Journal of Chromatography A 1695 (2023) 463935 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Selective extraction of levoglucosan and its isomers from complex matrices using ligand exchange-solid phase extraction for analysis by liquid chromatography-electrospray ionization-tandem mass spectrometry Nina Davtian a,∗ , Nuria Penalva a , Antoni Rosell-Melé a,b,1 , Joan Villanueva a,∗ a Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona, Bellaterra, Catalonia, Spain b Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia, Spain a r t i c l e i n f o a b s t r a c t Article history: The analysis of trace quantities of monosaccharide anhydrides (MAs) in sediments is complicated by the Received 30 December 2022 lack of fast and reliable technologies to selectively extract these water-soluble non-ionic compounds from Revised 14 March 2023 samples of complex composition. Here we describe a solid phase extraction method that takes advantage Accepted 15 March 2023 of the affinity between monosaccharide anhydrides (MAs) and immobilized Na+ ions related to ligand- Available online 17 March 2023 exchange processes (LE-SPE). The capacity factor of LE-SPE columns was enhanced by using non-aqueous Keywords: mobile phases such as DCM/MeOH mixtures. We have used the unique properties of LE-SPE columns to Biomass burning selectively extract MAs from lacustrine, coastal, and deep-sea oceanic sediment samples. The analytical Levoglucosan procedure produces extracts with low ion suppression effects (0–20%), resulting in ideal conditions for Sediment samples MAs quantification with LC-ESI-MS/MS systems irrespective of the sedimentary matrix and MAs concen- Liquid chromatography-electrospray tration. The analytical method yields repeatable concentration values (RSD of 9–23% for levoglucosan and ionization-tandem mass spectrometry 15–34% for mannosan and galactosan) and an IS recovery of 45–70%. The instrumental dynamic range is Ligand exchange-solid phase extraction 10–10 0 0 0 pg injected, but in practice, the methodological lower limit of quantification is constrained by sample contamination during processing. The combination of LE-SPE and LC-ESI-MS/MS has the potential to produce sensitive and reliable technologies to analyze saccharides and amino acids in environmental and biological samples. © 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 1. Introduction analytical approaches have been published using liquid (LC) or gas chromatography (GC) to separate and detect them at trace Levoglucosan (LEV), galactosan (GAL), and mannosan (MAN) are levels in aerosol and ice samples (see Table S1 for an overview). extensively used as selective tracers of fire-derived organic matter Published analytical methods typically involve an extraction step to the environment (see Fig. S1 for the chemical structures of LEV, followed by the direct analysis of crude extracts [12]. However, GAL, and MAN) [1–5]. These monosaccharide anhydrides (MAs) these methods produce unreliable results when applied to sam- are produced by pyrolysis of cellulose during natural and anthro- ples with complex matrices, such as soils and sediments. The pogenic fires and have been detected in polar regions (Greenland routine analysis of unpurified methanolic sediment extracts results [6] and Antarctica [7]), remote continental areas [8–10], and distal in matrix effects for LC-MS systems [13–15] and irreproducible oceanic areas [11]. derivatization of MAs and other low volatile compounds for The analysis of MAs in environmental samples has been sub- GC-MS systems [16–19]. However, techniques commonly used to ject to intense research over the last two decades and many purify and concentrate compounds of interest, such as solid phase extraction and liquid-liquid extraction, fail to selectively extract hydrophilic non-ionic compounds like MAs and other saccharides. ∗ Corresponding authors. Accordingly, new efficient and handy technologies are needed to E-mail addresses: nina.davtian@uab.cat (N. Davtian), juan.villanueva@uab.cat selectively concentrate trace amounts of saccharides from complex (J. Villanueva). 1 matrices. Prof. Antoni Rosell-Melé passed away in December 2021. https://doi.org/10.1016/j.chroma.2023.463935 0021-9673/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Monosaccharides have long been known to form co-ordination earth to completely fill the cell. On the top of the cell, a known complexes with metal ions in water [20,21] and in MeOH [22,23]. amount of the internal standard (IS) 13 C6 -Levoglucosan (10 ng, 50 Since 1961, ligand-exchange chromatography (LEC) on cation- μL of a solution of 200 ng/mL in MeOH) was added for quantifi- exchange resins in the metal form has been extensively used to cation. Sediment samples were previously extracted at 100°C with separate chiral isomers of saccharides [16,21,24]. In LEC, isomers hexane/acetone 1:1 using the solvent saver mode at 2 mL/min for are separated by the differential interaction of 2–3 hydroxyl groups 10 min to remove lipidic compounds [29]. Sediment samples were (with adequate spatial geometry) with immobilized metal ions then extracted at 100°C with MeOH using the solvent saver mode [20,21]. Since LEC is based on the ligand exchange of saccharide at 2 mL/min for 19 min to recover MAs. Methanolic extracts were and water molecules, compound separation is usually undertaken concentrated to dryness at 60°C under a N2 flow using a TurboVap in pure aqueous media [25–28]. However, the low energy involved (Biotage), and redissolved in 1 + 1 mL of MeOH. in ligand-exchange processes has limited their practical use for sample preparation purposes [27]. 2.4. LE-SPE purification method Here we describe a method to selectively separate MAs from sediment extracts, based on the interaction between MAs and The Amberchrom® 50WX8 resin was transformed in its sodium Na+ ions immobilized in a strong cation-exchange resin (Dowex form (Dowex-Na) by suspending it in a large excess of 5% NaCl 50WX8). We show that the strong affinity between MAs and im- in water. The supernatant was removed, and the resin was resus- mobilized ions also occurs in other solvents, such as MeOH and pended in MilliQ water. The resin was cleaned 2–3 times until dis- DCM/MeOH, in the absence of water. Since monosaccharides and coloration of the suspension liquid. A volume of 5 mL of resin in MeOH also form co-ordination complexes with cations [22,23], we water (equivalent to 4 g) was loaded into a 10 mL glass syringe. assume that the retention mechanism is also related to ligand- A glass-fiber filter and 0.4 mL of quartz wool were placed below exchange processes. We have optimized the conditions to concen- the resin to avoid loss of particles during sample processing. The trate quantitatively MAs using a ligand exchange-solid phase ex- column was conditioned sequentially with 30 mL of water, 15 mL traction column (LE-SPE). This novel technique has enabled us to of MeOH, and 10 mL of DCM/MeOH 9:1 with the occasional as- validate a sensitive and reproducible analytical protocol to quan- sistance of a mild vacuum suction. The replacement of water with tify MAs in lacustrine and oceanic sediments. MeOH resulted in a volume decrease of the resin from 5 mL to 3.8 mL. 2. Materials and methods The process of sample loading onto the LE-SPE column involved a sequential addition of 1 mL of DCM, 2 × 1 mL of methanolic 2.1. Standards, solvents, and reagents sediment extract, and 9 mL of DCM. The final solvent composi- tion (DCM/MeOH 5:1) was optimal to ensure quantitative reten- Pure standards of mannosan and galactosan were supplied by tion of MAs by the resin. The resulting mixture was percolated at Cayman Chemical Co. Levoglucosan standard was obtained from a controlled speed of <5 mL/min to ensure quantitative retention Merck, Spain. 13 C6 -Levoglucosan, 98%, was provided by Cambridge of MAs by the resin. The LE-SPE column was rinsed with 5 mL of Isotope Laboratories Inc. DCM/MeOH 9:1, and the resin was dried by applying vacuum and Methanol (MeOH), dichloromethane (DCM), n-hexane, and ace- N2 supply through the column for 3–6 min. The elution of MAs tone were of GC grade (SupraSolv®, Supelco). Acetonitrile (ACN) from the LE-SPE column involved a sequential addition of 2 mL of was of LC-MS analysis hypergrade (HiPerSolv Chromanorm® Ultra, MeOH and 10 mL of MilliQ water. The purified extract was then Supelco). MilliQ water was produced in house with a Millipore Co. collected into a polypropylene tube. The LE-SPE column was dis- system composed of an Elix prepurification unit and a MilliQ unit. carded after use (Text S1). Silica gel 60 (230–400 mesh), NH4 OH solution (25%), and NaCl (Suprapur grade) were provided by Merck Spain. The strong 2.5. Extract desalting and concentration cationic resin Amberchrom® 50WX8 (hydrogen form, 20 0–40 0 mesh, formerly Dowex® 50WX8) and the Mixed Ion Exchange Purified extracts were desalted using the AmberLite® Resin (AmberLite® MB20) were obtained from Merck Spain. The MB20 mixed-bed ion-exchange resin. Desalting columns were chelating resin Chelex® 10 0 (20 0–40 0 mesh, sodium form) was prepared by placing a quartz filter and 5 mL of the MB20 resin from Bio-Rad Laboratories Spain. (equivalent to 4 g) into 10 mL glass syringes and cleaned with 50 mL of MilliQ water before use. Sample extracts were let to 2.2. Sediment samples percolate through the column and residual retained MAs were recovered with 30 mL of MilliQ water. Desalted extracts were Sample EN651 Mix is a combination of deep-sea sediments collected into 50 mL polypropylene tubes and concentrated at from the tropical Atlantic (5–10°N, 21–36°W, 30 0 0–40 0 0 mbsl, 80°C under a N2 stream. Sample extracts were transferred to meters below sea level), with high CaCO3 (70%) and low TOC (0.6%) polypropylene injection vials and filtered with a 0.45 μm PTFE content. Sediment MERS ST7 was retrieved in the continental slope filter. The final sample volume was 200 μL of MeOH. off the Ebro River (Spain, 40.07°N, 1.53°E, 1476 mbsl) and has a low TOC (0.8%) and high CaCO3 (45%) content. Sample LUC (42.59°N, 2.6. LC-MS/MS analysis 7.11°W) is a TOC-rich lake sediment (TOC = 28%) with low CaCO3 content (<10%). MAs were detected using an LC-MS/MS system composed by a modular HPLC system (Agilent 1290 Infinity LC system with a 2.3. Extraction method quaternary pump, an automatic injector and a thermostated col- umn) coupled to a triple quadrupole detector (Agilent 6470A LC/TQ Sediments were freeze-dried and then homogenized with a model) using an electrospray ionization (ESI) interface. Both LC- Mixer Mill MM 400 (Retsch). Selective pressurized solvent extrac- ESI-MS/MS control and data acquisition were performed with the tion was performed using an ASE-350 (Dionex Thermo) equipped MassHunter Workstation software version 10.0 SR1 (Agilent). with 10 mL extraction cells and 60 mL collection vials. Extrac- Separation of LEV, MAN, and GAL was achieved with a Se- tion cells were filled from bottom to top with a glass-fiber filter, Quant® ZIC-HILICTM (150 × 2.1 mm, 3.5 μm particle size, Merck) 3 g of activated silica, 0.2–3.0 g of sediment, and diatomaceous column thermostated at 20°C with a mobile phase flow rate of 2 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 0.275 mL/min. The mobile phase was obtained by in-line mixing background MAs pollution using blank samples, 2) ion suppression MilliQ water (Solvent A) and ACN (Solvent B). A linear solvent gra- effect for MAN, GAL, LEV, and IS (Text S2), 3) total and mass recov- dient was programmed from 5% A to 20% A in 8 min [30]. At 8.1 ery of the IS, and 4) repeatability of the calculated MAs concentra- min, the column was cleaned from all injected compounds with tions. 80% A for 2 min. At 10.1 min, the column was reconditioned with only solvent B for 2 min and then under the original conditions 3. Results (5% A) for 13 min before the following analysis. The injected sam- ple volume was 10 μL and the sample vials were maintained at Results and processes used to optimize the instrumental and 16°C. LE-SPE conditions are summarized in Sections 3.1–3.4. Validation A post-column addition of 0.1 mL/min of a NH4 OH solution of the optimized conditions (as described in Sections 2.3–2.8) is (obtained by adding 1 mL of NH4 OH 25% to 1 L of MilliQ water) detailed in Section 3.5. was used to enhance the MS/MS signal. Data acquisition was pro- gramed between 4 and 8 min. The mobile phase was diverted to waste outside this runtime period. All MAs isomers (LEV, MAN, 3.1. LC-MS/MS detection setup and GAL) were detected in negative mode using the 161 → 101 m/z MS/MS transition. The IS (13 C6 -LEV) was detected using the Previously published methods used different adducts to detect 167 → 105 m/z MS/MS transition. The same MS/MS conditions MAs in both negative and positive mode (e.g., [M+F]− , [M+Cl]− , were used for all compounds: dwell time of 100 ms, fragmentor [M+CH3 COO]− , [M+HCOO]– , [M+Na]+ , and [M+Li]+ ) [30,32–36]. voltage of 110 V, collision energy of 35 V, and a cell acceleration We compared the sensitivity and selectivity of the LC-ESI-MS sys- voltage of 0 V. The conditions used at the ion source were the fol- tem upon post-column addition of HCOONH4 [30] and NH4 OH lowing: gas temperature of 300°C, gas flow of 4 L/min, nebulizer [7] to the mobile phase. The addition of HCOONH4 to the mo- pressure of 60 psi, sheath gas temperature of 380°C, sheath gas bile phase produced an intense [M+HCOO]– signal at m/z 207 flow of 12 L/min, and a capillary voltage of 50 0 0 V. but also provided poor MS/MS spectra, so we monitored the 207 → 207 transition with a low collision energy (10 V). When 2.7. Quantification of MAs and estimation of ion suppression adding NH4 OH to the eluent, we obtained optimal [M-H]– re- sponses using the 161 → 101 MS/MS transition. Both [M+HCOO]– All MAs concentrations were calculated using the peak area ra- and [M-H]– adducts yielded a similar sensitivity when analyzing tios between each analyte and the IS (13 C6 -LEV). The ratios were standard mixtures, resulting in lower limits of detection (LOD) converted to concentration values using 7 standard solutions of of 5–10 pg injected (Fig. 1a and b). However, we also found a MAs with IS. The calibration curve ranged from 1 to 10 0 0 ng/mL, poor selectivity and compromised sensitivity for the [M+HCOO]– corresponding to a dynamic range from 10 to 10 0 0 0 pg injected. adduct when analyzing sediment extracts (Fig. 1c). By contrast, we The calibration curve was calculated by adjusting a linear equation found a good selectivity for the [M-H]– adduct when analyzing the with the minimum squared method, weighted by the inverse of same sediment extracts (Fig. 1d). We thus selected the 161 → 101 the squared concentration (1/x2 ) [31]. The software used for peak MS/MS transition to quantify MAs for all subsequent experiments integration, run calibration, and quantification was the Quantita- and final method validation. tive Analysis Module of the MassHunter Workstation version 10.1 Quantification of the three MAs isomers requires their complete (Agilent). To account for changes in the LC-ESI-MS/MS response chromatographic separation because MS and MS/MS detectors are over an analytical run, several standard solutions were incorpo- unable to discriminate them. However, MAs show a low affinity rated. Several quality control analyses were distributed within the with commonly used reverse stationary phases (such as C18) and analytical run to monitor the stability of the LC-ESI-MS/MS detec- elute with the solvent front. Hopmans et al. [13] increased the tion system over time. retention of MAs by adding a weak ion-pair reagent to the mo- The estimates of lower limit of quantification (LOQ) were ob- bile phase (Et3 N). However, the use of ion-pair in LC-ESI-MS sys- tained by injecting between between 10 and 10 0 0 0 pg. The re- tems causes severe ion suppression issues, resulting in reduced ported LOQ was the lowest concentration that the linear cali- sensitivity. As shown in Table S1, a few published LC methods bration model could reproduce the nominal concentration within achieve adequate MAs isomer separation with LEC [28,37], HPAEC a tolerance of ±20%. Sedimentary MAs concentrations were ex- [14,35,38,40–43], and HILIC [30]. HPAEC requires the use of specific pressed in ng/gdw (grams of dry sediment weight). chromatographs, free of metallic parts, and uses MS-incompatible Ion suppression was estimated by measuring the peak area in- NaOH-containing mobile phases. Although specific equipment to crement associated to the standard addition to a sample solution. overcome these issues is commercially available, common LC-MS The signal increment was compared to the peak area measured for systems are not compatible with HPAEC and therefore we dis- the standard without sample matrix. Practical details are provided carded them for this method development. Since HILIC systems in Text S2. tend to produce optimal results when combined to ESI-MS detec- tors, we adapted the HILIC method described by Matějíček and 2.8. Method validation Vašíčková [30]. We achieved the same separation with a ZIC-HILIC column (instead of a ZIC-cHILIC column; Fig. 1). However, the re- To test the analytical robustness, the procedure was vali- sulting method suffered from a cumulative ion suppression effect dated with the three sediments described in Section 2.2, which when analyzing sediment samples. We solved the cumulative ion have different concentration ranges and matrix composition. Each suppression effect issue by implementing a cleaning step with wa- sediment sample was repeatedly analyzed in three independent ter/ACN 8:2 after MAs elution. batches with all three sediments in triplicate (n = 9). Triplicate blanks were included in each batch to determine MAs contami- 3.2. Ligand exchange-solid phase extraction (LE-SPE) nation issues during sample processing. The blanks were prepared similarly to sediment samples, but no sediment was packed in the Solid phase extraction (SPE) represents a practical approach for cell. Each of the three batches was analyzed on different days. sample preparation as it produces fast and reproducible meth- In addition to the calibration accuracy and the lack of chromato- ods to purify sample extracts and make them suitable for instru- graphic interferences, the following parameters were assessed: 1) mental analysis. We performed an initial screening to find the 3 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Fig. 1. Comparison of LC-MS/MS sensitivity and selectivity using post-column addition of HCOONH4 (m/z 207 → 207, left) and NH4 OH (m/z 161 → 101, right). (a) and (b) Standard mixtures close to the LOD and (c) and (d) sediment extract. An additional unidentified compound, systematically found in sediment samples, is indicated as “?” in panel 1d. Fig. 2. Percentage of MAs retained on Dowex 50WX8 and Chelex 100 ligand-exchange resins in different cationic forms after a suspension in DCM/MeOH 1:1. best combination of resin and counter cation capable of extract- SPE experiments, we loaded 5 mL of each in water into 10 mL ing MAs from a typical solvent extract in DCM/MeOH. To this end, glass syringes. Each chromatographic column was conditioned with aliquots of a strong cation-exchange resin (Dowex 50WX8, styrene the appropriate solvent, spiked with 1 μg of MAs, and then eluted divinylbenzene polymer with sulfonated groups) and a chelating with several solvent fractions of 2 mL. The experiment was re- resin (ChelexTM 100 styrene divinylbenzene polymer with iminodi- peated with three different solvent compositions (water, MeOH, acetic acid) were transformed to different cationic forms (Na+ , K+ , and DCM/MeOH 1:1) to identify the adequate loading and eluting Mg2+ , Ca2+ , Cu2+ , and Fe3+ ). Each resin (1 g of Dowex or 2 g of solvents. As shown in Fig. 3, MAs readily eluted from the Dowex- Chelex) was suspended in a solution of 1 μg of MAs in 24 mL Na resin when using water as the mobile phase. Dowex-Ca resin of DCM/MeOH 1:1 for 16 h. The proportion of MAs adsorbed to succeeded in partially retaining LEV and MAN with water. Remark- the resin in each experiment was quantified by LC-ESI-MS/MS. As ably, both Dowex-Na and Dowex-Ca resins displayed a far greater shown in Fig. 2, Dowex-Na and Dowex-Ca resins retained a signif- capacity to retain MAs when water is replaced with MeOH (Fig. 3c icant fraction of the MAs (15–32%). The Chelex resin retained less and d). The use of DCM/MeOH 1:1 further improved the retention than 10% of each MAs in most cases. In addition, transition metals of MAs by both Dowex-Na and Dowex-Ca resins (Fig. 3e and f). We (Fe and Cu) did not perform well in all cases in DCM/MeOH. thus conclude that the interaction of MAs with Ca2+ or Na+ ions To evaluate whether Dowex-Na and Dowex-Ca resins could pro- immobilized in the ion-exchange resin is strong enough to selec- vide enough retention capacity to quantitatively load/elute MAs in tively extract MAs from a DCM/MeOH solution using a SPE column. 4 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Fig. 3. Elution profiles of MAs in LE-SPE columns containing 5 mL of Dowex-Na (left) and Dowex-Ca (right) resins. Solvents tested were water (a) and (b), MeOH (c) and (d), and DCM/MeOH 1:1 (e) and (f). Values displayed at the top right of each graph indicate the cumulative recovery considering all fractions. We selected the Dowex-Na resin to further optimize SPE conditions ganic compounds, such as MeOH and saccharides is well docu- because MAs can be retrieved by eluting with a smaller volume of mented [21,22]. Therefore, the exchange of Na+ ligands (MeOH water (10 mL, Fig. 3a and b). and MAs) can account for the observed retention/elution of MAs The retention mechanism of MAs by the Dowex-Na resin is in MeOH-based solvents (Fig. 3). Increasing the hydrophobicity of mainly related to complexation of saccharides to immobilized Na+ the loading solvent with DCM further reduces MAs losses by low- (or Ca2+ ) ions in MeOH, in a similar process to that described ering the affinity of the mobile phase to hydrophilic MAs com- for water-based chromatographic systems, such as LEC (ligand ex- pounds (Fig. 4). After careful optimization, we concluded that the change chromatography) [20,21,25] and AEC (anion exchange chro- optimal solvents to load and elute MAs from LE-SPE Na+ columns matography) [27]. A comparison experiment showed that, un- were DCM/MeOH 5:1 and water, respectively. der the same experimental conditions, the Dowex-Na resin re- tained 78–87% of MAs, while the Dowex-H resin only retained 3.3. Optimization of the selective pressurized liquid extraction 15–18% of MAs. Therefore, only a small proportion of MAs re- tained by the Dowex-Na (and Dowex-Ca) resins in DCM/MeOH is A number of studies have used pressurized liquid extrac- related to hydrophilic/hydrophobic interactions with the styrene- tion using MeOH [10,13–15,40,44–46] or DCM/MeOH mixtures divinylbenzene copolymer of the resin. The existence of co- [9,11,47–51] to obtain crude MAs-containing sedimentary extracts. ordination complexes between Na+ /Ca2+ ions and hydroxylated or- However, the use of DCM/MeOH mixtures yields increased matrix 5 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Residual interfering ions in the sample extract thus compromised the sensitivity of the analysis and substantially biased the quan- tification of MAs. To address this issue, we implemented a sample desalting step with the Amberlite MB20 mixed-bed ion-exchange resin. This sample desalting step also eliminated interferences in LC-ESI-MS/MS chromatograms (Fig. S2) when analyzing CaCO3 -rich sediments. 3.5. Method validation 3.5.1. Linearity and sensitivity The LC-ESI-MS/MS method provided linear responses in the range from 10 to 10 0 0 0 pg injected (Table 1 and Fig. S3) with a LOD of approximately 5 pg injected. This instrumental sensitivity is comparable to the ones reported by Matějíček and Vašíčková [30] (LOD of 4–9 pg injected and LOQ of 10–30 pg injected) and Sanz Rodriguez et al. [35] (LOD of 5–25 pg injected and LOQ of 15–76 pg injected), which are among the best published instru- mental sensitivities in terms of injected MAs amounts (Table S1). Only Gambaro et al. [7], You et al. [39], and Yao et al. [36] achieved better instrumental sensitivities by two orders of magnitude (LODs Fig 4. Relationship between loading solvent composition and total MAs recovery of 0.30, 0.55, 0.20 pg injected, respectively), though at the expense using a LE-SPE column containing 5 mL of Dowex-Na. Error bars represent standard of LEV separation from MAN and GAL. deviations of duplicates. 3.5.2. Repeatability The analytical method provided repeatable MAs concentra- effects for LC-ESI-MS systems [13] and poor recoveries [46] com- tion results, irrespective of the sedimentary CaCO3 and TOC con- pared with methanolic sediment extracts. We further investigated tent, over a wide MAs concentration range (Table 2). The LUC these issues by testing hexane/acetone 1:1 and DCM/MeOH from sediment yielded the highest MAs concentrations (LEV/MAN/GAL 9:1 to 0:10 for selective pressurized liquid extraction (SPLE) 717/162/34.3 ng/gdw), the EN651 Mix sediment yielded the low- from spiked blanks and sediment samples from a TOC-rich lake est MAs concentrations (LEV/MAN/GAL 6.6/0.7/0.3 ng/gdw), and (TOC = 28 %). The other SPLE conditions are identical to those the MERS ST7 sediment yielded MAs concentrations twice as high described in Section 2.3. Spiked blanks proved that the acti- as those of the EN651 Mix sediment (LEV/MAN/GAL 13.6/1.4/0.7 vated silica gel quantitatively retains MAs and IS when using ng/gdw). Global RSDs for LEV ranged between 9% and 23%, with hexane/acetone 1:1, which enables a pre-purification step by the poorest repeatability for the deep-sea sediment (EN651 Mix). SPLE. Adequate MAs recoveries from the silica gel required highly Global RSDs for MAN and GAL were consistently poorer compared hydrophilic extraction solvents (20–100% MeOH, Fig. 5a). Similarly, to LEV (15–34%), likely related to the absence of isotopically la- the total IS recovery from the TOC-rich sediment maximized with beled IS for these MAs isomers. DCM/MeOH mixtures containing more than 40% MeOH (Fig. 5b). Total IS recoveries for CaCO3 -rich sediments remained consis- 3.5.3. Blank contributions tently low (10–30%, Fig. 5b) when using DCM/MeOH 6:4. Further Atmospheric MAs in the laboratory environment is a source of experiments involving sequential extraction with DCM/MeOH 6:4, contamination during sample processing [7,8,39,54–56]. We found MeOH, and water confirmed that MAs extraction with DCM/MeOH a systematic concentration background in the final extract in the 6:4 was not complete for CaCO3 -rich sediments. The reduced re- range of 5–30 ng/mL for LEV and 0.2–10 ng/mL for MAN and GAL. covery when extracting CaCO3 -rich sediments is likely related to Therefore, blank contributions should be considered for reliable strong interactions between MAs and cations present in the inor- MAN, GAL, and LEV quantifications (Table 2). During this valida- ganic matrix (i.e., Ca2+ ions). In situations where IS and MAs are tion exercise, we found that contamination during sample process- strongly retained by the stationary phase, adding the IS on top ing could contribute significantly (up to 29%) to calculated MAs of the sediment can lead to an underestimation of the IS recov- concentrations. Accordingly, these blank issues impose a practical ery. These low recoveries have been reported to produce signifi- lower limit of quantification of the analytical procedure (about 5– cant overestimation of concentration values in SPLE methods [52]. 10 ng/gdw in sediment), which is significantly higher than the in- Nevertheless, our CaCO3 -rich sediment extracts in DCM/MeOH 6:4 strumental LOQ (0.1 ng/gdw assuming 60% recovery and 3 g of and MeOH yielded consistent differences in total IS recoveries in sediment). Therefore, achieving extremely low analytical LOQ re- favor of the methanolic extracts, so we selected MeOH as the SPLE quires processing samples in a clean room [7,39] or under a lami- extraction solvent for MAs. nar flow bench [7,8,54–56]. 3.4. Extract desalting 3.5.4. Total and mass IS recovery and ion suppression effect Total recoveries (TR) were estimated by comparing the IS sig- During method development, we found that some extracts pro- nal obtained for processed samples with a standard solution. The duced reduced MAs signals due to ion suppression of the ESI- average TR values obtained during the validation exercise ranged MS/MS signal by interfering compounds. Signal losses occurred between 39% and 61% (Table 3). Remarkably, blank and sediment randomly with both blank and sediment extracts. We attributed samples provided similar TR, indicating that the relatively low TR this issue to accidental leaching of Na+ (and other ions) from is independent of the sample matrix (i.e., CaCO3 and TOC content). the Dowex-Na resin during LE-SPE treatment. These cations form The TR is a combination of the mass recovery (MR, the pro- extremely stable adducts at the ESI interface even at trace lev- portion of MAs recovered after sample processing) and the ma- els [30,36,43,53], resulting in severe signal suppression problems. trix effect (mainly related to ion suppression at the ESI interface; 6 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Fig. 5. Solvent composition effect on total MAs and IS recovery with selective pressurized extraction. (a) Total MAs and IS recovery from spiked blanks and (b) total IS recovery from sediment samples. Error bars represent standard deviations (n = 4–9). Table 1 Linearity and accuracy of the method. MAN GAL LEV Calibration standard mixtures Correlation coefficient (R2 )a 0.995–0.999 0.990–1.000 0.991–0.999 Accuracy (%) 89–111 94–118 91–113 Quality control standard mixture Batch 1 accuracy (mean ± SD, %)b 103 ± 12 95 ± 14 93 ± 3 Batch 2 accuracy (mean ± SD, %)b 106 ± 7 96 ± 8 101 ± 4 Batch 3 accuracy (mean ± SD, %)b 104 ± 6 99 ± 6 106 ± 6 Global accuracy (mean ± SD, %)c 105 ± 9 97 ± 10 100 ± 7 a Linear regression adjusted with the minimum squared method, weighted by the in- verse of the squared concentration (1/x2 ). b Calculated from 7 injections of 200 pg per independent batch. c Calculated from 21 injections of 200 pg over 3 independent batches. Table 2 Repeatability of calculated MAN, GAL, and LEV concentrations and blank contributions. MAs concentration [mean, ng/gdw (%RSD)]a Blank contribution (%)a Sample MAN GAL LEV MAN GAL LEV LUC (lacustrine sediment) 162.2 (15%) 34.3 (16%) 717.4 (9%) 1 3 2 MERS ST7 (coastal sediment) 1.4 (26%) 0.7 (30%) 13.6 (14%) 11 10 8 EN651 Mix (deep-sea sediment) 0.7 (34%) 0.3 (30%) 6.6 (23%) 29 23 18 a Calculated from 9 replicates over 3 independent batches. Table 3 Total and mass IS recoveries and ion suppression effect on MAN, GAL, LEV, and IS signals. IS recovery (mean ± SD, %)a Ion suppression effect (mean ± SD, %)a Sample Total recovery Mass recovery MAN GAL LEV IS Blank 42 ± 10 45 ± 10 3±5 2±5 12 ± 7 5±5 LUC (lacustrine sediment) 61 ± 7 70 ± 12 2±8 1 ± 10 4 ± 10 11 ± 7 MERS ST7 (coastal sediment) 55 ± 9 62 ± 7 12 ± 7 18 ± 7 7±9 11 ± 9 EN651 Mix (deep-sea sediment) 39 ± 11 47 ± 11 15 ± 7 19 ± 10 12 ± 8 16 ± 10 a Calculated from 9 replicates over 3 independent batches. 7 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 Text S2). Estimates of the ion suppression, obtained by measuring MAs in sediment extracts with a reasonable efficiency (mass recov- the signal increment caused by the addition of a known amounts eries of 45–70%, Table 3). However, we believe that other ligand- of MAs (and IS), remained at acceptable levels for all MAs in all exchange-based approaches can be developed to purify and con- samples considered (0–20%, Table 3). Low ion suppression values centrate other saccharides and amino acids in complex matrices, show that LE-SPE provides MAs extracts free of interferents for LC- such as biological and environmental samples. ESI-MS/MS. Therefore, the relatively low TR estimates are caused mainly by MAs losses during sample processing. 5. Conclusions The MR of the IS, as estimated by removing the ion suppression effect from the TR, ranged between 45% and 70% (Table 3), with We describe a solid phase extraction method that takes advan- the lowest values obtained for the blank samples. Further tests di- tage of the affinity between MAs and immobilized Na+ ions related rected at improving the MR showed that the final MR was the re- to ligand-exchange processes in non-aqueous, methanolic solvents. sult of small losses that occurred at every step of the analytical The capacity of LE-SPE columns to retain MAs can be enhanced setup and no significant improvement could be achieved. by increasing the hydrophobicity of the mobile phase with DCM. We have applied the unique properties of LE-SPE columns to se- 4. Discussion lectively extract MAs from samples of complex composition such as sediment extracts. The analytical procedure produces extracts Trace amounts of ions in the mobile phase can produce drastic with low matrix effect and is suitable for reliable MAs quantifica- changes in mass spectra of MAs in LC-ESI-MS systems. Although tion in sediments with LC-ESI-MS/MS systems. The resulting ana- this problem can be partially solved by promoting specific adduct lytical method provides robust and repeatable concentration values forms, such as the [M-H]– adduct by adding NH4 OH to the mobile for LEV, MAN, and GAL irrespective of the sample matrix and is ap- phase in this study, the instrumental response factor in multiple plicable to a wide concentration range. This analytical method rep- reaction monitoring or selected ion monitoring mode can still be resents a proof of concept, and other ligand-exchange-based SPE severely affected during the analysis of samples with high ionic protocols may be developed to quantify trace amounts of other compound concentrations. Analytical protocols based on LC-ESI- saccharides and amino acids in biological and environmental sam- MS require proper evaluation of the ion suppression by the sed- ples. imentary matrix, including the sensitivity, accuracy, and precision in real samples [57]. In principle, accuracy can be partially en- Declaration of competing interest sured by using an isotopically labelled IS that compensates for changes in the instrumental response factor. In the case of crude The authors declare that they have no known competing finan- sediments extracts, the analytical method suffered from major ion cial interests or personal relationships that could have appeared to suppression issues that hampered the detection of MAs. There- influence the work reported in this paper. fore, the analysis of MAs by LC-ESI-MS required exhaustive elim- ination of salts and other potential interferents from the sediment CRediT authorship contribution statement extract. Previous studies implemented crude desalting methods, such Nina Davtian: Writing – original draft, Investigation, Validation, as sample percolation through cotton wool [46], Na2 SO4 columns Formal analysis, Visualization. Nuria Penalva: Validation. Antoni [13,46], or with ACN [13] or DCM/MeOH 9:1 [46]. Another ex- Rosell-Melé: Funding acquisition. Joan Villanueva: Writing – orig- isting MAs purification protocol employed centrifugation and C18 inal draft, Writing – review & editing, Visualization, Methodology, columns [10]. However, we have shown that the use of apolar Supervision. solvents and Na2 SO4 columns may lead to significant MAs losses and introduce interfering ions. The efficiency of these purification Data availability methods cannot be evaluated because matrix effect and ion sup- pression were not reported. All data generated from this study are presented in Figs. 1–5, The hydrophilic character of saccharides makes impractical the S2, and S3 and Tables 1–3. Detailed method validation re- use of established methods to purify and concentrate sample ex- sults are available in Tables S2–S5. tracts. For example, liquid-liquid extraction fails to separate MAs from salts and other water-soluble interferents. Also, the low affin- Funding ity between saccharides and commercially available chromato- graphic stationary phases limits the applicability of SPE protocols. This work was supported by the European Research On the other hand, ligand-exchange chromatography is commonly Council (PALADYN Project #834934). This work contributes applied to separate and identify isomeric forms of saccharides and to the ICTA-UAB “María de Maeztu” Programme for Units amino acids, but these systems typically have low retention capac- of Excellence (CEX2019-0 0 0940-M) funded by MCIN/AEI/ ities that have hampered their applicability in sample purification 10.13039/50110 0 011033. ND is supported by the Juan de la processes [27]. We have demonstrated that the retention capacity Cierva-Formación grant FJC2020-044864-I funded by MCIN/AEI/ of ligand-exchange systems is enhanced in a methanolic context, 10.13039/50110 0 011033 and by “European Union NextGenera- compared to traditionally used water-based LC systems (Figure 3). tionEU/PRTR”. In this context, the elution capacity of the mobile phase can be fur- ther reduced by adding hydrophobic organic solvents to the mo- Acknowledgements bile phase, such as DCM. Ligand-exchange interactions between MAs and immobilized Na+ ions occurred even in highly hydropho- We thank the Editor and two anonymous reviewers for their bic solvents, such as DCM/MeOH 9:1 (Figure 4). Since ligand- help to substantially improve this manuscript. exchange interaction is highly specific to polyhydroxylated com- pounds such as saccharides, purified extracts were free of interfer- Supplementary materials ing compounds commonly found in sediments and soils, such as fulvic acids, lipids, and salts. In this study, we have demonstrated Supplementary material associated with this article can be that LE-SPE can be applied to selectively extract trace amounts of found, in the online version, at doi:10.1016/j.chroma.2023.463935. 8 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 References [24] V.A. Davankov, CHIRAL SEPARATIONS | ligand exchange chromatography, in: I.D. Wilson (Ed.), Encyclopedia of Separation Science, Academic Press, Oxford, [1] B.R.T. Simoneit, A review of biomarker compounds as source indicators and 20 0 0, pp. 2369–2380, doi:10.1016/B0- 12- 226770- 2/03111-2. tracers for air pollution, Environ. Sci. Pollut. Res 6 (1999) 159–169, doi:10.1007/ [25] H. Caruel, L. Rigal, A. Gaset, Carbohydrate separation by ligand-exchange liq- BF02987621. uid chromatography: correlation between the formation of sugar-cation com- [2] B.R.T. Simoneit, J.J. Schauer, C.G. Nolte, D.R. Oros, V.O. Elias, M.P. Fraser, plexes and the elution order, J. Chromatogr. A 558 (1991) 89–104, doi:10.1016/ W.F. Rogge, G.R. Cass, Levoglucosan, a tracer for cellulose in biomass burn- 0021- 9673(91)80114- V. ing and atmospheric particles, Atmos. Environ. 33 (1999) 173–182, doi:10.1016/ [26] H.F. Walton, Ligand-exchange chromatography: a brief review, Ind. Eng. Chem. S1352- 2310(98)00145- 9. Res. 34 (1995) 2553–2554, doi:10.1021/ie0 0 047a0 01. [3] G. Engling, P. Herckes, S.M. Kreidenweis, W.C. Malm, J.L. Collett, Composition of [27] M. Stefansson, D. Westerlund, Ligand-exchange chromatography of carbohy- the fine organic aerosol in Yosemite national park during the 2002 yosemite drates and glycoconjugates, J. Chromatogr. A 720 (1996) 127–136, doi:10.1016/ aerosol characterization study, Atmos. Environ. 40 (2006) 2959–2972, doi:10. 0 021-9673(95)0 0276-6. 1016/j.atmosenv.2005.12.041. [28] R.W. Dixon, G. Baltzell, Determination of levoglucosan in atmospheric aerosols [4] E.D. Vicente, C.A. Alves, An overview of particulate emissions from residential using high performance liquid chromatography with aerosol charge detec- biomass combustion, Atmos. Res. 199 (2018) 159–185, doi:10.1016/j.atmosres. tion, J. Chromatogr. A 1109 (2006) 214–221, doi:10.1016/j.chroma.2006.01. 2017.08.027. 021. [5] L. Marynowski, B.R.T. Simoneit, Saccharides in atmospheric particulate and [29] M. Raja, J. Villanueva, C. Moreu-Romero, M. Giaime, A. Rosell-Melé, Fast quan- sedimentary organic matter: status overview and future perspectives, Chemo- titative analysis of n-alkanes, PAHs and alkenones in sediments, Org. Geochem. sphere 288 (2022) 132376, doi:10.1016/j.chemosphere.2021.132376. 171 (2022) 104471, doi:10.1016/j.orggeochem.2022.104471. [6] N. Kehrwald, R. Zangrando, P. Gabrielli, J.-L. Jaffrezo, C. Boutron, C. Barbante, [30] D. Matějíček, J. Vašíčková, Hydrophilic interaction chromatography with A. Gambaro, Levoglucosan as a specific marker of fire events in Greenland Fourier transform mass spectrometry of monosaccharide anhydrides, J. Sep. snow, Tellus B: Chem. Phys. Meteorol. 64 (2012) 18196, doi:10.3402/tellusb. Sci. 39 (2016) 4748–4755, doi:10.1002/jssc.201601038. v64i0.18196. [31] H. Gu, G. Liu, J. Wang, A.-F. Aubry, M.E. Arnold, Selecting the correct weighting [7] A. Gambaro, R. Zangrando, P. Gabrielli, C. Barbante, P. Cescon, Direct deter- factors for linear and quadratic calibration curves with least-squares regres- mination of levoglucosan at the picogram per milliliter level in Antarctic sion algorithm in bioanalytical LC-MS/MS assays and impacts of using incor- ice by high-performance liquid chromatography/electrospray ionization triple rect weighting factors on curve stability, data quality, and assay performance, quadrupole mass spectrometry, Anal. Chem. 80 (2008) 1649–1655, doi:10. Anal. Chem. 86 (2014) 8959–8966, doi:10.1021/ac5018265. 1021/ac701655x. [32] E.C.H. Wan, J.Z. Yu, Analysis of sugars and sugar polyols in stmospheric [8] P. Yao, V.F. Schwab, V.-N. Roth, B. Xu, T. Yao, G. Gleixner, Levoglucosan concen- serosols by chloride sttachment in liquid chromatography/negative Ion electro- trations in ice-core samples from the Tibetan Plateau determined by reverse- spray mass spectrometry, Environ. Sci. Technol. 41 (2007) 2459–2466, doi:10. phase high-performance liquid chromatography–mass spectrometry, J. Glaciol. 1021/es062390g. 59 (2013) 599–612, doi:10.3189/2013JoG12J157. [33] C. Piot, J.-L. Jaffrezo, J. Cozic, N. Pissot, I. El Haddad, N. Marchand, J.-L. Be- [9] A. Callegaro, D. Battistel, N.M. Kehrwald, F. Matsubara Pereira, T. Kirchge- sombes, Quantification of levoglucosan and its isomers by high performance org, M. del, C. Villoslada Hidalgo, B.W. Bird, C. Barbante, Fire, vegetation, liquid chromatography – electrospray ionization tandem mass spectrometry and Holocene climate in a southeastern Tibetan lake: a multi-biomarker re- and its applications to atmospheric and soil samples, Atmos. Meas. Tech. 5 construction from Paru Co, Climate Past 14 (2018) 1543–1563, doi:10.5194/ (2012) 141–148, doi:10.5194/amt- 5- 141- 2012. cp- 14- 1543- 2018. [34] F. Mashayekhy Rad, S. Spinicci, S. Silvergren, U. Nilsson, R. Westerholm, Valida- [10] B. Aichner, B. Wünnemann, A. Callegaro, M.T.J. van der Meer, D. Yan, Y. Zhang, tion of a HILIC/ESI-MS/MS method for the wood burning marker levoglucosan C. Barbante, D. Sachse, Asynchronous responses of aquatic ecosystems to hy- and its isomers in airborne particulate matter, Chemosphere 211 (2018) 617– droclimatic forcing on the Tibetan Plateau, Commun Earth Environ 3 (2022) 623, doi:10.1016/j.chemosphere.2018.07.188. 1–11, doi:10.1038/s43247-021-00325-1. [35] E. Sanz Rodriguez, M.M.G. Perron, M. Strzelec, B.C. Proemse, A.R. Bowie, [11] L.T. Schreuder, E.C. Hopmans, J.-B.W. Stuut, J.S. Sinninghe Damsté, S. Schouten, B. Paull, Analysis of levoglucosan and its isomers in atmospheric samples by Transport and deposition of the fire biomarker levoglucosan across the tropical ion chromatography with electrospray lithium cationisation - triple quadrupole North Atlantic Ocean, Geochim. Cosmochim. Acta 227 (2018) 171–185, doi:10. tandem mass spectrometry, J. Chromatogr. A 1610 (2020) 460557, doi:10.1016/ 1016/j.gca.2018.02.020. j.chroma.2019.460557. [12] K. Janoszka, M. Czaplicka, Methods for the determination of levoglucosan [36] X. Yao, N. Wang, X. Zheng, Q. Li, E. Santos, L. Maharjan, J. Wang, Z. Guo, and other sugar anhydrides as biomass burning tracers in environmental J. Guo, H. Zhang, K. Zheng, J. Wu, Y. Li, Highly sensitive ultra-performance liq- samples – a review, J. Sep. Sci. 42 (2019) 319–329, doi:10.1002/jssc.20180 uid chromatography coupled with triple quadrupole mass spectrometry detec- 0650. tion method for levoglucosan based on Na+ enhancing its ionization efficiency, [13] E.C. Hopmans, R.A.L. dos Santos, A. Mets, J.S. Sinninghe Damsté, S. Schouten, RSC Adv. 13 (2023) 7030–7036, doi:10.1039/D2RA07419B. A novel method for the rapid analysis of levoglucosan in soils and sed- [37] A. Nouara, C. Panagiotopoulos, J. Balesdent, K. Violaki, E. Bard, Y. Fagault, iments, Org. Geochem. 58 (2013) 86–88, doi:10.1016/j.orggeochem.2013.02. D.J. Repeta, R. Sempéré, Liquid chromatographic isolation of individual carbo- 003. hydrates from environmental matrices for stable carbon analysis and radiocar- [14] T. Kirchgeorg, S. Schüpbach, N. Kehrwald, D.B. McWethy, C. Barbante, Method bon dating, Anal. Chim. Acta 1067 (2019) 137–146, doi:10.1016/j.aca.2019.03. for the determination of specific molecular markers of biomass burning in 028. lake sediments, Org. Geochem. 71 (2014) 1–6, doi:10.1016/j.orggeochem.2014. [38] S. Patil, J. Rohrer, A fast, single column high-performance anion exchange chro- 02.014. matography with pulsed amperometric detection method for the determina- [15] E. Norström, J. West, K. Kouli, C. Katrantsiotis, M. Hättestrand, R.H. Smit- tion of saccharides in atmospheric aerosol samples, J. Sep. Sci. 45 (2022) 3422– tenberg, Evaluation of anhydrosugars as a molecular proxy for paleofire ac- 3430, doi:10.1002/jssc.202100842. tivity: a case study on a Holocene sediment core from Agios Floros, Pelo- [39] C. You, L. Song, B. Xu, S. Gao, Method for determination of levoglucosan in ponnese, Greece, Org. Geochem. 153 (2021) 104193, doi:10.1016/j.orggeochem. snow and ice at trace concentration levels using ultra-performance liquid chro- 2021.104193. matography coupled with triple quadrupole mass spectrometry, Talanta 148 [16] S.C. Churms, Recent developments in the chromatographic analysis of carbo- (2016) 534–538, doi:10.1016/j.talanta.2015.11.030. hydrates, J. Chromatogr. A 500 (1990) 555–583, doi:10.1016/S0 021-9673(0 0) [40] S. Schüpbach, T. Kirchgeorg, D. Colombaroli, G. Beffa, M. Radaelli, 96092-8. N.M. Kehrwald, C. Barbante, Combining charcoal sediment and molecular [17] R.P. Evershed, Advances in Silylation, in: K. Blau, J.M. Halket (Eds.), Handbook markers to infer a Holocene fire history in the Maya Lowlands of Petén, of Derivatives for Chromatography, 2nd edition, John Wiley & Sons, Chichester, Guatemala, Quat. Sci. Rev. 115 (2015) 123–131, doi:10.1016/j.quascirev.2015.03. 1993, pp. 51–108. 004. [18] J.L. Little, Artifacts in trimethylsilyl derivatization reactions and ways to avoid [41] A. Caseiro, I.L. Marr, M. Claeys, A. Kasper-Giebl, H. Puxbaum, C.A. Pio, Deter- them, J. Chromatogr. A 844 (1999) 1–22, doi:10.1016/S0 021-9673(99)0 0267-8. mination of saccharides in atmospheric aerosol using anion-exchange high- [19] C.D. Simpson, R.L. Dills, B.S. Katz, D.A. Kalman, Determination of levoglucosan performance liquid chromatography and pulsed-amperometric detection, J. in atmospheric fine particulate matter, J. Air Waste Manage. Assoc. 54 (2004) Chromatogr. A 1171 (2007) 37–45, doi:10.1016/j.chroma.2007.09.038. 689–694, doi:10.1080/10473289.2004.10470945. [42] K. Saarnio, K. Teinilä, M. Aurela, H. Timonen, R. Hillamo, High-performance [20] R.W. Goulding, Liquid chromatography of sugars and related polyhydric alco- anion-exchange chromatography–mass spectrometry method for determina- hols on cation exchangers: the effect of cation variation, J. Chromatogr. A 103 tion of levoglucosan, mannosan, and galactosan in atmospheric fine par- (1975) 229–239, doi:10.1016/S0 021-9673(0 0)87215-5. ticulate matter, Anal. Bioanal. Chem 398 (2010) 2253–2264, doi:10.1007/ [21] J. Briggs, P. Finch, M.C. Matulewicz, H. Weigel, Complexes of copper(II), cal- s00216- 010- 4151- 4. cium, and other metal ions with carbohydrates: thin-layer ligand-exchange [43] D. Asakawa, Y. Furuichi, A. Yamamoto, Y. Oku, K. Funasaka, Rapid and sen- chromatography and determination of relative stabilities of complexes, Carbo- sitive quantification of levoglucosan in aerosols by high-performance anion- hydr. Res. 97 (1981) 181–188, doi:10.1016/S0 0 08-6215(0 0)80664-3. exchange chromatography with positive electrospray ionization mass spec- [22] J.A.Jr. Rendleman, Alkali metal complexes of carbohydrates. I. Interaction of al- trometry (HPAEC-positive ESI-MS), Atmos. Environ. 122 (2015) 183–187, doi:10. kali metal salts with carbohydrates in alcoholic media, J. Org. Chem. 31 (1966) 1016/j.atmosenv.2015.09.045. 1839–1845, doi:10.1021/jo01344a038. [44] R.A. Lopes dos Santos, P. De Deckker, E.C. Hopmans, J.W. Magee, A. Mets, [23] D.W.S. Wong, J.M. Randall, W.M. Camirand, R.H. Edwards, Sucrose extraction J.S. Sinninghe Damsté, S. Schouten, Abrupt vegetation change after the Late from beet by methanolic calcium chloride, JSBR 30 (1993) 57–69, doi:10.5274/ Quaternary megafaunal extinction in southeastern Australia, Nat. Geosci. 6 jsbr.30.1.57. (2013) 627–631, doi:10.1038/ngeo1856. 9 N. Davtian, N. Penalva, A. Rosell-Melé et al. Journal of Chromatography A 1695 (2023) 463935 [45] E. Dietze, K. Mangelsdorf, A. Andreev, C. Karger, L.T. Schreuder, E.C. Hopmans, [51] C. Häggi, E.C. Hopmans, E. Schefuß, A.O. Sawakuchi, L.T. Schreuder, D.J. Bertas- O. Rach, D. Sachse, V. Wennrich, U. Herzschuh, Relationships between low- soli Jr., C.M. Chiessi, S. Mulitza, H.O. Sawakuchi, P.A. Baker, S. Schouten, Neg- temperature fires, climate and vegetation during three late glacials and inter- ligible quantities of particulate low-temperature pyrogenic carbon reach the glacials of the last 430 kyr in northeastern Siberia reconstructed from Atlantic Ocean via the Amazon River, Global Biogeochem. Cycles 35 (2021) monosaccharide anhydrides in Lake El’gygytgyn sediments, Climate Past 16 e2021GB006990, doi:10.1029/2021GB006990. (2020) 799–818, doi:10.5194/cp- 16- 799- 2020. [52] S. Lundstedt, B. van Bavel, P. Haglund, M. Tysklind, L. Öberg, Pressurised liq- [46] Y. Ruan, M. Mohtadi, L.M. Dupont, D. Hebbeln, S. van der Kaars, E.C. Hop- uid extraction of polycyclic aromatic hydrocarbons from contaminated soils, J. mans, S. Schouten, E.J. Hyer, E. Schefuß, Interaction of fire, vegetation, and cli- Chromatogr. A 883 (20 0 0) 151–162, doi:10.1016/S0 021-9673(0 0)0 0419-2. mate in tropical ecosystems: a multiproxy study over the past 22,0 0 0 years, [53] S. Gao, D.A. Hegg, P.V. Hobbs, T.W. Kirchstetter, B.I. Magi, M. Sadilek, Water- Glob. Biogeochem. Cycles 34 (2020) e2020GB006677, doi:10.1029/2020GB00 soluble organic components in aerosols associated with savanna fires in south- 6677. ern Africa: identification, evolution, and distribution, J. Geophys. Res. 108 [47] E. Argiriadis, D. Battistel, D.B. McWethy, M. Vecchiato, T. Kirchgeorg, (2003), doi:10.1029/2002JD002324. N.M. Kehrwald, C. Whitlock, J.M. Wilmshurst, C. Barbante, Lake sediment [54] E. Barbaro, T. Kirchgeorg, R. Zangrando, M. Vecchiato, R. Piazza, C. Barbante, fecal and biomass burning biomarkers provide direct evidence for prehis- A. Gambaro, Sugars in Antarctic aerosol, Atmos. Environ. 118 (2015) 135–144, toric human-lit fires in New Zealand, Sci. Rep. 8 (2018) 12113, doi:10.1038/ doi:10.1016/j.atmosenv.2015.07.047. s41598- 018- 30606-3. [55] R. Zangrando, E. Barbaro, M. Vecchiato, N.M. Kehrwald, C. Barbante, A. Gam- [48] E. Dietze, D. Brykała, L.T. Schreuder, K. Jażdżewski, O. Blarquez, A. Brauer, baro, Levoglucosan and phenols in Antarctic marine, coastal and plateau M. Dietze, M. Obremska, F. Ott, A. Pieńczewska, S. Schouten, E.C. Hopmans, aerosols, Sci. Total Environ. 544 (2016) 606–616, doi:10.1016/j.scitotenv.2015. M. Słowiński, Human-induced fire regime shifts during 19th century industri- 11.166. alization: a robust fire regime reconstruction using northern Polish lake sedi- [56] M. Ahmed, X. Guo, X.-M. Zhao, Application of simple ultrasonic assisted ex- ments, PLoS One 14 (2019) e0222011, doi:10.1371/journal.pone.0222011. traction coupled with HPLC and GC/MS for the determination of surface active [49] L.T. Schreuder, E.C. Hopmans, I.S. Castañeda, E. Schefuß, S. Mulitza, J.S. Sin- compounds in atmospheric particulate matter, Microchem. J. 130 (2017) 400– ninghe Damsté, S. Schouten, Late Quaternary biomass burning in Northwest 411, doi:10.1016/j.microc.2016.10.023. Africa and interactions with climate, vegetation, and humans, Paleoceanogr. [57] K.E. Yttri, J. Schnelle-Kreis, W. Maenhaut, G. Abbaszade, C. Alves, A. Bjerke, Paleoclimatol. 34 (2019) 153–163, doi:10.1029/2018PA003467. N. Bonnier, R. Bossi, M. Claeys, C. Dye, M. Evtyugina, D. García-Gacio, [50] L.T. Schreuder, T.H. Donders, A. Mets, E.C. Hopmans, J.S. Sinninghe Damsté, R. Hillamo, A. Hoffer, M. Hyder, Y. Iinuma, J.-L. Jaffrezo, A. Kasper-Giebl, G. Kiss, S. Schouten, Comparison of organic and palynological proxies for biomass P.L. López-Mahia, C. Pio, C. Piot, C. Ramirez-Santa-Cruz, J. Sciare, K. Teinilä, burning and vegetation in a lacustrine sediment record (Lake Allom, Fraser R. Vermeylen, A. Vicente, R. Zimmermann, An intercomparison study of analyt- Island, Australia), Org. Geochem. 133 (2019) 10–19, doi:10.1016/j.orggeochem. ical methods used for quantification of levoglucosan in ambient aerosol filter 2019.03.002. samples, Atmos. Meas. Tech. 8 (2015) 125–147, doi:10.5194/amt- 8- 125- 2015. 10