Complementary Analytical Liquid Chromatography Methods for the

Article pubs.acs.org/ac

Complementary Analytical Liquid Chromatography Methods for the Characterization of Aqueous Phase from Pyrolysis of Lignocellulosic Biomasses Débora Tomasini,†,‡ Francesco Cacciola,§,∥ Francesca Rigano,† Danilo Sciarrone,† Paola Donato,†,∥,⊥ Marco Beccaria,† Elina B. Caramaõ ,‡ Paola Dugo,†,∥,⊥ and Luigi Mondello*,†,∥,⊥ †

Dipartimento di Scienze del Farmaco e Prodotti per la Salute, University of Messina, Viale Annunziata, 98168 Messina, Italy Institute of Chemistry-Universidade Federal do Rio Grande do Sul, Avenida Bento Gonçalves, 9500, 91501-960 Porto Alegre, Rio Grande do Sul, Brazil § Dipartimento di Scienze dell’Ambiente, della Sicurezza, del Territorio, degli Alimenti e della Salute, University of Messina, Viale Ferdinando Stagno d’Alcontres 31, 98166 Messina, Italy ∥ Chromaleont S.r.l., A Start-Up of the University of Messina, c/o Dipartimento di Scienze del Farmaco e Prodotti per la Salute, University of Messina, Viale Annunziata, 98168 Messina, Italy ⊥ Centro Integrato di Ricerca, University Campus Bio-Medico of Rome, Via Á lvaro del Portillo, 21, 00128 Rome, Italy ‡

S Supporting Information *

ABSTRACT: In this work, two analytical liquid chromatography methods were developed and compared for the characterization of aqueous phases from pyrolysis of lignocellulosic biomasses. NanoLC electron ionization-mass spectrometry (EI-MS) represents a novel and useful tool for both separation and identification of semi/nonvolatile and thermolabile molecules. The use of nanoscale flow rates, the highly reproducibility, and high detailed information on EI spectra are the principal advantages of this technique. On the other hand, comprehensive 2D-LC, providing a two-dimensional separation, increases the overall peak capacity lowering the occurrence of peak coelutions. Despite the use of reversed phase modes in both dimensions, a satisfactory degree of orthogonality was achieved by the employment of a smart design of gradient elution strategies in the second dimension in combination with photodiode array detection (PDA) and atmospheric pressure chemical ionization-mass spectrometry (APCIMS). Because of the absence of the preliminary extraction procedure, the fingerprint obtained for these samples results is independent of the extraction yield or contamination contrary to the gas chromatography-mass spectrometry (GC-MS) approach where a liquid−liquid extraction of the water phase is necessary. The main classes of identified compounds were phenols, ketones, furans, and alcohols. The synergistic information on the two powerful analytical approaches, e.g., NanoLC EI-MS and LC × LC, in the identification of such complex samples has never been investigated and fully benefit on the one hand from the superior degree of mass spectral information from EI-MS and on the other hand from enhanced LC × LC compound separation.

R

enewable energy resources can be one of the potential alternative solutions to fossil fuels and their derivatives. Biomass is one such promising alternative and has a worldwide abundance.1 The conversion of biomass into products with high added value is achieved by the pyrolysis that is a thermochemical process in the absence of air. The pyrolysis temperature of biomass ranges from 350 to 700 °C and the main products are bio-oil, gas, and solids.2 The pyrolysis causes thermal cracking of naturally occurring biomolecules (cellulose, hemicelluloses, lignin, etc.) in the biomass and converts them to simpler organic molecules.3 The liquid product obtained by pyrolysis can be separated into two fractions according to their water solubility. The water insoluble fraction is viscous and denser than the water-soluble, and is usually named bio-oil.4 © 2014 American Chemical Society

Bio-oil is a complex mixture of organic compounds with many oxygen-containing functional groups which include acids, aldehydes, ketones, phenols, furans, and sugars.5 The watersoluble phase predominately consists of low-molecular-weight acids and aldehydes, ketones with high reactivity, as well as phenolic compounds that provide smoky flavors.6,7 The aqueous phase of the bio-oil cannot be directly used as a fuel. Furthermore, some of the environmental parameters are in excess of sewer discharge limits and, therefore, this water must Received: July 28, 2014 Accepted: October 19, 2014 Published: October 20, 2014 11255

dx.doi.org/10.1021/ac5038957 | Anal. Chem. 2014, 86, 11255−11262

Analytical Chemistry

Article

undergo wastewater treatment before disposal.8 In the upgrading research, the aqueous phase is widely used for steam reforming and can generate a high hydrogen yield. In addition, in moderate hydrodeoxygenation and catalytic cracking processes, the aqueous phase has also been chosen as a raw material to produce hydrocarbons, alcohols, and olefins.6 The bio-oils from lignocellulosic biomass can be classified into four distinct fractions: (1) medium-polar monomers detectable by GC, 40% wt, (2) polar monomers detectable directly by HPLC or GC after derivatization, 12% wt, (3) water derived from reaction water and feedstock moisture, 28% wt, and (4) oligomeric material, 20 wt %.9 Many publications involving one-dimensional and comprehensive gas chromatography (GC × GC) analysis have been reported, with the most varied types of bio-oils.2,4,6,10,11 Obviously, such approaches required an extraction step due to the unsuitability of water samples in GC. This step, generally operated with dichloromethane, can greatly affect the qualitative and quantitative composition of the sample extracted from the water phase due to the different solvent affinity for each chemical family of the sample.12 The use of liquid chromatographic methods allows the analysis of the aqueous phases from bio-oils without extraction or derivatization methods such as in gas chromatography. Liquid chromatography also has been used to analyze bio-oils.13−16 However, a study of the aqueous phase of bio-oils by liquid chromatography has not been reported yet. Because of the complexity of the matrix, powerful analytical liquid chromatography methods with high selectivity are required. A very interesting approach is nanoLC. In fact, thanks to the reduced eluates by a nanoLC column, the liquid phase can be directly introduced into a common GC quadrupole mass spectrometer with electron ionization (EIMS), simply through a fused silica capillary tube of internal diameter smaller than 30 μm.17 The nanoliter level LC flow in fact generates a vaporized mobile phase volume compatible with the ion source volume and vacuum pump capability, preserving the vacuum level thus avoiding the ion suppression. Such approach represents a new and useful tool for the characterization of aqueous samples that cannot be directly injected in a GC-MS system. For this reason the nanoLC coupled to an EI-MS system is perfectly amenable to MS ionization. The interface shows a superior performance in the analysis of small-medium molecular weight compounds when compared to its atmospheric ionization techniques (API).17,18 A second valuable alternative could be two-dimensional liquid chromatography which is expected to be complementary to two-dimensional gas chromatography since compounds with high polarities are present in bio-oils, such as cracked sugars or molecules issued from the polycondensation of some oxygenated compounds.19 The development of comprehensive two-dimensional liquid chromatography (LC × LC) techniques has received much attention in recent years. The main advantage of these techniques compared to the traditional methods is the increased peak capacity due to different retention mechanisms in each dimension, which is beneficial for separation of components of complex real-world samples.20−23 As for GC × GC, in LC × LC two different columns with low correlation are normally used. Between them, a switching valve system (in most cases a 10 port-2 position valve) is placed for the collection of the fractions eluted from the first dimension (1D) column and subsequent reinjection in the second dimension (2D) where, usually, a quick separation is performed. In fact, the fractions injected into the second

column must be fully eluted before another transfer of the 1D eluate is performed. Recently, a method for the separation of compounds in the aqueous phase of a partially dehydroxygenated bio-oil by LC × LC in combination with photodiode array (PDA) detection was developed.19 However, the sample compounds were not identified, and a more informative detector such as a mass spectrometry detector would be required. The aim of this work was to characterize for the first time the aqueous phases from bio-oils obtained from three different biomasses exploiting the highly informative and synergistic data obtained by the two analytical liquid chromatography approaches, namely, nanoLC-EI-MS and LC × LC in combination with PDA and atmospheric pressure chemical ionization-mass spectrometry (APCI-MS). The NanoLC-EI-MS (method and instrumentation) approach is not meant to replace but to offer a complementary approach to LC × LC-APCI methods.

■

EXPERIMENTAL SECTION Samples and Sample Preparation. Biomasses from coconut fibers, sugar cane straw, and sugar cane bagasse were kindly provided by EMBRAPA-Tabuleiros Costeiros, SE, Brazil. The samples were previously ground in a knife mill (model MA340MF Marconi, SP, Brazil) and subjected to the drying process in an oven for a period of 48 h at 120 °C. The bio-oils were obtained from fast pyrolysis of biomasses in a fixed bed reactor. The conditions of the pyrolysis were previously optimized, and the following conditions were used: biomass = 400 g, temperature = 700 °C, heating rate = 100 °C min−1, and the final time = 5 min for the gas condensation and collection of liquid products. The aqueous phases were separated from the bio-oils by decantation and were collected without any further dilution, extraction, or pretreatment. The samples were submitted to a filtration through a 0.45 μm Acrodisc nylon membrane filter (Pall Life Sciences, Ann Arbor, MI) and stored in a 1.5 mL vial before the analysis. Reagents and Materials. For nanoLC analyses, water Trace select Ultra, for ultratrace analysis, and acetonitrile Trace select, for trace analysis, were both purchased from SigmaAldrich (St. Louis, MO). Chromatographic separations were carried out using a lab-made packed nanocolumn, ReproSyl-Pur C18 (300 mm × 0.075 mm length × i.d., 3 μm d.p.). For LC × LC-PDA-MS analyses, water and acetonitrile LC-MS Chromasolv were purchased from Sigma-Aldrich (St. Louis, MO). Chromatographic separations were carried out using different columns provided by Supelco (Bellefonte, PA): Ascentis RPAmide (250 mm × 1 mm, 5 μm d.p.) and Ascentis Express C8 (30 mm × 3.0 mm, 2.7 μm d.p.). Standards of 5(hydroxymethyl)furfural, furfural, 5-methylfurfural, 1-(2-furanyl)-ethanone, 2-hydroxy-3-methyl-2-cyclopenten-1-one, 1,4benzenediol, 4-methyl-1,2-benzenediol, 2,6-dimethoxyphenol, phenol, 4-methylphenol, 2-methylphenol, 4-hydroxy-3,5-dimethoxybenzaldehyde, 4-hydroxybenzaldehyde, and 2-methoxyphenol were provided by Sigma-Aldrich (St. Louis, MO). The standard mixture employed for the evaluation of the system orthogonality was prepared in acetonitrile with a concentration of 0.05 mg mL−1. NanoLC Instrumentation and Software. The instrument consisted of a nano prominence HPLC (Shimadzu, Kyoto, Japan) system coupled to a GCMS-QP2010nc Ultra system. The nanoLC configuration was equipped with two LC-20AD nano pumps, a CBM-20A controller, a DGU-20A3R degassing unit, and an FCV nanovalve six port-two position for the 11256



Article

extracted at a 280 nm wavelength. The LC × LC data were visualized in two and three dimensions using Chromsquare, version 2.0 (Shimadzu, Kyoto, Japan). LC × LC-PDA-MS Conditions. 1D separations were carried out using an Ascentis RP-Amide (250 mm × 1 mm, 5 μm d.p.) column (Supelco, Bellefonte, PA) using a flow rate of 10 μL min−1, which was precisely verified using a microsyringe and a stopwatch. The mobile phases employed were (A) water and (B) acetonitrile, eluted according to the following gradient: 0 min, 5% B; 120 min, 100% B; 150 min, 100% B. The injection volume was 2 μL, and the oven temperature was 40 °C. 2D separations were carried out using an Ascentis Express C8 (30 mm × 3.0 mm, 2.7 μm d.p.) column (Supelco, Bellefonte, PA). The mobile phase employed were (A) water and (B) acetonitrile, eluted according to the following gradient: (0− 80 min) 0 min, 2% B; 0.7 min, 10% B; 0.79 min, 10% B; 0.80 min, 2% B; 1.0 min, 2% B; (80−150 min) 0 min, 5% B; 0.79 min, 15% B; 0.80 min, 5% B; 1.0 min, 5% B. The flow rate employed was 3.0 mL min−1, and it was reduced to roughly 1 mL min−1 through a T-piece union prior to the MS detection. The oven temperature was 40 °C. MS acquisition was performed using the APCI interface operating in both positive and negative ionization modes, under the following conditions: mass spectral range, 50−400 m/z; event time, 0.3 s; scan speed, 1250 u/s; nebulizing gas (N2) flow, 2.0 L min−1; drying gas (N2) flow, 10 L min−1; heat block temperature, 200 °C; desolvation line (DL) temperature, 250 °C; DL voltage, −34 V; probe voltage, −4.5 kV; Qarray dc voltage, 1.0 V; Qarray rf voltage, 60 V; detection gain, 1.0 kV.

injection of nano quantity of sample. The valve was equipped with a 74 nL loop (peeksil tubing 15 cm × 25 μm i.d., SGE Analytical Science, Ringwood Victoria, Australia). The injection volume was regulated via software by controlling the switching time of the valve in relation to the mobile phase flow. LC conditions (gradient program, flow rate, valve switching) were controlled by Nano-Assist software version 1.00 (Shimadzu Co.) while the conditions relative to the GC-MS part was controlled by the GCMSsolution software version 2.70 (Shimadzu, Kyoto, Japan). AMDIS (Automated Mass Spectral Deconvolution and Identification System) software, developed by the National Institute of Standards and Technology (NIST), was applied to support peak identification. NanoLC-EI-MS Conditions. Analyses were performed by using a lab-made packed nanocolumn, ReproSyl-Pur C18, as described in the Reagents and Materials section. Water (solvent A) and acetonitrile (solvent B) were employed as mobile phase. The following gradient mode was used for all the samples analyzed: 0−30 min 0−25% B, 30−40 min 25−50%, held for 10 min, then to 0% in 2 min. The flow rate was 150 nL min−1, while the injection volume was 10 nL. The oven temperature was 40 °C. The interface between the LC and the GC mass spectrometer was achieved connecting the nano-LC column exit directly to the ion source of the mass spectrometer, by means of a 50 μm i.d. fused silica capillary tubing equipped with a final orifice smaller than 10 μm i.d., protruding about 1 mm into the ion source. The interface temperature was set at 60 °C while the ion source temperature was 300 °C. Mass spectrometer tuning and calibration were performed automatically at 300 °C, using perfluorotributylamine (PFTBA) and monitoring the ions at m/z 69, 219, and 502 as commonly done in a GC MS tuning procedure. The mobile phase was not allowed into the ion source during the tuning operation; sensitivity was adjusted with respect to the ion at m/z 131 that provided the best results taking into consideration that the highest molecular weight of the components analyzed was 154 Da. In this concern, a 80−200 m/z mass range was applied. The detector voltage was set relative to the tuning result; the ionization voltage was kept at 70 eV, thus allowing the comparison between the experimental mass spectra and commercially available EI mass spectral databases, namely, FFNSC 2.0 (Shimadzu), NIST 11, and WILEY 9. Analyses were all performed in full scan mode applying a 0.5 Hz acquisition rate in order to acquire a sufficient number of data points for an accurate peak reconstruction and to achieve the lower noise signal. LC × LC Instrumentation and Software. Comprehensive LC analyses were performed on a Shimadzu Prominence system (Shimadzu, Milan, Italy), consisting of a CBM-20A controller, two LC-20AD dual-plunger parallel-flow pumps (employed for the 1D separation), a LC-20AB solvent delivery module equipped with two dual-plunger tandem-flow pumps (2D), a DGU-20A3 online degasser, a CTO-20A column oven, a SIL-20AC autosampler, a SPD-M20A photo diode array detector (2.5 μL detector flow cell volume), and a LCMS-2020 single quadrupole mass spectrometer. Both dimensions were connected using an electronically controlled 2-position 10-port switching valve (Supelco, Bellefonte, PA) placed inside the column oven and equipped with two identical 10 μL sample loops. The system was controlled by the LabSolution software, version 5.41 SP1 (Shimadzu, Milan, Italy). Data were collected from 200 to 350 nm using a sampling rate of 12.5 Hz and 0.08 s during all the whole analysis time. The chromatograms were

■

RESULTS AND DISCUSSION Characterization of Aqueous Phase Samples by NanoLC-EI-MS. The goal of such an approach was the direct injection of the aqueous phase coming from the pyrolysis of three different biomasses avoiding any preliminary extraction procedure. As a matter of fact, the nonhomogenous extraction yield between the different chemical families, as for the most polar components, represents the major limitation of the GCMS analysis of water samples where a liquid−liquid extraction, normally operated with dichloromethane, is required. Almeida et al.11 recently determined the composition of a dichloromethane extract of the aqueous phase of coconut fibers by GCMS: 33 compounds were identified and among them only 15 were detected with a relative area higher than 1%. Phenols, ketones, and ethers were identified, unlike higher polarity compounds like furfurals, probably because they were not extracted from the water sample.12 As a consequence, the main advantage of the nanoLC-EI-MS analysis appeared to be the possibility of the direct injection of the water sample exploiting the EI-MS identification. In addition, an advantage arising from the use of the nanoLC approach is related to the considerable reduced solvent consumption; the solvent volume required for a single analysis, taking into consideration the reconditioning time, was only 10.8 μL. To the best of the authors’ knowledge, there has been no report in the literature about the aqueous phase of sugar cane straw and bagasse, although the results hereby reported are in accordance with their correspondent bio-oil composition previously reported.24−26 NanoLC-EI-MS analysis evidenced a similar composition for all the samples with compounds belonging to the ketones, phenols, and furans chemical families. According to Zhang et al.,27 the furan ring containing compounds and cyclic ketones are primarily produced from cellulose and hemicellulose. In fact, when the 11257



Article

spectral similarity higher than 90%, 2 between 82% and 84%. Particularly, 5-(hydroxymethyl)furfural (peak 1), 1,4-benzenediol (peak 2), and 2,6-dimethoxyphenol (peak 15) were confirmed by AMDIS that provided the capability to recognize, e.g., peak 1 and peak 4, otherwise not detectable. Further, it allowed one to resolve the spectral coelution of peak 15 with peak 14, responsible for the low spectral similarity (78%) arising from direct comparison with spectral databases. Among the tentatively identified components, assigned on the basis of literature data, 2-cyclopenten-1-one (peak 4) and 3-methyl-2cyclopenten-1-one (peak 6) were characterized by a high spectral similarity (99% and 96%, respectively); on the other hand, 4-ethyl-1,3-benzenediol (peak 17) presented a low spectral similarity, probably related to its poor MS signal, in all the samples, even though it was inferred by AMDIS (94%). Finally, 2-methoxy-4-methylphenol (peak 20), even if characterized by an 84% spectral similarity, was tentatively identified according to the presence in the sugar cane straw and bagasse samples, where it was identified with an 89% and 95% spectral similarity, respectively. In sugar cane straw (Figure 1B) 13 of the 19 components identified presented a spectral similarity higher than 90%, and 6 between 81 and 89%. 1,4-Benezenediol (peak 2), 1,2-benzenediol (peak 7), and 2-methyl-2-cyclopenten-2-one (peak 9) were identified only by AMDIS. The sample was characterized by the lowest content of components and by the presence of 3-ethyl-2-hydroxy-2-cyclopenten-1-one (peak 11) that was tentatively identified only in this sample. 2Hydroxy-3-methyl-2-cyclopenten-1-one (peak 4) was identified by standard injection: in such a case the low spectral similarity (81%) was related to the partial coelution with 2-cyclopenten1-one (peak 3) and 2-furanmethanol (peak 5). 4-Methyl-1,2benzenediol (peak 13) was assigned also on the basis of standard injection because of the low spectral similarity related to the reduced amount in the sample. In this case, AMDIS was able to provide a confirmation with a higher spectral similarity (94%). In sugar cane bagasse (Figure 1C), 17 of the 21 components identified presented a spectral similarity higher than 90%, 4 between 87 and 89%. The sample was characterized by an overall high spectral similarity for each component and by the presence of 4-ethyl-2-methoxyphenol (peak 23), identified only in this sample. As for sugar cane straw, 1,4-benzenediol (peak 2) and 2-methyl-2-cyclopenten-1one (peak 9) were only identified by AMDIS. The precision of the method was calculated in terms of average (n = 9), intraday (n = 6), and interday (n = 9) retention time RSD (Table 1): repeatability data were below 3.25% apart for peaks 2−4 and 6−8, which probably suffered from compound coelutions, e.g., the values obtained for peaks 4 and 7 can be inferred to the coelution with peaks 5 and 8, respectively. NanoLC-EI-MS was found to be suitable for the determination of aqueous phase of bio-oils: the highly informative electron impact spectra were a useful starting point for a complementary elucidation of the samples composition through the LC × LC-PDA-APCI-MS approach. Characterization of Aqueous Phase Samples by LC × LC-PDA-APCI-MS. After the preliminary data attained on the NanoLC-EI-MS system, an LC × LC system coupled to PDA and APCI-MS, for the three previously analyzed samples, was tuned. For this task, an RP-Amide column (RPA) in the 1D and a partially porous octylsilica column (C8) in the 2D were employed. Prior to separation of the samples, a standard mixture of 14 compounds was analyzed on the LC × LC system, for optimization purposes. In a previous work, the RPA

pyrolysis temperature is sufficiently high, cellulose and hemicellulose can directly decompose to some anhydrosugars, which can be converted to furans,28 while phenols, including methoxy-containing compounds as well as monofunctional phenols, are mainly produced from lignin. Figure 1 shows the profile of the three samples investigated which resulted very similar in composition with some differences in the relative concentrations.

Figure 1. NanoLC-EI-MS analyses of aqueous phase from pyrolysis of coconut fibers (A), sugar cane straw (B), and sugar cane bagasse (C).

Coconut fiber turned out to be the richest sample in phenol (peak 12), sugar cane bagasse the richest in furfural (peak 7), while sugar cane straw presented the lowest amount of almost all components. Table 1 reports the list of the identified compounds, precision data as intersamples average retention time, intraday (n = 6) and interday (n = 9) retention time RSD %, together with the experimental EI mass spectral similarity with the commercially available EI databases (Supplementary Figure 1, Supporting Information). In addition, AMDIS algorithm was applied for separating mass spectral signal recorded from coeluted substances (data reported in parentheses). Going into details, the list contains 23 identified compounds: 16 compounds, confirmed by standard injection and MS spectra, and 7 tentatively identified compounds due to the lack of reference standards, selected between possible candidates on the basis of literature data, the correspondence between the three samples and the employment of the AMDIS algorithm. The MS similarity value was high for almost all the compounds identified. Lower MS similarities were presumably related to a low MS signal or to coelutions issues. In such cases, the AMDIS algorithm was applied successfully. In coconut fiber (Figure 1A), 17 of the 19 identified components presented a 11258



Article

Table 1. Identified Compounds in Aqueous Phase of Bio-Oil Samples by NanoLC-EI-MSa spectral similarity no.

avg tR (min) n=9

RSD % intraday n=6

RSD % interday n=9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

9.39 9.99 10.97 11.98 13.02 14.49 15.55 15.72 17.99 18.93 20.90 22.49 23.50 25.94 27,58 28.97 33.51 36.04 37.84 38.15 44.33 44.79 45.50

2.71 3.46 3.04 3.40 2.87 2.74 2.14 2.22 0.61 1.69 1.15 1.14 1.44 1.11 0.68 0.87 0.71 0.55 0.45 0.74 0.27 0.38 0.30

3.18 4.39 4.70 6.59 2.87 4.92 5.63 3.93 2.93 3.24 1.15 2.80 3.25 1.77 2.01 1.74 1.49 1.23 1.02 0.84 0.61 0.51 0.30

a

composition

coconut fibers

sugar cane straw

sugar cane bagasse

5-(hydroxymethyl)furfural 1,4-benzenediol 2-cyclopenten-1-oneb 2-hydroxy-3-methyl-2-cyclopenten-1-one 2-furanmethanol 3-methyl-2-cyclopenten-1-oneb 1,2-benzenediolb furfural 2-methyl-2-cyclopenten-1-oneb 1-(2-furanyl)-ethanone 3-ethyl-2-hydroxy-2-cyclopenten-1-oneb 5-methylfurfural 4-methyl-1,2-benzenediol phenol 2,6-dimethoxyphenol 2-methoxyphenol 4-ethyl-1,3-benzenediolb 4-methylphenol 2-methylphenol 2-methoxy-4-methylphenol 4-ethylphenol 2,4-dimethylphenol 4-ethyl-2-methoxyphenolb

n.d. (82%)c 94%c 99% 89% (92%)c 98% 96% 97% 95% n.d. 95% n.d. 95% 92% (95%)c 99% 78% (91%)c 96% 85% (94%)c 96% 97% 84% n.d. 91% n.d.

n.d. 82%c 99% 81% (81%)c 98% 96% n.d. (87%)c 86% (96%)c 93%c 92% 87% (92%)c n.d. 77% (94%)c 95% 88% 92% 85% (89%)c 91% 93% 89% 91% n.d. n.d.

90% 92%c 99% 86% (89%)c n.d. 89% (91%)c 94% 98% 87%c 93% n.d. 96% n.d. (87%)c 95% 92% 97% 82% (95%)c 96% 97% 95% 96% 93% 89%

n.d. = not detectable. bTentatively identified compounds. cSimilarity value by AMDIS.

porous RP columns with comparable dimensions, and 6 s were sufficient to re-equilibrate the column at the initial conditions and to obtain perfectly repeatable retention times for the same peaks eluted in consecutive run.36−41 Figure 3 shows the LC × LC-PDA-APCI-MS chromatograms obtained for the aqueous phases from pyrolysis of coconut fibers (A), sugar cane straw (B), and sugar cane bagasse (C). For evaluating the system performance, the peak capacity (nc) values of the single dimensions were calculated using the well-known method defined by Neue.42 The theoretical peak capacity value, being multiplicative of the individual values obtained for the two dimensions (1nc × 2nc) was 865. Since the LC × LC separation space was not occupied by all peaks, in order to attain more realistic peak capacity values another approach was taken into consideration allowing to obtain “practical” peak capacity vaues corrected for both undersampling43 and orthogonality.44 First, the under-sampling effect was calculated considering the number of fractions effectively transferred from the 1D to the 2D. Such an approach, which takes into account the 2D cycle time (2tC = 1 min), and the average 1D peak width (1ω = 3 min) was estimated as 1.17:

column, probably due to the polar embedded RP phase, provided enough separation selectivity, in combination with a C18 column, making such a combination an interesting opportunity for RP-LC × RP-LC separations of complex samples.29 In this work, considering the polarity of compounds occurring in the tested samples, a C8 column was investigated for the 2D separations, as previously adopted for analysis of phenolic and flavone natural antioxidants in beer samples.30 Figure 2A shows the chemical structures for the standard compounds, whereas Figure 2B illustrates the resulting LC × LC chromatogram (extracted at 280 nm which is suitable for the different classes of compounds under investigation). As it can be seen, all the target compounds, which can be considered as representatives of the samples tested, are separated with a satisfactory degree of orthogonality. Optimization was carried out taking into consideration that each fraction injected onto the secondary column had to be completely eluted before the following transfer occurred thus keeping the 2D analysis time as short as possible (1 min), not to impair the separation achieved in the 1D.31 To this regard, for 2D analyses, a very high flow rate (3 mL min−1) was employed. Different gradients of increasing acetonitrile in water both in the 1D and the 2D dimensions were tested. The use of a segmented gradient turned out to be the most effective, yielding the highest peak capacity and increasing the orthogonality degree, in accordance with previous works dealing with RP-LC × RP-LC separations.32,33 In order to ensure peak focusing at the head of the 2 D column, a gradient program starting with 98% concentration of the weaker solvent (water) was adopted, considering the short reconditioning time used (12 s), which enabled the application of fast repetitive gradients (every 1 min).34,35 As demonstrated in previous studies, approximately one volume of column was sufficient to obtain reconditioning of partially

β=

⎛ 2t ⎞2 1 + 3.35⎜ 1 C ⎟ ⎝ ω⎠

Afterward, quantitative evaluation of orthogonality between the two dimensions was made, using the findings of a very recent work proposed by Camenzuli and Schoenmakers.44 Such a method uses a series of “asterix” equations based solely on the experimentally measured peak retention times, and it can be easily accomplished by using a simple Microsoft Excel software. These equations allow one to evaluate the spreading of peaks around four lines in the 2D plot. The application of this 11259



Article

Figure 2. Chemical structures of the reference materials (A) and RPLC × RP-LC-PDA (λ = 280 nm) contour plot of a standard mixture investigated in this study (B).

equations system led to an orthogonality value (Ao) of 47%. The “practical” nc value, considering both the under-sampling and the orthogonality effects on the system performance in a cumulative way, was determined by multiplying the theoretical nc (where the 1nc value was corrected for undersampling) by the Ao value yielding a value of 347. As far as identification is concerned, combined data, coming from the standard mixture, PDA and MS detection along with literature data were used.11,19,24−26 The identified compounds in the three samples are presented in Table 2, where the occurring oxygenated compounds such as furans, ketones, phenols, and alcohols are reported. The use of APCI-MS detection is hereby reported for the first time for the detection of polar and low molecular weight (m/z ≤ 179) compounds in aqueous phase of bio-oils by LC × LC; in fact, so far ESI-MS after sample derivatization has been commonly employed.45 Phenols and alcohols were detected by APCI-MS in negative ionization mode as [M − H]− because of the presence of the hydroxyl functional group, whereas most ketones and furans in the positive ionization mode, as [M]+ or [M + H]+. For all the samples, a similar composition was observed according to the identified compounds by nanoLC-EI-MS. For this reason, some compounds tentatively identified by LC × LC-PDA-MS can be confirmed by the use of EI-MS commercial libraries in the nanoLC-EI-MS system. The identified compounds for the aqueous phase from coconut sample are assigned in the 2D chromatogram in Figure 4. Quantitative evaluation of all the

Figure 3. RP-LC × RP-LC-PDA (λ = 280 nm) contour plots obtained for the aqueous phases from pyrolysis of coconut fibers (A), sugar cane straw (B), and sugar cane bagasse (C).

sample components was not carried out, at this level, due to limited availability of some reference materials. However, a rough estimation was performed since all compounds within each chemical class showed very similar molar absorbivities, through software integration of the corresponding blobs in the plot. From the calculated relative percentage areas attained, e.g., for the coconut sample, 2-hydroxy-3-methyl-2-cyclopenten-1one, furfural, phenol, and phenylpropanol were the most abundant among ketones, furans, phenol, and alcohols, respectively (Supplementary Table 1, Supporting Information). 11260



Article

Table 2. Identified Compounds in Aqueous Phase of Bio-Oil Samples Analyzed by LC × LC-PDA-MS samples no.

total tR (min)

RSD % (n = 9)

[M − H]−

1 2 3 4 5 6 7

22.16 23.15 29.20 33.24 42.16 49.28 52.38

3.14 3.08 2.56 4.59 2.74 2.02 1.90

96 (+) 127 (+) 97 125 109 96 (+) 111

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

56.43 64.32 66.46 67.36 74.55 74.72 79.79 80.75 87.54 87.60 87.77 92.63 93.50 104.80 107.82 108.39 111.83 115.91 126.56

2.04 0.04 0.86 0.02 1.34 1.35 0.73 0.65 0.66 0.66 1.15 0.60 0.06 1.34 1.33 1.65 0.52 0.48 0.75

137 (+) 123 109 109 110 (+) 125 149 179 123 135 153 123 93 151 133 137 107 107 121

a

tentative identification 3-methyl-2-cyclopenten-1-one furanyl-hydroxyethanone methyl-furanone 5-(hydroxymethyl)furfurala 1,4-benzenediola furfurala 2-hydroxy-3-methyl-2-cyclopenten-1onea hydroxyacetophenone methyl-1,4-benzenediol 1-(2-furanyl)-ethanonea 1,2-benzenediol 5-methylfurfurala 3-ethyl-2-hydroxy-2-cyclopenten-1-one coumaryl alcohol coniferyl alcohol 4-methyl-1,2-benzenediola phenylpropanol 2,6-dimethoxyphenola 2-methoxyphenola phenola dihydroxyacetophenone indanol 4-ethyl-1,3-benzenediol 4-methylphenola 2-methylphenola 4-ethylphenol

chemical class

coconut fibers

sugar cane straw

sugar cane bagasse

ketone furan ketone furan phenol furan ketone

+ + − + + + +

+ + − + + + +

− − + + + + +

ketone phenol furan phenol furan ketone alcohol alcohol phenol alcohol phenol phenol phenol ketone alcohol phenol phenol phenol phenol

+ − + + + + + + + + + + + + + − + + −

+ + + + + + + + + + + + + + − − + + −

+ + + + + + + + + + + + + − + + + + +

Confirmed compounds by standard injection.

techniques showed complementary behavior thanks to the enhanced identification capability of the EI-MS with respect to the APCI mode and to the increased separation afforded by the LC × LC system. In fact, the use of the nanoLC system, besides the identification capability because of commercially available EI-MS databases, allowed a “green chemistry” approach, due to the considerable solvent consumption reduction. On the other hand, the use of LC × LC, due to an increased peak capacity, allowed one to solve coelution issues thus leading to the detection of a higher number of compounds. In order to render more understandable for readers in a quantitative point of view the differences in detection between the two methods, a Venn diagram is included in Figure 5; even though in the nanoLC-EIMS approach some coelution issues could be solved through mass spectral deconvolution softwares, the LC × LC setup presents as main advantage the possibility to identify each

Figure 4. RP-LC × RP-LC-PDA (λ = 280 nm) contour plot obtained for the aqueous phase from pyrolysis of coconut fibers with the peak number of identified compounds. For the peak assignment, see Table 2

These compounds have important uses, e.g., ketones in chemical synthesis; phenols as disinfectants, resins, pesticides; and furans as lubricants, adhesives, plastics, and nylons.46−48

■

CONCLUSIONS Two highly informative analytical liquid chromatography methods were applied for the characterization of aqueous phases from pyrolysis of lignocellulosic biomasses. The two

Figure 5. Venn diagram containing all identified compounds by means of nanoLC-EI-MS and RP-LC × RP-LC-PDA techniques. 11261



Article

compound displayed in the 2D plot, corresponding to the first and second dimension retention times. With phenols, ketones, and furans as the principal classes of identified compounds, the aqueous phase obtained in the pyrolysis process after the separation of bio-oils may have several uses in the chemical industry. From a chromatographic stand-point, in the near future a nanocomprehensive LC setup will be adopted, aiming to take full advantage of the improved separation and identification power of nanoLC-MS.

■

(13) Andersson, T.; Hyotylainen, T.; Riekkola, M. J. Chromatogr., A 2000, 896, 343−349. (14) Tienpont, B.; David, F.; Pereira, A.; Sandra, P. J. Chromatogr., A 2011, 1218, 8303−8312. (15) Tessini, C.; Müller, N.; Mardones, C.; Meier, D.; Berg, A.; von Baer, D. J. Chromatogr., A 2012, 1219, 154−160. (16) Yanik, J.; Stahl, R.; Troeger, N.; Sinag, A. J. Anal. Appl. Pyrol. 2013, 103, 134−141. (17) Cappiello, A.; Famiglini, G.; Mangani, F.; Palma, P. Mass Spectrom. Rev. 2001, 20, 88−104. (18) Cappiello, A.; Famiglini, G.; Pierini, E.; Palma, P.; Trufelli, H. Anal. Chem. 2007, 79, 5364−5372. (19) Le Masle, A.; Angot, D.; Gouin, C.; D’Attoma, A.; Ponthus, J.; Quignard, A.; Heinisch, S. J. Chromatogr., A 2014, 1340, 90−98. (20) Dugo, P.; Cacciola, F.; Kumm, T.; Dugo, G.; Mondello, L. J. Chromatogr., A 2008, 1188, 208−215. (21) Stoll, D. R. Anal. Bional. Chem. 2010, 397, 979−986. (22) François, I.; Sandra, K.; Sandra, P. Anal. Chim. Acta 2009, 641, 14−31. (23) Donato, P.; Cacciola, F.; Tranchida, P. Q.; Dugo, P.; Mondello, L. Mass Spectrom. Rev. 2012, 31, 523−559. (24) Pattiya, A.; Sukkasi, S.; Goodwin, V. Energy 2012, 44, 1067− 1077. (25) Tsai, W. T.; Lee, M. K.; Chang, Y. M. J. Anal. Appl. Pyrol. 2006, 76, 230−237. (26) Mantilla, S. V.; Gauthier-Maradei, P.; Gil, P. A.; Cárdenas, S. T. J. Anal. Appl. Pyrol. 2014, 108, 284−294. (27) Zhang, H.; Xiao, R.; Wang, D.; He, G.; Shao, S.; Zhang, J.; Zhong, Z. Bioresour. Technol. 2011, 102, 4258−4264. (28) Carlson, T. R.; Jae, J.; Lin, Y. C.; Tompsett, G. A.; Huber, G. W. J. Catal. 2010, 270, 110−124. (29) Dugo, P.; Cacciola, F.; Donato, P.; Jacques, R. A.; Caramao, E. B.; Mondello, L. J. Chromatogr., A 2009, 1216, 7213−7221. (30) Hájek, T.; Škeřiková, V.; Č esla, P.; Výnhucalová, K.; Jandera, P. J. Sep. Sci. 2008, 31, 3309−3328. (31) Murphy, R. E.; Schure, M. R.; Foley, J. P. Anal. Chem. 1998, 70, 1585−1594. (32) Jandera, P. J. Chromatogr., A 2012, 1255, 112−129. (33) Leme, G. M.; Cacciola, F.; Donato, P.; Cavalheiro, A. J.; Dugo, P.; Mondello, L. Anal. Bioanal. Chem. 2014, 406, 4315−4324. (34) Schellinger, A. P.; Stoll, D. R.; Carr, P. W. J. Chromatogr., A 2005, 1064, 143−156. (35) Stoll, D. R.; Cohen, J. D.; Carr, P. W. J. Chromatogr., A 2006, 1122, 123−137. (36) Dugo, P.; Cacciola, F.; Herrero, M.; Donato, P.; Mondello, L. J. Sep. Sci. 2008, 31, 3297−3308. (37) Č esla, P.; Hájek, T.; Jandera, P. J. Chromatogr., A 2009, 1216, 3443−3457. (38) Jandera, P.; Hájek, T.; Č esla, P. J. Sep. Sci. 2010, 33, 1382−1397. (39) Russo, M.; Cacciola, F.; Bonaccorsi, I.; Dugo, P.; Mondello, L. J. Sep. Sci. 2011, 34, 681−687. (40) Dugo, P.; Cacciola, F.; Donato, P.; Airado-Rodriguez, D.; Herrero, M.; Mondello, L. J. Chromatogr., A 2009, 1216, 7483−7487. (41) Donato, P.; Cacciola, F.; Sommella, E.; Fanali, C.; Dugo, L.; Dach, M.; Campiglia, P.; Novellino, E.; Dugo, P.; Mondello, L. Anal. Chem. 2011, 83, 2485−2491. (42) Neue, U. W. J. Chromatogr., A 2005, 1079, 153−161. (43) Gu, H.; Huang, Y.; Carr, P. W. J. Chromatogr., A 2011, 1218, 64−73. (44) Camenzuli, M.; Schoenmakers, P. J. Anal. Chim. Acta 2014, 838, 93−101. (45) Prieto-Blanco, M. C.; Moliner-Martínez, Y.; López-Mahía, P.; Campíns-Falcó, P. Talanta 2013, 115, 876−880. (46) Surmont, R.; Verniest, G.; Kimpe, N. J. Org. Chem. 2010, 75, 5750−5753. (47) Jianhan, H. J. Appl. Polym. Sci. 2011, 121, 3717−3723. (48) Li, Q.; Lam, L. K. M.; Xun, L. Biodegradation 2011, 22, 1215− 1225.

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +39-090-6766536. Fax: +39-090-358220. E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Débora Tomasini thanks CAPES (Coordenaçaõ de Aperfeí Superior) for the scholarship çoamento de Pessoal de Nivel (Process Number BEX 0495/13-1) and EMBRAPA TC (SE, Brazil) for the biomass samples. The Project was funded by the “Italian Ministry for the University and Research (MIUR)” within the National Operative Project “Hi-Life Health Products from the Industry of Foods”, Project ID PON01_01499. We would like to thank Prof. Gasparrini from Sapienza University of Rome for the supply of the lab-made packed nanocolumn.

■

REFERENCES

(1) Jacobson, K.; Maheria, K. C.; Dalai, A. K. Renew. Sust. Energy Rev. 2013, 23, 91−106. (2) Cunha, M. E.; Schneider, J. K.; Brasil, M. C.; Cardoso, C. A.; Monteiro, L. R.; Mendes, F. L.; Pinho, A.; Jacques, R. A.; Machado, M. E.; Freitas, L. S.; Caramão, E. B. Microchem. J. 2013, 110, 113−119. (3) Kanaujia, P. K.; Sharma, Y. K.; Garg, M. O.; Tripathi, D.; Singh, R. J. Anal. Appl. Pyrol. 2014, 105, 55−74. (4) Bertero, M.; Gorostegui, H. A.; Orrabalis, C. J.; Guzmán, C. A.; Calandri, E. L.; Sedran, U. Fuel 2014, 116, 409−414. (5) Rover, M. R.; Johnston, P. A.; Lamsal, B. P.; Brown, R. C. J. Anal. Appl. Pyrol. 2013, 104, 194−201. (6) Wang, S.; Wang, Y.; Cai, Q.; Wang, X.; Jin, H.; Luo, Z. Sep. Purif. Technol. 2014, 122, 248−255. (7) Czernik, S.; Bridgwater, A. V. Energy Fuels 2004, 18, 590−598. (8) Xu, R.; Ferrante, L.; Briens, C.; Berruti, F. J. Anal. Appl. Pyrol. 2011, 91, 263−272. (9) Bayerbach, R.; Meier, D. J. Anal. Appl. Pyrol. 2009, 85, 98−107. (10) Djokic, M. R.; Dijkmans, T.; Yildiz, G.; Prins, W.; Van Geem, K. M. J. Chromatogr., A 2012, 1257, 131−140. (11) Almeida, T. M.; Bispo, M. D.; Cardoso, A. R. T.; Migliorini, M. V.; Schena, T.; de Campos, M. C. V.; Machado, M. E.; López, J. A.; Krause, L. C.; Caramão, E. B. J. Agric. Food Chem. 2013, 61, 6812− 6821. (12) de Oliveira, A. M.; Biajoli, A. F. P.; Fidelis, C. H. V.; Silva, R. G. C.; Augusto, F. In Trends in Sample Preparation; Arruda, M. A. Z., Ed.; Nova Science: New York, 2006; p 142. 11262


Complementary Analytical Liquid Chromatography Methods for the

Recommend Documents