characteristics and origin of char - shoucheng.pdf

Green Chemistry

PAPER

Cite this: Green Chem., 2013, 15, 3214

Received 3rd August 2013,Accepted 12th September 2013

DOI: 10.1039/c3gc41581c

www.rsc.org/greenchem

Characteristics and origin of char and coke from fastand slow, catalytic and thermal pyrolysis of biomassand relevant model compounds

Shoucheng Du, Julia A. Valla and George M. Bollas*

Char and coke from biomass catalytic pyrolysis have different origins. They cannot be lumped as one

since they occupy different locations on the catalyst surface and, thus, contribute differently to catalyst

deactivation. In this study, catalyst (ZSM-5) deactivation in the perspective of comparison of char and

coke from pyrolysis of different biomass types is investigated. Pine sawdust, glucose, and cellulose are

used as feedstocks in the pyrolysis experiments. Biomass char and coke samples produced via slow and

fast, thermal and catalytic pyrolysis are characterized with respect to their overall content, oxidation reac-

tivity, catalyst surface area, pore size distribution changes, bonding groups and their effect on catalyst

performance. In particular, it is shown that char forms as an external layer on the catalyst surface and in

its macropores, whereas coke forms inside the zeolite micropores via hydrogen transfer and addition

reactions. The catalyst effect on glucose and pine slow catalytic pyrolysis is minor compared with that on

cellulose slow catalytic pyrolysis, due to macropore blocking by char formation. In fast catalytic pyrolysis,

catalyst deactivation is mainly attributed to micropore blocking by coke formation. Char and coke are

shown to coexist on the catalyst surface after fast catalytic experiments, with the char content after

glucose fast catalytic pyrolysis being 30 wt% of the total solid residue. The origins of char and coke in the

cellulose, hemicellulose and lignin components of pine are identified and mechanisms for their for-

mation are proposed.

Introduction

Biomass has received considerable attention as a close to CO2

neutral and sustainable feedstock that can replace fossil fuelsfor energy generation.1–5 Lignocellulosic biomass is a low-costfeedstock that is uniquely suited for the production of sustain-able liquid fuels.4,6,7 However, lignocellulosic biomass isdifficult to deconstruct into hydrocarbon-containing sub-fractions, because of its heterogeneous composition.7 Theapproaches used for lignocellulose deconstruction can bebroadly lumped into combustion, gasification, liquefaction,and pyrolysis.8,9 Among these, pyrolysis is developing rapidlyand can play a very important role in the future of renewableenergy production.10 Depending on the heating rate, biomasspyrolysis can be separated into two categories: slow pyrolysis(0.1–1 K s−1) and fast pyrolysis (10–1000 K s−1 or higher,including flash pyrolysis).11 Slow pyrolysis produces largeamounts of carbonaceous residues, which can be used as a

solid fuel or fertilizer, whereas fast pyrolysis produces highyields of bio-oil.6 Fast catalytic pyrolysis enhances selectivity tohydrocarbons, particularly aromatics.12–15 The reason for intro-ducing a catalyst to the pyrolysis process is primarily becauseof its ability to improve the quality of bio-oil.16

In biomass pyrolysis, slow or fast, catalytic or thermal,there is always some solid residue (typically, a mixture of cokeand char, depending on the pyrolysis process) produced inparallel to non-condensable products and condensable bio-oilplus water.9,17,18 However, the definition of coke and charvaries among different studies. Elordi et al.19 studied the fastcatalytic pyrolysis of polyethylene in a spouted bed reactor at500 °C, using HZSM-5, HY, Hβ catalysts. They defined coke asthe carbonaceous material deposited on the catalyst. Theyclaimed that the combustion of coke in the meso- and macro-pores of the catalyst shows a temperature programmed oxi-dation (TPO) peak at lower temperatures, compared with thecoke located inside the zeolite crystal channels due to differ-ences in composition. They also observed coke outside thezeolite crystals with heterogeneous sizes between 10 and50 nm, using transmission electron microscopy (TEM). In thereview of biofuel production by Huber and Corma,20 coke wasdefined as the organic fraction that could only be removed

Department of Chemical & Biomolecular Engineering, University of Connecticut,

Storrs, 191 Auditorium Road, Unit 3222, Storrs, CT 06269-3222, USA.

E-mail: [email protected]; Tel: +1-860-486-4602

3214 | Green Chem., 2013, 15, 3214–3229 This journal is © The Royal Society of Chemistry 2013

Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article OnlineView Journal | View Issue

www.rsc.org/greenchem

http://dx.doi.org/10.1039/c3gc41581c

http://pubs.rsc.org/en/journals/journal/GC

http://pubs.rsc.org/en/journals/journal/GC?issueid=GC015011

from the catalyst via calcination. Char was defined as theorganics deposited in the reactor as a result of thermaldecomposition, but not on the catalyst. Triantafyllidis et al.21

studied the fast catalytic pyrolysis of beech wood in a fixed bedreactor at 500 °C, using mesoporous aluminosilicate and con-ventional Al-MCM-41 catalysts. They considered coke as alump of the solid carbonaceous residues produced thermallyin the reactor as a separate phase to the catalyst, as well as thesolid residues deposited on the catalyst surface due to thermaland catalytic cracking. Generally, coke is considered as the cata-lytic product, whereas char is the residue formed via thermaldeconstruction. This definition is widely accepted, and manyresearchers describe the solid residue after thermal pyrolysisas char.22–24 Based on this definition, primary decompositionand secondary polymerization contribute to char for-mation,25,26 while coke formation is mainly attributed to cata-lytic polymerization of small biomass molecules inside catalystpores.27

Generally, coke formation leads to catalyst deactivation andresults in undesirable product selectivity in biomass pyrolysis,6

whereas char may or may not deactivate the catalyst, depend-ing on the location of its formation. However, the real reasonfor catalyst deactivation due to coke and char is not well under-stood. Carlson et al.28 studied the effect of ZSM-5 deactivation(due to coking) on the selectivity of glucose fast catalyticpyrolysis in a pyroprobe reactor. Coke yields of the order of33–45 mol% (moles of carbon in coke per total moles ofcarbon) were measured, which translates to about 15 wt% cokeyield on glucose mass basis. In experiments with zeolite toglucose ratios of 19 (thus, coke on catalyst of about 0.79 wt%),they observed positive effects of coke formation on biomasspyrolysis selectivity. On the contrary, Aho et al.29 performedfast catalytic pyrolysis experiments with pine and ZSM-5 in afluidized bed reactor at 450 °C, but with low zeolite tobiomass ratio (0.4), showing that coking of the zeolite leadsto a significant decrease in catalytic activity. Also, Chengand Huber13 investigated the conversion of furan over HZSM-5in a fixed-bed reactor at 600 °C. A continuous loss ofcatalytic activity was observed as the amount of coke increasedon the catalyst surface. In most studies, carbon and char arelumped as “coke”, but according to Aho et al.29 the char tocoke ratio after pyrolysis of pine sawdust with ZSM-5 isabout 2 : 1.

As indicated above, significant effort has been devoted onthe characterization of biomass char and coke. However, mostof the published work focuses on thermal chars,30–32 leadingto a lack of understanding of the mechanisms and effects ofchar and coke formation when catalyst is introduced in thepyrolysis. This work focuses on studying catalyst deactivationand the effectiveness of model compounds, in particularglucose and cellulose, as compared to pine from the perspec-tive of comparing their pyrolysis char and coke characteristics.The objective of this study is to explore the difference betweencoke and char residues and explore the origin of their for-mation and their contribution to catalyst deactivation duringbiomass catalytic pyrolysis.

Experimental sectionExperimental setup

Two experimental setups are used in the pyrolysis experiments.Slow pyrolysis is performed in a fixed bed quartz reactor(1 inch o.d. and 24 inch length), which is heated in a horizon-tal tube furnace. Fast pyrolysis is studied in a speciallydesigned spouted bed reactor (Fig. 1). In the spouted bedreactor, biomass is fed with an auger feeder at the bottom andenters the reactor via entrainment with 4–10 l min−1 N2 flow.The reactor operates at temperatures up to 1000 °C, 600 °C inthe experiments discussed here. The products pass through siximpingers, with three of them filled with 5 ml methanol each,to collect the liquid products. A cooling jacket is used at thebottom of the inlet tubing of the reactor, to prevent thermalpyrolysis at lower temperatures (or temperature gradients)before entering the spouting zone. As shown by Ferdouset al.,33 Nowakowski et al.,34 and Sharma et al.,35 lignin canmelt at lower temperatures before entering the hot reactorzone, which makes it very difficult to study the effect of ligninon the pyrolysis of lignocellulosic biomass, as most of thelignin is never fed to the reactor. This is addressed in thecurrent setup with the aforementioned cooling jacket, keepingthe feeding pipe under 100 °C, and the high gas velocities inthe feeding pipe, which prohibit early decomposition ofbiomass components.

The selection of a spouted bed reactor for biomass pyrolysiswas based on previous work by Cui and Grace,2 Atutxa et al.16

and Bilbao and co-workers.36–38 Spouted bed reactors are idealfor the characteristics of biomass: they can handle large par-ticle size distributions, larger particles, differences in particledensities, and provide excellent mixing.2 Their fountain-typehydrodynamic regime can decouple residence times of gas andsolids; thus, reducing unwanted secondary reactions,37 andprovides high heating rates and isothermality.38 The detailedreactor dimensions, design equations and calculations of the

Fig. 1 Experiment setup for biomass fast (catalytic) pyrolysis.

Green Chemistry Paper

This journal is © The Royal Society of Chemistry 2013 Green Chem., 2013, 15, 3214–3229 | 3215

Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


reactor operating variables are discussed in Du and Bollas(2013).39

Feedstock and catalyst

In order to study the mechanism of biomass pyrolysis, variousbiomass feedstocks and model compounds are being used.Among them, pine sawdust is widely used in bio-fuel pro-duction studies.12,38,40 Cellulose and glucose are popularbiomass model compounds. Cellulose is the major componentin cellulosic biomass (typically, 23–32% in lignocellulose);therefore, the study of cellulose pyrolysis has been consideredto be critical for the study of biomass pyrolysis mecha-nisms.22,41,42 Glucose, the monomer of cellulose, is often usedas a model compound for cellulose to simplify simulations.26

Recently, Mettler et al.43 studied the fast thermal pyrolysis ofglucose, cellodextrins and cellulose in a thin-film pyrolysisreactor at 500 °C. They reported very different product dis-tributions between glucose (with –OH groups instead ofglycosidic linkages) and cellulose pyrolysis, showing the in-appropriateness of using glucose as a model compound forcellulose.

According to the above discussion, pine, cellulose andglucose were chosen as the feedstocks in this study. Sawdustfrom pine bark (50.81 wt% C, 5.95 wt% H, 42.96 wt% O,0.28 wt% N), α-cellulose from Sigma-Aldrich and α-D-glucosefrom Acros Organics were used as biomass and correspondingmodel compounds. The grinded pine sawdust and glucose andcellulose powders were sieved to <350 μm in particle size.

In biomass catalytic pyrolysis, zeolite catalysts, includingZSM-5, Beta zeolite, Y zeolite, Mordenite and several meso-porous materials, have been widely studied.21,29,44,45 In themajority of these studies, ZSM-5 has been proven to be veryeffective in catalytic (fast) pyrolysis due to its proper pore mor-phology. Jae et al.44 investigated the shape selectivity of zeolitecatalysts for glucose conversion in fast catalytic pyrolysis in apyroprobe reactor at 600 °C, using different zeolite catalysts.They showed that medium pore zeolites, such as ZSM-5 hadthe highest aromatic yield and minimum coke formation.Similarly, Carlson et al.46 tested different zeolite catalysts forthe conversion of glucose, xylitol, cellobiose and cellulose toaromatics using a pyroprobe reactor at 600 °C. They found thatZSM-5 had the highest aromatic yields and the lowest cokeselectivity. Thus, in this study, a commercial ZSM-5 catalyst

(synthesized by W.R. Grace & Co. in a macroporous matrix ofmean particle size of 75 μm) was used for the catalytic studies.

Experimental procedure

All experiments presented in this work were performed with abiomass to catalyst weight ratio of 1. The relatively low ratiowas chosen to maximize char/coke to catalyst ratios, and thus,exemplify the results. In the slow catalytic pyrolysis experi-ments, biomass and catalyst were well mixed before the experi-ment. Then, the mixture was pyrolyzed in a N2 environment.Before starting the experiment, N2 flow (20 ml min−1) was keptfor 1 h to purge the air inside the reactor. The feedstock wasfirst dried at 120 °C for 30 min and then the temperature wasramped to 600 °C at a rate of 10 K min−1. In fast pyrolysis,1–2 g of biomass was dried at 120 °C overnight in a separatefurnace. The desired reactor temperature (600 °C) was reachedbefore feeding the catalyst and biomass. Catalyst was fed viaentrainment from the bottom of the reactor with 4 l min−1 N2

flow. When the catalyst bed temperature reached 600 °C,biomass was fed from the bottom as well. With the high N2

flow rate, biomass reaches the spouting zone within very shorttimes (∼0.1 ms) and contacts with the hot catalyst particles.After the reaction, char/coke samples were collected viaentrainment with 19 l min−1 N2 flow. Quartz beads (sieved<180 μm) were used for the fast thermal pyrolysis experiments,assuming zero catalytic activity.

Char/coke characterization

Scanning electron microscopy (SEM). Char/coke sampleswere observed using a FEI Quanta FEG 250 scanning electronmicroscopy (SEM) under high vacuum, to distinguish thedifferences in morphology of char samples produced indifferent experiments and visualize the coke/char depositionon the surface of catalyst pellets. Before each experiment, charsamples were coated with gold to inhibit charging when themagnification is high.

Focused ion beam (FIB)/energy dispersive X-ray spectroscopy(EDX) analysis. The FIB in situ sample preparation and EDXelement mapping was performed in a FEI Strata 400 STEMDual Beam system, a fully digital Field Emission ScanningElectron Microscope (FE-SEM) equipped with focused ionbeam (FIB) technology and flip stage/STEM assembly. A briefdescription of the sample preparation is illustrated in Fig. 2.

Fig. 2 Process of sample preparation in focused ion beam (FIB)/energy dispersive X-ray spectroscopy (EDX) analysis.

Paper Green Chemistry


Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


A catalyst particle was selected and milled on both sides,leaving a thin layer along the equatorial plane of the particle.The thin catalyst layer was lifted out with an Omniprobemicro-manipulator and imaged under scanning transmissionelectron microscopy (STEM).

Thermal gravimetric analysis (TGA). Comparison of thermalproperties of char samples was performed in a Q-500 thermo-gravimetric analyzer from TA Instruments. Platinum cruciblesinstead of alumina crucibles were used for the oxidation ofchars, to prevent catalyst sintering with the crucibles. Thesamples were first dried at 120 °C for 30 min and heated up to900 °C in air flow of 60 ml min−1. Miura and Silveston47

studied non-catalytic gas–solid reactions using the tempera-ture-programmed reaction (TPR) technique. They showed arelative unreliability of rate parameters obtained based on TPRat only one heating rate. Therefore, in this study three differentheating rates (5/10/15 K min−1) were used in order to get accu-rate results in TPR data processing. All the TGA results pre-sented in this paper were normalized to char weight loss(catalyst weight excluded) to compare thermal with catalyticresidues.

Nitrogen adsorption. The surface area and pore size distri-bution of catalyst and coked catalyst/char mixtures were deter-mined in a Micrometitics ASAP 2020 Accelerated Surface Areaand Porosimetry System. The isotherms of N2 at 77 K wereobtained from physisorption. Before analysis, all the char/catalyst samples were degassed at 250 °C under vacuum for12 h to remove the surface contaminants. The pore size distri-bution of the char/catalyst samples was determined from theN2 adsorption isotherms at 77 K, using the Barrett–Joyner–Halenda (BJH) method.

Fourier transform infrared spectroscopy (FTIR). FTIRmeasurements to identify functional groups in char/coke wereperformed in a Nicolet MAGNA-IR 560 spectrometer with aDTGS detector, operated at 4 cm−1 resolution and 132 scans.

Raman spectroscopy. Raman spectra were obtained in aRenishaw 2000 Ramanscope, operated with a 514 nm laserexcitation source, at 1–25% power and 16–32 exposure times,to avoid detector saturation. For each sample, laser focus wasset to 40% to prevent local damage and 3 different positionswere analysed to verify the spectra.

Results and discussionChar/coke yields

Table 1 presents a brief literature review and experimentalresults of the current study for the yields of char/coke frompyrolysis of glucose, cellulose and pine. The effect of heatingrates and catalyst are investigated in this study, although otherfactors, such as temperature, also play an importantrole.17,31,32 The char/coke yields are significantly decreasingwhen fast heating rates are applied to the pyrolysis comparedwith slow heating rates. The lower char/coke yields from fastheating rates can be explained by the enhancement of thebond-scission reactions of the biomass to form tar fragments,

which, to some extent, limits the secondary pyrolysis (polymer-ization) of the volatiles.60 Moreover, char/coke yields vary sig-nificantly between different feedstocks. To be specific, in slowthermal pyrolysis, pine produces the highest yield of char/coke; whereas cellulose produces less char/coke than glucose.Similar results were also obtained by other researchers.58,61 Inslow catalytic pyrolysis, fast thermal pyrolysis and fast catalyticpyrolysis, char/coke yield from pyrolysis of pine outweighs thatfrom cellulose, while glucose produces the lowest yield of char/coke. Furthermore, in order to study the catalyst effect, char/coke yields between thermal pyrolysis and catalytic pyrolysisare compared. For all the pyrolysis conditions performed inthis study, catalytic pyrolysis produces more carbonaceousresidues than the corresponding thermal pyrolysis due to thepresence of catalyst. However, the reason for the increase ofcarbonaceous solid residues is not certain at this point. Inother words, the solid residues can be char (a non-catalyticproduct), catalytic coke, or both. In the following, experimentalresults are analysed, focusing on the formation of char andcoke.

Char/coke morphologies

In Fig. 3 the surface morphological characteristics of purebiomass feedstocks (a–c) and corresponding chars with orwithout catalyst, collected after slow thermal (d–f ), slow cata-lytic (g–i), fast thermal ( j–l), and fast catalytic (m–o), pyrolysisof the three types of biomass are investigated using SEM. Themorphology of char derived from the slow thermal pyrolysis ofglucose (d), cellulose (e), and pine (f), retains a similarity tothe original structure of the feedstock (a–c). In the case of slowcatalytic pyrolysis, the glucose char (g), is surrounding thecatalyst particles forming catalyst agglomerates due to lowtemperature primary pyrolysis and melting. The slow catalyticcellulose char (h), has a spiral type structure similar to therespective char produced from slow thermal pyrolysis, andappears to be formed as a separate phase to the catalyst par-ticles. Pine slow catalytic char (i), consist of irregular, largeparticles with slit-shaped surfaces surrounding the catalyst

Table 1 Comparison of char/coke yields between slow and fast, thermal andcatalytic pyrolysis

Char/coke yield wt% Glucose Cellulose Pine

Slow thermal (literature) 1848 32.6752

1048 18.7–24.449

1049 29.653

750 24–2754

20.151 29.5–32.655

21.551

Slow thermal (this work) 20.1 16.1 31.2Slow catalytic (literature) 2356 — 19.6–27.857

Slow catalytic (this work) 22.9 24.6 35.7Fast thermal (literature) 23.743 5.3558 22.616

9.8458

Fast thermal (this work) 7.6 8.7 14.9Fast catalytic (literature) 1528 1246 17.6–18.816

1446 14.259

Fast catalytic (this work) 8.7 9.3 15.5



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


particles. In fast thermal pyrolysis, the morphologies of charsfrom glucose ( j), cellulose (k), and pine (l), retain the originalstructure to some extent, but in a different way to the slowthermal pyrolysis, due to the different heating rates. A sig-nificant difference in morphology can be seen in pine fastthermal char (l), compared with the slow thermal pyrolysis.

The fast volatile release during fast pyrolysis enlarges theinternal cavities, resulting in a more open structure. Therefore,the macroporosity of chars increases with increasing heatingrate.17 In fast catalytic pyrolysis, there is no catalyst agglomera-tion occurring in the glucose case (m), reflecting the benefit offast heating rates to prevent excessive char formation. In thecase of fast catalytic pyrolysis of cellulose (n), and pine (o),char generally does not retain the structure of the originalfeedstock indicating that repolymerization may be the domi-nant mechanism for its formation.62 However, it shouldbe noted that in all the pyrolysis experiments, the formationof coke is considered to occur inside the catalyst pores due tocatalytically enhanced reactions of small molecule productsand intermediates in the secondary pyrolysis of volatile

matters.62 Thus, the formation of coke cannot be visualizedwith SEM.

The SEM results imply that the catalyst effect on pyrolysis isdifferent for the biomass feedstocks studied, as far as for-mation of char on the outer surface of catalyst particles is con-cerned. Evidently, in slow catalytic pyrolysis, glucose and pinefirst melt, wetting the catalyst surface, and then pyrolyze,leaving particle/char aggregates. It is reasonable to assumethat the catalyst quickly deactivates in these conditions, bylosing its accessibility due to the rapid surface coverage of aliquid pyrolysis intermediate. Therefore, only minimal differ-ences between the characteristics of the char/catalyst mixturesafter slow pyrolysis of glucose and pine should be expected.

Char/coke deposition on the outer surface and the equatorialplane of the catalyst

Fig. 4 is the EDX elemental mapping of coked catalysts frompyrolysis of different biomass feedstocks. Clear char “foot-prints” can be seen in the slow catalytic pyrolysis of glucoseand sawdust. The catalyst coupon from slow catalytic pyrolysisof glucose has a clear char edge, which is consistent with SEMpictures. It also has a large amount of char inside the pellet,which must be the result of melting inside catalyst macroporesat intermediate temperatures. The coupon from slow catalyticpyrolysis of pine also has a char edge, but it is much thinnerthan that of glucose. Inside the catalyst pellet, less carbon isobserved. The catalyst after slow catalytic pyrolysis of cellulosehas almost no edges and only some small red spots can beseen, which indicates the formation of carbon inside thezeolite. In all the fast catalytic pyrolysis cases, the catalyst haschar deposition both on the outer surface of the catalyst andinside the catalyst. Specifically, in fast catalytic pyrolysis ofglucose, there exists a thinner and more homogeneous carbonlayer on the surface of the catalyst pellet compared with slowcatalytic pyrolysis of glucose, which verifies the improvementin catalyst accessibility, concluded by the SEM analysis.Overall, the carbon mapping of the coked catalyst shows

Fig. 3 SEM morphology of biomass feedstocks: (a) glucose, (b) cellulose, (c)pine; and chars obtained by: (d) glucose slow thermal, (e) cellulose slowthermal, (f ) pine slow thermal, (g) glucose slow catalytic, (h) cellulose slow cata-lytic, (i) pine slow catalytic, ( j) glucose fast thermal, (k) cellulose fast thermal, (l)pine fast thermal, (m) glucose fast catalytic, (n) cellulose fast catalytic, (o) pinefast catalytic pyrolysis.

Fig. 4 FIB images of the equatorial plane of the catalyst particle after slow andfast catalytic pyrolysis. (a) glucose slow catalytic, (b) cellulose slow catalytic, (c)pine slow catalytic, (d) glucose fast catalytic, (e) cellulose fast catalytic, (f ) pinefast catalytic (green = Al, yellow = Si, red = carbon).



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


significant differences between slow catalytic pyrolysis ofglucose/sawdust and cellulose. The deposition of char on theouter surface of the catalyst might affect the accessibility of thecatalyst at the beginning of the pyrolysis. Thus, it can preventvolatile matters from accessing the zeolite and prohibit sec-ondary pyrolysis, where coke is produced.

Comparison of surface areas and pore size distributions

Fig. 5 shows the isotherms of N2 adsorption and corres-ponding pore size distribution plots for pure ZSM-5 catalystand the solid residuals obtained after slow and fast catalyticpyrolysis. The isotherm profile of the catalyst after glucoseslow catalytic pyrolysis exhibits very characteristic profiles.Microporosity (<2 nm) increases, which can be attributed tothe porosity of char formed, also evident in the desorptionprofiles that indicate ink bottle shaped pores. Moreover, thecatalyst macropores are significantly diminished, while thehysteresis loop during N2 desorption reflects the creation of amesoporous network in the form of ink bottle type pores(hysteresis type H2 in IUPAC classification), as shown in theisotherm and pore size distribution plots. In the isotherm ofthe solid residue after cellulose slow catalytic pyrolysis, thedecrease of the micropore volume reflects the enhanced cokeformation and micropore blocking due to coking. In the iso-therm of the solid residue after pine slow catalytic pyrolysis,the micropore volume increases slightly, but the macroporevolume does not have a significant change (shown also in thepore size distribution plot). In fast catalytic pyrolysis, however,

a clear loss of microporosity is observed in the isotherms forall the cases, which can be attributed to the formation of cokeinside the zeolite micropores. Table 2 shows a summary of themicropore area and total surface area of the catalyst and thecatalytic chars (catalyst included). The micropore area ofglucose and pine slow catalytic chars increased by 24% and11% respectively, compared to pure catalyst. The microporosityof cellulose slow catalytic chars decreased by 13%, a smalleramount compared with the microporosity of all the fast cata-lytic chars, which decreased by 52% (glucose), 38% (cellulose),and 22% (pine). The glucose slow catalytic char has thehighest total surface area (highest microporosity), whereas theglucose fast catalytic char has the lowest microporosity.

This result reveals that the catalyst porosity change is mostsignificant in the pyrolysis of glucose when high heating rateis introduced in the experiment. Combining the results from

Fig. 5 N2 adsorption/desorption isotherms measured at −196 °C and the pore size distributions of pure catalyst and deactivated catalyst produced from (a,d)glucose, (b,e) cellulose, (c,f ) pine, calculated from adsorption isotherms by using the BJH method.

Table 2 Micropore and total surface area of the fresh ZSM-5 catalyst andcoked catalyst/char mixtures after catalytic pyrolysis (m2 g−1)

Micropore area (<2 nm) Total surface area

Pure catalyst 98.98 124.3Glucose slow catalytic 122.4 162.1Glucose fast catalytic 47.55 68.33Cellulose slow catalytic 86.30 115.0Cellulose fast catalytic 60.93 90.05Pine slow catalytic 109.7 141.5Pine fast catalytic 76.94 99.50



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


SEM, FIB-STEM, BET and TPO (discussed in the next section),the significant reduction of macropores in the case of glucoseand pine slow catalytic char might be attributed to macroporeblocking by char formation on the catalyst outer surface.Macropore blockage is the reason for the minor catalyst effecton the properties of char and coke (e.g., oxygen content) inslow catalytic pyrolysis, which are responsible for TPO profilessimilar with those of slow pyrolysis char. Furthermore, thedecrease of the number of micropores after cellulose slow cata-lytic pyrolysis is attributed to coking, showing catalytic activity,even at slow heating rates. The smaller quantity of char pro-duced in this case is because of the catalytic enhancement ofdepolymerization of biomass to liquid and gas products.

Thermal gravimetric analysis of char/coke

In order to verify the effect of char/coke formation on the cata-lyst performance, temperature programmed oxidation (TPO) ofthe char/coke samples are performed in air using 3 differentheating rates.63 Before analysing the char/coke samples, thecatalyst effect on the oxidation itself was studied. TPOs of puregraphitic carbon and carbon/catalyst mixtures showed thatthere is no significant effect of the catalyst on the char/cokeoxidation. Fig. 6 presents differential weight loss profiles inTPO of the solid residues of thermal and catalytic (slow andfast) pyrolysis. TPO profiles are normalized, to exclude thecatalyst weight and the first derivative of weight loss (DTG)versus temperature is shown. In the comparison of slowthermal and slow catalytic char/coke from the three feedstocks,the DTG peak is at ∼550 °C for the glucose thermal char andat ∼500 °C for the cellulose and pine thermal char. It has beendocumented in the literature62 that glucose char is producedvia polymerization reactions, whereas pyrolytic decompositionis the main pathway for cellulose char formation. Evidently,these two different mechanisms are responsible for differentchar structures and/or char composition, which correspond-ingly react at different temperatures.19 The char formed frompine thermal pyrolysis exhibits a DTG maximum identical tothat of cellulose, whereas it also gives two additional DTGpeaks. The three DTG peaks of pine thermal pyrolysis charreflect char structures of different origins, corresponding to itshemicellulose, cellulose and lignin compounds.64

Interestingly, the addition of catalyst in slow pyrolysis con-ditions has only minor effect on the oxidation profiles of thesolid residues from glucose and pine. In the case of glucose,the DTG peak is shifted slightly to the left, indicating a slightcatalytic activity affecting the polymerization reactions, respon-sible for glucose char/coke formation. Overall and in agree-ment with the SEM images, the catalytic effect in slowpyrolysis conditions is minor, which can be attributed to thecatalyst surface coverage by the intermediate liquid and, there-fore, rapid catalyst deactivation. Small catalyst effect in slowpyrolysis of pine has also been observed in the literature,65

where the pyrolysis of wood was performed in TGA. Cellulosepyrolysis exhibits a rather different behaviour. The DTG peakmoves to the right, indicating a more significant catalyticactivity towards solid residues of lower oxygen content. This

Fig. 6 Oxidation of slow thermal (solid line) slow catalytic (dotted line), fastthermal (solid disc) and fast catalytic (empty disc) pyrolysis char/coke from (a)glucose, (b) cellulose and (c) pine in TGA in three heating rates (5 K min−1

(green), 10 K min−1 (blue), 15 K min−1 (red)).



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


indicates a formation of an oxygen poorer phase that can beattributed to coke formation or formation of gaseous inter-mediates that react with the cellulose residues forming loweroxygen content char. Moreover, the oxidation peak of theglucose (catalytic and thermal) char/coke sample is more sym-metrical than that of cellulose. In the latter case the hightemperature side of the DTG profile displays a faster decreasein reaction rate, which can be attributed to a lower reactionorder for the cellulose char/coke oxidation.66 TPO of fast cata-lytic pyrolysis samples show a broad DTG peak for glucose andmultiple DTG peaks for the cellulose and pine solid residues.The broad DTG peak in the case of glucose can be deconvo-luted into multiple peaks that represent different oxidationsteps, consistent with cellulose and pine. Generally, the firstchar/coke oxidation peak at 400–450 °C corresponds to charformation due to thermal decomposition. The second DTGpeak at 550–600 °C corresponds to catalytic coke formation.The wider shape of the DTG curves in the case of fast catalyticpyrolysis char/coke indicates a lower char/coke apparent oxi-dation reaction rate or a multiplicity of compounds, being oxi-dized at different rates. Compared with fast catalytic pyrolysis,fast thermal pyrolysis produces higher oxygen content char,leading to a lower temperature peak in the case of glucose andpine. Moreover, there is no secondary DTG peak at highertemperature in the fast thermal pyrolysis, reflecting the infeasi-bility of coke formation due to non-catalytic reactions. It isclearly evident that the catalytic effect is much stronger in fastpyrolysis experiments for each biomass feedstock comparedwith that in slow pyrolysis experiments.

In summary, many hypotheses derived from SEM, STEMand BET were verified with TPO results. Cellulose exhibits anentirely different behaviour, showing smaller catalyst surfacecoverage at slow pyrolysis conditions. TPO confirms this obser-vation, showing a clear contribution of the catalyst to the oxi-dation temperature of cellulose slow catalytic char. Thecatalyst/char samples after fast catalytic pyrolysis of all thefeedstocks show smaller amounts of char and intermediatesurface coverage of the catalyst, indicating a catalytic mechan-ism for coke formation on the catalyst surface.

Deconvolution of DTG thermogravimetric measurements

The TPO results of Fig. 6 were further analysed using model-based and statistical deconvolution methods. For this illus-tration the TPO experiments of the glucose char/coke and the

chars of the thermal pyrolysis of pine were analysed. In prin-ciple, glucose is an ideal candidate for a model-based analysisof its coke TPO, since there exists mainly one mechanism (i.e.,polymerization) for its formation.62 The random pore model(RPM) by Bhatia and Perlmutter67 was utilized for this analy-sis. This model considers the overlapping of pore surfaces andthe competing effects of pore growth during gasification, andthe destruction of the pores due to the coalescence of neigh-bouring pores by oxidation. The RPM models solid conversion,X, according to:

dXdt

¼ kRPMð1� XÞffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið1� ψÞ lnð1� XÞ

p; ð1Þ

where kRPM is the reaction rate constant and ψ a pore structurefactor of the unreacted sample:

kRPM ¼ kDB�ERT ; ð2Þ

ψ ¼ 4πLDð1� EDÞSD2 ; ð3Þ

with SD, LD and ED the pore surface area, pore length, andsolid porosity, respectively. Table 3 summarizes the parametersof the best fit of the RPM for the char/coke obtained fromglucose pyrolysis. As shown in Fig. 7, the RPM is capable ofrepresenting the TPO of chars from glucose slow pyrolysis. Thesmall ψ factors estimated should be attributed to the highsurface area of glucose chars and the presumably small porelength (indicated by the EDX elemental mapping of Fig. 4).

In agreement with the previous discussion about the lack ofcatalytic contribution in glucose slow pyrolysis, the RPM iscapable of modelling the DTG profile of the glucose slow cata-lytic char with the parameters of the thermal pyrolysis fit, byonly adjusting the RPM shape factor (Fig. 7(c)). This clearlyindicates that the only contribution of the catalyst at slowpyrolysis conditions is on the shape, surface area andpore length of the char formed, but not on its chemicalcomposition.

Fig. 8 presents the RPM fit of the glucose chars from fastpyrolysis. The quality of fit is again very good for the thermalpyrolysis char, whereas the catalytic pyrolysis char exhibits twoconvoluted DTG peaks that cannot be represented by the RPM(Fig. 8(b)). Fig. 8(c) presents the result of the RPM whenincluding the contribution of the fast thermal char, adjusting

Table 3 Kinetic parameters of the RPM for the glucose char samples during TPO at three heating rates (5, 10 and 15 K min−1)

ko (s−1) E ( J mol−1) ψ R2

Glucose slow thermal 1.870 × 106 1.361 × 105 0.268 9.918 × 10−1

Glucose slow catalytic char fraction = 0a 8.302 × 105 1.292 × 105 0 9.935 × 10−1

Glucose slow catalytic char fraction = 1b 1.870 × 106 1.361 × 105 0.193 9.621 × 10−1

Glucose fast thermal 3.665 × 102 0.768 × 105 0.898 9.956 × 10−1

Glucose fast catalytic char fraction = 0c 8.066 5.932 × 105 0 9.500 × 10−1

Glucose fast catalytic char fraction = 0.3d 6.525 × 101 0.755 × 105 0 9.968 × 10−1

a Assumes a dominant contribution of catalytic reactions. b Assumes a negligible contribution of catalytic reactions. c Assumes only one DTGpeak. d Includes the DTG contribution of the thermal char.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


only the RPM shape factor, to account for the effect of the cata-lyst on shape and surface area differences. For a more detaileddiscussion on the application of the RPM to multicomponentsolids refer to the works by Miura et al.68 and Fermoso et al.63

The inclusion of the RPM of the fast thermal char of glucose iscapable of deconvoluting the fast catalytic DTG, identifyingtwo solid residues of different origins. Therefore, it is reason-able to assume that, in parallel to the thermal polymerizationreactions (char), there is a significant contribution from cata-lytically enhanced reactions yielding solid residues of differentcomposition (coke). This is an interesting finding, because itindicates that we can deconvolute experimentally the extent towhich the catalyst contributes to catalytic pyrolysis reactions.The RPM predicts that 30 wt% of the solid residue after fastcatalytic pyrolysis of glucose is fast thermal char. It should benoted that application of the RPM is superior to statisticaldeconvolution methods, as kinetic constants are extracted andthe deconvolution is performed with the same constants fordifferent TPO heating rates. This is valid for as long as the fitsare of statistical significance, which is the case when fitting asmall number of DTG peaks (small number of solid com-ponents being oxidized).

However, chars from thermal pyrolysis of pine show fourinflection points in their DTG curves, with additional peaks

measurable after catalytic pyrolysis, due to the formation ofcatalytic coke. Therefore, a more conventional statisticaldeconvolution approach was employed; using iterative least-square fits of exponentially modified Gaussian (EMG) dis-tribution function to the pine DTG signals. Application ofEMG is very common in deconvolution of chromatographicpeaks69 and was selected for this analysis due to its ability torepresent fronted (or tailed) distributions. Fronted distri-butions are evident in all the TPOs of Fig. 6, particularly in theTPO of cellulose. Fig. 9 shows the results of the deconvolutionfor the slow and fast pyrolysis chars from pine. For each case,the four deconvoluted peaks represent different chars from thethree main components in pine (hemicellulose, cellulose andlignin), which means that these precursors participatein different reactions (or same type of reactions, but withdifferent precursors). The last small peak at T > 500 °C isattributed to the oxidation of extractives and other hetero-geneous components of pine and should be considered part ofthe lignin char. These high temperature peaks were evident inTPO of the original pine (not shown here); therefore, they arenot a pyrolysis product. By comparing the deconvolutionresults with those from TPO of pine and cellulose, it is reason-able to assume that peaks #1, #2, #3 represent chars fromhemicellulose, lignin and cellulose, respectively; which is

Fig. 7 Application of the RPM on the TPO of chars from: (a) glucose slowthermal pyrolysis; (b) glucose slow catalytic pyrolysis; and (c) glucose slow cataly-tic pyrolysis using the RPM parameters of (a).

Fig. 8 Application of the RPM on the TPO of chars from: (a) glucose fastthermal pyrolysis; (b) glucose fast catalytic pyrolysis; and (c) glucose fast catalyticpyrolysis using the combined RPM models of (a) and (b).



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


consistent with the location of peaks of the three componentsin the deconvolution of pyrolysis of pine.70 In slow pyrolysis ofpine, the ratio of hemicellulose (peak #1) and cellulose (peak#3) chars is ∼2, which does not agree well with the initial pinecomposition (hemicellulose: 23–32%; cellulose: 38–50%;lignin: 15–25%). Moreover, in fast pyrolysis of pine, the frac-tion of char from hemicellulose increased significantly,whereas the fraction of char from cellulose decreased, whichreflects that fast heating rates favour the formation of hemi-cellulose char. The ratio of cellulose to lignin char is about6/10 in all cases, which is in good agreement with their initialfractions (∼40 wt% and ∼20 wt%, respectively) times their charyield (∼15 wt% and ∼50 wt%,71 respectively). Fig. 9 shows thatin the research of catalytic options for the minimization ofchar formation in biomass pyrolysis, we need to focus on thereaction pathways of hemicellulose, since it appears to be theleast affected by the fast heating rates. The char fractions oflignin and cellulose (peaks #2 and #3, respectively) aredecreasing with increasing pyrolysis heating rates.

FTIR and Raman analysis of char and coke

The char/coke chemical composition with respect to theirbonding groups was studied in FTIR. Fig. 10 shows the spec-trum 1500–1800 cm−1 of the glucose char/coke, produced atdifferent heating rates with and without catalyst. The CvCstretch (1620–1680 cm−1) appears in the char/coke producedfrom all the pyrolysis experiments. Interestingly, the char/coke

after fast catalytic pyrolysis contains minor amounts of carbo-nyl (CvO) groups, which reflects the efficiency of oxygenremoval in fast catalytic pyrolysis and the dominance of adifferent mechanism producing the catalytic coke. Consistentwith the TPO results, the fast heating rates of the spouted bedreactor enhance the catalytic effect in pyrolysis, thus favouringcatalytic coke formation. The diminished CvO in fast catalyticpyrolysis compared to that in slow catalytic pyrolysis revealsthat CvO is contained mostly in char instead of coke.

Fig. 10 FTIR spectra of glucose chars from slow thermal pyrolysis, slow catalyticpyrolysis, fast thermal pyrolysis and fast catalytic pyrolysis.

Fig. 9 Exponentially modified Gaussian peak deconvolution of chars obtained from: (a–c) pine slow thermal pyrolysis, 5/10/15 K min−1; and (d–f ) pine fastthermal pyrolysis, 5/10/15 K min−1.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


The spectrum at 2500–3600 cm−1 (not shown here), exhibitsa similar broad peak for all the chars studied, showing thatO–H bonds (free hydroxyl bonded, hydrogen bonded and incarboxyl group) exist consistently. The FTIR analysis showsthat coke and char are not graphitic carbon, and should not betreated as such in the modelling of pyrolysis mechanisms.72,73

Further verification of this observation was performed in TPOexperiments with mixtures of graphitic carbon and coked cata-lysts (not shown here), in which the TPO shows one clear DTGpeak for the carbon (at ∼800 °C), well-separated from those ofchar and coke.

In-depth study of the char/coke composition was also per-formed in Raman. Fig. 11 shows the Raman spectra of thechar/coke from glucose pyrolysis. The deconvolution of thespectra is based on observation of the main peaks, shoulders,valleys, and tails. Thus for the slow pyrolysis, the spectra hasbeen deconvoluted into 6 Gaussian peaks: #1, ca. 1700 cm−1

(D2 band, corresponding to carbonyl groups); #2, ca. 1600 cm−1

(G band, corresponding to well-structured aromatic rings); #3,1550–1570 cm−1 (G′ band, corresponding to coexistence of awell-structured and a not well-structured carbons); #4,ca. 1450 cm−1 (D3 band, corresponding to structural defects ofaromatic clusters); #5, 1360–1370 cm−1 (D band, corres-ponding to not well-structured aromatics); and #6,ca. 1270 cm−1 (corresponding to C–H vibrations). For the fastpyrolysis, 5 peaks were sufficient for fitting the spectra: #1 (D2

band); #2 (G band); #3 (G′ band); #4 (D band); and #5 (C–Hvibrations). Detailed description of the interpretation of eachband identified by Raman can be found elsewhere.74–76 Com-parison of the Raman spectra of the slow and fast pyrolysisshows that the second broad peaks (D3 + D + C − H in slowpyrolysis; D + C − H in fast pyrolysis) decrease significantlywhen fast heating rates are applied in pyrolysis. The fast cata-lytic pyrolysis char/coke has the lowest second peak (0.21 inarea fraction) among the four. It appears that the formation ofdefect aromatics (mainly D band and D3 band) is not favouredby the fast heating rates. Combining the results from TPO andRaman implies that the second peak is characteristic of thechar formed. This is in agreement with Sheng,75 who claimedthat as the defect bands increase, the ordering of the char

decreases and the char becomes more reactive. Moreover, thesum of areas of the G and G′ bands increases for the fast pyro-lysis chars and is highest for the fast catalytic pyrolysis char/coke. The fraction of the areas of the G and G′ bands exhibitsimilar trends to those observed from TPO, indicating anincrease in catalytic coke. Furthermore, the total peak area ofthe defect band changes only slightly between slow thermaland slow catalytic pyrolysis, compared with that between fastthermal and fast catalytic pyrolysis, which is very consistent tothe TPO results (lack of significant catalytic activity at slowpyrolysis conditions). It is not reasonable to define coke asthe well-structured aromatics and char as defect aromatics,since G and D bands coexist in all the Raman spectra, but thechanges of their area fractions are indicative of the sameobservations in TPO, STEM/FIB and FTIR.

Reaction pathways for coke and char formation

Dominant mechanisms for the formation of coke and char infast catalytic pyrolysis of glucose28 and fast thermal pyrolysisof cellulose41 have been postulated in the literature. Insummary, the formation of char and coke can be described asthe result of polymerization, dehydration, decarboxylation,decarbonylation of three origins: (1) anhydrosugars, (2) furaniccompounds, (3) fragmented oxygenates and/or olefins. Accord-ing to Carlson et al.,28 although these reactions are promotedby acid catalysts, they can still occur in the absence of catalyst.Thus, the dominant mechanisms in catalytic pyrolysis aresimilar (and co-exist) with those of thermal pyrolysis and themechanism of formation of coke (defined here as a catalyticproduct) should be similar to that of char (defined as athermal product).

Analysis of the nature of coke deposited on HZSM-5 inpyrolysis of bio-oil (produced from pine) performed by Valleet al.76 and Ibáñez et al.77 showed that catalyst deactivation iscaused by coke deposition. They identified coke of two origins:thermal and catalytic; and observed catalytic coke to be de-posited mainly inside the zeolite crystal channels; whereasthermal coke deposited on the outer surface of the catalyst(meso-, macro-pores). This is in agreement with Cheng andHuber,13 who postulated that coke from furan pyrolysis

Fig. 11 Raman spectra of glucose char from (a) slow thermal, (b) slow catalytic, (c) fast thermal and (d) fast catalytic pyrolysis. The calculated area fractions ofdeconvoluted peaks (#1–6 for slow pyrolysis, #1–5 for fast pyrolysis) in each case are shown.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


comprises polycyclic oxygenates. The small fraction of solublecoke that they were able to extract from the organics retainedinside the catalyst had molecular weights as high as 204 gmol−1, mostly containing aromatics rings and carbonyl endgroups. Pure polyaromatics were also identified. In summary,what we know from the relevant literature and this study isthat char and coke are not graphite-type carbons. Char showslow TPO temperature peaks, significant amounts of carbonylgroups and a large number of Raman defects. Coke exhibitsmuch higher TPO peaks, shows smaller amounts of carbonylgroups and significantly smaller Raman defects. Of course,coke and char co-exist in the solid residue of fast catalyticpyrolysis, but as shown in Fig. 8, coke is the dominant carbonform after fast catalytic pyrolysis (ratio of 7/3). Brewer et al.78

performed 13C-NMR analysis on the chars obtained from fastand slow thermal pyrolysis of switchgrass, showing that aro-matic clusters of 7–8 rings terminated by carbonyl andhydroxyl groups are the representative composition of thermalchar (a structure similar to the one shown in Fig. 12). For-mation of char is typically observed in concert with CO andCO2, which can be explained by the decarbonylation of fura-nics postulated by Huber and co-workers.13,79

Another interesting observation was reported by Mettleret al.,62 who compared the char yield from pyrolysis of glucose-based carbohydrates and the fraction of carbonyl groupspresent in the volatile products and observed a very strongrelation. They postulated that aldol condensation chemistry issignificant in thermal pyrolysis char formation. In parallel tothe aforementioned studies, there is a wealth of informationand published work on the nature and possible mechanisms

of coke formation in the catalytic cracking of gasoil and otherhydrocarbons.80–83 Hydrogen transfer and carbenium ionchemistry is shown to be dominant in the mechanisms of cata-lytic coke formation, having small olefins and single ringaromatics as its origin. The production of aromatics isestablished in catalytic pyrolysis of biomass, while smallolefins are mostly postulated as responsible for their for-mation via the hydrocarbon pool mechanism proposed byHuber and coworkers.28 Therefore, it is reasonable to assumethat the mechanisms accepted in catalytic pyrolysis of hydro-carbons may well be relevant to the catalytic pyrolysis ofbiomass. In that view, there should be a competition betweenthe reactions forming aromatics and the reactions responsiblefor catalytic coke, given the hydrogen-poor environment in thehydrocarbon pool and the small evidence of larger olefinsobserved in biomass catalytic pyrolysis experiments. On theother hand, char formation reactions can be surmised toproceed via reactions of oxygenates (aldol and Diels–Alder),but given the FTIR and Raman evidence they must lead tolarge polyaromatic rings terminated with carbonyl andhydroxyl groups. One other aspect to consider is the steric con-straints of the ZSM-5 catalyst, which impose a requirement forthe catalytic coke to be smaller (in its number of aromaticrings) than the char formed. The largest polyaromatic that canbe accommodated and trapped inside the channel intersec-tions of MFI zeolites is methylpyrene (shown with molecularmodelling by Guisnet et al.84). The formation of polyaromaticrings is accepted to proceed via a series of alkylation andhydrogen transfer steps.84 Marin and co-workers83 show thatcoke formation proceeds via hydrogen transfer, alkylation and

Fig. 12 Possible reaction pathways for coke and char formation. (a) RA82 – toluene self-alkylation via RB.1 (alkylation), RB.2 (dehydrogenative coupling), RB.3 (iso-merization), and RB.4 (hydrogen transfer and repetition of RB.1–RB.5); (b) RB83 – coke formation via RA.1 (alkylation on the nucleus with carbenium ions), RA.2 (sidealkylation and isomerization), RA.3 (cyclization), and RA.4 (repetition of RA.2, RA.3); (c) RC – char formation from furfural via RC.1 (Diels–Alder with propylene), andaldol condensations (RC.2–RC.7); RD – Diels Alder cycloadditions of C-5 and C-6 anhydrosugars via RD.1 (Diels Alder self- or hetero-cycloaddition), RD.2 (Diels Aldercycloaddition) of the products of RD.1 with the original anhydrosugars, and RD.3 (repetition of RD.2, followed by enol–keto tautomerization to produce carbonylending groups and condensation to fused polyaromatic rings terminated by carbonyl and hydroxyl groups). MPD is the minimum projection diameter of each mole-cule; KD is the kinetic diameter reported by Jae et al.44



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


ring closing, with toluene and propylene as possible cokeprecursors. Cheng and Huber,79 accept a mechanism that pro-duces allene and/or methylacetylene (formed by decarbonyla-tion of furanics), which undergo oligomerization to form aseries of olefins and single ring aromatics. Accepting thethermodynamic stability of polyaromatics, it is straight-forward to envision that hydrogen transfer chemistry is rele-vant to coke formation in biomass catalytic pyrolysis.

On the basis of the aforementioned possible reaction path-ways for the formation of coke and char in the catalytic pyro-lysis of biomass, we suggest representative reaction schemesthat are relevant to the pyrolysis of glucose and cellulose. Fourreaction pathways (noted as RA, RB, RC, and RD) for coke andchar formation are illustrated in Fig. 12, together with theminimum projection diameter (MPD) for each componentcalculated in ChemAxon Marvin.85 Toluene and furfural arechosen as the starting molecules for showing the mechanisms,since their presence as biomass (glucose) pyrolysis products isconfirmed in our liquid product analysis and the study byCarlson et al.28 (1E,3E)-Penta-1,3-diene-1,3,5-triol and (2E,4Z)-hepta-2,4-diene-1,4,7-triol are used as char precursors, as theyare (according to the glucose/cellulose pyrolysis mechanismproposed by Vinu and Broadbelt73) among the postulatedanhydrosugars responsible for carbon formation (char istracked as carbon in that model). The reactions of Fig. 12 arein no way exhaustive and can be considered only as representa-tive of the mechanisms involved in coke and char formation.Moreover, there are alternative pathways that can co-exist (forinstance, dehydration and tautomerization of the alcoholsproduct in pathways RC and RD), which are omitted. Thefocus of this analysis is to explore the likely mechanisms thatcan produce fused or linear polyaromatic rings with or withoutoxygen containing ending groups.

Mechanisms RA and RB are similar to those proposed inCerqueira et al.82 and Quintana-Solórzano et al.,83 respectively,and present common coke formation reactions (with hydro-carbon precursors). According to Guisnet & Magnoux,80 thecomposition of high temperature (400–600 °C) coke is practi-cally independent of the reactant and is considered to com-prise polyaromatics, formed not only by condensation andrearrangement steps but also via various hydrogen transfersteps on acid catalysts. Methylpyrene is a major coke com-ponent with a molecular size intermediate between that of theMFI supercages and of its pore apertures. Therefore, the inter-mediate size aromatic rings derived via pathways RA and RBare sterically blocked inside the supercages of the ZSM-5,while they are thermodynamically resistant to cracking.Further alkylation and aromatization of the representativefinal polyaromatic of the RA mechanism is not feasible, as it issterically constrained by the size of the ZSM-5 cavities. Firstet al.86 report a cavity size of 7.5 Å for calcined ZSM-5,87 whichcan tightly accommodate the pyrene molecule (∼9 Å), giventhat the ZSM-5 cage and pyrene are not spheres.

Methyl- or ethyl-anthracene (the final product of pathwayRB, ∼8 Å diameter) cannot escape the ZSM-5 pore (with porelimiting diameter of 4.5 Å and maximum crystallographic

diameter of ∼6 Å). Therefore, polyaromatics once formedinside the zeolite cage are bound to stay in it and undergofurther aromatization leading to ring structures with amaximum of ca. 4 rings. Given their relatively larger residencetime inside the zeolite (and the reactor) these polyaromaticsmight partially condense to heavier forms, leading to insolublecoke that is difficult to analyse. In summary, we postulate thatthe dominant coke formation mechanisms of catalytic crack-ing of hydrocarbons are relevant in biomass (glucose in thiscase) pyrolysis as well. The largest coke form possible shouldbe defined by the ZSM-5 steric constraints, through a more in-depth analysis is required, such as the work by First et al.,86

describing the interactions between the guest and host atomsthat capture the particular shape of the molecule subject to itspossible rotations. Molecular projection (MPD) and kineticdiameters (KD) are of little use in this regard.

Nonetheless, the coke from glucose pyrolysis is quantitat-ively much larger than what one would anticipate from hydro-carbons. Hence, there must be a significant contribution fromoxygenates in the formation of coke (besides, their obviouscontribution to char, discussed later). According to the resultsobtained in the char/coke analysis of this study and the postu-lated mechanisms for coke formation discussed previously,the most important mechanisms for coke formation withorigins in oxygenated compounds are Diels–Alder and aldolreactions. A scheme involving these reactions is depicted inpathway RC of Fig. 12. Furfural acting as a diene can react witha small olefin (from the hydrocarbon pool) yielding tolu-aldehydes, which can undergo aldol condensation reactions(inside or outside the ZSM-5 pore) to form polyaromatics ofthe form proposed by Brewer et al.78 The aldol is indeed aninteresting pathway, because it can lead to polyaromatics withcarbonyl and hydroxyl end groups which are clearly evident inthe FTIR and Raman analyses. It can proceed inside the cata-lyst pore forming oxo-aromatics (Valle et al.76) or continueoutside of the zeolite, blocking both the micropores and themacropores of the catalyst (as observed in the BET analyses,Table 2).

Finally, in regards to char formation (besides the aforemen-tioned aldol route), we look at the molecular structure of(1E,3E)-penta-1,3-diene-1,3,5-triol and (2E,4Z)-hepta-2,4-diene-1,4,7-triol, proposed by Vinu and Broadbelt73 as major charprecursors. These compounds (and other similar obtained byenol–keto tautomerizations) can act as dienes and dienophiles,which points strongly towards a Diels–Alder route for char for-mation. In Fig. 12 we enumerate all the possible products oftheir Diels–Alder reactions, omitting the possible tautomeriza-tion and condensation steps (which would not significantlyaffect the aromatic structure of these intermediates). The firstproducts of reaction RD.1 are all significantly larger (>9 ÅMPD) than the ZSM-5 pores or cavities. Hence, their formationcan only occur in the catalyst macropores and on the catalystsurface. In the absence of catalyst they can lead to structuressimilar to those proposed by Brewer et al.78 and it is likely thatthey are promoted at lower temperatures and slow heatingrates, as indicated by the larger transmittance intensity of the



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


FTIR carbonyl group in the glucose char from slow thermalpyrolysis. If we take into account the relatively narrow TPODTG curves from slow thermal and slow catalytic pyrolysis ofglucose and cellulose and the wide defect bands observed inthe Raman spectra of the chars from slow thermal and slowcatalytic pyrolysis of glucose, it is reasonable to assume thatchar is composed of a few oxygen-containing polyaromatics,that are preferably formed at lower temperatures (or their kine-tics are slower). Knowledge of the type (char or coke), location(macropores or micropores), and amount (weight fraction) ofthe carbonaceous residues during pyrolysis can shed light tothe extent and significance of the two domains shown inFig. 12 and help to extend the excellent work by Vinu andBroadbelt,73 Cho et al.,22,88 Lin et al.,41 Cai et al.,89 Mettleret al.,43,62 and Rangarajan et al.,90 to account for thermal andcatalytic reactions leading to char and coke.

Conclusions

In summary, the properties and the characteristics of the charand coke derived from thermal and catalytic pyrolysis aredifferent. The formation of char and coke strongly depends onthe biomass source and also on the pyrolysis conditions (e.g.,heating rates). This study reveals that when glucose and pineare used as biomass feed in slow catalytic pyrolysis, catalyst de-activation due to formation of char and the corresponding lossof accessibility (surface coverage and macropore blocking)becomes dominant. On the other hand, in the cellulose slowcatalytic pyrolysis and all the fast catalytic pyrolysis experi-ments, the formation of coke is attributed to catalyst micro-pore blocking. In fast catalytic pyrolysis of glucose and pine,formation of catalytic coke proceeds in parallel to the for-mation of thermal char. We identified the char to coke ratio tobe 3/7 (mass basis) for the case of glucose and it is theoreti-cally feasible to do so for other biomass model compounds.The following conclusions were drawn in this work:

• In slow thermal pyrolysis, TPO results show that, forglucose and pine, similar oxidation reactivities of the char/coke products are obtained, compared to the correspondingslow catalytic char/coke. For cellulose, the oxygen content ofthe char in slow thermal pyrolysis is higher than that in slowcatalytic pyrolysis. Glucose char is likely produced via thermalDiels–Alder and aldol reactions, while the origin of cellulosechar also includes pyrolytic decomposition. Analysis of theTPO DTG peak shapes shows that cellulose char oxidation hasa lower reaction order than pine and glucose; indicating theformation of a very narrow distribution of hydrocarbons and/or oxygenates.

• In slow catalytic pyrolysis, SEM, STEM and BET resultsconfirm that macropore blocking occurs for glucose and pine,leading to minor accessibility of the volatiles to the catalyst,while micropore blocking occurs in the case of cellulose due tocoking. Similar reaction orders for the char/coke oxidation areobserved in TPO, compared to slow thermal pyrolysis. Appli-cation of the random pore model shows that, at slow catalytic

pyrolysis of glucose, catalyst contributes only on the shape,surface area and pore length of the char formed, but not on itschemical composition.

• In fast thermal pyrolysis, with TPO DTG peaks shifting tolower temperatures, glucose and pine char/coke have higheroxygen content than the corresponding fast catalytic ones.Comparison of the TPO results of fast thermal pyrolysis charto those from fast catalytic pyrolysis shows that there is nocoke formation in fast thermal pyrolysis. Hemicellulose is pro-posed to be the most significant char formation precursor infast thermal pyrolysis of pine, on the basis of TPO deconvolu-tion results.

• In fast catalytic pyrolysis, as shown by BET, microporeblocking occurs for all the feedstocks, reflecting good accessi-bility to the zeolite, and thus a stronger overall catalyst effectthan that of slow catalytic pyrolysis. The wide shapes of theTPO curves indicate that catalytic coke corresponds to a rangeof hydrocarbons or oxygenates, formed via different mecha-nisms. Common catalytic coke formation mechanisms are pro-posed for the glucose coke formation (on the basis of FTIRand Raman observations of diminishing carbonyl groups anddecreasing defect bands, respectively), which are assumed toproceed in parallel to aldol and Diels–Alder reactions ofoxygen containing biomass intermediates. The catalytic cokereactions are of the same type and antagonistic in nature tothe reactions leading to aromatics, resulting in consumptionof single-ring aromatics for the production of coke inside thezeolite cage.

Acknowledgements

This material is based upon work partly supported by theUCONN Faculty Large Grant. The authors thank Dr RogerRistau for help with FIB-STEM analysis. The authors alsothank W. R. Grace & Co. for supplying the ZSM-5 catalysts usedin this work. GMB thanks Prof. Christodoulos A. Floudas foruseful discussions on the use of ZEOMICS and for modellingthe calcined ZSM-5.

References

1 D. M. Alonso, J. Q. Bond and J. A. Dumesic, Green Chem.,2010, 12, 1493–1513.

2 H. Cui and J. R. Grace, Bioresour. Technol., 2008, 99, 4008–4020.

3 A. V. Bridgwater and G. V. C. Peacocke, Renewable Sustain-able Energy Rev., 2000, 4, 1–73.

4 A. C. Kokossis and A. Yang, Comput. Chem. Eng., 2010, 34,1397–1405.

5 J. Adam, E. Antonakou, A. Lappas, M. Stöcker,M. H. Nilsen, A. Bouzga, J. E. Hustad and G. Øye, Micro-porous Mesoporous Mater., 2006, 96, 93–101.

6 G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006,106, 4044–4098.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


7 G. W. Huber, National Science Foundation – Division ofChemical, Bioengineering, Environmental, and TransportSystems, 2008.

8 S. N. Naik, V. V. Goud, P. K. Rout and A. K. Dalai, RenewableSustainable Energy Rev., 2010, 14, 578–597.

9 D. Mohan, C. U. Pittman and P. H. Steele, Energy Fuels,2006, 20, 848–889.

10 F. A. Agblevor, O. Mante, N. Abdoulmoumine andR. McClung, Energy Fuels, 2010, 24, 4087–4089.

11 A. V. Bridgwater, D. Meier and D. Radlein, Org. Geochem.,1999, 30, 1479–1493.

12 A. J. Foster, J. Jae, Y.-T. Cheng, G. W. Huber and R. F. Lobo,Appl. Catal., A, 2012, 423–424, 154–161.

13 Y.-T. Cheng and G. W. Huber, ACS Catal., 2011, 1, 611–628.14 J. Jae, G. A. Tompsett, Y. Lin, T. R. Carlson, J. Shen,

T. Zhang, B. Yang, C. E. Wyman, W. C. Conner andG. W. Huber, Energy Environ. Sci., 2010, 3, 358–365.

15 T. R. Carlson, Y.-T. Cheng, J. Jae and G. W. Huber, EnergyEnviron. Sci., 2011, 4, 145.

16 A. Atutxa, R. Aguado, A. G. Gayubo, O. Martinand andJ. Bilbao, Energy Fuels, 2005, 19, 765–774.

17 M. Guerrero, M. P. Ruiz, M. U. Alzueta, A. R. Bilbao andA. Millera, J. Anal. Appl. Pyrolysis, 2005, 74, 307–314.

18 J. Cho, J. M. Davis and G. W. Huber, ChemSusChem, 2010,3, 1162–1165.

19 G. Elordi, M. Olazar, G. Lopez, P. Castaño and J. Bilbao,Appl. Catal., B, 2011, 102, 224–231.

20 G. W. Huber and A. Corma, Angew. Chem., Int. Ed., 2007,46, 7184–7201.

21 K. S. Triantafyllidis, E. F. Iliopoulou, E. V. Antonakou,A. A. Lappas, H. Wang and T. J. Pinnavaia, MicroporousMesoporous Mater., 2007, 99, 132–139.

22 J. Cho, J. M. Davis and G. W. Huber, ChemSusChem, 2010,3, 1162–1165.

23 D. J. Nowakowski, a. V. Bridgwater, D. C. Elliott, D. Meierand P. de Wild, J. Anal. Appl. Pyrolysis, 2010, 88, 53–72.

24 M. Al-Haddad, E. Rendek, J.-P. Corriou and G. Mauviel,Energy Fuels, 2010, 24, 4689–4692.

25 M. J. Antal and M. Grønli, Ind. Eng. Chem. Res., 2003, 42,1619–1640.

26 C. Di Blasi, Prog. Energy Combust. Sci., 2008, 34, 47–90.27 H. Zhang, T. R. Carlson, R. Xiao and G. W. Huber, Green

Chem., 2012, 14, 98.28 T. R. Carlson, J. Jae, Y.-C. Lin, G. A. Tompsett and

G. W. Huber, J. Catal., 2010, 270, 110–124.29 A. Aho, N. Kumar, K. Eranen, T. Salmi, M. Hupa and

D. Murzin, Fuel, 2008, 87, 2493–2501.30 M. Asadullah, S. Zhang, Z. Min, P. Yimsiri and C.-Z. Li,

Bioresour. Technol., 2010, 101, 7935–7943.31 M. S. Hasan Khan Tushar, N. Mahinpey, A. Khan,

H. Ibrahim, P. Kumar and R. Idem, Biomass Bioenergy,2012, 37, 97–105.

32 C. Di Blasi, Prog. Energy Combust. Sci., 2009, 35, 121–140.

33 D. Ferdous, A. K. Dalai, S. K. Bej and R. W. Thring, EnergyFuels, 2002, 16, 1405–1412.

34 D. J. Nowakowski, A. V. Bridgwater, D. C. Elliott,D. Meier and P. de Wild, J. Anal. Appl. Pyrolysis, 2010,88, 53–72.

35 R. K. Sharma, J. B. Wooten, V. L. Baliga, X. Lin,W. Geoffrey Chan and M. R. Hajaligol, Fuel, 2004, 83,1469–1482.

36 M. Olazar, R. Aguado, M. J. S. José and J. Bilbao, J. Chem.Technol. Biotechnol., 2001, 76, 469–476.

37 R. Aguado, M. Olazar, M. J. S. José, G. Aguirre andJ. Bilbao, Ind. Eng. Chem. Res., 2000, 39, 1925–1933.

38 M. Olazar, R. Aguado, J. Bilbao and A. Barona, AIChE J.,2000, 46, 1025–1033.

39 S. Du and G. M. Bollas, Fuel, 2013, in review.40 A. Aho, N. Kumar, A. V. Lashkul, K. Eränen, M. Ziolek,

P. Decyk, T. Salmi, B. Holmbom, M. Hupa andD. Y. Murzin, Fuel, 2010, 89, 1992–2000.

41 Y.-C. Lin, J. Cho, G. A. Tompsett, P. R. Westmoreland andG. W. Huber, J. Phys. Chem. C, 2009, 113, 20097–20107.

42 P. R. Patwardhan, J. A. Satrio, R. C. Brown andB. H. Shanks, Bioresour. Technol., 2010, 101, 4646–4655.

43 M. S. Mettler, A. D. Paulsen, D. G. Vlachos andP. J. Dauenhauer, Green Chem., 2012, 14, 1284.

44 J. Jae, G. A. Tompsett, A. J. Foster, K. D. Hammond,S. M. Auerbach, R. F. Lobo and G. W. Huber, J. Catal., 2011,279, 257–268.

45 E. M. Sulman, V. V. Alferov, Y. Y. Kosivtsov, A. I. Sidorov,O. S. Misnikov, A. E. Afanasiev, N. Kumar, D. Kubicka,J. Agullo, T. Salmi and D. Y. Murzin, Chem.–Eng. J., 2007,134, 162–167.

46 T. R. Carlson, G. A. Tompsett, W. C. Conner andG. W. Huber, Top. Catal., 2009, 52, 241–252.

47 K. Miura and P. L. Silveston, Energy Fuels, 1989, 3, 243–249.

48 T. E. Mcgrath, W. G. Chan and R. Hajaligol, J. Anal. Appl.Pyrolysis, 2003, 66, 51–70.

49 P. T. Williams and S. Besler, Renewable Energy, 1996, 7,233–250.

50 H. Yang, R. Yan, H. Chen, C. Zheng, D. H. Lee andD. T. Liang, Energy Fuels, 2006, 20, 388–393.

51 A. Aho, N. Kumar, K. Eränen, B. Holmbom, M. Hupa,T. Salmi and D. Y. Murzin, Int. J. Mol. Sci., 2008, 9, 1665–1675.

52 A. Zabaniotou, O. Ioannidou, E. Antonakou and A. Lappas,Int. J. Hydrogen Energy, 2008, 33, 2433–2444.

53 E. Apaydın-Varol and A. E. Pütün, J. Anal. Appl. Pyrolysis,2012, 98, 29–36.

54 T. Pattanotai, H. Watanabe and K. Okazaki, Fuel, 2013, 104,468–475.

55 A. Zhurinsh, J. Zandersons and G. Dobele, J. Anal. Appl.Pyrolysis, 2005, 74, 439–444.

56 H. Zhang, Y.-T. Cheng, T. P. Vispute, R. Xiao andG. W. Huber, Energy Environ. Sci., 2011, 4, 2297.

57 Z. Wang, F. Wang, J. Cao and J. Wang, Fuel Process.Technol., 2010, 91, 942–950.

58 P. R. Patwardhan, J. A. Satrio, R. C. Brown andB. H. Shanks, J. Anal. Appl. Pyrolysis, 2009, 86, 323–330.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


59 A. Aho, N. Kumar, K. Eränen, T. Salmi, M. Hupa andD. Y. Murzin, Fuel, 2008, 87, 2493–2501.

60 J. Shen, X.-S. Wang, M. Garcia-Perez, D. Mourant,M. J. Rhodes and C.-Z. Li, Fuel, 2009, 88, 1810–1817.

61 M. S. Mettler, S. H. Mushrif, A. D. Paulsen, A. D. Javadekar,D. G. Vlachos and P. J. Dauenhauer, Energy Environ. Sci.,2012, 5414.

62 M. S. Mettler, D. G. Vlachos and P. J. Dauenhauer, EnergyEnviron. Sci., 2012, 5, 7797–7809.

63 J. Fermoso, M. V. Gil, C. Pevida, J. J. Pis and F. Rubiera,Chem.–Eng. J., 2010, 161, 276–284.

64 K. Slopiecka, P. Bartocci and F. Fantozzi, Appl. Energy,2012, 97, 491–497.

65 J. Adam, M. Blazsό, E. Mésázros, M. Stöcker, M. H. Nilsen,A. Bouzga, J. E. Hustad, M. Grønli and G. Øye, Fuel, 2005,84, 1494–1502.

66 A. K. Burnham and R. L. Braun, Energy Fuels, 1999, 13,1–22.

67 S. K. Bhatia and D. D. Perlmutter, AIChE J., 1980, 26, 379–386.

68 K. Miura, H. Nakagawa, S. Nakai and S. Kajitani, Chem.Eng. Sci., 2004, 59, 5261–5268.

69 V. B. Di Marco and G. G. Bombi, J. Chromatogr., A, 2001,931, 1–30.

70 D. Vamvuka, E. Kakaras, E. Kastanaki and P. Grammelis,Fuel, 2003, 82, 1949–1960.

71 S. Xin, H. Yang, Y. Chen, X. Wang and H. Chen, Fuel, 2013,113, 266–273.

72 Y. Zhang, S. Kajitani, M. Ashizawa and Y. Oki, Fuel, 2010,89, 302–309.

73 R. Vinu and L. J. Broadbelt, Energy Environ. Sci., 2012, 5,9808.

74 B. Guichard, M. Roy-Auberger, E. Devers, B. Rebours,A. A. Quoineaud and M. Digne, Appl. Catal., A, 2009, 367,1–8.

75 C. Sheng, Fuel, 2007, 86, 2316–2324.76 B. Valle, P. Castaño, M. Olazar, J. Bilbao and A. G. Gayubo,

J. Catal., 2012, 285, 304–314.77 M. Ibáñez, B. Valle, J. Bilbao, A. G. Gayubo and P. Castaño,

Catal. Today, 2012, 195, 106–113.78 C. E. Brewer, K. Schmidt-Rohr, J. A. Satrio and R. C. Brown,

Environ. Prog. Sustainable Energy, 2009, 28, 386–396.79 Y.-T. Cheng and G. W. Huber, Green Chem., 2012, 14, 3114.80 M. Guisnet and P. Magnoux, Appl. Catal., A, 2001, 212,

83–96.81 C. H. Bartholomew, Appl. Catal., A, 2001, 212, 17–60.82 H. S. Cerqueira, G. Caeiro, L. Costa and F. Ramôa Ribeiro,

J. Mol. Catal. A: Chem., 2008, 292, 1–13.83 R. Quintana-Solórzano, J. W. Thybaut, G. B. Marin,

R. Lødeng and A. Holmen, Catal. Today, 2005, 107–108,619–629.

84 M. Guisnet, L. Costa and F. Ramôa, J. Mol. Catal., 2009,305, 69–83.

85 Marvin Sketch, ChemAxon, https://www.chemaxon.com/products/marvin/, 2010.

86 E. L. First, C. E. Gounaris and C. A. Floudas, Langmuir,2013, 29, 5599–5608.

87 http://helios.princeton.edu/zeomics/88 J. Cho, S. Chu, P. J. Dauenhauer and G. W. Huber, Green

Chem., 2012, 14, 428.89 J. Cai, W. Wu, R. Liu and G. W. Huber, Green Chem., 2013,

15, 1331.90 S. Rangarajan, A. Bhan and P. Daoutidis, Ind. Eng. Chem.

Res., 2010, 49, 10459–10470.



Publ

ishe

d on

13

Sept

embe

r 20

13. D

ownl

oade

d by

Nat

iona

l Don

g H

wa

Uni

vers

ity L

ibra

ry o

n 29

/03/

2014

15:

24:3

8.

View Article Online


characteristics and origin of char - shoucheng.pdf

Documents