Impact of combined acid washing and acid impregnation on the pyrolysis of Douglas fir wood

Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

Contents lists available at ScienceDirect

Journal of Analytical and Applied Pyrolysis journal homepage: www.elsevier.com/locate/jaap

Impact of combined acid washing and acid impregnation on the pyrolysis of Douglas fir wood Brennan Pecha a , Pablo Arauzo a,b , Manuel Garcia-Perez a,∗ a

Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA Thermochemical Conversion Group (GPT), Aragon Institute of Engineering Research (13A), University of Zaragoza, Maria de Luna, 3, Torres-Quevedo Bld., 50018 Zaragoza, Spain b

a r t i c l e

i n f o

Article history: Received 7 March 2015 Received in revised form 18 May 2015 Accepted 25 May 2015 Available online 28 May 2015 Keywords: Pyrolysis Mineral removal Acid wash Acid infusion pretreatment Levoglucosan Py–GC/MS

a b s t r a c t This study reports the impact of acid washing (to remove alkali and alkaline earth metals) followed by mild acid impregnation on the pyrolysis of Douglas fir wood. Dilute nitric acid was used in the washing and sulfuric acid, acetic acid, nitric acid, and phosphoric acid at acid loading levels of 0.05, 0.1, 0.3, and 0.5 wt.% were used for the acid impregnation to improve the yield of levoglucosan. A Py–GC/MS instrument was used to semi-quantitatively measure the impact of various acids and concentrations on the yield of low molecular weight compounds released during pyrolysis at 500 ◦ C. The nitric acid wash removed 56% of the metal content. The results confirmed all the acids studied increased the production of levoglucosan, likely due to the mitigation of undesirable interactions between the cellulose and the other constituents of the lignocellulosic matrix. The highest yields of levoglucosan were achieved with the strong acids: sulfuric acid loading of 0.05 wt.%, nitric acid 0.05 wt.% and phosphoric acid 0.3 wt.%. Sulfuric and phosphoric acid also enhanced dehydration reactions in cellulose products and decreased the production of methoxylated phenolics from lignin. The very small range of concentration at which these acids increase levoglucosan yield makes it very difficult to control the process. In the case of acetic acid, it is not strong enough to catalyze dehydration reactions. Consequently a much wider range of concentrations can be used thus facilitating the control of the process. Acetic acid also does not affect the yield of lignin products. Therefore, acetic acid appears to be the most practical for acid impregnation (following mineral removal) in wood pyrolysis. Published by Elsevier B.V.

1. Introduction Pyrolysis is an ancient technology that can convert lignocellulosic materials like wood and straw into char, oil, and gas. Today, pyrolysis is considered one of the few carbon negative processes for producing chemicals and fuel from biological material. Originally used primarily to make low-smoking charcoal for indoor cooking, the processing parameters have been continually tweaked by scientists and engineers throughout the 20th and 21st century to maximize the oil production. Fast pyrolysis is a simple method primarily used for processing lignocellulosic material, such as wood, by which the material is heated to temperatures between 350 and 600 ◦ C in the absence or near absence of oxygen.

∗ Corresponding author at: Department of Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA. Tel.: +1 509 335 7758. E-mail address: [email protected] (M. Garcia-Perez). http://dx.doi.org/10.1016/j.jaap.2015.05.014 0165-2370/Published by Elsevier B.V.

Using high heating rates above 1000 ◦ C/s at atmospheric pressure, fast pyrolysis of softwood typically yields a crude bio-oil (∼75 wt.%), gases (10–20 wt.%), and char (10–15 wt.%) [1]. Bio-oil is made up of a mixture of water and oxygenated organic compounds which have a variety of industrial and food related uses. Natural lignin in the wood (around 25 wt.% of the material) can typically be converted into monomeric and oligomeric phenols at a 70% efficiency. These compounds have known uses for fuels and fuel additives [2]. However, pyrolysis systems typically only recover 10–20 wt.% of cellulose (which comprises 40–50% of wood) in the form of recoverable sugars, the desired products [3]. It is understood that the first reaction of cellulose during pyrolysis makes mono- and oligo-anhydrosugars (mainly levoglucosan and cellobiosan); however, secondary reactions in the liquid intermediate phase (still on the char) or in the gas phase (while leaving the reactor) convert the sugars into C1–C4 molecules and char through various forms of dehydration reactions followed by fragmentation, elimination, and transglycosylation reactions [4–8]. It is greatly desired to preserve the pyrolytic anhydrosugars, which

128

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

have many known uses for polymer chemistry, pharmaceuticals, fermentation, and food applications [9–14]. Although the levoglucosan can be produced in large yields (60+ wt.%) from the pyrolysis of cellulose [7,15,16], yields of only around 20 wt.% (on cellulose basis) are obtained typically during the pyrolysis of lignocellulosic materials [3]. Secondary reactions are also well-known to affect lignin products, which break down the primary product oligomers (pyrolytic lignin) into monomers [17,18]. In fast pyrolysis, where vapor-phase reactions are minimized by a fast-flowing carrier gas, interactions in the liquid intermediate phase have a large impact on the chemical nature of collectible oils and char [7,19]. One well-understood cause is the catalytic properties of metals naturally present in the lignocellulose matrix. Douglas fir wood, for example, has measurable quantities of Na, K, Mg, Ca, Al, Mn, Co, Ni, Cu, Zn, and Ba. While the total mineral content in most woods is in the 1% range, it can be in the 10% range or higher for straws, grasses, and leaves. Most interest lies in the alkaline and alkaline earth metals (AAEMs) due to their high abundance relative to other metals. Again and again, researchers find that sodium, potassium, and calcium increase the gas yield (particularly CO2 and CO), decrease both the yield and molecular weight of the oil, and decrease the surface area of the char [20–25]. The influence of inorganics extends beyond the pyrolysis reactor. Their presence continues to degrade products during storage of pyrolysis oil [26,27]. These metals behave differently: with regards to their impact on pyrolysis by concentration, K+ > Na+ > Li+ > Ca2+ > Mg2+ [28–30]. Non-metals have a far less significant impact on pyrolysis, but experimentalists take care to compare metals using the same conjugate ions due to varying dissociation strengths, where PO4 3− ≈ CO3 2− > OH− ≈ NO3 − > Cl− [28]. The impact on the chemical makeup of the oil is best understood by the effects that AAEMs have on each biomass constituent: cellulose, hemicellulose, and lignin. The most significant impact of AAEMs on cellulose pyrolysis is the dramatic reduction of the product levoglucosan [28]. Instead of levoglucosan, enhanced production of polymer, furfural, 5-5-hydroxymethylfurfural, glycoaldehyde, hydroxyacetone, formic acid, CO2 , CO, and CH4 is observed [25,28,30–33]. Except for cross-linked polymers, these products have lower molecular weights than sugars. Many hypotheses have been presented to explain the impact of AAEMs on cellulose pyrolysis. The current understanding, based on molecular modeling techniques, explains that AAEMs alter the electronic structure of the carbohydrate by interacting with oxygen, affecting the stereochemistry of the molecules during reactions; because of that, rearrangement and dehydration reactions are enhanced, followed by fragmentation reactions [34,35]. It is thought that AAEMs can directly attack the cellulose chain before and during depolymerization reactions as well as catalyze reactions in the liquid intermediate phase [34,36]. Hemicellulose, a carbohydrate polymer with a backbone made up of xylose, arabinose, glucose, galactose, and non-hydrolyzable sugars, is also influenced by metal content during pyrolysis. AAEMs increased production of char CO2 , water, 2-furaldehyde, and acetaldehyde. They did not affect acetic acid production, but they decreased production of dehydrated sugars, formic acid, and acetol [37]. Lignin products are also affected by minerals. It was found that the suppression of AAEMs with sulfuric acid actually reduced the production of methoxylated phenolic products as well as lignin oligomers [38]. Di Blasi et al. found that the addition sodium, potassium, and calcium in wood pyrolysis actually enhanced the production of phenols such as guaiacol, cresol, phenol, and isoeugenol [29,39] and greatly reduced the production of lignin oligomers (pyrolytic lignin) [22].

Fortunately, removal of AAEMs is relatively easy. Acid washing to remove a majority of the metals from lignocellulosic biomass has been shown to double the sugar yield [40,41]. This procedure can dramatically reduce the metal content but has trouble with calcium, which is more tightly bound to the chemical structure of lignocellulose [22,42]. Similarly, adding small amounts of acid to lignocellulosic materials without washing can increase the levoglucosan yield [23,40,43]. In 2012, Kuzhiyil et al. studied the potential effect of infusing switchgrass with various acids to passivate the catalytic effect of alkalines [23]. The authors found that sulfuric acid performed the best. Our group performed a similar study with a 0.05 wt.% sulfuric acid impregnation with Douglas fir in an oxygenfree auger reactor and achieved a 12 wt.% levoglucosan yield [43]. More recently, Kim et al., with red oak with a 0.4 wt.% sulfuric acid impregnation in a partial oxidation fluidized bed reactor, was able to reach a levoglucosan yield close to 20 wt.% [44]. The combination of acid washing and acid impregnation has also been explored. Levoglucosan yields of nearly 26 wt.% from wood (58% on a cellulose basis) were observed in the 1980s with Douglas fir pretreated with organic extraction, acid washing, and a mild acid addition followed by 400 ◦ C pyrolysis at atmospheric pressure. In the same study, a water wash and a mild acid wash followed by 0.1 wt.% acid impregnation gave 14 and 19 wt.% levoglucosan yields, respectively [40,45]. This effect was not observed when these experiments were conducted with cellulose or with holocellulose (cellulose + hemicellulose). The authors explained these results with the hypothesis that this effect is caused by the effect of these acids on cellulose–lignin interactions. The nature of these interactions is still not well understood, but Matsuoka et al. recently proposed the hypothesis that they are associated with hydrogen bonding between the cellulose and lignin structures [46]. Similar experiments conducted by Zhou et al. reached the same conclusions [43,47]. These authors found that there is a concentration of sulfuric acid at which the production of levoglucosan is maximized. Other research comparing acid-washed material with combined pyrolysis of cellulose, hemicellose, and lignin indicate interactions between these constituents that are unique to the wood polymer properties, supporting the hypothesis discussed herein [48]. Table 1 highlights significant examples of acid pretreatment methods to increase levoglucosan yield. Based on our preliminary results and the literature, a combination of acid-wash and acidimpregnation seems to be the most viable approach to increasing levoglucosan yields, though it has not been fully explored. In this paper, we hypothesize that this combination of acid washing and acid impregnation could also work for other acids. Nitric acid was chosen for the acid wash to remove metals before the water wash. Four different acids are considered for impregnation into the biomass: sulfuric acid, acetic acid, phosphoric acid, and nitric acid. Py–GC/MS will be used to parametrically study acid impregnations ranging from 0.05 to 0.5 wt.% relative to the feedstock to determine the impact of each acid on the production of volatiles in the pyrolysis of Douglas fir wood. The removal of minerals will be studied by measuring the ash content before and after the acid wash as well as quantitative measurement of metals via induced coupled mass spectrometry (ICP-MS).

2. Materials and methods 2.1. Wood grinding, acid washing, and acid addition All tests were conducted as illustrated in Fig. 1. Bark-free Douglas fir wood chips (Pseudotsuga menziesii) were provided by Herman Brothers Logging and Construction (Port Angeles, WA), harvested

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

129

Table 1 Important chemical pretreatments of cellulose and biomass to increase pyrolytic sugars from wood. Pretreatment method

Biomass; pyrolysis

LG wt.% yield from starting material, without acid added and with acid added

Explanation provided

Ref.

Water wash, 0.1% (feedstock) H2 SO4 impregnation Acid wash & H2 O rinse, 0.1% (feedstock) H2 SO4 impregnation Organic extraction (8 h toluene–ethanol) followed by hydrolysis (0.1 M H2 SO4 , >100 ◦ C), rinse w/H2 O, then 0.1% (feedstock) H2 SO4 impregnation Sulfuric acid wash (5%) 90 ◦ C 5.5 h followed by water wash to pH 6.6 0.2 mmol/g various acids, 15 g Aq. w/5 g biomass, dry

Douglas fir (46% cellulose); tube furnace (400 ◦ C) Douglas fir (46% cellulose); tube furnace (400 ◦ C)

7 w/o 14 w/ 9 w/o 19 w/

Suppressed lignin–cellulose reactions and inorganics Ash removal

Shafizadeh and Stevenson [40]

Douglas fir (46% cellulose); tube furnace (400 ◦ C)

12 w/o 26 w/

Extractives, hemicellulose, ash removal; suppressed lignin–cellulose reactions

Shafizadeh and Stevenson [40]

Poplar wood (50% cellulose); fluidized bed (500 ◦ C)

3.04 w/o 30.42 w/

Piskorz, et al. [45]

Switchgrass (33.3% cellulose); Py–GC/MS (500 ◦ C)

2 w/o acid 1.5 Acetic/formic acid 3 Nitric acid 5HCl 12H3 PO4 16 Sulfuric acid 3.4 w/o 18 Wash + rinse 11.1 Wash, no rinse 5.5 w/o 12 w/ 20.62 w/

Hydrolysis of cellulose breaks glycosidic bonds; mineral removal H2 SO4 > H3 PO4 > HCl > HNO3 Chlorides, phosphates, and sulfates form thermally stable salts (inorganic suppression)

Kuzhiyil, et al. [23]

Mineral removal

Oudenhoven, et al. [49]

Alkaline inorganics suppression Alkaline suppression by acids, improved vapor release by oxygen

Zhou, et al. [43]

10% Acetic acid, 3.75% acetone, 3.75%, 1.5% propionic acid, 1.5% guiacol; rinse with DI water 0.05% H2 SO4 infusion in water followed by drying 0.4% H2 SO4 infusion in water followed by drying

Pine; fluidized bed (530 ◦ C)

Douglas fir (45% cellulose); augur reactor (500 ◦ C) Red oak (50% cellulose); fluidized bed with 2.1% oxygen (500 ◦ C)

Shafizadeh and Stevenson [40]

Kim et al. [44]

Dry Douglas fir powder

Acid wash

E-pure water wash

Acid wash

E-pure water wash

PyGC/MS Addition of acids (nitric, acetic, sulfuric, phosphoric)

Fig. 1. Diagram of feedstock preparation and acid addition procedure.

in the Cascade Mountains of Washington State. The wood was first hammer milled to less than a 2 mm diameter sieve size (Bliss Industries: Model 400HD, serial 2404) in the Composite Materials and Engineering Center at WSU. After hammer milling, the wood was ball-milled at 300 rpm for 30 h with ceramic balls in a 100 mL jar; ground ceramic makes up 1.1% of the material after this step (Across International PQ-N2). Following milling, the wood was soaked in a of 1 wt.% solution of nitric acid (pH 3.8) in 18 M-cm E-Pure water (Barnstead, Thermoscientific, Waltham, MA) for 24 h at room temperature without agitation. Approximately 10 g of acid solution per gram of wood was used. Following the soaking, the acid solution was removed by centrifugation (5 min at 3100 rpm) followed by decanting and washing with E-Pure water until the electrical resistivity of the water mixed with wood read nearly 18 M-cm. This required about 20 water wash-centrifuge steps. After washing was complete, residual water was removed by vacuum for over one week to avoid heating. To the acid-washed, dried biomass was added various quantities of the acids in methanol solutions: sulfuric acid, acetic acid, phosphoric acid, and nitric acid. The acids used for the experiments were: HNO3 (JT Baker 69–70%), CH3 COOH (JT Baker 100%), H2 SO4 (Sigma–Aldrich 95–98%), and H3 PO4 (EMD 85%). The acids were diluted to 1 wt.% in the methanol to avoid high concentrations of

acid on the material. One gram of acid washed biomass was blended with 0.5, 0.3, 0.1, and 0.05 gram of the acids studied. Methanol was used for rapid removal via vacuum. Acid deposition masses are reported on a gram acid per gram biomass basis. Following acid addition, the material was dried in a vacuum desiccator for over two days. It is likely that some acid evaporated during this time if it was not strongly bound to the material. 2.2. Ash content Ash content was measured using a standard method [50]. In short, oven-dried biomass was added to dry aluminum crucibles and heated in a muffle furnace for 24 h at 550 ◦ C. The ash content was measured for the untreated wood as well as acid-washed wood for comparison. Ceramic content from ball-milling was subtracted from the reported ash contents, which are on a dry, ceramic-free basis. 2.3. ICP-MS Metal analysis was conducted for the Douglas fir in an ICP-MS (Agilent 7500cx). Ball-milled Douglas fir and acid-washed Douglas fir samples (∼0.15 g) were with 3 mL concentrated nitric acid

130

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

Table 2 Py–GC/MS parameters. Parameter

Details

Sample mass Pyroprobe chamber temperature Pyroprobe coil temperature GC column

300–500 ␮g 250 ◦ C 500 ◦ C, hold 1 min Agilent HP-5MS, 30 m long, internal diameter 250 ␮m, thickness 0.25 ␮m 250 ◦ C 150 230 280 Start at 40 ◦ C, hold 1 min Heat 40–280 at 6 ◦ C/min, hold 15 min 48.768 kPa 1 mL/min 50 20 mL/min at 2 min 30 400 4 100

Oven inlet temperature MSD setpoint quad MSD setpoint source MSD temperature GC temperature program He gas inlet pressure He flow rate Split ratio Gas saver Low mass scan m/z High mass scan m/z Samples scanned Scan threshold

and 2 mL 30% H2 O2 in a microwave digester (SP-D, CEM corporation) at 300 ◦ C and 250 psi for 5 min. A five minute ramp was used to reach digestion conditions. Digested samples were diluted to 100 mL using E-pure water. 1 mL internal standard was added (10 ␮g/mL Bi, Ge, In, Li-6, Sc, Tb, Y; Accustandard, Inc., AG-INT-ASL5, lot 212025167). Operational power was set to 1580–1600 W. Argon carrier gas was set to 0.9 L min−1 for the nebulization and 0.25 L min−1 for the make-up. Parameters were adjusted to obtain good sensitivity for Li6+, Y, and Tl at a 1 ppb concentration. Calibration was done with multi-element standards (Accustandard Inc., AG-CAL-ASL-5, lot 211115029) ranging from 100 ppb to 100 ppm for Na, K, Mg, Ca and Fe and 1 ppb to 1 ppm for all other elements. All standards samples included the internal standard. Corrected calibration curves showed a linearity with R2 of at least 0.995 for the elements studied. 2.4. Py–GC/MS The pyrolysis studies were all carried out at 500 ◦ C. The pyrolysis-gas chromatography/mass spectrometry (Py–GC/MS) system in the Analytical Chemistry Service Center at WSU consists of a CDS Pyroprobe 5000 (CDS Analytical Inc., Oxford, PA) connected to a 6890 N Network GC system with a 5975B mass spectrometer (Agilent Technologies, Santa Clara, CA). The GC column used was an Agilent HP-5MS, 30 m long, internal diameter 250 ␮m, thickness 0.25 ␮m. Samples were weighed out in a micro-scale between 300 and 500 ␮g. Other GC/MS parameters can be seen in Table 2. The samples were loaded into quartz tubes of 25 mm length and 1.8 mm diameter, held in the center of the tube with a plug of quartz wool at the bottom. The tubes with the quartz wool were cleaned at 700 ◦ C before sample loading to remove any impurity that may be accumulated on it. The pyroprobe was set at 500 ◦ C with a 1 min hold for complete pyrolysis. The pyroprobe chamber was held at 250 ◦ C during this time. The pyroprobe temperature was calibrated with a type-K thermocouple placed in the center of the quartz tube, touching the quartz wool. Due to the nature of the pyroprobe system attached to the Py–GC/MS, it was not possible to accurately calibrate for individual compounds. However, integration does provide semi-quantitative information in the form of peak area per mass of sample; this allows for real comparison between each trial condition. Integrations were performed in MSD Chemstation D.03.00.611 (Agilent Technologies, Santa Clara, CA). Peaks were integrated using selected ion integration, which dramatically reduces noise. Each peak was integrated above the background noise by hand. Compounds were identified

Table 3 Metal composition by ICP-MS for raw Douglas fir, ball-milled Douglas fir, and acidwashed Douglas fir. The acid washing procedure dramatically reduced the content of all metals except for aluminum, whose primary presence is due to the ball-milling procedure. Units are in ppm. N.B.

Na Mg K Ca Fe Al Mn Zn Ba

Raw DFW

Ballmilled DFW

Acid washed, ballmilled DFW

* 80.18 464.3 571.5 22.86 10.47 21.44 3.965 14.41

20.45 112.6 497.0 638.2 32.46 1031. 21.51 5.346 64.42

0.8544 23.96 36.66 27.35 5.979 803.9 0.3083 0.3702 1.530

*Below quantification limit.

using the NIST Mass Spectral Search Program (2.0) coupled with the NIST/EPA/NIH mass spectral database (V 2.0, 2005). Table 4 contains compounds, their peak centers, and the ions selected for integration. 3. Results 3.1. Metal and ash composition before and after acid washing The ash content for the Douglas fir wood feedstock was reduced from 0.319 wt.% to 0.141 wt.% using the 1 wt.% nitric acid washing procedure. It should be noted that the ash content value was corrected for the mass of ceramic added by the ball-milling procedure, which contributed 1.12 wt. %. This is significant because the acid-washing was performed on ball-milled wood. However, to truly understand the effect of the acid washing, ICP-MS analysis was performed on the wood samples to provide selected metal compositions (see Table 3). It should be noted that the ball milling significantly increased the content of aluminum, copper, and barium due to the ceramic added through the grinding. Most importantly, the acid-washing procedure reduced the concentrations of Mg, K, Ca, Mn, Ni, Cu, Zn, and Pb to quantification limits from their respective concentrations in the milled Douglas fir. 3.2. GC/MS characterization The purpose of this study was to understand the effect of acid washing and acid addition on the compounds detectable by GC/MS. This limits the compounds to a mass range of about 32–200 amu. The 35 most abundant compounds found in our study are shown in Table 3. These compounds have been previously identified and reported by many other researchers [47,51]. It is estimated that there are thousands of unique compounds released during biomass pyrolysis [52], but there is clearly a countable number of main players which are favorably formed by the natural kinetics of the process. Furthermore, many of the products are too large to be seen by the Py–GC/MS system. Of these compounds we chose 17 for integrating. These compounds can be categorized by their sources in the biomass: cellulose, hemicellulose, lignin, wood extractives, and protein derived products. Fig. 2 shows the total ion chromatograms (TICs) divided by their respective sample masses for the dried wood, acid washed wood, and 0.05 wt.% acid loading for sulfuric acid, acetic acid, phosphoric acid, and nitric acid. The broad, large peak at 22 min is levoglucosan. It is clear at first glance that the acid-wash dramatically increases its production, and the acid loading further enhances it. The following results section will discuss the semiquantitative results by their probable chemical sources within the wood.

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

131

Table 4 Most abundant compounds with their respective retention times, molecular weights, and ions used for integrations. Compounds for which the extract ions are listed were used for integration across all acids and acid loadings performed. No.

Compound name

Retention time (min)

Molecular weight (amu)

Extract ion (m/z)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

Carbon dioxide Acetone Hydroxyacetaldehyde Methyl vinyl ketone 2-Butanone Acetic acid 1-Hydroxy-2-propanone a.k.a. Acetol 1,2-Ethanediol, monoacetate 2-Oxo propanoic acid methyl ester 3-Furaldehyde Furfural 2-Propylfuran Ethanone, 1-(2-furanyl)2-Hydroxy-2-cyclopenten-1-one 2,2-Diethyl-3-methyl-oxazolidine 2-methoxy phenol Levoglucosenone 2-methoxy-4-methyl phenol 1,4:3,6-Dianhydro-␣-d-glucopyranose Hydroxymethylfurfural a.k.a. HMF 4-Ethylguaiacol 2-Methoxy-4-vinylphenol Unknown compound (likely a sugar) Eugenol Vanillin 2-Methoxy-4-(1-propenyl)-, (E)-phenol 2-Methoxy-4-propyl-phenol 1-(4-Hydroxy-3-methoxyphenyl)-ethanone Unknown compound Levoglucosan 1,6-Anhydro-␣-d-galactofuranose 4-Hydroxy-2-methoxycinnamaldehyde n-Hexadecanoic acid Octadecanoic acid 10,11-Dianhydro-10-hydroxydibenz(b,f) oxepin

1.6 1.8 2.1 2.1 2.2 2.3 2.7 4.0 4.3 4.8 5.2 5.8 6.9 7.3 9.0 11.3 11.9 13.8 14.4 15.5 15.8 16.6 16.9 17.5 18.6 19.5 19.9 20.4 20.6 21.8 23.9 25.3 28.7 31.9 38.6

44 58 60 70 72 60 74 104 102 96 96 110 110 98 143 124 126 138 144 126 152 150

44 43 60

45 43 73

96

114 109 98 69 97

164 152 164 166 166

164 151

162 162 178 256 284 272

57 178 284

One assumption that we will make during these discussions is that each peak area is proportional its yield from the wood during pyrolysis. For comparisons, we will consider, for example, that if the area per mass of biomass pyrolyzed has doubled, the yield has also doubled. We believe this is a reasonable assumption based on our past experiences with quantification of these compounds, which nearly always have linear calibrations. To effectively compare acid conditions, each integrated peak area is divided by the mass of the pyrolyzed wood sample. Of the compounds identified we were only quantified the areas of 17 molecules. For all these molecules we conducted a semiquantitative analysis to evaluate the effect of additive on their production. A semiquantitative estimation of the yield of products versus the concentration of the acids studied is shown in Figs. 3–8. Fig. 7. Carbon dioxide. Error bars represent the 95% confidence interval. The curves were drawn to indicate trends.

3.2.1. Cellulose derived compounds The chemical species that derive from cellulose detectable by a GC/MS are often described as low molecular weight compounds. During pyrolysis, it is understood that the primary reaction that cellulose undergoes produces a continuum of anhydrosugars ranging from the single unit (levoglucosan) up through cellobiosan, cellotriosan, etc. However, under atmospheric pressure conditions, typically only levoglucosan, cellobiosan and some cellotriosan are observed analytically due to the thermodynamic properties of evaporation [53] as well as physical thermal ejection of the larger sugars [54]. Furthermore, levoglucosan is the only sugar typically observed by this instrument without using special preparation techniques like silation, which would be impossible with the pyroprobe unit. Nevertheless, levoglucosan and a multitude of dehydrated rearranged compounds are seen and provide insight

into the effect of mineral removal and mild acid loading in the pyrolysis of soft wood. Fig. 3 presents the plots and of these compounds. The first effect studied is the acid-washing. It is clear that the nitric acid wash alone greatly enhances the yield of levoglucosan. Following our assumptions, the acid washing appears to double the yield of this sugar, considering the error bars. This is consistent with quantitative results obtained by others who saw their sugar yield from wood increase by over 10% with organic acid washing [49] or nearly 30% with strong sulfuric acid washing [45]. This phenomenon is well-studied and it is understood that alkali and alkaline earth metals, particularly potassium, sodium, and calcium, can inhibit the unzipping pyrolysis reaction of cellulose and promote homolytic fission within glucose rings, leading to glycoaldehyde, formic acid, acetol, 2-furaldehyde, and

132

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

Fig. 2. Total ion chromatograms divided by their respected sample masses for dry, unwashed Douglas fir, acid-washed Douglas fir, and 0.05 wt.% acid loaded washed samples with sulfuric acid, acetic acid, nitric acid, and phosphoric acid. The most notable change is the broad levoglucosan peak at 22 min, which dramatically increases from the acid-washed sample to the acid-loaded samples. Peak identities can be seen in Table 4.

5-hydroxymethyl furfural (HMF) [28]. Our results support their hypothesis, as seen in the plots for hydroxymethyl furfural and acetol as well as hydroxyacetaldehyde and 1,2-ethanediolacetate in Fig. 3. The most notable results of this study is that after the acid wash impregnation with small quantities of sulfuric acid, acetic acid, and phosphoric acid further enhance the production of levoglucosan. Sulfuric acid, which provided the best sugar yield at a 0.05% loading (mass acid/mass wood), appears to increase the sugar yield by over 85% from the acid washed material and nearly 400% from the unwashed material. Future work will certainly need to be done to quantify the sugar yield at this concentration. However, sulfuric acid was not the only performer. Nitric acid and acetic acid also enhanced the sugar yield by 50% compared with acid washed material and 300% from the unwashed material. The increase in the production of levoglucosan at low concentrations can be explained by the effect of this acid mitigating the interactions between cellulose and the other polymers (hemicellulose and lignin) in the lignocellulosic matrix. As the concentrations of sulfuric acid increase the dehydration reactions also increase with the consequent reduction in the production of levoglucosan and the increase in the production of levoglucosenone. In the case of phosphoric acid, we can see a similar effect but the dehydration occurs to a lesser degree. Notably, the acetic acid has a desirable feature of being rather insensitive to the amount of acid added to the wood. While all the other acids begin to drop the yield of levoglucosan beyond a loading of 0.05 wt.%, acetic acid remains constant up to our highest loading of 0.5 wt.%. Sulfuric acid, in fact, reduces the levoglucosan yield to below levels of the raw biomass by 0.5 wt.% acid load-

ing, and nitric and phosphoric acid appear to follow that path as well but at a slower rate. The nitric and acetic acid do not seem to contribute much to the dehydration of levoglucosan to produce levoglucosenone. Among these acids, acetic acid is the one that allows to obtain higher concentrations of levoglucosan and the most important outcome is that there is a relatively large range of concentrations that allows to obtain high yields of levoglucosan. It is important to remember that the acetic acid is an important product of biomass pyrolysis and as such it is natural to think that the aqueous phase rich in acetic acid could be an excellent additive to improve the production of levoglucosan [49]. So, where does the levoglucosan go with the higher inorganic acid loads? Evidently, the sugar converts to levoglucosenone and 1,4:3,6-dianhydro-␣-d-glucopyranose (DHGP) (Fig. 3). Both of these compounds are products of dehydration reactions that are thought to form during secondary liquid phase reactions from anhydrosugars, like levoglucosan. In fact, the formation of these sugars is well known to be greatly enhanced by sulfuric acid and phosphoric acid [55,56]. Another trend of interest is seen in furfural (Fig. 3). Acid washing does not significantly affect furfural production. While nitric acid, and phosphoric acid do not significantly alter the furfural production beyond the experimental error, sulfuric acid dramatically enhances the furfural production by around 400% at 0.05 and 0.1 wt.% acid loadings. However, its yield returns to starting levels by 0.3 wt.% sulfuric acid loading. Acetic acid appears to enhance furfural production by up to 100% across all concentrations. Furfural is thought to be produced via fragmentation of the pyranose sugar ring and dehydration reactions [57].

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

133

Fig. 3. Low molecular weight compounds from cellulose or hemicellulose. Error bars represent the 95% confidence interval. The curves were drawn to indicate trends.

A more recent study showed that furans like hydroxymethyl furfural (HMF) and furfural can be formed directly from the cellulose chain, without passing through a levoglucosan or other intermediate [5]. This explains why the furan compounds we studied do not follow the same trend as levoglucosenone, which is likely formed as a secondary product from levoglucosan. However, it is important to

note that acetic acid and nitric acid enhance the formation of HMF, while sulfuric acid and phosphoric acid have almost no significant impact. Acetol, 1,2-ethanediol acetate, and hydroxyacetaldehyde are fragmentation products from cellulose pyrolysis, which are all catalyzed by the presence of AAEMs [22]. Therefore it is logical that the

134

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

350 Raw (dried, unwashed)

Peak area/μg sample (in thousands)

300

Acid washed Acid washed + MeOH

250

200

150

100

50

0 Fig. 4. Levoglucosan raw, acid washed, and acid washed with methanol. Error bars represent the 95% confidence interval.

Fig. 5. Low molecular weight compounds from hemicellulose. Error bars represent the 95% confidence interval. The curves were drawn to indicate trends.

acid washing dramatically reduced their formation, as exhibited in Fig. 3, and is consistent with previous observations by our group [43]. Acetic acid and nitric acid have nearly no significant impact on the yields of these fragmentation products. However, sulfuric acid and phosphoric acid reduce their production – most notably for hydroxyacetaldehyde. It is thought that hydroxyacetaldehyde is formed through free-radical rearrangement and cleavage of the glycosidic bond directly on the cellulose chain [5]. It is quite possible that the higher molecular weights of sulfuric acid and phosphoric acid compared with nitric and acetic acid allow them to catalyze ionic dehydration reactions to form levoglucosenone and DHGP at lower temperatures than high-temperature free-radical reactions. This hypothesis could be later tested using quantum mechanical simulations. A possible explanation for the increase in the levoglucosan yield was that the methanol, which was used to solubilize the acids for easy evaporation, has some effect on the process. However, Fig. 4 shows that the yields of levoglucosan for the acid washed material and that material with methanol added then evaporated are within each other’s 95% confidence interval. It can thus be concluded that the methanol did not significantly affect the products derived from cellulose.

As can be observed in the figure, the only acid that has an effect on the increase in the production of furfural is sulfuric acid. This result can be explained by the very strong dehydrating capacity of this acid. 3.2.2. Hemicellulose derived compounds Acetic acid production (Fig. 5) is significantly reduced through mineral removal, as consistent with previous work by our group [43]. This compound is well-known to be derived primarily from the pyrolysis of hemicellulose, particularly from the O-acetylxylan and 4-O-methylglucuronic acid units in the biopolymer [58]. The production of acetic acid is significantly reduced by the acid washing alone, due to the fact that its mechanism is likely free-radical in its nature and AAEMs can catalyze free-radical reactions. Trends appear with various acid loadings, but the experimental error is too large to call them significant. Naturally, the acetic acid yield is higher for those samples impregnated with that compound. Acetone (Fig. 5) is a bit more complicated to draw conclusions from because it can be derived from both cellulose and hemicellulose. It is immediately apparent that the removal of AAEMs reduces acetone production by about 40%. Acetic acid and nitric acid impregnations do not significantly affect the acetone yield. However, sulfuric acid and phosphoric acid reduce acetone produc-

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

135

Fig. 6. Phenolic compounds from lignin. Error bars represent the 95% confidence interval. The curves were drawn to indicate trends.

Fig. 8. Octodecanoic acid and 2,2-diethy-3-metholoxazolidine. Error bars represent the 95% confidence interval. The curves were drawn to indicate trends.

tion by about 50%. Acetone is thought to be formed by free radical mechanisms from the xylan unit of hemicellulose through primary reactions on the chains or secondary reactions of the molecules already released [58]. The behavior of the acetic acid molecule is similar to the behavior observed for CO2 (see Fig. 7). The addition of phosphoric acid does not seem to affect the production of this molecule. The

addition of acetic acid is reflected in the gradual increase in the concentration of this molecule.

3.2.3. Lignin derived compounds Vanillin, isoeugenol, 4-hydroxy-methoxycinnamaldehyde, and 2-methoxy phenol follow the same trend for the effect of acid wash-

136

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

ing, as seen in Fig. 6. The mineral removal through acid washing appears to decrease the yields of these compounds by about 50%. However, the story becomes more complex with acid impregnation. Acetic acid and nitric enhance the yields of vanillin and isoeugenol. Sulfuric acid and phosphoric acid initially have no statistically significant impact, but reduce the production of vanillin, isoeugenol, and 4-hydroxy-2-methoxycinnamaldehyde. This result is observed most dramatically with 4-hydroxymethoxycinnamaldehyde, whose yield is reduced to the low detection limit for sulfuric acid additions at or above 0.3 wt.% loading. The 2-methoxyphenol yield is not significantly affected by the acids, but is slightly enhanced at the low concentrations (<0.1 wt.% loading) of sulfuric acid. These results are consistent with a previous study performed by our group which found that sulfuric acid can catalyze some dehydration reactions in compounds with methoxyl groups, like those which we present here [38]. 2-Methoxyphenol is unchanged with acid addition, possibly because it has a lower boiling point due to its molecular weight. This indicates that the reactions that form lignin monomers occur in the liquid intermediate phase, where the acids are likely to have an impact. The new finding here is that sulfuric and phosphoric acid appear to have the strongest impact on the lignin monomeric product formation compared with acetic and nitric acid. In the case of vanillin we can observe an increase in yield when acetic acid and nitric acids were used. In the case of the 4-hydroxy2-methoxycinnamaldehyde the only acid that contributes to its formation is the nitric acid. 3.2.4. Other notable compounds Carbon dioxide, seen in Fig. 7, is suppressed by the removal of AAEMs. It is understood that CO2 is formed by decarboxylation reactions, and generally correlates with a decrease in the yield of lignin products that contain carboxylic acids. There are no significant trends with the production of CO2 , which corresponds well with the theory that CO2 is directly released from the lignin primary reactions with fragmentation of carboxylic acid groups, likely through a free-radical mechanism. Fatty acids and nitrogenous compounds have not been well discussed in the literature for woody biomass. However, it is known that octadecanoic acid is derived from the fatty acids in wood extractives. This study found that the yields of this compound were not significantly affected by acid impregnation. 2,2-Diethyl-3-methyloxazolidine is a heterocyclic nitrogenous compound formed from either a protein or nucleic acid source. Although it has been observed by others via Py–GC/MS, it has not been well-discussed [47]. However, as can be seen in Fig. 8, acid washing has no significant effect but excess sulfuric acid suppresses its production. Based on these results it can be concluded that sulfuric acid performs the best for impregnation of the pre-washed wood to improve the sugar yield. However, beyond a very small added level of about 0.1% loading, the sugar yields decrease and the lignin product quality also decreases. Of most interest, however, is acetic acid, which does not appear to catalyze deleterious reactions at higher concentrations. However, another question remains: why does acid impregnation further improve the levoglucosan yield after mineral removal? One hypothesis could be that the metals suppress residual metal content that remains after the washing. To check this we calculated the total moles based on ICP results presented in Table 3 of Na, Mg, K, and Ca, the metals with the largest impact on pyrolysis products (see Table 5). Using this value, we calculated the theoretical optimum impregnation of the four acids used in this study where H2 SO4 and H3 PO4 are assumed to have two metal complexing sites; however, the optimal quantity of impregnated acid we

Table 5 Moles per gram of significant AAEMs in acid washed Douglas fir, based on ICP information from Table 3. AAEM species

mol/g wood

Na Mg K Ca total (mol/g wood)

3.71E-08 9.86E-07 9.38E-07 6.84E-07 2.64E-06

Table 6 Theoretical optimum acid concentrations based on moles of AAEMS (Table 5), compared with observed optimum acid concentrations. The observed optimum acid concentrations were much higher than the calculated values, signifying phenomena beyond metal complexing enhanced the levoglucosan yield. Acids

Theoretical optimum concentration, g/100 g wood

Observed optimum concentration, g/100 g wood

H2 SO4 (2 sites) HNO3 H3 PO4 (2 sites) Acetic acid

0.0130 0.0167 0.0130 0.0159

0.05 0.05 0.05 0.3+

observed was approximately four times that which was calculated (see Table 6). Therefore, to conclude that the impregnated acids only enhanced the levoglucosan yield via metal complexing, as has been reported elsewhere, is likely insufficient. Another hypothesis for the enhancement of the sugar yield at higher acid concentrations is that the acid inhibits deleterious cellulose–lignin interactions. It has been shown that lignin can enhance the production of small compounds when pyrolyzed with neat cellulose [59]. However, a more recent study observed that in wood, cellulose-lignin interactions do not significantly enhance depolymerization reactions compared with switchgrass [60]. These results were attributed to the smaller quantity of cellulose–lignin covalent bonds in wood compared with the grass. It is known that cellulose depolymerization reactions are inhibited by ether groups, which act as base catalysts, especially when electrons are stabilized by an adjacent aromatic ring like in lignin [61,62]. Furthermore, it has been proposed that the depolymerization reaction of cellulose is catalyzed by proton donation [46,63,64], be it from cellulose or any other source. Therefore, we propose that the presence of a very small quantity of acid is able to ward of some of the cellulose–lignin interactions that typically limit the production of 1,6-anhydrosugars in the pyrolysis of woody biomass. It is possible that the acids react with the slightly basic ether groups. Furthermore, strong acids catalyzed dehydration reactions beyond this concentration, but weak acids like acetic acid did not. 4. Conclusions ICP-MS studies of the biomass after acid wash indicate that most of the alkalines were removed by this procedure. Semi-quantitative Py–GC/MS results suggest that sulfuric acid is the acid-additive studied with the best capacity to increase the production of levoglucosan. However, sulfuric acid like phosphoric acid significantly catalyzed dehydration reactions at concentrations higher than optimal 0.05 wt.% impregnation. Acetic acid could prove much more valuable for industrial application due to issues with the very small concentration range in which sulfuric acid has a positive effect. The acetic acid does not seem to catalyze dehydration reactions responsible for the decrease in levoglucosan yields. Using the nitric acid wash with nitric or acetic acid impregnation, up to a 3-fold increase in the production of levoglucosan compared with raw Douglas fir using acetic or nitric acid. Removal of the AAEMs

B. Pecha et al. / Journal of Analytical and Applied Pyrolysis 114 (2015) 127–137

and acid impregnation decreased the production of methoxylated phenolic compounds from lignin, likely due to the acceleration of dehydration reactions by the presence of protons in the system. Acknowledgements This project was financially supported by the US National Science Foundation (CBET-1434073, CAREER CBET CBET-1150430). The authors are also very thankful to the Agricultural Research Center (NIFA-Hatch-WNP00701) and Analytical Chemistry Service Center (ACSC) in the Department of Biological Systems Engineering at Washington State University. Matthew Smith generously assisted with the ICP-MS studies. Reference: [1] D. Mohan, C.U. Pittman Jr., P.H. Steele, Energy Fuels 20 (2006) 848, http://dx. doi.org/10.1021/ef0502397 [2] J. Piskorz, P. Majerski, D. Radlein, D.S. Scott, Energy Fuels 3 (1989) 723, http:// dx.doi.org/10.1021/ef00018a011 [3] P. Scott, J. Piskorz, D. Radlein, J. Anal. Appl. Pyrolysis 51 (1999) 23, http://dx. doi.org/10.1016/S0165-2370(99)6-6 [4] V. Mamleev, S. Bourbigot, M. Le, Bras, J. Yvon, J. Anal. Appl. Pyrolysis 84 (2009) 1, http://dx.doi.org/10.1016/j.jaap.2008.10.014 [5] M.S. Mettler, S.H. Mushrif, A.D. Paulsen, A.D. Javadekar, D.G. Vlachos, P.J. Dauenhauer, Energy Environ. Sci. 5 (2012) 5414, http://dx.doi.org/10.1039/ C1EE02743C [6] W. Chaiwat, I. Hasegawa, T. Tani, K. Sunagawa, K. Mae, Energy Fuels 23 (2009) 5765, http://dx.doi.org/10.1021/ef900674b [7] T. Hosoya, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 83 (2008) 64, http:// dx.doi.org/10.1016/j.jaap.2008.06.008 [8] H.B. Mayes, L.J. Broadbelt, J. Phys. Chem. A 116 (2012) 7098, http://dx.doi.org/ 10.1021/jp300405x [9] E. Berman, Symposium on Levoglucosenone and Levoglucosans Chemistry and Applications, 204th National ACS Meeting, in: Z.J. Witczak (Ed.), ATL, Press, Inc., Washington, D.C, 1994, p. 189. [10] D.S. Scott, J. Piskorz, D. Radlein, Symposium on Levoglucosenone and Levoglucosans Chemistry and Applications, 204th National ACS Meeting, in: Z.J. Witczak (Ed.), ATL, Press, Inc., Washington, D.C, 1994, p. 179. [11] N.M. Bennett, S.S. Helle, S.J.B. Duff, Bioresource Technol. 100 (2009) 6059, http://dx.doi.org/10.1016/j.biortech.2009.06.067 [12] E.M. Prosen, D. Radlein, J. Piskorz, D.S. Scott, R.L. Legge, Biotechnol. Bioeng. 42 (1993) 538, http://dx.doi.org/10.1002/bit.260420419 [13] J.N. Lian, S.L. Chen, S.A. Zhou, Z.H. Wang, J. O’Fallon, C.Z. Li, M. Garcia-Perez, Bioresource Technol. 101 (2010) 9688, http://dx.doi.org/10.1016/j.biortech. 2010.07.071 [14] J. Lian, M. Garcia-Perez, S. Chen, Bioresource Technol. (2013), http://dx.doi. org/10.1016/j.biortech.2013.01.031 [15] D. Chiaramonti, A. Oasmaa, Y. Solantausta, Renew. Sustain. Energy Rev. 11 (2007) 1056, http://dx.doi.org/10.1016/j.rser.2005.07.008 [16] G.J. Kwon, D.Y. Kim, S. Kimura, S. Kuga, J. Anal. Appl. Pyrolysis 80 (2007) 1, http://dx.doi.org/10.1016/j.jaap.2006.12 [17] S. Zhou, M. Garcia-Perez, B. Pecha, A.G. McDonald, S.R.A. Kersten, R.J.M. Westerhof, Energy Fuels 27 (2013) 1428, http://dx.doi.org/10.1021/ef3019832 [18] M. Garcia-Perez, S. Wang, J. Shen, M. Rhodes, W.J. Lee, C.-Z. Li, Energy Fuels 22 (2008) 2022, http://dx.doi.org/10.1021/ef7007634 [19] T.J. Haas, M.R. Nimlos, B.S. Donohoe, Energy Fuels 23 (2009) 3810, http://dx. doi.org/10.1021/ef900201b [20] M. Nik-Azar, M.R. Hajaligol, M. Sohrabi, B. Dabir, Fuel Process. Technol. 51 (1997) 7, http://dx.doi.org/10.1016/S0378-3820(96)1074-0 [21] K. Raveendran, A. Ganesh, K.C. Khilar, Fuel 74 (1995) 1812, http://dx.doi.org/ 10.1016/0016-2361(95)80013-8 [22] D. Mourant, Z. Wang, M. He, X.S. Wang, M. Garcia-Perez, K. Ling, C.-Z. Li, Fuel 90 (2011) 2915, http://dx.doi.org/10.1016/j.fuel.2011.04.033 [23] N. Kuzhiyil, D. Dalluge, X. Bai, K.H. Kim, R.C. Brown, ChemSusChem 5 (2012) 2228, http://dx.doi.org/10.1002/cssc.201200341 [24] D.S. Scott, J. Piskorz, D. Radlein, Ind. Eng. Chem. Process Des. Dev. 24 (1985) 581, http://dx.doi.org/10.1021/i200030a011 [25] W.-P. Pan, G.N. Richards, J. Anal. Appl. Pyrolysis 16 (1989) 117. [26] A. Oasmaa, E. Kuoppala, Y. Solantausta, Energy Fuels 17 (2003) 433, http://dx. doi.org/10.1021/ef020206g [27] A. Oasmaa, E. Kuoppala, Energy Fuels 17 (2003) 1075.

137

[28] P.R. Patwardhan, J.A. Satrio, R.C. Brown, B.H. Shanks, Bioresource Technol. 101 (2010) 4646, http://dx.doi.org/10.1016/j.biortech.2010.01.112 [29] C. Di Blasi, A. Galgano, C. Branca, Ind. Eng. Chem. Res. 48 (2009) 3359, http:// dx.doi.org/10.1021/ie801468y [30] N. Shimada, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 81 (2008) 80, http:// dx.doi.org/10.1016/j.jaap.2007.09.005 [31] M.R. Gray, W.H. Corcoran, G.R. Gavalas, Ind. Eng. Chem. Process Des. Dev. 24 (1985) 646, http://dx.doi.org/10.1021/i200030a020 [32] K. Tamaru, Bull. Chem. Soc. Jpn. 24 (1951) 164, http://dx.doi.org/10.1246/bcsj. 24.164 [33] G. Varhegyi, M.J. Antal Jr., T. Szekely, F. Till, E. Jakab, Energy Fuels 2 (1988) 267. [34] H.B. Mayes, M.W. Nolte, G.T. Beckham, B.H. Shanks, L.J. Broadbelt, ACS Catal. 5 (2014) 192, http://dx.doi.org/10.1021/cs501125n [35] M.R. Nimlos, S.J. Blanksby, G.B. Ellison, R.J. Evans, J. Anal. Appl. Pyrolysis 66 (2003) 3. [36] H. Kawamoto, D. Yamamoto, S. Saka, J. Wood Sci. 54 (2008) 242, http://dx.doi. org/10.1007/s10086-007-0930-8 [37] P.R. Patwardhan, R.C. Brown, B.H. Shanks, ChemSusChem 4 (2011) 636. [38] S. Zhou, N.B. Osman, H. Li, A.G. McDonald, D. Mourant, C.Z. Li, M. Garcia-Perez, Fuel 103 (2013) 512, http://dx.doi.org/10.1016/j.fuel.2012.07.052 [39] D.J. Nowakowski, J.M. Jones, R.M. Brydson, A.B. Ross, Fuel 86 (2007) 2389. [40] F. Shafizadeh, T.T. Stevenson, J. Appl. Polym. Sci. 27 (1982) 4577, http://dx.doi. org/10.1002/app.1982.070271205 [41] S.R.G. Oudenhoven, R.J.M. Westerhof, N. Aldenkamp, D.W.F. Brilman, S.R.A. Kersten, J. Anal. Appl. Pyrolysis (2013), http://dx.doi.org/10.1016/j.jaap.2012. 10.002 [42] I.-Y. Eom, J.-Y. Kim, S.-M. Lee, T.-S. Cho, H. Yeo, J.-W. Choi, Bioresource Technol. 128 (2013) 664. [43] S. Zhou, D. Mourant, C. Lievens, Y. Wang, C.-Z. Li, M. Garcia-Perez, Fuel 104 (2013) 536, http://dx.doi.org/10.1016/j.fuel.2012.06.010 [44] K.H. Kim, R.C. Brown, X. Bai, Fuel 130 (2014) 135, http://dx.doi.org/10.1016/j. fuel.2014.04.044 [45] J. Piskorz, D.S. Radlein, D.S. Scott, S. Czernik, J. Anal. Appl. Pyrolysis 16 (1989) 127, http://dx.doi.org/10.1016/0165-2370(89)85012-0 [46] S. Matsuoka, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 106 (2014) 138. [47] S. Zhou, Z. Wang, S.-S. Liaw, C.-Z. Li, M. Garcia-Perez, J. Anal. Appl. Pyrolysis 104 (2013) 117, http://dx.doi.org/10.1016/j.jaap.2013.08.013 [48] T. Hosoya, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 78 (2007) 328, http:// dx.doi.org/10.1016/j.jaap.2006.008 [49] S.R.G. Oudenhoven, R.J.M. Westerhof, N. Aldenkamp, D.W.F. Brilman, S.R.A. Kersten, J. Anal. Appl. Pyrolysis 103 (2013) 112, http://dx.doi.org/10.1016/j. jaap.2012.10.002 [50] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, Natl. Renew. Energy Lab. (2008) 1. [51] X. Zhou, M.W. Nolte, B.H. Shanks, L.J. Broadbelt, Ind. Eng. Chem. Res. (2014), http://dx.doi.org/10.1021/ie502260q [52] T. Kekäläinen, T. Venäläinen, J. Jänis, Energy Fuels (2014), http://dx.doi.org/10. 1021/ef500849z [53] T. Shoji, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis (2014), http://dx.doi. org/10.1016/j.jaap.2014.06.014 [54] A.R. Teixeira, K.G. Mooney, J.S. Kruger, C.L. Williams, W.J. Suszynski, L.D. Schmidt, D.P. Schmidt, P.J. Dauenhauer, Energy Environ. Sci. 4 (2011) 4306, http://dx.doi.org/10.1039/c1ee01876k [55] A. Broido, M. Evett, C.C. Hodges, Carbohydr. Res. 44 (1975) 267, http://dx.doi. org/10.1016/S0008-6215(00)84170-1 [56] H. Kawamoto, S. Saito, W. Hatanaka, S. Saka, J. Wood Sci. 53 (2007) 127, http://dx.doi.org/10.1007/s10086-006-0835-y [57] J.B. Paine Iii, Y.B. Pithawalla, J.D. Naworal, J. Anal. Appl. Pyrolysis 83 (2008) 37, http://dx.doi.org/10.1016/j.jaap.2008.05.008 [58] D.K. Shen, S. Gu, A.V. Bridgwater, J. Anal. Appl. Pyrolysis 87 (2010) 199, http:// dx.doi.org/10.1016/j.jaap.2009.12.001 [59] T. Hosoya, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 80 (2007) 118, http:// dx.doi.org/10.1016/j.jaap.2007.01.006 [60] J. Zhang, Y.S. Choi, C.G. Yoo, T.H. Kim, R.C. Brown, B.H. Shanks, ACS Sustain. Chem. Eng. 3 (2014) 293, http://dx.doi.org/10.1021/sc500664h [61] S. Matsuoka, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 93 (2012) 24, http://dx.doi.org/10.1016/j.jaap.2011.09.005 [62] T. Hosoya, H. Kawamoto, S. Saka, Carbohydr. Res. 341 (2006) 2293, http://dx. doi.org/10.1016/j.carres.2006.06.014 [63] S. Matsuoka, H. Kawamoto, S. Saka, Carbohydr. Res. 346 (2011) 272, http://dx. doi.org/10.1016/j.carres.2010.10.018 [64] S. Matsuoka, H. Kawamoto, S. Saka, Polym. Degrad. Stab. 96 (2011) 1242, http://dx.doi.org/10.1016/j.polymdegradstab.2011.04.009.