to an improved detection of high MW species when an ion mobility (IM) feature was employed with ESI-HR TOF-MS. The determined average MW values of lignin were compared to those obtained by laser desorption/ionization (LDI)-HR TOF-MS and gel permeation chromatography (GPC).

Experimental

Materials and Reagents

All standards used in this study as lignin model compounds are listed in Table 2, along with their acronyms, while details are provided in Supplementary Table 1. Alkali lignin (CAS Number: 8068-05-1), cesium iodide (99.999%), ammonium acetate (≥ 99%), sodium iodide (99.999%), lithium chloride (99.999%), lithium hydroxide (99.998%), and sodium hydroxide (≥ 98%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ammonium hydroxide and trifluoroacetic, acetic, and formic acids (puriss p.a., eluent additive for LC-MS) were obtained from Fluka.

Table 2 List of Lignin Model Compounds Used in This Study
Full size table

MALDI matrices, i.e., 2-(4′-hydroxybenzeneazo)benzoic acid (HABA, ≥ 99.5%) and α-cyano-4-hydroxycinnamic acid (CHCA, ≥ 99%), were purchased from Sigma-Aldrich. HPLC-MS grade water, acetonitrile (ACN), unstabilized THF, and methanol (MeOH) were obtained from Fisher Scientific (Fair Lawn, NJ, USA). All chemicals were used as received, without further purification. Deionized water was obtained using a Direct-Q® 3 system, Millipore, Billerica, MA, USA.

Several dimeric lignin model compounds, i.e., D2V [65, 66], G-β-2 [63], ET2 [64], EST2 [64], ALC2 [64], ALK2 [64], and ET3 [64], featuring different functional groups and linkages (Table 2) were synthesized according to the procedures published earlier, with an addition of column chromatography and recrystallization purification steps. A detailed description of the syntheses can be found under Supplementary Material. These compounds were characterized by 1H NMR, GC-MS, and direct infusion ESI-HR TOF-MS.

Stock solutions of lignin mono-, di-, and trimeric standards were prepared in 50% MeOH/water with a final concentration of 100 ppm w/v. For kraft alkali lignin analysis, it was essential to dissolve the polymer while avoiding the use of aggressive solvents such as dimethyl sulfoxide or N,N-dimethylformamide. We have shown that alkali lignin may be completely dissolved in ACN-water (1:1) or THF-water (1:1) mixtures, with concentrations up to 10,000 and 50,000 ppm (w/v), respectively. It is of note that neither pure organic solvents (ACN, THF) nor water dissolved any amounts of lignin to form a true solution. For direct infusion analysis of lignin, alkali lignin was completely dissolved in either water/ACN (1:1) or water/THF (1:1) at a final concentration of 100 or 1000 ppm, respectively. The solutions were diluted to a final lignin concentration of 80 or 90 ppm prior to the analysis. Neither of the utilized electrolytes caused lignin precipitation. All samples and stock solutions were stored in a refrigerator at 4 °C prior to analysis.

To address the solubility issue in THF, the effect of acetylation was evaluated [67]. Acetylation of lignin samples was performed by a conventional method [68]. In brief, about 50 mg of the sample was completely dissolved in 500 μL of dry pyridine and reacted with 500 μL of acetic anhydride. The reaction mixture was stirred for 12 h at room temperature. Then, 200 μL of methanol was added to the reaction mixture to terminate the reaction. Solvents were evaporated under a stream of nitrogen and the residue was dried in a vacuum oven at 30 °C overnight. Acetylated lignin was completely dissolved in 5.5 mL of THF resulting in a 10,000-ppm w/v solution, which was further diluted with water/ACN (1:1) or water/THF (1:1) mixtures for the ESI-HR TOF-MS analysis.

MS Analysis: Ionization

An Agilent 6210 ESI-HR TOF-MS system was used throughout the study for method development and parameter optimization. An initial optimization of MS conditions included selection of the ionization polarity, electrospray (e.g., capillary) and collision-induced dissociation (e.g., fragmentor) potentials, nebulization temperature, nebulizing gas flow rate, and nebulization pressure. Samples were introduced via a direct infusion with a syringe pump at a flow of 5 μL min−1 for the initial optimization. ESI potentials were optimized between 2000 and 5000 V. Nebulization pressure, gas flow, and vaporizer temperature were varied between 18 and 40 psi, 4 and 12 L min−1, and 250 and 400 °C, respectively. The full range of electrolytes specified in the “Materials and Reagents” section was evaluated. In the experiments involving THF, all PEEK tubings were replaced with stainless steel.

To optimize the electrolyte concentration, a flow injection analysis (FIA) was performed employing an Agilent 1100 Series HPLC. An aliquot (20 μL) of the prepared solution was injected into a mobile phase consisting of 50% ACN or MeOH in water at a flow rate of 0.2–1.0 mL min−1 and delivered directly to the ESI-HR TOF-MS system (no LC column was installed). In this study, we optimized the electrolyte concentration in the mobile phase by doping it only into the sample as we did previously in our work [69]. We experimentally confirmed that the final electrolyte concentration in the mobile phase after the injection of an electrolyte-doped sample remained the same (Supplementary Table 2).

The ESI-HR TOF-MS system was calibrated with an ESI (50–3500 m/z) tuning mixture purchased from Agilent. For higher m/z measurements (i.e., analysis of kraft alkali lignin in a range of 150–10,000 m/z), the calibration was performed in the positive mode while using [(CsI) n +Cs]+ clusters formed by an introduction of cesium iodide [30 mmol L−1 solution in ACN/water 1:1 (v/v)] via direct infusion at a flow rate of 5 μL min−1 (Supplementary Fig. 1). Agilent 6560 IM HR Q-TOF-MS system equipped with an ESI source was used to acquire IM mass spectra under the optimized conditions determined with 6210 TOF MS system.

A MALDI SYNAPT G2-Si Q-TOF MS System (Waters, Milford, MA, USA) with CryLaS FTS355-Q laser (a repetition rate of 2.5 kHz, wavelength 355 nm) was employed to acquire MALDI MS spectra in the range 50–8000 m/z. The system was manually calibrated with red phosphorus, and the experiments were performed in the positive resolution mode (20,000). The laser energy was set to 350 arbitrary intensity units. Typically, HABA [29, 36, 38, 39, 41,42,43,44], CHCA [37, 40], retinoic [38], or sinapinic [39] acids were used as matrices for lignin analysis. Contradicting results were reported on the analysis with no matrix used (LDI); some studies claim successful ionization [35, 42], while the others report it to be ineffective [40]. In this study, we evaluated lignin ionization with CHCA and HABA used as matrices and without any matrix. The best ionization effectiveness was achieved when no matrix was used.

MS Characteristics and Data Processing

The 6210 HR TOF-MS system with a mass resolution of > 13,000 (at m/z 2722) and mass accuracy < 2 ppm (m/z 609.2807) was utilized. A 6560 IM Q-TOF-MS system used for IM mass imaging had a resolution of > 42,000 (at m/z 2722) and mass accuracy < 2 ppm. The resolving power of SYNAPT G2-Si MS system was 50,000 and the mass accuracy was under 1 ppm.

Mass Hunter software packages, B.02.00 and B.07.00, were used for data processing. The spectra of kraft alkali lignin recorded in the positive mode, which included multiply charged ions, were deconvoluted using a built-in tool utilizing an unbiased isotope model with a peak spacing tolerance of 0.0025 m/z. The maximal assigned charge state was not limited. Both hydrogen and sodium were considered as the charge carriers. The peaks selected for deconvolution were filtered based on their absolute height (≥ 100 counts) and the relative height of the largest peak, which was set to ≥ 0.1% of the largest peak unless otherwise stated. The maximum number of peaks was not specified. The deconvolution data were further supported by independent calculations in MS Excel.

Equation 1 was used for the MW (M) calculation of the multiply charged species:

$$ \frac{m}{z}\cdotp z-\left[\left(\mathrm{atomic}\ \mathrm{mass}\ \mathrm{of}\ \mathrm{charge}\ \mathrm{carrier}\ \left({}^1\mathrm{H}\ \mathrm{or}\ {}^{23}\mathrm{Na}\right)-\mathrm{electron}\ \mathrm{rest}\ \mathrm{mass}\right)\cdotp z\right]=M,\mathrm{Da} $$
(1)

Solver function, an Excel add-on, was utilized to propose the possible chemical structures for the corresponding MW. Masses of the typical lignin structural units and the most common types of linkages observed in lignin were the input parameters, which restrained the proposed structures.

An open source alternative software for mass spectrometric data analysis mMass Data Miner [70] was used for MALDI data processing.

To qualitatively assess the MW distribution of lignin utilizing the MS data, we applied an approach used in our previous work for MALDI MS data interpretation [11]. To calculate the number average (M n ), weight average (Mw), and z-average (M z ) MW of lignin samples, Eqs. 2–4, where I i is the absolute abundance of the deconvoluted species of a given MW (M i ), were used [71].

$$ {M}_n=\frac{\sum {I}_i{M}_i}{\sum {I}_i} $$
(2)
$$ {M}_{\mathrm{w}}=\frac{\sum {I}_i{M}_i^2}{\sum {I}_i{M}_i} $$
(3)
$$ {M}_z=\frac{\sum {I}_i{M}_i^3}{\sum {I}_i{M}_i^2}. $$
(4)

Results and Discussion

Electrolyte Screening: Effect on the Representative Model Compounds

We optimized the ESI-HR TOF-MS ionization targeting a broad range of electrolytes using two representative dimers as model compounds (Figure 1). Two dimers, G-β-2 and ET2, were selected for this initial screening, the former featuring both aromatic and aliphatic hydroxyl groups, while the latter does not have any hydroxyl groups. Both of these standards exhibited the most efficient ionization while forming sodium adducts in the positive mode in the presence of formic or acetic acids at ≤ 10 mmol L−1 concentration (Figure 1).

Figure 1
figure 1

ESI-HR TOF-MS response obtained in the presence of different electrolytes via a direct infusion for two representative dimers (a) G-β-2; (b) ET2 in the positive and negative ionization modes. For most of the electrolytes used, the response for [M+Na]+ and deprotonated molecular ions is shown, except for LiCl and LiOH whose application resulted in the formation of [M+Li]+. The electrolyte concentration was 1.0 mmol L−1 unless specified otherwise

Full size image

A preferable ionization of ET2 in the positive ESI mode was expected, due to the lack of hydroxyl groups in its structure. For the hydroxylated compound G-β-2, contrary to expectations, the ionization efficiency improved in the positive ESI mode resulting in an abundant sodium adduct ion (Figure 1a). The formation of sodium adduct ions even when sodium was not purposely added to the samples is known to occur because of traces of sodium leaching from glassware [72].

An effective ionization was also observed with no electrolyte present (Figure 1a), the feature frequently observed in ESI-HR TOF-MS [73]. However, these conditions were deemed to be nonoptimal as the lack of a buffer could result in pH instability and consequently cause the dependence of ionization on the sample composition and, as a result, irreproducible data [73].

Sodium hydroxide, which was previously claimed as an effective ionization agent for lignin model compounds in both positive and negative ESI modes [59], did not promote the formation of sodium or protonated adducts for G-β-2 as much as the majority of other evaluated electrolytes (Figure 1a). Moreover, in contrast to the previous work [59], excessive fragmentation was observed in the presence of sodium hydroxide in the positive (Figure 2c) and negative modes (Supplementary Fig. 2), whereas a “clean” mass spectrum was obtained when either ammonium acetate or acetic acid was used (Figure 2a, b). Perhaps, the reason for a different fragmentation pattern upon ionization of G-β-2 was a varied stability (fragmentation) of ions in different mass analyzers used in previous studies, as Haupert et al. [59] utilized a linear quadrupole ion trap MS, whereas in our study we used TOF.

Figure 2
figure 2

Positive ESI-HR TOF-MS of 5 ppm G-β-2 afforded the sodiated adduct ([M+Na]+ at m/z 343.1152) in the presence of 1.0 mmol L−1 of (a) ammonium acetate (mass accuracy error 5 ppm), (b) acetic acid (mass accuracy error 7 ppm), and (c) sodium hydroxide (mass accuracy error 6.7 ppm)

Full size image

ESI: Impact of Oxygenated Functional Groups on Ionization

Based on the screening experiments (cf. Figure 1), formic and acetic acids appeared to be the most efficient electrolytes. Thus, we compared these acids to the frequently used ammonium acetate [74], to evaluate the impact of these electrolytes on the ionization effectiveness of a broader suite of mono-, di-, and triarene lignin model compounds featuring different linkages and functional groups (Table 2), and investigated the contribution of various oxygenated functional groups: The results are shown in Table 3. Contrary to the previously preferred negative ionization mode [34, 57], we showed ionization of all considered compounds (with hydroxyl, methoxy, and carboxyl groups) in the positive ESI mode with both formic/acetic acids and ammonium acetate, although some selectivity toward specific oxygenated functional groups was observed.

Table 3 ESI-HR TOF-MS Response with Acids (Either Formic or Acetic) and Ammonium Acetate as ESI Electrolytes for Representative Lignin Mono- to Trimeric Structure Model Compounds in Both Positive and Negative Ionization Modes
Full size table

The compounds without phenolic hydroxyl groups, with multiple methoxy groups, or with aliphatic hydroxyl groups, were preferentially ionized in the positive mode (the upper portion of Table 3) corroborating the results obtained earlier for nonacidic lignin model compounds [59, 60]. Thus, the positive ESI mode is preferable as similar structural features, i.e., prevailing methoxy over phenolic hydroxyl functional groups, are also characteristic for alkali lignin (4.6 vs. 3.6 mol/1000 g as claimed by the supplier).

While both electrolyte systems seemed to show satisfactory performance, acids were more effective for the majority of species with no hydroxyl groups and prevailing methoxy groups (Table 3). It is of note that some of the standards showed low response or could not be detected in the negative ESI mode at all, e.g., VER, EST2, ET2, ET3-1, SA, ET3-2, P2, and G-β-2. This could perhaps be explained by the absence of hydroxyl groups or their steric hindrance (the structure motifs occurring in these molecules are shown in Table 2). For example, in the case of syringaldehyde (SA), two MeO groups in the ortho-position to the hydroxyl moiety made the deprotonation of these compounds difficult.

Highly hydroxylated monomeric phenolic standards with no more than one methoxy group (the bottom portion of Table 3), those containing carboxyl groups as well as other compounds of high acidity such as VA, V, and D2V, showed a higher ionization efficiency in the negative mode, as expected (Table 3). Nevertheless, as mentioned above, such a high ratio of hydroxylation/methoxylation is not characteristic for kraft lignin and thus does not seem to be suitable for selection of ionization conditions.

Lignin ESI: Electrolyte Selection

Similar to lignin model compounds, electrolyte screening was performed for efficient ionization of lignin itself over a broad range of commonly used salts and acids in both the positive and negative modes. The urgent goal was to ensure an effective ionization of higher MW species forming high m/z ions (Figure 3). Formic acid at its highest concentrations of 100 and 200 mmol L−1 in the positive ESI mode enhanced the formation of multiply charged species. Following the deconvolution (as explained in the next section), this protocol allowed for detecting the masses of up to 9000 Da (Figure 3b).

Figure 3
figure 3

The highest mass-to-charge (m/z) values of the ions detected in the mass spectrum of (a) 15–90 ppm kraft alkali lignin and (c) 15 ppm acetylated alkali lignin dissolved in ACN-water 1:1 in the positive and negative ionization modes while using different electrolytes; MW (Da) obtained after spectrum deconvolution of (b) 15–90 ppm kraft alkali lignin and (d) 15 ppm acetylated alkali lignin. The electrolyte concentration was 1.0 mmol L−1 unless specified otherwise under the horizontal axes of panels a-b. (a) kraft alkali lignin. The highest observed m/z prior to deconvolution. (b) kraft alkali lignin. The highest MW (Da) after deconvolution of the mass spectra. (c) Acetylated alkali lignin. The highest m/z prior to deconvolution. (d) Acetylated alkali lignin. The highest MW (Da) after deconvolution of the mass spectra. The horizontal lines are in each chart for the scale comparison in panels a-d and mark an m/z and M values of 4,000.

Full size image

Under these conditions, the ionization of lower MW lignin model compounds was somewhat suppressed, yet sufficient for their effective detection (cf. Figure 1), thus providing more balanced mass spectra.

Since acetylation is used in a number of studies to enhance lignin’s solubility, we investigated its ESI ionization as well. Similar to kraft alkali lignin, the positive mode was preferred for derivatized lignin when the hydroxyl groups were substituted with acetyloxy (CH3COO-) groups (Figure 3c). Lignin acetylation prevented a facile deprotonation in the negative mode (Figure 3c). The acetylated lignin spectrum also featured multiply charged species allowing for subsequent deconvolution (Figure 3d).

Lignin ESI: Mass Spectrum Deconvolution

ESI-HR TOF-MS of lignin in the positive mode in the presence of 100 mmol L−1 formic acid allowed for minimizing the ionization discrimination. Thus, lower MW species were mainly observed as singly charged ions (Supplementary Fig. 4), whereas high MW lignin constituents carried multiple charges. For example, vanillin was detected as a singly charged ion of 153.0570 m/z (protonated adduct [C8H8O3+H]+, 16 ppm mass accuracy error) (Figure 4a). The corresponding deconvoluted species possessed a mass of 152.05 Da (Figure 4b) calculated based on Eqs. 1 and 5, which was equal to vanillin’s MW.

$$ 153.0570\cdotp 1-\left[\left(1.007825-0.000549\right)\cdotp 1\right]=152.05\ \mathrm{Da} $$
(5)
Figure 4
figure 4

Positive ESI-HR TOF-MS of an 80-ppm solution of kraft alkali lignin in ACN-water (1:1) with 100 mmol L−1 formic acid: (a), (c), and (e) zoomed and full original mass spectra; (b), (d), and (f) zoomed and full mass spectra after deconvolution

Full size image

High MW species were predominantly observed as multiply charged ions; their occurrence was confirmed with IM ESI-HR TOF-MS as shown below in the section addressing ion mobility ESI MS. To illustrate this approach, the 597.2073 m/z ion carried a charge of + 7 (the spectrum is shown in Figure 4c). The mass spectrum deconvolution resulted in a calculated MW of 4173.40 for the corresponding species assuming the formation of exclusively protonated ions (Eq. 6), and the spectrum is shown in Figure 4d. There was a variety of lignin structures, which could be proposed for an MW of 4173.4 Da, e.g., (1) G14H8S1 linked through 7 linkages of β-β, β-5, β-1, 5-5 types and 16 linkages of 4-O-5 or β-O-4 types; (2) G20H4 linked through 22 linkages of β-β, β-5, β-1, 5-5 types and 1 linkage of 4-O-5 or β-O-4 types; and (3) G14H7S3 linked through 22 linkages of β-β, β-5, β-1, 5-5 types and 1 linkage of 4-O-5 or β-O-4 types, where G, H, and S stand for coniferyl alcohol, p-coumaryl alcohol, and sinapyl alcohol phenylpropanoid units, respectively. More examples of the multiply charges species are provided in Supplementary Fig. 5. Also, the increase in the allowed mass error tolerance and accounting for various charge carriers would significantly increase the number of possible structures.

It is of note that the formation of multiply charged ions is likely the result of both protonation and sodiation, and the profile of the deconvoluted mass spectrum remains the same if either proton or sodium is treated as a charge carrier. The difference up to 22 Da does not affect the MW distribution values described in the following section.

$$ 597.2073\cdotp 7-\left[\left(1.007825-0.000549\right)\cdotp 7\right]=4173.40\ \mathrm{Da} $$
(6)

Thus, analysis of lignin under the suggested ESI conditions allowed for a simultaneous determination of lignin constituents varying in size and structure via direct infusion analysis, with minimal sample preparation. An introduced feature of the multiply charged ion formation allowed for an efficient ionization of high MW lignin species, which, to our knowledge, was not previously reported. This approach is an essential contribution to the lignomics toolkit allowing for analysis of higher MW species, as shown in the next section.

Lignin MW Determination by ESI-HR TOF-MS

To assess lignin MW, we previously adapted an approach from NIST [75] employing Eqs. 2–4 to the deconvoluted spectral data [11]. This calculation is typically used for evaluating GPC data rather than MS; yet, the obtained values, 1480, 2520, and 3790 Da for M n , Mw, and M z , respectively, were in good agreement with the MW values determined earlier by GPC (Table 4) [11]. Deconvolution performed for protonated and sodiated ions resulted in the same MW values. Interestingly, similar MW values were determined for acetylated lignin, i.e., 1570, 2440, and 3530 Da for M n , Mw, and M z , respectively. This could be explained either by fragmentation with the loss leaving the most abundant molecular ions as ions [M-CH3CO+] or incomplete acetylation of lignin due to sterical hindrance, thus still detecting mostly unmodified lignin. In either case, it appears that the acetylated lignin MS data could also serve for the MW determination of the original, native lignin without a typical mass correction on acetylation.

Table 4 Number Average, Weight Average, and z-Average Molecular Weight of Lignin Determined by ESI-HR TOF-MS, GPC, and MALDI-HR TOF-MS
Full size table

Ionization: MALDI-HR TOF-MS Vs. ESI-HR TOF-MS

Similar mass spectra of kraft alkali lignin were recorded with MALDI and ESI as ionization sources (Figure 5 compared to Figure 4f). MALDI MS was selected as a reference technique frequently used for polymer MW determination [14, 38, 40, 42, 76,77,78]. To achieve the optimal lignin ESI, we evaluated the ionization efficiency with CHCA and HABA matrices and also without a matrix (Supplementary Fig. 6). The mass spectrum recorded with no matrix was of higher clarity, showing well-resolved peaks of higher intensity above 2500 m/z. Apparently, the matrix structurally similar to lignin increased the background and complexity of spectra, perhaps, due to the undesired association effects occurring during the sample co-crystallization with a matrix [30]. The recorded LDI mass spectrum (i.e., without a matrix) demonstrated several local maxima at 500, 1000, 2900, 4500, and 6200 m/z (Figure 5) with the decreasing signal at m/z values of 7000. The trends in the observed MALDI (LDI) and ESI-HR TOF mass spectra were similar; however, ESI-HR TOF-MS allowed for obtaining a cleaner deconvoluted spectra for the high MW species.

Figure 5
figure 5

LDI (no matrix)-HR TOF-MS of alkali lignin. The insert shows a zoomed in part of the spectrum in the range 2800–8000 m/z

Full size image

We evaluated the LDI MS data using the calculations shown in Eqs. 2–4. The determined MW values agreed with the results obtained with GPC and ESI-HR TOF-MS (Table 4). The MW elucidated while employing MALDI was shifted toward lower values, perhaps due to the suppression of high MW species ionization because their detection was limited by the predominant formation of singly charged ions.

Ion Mobility ESI MS: Confirmation of the Multiply Charged Species Formation

To confirm the formation of multiply charged species, kraft alkali lignin was analyzed with IM ESI-HR Q-TOF-MS (Figure 6a). Five distinguishable regions observed in the two-dimensional spectrum (drift time (ms) vs. m/z) suggested the occurrence of species carrying either varied charges or structural conformations (Figure 6a). The charge state seemed to be the driving force of the ion mobility separation as the species of different charge state were detected in the mass spectrum regions (Supplementary Fig. 7a). Thus, the mass spectrum extracted for the first region did not feature any multiply charged species, whereas in the second region, the sum of the intensities of the double-charged ions accounted for 8% of the total sum of the intensities of all ions observed in the spectrum (Supplementary Fig. 7b). The increase of the relative abundance of multiply charged species was observed in the third through fifth regions in the IM ESI-HR TOF mass spectrum (Supplementary Fig. 7c–f). In this case, both protons and sodium ions can serve as charge carriers resulting in the formation of multiply charged species.

Figure 6
figure 6

(a) Ion mobility image of a 100-ppm solution of kraft alkali lignin in ACN-water (1:1) with 100 mmol L−1 formic acid, recorded in the positive ESI mode. (b) Original and deconvoluted mass spectra of a 100-ppm solution of kraft alkali lignin in ACN-water (1:1) with 100 mmol L−1 formic acid recorded in the positive ESI mode. The blank spectrum was subtracted before deconvolution. An accurate deconvolution algorithm is described in Eq. 1

Full size image

These results have confirmed the formation of multiply charged lignin species and their separation in the drift tube resulting in the distinguishable regions in the mass spectrum (Figure 6a). Nevertheless, this assignment is not entirely unambiguous, and one of the series could potentially be related to a different class of lignin-related species such as glycolignins. However, contradicting to this speculation would be the stated relatively high purity of alkali lignin.

The MW distribution values determined by IM ESI-HR TOF-MS were in good agreement with the results obtained with ESI-HR TOF-MS when the IM feature was not utilized, e.g., 1460 Da compared to 1480 Da for M n with two methods, respectively. Species with a higher MW featured a higher abundance when analyzed by IM ESI-HR TOF-MS, as expected due to a better ion focusing typical for this technique. Notably, both deconvoluted spectra featured the same ion species observed after deconvolution.

With IM ESI-HR TOF-MS, high MW species with masses up to 20,000 Da were observed, while ESI-HR TOF-MS allowed for the detection of species only up to 10,000 Da. This difference (compared to HR TOF-MS with Agilent 6210) may be due to a higher sensitivity of the newer IM ESI Q-TOF instrument (Agilent 6560), possibly due to the IM feature allowing for a higher resolution and ion focusing.

Conclusions

An ESI-HR TOF-MS method for kraft alkali lignin analysis has been developed allowing for a simultaneous detection of both low and high MW species via direct infusion with minimal sample preparation. The most efficient ionization conditions were achieved in the positive ESI mode with 100 mmol L−1 formic acid as an electrolyte. For the first time, the formation of multiply charged ions promoting the ionization of high MW lignin species was shown. Determination of multiply charged ions was possible due to an inherently high resolving power of an applied HR TOF mass analyzer. To elucidate MW, the mass spectrum was deconvoluted. The obtained M n , Mw, and M z values of 1480, 2520, and 3790 Da, respectively, were in good agreement with those determined previously for similar samples by gel permeation chromatography. The presence of multiply charged lignin ions was confirmed by IM MS using IM ESI-HR TOF-MS. The developed method may extend the lignomics toolkit while targeting higher MW species.