- Email: [email protected]
A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin
Accepted Manuscript Title: A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin Author: Stylianos D. Stefanidis Konstantinos G. Kalogiannis Eleni F. Iliopoulou Chrysoula M. Michailof Petros A. Pilavachi Angelos A. Lappas PII: DOI: Reference:
S0165-2370(13)00237-4 http://dx.doi.org/doi:10.1016/j.jaap.2013.10.013 JAAP 3076
To appear in:
J. Anal. Appl. Pyrolysis
Received date: Revised date: Accepted date:
13-9-2013 23-10-2013 28-10-2013
Please cite this article as: S.D. Stefanidis, K.G. Kalogiannis, E.F. Iliopoulou, C.M. Michailof, P.A. Pilavachi, A.A. Lappas, A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin, Journal of Analytical and Applied Pyrolysis (2013), http://dx.doi.org/10.1016/j.jaap.2013.10.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin Stylianos D. Stefanidisa, b, *, Konstantinos G. Kalogiannisa, Eleni F. Iliopouloua, Chrysoula M. Michailofa, Petros A. Pilavachib, Angelos A. Lappasa Chemical Process and Energy Resources Institute, Center for Research and Technology Hellas, 57001 Thessaloniki, Greece b
ip t
a
Department of Mechanical Engineering, University of Western Macedonia, 50100 Kozani, Greece
cr
* Corresponding author. Tel: +30 2310 498369, Fax: +30 2310 498380, e-mail: [email protected]
us
Abstract
In this study, thermogravimetric (TG) analyses, along with thermal and catalytic
an
fast pyrolysis experiments of cellulose, hemicellulose, lignin and their mixtures were carried out in order to investigate their pyrolysis products and whether the prediction
M
of the pyrolysis behavior of a certain lignocellulosic biomass feedstock is possible, when its content in these three constituents is known. We were able to accurately predict the final solid residue of mixed component samples in the TG analyses but the
d
differential thermogravimetric (DTG) curves indicated limited heat transfer when
te
more than one component was present in the pyrolyzed sample. The limited heat transfer did not have a significant effect on the TG curves but it affected the product
Ac ce p
distribution in the fast pyrolysis experiments, which resulted in inaccurate calculation of the product yields, when using a simple additive law. In addition, the pyrolysis products of each biomass constituent were characterized in order to study their contribution to the yield and composition of the products from whole biomass pyrolysis. An investigation into the pyrolysis reaction pathways of each component was also carried out, using the bio-oil characterization data from this study and data found in the literature.
Keywords: Biomass; Pyrolysis; Depolymerization; Biomass constituents; Reaction pathways 1. Introduction Increased world energy demand and environmental concerns have driven research on alternative energy sources. Lignocellulosic biomass is considered a promising
1
Page 1 of 34
alternative and renewable energy source that can be converted via the biomass fast pyrolysis process into a liquid product, known as bio-oil, which is considered to be a promising biofuel/bioenergy carrier. The bio-oil is a complex mixture of oxygenated compounds and its composition and quality is heavily dependent on the composition
ip t
of the biomass feedstock. Lignocellulosic biomass is composed mainly of three basic structural components; cellulose, hemicellulose and lignin. The content of these components in biomass varies depending on the biomass type. Woody plant species
cr
have tightly bound fibers and are richer in lignin while herbaceous plants have more
loosely bound fibers, a fact that indicates lower lignin content. Usually, cellulose,
us
hemicellulose and lignin constitute 40-50 wt.%, 20-40 wt.% and 10-40 wt.% of the plant material respectively [1]. In addition to these components, lignocellulosic
an
biomass also contains a small amount of inorganic material (ash) and extractives. It has been suggested that since all lignocellulosic biomass is mainly composed of these basic structural components, the aggregative behavior during pyrolysis of each
M
of those independent components would describe the behavior of any lignocellulosic feed [2]. Therefore, it should be possible to predict the yield and composition of the pyrolysis products of any feedstock, when its composition is known. In order for that
d
to be true however, the pyrolysis behavior of each individual component must be
te
independent of the presence of other components. Otherwise, if synergistic effects were to take place, the prediction of the behavior of a biomass feedstock would be a
Ac ce p
more complex task. In addition, since biomass pyrolysis proceeds through an extremely complex network of reaction mechanisms, the study of the biomass pyrolysis chemistry can be simplified by studying the pyrolysis reactions of cellulose, hemicellulose and lignin individually. Several groups in the literature have studied the pyrolysis of biomass on the basis
of these three main components. Raveendran et al. [2] studied the pyrolysis characteristics of biomass components in a thermogravimetric analyzer and a packedbed pyrolyzer and found no detectable interactions among the components during pyrolysis in either experimental setup. Yang et al. [3] also observed negligible interactions among the three biomass components in their study, when using a thermogravimetric analyzer. On the other hand, Worasuwannarak et al. [4] studied the pyrolysis behavior of cellulose, xylan, lignin and mixtures by TG-MS technique and observed significant interactions between cellulose and lignin that caused a suppression of liquid product formation and an increase in solid residue yield. Wang
2
Page 2 of 34
et al. [5] also observed cellulose-lignin interactions, as well as hemicellulose-lignin interactions, while they reported that hemicellulose and lignin did not seem to affect each other during pyrolysis in a thermogravimetric analyzer. More recently, Wang et al. [6] studied the interactions of the biomass components in both a TG-FTIR and an
ip t
experimental pyrolyzer. They reported no significant differences between the experimental and calculated TG/DTG curves, when using biomass component
mixtures, but differences in the evolution curves of the main products (levoglucosan,
cr
2-furfural, acetic acid and 2,6-dimethoxy phenol) were apparent. In addition, the
mixed samples exhibited a common tendency to form less liquid and more gas
us
products than what the calculations predicted. To the best of our knowledge, the pyrolysis of mixed biomass components with catalytic upgrading of the pyrolysis
an
vapors has not been studied yet.
In this study, the thermal and catalytic pyrolysis of biomass components (cellulose, xylan and lignin) was studied, both in a TG apparatus and in a pyrolysis
M
reactor and the contribution of each component on the yield and the composition of the biomass pyrolysis products was investigated. Moreover, the pyrolysis of mixtures of cellulose, hemicellulose and lignin was studied in order to investigate the
d
contradictory literature results concerning the interactions between cellulose,
te
hemicellulose and lignin during pyrolysis. Finally, based on the results from the pyrolysis of the individual components and reports from the literature, an elucidation
Ac ce p
of the reaction pathways in the absence or presence of a catalyst was attempted. 2. EXPERIMENTAL
2.1. Biomass Feedstock properties The cellulose used in this study was purchased from JRS Pharma with the
commercial name VIVAPUR Type 200 Microcrystalline Cellulose. The lignin used was a kraft lignin isolated from a commercial pulp mill using predominantly Norway spruce as a raw material and was purchased from Sigma-Aldrich (Aldrich 370959). Because hemicellulose was not commercially available, xylan purchased from SigmaAldrich was used instead (Sigma X4252). Xylan is considered to be an adequate replacement for hemicellulose and it is widely used in the literature for this purpose [3]. All biomass samples were dried at 105 °C for 4 h before use and therefore, all
3
Page 3 of 34
results reported here are on a dry basis. The elemental composition of the dry biomass samples is presented in Table 1. 2.2. TG experiments The TG experiments were performed in a TA Instruments SDT2960
ip t
thermogravimetric analyzer to study the decomposition curves upon pyrolysis of the
samples. Experiments were carried out with sample masses of about 10 mg using a
cr
linear heating rate of 10 °C/min within the temperature range of 28 °C to 840 °C and a steady nitrogen flow rate of 100 cm3/min.
us
2.3. Pyrolysis experiments
The pyrolysis experiments were performed at 500 ºC in a bench-scale fixed bed
an
biomass pyrolysis reactor. The reactor was made of stainless steel 316 and measured 14 mm in internal diameter and 41 cm in height. Heat was provided by a three-zone
M
furnace and the temperature of each zone was independently controlled using temperature controllers. The catalyst bed temperature was considered as the pyrolysis temperature and was monitored with a thermowell. The biomass sample (1.5 g) was
d
loaded in a specially designed piston mounted on top of the reactor, outside of the
te
furnace. After the desired pyrolysis temperature in the rector was reached, the piston was used to introduce the biomass feedstock into the hot reactor. A constant stream of
Ac ce p
nitrogen (100 cm3/min) was fed from the top of the reactor for the continuous withdrawal of the pyrolysis vapors and maintenance of the inert atmosphere during the experiment. The vapors were carried through the catalyst bed (0.7 g), exited from the bottom of the reactor in gaseous form and were condensed in a glass receiver submerged in a cooling bath that was kept at -17 °C. Non-condensable gases were collected in a gas collection system, while a glass wool filter placed between the glass receiver and the gas collection system recovered any light condensable liquids that were not condensed in the receiver. The above described catalytic pyrolysis experiments can be referred to as in situ upgrading of pyrolysis vapors and are of the “ex-bed” type (i.e. there was no mixing of solid biomass with the solid catalyst). In addition, all the experimental parameters (i.e., fast heating of biomass, low residence time, fast cooling of products) resemble those of the biomass fast pyrolysis type of experiments. A schematic representation of the experimental setup can be found elsewhere [7].
4
Page 4 of 34
Both thermal and catalytic pyrolysis experiments were performed. Inert silica sand was used in place of the catalyst bed in the thermal pyrolysis experiments, while a commercial ZSM-5 formulation diluted with silica alumina (30 wt.% crystalline zeolite, 138 m2/g surface area, 4 nm average pore diameter, 0.037 cm3/g
ip t
microporosity and 0.018 cm3/g mesoporosity) was used for the catalytic pyrolysis experiments.
The liquid products were collected and quantitatively measured in the pre-
cr
weighted glass receiver. The pyrolytic vapors, upon their condensation in the glass receiver, formed multiple phases; an aqueous phase, a liquid organic phase and
us
viscous organic deposits on the receiver walls. Extensive effort has been put in the development of a method for the collection of a representative bio-oil sample for
an
analysis [7]. Towards this goal, the bio-oil was first fully homogenized inside the receiver using ethyl lactate as the solvent and then collected as a solution, which was then submitted for analysis. Ethyl lactate was chosen for its non-volatility, which
M
minimizes errors during weighing. It also proved to be a good solvent for all the biooil samples in this study.
d
2.4. Analysis methods
te
The elemental composition of the biomass samples and the bio-oil was determined with an elemental CHN LECO-800 analyzer. The water content of the
Ac ce p
bio-oil was measured with Karl-Fischer titration (ASTM E203-08). The water/aqueous phase present in the bio-oil was then separated from the organic phase using an organic solvent (dichloromethane) and the organic phase of the bio-oil was analyzed by GC-MS using an Agilent 7890A/5975C gas chromatograph-mass spectrometer system (Electron energy 70 eV, Emission 300 V, Helium flow rate: 0.7 cc/min, Column: HP-5MS 30 m x 0.25 mm ID x 0.25 μm). Internal libraries were used for the identification of the compounds found in the bio-oil and their categorization into main functional groups. The gaseous products were collected and measured by the water displacement method. The gaseous products were analyzed in a HP 5890 Series II gas chromatograph, equipped with four columns (Precolumn: OV-101, Columns: Porapak N, Molecular Sieve 5A and Rt-Qplot 30 m x 0.53 mm ID) and two detectors (TCD and FID). The amount of the solid residue left in the reactor and deposited on the catalyst surface, consisted mainly of charcoal and coke-
5
Page 5 of 34
on-catalyst formed by thermal and/or catalytic cracking and was determined by direct weighting. 3. RESULTS AND DISCUSSION
ip t
3.1. TG analyses 3.1.1. TG analyses of separate components
cr
The weight loss and derivative weight loss curves of the separate biomass samples are given in Fig. 1. As it can be seen, cellulose decomposed over a narrow
us
temperature range, between 280 °C and 360 °C with the highest decomposition rate being observed at 339 °C and the total solid residue at 500 °C and 800 °C being 10.7
an
and 7.4 wt.% respectively. This very well-defined decomposition temperature range is due to cellulose’s very homogeneous unbranched crystalline structure of linked Dglucose units. Xylan on the other hand, which is also a polysaccharide, decomposed at
M
lower temperatures, over a slightly wider temperature range, between 200 °C and 320 °C, with two distinct peaks; one at 246 °C and one at 295 °C. The lower
d
decomposition temperature range of xylan is attributed to its structure, which is amorphous with many branched units that have low activation energy [8]. The xylan
te
residue at 500 °C and 800 °C was significantly higher than that of cellulose, about 30 and 25 wt.% respectively. This can be attributed, in part, to the high ash content of the
Ac ce p
commercial xylan sample that was used in this study (Table 1). However, Patwardhan et al. [9] also reported higher solid residue yields from xylan pyrolysis, despite using an ash-free sample. They suggested that, due to the different sugars that make up xylan (pentoses and hexoses), pyrolysis proceeds via a different mechanism from that of cellulose, which may account for the higher solid residue formation. Lignin decomposed over a very wide temperature range, from 140 °C to 600 °C,
with a very low-intensity peak around 380 °C. The lignin residue at 500 °C was the highest observed, about 53.4 wt.% and even at 800 °C the residue was 41.2 wt.%. This high residue yield is attributed to lignin’s structure, which consists of a complex network of cross-linked aromatic molecules that are difficult to decompose and therefore have high thermal stability [3].
6
Page 6 of 34
3.1.2. TG analyses of mixtures In order to examine any synergistic effects between the three different biomass components during pyrolysis, two mixed samples were prepared by simple mechanical mixing. For simplicity, one sample contained ½ cellulose and ½ xylan
ip t
(Mix 1), while the other sample also contained lignin and consisted of ⅓ cellulose, ⅓
xylan and ⅓ lignin (Mix 2). TG analysis was performed for both mixed samples at the
cr
same conditions as the separate biomass components and the experimental TG and DTG curves were compared to calculated curves. The calculated curves were
us
determined assuming that the overall pyrolysis behavior of a mixture is the weighted
,
,
and
M
where
an
sum of the partial contributions of its components [2], according to the formulas: (1) (2)
are the weight losses at
temperature T for the mixture, cellulose, xylan and lignin respectively, while Xcel, Xxyl
,
,
and
are the derivative weight
te
Accordingly,
d
and Xlig are the fractions of cellulose, xylan and lignin that are present in the mixture.
losses at temperature T for the mixture, cellulose, xylan and lignin respectively. The
Ac ce p
experimental and calculated TG and DTG curves for Mix 1 are presented in Fig. 2, while the corresponding results for Mix 2 are presented in Fig 3. For Mix 1, which was composed from equal parts of cellulose and hemicellulose,
the experimental and calculated TG curves agreed quite well, but there were obvious differences in the DTG curves. The xylan peaks, that were calculated at 247 °C and 302 °C, turned out as expected, with only a slight drop in intensity, but the cellulose peak had significantly lower intensity than expected and shifted to a slightly higher temperature (349 °C instead of 339 °C). In addition, the cellulose peak was wider than expected. The biomass residue at temperatures greater than 400 °C was predicted with very good accuracy and was found to be about 20 wt.% at 500 °C. In Mix 2 (Fig. 3), the two TG curves agreed quite well too, but the differences between the experimental and the calculated DTG curves were again obvious and more significant than in the case of Mix 1. The drop in the intensity of the cellulose
7
Page 7 of 34
peak was more pronounced in Mix 2, compared to Mix 1. The shift in the temperature of the cellulose peak was also more pronounced, as it was expected at 339 °C but it appeared at 353 °C. In addition, the experimental cellulose peak was wider than the calculated. No visible lignin peak was observed in the experimental DTG curve of
ip t
Mix 2, due to its low intensity and the overlapping cellulose peak. The biomass residue at temperatures greater than 400 °C was predicted with very good accuracy and it was found to be about 31% at 500 °C.
cr
Overall, the experimental and calculated TG curves of both mixtures agreed quite
well and were able to predict the thermal behavior of the mixed biomass samples with
us
accuracy. The DTG curves however, were more suitable to demonstrate the subtle but important differences in the thermal behavior of the mixed samples. While the
an
thermal decomposition of xylan did not seem to be significantly affected by the presence of the other components, especially in Mix 1, the cellulose peaks indicated a change in the thermal behavior of cellulose. More specifically, in both mixed samples,
M
the maximum weight loss rate of cellulose decreased significantly and shifted to a higher temperature. Moreover, the temperature range of the cellulose decomposition was wider in the mixed samples. These observations indicate a hindrance in the heat
d
transfer from the heated plate of the TG instrument towards the cellulose particles,
te
which appeared to decompose at higher temperatures and over a wider range due to the possible difference between the temperature that was recorded by the TG
Ac ce p
instrument and the actual temperature of the mixed biomass sample (that was in fact lower). The hemicellulose particles, which decomposed at lower temperatures, formed a charred film that wrapped around the cellulose particles and acted as a heat barrier. Hosoya et al. [10] and Wang et al. [6] made similar observations about the charred hemicellulose forming around cellulose particles and suggested that it could limit mass transfer during the volatilization of the cellulose particles. Our results indicate limited heat transfer, in addition to possible limited mass transfer that was suggested by the other research groups [6, 10]. From our experience, a similar phenomenon occurs with lignin, which starts to melt and form agglomerates at temperatures a little over 100 °C. Such alterations in the heat and mass transfer rates were not of significant importance in the experimental conditions of the TG experiments, hence the non-obvious differences between the experimental and the calculated TG curves. During the fast pyrolysis process however, rapid heat transfer rates and quick removal
8
Page 8 of 34
of the pyrolysis vapors are crucial parameters and alteration of these rates would result in different product yield distributions. Another interesting observation was that the presence of ash in the mixed samples did not seem to have an effect on the pyrolysis behavior of cellulose. The effect of ash
ip t
in the pyrolysis of biomass has been extensively studied and is known to shift the decomposition of the sample towards lower temperatures [13]. Therefore, the
cellulose peak in Mix 1 and Mix 2 was expected to shift to a lower temperature, due
cr
to the presence of xylan in the sample, which had high ash content. No such shift was observed however. It seems that in the case of simple mechanical mixing that was
us
used to produce the mixed samples, the presence of ash in hemicellulose and lignin did not affect the decomposition of cellulose. This is in agreement with Couhert et al.
an
[11] who observed that ash only slightly affected pyrolysis behavior when simple mixing with biomass was applied.
M
3.2. Pyrolysis experiments 3.2.1. Pyrolysis of separate components
d
The product yield distributions from the thermal and catalytic pyrolysis of cellulose, xylan and lignin are given in Fig. 4. The pyrolysis of cellulose yielded the
te
most bio-oil, with high organic fraction and low water content. As in the TG
Ac ce p
experiments (section 3.1.1), out of the three components studied, cellulose yielded the lowest solid residue (21 wt.%). Xylan on the other hand, gave high yields of gas products and bio-oil rich in water and poor in organic fraction (Fig. 4). As seen in the TG experiments (Fig. 1), xylan decomposed at lower temperatures than the temperature at which the pyrolysis experiments were performed (500 °C). At such high temperature, the pyrolysis vapors of xylan underwent significant secondary cracking and gave high yields of water, gas products and coke. The significant cracking of the xylan pyrolysis vapors could also be attributed to the fact that the commercial xylan sample used in this study had high ash content, which is known to promote gas formation at the expense of liquid products [12]. The pyrolysis of lignin gave moderate yields of bio-oil, very low gas yields and remarkably high solid residue yields, in accordance with our TG experimental results (Fig. 1). When using ZSM-5 catalyst for the upgrading of the pyrolysis vapors, the water, gas and solid residue yields increased, while the organic and total liquid product
9
Page 9 of 34
yields decreased, as was observed with regular biomass in our previous study [7]. The decrease in the total liquid product yield and the increase in the water and gas yields are due to the deoxygenation reactions that occur on the catalyst, which remove oxygen from the pyrolysis vapors as CO, CO2 and H2O. The solid product yield
ip t
increased due to the formation of coke deposits on the surface and the pores of the catalyst.
The chemical composition of the organic phase of the bio-oil obtained from both
cr
thermal and catalytic pyrolysis of cellulose, xylan and lignin, was analyzed with a
GC-MS system and is presented in Fig. 5, Fig. 6 and Fig. 7 respectively. Usually,
us
hundreds of compounds can be identified in the organic phase of the bio-oil and in order to make the comparison of the results easier, the identified compounds were
an
categorized in groups. These groups were aromatic hydrocarbons (AR), aliphatic hydrocarbons (ALI), phenolic compounds (PH), furans (FUR), acids (AC), esters (EST), alcohols (AL), ethers (ETH), aldehydes (ALD), ketones (KET), polycyclic
M
aromatic hydrocarbons (PAH), sugars (SUG) and compounds that contain nitrogen (NIT).
The organic phase of the bio-oil from the thermal pyrolysis of cellulose could not
d
be separated from the aqueous phase and was analyzed as it was. The main products
te
from the thermal pyrolysis of cellulose were dehydrated sugars. A wide peak identified as 1,6-Anhydro-β-D-glucopyranose (levoglucosan) covered a large portion
Ac ce p
of the chromatogram. A sharp peak of 1,4:3,6-Dianhydro-α-d-glucopyranose was also identified in the samples. Other products that covered a significantly smaller portion of the chromatogram were identified as well, such as some simple phenols (phenol and phenol molecules with methyl substitutes), ketones (mainly cyclic ketones), aldehydes and alcohols. The cellulose catalytic bio-oil had higher water content and its organic phase was easily separated from the aqueous phase before GC-MS analysis. Sugars were considerably reduced after catalysis. Because sugars are watersoluble, the remarkable drop in identified sugars could be attributed, in part, to their removal from the sample, along with the aqueous phase. Levoglucosan and 1,4:3,6Dianhydro-α-d-glucopyranose were still present in the organic phase, but their total area percentage in the chromatogram was considerably lower. Many new products were identified in the catalytic bio-oil, most of which were aromatic hydrocarbons, such as xylene, toluene, indane and indene and PAHs, such as naphthalene and naphthalene molecules with methyl substitutes.
10
Page 10 of 34
As shown in Fig. 6, the main products from the thermal pyrolysis of xylan are ketones and phenolic compounds. Acetic and propanoic acid, as well as a few aromatic hydrocarbons were also present. The ketones were mainly 1-hydroxy-2propanone (hydroxyacetone) and various cyclopentenones with methyl- and ethyl-
ip t
substitutes. The phenolic compounds were mainly simple phenols with methylsubstitutes. After catalysis with ZSM-5, the ketones in the organic phase were drastically reduced, while the phenols and the aromatic hydrocarbons were
cr
significantly increased. Formation of PAHs (napthalenes) was also observed. The
aromatic hydrocarbons that were formed after catalysis were toluene, xylene and other
us
substituted benzene ring compounds, as well as indanes and indenes. The phenolic compounds identified were mostly simple phenols with methyl- substitutes.
an
Hydroxyacetone and various cyclopentenones were still present in the bio-oil after catalysis but their peaks occupied a significantly smaller area of the chromatogram. The products derived from the thermal pyrolysis of lignin were almost exclusively
M
phenols, as shown in Fig. 7. Some heavy acids were also detected. The phenolic compounds derived from lignin were different from those derived from cellulose and xylan and were mostly benzenediols (catechols), phenols with methoxy- substitutes
d
and more complex phenolic compounds with higher molecular weights. An
te
interesting observation is that after catalysis, besides the drop in the acids content and the slight increase of PAHs, the composition of the bio-oil remained almost unaltered.
Ac ce p
The catalytic bio-oil from lignin was still, almost exclusively, composed of benzenediols and phenols with methoxy- substitutes. This confirms previous reports that the C-O bond between the aromatic ring and the hydroxyl group in the phenol molecule is refractory to ZSM-5 [13]. 3.2.2. Pyrolysis of mixtures
As was seen in the TG experiments with cellulose, xylan and lignin mixtures, the
behavior of each mixture during pyrolysis could be predicted quite well by assuming that the overall pyrolysis behavior is the weighted sum of the partial contributions of its components. In fact, the final solid residue at temperatures above 400 °C was predicted with very good accuracy (see section 3.1.2). In order to verify whether such predictions for the pyrolysis product yields and composition can be achieved under fast pyrolysis conditions, a set of thermal and catalytic experiments with ZSM-5 was performed with Mixture 1 and Mixture 2 in the bench scale biomass pyrolysis unit.
11
Page 11 of 34
The results were compared to calculated values that were based on the experimental results from the thermal and catalytic pyrolysis of the separate biomass components (see section 3.2.1). The experimental results and calculated values for the product yields of the
ip t
thermal and the catalytic pyrolysis of both mixtures are presented in Fig. 8 and Fig. 9 respectively. From the data presented, it is obvious that the calculations could not
accurately predict the product yields; differences between the experimental and
cr
calculated values were observed. In particular, gas and solid product yields were
higher than expected, while bio-oil and organic phase yields were lower. This trend
us
was observed both for Mix 1 and Mix 2 and both for thermal and catalytic pyrolysis. As seen in the TG experiments of the biomass component mixtures, the derivative
an
weight loss peaks of cellulose in the mixed biomass samples were wider, had lower intensity and were shifted towards higher temperatures, while the xylan peaks were not significantly affected (Fig. 2 and 3). This behavior was attributed to limited heat
M
and mass transfer due to the initial decomposition of xylan and lignin that melted and formed a solid residue around cellulose particles. This seems to be well supported by our fast pyrolysis experimental results. Poor heat and mass transfer in fast pyrolysis
te
gas yields.
d
conditions is known to result in higher biomass residue yields, lower liquid and higher The composition of the organic phase from the thermal and catalytic pyrolysis of
Ac ce p
biomass mixtures is presented in Fig. 10 and Fig. 11 respectively. It can be seen that there can be a significant difference between the experimental and calculated composition of the liquid product. In the thermal pyrolysis experiments (Fig. 10), the chromatogram area of sugars (mainly levoglucosan) was significantly smaller than the calculations, while ketones in Mix 1 and phenols in Mix 2 were more abundant. Closer examination of the GC-MS results revealed that the increased phenolic compounds in Mix 2 were phenols with oxygenated substitutes, which are characteristic products of the decomposition of lignin. In the catalytic experiments, the most striking difference between the experimental and the calculated results is observed in the total chromatogram area % of phenols, which in the experimental runs was significantly higher than expected for both Mix 1 and Mix 2. As evident from the significant differences between the experimental and the calculated results, the prediction of the chemical composition of the bio-oil was even more challenging than the prediction of the pyrolysis yields. Due to its complexity,
12
Page 12 of 34
bio-oil is difficult to analyze by conventional GC-MS. Compounds that can be identified in one sample may not be identified in another, more complex sample because of convoluted peaks. In addition, the analysis of the organic phase was semiquantitative and the area covered by a peak did not necessarily correspond to a mass
ip t
fraction. For these reasons, it would not be safe to elaborate on these results. In a similar study, Wang et al. [6] also observed significant differences between the experimental and calculated values of the peak area % in their chromatograms.
cr
3.3. Reaction pathways
us
Based on our results from the analyses of the organic phase of the fast pyrolysis bio-oil of cellulose, xylan and lignin, an elucidation of the reaction pathways during thermal pyrolysis was attempted and is described in this section. In the catalytic
an
experiments, the biomass particles did not come in contact with the catalyst particles, therefore the pyrolysis vapors that reached the catalyst bed were assumed to be the
M
same as in the non-catalytic experiments. When the vapors passed through the catalyst, the compounds of the thermal pyrolysis were catalytically converted to other products. Therefore, by comparing the product distribution from the thermal and
te
possible.
d
catalytic experiments, an elucidation of the catalytic reaction pathways was also
Ac ce p
3.3.1. Pyrolysis pathways of cellulose
The main product from the thermal pyrolysis of cellulose in our study was
levoglucosan. At fast pyrolysis conditions, pure cellulose is known to depolymerize and form levoglucosan via a transglycosylation mechanism [14-16]. Other anhydrosugars, such as levoglucosenone, 1,4:3,6-Dianhydro-α-D-glucopyranose and Ethyl α-D-glucopyranoside, were also identified in the organic phase. Such anhydrosugars can be produced either by the primary pyrolysis of cellulose, or by the secondary pyrolysis of levoglucosan, as showed by Kawamoto et al. [17]. Kawamoto et al. [17] also showed that levoglucosan, through ring-opening polymerization, can form polysaccharides, which eventually form carbonized products (solid residue) and Lin et al. [18] proposed that the majority of the cellulose solid residue is formed through this mechanism and not from the cellulose itself. Apart from anhydrosugars, small peaks of phenolic compounds were identified as well. Other researchers [16, 19, 20] also detected phenolic compounds in the cellulose pyrolysis products. Evans and
13
Page 13 of 34
Milne [16] showed that these aromatic products are not derived from the primary pyrolysis of the cellulose, but are formed from secondary reactions (probably from the gas phase polymerization of unsaturated species such as propylene, butadiene and butene). The acetic acid, which was also detected in our experimental runs, is
ip t
considered to derive from hydroxyacetaldehyde by dehydration to form ketene and subsequent hydration to acetic acid [14, 15]. Due to its early elution, hydroxyacetaldehyde was not detected in our analyses, but is widely reported as a
cr
cellulose pyrolysis product in the literature [14, 15, 18, 21, 22]. Acetic acid is also formed from the thermal decomposition of 5-hydroxymethylfurfural (5-HMF) [23],
us
which is formed either via the direct ring-opening and rearrangement reactions of cellulose unit molecules or via the secondary reactions of levoglucosan [15, 17].
an
Finally, our analyses also showed the presence of cyclic ketones in the thermal bio-oil from cellulose, which were also reported by Wang et al. [23]. More peaks were present in our chromatogram but could not be identified by the GC-MS system.
M
In the catalytic experiments, the levoglucosan content in the organic phase of the bio-oil from cellulose was significantly reduced, while other compounds were formed. The most notable compounds observed after catalysis were aromatic hydrocarbons,
d
phenols, furans and PAHs. The aromatic hydrocarbons were benzene compounds with
te
linear substitutes, as well as indanes and indenes. PAHs were naphthalene compounds with methyl- substitutes. Phenolic compounds were also detected before catalysis, but
Ac ce p
their relative abundance (area %) increased significantly after catalysis. Both before and after catalysis, phenolic compounds were phenols with methyl- substitutes. Finally, we also detected furan compounds that were mostly benzofurans and furans with methyl- substitutes.
Horne and Williams [13] showed that benzenes and napthalenes can be found in
the catalytic oil product from cyclic ketones (cyclopentanone), as well as from furfural. They also showed that benzofurans can form from the catalytic transformation of furfural and indanes/indenes from the catalytic transformation of cyclopentanone. Carlson et al. [24] reported the possible contribution of furan compounds (furfural, 5-HMF and furfuryl alcohol) to the formation of hydrocarbons as well. Grandmaison et al. [25] also reported the formation of aromatic hydrocarbons, as well as benzofurans during the conversion of furanic compounds (furfural and furan) over ZSM-5. Simple aromatic hydrocarbons and phenols may also form in the gas phase by secondary polymerization of unsaturated light compounds,
14
Page 14 of 34
as suggested by Evans and Milne [16], which are expected to be more abundant in the catalytic vapors due to the cracking effect of the ZSM-5 catalyst. Also, part of the phenols that ended up in the catalytic pyrolysis oil, were derived from the thermal pyrolysis of cellulose due to the fact that the C-O bond between the benzene ring and form thermal coke, or pass unconverted through the catalyst [26].
cr
3.3.2. Pyrolysis pathways of xylan
ip t
the methoxy group is refractory to ZSM-5 [13] and the phenolic molecules either
The majority of the products observed from the thermal pyrolysis of xylan were
us
phenols and cyclic ketones. Phenols are likely derived from the cleavage of the ferulic acid ester branch of xylan, as well as from the gas phase polymerization of unsaturated light species [16]. The cyclic ketones are likely derived from the xylan
an
main stem by cleavage of the o-glucosidic bonds and subsequent removal of the hydroxyl groups of the xylose rings. Another major product from the thermal
M
pyrolysis of xylan was hydroxyacetone. Propanoic and acetic acid were also observed. Acetic acid is formed by cleavage of the acetyl group of the xylan structure with simultaneous release of CO2 [27].
d
After catalysis, aromatic hydrocarbons, phenolic compounds and PAHs were
te
considerably increased, while acids and especially ketones were reduced. Aromatic hydrocarbons were mostly benzenes with methyl-/ethyl- substitutes, as well as
Ac ce p
indenes/indanes and PAHs were napthalenic compounds. As in thermal pyrolysis, phenolic compounds were simple phenols with methyl-/ethyl- substitutes. This product distribution in the catalytic bio-oil was expected, since the compounds formed from the thermal pyrolysis were mainly linear ketones, cyclic ketones and phenols. Horne and Williams [13] showed that cyclic ketones (cyclopentanone) form simple aromatic compounds when passed over a ZSM-5 catalyst and are also responsible for the formation of indanes/indenes and napthalenes. Hydroxyacetone, which was a major product in the thermal pyrolysis vapors from xylan, tentatively followed a reaction pathway proposed by Gayubo et al. [28] and transformed into olefins, paraffins and eventually, aromatic hydrocarbons. As mentioned previously, the C-O bond in the phenolic compounds is refractory to ZSM-5 and the phenols from the thermal pyrolysis vapors either formed thermal coke or passed unconverted through the catalyst. Additional phenols, as well as simple aromatic hydrocarbons, can form in the gas phase by polymerization of light unsaturated species [16], which are expected
15
Page 15 of 34
to be more abundant during catalytic pyrolysis due to the cracking properties of the ZSM-5 catalyst. Finally, Adjaye and Bakhshi [29] suggested that carboxylic acids are easily converted over ZSM-5 and form mostly coke, gases and water, while a small fraction is converted to aromatic hydrocarbons.
ip t
3.3.3. Pyrolysis pathways of lignin The identified compounds in the thermal pyrolysis oil from lignin were almost
cr
entirely phenolic compounds, which were more complicated than the phenols from cellulose and xylan, as they were mostly phenols with methoxy- substitutes,
us
benzenediols and polysubstituted phenols. The complex structures of these phenolic compounds indicated that they were derived directly from the lignin matrix. Simpler chromatogram was significantly smaller.
an
phenols with methyl- substitutes were also observed but their total area in the Kawamoto et al. and Asmadi et al. recently investigated the thermal pyrolysis of
M
lignin using representative model dimers [30], guaiacol and syringol [31] and other model compounds [32]. Overall, their studies showed that the thermal decomposition of lignin begins by cleavage of the weak α-ether and β-ether bonds of the lignin
d
structure, which releases guaiacyl- and syringyl-type aromatics, depending on the
te
lignin source. Softwood lignins have only guaiacyl-type nuclei, while hardwood lignins have both [30]. These aromatics further reacted, either by homolysis of the O-
Ac ce p
CH3 bonds to produce catechols and pyrogallols, or by radical induced rearrangement (ipso-substitution) reactions to produce cresols and xylenols [31]. Further decomposition of catechols and pyrogallols yields gases, coke and few PAHs, while decomposition of cresols and xylenols yields phenol and cresols by demethylation [32].
Since the C-O bond in the phenol molecule is refractory to ZSM-5, the presence
of the catalyst was not expected to have any significant impact on the composition of the catalytic bio-oil. Although there was a slight increase in the yields of gases, solid products (coke) and water, an indication of the catalytic effect of the ZSM-5, the composition of the bio-oil was not substantially altered. Methoxy- substituted phenols, benzenediols and poly-substituted phenols were still the prevalent compounds. The increase in gas and water yields was most likely due to cleavage of the side groups of the phenolic units as they passed through the catalyst. The aromatic ring and the C-O bond were not affected.
16
Page 16 of 34
4. Conclusions Thermal and catalytic pyrolysis of biomass components (cellulose, xylan and lignin) and mixtures thereof was performed in both a TG apparatus and a bench-scale pyrolysis reactor. The pyrolysis of cellulose gave high yields of bio-oil rich in
ip t
levoglucosan and other anhydrosugars with minimal coke formation. Xylan gave high
gas yields and moderate yields of bio-oil, rich in water, phenols and ketones. Lignin
cr
gave the highest solid residue yield and produced a bio-oil that was rich in phenolic
compounds. The pyrolysis of biomass component mixtures was studied in order to
us
investigate whether the calculation of the product distribution and quality from the pyrolysis of a biomass with known composition is possible. The calculation of the final biomass residue yield at 500 °C was achieved with very good accuracy in the TG
an
experiments, but differences between the calculated and experimental DTG curves were observed, which were attributed to limited heat and mass transfer in the biomass
M
mixtures. The effects of the limited heat and mass transfer were more evident in the pyrolysis experiments, where higher than expected solid residue and gas yields were obtained, at the expense of liquid products. The composition of the bio-oil produced
d
from the biomass mixtures was also affected. Even though characteristic chemical
te
groups of each biomass component were present in the bio-oil, their relative abundance was different than what would be expected.
Ac ce p
An elucidation of the reaction pathways before and after catalysis was also attempted, based on the observation of the products from the pyrolysis of the individual components and reports from the literature. Cellulose formed mainly levoglucosan via transglycosylation, which was easily converted catalytically to other products, mainly aromatic hydrocarbons, phenols, furans and PAHs. Xylan formed a wider array of products, mainly phenols and ketones as well as some acids. Acids and ketones were reduced after catalysis and formed aromatic hydrocarbons, phenols and PAHs. Due to its aromatic structure, lignin formed mainly phenolic products that could not be catalytically converted to other chemical groups.
Acknowledgements
This research has been co-financed by European Union (European Social Fund) and Greek national funds through Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF)-
17
Page 17 of 34
Research Funding Program: THALES-Investing in knowledge society through the European Social Fund (MIS 380405). References
ip t
[1]. P. McKendry, Energy production from biomass (part 1): overview of biomass, Bioresource Technol. 83 (2002) 37–46. and biomass components, Fuel 75 (1996) 987-998.
cr
[2]. K. Raveendran, A. Ganesh, K.C. Khilar, Pyrolysis characteristics of biomass
[3]. H. Yang, R. Yan, H. Chen, C. Zheng, D. Lee, In-depth investigation of biomass
us
pyrolysis based on three major components: hemicellulose, cellulose and lignin. Energ. Fuel. 20 (2006) 388-393.
an
[4]. N. Worasuwannarak, T. Sonobe, W. Tanthapanichakoon, Pyrolysis behaviors of rice straw, rice husk, and corncob by TG-MS technique, J. Anal. Appl. Pyrol. 78 (2007) 265–271.
M
[5]. G. Wang, W. Li, B. Li, H. Chen, TG study on pyrolysis of biomass and its three components under syngas, Fuel 87 (2008) 552–558.
d
[6]. S. Wang, X. Guo, K. Wang, Z. Luo, Influence of the interaction of components on the pyrolysis behavior of biomass, J. Anal. Appl. Pyrol. 91 (2011) 183–189.
te
[7]. S.D. Stefanidis, K.G. Kalogiannis, E.F. Iliopoulou, A.A. Lappas, P.A. Pilavachi, In-situ upgrading of biomass pyrolysis vapors: Catalyst screening on a fixed bed
Ac ce p
reactor, Bioresource Technol. 102 (2011) 8261–8267.
[8]. H. Yang, R. Yan, H. Chen, D. Lee, C. Zheng. Characteristics of hemicellulose, cellulose and lignin pyrolysis, Fuel 86 (2007) 1781-1788.
[9]. P.R. Patwardhan, R.C. Brown, B.H. Shanks, Product distribution from the fast pyrolysis of hemicellulose. ChemSusChem 4 (2011) 636–643.
[10]. T. Hosoya, H. Kawamoto, S. Saka, Cellulose–hemicellulose and cellulose– lignin interactions in wood pyrolysis at gasification temperature, J. Anal. Appl. Pyrol. 80 (2007) 118–125. [11]. C. Couhert, J-M. Commandre, S. Salvador, Is it possible to predict gas yields of any biomass after rapid pyrolysis at high temperature from its composition in cellulose, hemicellulose and lignin?, Fuel 88 (2009) 408–417.
18
Page 18 of 34
[12]. R. Fahmi, A.V. Bridgwater, L.I. Darvell, J.M. Jones, N. Yates, S. Thain, I.S. Donnison, The effect of alkali metals on combustion and pyrolysis of Lolium and Festuca grasses, switchgrass and willow, Fuel 86 (2007) 1560–1569 [13]. P. Horne, P. Williams, Reaction of oxygenated biomass pyrolysis model
ip t
compounds over a ZSM-5 catalyst, Renew. Energ. 7 (1996) 131–144. [14]. J. Piskorz, D. Radlein, D.S. Scott, On the mechanism of the rapid pyrolysis of cellulose, J. Anal. Appl. Pyrol. 9 (1986) 121–137.
cr
[15]. D. Shen, S. Gu, The mechanism for thermal decomposition of cellulose and its main products, Bioresource Technol. 100 (2009) 6496–6504.
us
[16]. R.J. Evans, T.A. Milne, Molecular characterization of the pyrolysis of biomass, Energ. Fuel. 1 (1987) 123–137.
an
[17]. H. Kawamoto, M. Murayama, Pyrolysis behavior of levoglucosan as an intermediate in cellulose pyrolysis: polymerization into polysaccharide as a key reaction to carbonized product formation, J. Wood Sci. 49 (2003) 469-473.
M
[18]. Y-C. Lin, J. Cho, G.A. Tompsett, P.R. Westmoreland, G.W. Huber, Kinetics and mechanism of cellulose pyrolysis. J. Phy. Chem. C 113 (2009) 20097– 20107.
d
[19]. Y.F. Liao, Mechanism study of cellulose pyrolysis, PhD Thesis, ZheJiang
te
University, HangZhou, China, 2003.
[20]. R. Alen, E. Kuoppala, P. Oesch, Formation of the main degradation compound
Ac ce p
groups from wood and its components during pyrolysis, J. Anal. Appl. Pyrol. 36 (1996) 137–148.
[21]. Z.Y. Luo, S.R. Wang, Y.F. Liao, K.F. Cen, Mechanism study of cellulose rapid pyrolysis, Ind. Eng. Chem. Res. 43 (2004) 5605–5610.
[22]. J. Banyasz, S. Li, J. Lyons-Hart, K. Shafer, Gas evolution and the mechanism of cellulose pyrolysis, Fuel 80 (2001) 1757–1763.
[23]. S. Wang, X. Guo, T. Liang, Y. Zhou, Z. Luo, Mechanism research on cellulose pyrolysis by Py-GC/MS and subsequent density functional theory studies, Bioresource Technol. 104 (2011) 722-728. [24]. T.R. Carlson, J. Jae, Y-C. Lin, G.A. Tompsett, G.W. Huber, Catalytic fast pyrolysis of glucose with HZSM-5: The combined homogeneous and heterogeneous reactions, J. Catal. 270 (2010) 110–124. [25]. J.L. Grandmaison, P.D. Chantal, S.C. Kaliaguine, Conversion of furanic compounds over H-ZSM-5 zeolite, Fuel 69 (1990) 1058–1061.
19
Page 19 of 34
[26]. P. Chantal, S. Kaliaguine, J. Grandmaison, Reactions of phenolic compounds over HZSM-5, Appl. Catal. 18 (1985) 133–145. [27]. D.K. Shen, S. Gu, A.V. Bridgwater, Study on the pyrolytic behaviour of xylanbased hemicellulose using TG-FTIR and Py-GC-FTIR, J. Anal. Appl. Pyrol. 87
ip t
(2010) 199–206. [28]. A. Gayubo, A. Aguayo, A. Atutxa, Transformation of oxygenate components of
biomass pyrolysis oil on a HZSM-5 zeolite. II. Aldehydes, ketones, and acids,
cr
Ind. Eng. Chem. Res. 43 (2004) 2619-2626.
[29]. J. Adjaye, N. Bakhshi, Catalytic conversion of a biomass-derived oil to fuels
us
and chemicals. 1. Model-compound studies and reaction pathways, Biomass Bioenerg. 8 (1995) 131–149.
an
[30]. H. Kawamoto, S. Horigoshi, S. Saka, Pyrolysis reactions of various lignin model dimers, J. Wood Sci. 53 (2006) 168–174.
[31]. M. Asmadi, H. Kawamoto, S. Saka, Thermal reactions of guaiacol and syringol [32]. M.
Asmadi,
H.
M
as lignin model aromatic nuclei, J. Anal. Appl. Pyrol. 92 (2011) 88–98. Kawamoto,
S.
Saka,
Thermal
reactivities
of
catechols/pyrogallols and cresols/xylenols as lignin pyrolysis intermediates, J.
Ac ce p
te
d
Anal. Appl. Pyrol. 92 (2011) 76–87.
20
Page 20 of 34
Figure Captions Fig. 1. Weight loss and derivative weight loss curves for cellulose, xylan and lignin Fig. 2. Experimental and calculated TG and DTG curves for Mix 1 Fig. 3. Experimental and calculated TG and DTG curves for Mix 2
ip t
samples
cr
Fig. 4. Product yield distribution from the thermal and catalytic pyrolysis of cellulose, xylan and lignin
us
Fig. 5. Chemical composition of the organic phase of thermal and catalytic bio-oil from cellulose
Fig. 6. Chemical composition of the organic phase of thermal and catalytic bio-oil
an
from xylan
Fig. 7. Chemical composition of the organic phase of thermal and catalytic bio-oil
M
from lignin
Fig. 8. Comparison between experimental (exp.) and calculated (calc.) product yield values of the thermal pyrolysis of biomass mixtures
d
Fig. 9. Comparison between experimental (exp.) and calculated (calc.) product yield
te
values of the catalytic pyrolysis of biomass mixtures Fig. 10. Experimental (exp.) and predicted (calc.) composition of the organic phase
Ac ce p
from the thermal pyrolysis of biomass mixtures Fig. 11. Experimental (exp.) and predicted (calc.) composition of the organic phase from the catalytic pyrolysis of biomass mixtures
21
Page 21 of 34
Table 1. Composition of the biomass feedstocks used in this study (on dry basis).
C (wt.%) H (wt.%) O (wt.%)1 Ash (wt.%) 42.27 6.40 51.33 0.00 42.07 5.82 45.75 6.36 63.41 5.89 28.42 2.28 1 By difference (oxygen% = 100% - carbon% - hydrogen% - ash%)
Ac ce p
te
d
M
an
us
cr
ip t
Sample Cellulose Xylan Lignin
22
Page 22 of 34
Highlights Pyrolysis of biomass component mixtures in both TGA and pyrolysis reactor setups Comparison between experimental and calculated product yields and composition
ip t
Heat transfer lag was evident during the pyrolysis of biomass component mixtures Elucidation of thermal and catalytic reaction pathways of each biomass
Ac ce p
te
d
M
an
us
cr
component
23
Page 23 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1
Page 24 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 2
Page 25 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3
Page 26 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4
Page 27 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 5
Page 28 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 6
Page 29 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 7
Page 30 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 8
Page 31 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 9
Page 32 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 10
Page 33 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 11
Page 34 of 34
S0165-2370(13)00237-4 http://dx.doi.org/doi:10.1016/j.jaap.2013.10.013 JAAP 3076
To appear in:
J. Anal. Appl. Pyrolysis
Received date: Revised date: Accepted date:
13-9-2013 23-10-2013 28-10-2013
Please cite this article as: S.D. Stefanidis, K.G. Kalogiannis, E.F. Iliopoulou, C.M. Michailof, P.A. Pilavachi, A.A. Lappas, A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin, Journal of Analytical and Applied Pyrolysis (2013), http://dx.doi.org/10.1016/j.jaap.2013.10.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A study of lignocellulosic biomass pyrolysis via the pyrolysis of cellulose, hemicellulose and lignin Stylianos D. Stefanidisa, b, *, Konstantinos G. Kalogiannisa, Eleni F. Iliopouloua, Chrysoula M. Michailofa, Petros A. Pilavachib, Angelos A. Lappasa Chemical Process and Energy Resources Institute, Center for Research and Technology Hellas, 57001 Thessaloniki, Greece b
ip t
a
Department of Mechanical Engineering, University of Western Macedonia, 50100 Kozani, Greece
cr
* Corresponding author. Tel: +30 2310 498369, Fax: +30 2310 498380, e-mail: [email protected]
us
Abstract
In this study, thermogravimetric (TG) analyses, along with thermal and catalytic
an
fast pyrolysis experiments of cellulose, hemicellulose, lignin and their mixtures were carried out in order to investigate their pyrolysis products and whether the prediction
M
of the pyrolysis behavior of a certain lignocellulosic biomass feedstock is possible, when its content in these three constituents is known. We were able to accurately predict the final solid residue of mixed component samples in the TG analyses but the
d
differential thermogravimetric (DTG) curves indicated limited heat transfer when
te
more than one component was present in the pyrolyzed sample. The limited heat transfer did not have a significant effect on the TG curves but it affected the product
Ac ce p
distribution in the fast pyrolysis experiments, which resulted in inaccurate calculation of the product yields, when using a simple additive law. In addition, the pyrolysis products of each biomass constituent were characterized in order to study their contribution to the yield and composition of the products from whole biomass pyrolysis. An investigation into the pyrolysis reaction pathways of each component was also carried out, using the bio-oil characterization data from this study and data found in the literature.
Keywords: Biomass; Pyrolysis; Depolymerization; Biomass constituents; Reaction pathways 1. Introduction Increased world energy demand and environmental concerns have driven research on alternative energy sources. Lignocellulosic biomass is considered a promising
1
Page 1 of 34
alternative and renewable energy source that can be converted via the biomass fast pyrolysis process into a liquid product, known as bio-oil, which is considered to be a promising biofuel/bioenergy carrier. The bio-oil is a complex mixture of oxygenated compounds and its composition and quality is heavily dependent on the composition
ip t
of the biomass feedstock. Lignocellulosic biomass is composed mainly of three basic structural components; cellulose, hemicellulose and lignin. The content of these components in biomass varies depending on the biomass type. Woody plant species
cr
have tightly bound fibers and are richer in lignin while herbaceous plants have more
loosely bound fibers, a fact that indicates lower lignin content. Usually, cellulose,
us
hemicellulose and lignin constitute 40-50 wt.%, 20-40 wt.% and 10-40 wt.% of the plant material respectively [1]. In addition to these components, lignocellulosic
an
biomass also contains a small amount of inorganic material (ash) and extractives. It has been suggested that since all lignocellulosic biomass is mainly composed of these basic structural components, the aggregative behavior during pyrolysis of each
M
of those independent components would describe the behavior of any lignocellulosic feed [2]. Therefore, it should be possible to predict the yield and composition of the pyrolysis products of any feedstock, when its composition is known. In order for that
d
to be true however, the pyrolysis behavior of each individual component must be
te
independent of the presence of other components. Otherwise, if synergistic effects were to take place, the prediction of the behavior of a biomass feedstock would be a
Ac ce p
more complex task. In addition, since biomass pyrolysis proceeds through an extremely complex network of reaction mechanisms, the study of the biomass pyrolysis chemistry can be simplified by studying the pyrolysis reactions of cellulose, hemicellulose and lignin individually. Several groups in the literature have studied the pyrolysis of biomass on the basis
of these three main components. Raveendran et al. [2] studied the pyrolysis characteristics of biomass components in a thermogravimetric analyzer and a packedbed pyrolyzer and found no detectable interactions among the components during pyrolysis in either experimental setup. Yang et al. [3] also observed negligible interactions among the three biomass components in their study, when using a thermogravimetric analyzer. On the other hand, Worasuwannarak et al. [4] studied the pyrolysis behavior of cellulose, xylan, lignin and mixtures by TG-MS technique and observed significant interactions between cellulose and lignin that caused a suppression of liquid product formation and an increase in solid residue yield. Wang
2
Page 2 of 34
et al. [5] also observed cellulose-lignin interactions, as well as hemicellulose-lignin interactions, while they reported that hemicellulose and lignin did not seem to affect each other during pyrolysis in a thermogravimetric analyzer. More recently, Wang et al. [6] studied the interactions of the biomass components in both a TG-FTIR and an
ip t
experimental pyrolyzer. They reported no significant differences between the experimental and calculated TG/DTG curves, when using biomass component
mixtures, but differences in the evolution curves of the main products (levoglucosan,
cr
2-furfural, acetic acid and 2,6-dimethoxy phenol) were apparent. In addition, the
mixed samples exhibited a common tendency to form less liquid and more gas
us
products than what the calculations predicted. To the best of our knowledge, the pyrolysis of mixed biomass components with catalytic upgrading of the pyrolysis
an
vapors has not been studied yet.
In this study, the thermal and catalytic pyrolysis of biomass components (cellulose, xylan and lignin) was studied, both in a TG apparatus and in a pyrolysis
M
reactor and the contribution of each component on the yield and the composition of the biomass pyrolysis products was investigated. Moreover, the pyrolysis of mixtures of cellulose, hemicellulose and lignin was studied in order to investigate the
d
contradictory literature results concerning the interactions between cellulose,
te
hemicellulose and lignin during pyrolysis. Finally, based on the results from the pyrolysis of the individual components and reports from the literature, an elucidation
Ac ce p
of the reaction pathways in the absence or presence of a catalyst was attempted. 2. EXPERIMENTAL
2.1. Biomass Feedstock properties The cellulose used in this study was purchased from JRS Pharma with the
commercial name VIVAPUR Type 200 Microcrystalline Cellulose. The lignin used was a kraft lignin isolated from a commercial pulp mill using predominantly Norway spruce as a raw material and was purchased from Sigma-Aldrich (Aldrich 370959). Because hemicellulose was not commercially available, xylan purchased from SigmaAldrich was used instead (Sigma X4252). Xylan is considered to be an adequate replacement for hemicellulose and it is widely used in the literature for this purpose [3]. All biomass samples were dried at 105 °C for 4 h before use and therefore, all
3
Page 3 of 34
results reported here are on a dry basis. The elemental composition of the dry biomass samples is presented in Table 1. 2.2. TG experiments The TG experiments were performed in a TA Instruments SDT2960
ip t
thermogravimetric analyzer to study the decomposition curves upon pyrolysis of the
samples. Experiments were carried out with sample masses of about 10 mg using a
cr
linear heating rate of 10 °C/min within the temperature range of 28 °C to 840 °C and a steady nitrogen flow rate of 100 cm3/min.
us
2.3. Pyrolysis experiments
The pyrolysis experiments were performed at 500 ºC in a bench-scale fixed bed
an
biomass pyrolysis reactor. The reactor was made of stainless steel 316 and measured 14 mm in internal diameter and 41 cm in height. Heat was provided by a three-zone
M
furnace and the temperature of each zone was independently controlled using temperature controllers. The catalyst bed temperature was considered as the pyrolysis temperature and was monitored with a thermowell. The biomass sample (1.5 g) was
d
loaded in a specially designed piston mounted on top of the reactor, outside of the
te
furnace. After the desired pyrolysis temperature in the rector was reached, the piston was used to introduce the biomass feedstock into the hot reactor. A constant stream of
Ac ce p
nitrogen (100 cm3/min) was fed from the top of the reactor for the continuous withdrawal of the pyrolysis vapors and maintenance of the inert atmosphere during the experiment. The vapors were carried through the catalyst bed (0.7 g), exited from the bottom of the reactor in gaseous form and were condensed in a glass receiver submerged in a cooling bath that was kept at -17 °C. Non-condensable gases were collected in a gas collection system, while a glass wool filter placed between the glass receiver and the gas collection system recovered any light condensable liquids that were not condensed in the receiver. The above described catalytic pyrolysis experiments can be referred to as in situ upgrading of pyrolysis vapors and are of the “ex-bed” type (i.e. there was no mixing of solid biomass with the solid catalyst). In addition, all the experimental parameters (i.e., fast heating of biomass, low residence time, fast cooling of products) resemble those of the biomass fast pyrolysis type of experiments. A schematic representation of the experimental setup can be found elsewhere [7].
4
Page 4 of 34
Both thermal and catalytic pyrolysis experiments were performed. Inert silica sand was used in place of the catalyst bed in the thermal pyrolysis experiments, while a commercial ZSM-5 formulation diluted with silica alumina (30 wt.% crystalline zeolite, 138 m2/g surface area, 4 nm average pore diameter, 0.037 cm3/g
ip t
microporosity and 0.018 cm3/g mesoporosity) was used for the catalytic pyrolysis experiments.
The liquid products were collected and quantitatively measured in the pre-
cr
weighted glass receiver. The pyrolytic vapors, upon their condensation in the glass receiver, formed multiple phases; an aqueous phase, a liquid organic phase and
us
viscous organic deposits on the receiver walls. Extensive effort has been put in the development of a method for the collection of a representative bio-oil sample for
an
analysis [7]. Towards this goal, the bio-oil was first fully homogenized inside the receiver using ethyl lactate as the solvent and then collected as a solution, which was then submitted for analysis. Ethyl lactate was chosen for its non-volatility, which
M
minimizes errors during weighing. It also proved to be a good solvent for all the biooil samples in this study.
d
2.4. Analysis methods
te
The elemental composition of the biomass samples and the bio-oil was determined with an elemental CHN LECO-800 analyzer. The water content of the
Ac ce p
bio-oil was measured with Karl-Fischer titration (ASTM E203-08). The water/aqueous phase present in the bio-oil was then separated from the organic phase using an organic solvent (dichloromethane) and the organic phase of the bio-oil was analyzed by GC-MS using an Agilent 7890A/5975C gas chromatograph-mass spectrometer system (Electron energy 70 eV, Emission 300 V, Helium flow rate: 0.7 cc/min, Column: HP-5MS 30 m x 0.25 mm ID x 0.25 μm). Internal libraries were used for the identification of the compounds found in the bio-oil and their categorization into main functional groups. The gaseous products were collected and measured by the water displacement method. The gaseous products were analyzed in a HP 5890 Series II gas chromatograph, equipped with four columns (Precolumn: OV-101, Columns: Porapak N, Molecular Sieve 5A and Rt-Qplot 30 m x 0.53 mm ID) and two detectors (TCD and FID). The amount of the solid residue left in the reactor and deposited on the catalyst surface, consisted mainly of charcoal and coke-
5
Page 5 of 34
on-catalyst formed by thermal and/or catalytic cracking and was determined by direct weighting. 3. RESULTS AND DISCUSSION
ip t
3.1. TG analyses 3.1.1. TG analyses of separate components
cr
The weight loss and derivative weight loss curves of the separate biomass samples are given in Fig. 1. As it can be seen, cellulose decomposed over a narrow
us
temperature range, between 280 °C and 360 °C with the highest decomposition rate being observed at 339 °C and the total solid residue at 500 °C and 800 °C being 10.7
an
and 7.4 wt.% respectively. This very well-defined decomposition temperature range is due to cellulose’s very homogeneous unbranched crystalline structure of linked Dglucose units. Xylan on the other hand, which is also a polysaccharide, decomposed at
M
lower temperatures, over a slightly wider temperature range, between 200 °C and 320 °C, with two distinct peaks; one at 246 °C and one at 295 °C. The lower
d
decomposition temperature range of xylan is attributed to its structure, which is amorphous with many branched units that have low activation energy [8]. The xylan
te
residue at 500 °C and 800 °C was significantly higher than that of cellulose, about 30 and 25 wt.% respectively. This can be attributed, in part, to the high ash content of the
Ac ce p
commercial xylan sample that was used in this study (Table 1). However, Patwardhan et al. [9] also reported higher solid residue yields from xylan pyrolysis, despite using an ash-free sample. They suggested that, due to the different sugars that make up xylan (pentoses and hexoses), pyrolysis proceeds via a different mechanism from that of cellulose, which may account for the higher solid residue formation. Lignin decomposed over a very wide temperature range, from 140 °C to 600 °C,
with a very low-intensity peak around 380 °C. The lignin residue at 500 °C was the highest observed, about 53.4 wt.% and even at 800 °C the residue was 41.2 wt.%. This high residue yield is attributed to lignin’s structure, which consists of a complex network of cross-linked aromatic molecules that are difficult to decompose and therefore have high thermal stability [3].
6
Page 6 of 34
3.1.2. TG analyses of mixtures In order to examine any synergistic effects between the three different biomass components during pyrolysis, two mixed samples were prepared by simple mechanical mixing. For simplicity, one sample contained ½ cellulose and ½ xylan
ip t
(Mix 1), while the other sample also contained lignin and consisted of ⅓ cellulose, ⅓
xylan and ⅓ lignin (Mix 2). TG analysis was performed for both mixed samples at the
cr
same conditions as the separate biomass components and the experimental TG and DTG curves were compared to calculated curves. The calculated curves were
us
determined assuming that the overall pyrolysis behavior of a mixture is the weighted
,
,
and
M
where
an
sum of the partial contributions of its components [2], according to the formulas: (1) (2)
are the weight losses at
temperature T for the mixture, cellulose, xylan and lignin respectively, while Xcel, Xxyl
,
,
and
are the derivative weight
te
Accordingly,
d
and Xlig are the fractions of cellulose, xylan and lignin that are present in the mixture.
losses at temperature T for the mixture, cellulose, xylan and lignin respectively. The
Ac ce p
experimental and calculated TG and DTG curves for Mix 1 are presented in Fig. 2, while the corresponding results for Mix 2 are presented in Fig 3. For Mix 1, which was composed from equal parts of cellulose and hemicellulose,
the experimental and calculated TG curves agreed quite well, but there were obvious differences in the DTG curves. The xylan peaks, that were calculated at 247 °C and 302 °C, turned out as expected, with only a slight drop in intensity, but the cellulose peak had significantly lower intensity than expected and shifted to a slightly higher temperature (349 °C instead of 339 °C). In addition, the cellulose peak was wider than expected. The biomass residue at temperatures greater than 400 °C was predicted with very good accuracy and was found to be about 20 wt.% at 500 °C. In Mix 2 (Fig. 3), the two TG curves agreed quite well too, but the differences between the experimental and the calculated DTG curves were again obvious and more significant than in the case of Mix 1. The drop in the intensity of the cellulose
7
Page 7 of 34
peak was more pronounced in Mix 2, compared to Mix 1. The shift in the temperature of the cellulose peak was also more pronounced, as it was expected at 339 °C but it appeared at 353 °C. In addition, the experimental cellulose peak was wider than the calculated. No visible lignin peak was observed in the experimental DTG curve of
ip t
Mix 2, due to its low intensity and the overlapping cellulose peak. The biomass residue at temperatures greater than 400 °C was predicted with very good accuracy and it was found to be about 31% at 500 °C.
cr
Overall, the experimental and calculated TG curves of both mixtures agreed quite
well and were able to predict the thermal behavior of the mixed biomass samples with
us
accuracy. The DTG curves however, were more suitable to demonstrate the subtle but important differences in the thermal behavior of the mixed samples. While the
an
thermal decomposition of xylan did not seem to be significantly affected by the presence of the other components, especially in Mix 1, the cellulose peaks indicated a change in the thermal behavior of cellulose. More specifically, in both mixed samples,
M
the maximum weight loss rate of cellulose decreased significantly and shifted to a higher temperature. Moreover, the temperature range of the cellulose decomposition was wider in the mixed samples. These observations indicate a hindrance in the heat
d
transfer from the heated plate of the TG instrument towards the cellulose particles,
te
which appeared to decompose at higher temperatures and over a wider range due to the possible difference between the temperature that was recorded by the TG
Ac ce p
instrument and the actual temperature of the mixed biomass sample (that was in fact lower). The hemicellulose particles, which decomposed at lower temperatures, formed a charred film that wrapped around the cellulose particles and acted as a heat barrier. Hosoya et al. [10] and Wang et al. [6] made similar observations about the charred hemicellulose forming around cellulose particles and suggested that it could limit mass transfer during the volatilization of the cellulose particles. Our results indicate limited heat transfer, in addition to possible limited mass transfer that was suggested by the other research groups [6, 10]. From our experience, a similar phenomenon occurs with lignin, which starts to melt and form agglomerates at temperatures a little over 100 °C. Such alterations in the heat and mass transfer rates were not of significant importance in the experimental conditions of the TG experiments, hence the non-obvious differences between the experimental and the calculated TG curves. During the fast pyrolysis process however, rapid heat transfer rates and quick removal
8
Page 8 of 34
of the pyrolysis vapors are crucial parameters and alteration of these rates would result in different product yield distributions. Another interesting observation was that the presence of ash in the mixed samples did not seem to have an effect on the pyrolysis behavior of cellulose. The effect of ash
ip t
in the pyrolysis of biomass has been extensively studied and is known to shift the decomposition of the sample towards lower temperatures [13]. Therefore, the
cellulose peak in Mix 1 and Mix 2 was expected to shift to a lower temperature, due
cr
to the presence of xylan in the sample, which had high ash content. No such shift was observed however. It seems that in the case of simple mechanical mixing that was
us
used to produce the mixed samples, the presence of ash in hemicellulose and lignin did not affect the decomposition of cellulose. This is in agreement with Couhert et al.
an
[11] who observed that ash only slightly affected pyrolysis behavior when simple mixing with biomass was applied.
M
3.2. Pyrolysis experiments 3.2.1. Pyrolysis of separate components
d
The product yield distributions from the thermal and catalytic pyrolysis of cellulose, xylan and lignin are given in Fig. 4. The pyrolysis of cellulose yielded the
te
most bio-oil, with high organic fraction and low water content. As in the TG
Ac ce p
experiments (section 3.1.1), out of the three components studied, cellulose yielded the lowest solid residue (21 wt.%). Xylan on the other hand, gave high yields of gas products and bio-oil rich in water and poor in organic fraction (Fig. 4). As seen in the TG experiments (Fig. 1), xylan decomposed at lower temperatures than the temperature at which the pyrolysis experiments were performed (500 °C). At such high temperature, the pyrolysis vapors of xylan underwent significant secondary cracking and gave high yields of water, gas products and coke. The significant cracking of the xylan pyrolysis vapors could also be attributed to the fact that the commercial xylan sample used in this study had high ash content, which is known to promote gas formation at the expense of liquid products [12]. The pyrolysis of lignin gave moderate yields of bio-oil, very low gas yields and remarkably high solid residue yields, in accordance with our TG experimental results (Fig. 1). When using ZSM-5 catalyst for the upgrading of the pyrolysis vapors, the water, gas and solid residue yields increased, while the organic and total liquid product
9
Page 9 of 34
yields decreased, as was observed with regular biomass in our previous study [7]. The decrease in the total liquid product yield and the increase in the water and gas yields are due to the deoxygenation reactions that occur on the catalyst, which remove oxygen from the pyrolysis vapors as CO, CO2 and H2O. The solid product yield
ip t
increased due to the formation of coke deposits on the surface and the pores of the catalyst.
The chemical composition of the organic phase of the bio-oil obtained from both
cr
thermal and catalytic pyrolysis of cellulose, xylan and lignin, was analyzed with a
GC-MS system and is presented in Fig. 5, Fig. 6 and Fig. 7 respectively. Usually,
us
hundreds of compounds can be identified in the organic phase of the bio-oil and in order to make the comparison of the results easier, the identified compounds were
an
categorized in groups. These groups were aromatic hydrocarbons (AR), aliphatic hydrocarbons (ALI), phenolic compounds (PH), furans (FUR), acids (AC), esters (EST), alcohols (AL), ethers (ETH), aldehydes (ALD), ketones (KET), polycyclic
M
aromatic hydrocarbons (PAH), sugars (SUG) and compounds that contain nitrogen (NIT).
The organic phase of the bio-oil from the thermal pyrolysis of cellulose could not
d
be separated from the aqueous phase and was analyzed as it was. The main products
te
from the thermal pyrolysis of cellulose were dehydrated sugars. A wide peak identified as 1,6-Anhydro-β-D-glucopyranose (levoglucosan) covered a large portion
Ac ce p
of the chromatogram. A sharp peak of 1,4:3,6-Dianhydro-α-d-glucopyranose was also identified in the samples. Other products that covered a significantly smaller portion of the chromatogram were identified as well, such as some simple phenols (phenol and phenol molecules with methyl substitutes), ketones (mainly cyclic ketones), aldehydes and alcohols. The cellulose catalytic bio-oil had higher water content and its organic phase was easily separated from the aqueous phase before GC-MS analysis. Sugars were considerably reduced after catalysis. Because sugars are watersoluble, the remarkable drop in identified sugars could be attributed, in part, to their removal from the sample, along with the aqueous phase. Levoglucosan and 1,4:3,6Dianhydro-α-d-glucopyranose were still present in the organic phase, but their total area percentage in the chromatogram was considerably lower. Many new products were identified in the catalytic bio-oil, most of which were aromatic hydrocarbons, such as xylene, toluene, indane and indene and PAHs, such as naphthalene and naphthalene molecules with methyl substitutes.
10
Page 10 of 34
As shown in Fig. 6, the main products from the thermal pyrolysis of xylan are ketones and phenolic compounds. Acetic and propanoic acid, as well as a few aromatic hydrocarbons were also present. The ketones were mainly 1-hydroxy-2propanone (hydroxyacetone) and various cyclopentenones with methyl- and ethyl-
ip t
substitutes. The phenolic compounds were mainly simple phenols with methylsubstitutes. After catalysis with ZSM-5, the ketones in the organic phase were drastically reduced, while the phenols and the aromatic hydrocarbons were
cr
significantly increased. Formation of PAHs (napthalenes) was also observed. The
aromatic hydrocarbons that were formed after catalysis were toluene, xylene and other
us
substituted benzene ring compounds, as well as indanes and indenes. The phenolic compounds identified were mostly simple phenols with methyl- substitutes.
an
Hydroxyacetone and various cyclopentenones were still present in the bio-oil after catalysis but their peaks occupied a significantly smaller area of the chromatogram. The products derived from the thermal pyrolysis of lignin were almost exclusively
M
phenols, as shown in Fig. 7. Some heavy acids were also detected. The phenolic compounds derived from lignin were different from those derived from cellulose and xylan and were mostly benzenediols (catechols), phenols with methoxy- substitutes
d
and more complex phenolic compounds with higher molecular weights. An
te
interesting observation is that after catalysis, besides the drop in the acids content and the slight increase of PAHs, the composition of the bio-oil remained almost unaltered.
Ac ce p
The catalytic bio-oil from lignin was still, almost exclusively, composed of benzenediols and phenols with methoxy- substitutes. This confirms previous reports that the C-O bond between the aromatic ring and the hydroxyl group in the phenol molecule is refractory to ZSM-5 [13]. 3.2.2. Pyrolysis of mixtures
As was seen in the TG experiments with cellulose, xylan and lignin mixtures, the
behavior of each mixture during pyrolysis could be predicted quite well by assuming that the overall pyrolysis behavior is the weighted sum of the partial contributions of its components. In fact, the final solid residue at temperatures above 400 °C was predicted with very good accuracy (see section 3.1.2). In order to verify whether such predictions for the pyrolysis product yields and composition can be achieved under fast pyrolysis conditions, a set of thermal and catalytic experiments with ZSM-5 was performed with Mixture 1 and Mixture 2 in the bench scale biomass pyrolysis unit.
11
Page 11 of 34
The results were compared to calculated values that were based on the experimental results from the thermal and catalytic pyrolysis of the separate biomass components (see section 3.2.1). The experimental results and calculated values for the product yields of the
ip t
thermal and the catalytic pyrolysis of both mixtures are presented in Fig. 8 and Fig. 9 respectively. From the data presented, it is obvious that the calculations could not
accurately predict the product yields; differences between the experimental and
cr
calculated values were observed. In particular, gas and solid product yields were
higher than expected, while bio-oil and organic phase yields were lower. This trend
us
was observed both for Mix 1 and Mix 2 and both for thermal and catalytic pyrolysis. As seen in the TG experiments of the biomass component mixtures, the derivative
an
weight loss peaks of cellulose in the mixed biomass samples were wider, had lower intensity and were shifted towards higher temperatures, while the xylan peaks were not significantly affected (Fig. 2 and 3). This behavior was attributed to limited heat
M
and mass transfer due to the initial decomposition of xylan and lignin that melted and formed a solid residue around cellulose particles. This seems to be well supported by our fast pyrolysis experimental results. Poor heat and mass transfer in fast pyrolysis
te
gas yields.
d
conditions is known to result in higher biomass residue yields, lower liquid and higher The composition of the organic phase from the thermal and catalytic pyrolysis of
Ac ce p
biomass mixtures is presented in Fig. 10 and Fig. 11 respectively. It can be seen that there can be a significant difference between the experimental and calculated composition of the liquid product. In the thermal pyrolysis experiments (Fig. 10), the chromatogram area of sugars (mainly levoglucosan) was significantly smaller than the calculations, while ketones in Mix 1 and phenols in Mix 2 were more abundant. Closer examination of the GC-MS results revealed that the increased phenolic compounds in Mix 2 were phenols with oxygenated substitutes, which are characteristic products of the decomposition of lignin. In the catalytic experiments, the most striking difference between the experimental and the calculated results is observed in the total chromatogram area % of phenols, which in the experimental runs was significantly higher than expected for both Mix 1 and Mix 2. As evident from the significant differences between the experimental and the calculated results, the prediction of the chemical composition of the bio-oil was even more challenging than the prediction of the pyrolysis yields. Due to its complexity,
12
Page 12 of 34
bio-oil is difficult to analyze by conventional GC-MS. Compounds that can be identified in one sample may not be identified in another, more complex sample because of convoluted peaks. In addition, the analysis of the organic phase was semiquantitative and the area covered by a peak did not necessarily correspond to a mass
ip t
fraction. For these reasons, it would not be safe to elaborate on these results. In a similar study, Wang et al. [6] also observed significant differences between the experimental and calculated values of the peak area % in their chromatograms.
cr
3.3. Reaction pathways
us
Based on our results from the analyses of the organic phase of the fast pyrolysis bio-oil of cellulose, xylan and lignin, an elucidation of the reaction pathways during thermal pyrolysis was attempted and is described in this section. In the catalytic
an
experiments, the biomass particles did not come in contact with the catalyst particles, therefore the pyrolysis vapors that reached the catalyst bed were assumed to be the
M
same as in the non-catalytic experiments. When the vapors passed through the catalyst, the compounds of the thermal pyrolysis were catalytically converted to other products. Therefore, by comparing the product distribution from the thermal and
te
possible.
d
catalytic experiments, an elucidation of the catalytic reaction pathways was also
Ac ce p
3.3.1. Pyrolysis pathways of cellulose
The main product from the thermal pyrolysis of cellulose in our study was
levoglucosan. At fast pyrolysis conditions, pure cellulose is known to depolymerize and form levoglucosan via a transglycosylation mechanism [14-16]. Other anhydrosugars, such as levoglucosenone, 1,4:3,6-Dianhydro-α-D-glucopyranose and Ethyl α-D-glucopyranoside, were also identified in the organic phase. Such anhydrosugars can be produced either by the primary pyrolysis of cellulose, or by the secondary pyrolysis of levoglucosan, as showed by Kawamoto et al. [17]. Kawamoto et al. [17] also showed that levoglucosan, through ring-opening polymerization, can form polysaccharides, which eventually form carbonized products (solid residue) and Lin et al. [18] proposed that the majority of the cellulose solid residue is formed through this mechanism and not from the cellulose itself. Apart from anhydrosugars, small peaks of phenolic compounds were identified as well. Other researchers [16, 19, 20] also detected phenolic compounds in the cellulose pyrolysis products. Evans and
13
Page 13 of 34
Milne [16] showed that these aromatic products are not derived from the primary pyrolysis of the cellulose, but are formed from secondary reactions (probably from the gas phase polymerization of unsaturated species such as propylene, butadiene and butene). The acetic acid, which was also detected in our experimental runs, is
ip t
considered to derive from hydroxyacetaldehyde by dehydration to form ketene and subsequent hydration to acetic acid [14, 15]. Due to its early elution, hydroxyacetaldehyde was not detected in our analyses, but is widely reported as a
cr
cellulose pyrolysis product in the literature [14, 15, 18, 21, 22]. Acetic acid is also formed from the thermal decomposition of 5-hydroxymethylfurfural (5-HMF) [23],
us
which is formed either via the direct ring-opening and rearrangement reactions of cellulose unit molecules or via the secondary reactions of levoglucosan [15, 17].
an
Finally, our analyses also showed the presence of cyclic ketones in the thermal bio-oil from cellulose, which were also reported by Wang et al. [23]. More peaks were present in our chromatogram but could not be identified by the GC-MS system.
M
In the catalytic experiments, the levoglucosan content in the organic phase of the bio-oil from cellulose was significantly reduced, while other compounds were formed. The most notable compounds observed after catalysis were aromatic hydrocarbons,
d
phenols, furans and PAHs. The aromatic hydrocarbons were benzene compounds with
te
linear substitutes, as well as indanes and indenes. PAHs were naphthalene compounds with methyl- substitutes. Phenolic compounds were also detected before catalysis, but
Ac ce p
their relative abundance (area %) increased significantly after catalysis. Both before and after catalysis, phenolic compounds were phenols with methyl- substitutes. Finally, we also detected furan compounds that were mostly benzofurans and furans with methyl- substitutes.
Horne and Williams [13] showed that benzenes and napthalenes can be found in
the catalytic oil product from cyclic ketones (cyclopentanone), as well as from furfural. They also showed that benzofurans can form from the catalytic transformation of furfural and indanes/indenes from the catalytic transformation of cyclopentanone. Carlson et al. [24] reported the possible contribution of furan compounds (furfural, 5-HMF and furfuryl alcohol) to the formation of hydrocarbons as well. Grandmaison et al. [25] also reported the formation of aromatic hydrocarbons, as well as benzofurans during the conversion of furanic compounds (furfural and furan) over ZSM-5. Simple aromatic hydrocarbons and phenols may also form in the gas phase by secondary polymerization of unsaturated light compounds,
14
Page 14 of 34
as suggested by Evans and Milne [16], which are expected to be more abundant in the catalytic vapors due to the cracking effect of the ZSM-5 catalyst. Also, part of the phenols that ended up in the catalytic pyrolysis oil, were derived from the thermal pyrolysis of cellulose due to the fact that the C-O bond between the benzene ring and form thermal coke, or pass unconverted through the catalyst [26].
cr
3.3.2. Pyrolysis pathways of xylan
ip t
the methoxy group is refractory to ZSM-5 [13] and the phenolic molecules either
The majority of the products observed from the thermal pyrolysis of xylan were
us
phenols and cyclic ketones. Phenols are likely derived from the cleavage of the ferulic acid ester branch of xylan, as well as from the gas phase polymerization of unsaturated light species [16]. The cyclic ketones are likely derived from the xylan
an
main stem by cleavage of the o-glucosidic bonds and subsequent removal of the hydroxyl groups of the xylose rings. Another major product from the thermal
M
pyrolysis of xylan was hydroxyacetone. Propanoic and acetic acid were also observed. Acetic acid is formed by cleavage of the acetyl group of the xylan structure with simultaneous release of CO2 [27].
d
After catalysis, aromatic hydrocarbons, phenolic compounds and PAHs were
te
considerably increased, while acids and especially ketones were reduced. Aromatic hydrocarbons were mostly benzenes with methyl-/ethyl- substitutes, as well as
Ac ce p
indenes/indanes and PAHs were napthalenic compounds. As in thermal pyrolysis, phenolic compounds were simple phenols with methyl-/ethyl- substitutes. This product distribution in the catalytic bio-oil was expected, since the compounds formed from the thermal pyrolysis were mainly linear ketones, cyclic ketones and phenols. Horne and Williams [13] showed that cyclic ketones (cyclopentanone) form simple aromatic compounds when passed over a ZSM-5 catalyst and are also responsible for the formation of indanes/indenes and napthalenes. Hydroxyacetone, which was a major product in the thermal pyrolysis vapors from xylan, tentatively followed a reaction pathway proposed by Gayubo et al. [28] and transformed into olefins, paraffins and eventually, aromatic hydrocarbons. As mentioned previously, the C-O bond in the phenolic compounds is refractory to ZSM-5 and the phenols from the thermal pyrolysis vapors either formed thermal coke or passed unconverted through the catalyst. Additional phenols, as well as simple aromatic hydrocarbons, can form in the gas phase by polymerization of light unsaturated species [16], which are expected
15
Page 15 of 34
to be more abundant during catalytic pyrolysis due to the cracking properties of the ZSM-5 catalyst. Finally, Adjaye and Bakhshi [29] suggested that carboxylic acids are easily converted over ZSM-5 and form mostly coke, gases and water, while a small fraction is converted to aromatic hydrocarbons.
ip t
3.3.3. Pyrolysis pathways of lignin The identified compounds in the thermal pyrolysis oil from lignin were almost
cr
entirely phenolic compounds, which were more complicated than the phenols from cellulose and xylan, as they were mostly phenols with methoxy- substitutes,
us
benzenediols and polysubstituted phenols. The complex structures of these phenolic compounds indicated that they were derived directly from the lignin matrix. Simpler chromatogram was significantly smaller.
an
phenols with methyl- substitutes were also observed but their total area in the Kawamoto et al. and Asmadi et al. recently investigated the thermal pyrolysis of
M
lignin using representative model dimers [30], guaiacol and syringol [31] and other model compounds [32]. Overall, their studies showed that the thermal decomposition of lignin begins by cleavage of the weak α-ether and β-ether bonds of the lignin
d
structure, which releases guaiacyl- and syringyl-type aromatics, depending on the
te
lignin source. Softwood lignins have only guaiacyl-type nuclei, while hardwood lignins have both [30]. These aromatics further reacted, either by homolysis of the O-
Ac ce p
CH3 bonds to produce catechols and pyrogallols, or by radical induced rearrangement (ipso-substitution) reactions to produce cresols and xylenols [31]. Further decomposition of catechols and pyrogallols yields gases, coke and few PAHs, while decomposition of cresols and xylenols yields phenol and cresols by demethylation [32].
Since the C-O bond in the phenol molecule is refractory to ZSM-5, the presence
of the catalyst was not expected to have any significant impact on the composition of the catalytic bio-oil. Although there was a slight increase in the yields of gases, solid products (coke) and water, an indication of the catalytic effect of the ZSM-5, the composition of the bio-oil was not substantially altered. Methoxy- substituted phenols, benzenediols and poly-substituted phenols were still the prevalent compounds. The increase in gas and water yields was most likely due to cleavage of the side groups of the phenolic units as they passed through the catalyst. The aromatic ring and the C-O bond were not affected.
16
Page 16 of 34
4. Conclusions Thermal and catalytic pyrolysis of biomass components (cellulose, xylan and lignin) and mixtures thereof was performed in both a TG apparatus and a bench-scale pyrolysis reactor. The pyrolysis of cellulose gave high yields of bio-oil rich in
ip t
levoglucosan and other anhydrosugars with minimal coke formation. Xylan gave high
gas yields and moderate yields of bio-oil, rich in water, phenols and ketones. Lignin
cr
gave the highest solid residue yield and produced a bio-oil that was rich in phenolic
compounds. The pyrolysis of biomass component mixtures was studied in order to
us
investigate whether the calculation of the product distribution and quality from the pyrolysis of a biomass with known composition is possible. The calculation of the final biomass residue yield at 500 °C was achieved with very good accuracy in the TG
an
experiments, but differences between the calculated and experimental DTG curves were observed, which were attributed to limited heat and mass transfer in the biomass
M
mixtures. The effects of the limited heat and mass transfer were more evident in the pyrolysis experiments, where higher than expected solid residue and gas yields were obtained, at the expense of liquid products. The composition of the bio-oil produced
d
from the biomass mixtures was also affected. Even though characteristic chemical
te
groups of each biomass component were present in the bio-oil, their relative abundance was different than what would be expected.
Ac ce p
An elucidation of the reaction pathways before and after catalysis was also attempted, based on the observation of the products from the pyrolysis of the individual components and reports from the literature. Cellulose formed mainly levoglucosan via transglycosylation, which was easily converted catalytically to other products, mainly aromatic hydrocarbons, phenols, furans and PAHs. Xylan formed a wider array of products, mainly phenols and ketones as well as some acids. Acids and ketones were reduced after catalysis and formed aromatic hydrocarbons, phenols and PAHs. Due to its aromatic structure, lignin formed mainly phenolic products that could not be catalytically converted to other chemical groups.
Acknowledgements
This research has been co-financed by European Union (European Social Fund) and Greek national funds through Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF)-
17
Page 17 of 34
Research Funding Program: THALES-Investing in knowledge society through the European Social Fund (MIS 380405). References
ip t
[1]. P. McKendry, Energy production from biomass (part 1): overview of biomass, Bioresource Technol. 83 (2002) 37–46. and biomass components, Fuel 75 (1996) 987-998.
cr
[2]. K. Raveendran, A. Ganesh, K.C. Khilar, Pyrolysis characteristics of biomass
[3]. H. Yang, R. Yan, H. Chen, C. Zheng, D. Lee, In-depth investigation of biomass
us
pyrolysis based on three major components: hemicellulose, cellulose and lignin. Energ. Fuel. 20 (2006) 388-393.
an
[4]. N. Worasuwannarak, T. Sonobe, W. Tanthapanichakoon, Pyrolysis behaviors of rice straw, rice husk, and corncob by TG-MS technique, J. Anal. Appl. Pyrol. 78 (2007) 265–271.
M
[5]. G. Wang, W. Li, B. Li, H. Chen, TG study on pyrolysis of biomass and its three components under syngas, Fuel 87 (2008) 552–558.
d
[6]. S. Wang, X. Guo, K. Wang, Z. Luo, Influence of the interaction of components on the pyrolysis behavior of biomass, J. Anal. Appl. Pyrol. 91 (2011) 183–189.
te
[7]. S.D. Stefanidis, K.G. Kalogiannis, E.F. Iliopoulou, A.A. Lappas, P.A. Pilavachi, In-situ upgrading of biomass pyrolysis vapors: Catalyst screening on a fixed bed
Ac ce p
reactor, Bioresource Technol. 102 (2011) 8261–8267.
[8]. H. Yang, R. Yan, H. Chen, D. Lee, C. Zheng. Characteristics of hemicellulose, cellulose and lignin pyrolysis, Fuel 86 (2007) 1781-1788.
[9]. P.R. Patwardhan, R.C. Brown, B.H. Shanks, Product distribution from the fast pyrolysis of hemicellulose. ChemSusChem 4 (2011) 636–643.
[10]. T. Hosoya, H. Kawamoto, S. Saka, Cellulose–hemicellulose and cellulose– lignin interactions in wood pyrolysis at gasification temperature, J. Anal. Appl. Pyrol. 80 (2007) 118–125. [11]. C. Couhert, J-M. Commandre, S. Salvador, Is it possible to predict gas yields of any biomass after rapid pyrolysis at high temperature from its composition in cellulose, hemicellulose and lignin?, Fuel 88 (2009) 408–417.
18
Page 18 of 34
[12]. R. Fahmi, A.V. Bridgwater, L.I. Darvell, J.M. Jones, N. Yates, S. Thain, I.S. Donnison, The effect of alkali metals on combustion and pyrolysis of Lolium and Festuca grasses, switchgrass and willow, Fuel 86 (2007) 1560–1569 [13]. P. Horne, P. Williams, Reaction of oxygenated biomass pyrolysis model
ip t
compounds over a ZSM-5 catalyst, Renew. Energ. 7 (1996) 131–144. [14]. J. Piskorz, D. Radlein, D.S. Scott, On the mechanism of the rapid pyrolysis of cellulose, J. Anal. Appl. Pyrol. 9 (1986) 121–137.
cr
[15]. D. Shen, S. Gu, The mechanism for thermal decomposition of cellulose and its main products, Bioresource Technol. 100 (2009) 6496–6504.
us
[16]. R.J. Evans, T.A. Milne, Molecular characterization of the pyrolysis of biomass, Energ. Fuel. 1 (1987) 123–137.
an
[17]. H. Kawamoto, M. Murayama, Pyrolysis behavior of levoglucosan as an intermediate in cellulose pyrolysis: polymerization into polysaccharide as a key reaction to carbonized product formation, J. Wood Sci. 49 (2003) 469-473.
M
[18]. Y-C. Lin, J. Cho, G.A. Tompsett, P.R. Westmoreland, G.W. Huber, Kinetics and mechanism of cellulose pyrolysis. J. Phy. Chem. C 113 (2009) 20097– 20107.
d
[19]. Y.F. Liao, Mechanism study of cellulose pyrolysis, PhD Thesis, ZheJiang
te
University, HangZhou, China, 2003.
[20]. R. Alen, E. Kuoppala, P. Oesch, Formation of the main degradation compound
Ac ce p
groups from wood and its components during pyrolysis, J. Anal. Appl. Pyrol. 36 (1996) 137–148.
[21]. Z.Y. Luo, S.R. Wang, Y.F. Liao, K.F. Cen, Mechanism study of cellulose rapid pyrolysis, Ind. Eng. Chem. Res. 43 (2004) 5605–5610.
[22]. J. Banyasz, S. Li, J. Lyons-Hart, K. Shafer, Gas evolution and the mechanism of cellulose pyrolysis, Fuel 80 (2001) 1757–1763.
[23]. S. Wang, X. Guo, T. Liang, Y. Zhou, Z. Luo, Mechanism research on cellulose pyrolysis by Py-GC/MS and subsequent density functional theory studies, Bioresource Technol. 104 (2011) 722-728. [24]. T.R. Carlson, J. Jae, Y-C. Lin, G.A. Tompsett, G.W. Huber, Catalytic fast pyrolysis of glucose with HZSM-5: The combined homogeneous and heterogeneous reactions, J. Catal. 270 (2010) 110–124. [25]. J.L. Grandmaison, P.D. Chantal, S.C. Kaliaguine, Conversion of furanic compounds over H-ZSM-5 zeolite, Fuel 69 (1990) 1058–1061.
19
Page 19 of 34
[26]. P. Chantal, S. Kaliaguine, J. Grandmaison, Reactions of phenolic compounds over HZSM-5, Appl. Catal. 18 (1985) 133–145. [27]. D.K. Shen, S. Gu, A.V. Bridgwater, Study on the pyrolytic behaviour of xylanbased hemicellulose using TG-FTIR and Py-GC-FTIR, J. Anal. Appl. Pyrol. 87
ip t
(2010) 199–206. [28]. A. Gayubo, A. Aguayo, A. Atutxa, Transformation of oxygenate components of
biomass pyrolysis oil on a HZSM-5 zeolite. II. Aldehydes, ketones, and acids,
cr
Ind. Eng. Chem. Res. 43 (2004) 2619-2626.
[29]. J. Adjaye, N. Bakhshi, Catalytic conversion of a biomass-derived oil to fuels
us
and chemicals. 1. Model-compound studies and reaction pathways, Biomass Bioenerg. 8 (1995) 131–149.
an
[30]. H. Kawamoto, S. Horigoshi, S. Saka, Pyrolysis reactions of various lignin model dimers, J. Wood Sci. 53 (2006) 168–174.
[31]. M. Asmadi, H. Kawamoto, S. Saka, Thermal reactions of guaiacol and syringol [32]. M.
Asmadi,
H.
M
as lignin model aromatic nuclei, J. Anal. Appl. Pyrol. 92 (2011) 88–98. Kawamoto,
S.
Saka,
Thermal
reactivities
of
catechols/pyrogallols and cresols/xylenols as lignin pyrolysis intermediates, J.
Ac ce p
te
d
Anal. Appl. Pyrol. 92 (2011) 76–87.
20
Page 20 of 34
Figure Captions Fig. 1. Weight loss and derivative weight loss curves for cellulose, xylan and lignin Fig. 2. Experimental and calculated TG and DTG curves for Mix 1 Fig. 3. Experimental and calculated TG and DTG curves for Mix 2
ip t
samples
cr
Fig. 4. Product yield distribution from the thermal and catalytic pyrolysis of cellulose, xylan and lignin
us
Fig. 5. Chemical composition of the organic phase of thermal and catalytic bio-oil from cellulose
Fig. 6. Chemical composition of the organic phase of thermal and catalytic bio-oil
an
from xylan
Fig. 7. Chemical composition of the organic phase of thermal and catalytic bio-oil
M
from lignin
Fig. 8. Comparison between experimental (exp.) and calculated (calc.) product yield values of the thermal pyrolysis of biomass mixtures
d
Fig. 9. Comparison between experimental (exp.) and calculated (calc.) product yield
te
values of the catalytic pyrolysis of biomass mixtures Fig. 10. Experimental (exp.) and predicted (calc.) composition of the organic phase
Ac ce p
from the thermal pyrolysis of biomass mixtures Fig. 11. Experimental (exp.) and predicted (calc.) composition of the organic phase from the catalytic pyrolysis of biomass mixtures
21
Page 21 of 34
Table 1. Composition of the biomass feedstocks used in this study (on dry basis).
C (wt.%) H (wt.%) O (wt.%)1 Ash (wt.%) 42.27 6.40 51.33 0.00 42.07 5.82 45.75 6.36 63.41 5.89 28.42 2.28 1 By difference (oxygen% = 100% - carbon% - hydrogen% - ash%)
Ac ce p
te
d
M
an
us
cr
ip t
Sample Cellulose Xylan Lignin
22
Page 22 of 34
Highlights Pyrolysis of biomass component mixtures in both TGA and pyrolysis reactor setups Comparison between experimental and calculated product yields and composition
ip t
Heat transfer lag was evident during the pyrolysis of biomass component mixtures Elucidation of thermal and catalytic reaction pathways of each biomass
Ac ce p
te
d
M
an
us
cr
component
23
Page 23 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1
Page 24 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 2
Page 25 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3
Page 26 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4
Page 27 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 5
Page 28 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 6
Page 29 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 7
Page 30 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 8
Page 31 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 9
Page 32 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 10
Page 33 of 34
Ac
ce
pt
ed
M
an
us
cr
i
Figure 11
Page 34 of 34