Renewable Energy 185 (2022) 552e563

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Fast torrefaction of large biomass particles by superheated steam:

Enhanced solid products for multipurpose production
Qi Niu a, b, Frederik Ronsse b, Zhiyong Qi a, Dongdong Zhang a, *
a
Institute of Urban Agriculture, Chinese Academy of Agricultural Sciences, Chengdu, China
b
Thermochemical Conversion of Biomass Research Group, Ghent University, Coupure Links, 653, 9000, Ghent, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Conventional torrefaction technology upgrade biomass for improved storage or as pretreatment prior to
Received 10 September 2021 combustion or pyrolysis, and are generally performed on small-scale particles. Superheated steam (SHS)
Received in revised form as the carrier gas and sole heat source was proposed to heat biomass rapidly and evenly while generating
12 December 2021
intensively torrefied biomass. This work revealed the potential of SHS torrefaction to upgrade large wood
Accepted 15 December 2021
Available online 20 December 2021
particles (F ¼ 30 mm, H ¼ 60 mm), to improve the feedstock suitability and expand products diversity in
torrefaction technology. The effects of SHS torrefaction at different depths in large particles were
investigated. A torrefaction time of 30 min was deemed capable of upgrading even the inner sections of
Keywords:
Superheated steam torrefaction
large particle to have appropriate fuel properties. The HHV of the torrefied biomass enhanced by 50%
Large biomass particles reaching 28.76 MJ kg1. Combustion and pyrolysis behaviour of torrefied biomass demonstrated their
Multipurpose torrefied biomass properties as benign fuel and pyrolysis feedstock, respectively. Moisture absorption (from 13.52 to
Hydrophobicity 3.42 wt%) and grindability results (386.96e10.52 kWh ton1) indicated that large particles from SHS
Grindability torrefaction were easy for handling, transportation, and storage. The investigation of physicochemical
properties at different depths of the torrefied biomass helped to understand SHS torrefaction behaviour
in large biomass particles, hence, to control the reaction extent in the desired end product.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction [9e12].
Torrefaction is usually performed in an inert atmosphere at a
The depletion of conventional fossil fuel reserves along with temperature between 200 and 300  C, with the torrefaction time
massive greenhouse gas emissions and environmental pollution ranging from minutes to hours [13,14]. Hemicellulose and extrac-
have prompted the investigation of sustainable, efficient and cost- tives are easily degraded during torrefaction process [15]. A small
effective energy sources [1e3]. The utilization of biomass has part of cellulose and lignin (mainly short chain lignin compounds)
attracted a lot of interests [4]. However, disadvantages such as a low is also decomposed during torrefaction process [16,17]. The torre-
energy density (8e16 MJ kg1) and a low bulk density fied biomass retains most of the chemical energy from the raw
(60e100 kg m3) of biomass have limited the commercial and in- biomass [18]. The energy density is enhanced, resulting from the
dustrial utilization [5,6]. Additional problems concerning stable loss of low calorific compounds. Torrefaction increases hydropho-
storage are caused by the hydrophilic nature of biomass [7]. Diffi- bicity due to the loss of surface eOH groups, mainly from hemi-
culties in combustion are attributed to the low burning rate and cellulose. Torrefaction also increases the grindability of biomass by
low adiabatic flame temperature of biomass [8]. In order to increase changing its microstructure. The improved energy density, hydro-
the energy conversion efficiency, biomass can be pretreated in or- phobicity and grindability benefit the handling and storage of tor-
der to improve its fuel qualities. Torrefaction is a promising method refied biomass [19].
to convert low-quality biomass into high energy density feedstock, Nevertheless, the particle size of torrefied feedstock was mostly
suitable for long-distance transportation, long-term storage, having reported to be smaller than 2 mm [20e22]. Studies with large
easier grindability and attaining higher efficiency in combustion particles were in general scarce. Peng et al. [23] and Wang et al. [24]

* Corresponding author.
E-mail address: [email protected] (D. Zhang).

https://doi.org/10.1016/j.renene.2021.12.070
0960-1481/© 2021 Elsevier Ltd. All rights reserved.
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

conducted the effects of particle size on torrefaction in a range of 2. Methods and materials
0.2e6.35 mm. Basu et al. [25] investigated the torrefaction of large
particles (diameter of 25.4 mm and length of 65 mm). Those studies 2.1. Materials
mainly focused on mass and energy yield and were short of
investigation on how torrefaction affected different depths in large A tree trunk from a 20 years old Cinnamomum camphora
particles. In practice, torrefaction of larger particles requires longer (Jiangxi, China) was used to make wooden cylinders
reaction time due to heat transfer limitations and uneven heating, (diameter ¼ 30 mm, long ¼ 60 mm) as feedstock for torrefaction. A
otherwise products have usually nonuniform properties resulting wood-working factory helped to shape the wooden cylinders.
in poor performance of grindability and hydrophobicity. Milling the Wooden cylinders were taken from similar trunk parts to ensure
feedstock to small size prior to torrefaction is inevitable which uniform physiochemical properties.
requires large amounts of milling energy. Superior technology to
torrefy large biomass particles would be of interests as milling after 2.2. Torrefaction experiments
the torrefaction process considerably requires less energy than
milling the raw feedstock prior to torrefaction. The SHS torrefaction setup has been described in our previous
Superheated steam (SHS) torrefaction is a thermochemical work [10]. A brief diagram is given in Fig. 1 a. SHS (2.2 kg h1) is
process that uses SHS as the heating medium to upgrade biomass used as the carrier gas and the sole heat source to heat the reactor
under atmospheric pressure. As demonstrated in our previous chamber to the set temperature before experiment. Six raw wood
study [10], SHS is superior compared to other heat transfer media cylinders are filled into a cylindrical vessel (volume of 400 mL) for
(like hot nitrogen or combusted torrefaction off-gases) as it has a each test. The vessel was made of stainless-steel mesh with holes of
large heat capacity with high heat transfer rates and avoids mass 0.9 mm N2 gas is used to remove air from the vessel before the
transfer limitations [26]. Moreover, SHS can directly contact the vessel is put into the reactor chamber. In this study, 300  C was
biomass. The heating rate of biomass in a SHS torrefaction reactor chosen as the reaction temperature, in which extensive degrada-
can reach 100  C min1, being 20 times faster than in an indirectly tion of hemicellulose and cellulose occurs, while minor depoly-
heated tubular reactor [10]. SHS has been performed to torrefy merization of lignin happens [29]. The biomass temperature is
biomass in a heavily loaded fixed bed reactor (160 g in 300 mL). A determined by the average temperature measured by thermocou-
heating rate of 15  C min1 to reach 325  C and followed by a ples above and underneath the vessel (i.e., TReactor1 and TReactor2,
holding time of 2 min improved the energy density by 45%; com- respectively). Heating rate is measured to be higher than 100  C
bustion indices and behaviours showed that the torrefied biomass min1, thus the heating time is included in the torrefaction time.
could be used as either co-firing or fuel [27]. Thus, SHS enables fast When the set torrefaction time has been reached, the vessel con-
and homogeneous biomass heating, and can heat the inner parts of taining the torrefied biomass is taken out and placed in a container
large (and porous) biomass particles quickly. which is flushed with N2 gas. The vessel and its contents are cooled
As a result, SHS torrefaction of large biomass particles is pro- to room temperature in about 1 min. Both liquid and gas products
posed, which could save significant amounts of efforts and energy in the effluent are safely disposed.
that would otherwise have to be spent to prepare (mill) the biomass
to small particles prior to torrefaction using standard means of 2.3. Sample preparation
heating (i.e., by contact or using hot non-condensable gases)
[21,22,28]. Hence, feedstock suitability is improved. Moreover, tor- After torrefaction, the biomass particle shrunk depending on the
refaction duration of large biomass particles by SHS is supposed to be torrefaction severity (25.4e30.0 mm diameter, 55.8e60.0 mm long).
much shorter than in a conventionally heated process. On the other The torrefied biomass was divided into three sections: core, middle,
hand, according to our previous findings [10], the surface of the SHS and outer (Fig. 1b). The radius and length of the torrefied biomass
torrefied biomass was smooth with limited pores and contained no particle were measured as ‘Xi’ and ‘Yi’. The core section was defined
eOH groups, making it highly hydrophobic. Even more, torrefaction to be the inner part with a radius of ‘Xi/3’ and length of ‘Yi 4/3 Xi’.
of the surface of large-sized biomass particles could form a hydro- The middle section was defined as having a radius of ‘2/3 Xi’ and
phobic outer layer and keep the inner section dry enough for long length of ‘Yi 2/3 Xi’ and excluding the core section. The outer section
time storage purposes. Besides, SHS torrefied biomass has higher was the remaining part without middle and core sections. The outer
HHV than most products from conventional torrefaction processes section was firstly removed by sandpaper or knife to take samples. A
[10]. The SHS torrefaction of large biomass particles could generate vernier caliper was used to measure the particle sizes carefully. Af-
products suitable for multiple applications. terwards, the middle section was collected in the same way. The core
This work studies SHS torrefaction of large-sized biomass par- section can be obtained after stripping the outer and middle sections.
ticles for multiple potential applications and investigates the SHS A planetary ball mill milled the collected samples to <0.075 mm
torrefaction reactions at different depths in the particle. This prior to analysis. The samples were labeled as TO20, TM20, TC20,
investigation of large particle is vital, because a good understanding TO30, TM30, TC30, TO60, TM60, and TC60 (where T means torre-
of the physicochemical characteristics at different depths helps to faction, O, M and C means outer, middle and core section, respec-
understand the mechanism of SHS torrefaction in upgrading tively, and numbers means the torrefaction time).
biomass, and hence to control the extent of reaction and develop
desired properties in the end products (i.e., fuel, pyrolysis feedstock, 2.4. Analytical techniques and calculations
storage, grindability properties). Torrefaction times of 20, 30 and
60 min at 300  C were studied to torrefy Cinnamomum camphora 2.4.1. Proximate and ultimate analysis
wood cylinders. The torrefied biomass particle is classified into Ash, volatile matter (VM), fixed carbon (FC) and moisture con-
three different sections representing different depths. Physico- tent of biomass and products were analyzed according to the ASTM
chemical characteristics along with the fuel properties have been standard method (ASTM-D5142-04). An elemental analyzer (Ele-
determined. This analysis will enable to understand the behaviours mentar Vario El Cube) was used to measure C, H, N and S content,
of biomass at different depths undergoing SHS torrefaction. Based and oxygen content was determined by difference. The higher
on the above discussion, suggestions for commercial applications heating value (HHV) was measured by an oxygen bomb calorimeter
were proposed to provide insights into biomass torrefaction. (XRY-1A).
553
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Fig. 1. (a) Diagram of the reactor and (b) different designated parts of the torrefied biomass cylinder.

2.4.2. Chemical compounds follows [32]:

Cellulose, hemicellulose and lignin concentration were analyzed
according to NREL methods [10,30]. S ¼ (Rmax * Rmean) / (T2i * Tb) (1)

where Rmax refers to the maximum weight loss rate, in % min1;

2.4.3. FTIR analysis
Rmean refers to the mean weight loss rate, in % min1.
The surface functional groups of the raw and torrefied biomass
The Coats-Redfern integral method has been widely applied to
samples was employed to determine the change in biomass con-
assess the kinetic parameters in the non-isothermal combustion
stituents during torrefaction by Fourier Transform Infrared spec-
process [32]. The activation energy (Ea), frequency factor (A) and
troscopy (Nicolet iS10 FT-IR spectrometer). Around 1e2 mg sample
reaction order (n) can be obtained from the differential and integral
was analyzed with a wavenumber in the range of 4000e800 cm1
methods.
with a resolution of 4 cm1 [10].
The reaction rate is expressed as:

2.4.4. Combustion and pyrolysis characteristics and kinetic analysis da

¼ kð1  aÞn (2)
of samples dt
Thermogravimetric analysis was carried out by a Mettler Toledo
TG-DSC 3þ. Around 10 mg of ground sample was analyzed using a where a refers to the sample fraction decomposed at time t, n refers
heating rate of 10  C min1 in the temperature range of room to the order of reaction, and k refers to the rate constant given by
temperature to 800  C. Air and nitrogen gas with a flow rate of the Arrhenius expression:
70 mL min1 were set as carrier gases to investigate the combustion
Ea
and pyrolysis characteristics, respectively. k ¼ A,eR,T (3)
Several methods have been applied to evaluate the combustion
characteristics according to the thermogravimetric analysis results. where A is frequency factor, in mol1, and Ea is activation energy of
The parameters used to assess the combustion performance have the reaction, in J mol1.
been defined as follows: ignition temperature (Ti) refers to the The Coats-Redfern equation can be expressed as:
temperature where the weight loss rate increases to 1% min1 at
the early stage of the combustion process, and the burnout tem- lnð1  xÞ AR Ea
lnf g ¼ lnð Þ ðfor n ¼ 1Þ (4)
perature (Tb) refers to the temperature where the weight loss rate T2 bEa RT
decreases to 1% min1 in the late stage of the combustion process
[31]. R1 represents the weight loss rate of the first peak in the DTG and:
curve, and T1 is defined as the first peak temperature corresponding
to this first peak. R2 represents the weight loss rate of the second 1  ð1  xÞ1n AR Ea
peak in the DTG curve, and T2 is defined as the second peak tem- lnf g ¼ lnð Þ ðfor n s 1Þ (5)
T 2 ð1  nÞ bEa RT
perature corresponding to this second peak. Furthermore, to eval-
uate the combustion reactivity of the torrefied samples, the index of Similarly, the pyrolysis kinetics parameters were obtained using
comprehensive combustibility (S) is applied, which is defined as the CoatseRedfern method.
554
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

2.4.5. SEM analysis function of torrefaction time at 300  C. The values are listed in
Samples were collected by cutting the torrefied biomass parti- Table A1. Fig. 2 a shows that the weight loss of the solid product
cles in half and taking pieces in the corresponding sections. The raw increase steadily with the increase in torrefaction time. The yields
and torrefied biomass cuttings with a size range of 0.5e1.5 mm in the solid residue decline dramatically from 62.3% to 37.4%, when
were analyzed by scanning electron microscopy (Hitachi SU4800) the torrefaction time is extended from 20 to 60 min. The weight loss
[10]. in the studied large biomass particles is comparable to that of
small-sized biomass particles in dry torrefaction at same temper-
2.4.6. Moisture absorption tests ature of 300  C [32], implying that SHS could lead to severe tor-
Both raw and torrefied biomass were kept in a closed environ- refaction of large biomass particles. As torrefied biomass keeps its
ment with a temperature of 25  C and relative humidity of 55%. All original shape and only modestly shrinks, the bulk density de-
samples were weighed every 24 h. The moisture absorption was creases significantly and reaches a steady-state at torrefaction
calculated as: times of 30 min onwards, see Fig. 2 b.
HHV is an important parameter to determine the biomass
Wi  W0 characteristic as a fuel. Although certain amount of energy is lost
Mositure absorption; % ¼  100 (6)
W0 due to partial devolatilization of the sample upon SHS torrefaction,
torrefied biomass still obtains a high HHV under severe conditions.
Where Wi , in g is the weight of the taken sample, and W0 , in g is the Fig. 2 c illustrates the HHV at different depths of the torrefied
weight of the initial sample. biomass particles. All torrefied samples show significant increase in
HHV along with the torrefaction time. However, the HHV increases
2.4.7. Grinding energy consumption differently at different depths of the large particle. The outer sec-
Two cutting mills were performed to determine the grindability tion is the most affected, probably caused by direct contact be-
of torrefied and raw biomass in two stages. Samples were first tween SHS and biomass, where heat is rapidly transferred by
ground by IKA MultiDrive basic to a smaller size that can be fed into convective heat transfer. Hence, TO60 had the highest HHV
IKA MF 10.1 for fine grinding. The MF 10.1 mill was equipped with a (29.21 MJ kg1). Nevertheless, the higher heat transfer of SHS also
2 mm bottom sieve. The electricity consumption of the two cutting offers a good distribution of heat to the inner sections of the
mills with an empty load and during grinding were recorded. The biomass particle, which improves the homogeneity of the product
energy consumption for biomass grinding was obtained by sub- after torrefaction. A torrefaction time of 30 min enabled the middle
tracting the energy consumption with an empty load from that of section to have an HHV comparable to that in the outer section. The
the sample grinding stage. Then, energy consumption in both core section is the most difficult to heat due to limited conductive
grinding stages were summed to obtain the total grinding energy. heat transfer, resulting in a longer torrefaction time of 60 min to
increase its HHV to a similar level as the outer section of 30 min. As
the core section only weights around 2e8% of the whole torrefied
3. Results and discussion
biomass, SHS torrefaction with a holding time of 30 min is
considered sufficient for upgrading large particles of the studied
3.1. Fuel characters of torrefied biomass
size (reaching an overall HHV of 28.99 MJ kg1). As pointed out by
Anna et al. [22], the highest HHV from dry torrefaction (300  C, 30
Fig. 2 shows the yields and the basic fuel characteristics of raw
min) of large particles was about 23 MJ kg1. Energy enhancement
and torrefied biomass at different depths (or as a whole) as a

Fig. 2. (a) Weight loss, (b) density, (c) HHV and (d) energy enhancement of raw and SHS (300  C) torrefied biomass.

555
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Table 1
Composition Properties of raw and torrefied biomass (wt%, daf basis).

Samples Cellulose Hemicellulose Lignin Others

Raw biomass 31.76 ± 0.71 8.53 ± 1.14 26.25 ± 0.55 33.46 ± 0.23

20 min Outer 16.93 ± 0.70 0.19 ± 0.01 73.49 ± 2.45 9.39 ± 1.08
Middle 29.48 ± 0.97 1.82 ± 0.11 48.24 ± 1.34 20.46 ± 0.65
Core 34.99 ± 1.94 5.02 ± 0.14 37.20 ± 0.88 22.78 ± 0.95
30 min Outer 0.00 0.00 96.07 ± 0.83 3.93 ± 0.83
Middle 16.17 ± 0.14 0.00 69.75 ± 2.81 14.08 ± 1.46
Core 39.30 ± 2.03 0.91 ± 0.03 45.13 ± 1.06 14.66 ± 1.11
60 min Outer 0.00 0.00 98.05 ± 0.69 1.95 ± 0.69
Middle 0.00 0.00 97.25 ± 0.54 2.75 ± 0.54
Core 0.27 ± 0.02 0.00 95.43 ± 0.41 4.30 ± 0.40

of SHS torrefaction is generally higher than conventional torre- transfers more heat to the middle and core section. As a result, the
faction [13], Fig. 2 d. The trend of the energy enhancement at fixed carbon content of the middle and the core sections reach the
different depths is consistent with the variation in weight loss and same value as that of the outer section at 60 min of torrefaction
HHV. time. The slight increase in ash content from the outer to core
section demonstrates the loss (devolatilization) in organic com-
ponents. Compared to dry torrefaction of large particles with
3.2. Chemical characteristics of torrefied biomass
distinct external and internal sections [22,25], SHS leads to a more
homogeneously torrefied biomass within same torrefaction times.
The changes in elemental and proximate results of raw and
The Van Krevelen diagram is shown in Fig. 4. O/C and H/C ratios
torrefied biomass at different conditions (and different sections)
of all outer and middle sections of torrefied samples are in the
are displayed in Fig. 3. Generally, the carbon fraction increases
range of coal which is promising for fuel applications [31]. The core
considerably with the extended torrefaction time, whereas the
section undergoing 60 min is in the range of coal as well, indicating
hydrogen and oxygen contents decrease, resulting from the loss of
that SHS torrefaction with a reasonable holding time can fully up-
eOH groups. The higher carbon content is in agreement with the
grade large-size particles. Starting from the raw biomass, all tor-
HHV of torrefied samples [8]. Regarding the carbon content at
refied samples follow a line a bit above the dehydration line during
different depths, the outer and middle sections have similar values,
the torrefaction process. This implies that oxygen removal from the
which are slightly higher than the core section. It implies that the
samples is not only in the form of water, but also in the form of
core section is less torrefied because of the delayed heat transfer.
some non-condensable gases and light volatiles [33].
Though elemental compositions of the outer and middle sections
are more or less the same, the proximate analysis tells a different
story. The fixed carbon content in the outer section is much higher 3.3. Chemical components and functional groups
than that in the middle section undergoing torrefaction times of 20
and 30 min. This is because SHS heats the outer section faster, Fig. 5 illustrates the FTIR results of raw and torrefied biomass.
leading to a higher loss of volatile matter. The content of volatile Biochemical composition of the raw and torrefied biomass are
matter in different sections gives an insight in the heat transfer that displayed in Table 1. The band around 3385 cm1 is assigned to
occurs in large particles. With the extended torrefaction time, SHS eOH stretching vibration [34]. The significant difference from

Fig. 3. Ultimate and proximate analysis results (in wt%) of raw and torrefied biomass in different sections.

556
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Fig. 4. Van Krevelen diagram of raw and torrefied biomass.

different sections can be attributed to dehydration during the tor- the hemicellulose content as hemicellulose were fully decomposed
refaction process, which agrees with the elemental analysis results. at 30 min in outer and middle section. Enhanced decomposition of
Dehydration decreases in the order from the outside to the core hemicellulose occurred at all sections at 60 min. The bands at 1424,
section. The extension of the torrefaction time from 30 min to 1463, 1507 and 1595 cm1 are ascribed to C]C stretching of the
60 min does not significantly improve the dehydration efficiency of aromatic skeletal vibrations in lignin [35]. The decomposition of
the core section. lignin is proportional to the extent of torrefaction time. However,
The band around 2920 m1 could be symmetric aliphatics [15]. lignin decomposition is less severe than hemicellulose, which is
The intensity of aliphatic compounds decreased as the torrefaction consistent with the inherent structural complexity of lignin, which
time extended. The decreasing torrefaction severity from outer to requires higher energies (i.e. temperatures above 300  C) to be
core can be correlated to the biochemical composition from Table 1. decomposed [36]. The band around 1329 cm1 is a characteristic
The band at 1738 cm1 is attributed to the C]O vibration of esters, band for syringyl, guaiacyl and phenol groups within lignin [35].
ketones and aldehydes in hemicellulose [34]. The intensity of this The constant intensity of this band is similar to other bands asso-
band decreased with prolonging the torrefaction time from 20 to ciated to lignin. The band at 1374 cm1 which corresponds to alkyl
30 min. No significant intensity changes were observed from CeH group bending from lipids, carbohydrates and aliphatic chains
different sections at 30 min compared to 60 min, which agrees with varies evidently with the extension of torrefaction time [15]. This

Fig. 5. FTIR results of raw and torrefied biomass.

557
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

agrees with the content in other organics (extractives) in Table 1. prolonged from 20 to 60 min, more marked and deeper cracks are
The bands around 1100-1242 cm1 are the characteristic bands for observed in the external and internal sections. It should be noted
hemicellulose: the band around 1242 cm1 is assigned to glucose that the formation of pores and cracks are the consequence of the
ring stretching vibration [37]; the band around 1161 cm1 is release of volatiles from samples, and the amounts of pores and
attributed to CeOeC glycosidic linkages in both cellulose and cracks increased with longer torrefaction time, which is consistent
hemicellulose; the band around 1108 cm1 owes to CeO stretching with the decrease in volatile matter from proximate analysis [38].
vibration of xyloglucan in hemicellulose. The slightly decreased Pore-like structures were clearly exhibited in 60 min treated sam-
absorption intensity of these bands from a torrefaction time of ples (both external and internal sections), which owes to the
20e60 min is in line with the decomposition of hemicellulose. The degradation, devolatilization, and escape of volatiles produced
bands at 1031-1056 cm1 refer to CeO stretching vibrations in from hemicellulose, amorphous cellulose, and parts of lignin. The
cellulose and hemicelluloses, and aromatic CeH deformation in changes in microstructure can also explain the improved grind-
lignin [35]. Those peaks disappear when raw biomass is torrefied at ability of torrefied biomass compared to that of raw biomass [39].
300  C for 60 min, demonstrating the reduction of those com-
pounds at this torrefaction time. 3.5. Combustion behaviour of torrefied biomass

3.4. SEM analysis The properties of torrefied samples as fuel are evaluated ac-
cording to results of TG and DTG under air atmosphere (Fig. 7). The
The SEM micrographs of the raw and the torrefied samples at combustion kinetic study is discussed thereafter. Table 2 demon-
different sections are shown in Fig. 6. It can be noticed that the strates the combustion characteristic parameters of all samples
particle structure of the raw material is compact, homogeneous and derived from TG and DTG curves. The combustion process of
has a smooth fibrous organization. Visible damage of the external biomass can be divided into two stages, i.e., devolatilization and
structure (outer section) from torrefied biomass can be observed in combustion of volatiles (200e350  C) and combustion of fixed
the sample produced at 300  C and 20 min, while its internal sec- carbon (350e800  C). The weight loss of raw biomass at the
tion (core section) is similar to raw biomass, indicating only slight devolatilization and volatile combustion zone was 67.16% within
torrefaction. The cracks and pores are generated by the devolatili- the temperature range of 200e350  C. The weight loss in the char
zation of polymers with poor thermal stability. As torrefaction time combustion zone accounted for 28.86%. In torrefied biomass

Fig. 6. SEM images of (a) raw biomass, (b) TO20, (c) TO30, (d) TO60, (e) TM20, (f)TM30, (g) TM60, (h) TC20, (i) TC30 and (j) TC60.

558
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Fig. 7. TG and DTG curves of different samples under air atmosphere.

Table 2
Combustion characteristic parameters for different samples at a heating rate of 10  C min1.

Samples Ti ( C) Tb ( C) DT ( C) T1 ( C) R1 (% min1) T2 ( C) R2 (% min1) Rmax (% min1) Rmean (% min1) S (107%2  C-3 min2)

Raw biomass 248.13 437.15 189.02 314.24 8.98 425.77 8.69 8.98 1.249 4.17

20 min Outer 294.30 452.21 157.91 333.81 3.26 422.97 18.04 18.04 1.252 5.77
Middle 287.34 474.79 187.45 322.63 4.16 420.18 13.38 13.38 1.278 4.36
Core 284.62 472.15 187.53 330.09 9.24 434.06 10.22 10.22 1.314 3.51
30 min Outer 292.90 446.82 153.92 339.40 3.12 411.63 21.23 21.23 1.244 7.05
Middle 291.26 444.63 153.37 339.59 3.24 420.21 20.19 20.19 1.258 6.73
Core 277.28 464.79 187.51 324.40 7.45 421.07 13.80 13.80 1.448 5.59
60 min Outer 309.72 470.50 160.78 335.28 1.98 418.84 21.02 21.02 1.262 5.88
Middle 298.91 453.52 154.61 329.15 2.48 411.59 17.03 17.03 1.290 5.42
Core 289.27 443.87 154.6 325.07 3.01 422.06 20.56 20.56 1.258 6.96

produced at 60 min, the first weight loss peak is almost invisible in combustion characteristics are [31]. An optimum S value of
any of the sections, attributed to the removal of volatile matter and 7.05  107%2  C-3 min2 is obtained from TO30. The above results
conversion into fixed carbon [32]. show the enhanced combustion performance of torrefied samples.
As shown in Table 2, the ignition temperature of torrefied Table 3 shows the combustion kinetic parameters of different
samples increased compared to that of the raw material, and the samples. The regression coefficients (R2) are all higher than 0.9,
increasing trend is enhanced with the extension in torrefaction which is acceptable in kinetics studies. The activation energies (Ea)
time. However, the ignition temperatures decrease from the outer of torrefied samples at the devolatilization and volatile combustion
to the core section, attributed to the varying degree of volatile stages and at the char combustion stages are lower than those in
matter removal during the torrefaction process [32]. The difference raw biomass. This indicates that SHS torrefaction is conducive to
between ignition and burnout temperature narrows after torre- the combustion process. This is in accordance with the combustion
faction. The mass loss rates of the first and second peaks are applied characteristic parameters shown in Fig. 7 and Table 2.
to evaluate the combustion characteristics. The decreased R1 in
torrefied samples can be attributed to the lower decomposition of 3.6. Thermal behaviour of torrefied biomass
volatiles after the torrefaction process, which illustrates that the
volatile matter has been reduced after torrefaction. As shown in The properties of torrefied samples as pyrolysis feedstock are
Table 2, the R2 values increased compared with the raw samples, evaluated by TG and DTG curves under N2 atmosphere (Fig. 8). The
indicating that the char combustion is enhanced in torrefied whole pyrolysis process can be divided into three stages. The
biomass. The maximum combustion rate is 21.23% min1 in TO30. temperature range of the first decomposition stage is from room
Extending the torrefaction time from 30 to 60 min has no signifi- temperature to 200  C, and the weight loss at this stage is due to
cant effect on the combustion rate of the outer section, unlike for dehydration and free water evaporation [41]. A significant mass loss
the core section, indicating the almost full torrefaction of the outer is observed between 200 and 500  C, which is attributed to the
section. The combustion rate of the core section material can only depolymerization and decomposition of lignocellulose. The small
approach a value comparable to the outer section when torrefaction shoulder around 300  C is related to the decomposition of hemi-
time reached 60 min. The comprehensive combustibility index (S) cellulose. The prominent peak around 350  C represents the
can evaluate the fuel ignition and combustion performance of cracking of cellulose and lignin. The decomposition of lignin con-
different samples [40]. The higher the S value is, the better the tinues into the third stage [42].
559
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Table 3
Combustion kinetic parameters of different samples.

Samples Devolatilization and volatile combustion Char combustion

R2 A (min1) Ea (kJ mol1) R2 A (min1) Ea (kJ mol1)

Raw biomass 0.9527 2.61E-04 16.226 0.9242 2.96E-15 108.805

20 min Outer 0.9867 3.59E-03 10.301 0.9361 7.57E-12 83.522

Middle 0.9949 1.86E-03 12.442 0.9423 8.36E-10 65.746
Core 0.9709 4.18E-04 15.618 0.9639 2.61E-10 70.269
30 min Outer 0.9894 3.65E-03 10.012 0.933 1.40E-13 95.769
Middle 0.9895 3.53E-03 10.240 0.9304 3.14E-15 100.684
Core 0.9468 2.04E-03 11.340 0.9522 1.63E-10 70.984
60 min Outer 0.9648 5.41E-03 8.808 0.9536 3.82E-12 86.116
Middle 0.9656 4.71E-03 9.021 0.9574 2.90E-12 85.318
Core 0.9683 2.64E-03 10.986 0.9552 6.95E-16 116.812

Fig. 8. TG and DTG curves of different samples under N2 atmosphere.

The weight loss rate of the second stage diminishes after tor- outer, middle and to the core section, reaching a 16.8% reduction
refaction, and longer torrefaction times lead to lower weight loss (compared to raw biomass) in TC60 which is comparable to TO20,
rates. This is attributed to the decomposition of volatiles during the indicating the improved properties of torrefied large sized
torrefaction process. Compared to the raw biomass, the maximum biomass as pyrolysis feedstock. In addition, the pre-exponential
mass loss rate of torrefied samples slightly shifts to higher tem- factor (A), which is used to estimate the reaction intensity, de-
perature. The shoulder of the hemicellulose peak became more creases with the increase in torrefaction time. The coefficient of
difficult to distinguish with the extension of torrefaction time, and determination (R2) of the Coats-Redfern method exceeds 0.9 for
the mass loss is significantly decreased due to the degradation of both the raw and torrefied samples, which is deemed acceptable
hemicellulose, and to a minor extent of cellulose and lignin. The in kinetics studies.
shapes of TG and DTG curves from different sections are consid-
erably different. As can be seen from the TG curves, the weight loss 3.7. Moisture absorption and grindability for biomass storage
declines from the outer to the core section. Significant mass loss of
the core section is observed even as the torrefaction time was SHS torrefaction of large biomass particle quickly form a hy-
extended up to 60 min. However, the DTG curves of the outer and drophobic surface, which guarantees stability upon long storage
middle sections at the second pyrolysis stage are relatively flat, periods. Fig. 9 a presents the moisture content in raw and torrefied
which can be explained by the severe decomposition that already biomass that has been stored in a closed environment for up to
took place during the SHS torrefaction process. 168 h. Trials with a torrefaction time of 10 min were also included.
Kinetics parameters of raw and torrefied biomass pyrolysis The low moisture absorption in torrefied biomass reveals that
obtained by the CoatseRedfern model at a heating rate of 10  C torrefied biomass has a lower moisture absorption rate [31]. This is
min1 are presented in Table 4. The Coats-Redfern method is used attributed to the decreased contents of hydrophilic hydroxyl groups
to determine the average activation energy of the last two on its surface, resulting from the thermal decomposition of hemi-
decomposition stages, which are the main pyrolysis stages [32]. cellulose in the torrefaction process [43]. Biomass undergoing tor-
The activation energy (Ea) decreases after torrefaction due to the refaction times longer than 20 min slightly absorbs moisture
cracking of large organic components in the biomass matrix. (<4.0 wt%) and its absorption varied negligibly with prolonged
Notable reduction of Ea is observed, in ascending order, from storage time, indicating that the dehydroxylation in the outer
560
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

Table 4
Kinetics parameters for raw and torrefied biomass pyrolysis obtained by the CoatseRedfern model at a heating rate of 10  C min1.

Samples Ea (kJ mol1) A (min1) R2

Raw biomass 14.231 5.92E-04 0.9884

20 min Outer 11.931 4.21E-03 0.9649

Middle 12.478 3.43E-03 0.9756
Core 14.027 1.37E-03 0.9806
30 min Outer 10.688 5.54E-03 0.9617
Middle 11.188 4.93E-03 0.9636
Core 13.404 1.92E-03 0.9857
60 min Outer 9.895 7.08E-03 0.9488
Middle 10.502 6.35E-03 0.9503
Core 11.845 4.50E-03 0.9688

section alone helps to make the large particles resistant to moisture constituents [46]. Hence, the energy requirements are significantly
absorption upon storage. decreased for grinding torrefied samples. The grinding energy of
Conventional torrefaction needs massive milling energy (prior torrefied biomass undergoing 60 min decreased to only 10.52 kWh
to torrefaction) because larger particles would have an inefficient t1. Longer torrefaction times could further decrease the grinding
heat transfer with the conventional heating methods applied; energy, yet the time and energy needed for deeper torrefaction are
hence prior milling is required. Fortunately, SHS enables fast tor- uneconomic. Thus, SHS torrefaction times longer than 20 min were
refaction of large particles and reduces the need for energy in deemed sufficient for significant savings on grinding energy.
intensive milling steps prior to torrefaction, rendering SHS torre-
faction more energy-efficient. The energy consumption of milling
3.8. Discussion of torrefaction mechanisms
raw samples in the form of a cylinder of 30 mm diameter with
60 mm height down to pieces <2 mm is demonstrated in Fig. 9 a.
SHS offers a high heat transfer coefficient that can heat large
Practically, the size of 2 mm is a common feedstock size for con-
particles uniformly from the outer section to the core. Moreover,
ventional torrefaction. The energy for milling raw biomass is
steam can penetrate the pores of the biomass particles, as this will
comparable to the energy needed for conventional torrefaction, as
no doubt help the more rapid heating in the center fraction of the
calculated in supplementary materials. The grindability of torrefied
particle. The SHS torrefaction facilitates the decomposition of
biomass is evaluated by determining the energy required for
biomass components from the outer to the core section as the
grinding large torrefied particles into 2 mm sized particles. The
torrefaction time extended from 20 to 60 min at 300  C. The
energy required for grinding torrefied biomass decreases signifi-
compositional changes in the internal sections imply that heat
cantly with an increase in torrefaction time. The grinding energy of
transfer in this work was quick. Torrefaction involves the extensive
torrefied biomass is reduced from 386 (raw samples) to 14.59 kWh
decomposition of hemicellulose and the slight lignin degradation at
t1 at 300  C for 30 min, which is comparable with corn (19.8 kWh
300  C from the outer to the core section. Hemicellulose degrada-
t1) and barley straw (27.1 kWh t1) [31]. Enhanced grindability of
tion leads to the loss of eOH groups accompanied by the release of
torrefied biomass can be linked to the relatively increased brittle-
light gases such as CO2, CO, and water. The decomposition of cel-
ness and its degradation of the biomass fibrous structure during
lulose and lignin includes depolymerization, cleavage of glycosidic
torrefaction [31,44]. The lignocellulose network is formed by link-
bonds and demethoxylation [29,47]. Cellulose is extensively
ing and embedding cellulose microfibrils and macro-fibrils in a
degraded in SHS torrefaction which is not expected at the given
matrix of hemicellulose and lignin, resulting in a tough layer (cell
temperature. The cellulose content of the core section undergoing
wall) that protects the wood structure against mechanical forces
60 min demonstrates that heat has been well transferred to the
[45]. Loss of mechanical strength of cell walls and wood structure
core. As a result of the above-mentioned thermal reactions, the O/C
occurred after torrefaction due to the destruction of the interlinked
and H/C ratios of torrefied biomass decreased while the carbon
cellulose-hemicellulose-lignin matrix and decomposition of these
content and the heating value increased. The loss of hydroxyl

Fig. 9. (a) Moisture absorption of raw and torrefied biomass that have been stored for up to 168 h; (b) Grinding energy consumption of raw and torrefied samples from a cylinder of
diameter 30 mm and with a height of 60 mm to particles <2 mm.

561
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

groups results in a hydrophobic outer section. Also, the fibrous [2] M.-F. Li, et al., Wet torrefaction of bamboo in hydrochloric acid solution by
microwave heating, ACS Sustain. Chem. Eng. 3 (9) (2015) 2022e2029.
structure becomes brittle during torrefaction, significantly
[3] O. Winjobi, D.R. Shonnard, W. Zhou, Production of hydrocarbon fuel using
improving grindability. two-step torrefaction and fast pyrolysis of pine. Part 1: techno-economic
analysis, ACS Sustain. Chem. Eng. 5 (6) (2017) 4529e4540.
[4] L. Wang, et al., Effect of torrefaction on physiochemical characteristics and
4. Conclusion grindability of stem wood, stump and bark, Appl. Energy 227 (2018) 137e148.
[5] N. Mi
sljenovi c, et al., Torrefaction influence on pelletability and pellet quality
The effects of SHS torrefaction on composition, chemical prop- of Norwegian forest residues, Energy Fuels 28 (4) (2014) 2554e2561.
[6] M. Tan, et al., Comparison of atmospheric and gas-pressurized oxidative tor-
erties, combustion behaviors, pyrolysis characteristics, grindability,
refaction of heavy-metal-polluted rice straw, J. Clean. Prod. 283 (2021)
unique microstructure and morphology of different sections in 124636.
large particles were investigated. The HHV and fixed carbon con- [7] M.J.C. van der Stelt, et al., Biomass upgrading by torrefaction for the produc-
tion of biofuels: a review, Biomass Bioenergy 35 (9) (2011) 3748e3762.
tent of this work are higher than those of most conventional tor-
[8] M.J. Prins, K.J. Ptasinski, F.J.J.G. Janssen, More efficient biomass gasification via
refaction (16e29 MJ kg1) with smaller sized wood feedstock [20]. torrefaction, Energy 31 (15) (2006) 3458e3470.
The middle section of the large particle has similar fuel character- [9] N.-E.E. Mansouri, J. Salvado  , Structural characterization of technical lignins for
istics (above 28 MJ kg1) as the outer section with a torrefaction the production of adhesives: application to lignosulfonate, kraft, soda-
anthraquinone, organosolv and ethanol process lignins, Ind. Crop. Prod. 24
time of 30 min, indicating that SHS (at 300  C) can transfer heat (1) (2006) 8e16.
rapidly to the inner section of large particles. Combustion and py- [10] D. Zhang, et al., Superheated Steam as Carrier Gas and the Sole Heat Source to
rolysis behaviours indicate that SHS torrefaction can upgrade large Enhance Biomass Torrefaction, Bioresource Technology, 2021, p. 124955.
[11] S. Singh, J.P. Chakraborty, M.K. Mondal, Pyrolysis of torrefied biomass: opti-
biomass particles as fuel or pyrolysis feedstock. SHS torrefaction of mization of process parameters using response surface methodology, char-
large particles at 300  C for 20 min ensures long storage time of acterization, and comparison of properties of pyrolysis oil from raw biomass,
products due to the hydrophobic surface. On the other hand, SHS J. Clean. Prod. 272 (2020) 122517.
[12] Y.Y. Gan, et al., Microwave-assisted wet torrefaction of microalgae under
can save grinding energy and efforts by milling large particles after various acids for coproduction of biochar and sugar, J. Clean. Prod. 253 (2020)
torrefaction whereas conventional torrefaction requires milling the 119944.
feedstocks before. The grindability of torrefied samples is sub- [13] S. Negi, et al., Torrefaction: a sustainable method for transforming of agri-
wastes to high energy density solids (biocoal), Rev. Environ. Sci. Biotechnol.
stantially improved with the grinding energy consumption reduced 19 (2) (2020) 463e488.
from 386 to 10.5 kWh t1. These results demonstrate the superior [14] A. Panahi, et al., Combustion details of raw and torrefied biomass fuel particles
properties of large particles after SHS torrefaction for easier with individually-observed size, shape and mass, Combust. Flame 207 (2019)
327e341.
handling, transportation, storage and as feedstock for thermal n~ ez-Rodríguez, et al., Effect of torrefaction conditions on greenhouse crop
[15] I. Ia
processes. residue: optimization of conditions to upgrade solid characteristics, Bioresour.
Technol. 244 (2017) 741e749.
[16] Y. Niu, et al., Biomass torrefaction: properties, applications, challenges, and
CRediT authorship contribution statement economy, Renew. Sustain. Energy Rev. 115 (2019) 109395.
[17] L. Li, et al., Uncovering structureereactivity relationships in pyrolysis and
Qi Niu: Conceptualization, Software, Investigation, Data cura- gasification of biomass with varying severity of torrefaction, ACS Sustain.
Chem. Eng. 6 (5) (2018) 6008e6017.
tion, Formal analysis, Writing e original draft, Writing e review &
[18] X. Chen, et al., Effects of aqueous phase recirculation in hydrothermal
editing. Frederik Ronsse: Writing e original draft, Writing e re- carbonization of sweet potato waste, Bioresour. Technol. 267 (2018) 167e174.
view & editing. Zhiyong Qi: Software, Writing e review & editing. [19] Y.-L. Lin, N.-Y. Zheng, C.-H. Hsu, Torrefaction of fruit peel waste to produce
environmentally friendly biofuel, J. Clean. Prod. 284 (2021) 124676.
Dongdong Zhang: Conceptualization, Methodology, Formal anal-
[20] W.-H. Chen, et al., Progress in biomass torrefaction: principles, applications
ysis, Writing e original draft, Writing e review & editing, Funding and challenges, Prog. Energy Combust. Sci. 82 (2021) 100887.
acquisition. [21] W.A. Campbell, A. Coller, R.W. Evitts, Comparing severity of continuous tor-
refaction for five biomass with a wide range of bulk density and particle size,
Renew. Energy 141 (2019) 964e972.
Declaration of competing interest [22] A. Trubetskaya, et al., The effect of particle size, temperature and residence
time on the yields and reactivity of olive stones from torrefaction, Renew.
Energy 160 (2020) 998e1011.
The authors declare that they have no known competing
[23] J.H. Peng, et al., A study of particle size effect on biomass torrefaction and
financial interests or personal relationships that could have densification, Energy Fuels 26 (6) (2012) 3826e3839.
appeared to influence the work reported in this paper. [24] Z. Wang, et al., Effects of temperature and particle size on biomass torre-
faction in a slot-rectangular spouted bed reactor, Bioresour. Technol. 244 (Pt
1) (2017) 281e288.
Acknowledgements [25] P. Basu, S. Rao, A. Dhungana, An investigation into the effect of biomass
particle size on its torrefaction, Can. J. Chem. Eng. 91 (3) (2013) 466e474.
[26] L.E. Arteaga-Pe rez, et al., Steam torrefaction of Eucalyptus globulus for pro-
This work was supported by China Scholarship Council
ducing black pellets: a pilot-scale experience, Bioresour. Technol. 238 (2017)
(201706250029). The Project Sponsored by the Scientific Research 194e204.
Foundation for the Returned Overseas Chinese Scholars, Depart- [27] D. Zhang, et al., Fuel properties and combustion behaviors of fast torrefied
pinewood in a heavily loaded fixed-bed reactor by superheated steam, Bio-
ment of Human Resources and Social Security of Sichuan Province,
resour. Technol. 342 (2021) 125929.
China, and Technology Innovation Program of Chinese Academy of [28] X. Shi, F. Ronsse, J.G. Pieters, Space-time integral method for simplifying the
Agricultural Sciences (ASTIP-CAAS 34-IUA-04 and ASTIP-IUA- modeling of torrefaction of a centimeter-sized biomass particle, J. Anal. Appl.
2021005). Pyrol. 124 (2017) 486e498.
[29] S. Yu, et al., Improving energy density and grindability of wood pellets by dry
torrefaction, Energy Fuels 33 (9) (2019) 8632e8639.
Appendix A. Supplementary data [30] A. Sluiter, et al., Determination of Structural Carbohydrates and Lignin in
Biomass, Laboratory analytical procedure, 2012, pp. 1e16.
[31] M. Phanphanich, S. Mani, Impact of torrefaction on the grindability and fuel
Supplementary data to this article can be found online at characteristics of forest biomass, Bioresour. Technol. 102 (2) (2011)
https://doi.org/10.1016/j.renene.2021.12.070. 1246e1253.
[32] Q. Guo, et al., Optimal strategy for clean and efficient biomass combustion
based on ash deposition tendency and kinetic analysis, J. Clean. Prod. 271
References (2020) 122529.
[33] L. Wang, J.E. Hustad, M. Grønli, Sintering characteristics and mineral trans-
[1] Q. Niu, et al., Comparative study of different algae pyrolysis using photoion- formation behaviors of corn cob ashes, Energy Fuels 26 (9) (2012) 5905e5916.
ization mass spectrometry and gas chromatography/mass spectrometry, [34] M.-C. Popescu, et al., Structural modifications of Tilia cordata wood during
J. Anal. Appl. Pyrol. 155 (2021) 105068. heat treatment investigated by FT-IR and 2D IR correlation spectroscopy,

562
Q. Niu, F. Ronsse, Z. Qi et al. Renewable Energy 185 (2022) 552e563

J. Mol. Struct. 1033 (2013) 176e186. J. Anal. Appl. Pyrol. 152 (2020) 104946.
€ Ozgenç,
[35] O. € et al., Determination of chemical changes in heat-treated wood [42] F. Abnisa, W.M.A.W. Daud, J.N. Sahu, Pyrolysis of mixtures of palm shell and
using ATR-FTIR and FT Raman spectrometry, Spectrochim. Acta Mol. Biomol. polystyrene: an optional method to produce a high-grade of pyrolysis oil,
Spectrosc. 171 (2017) 395e400. Environ. Prog. Sustain. Energy 33 (3) (2014) 1026e1033.
[36] H. Yang, et al., Thermogravimetric AnalysisFourier Transform infrared [43] B. Mi, et al., Investigating pyrolysis and combustion characteristics of torrefied
analysis of palm oil waste pyrolysis, Energy Fuels 18 (6) (2004) 1814e1821. bamboo, torrefied wood and their blends, Bioresour. Technol. 209 (2016)
[37] M. Safar, et al., Catalytic effects of potassium on biomass pyrolysis, combus- 50e55.
tion and torrefaction, Appl. Energy 235 (2019) 346e355. [44] B. Arias, et al., Influence of torrefaction on the grindability and reactivity of
[38] D.A. Granados, et al., Study of reactivity reduction in sugarcane bagasse as woody biomass, Fuel Process. Technol. 89 (2) (2008) 169e175.
consequence of a torrefaction process, Energy 139 (2017) 818e827. [45] W.-H. Chen, P.-C. Kuo, Torrefaction and co-torrefaction characterization of
[39] W.-H. Chen, et al., Non-oxidative and oxidative torrefaction characterization hemicellulose, cellulose and lignin as well as torrefaction of some basic con-
and SEM observations of fibrous and ligneous biomass, Appl. Energy 114 stituents in biomass, Energy 36 (2) (2011) 803e811.
(2014) 104e113. [46] A. Khelfa, et al., Influence of some minerals on the cellulose thermal degra-
[40] Q. Guo, et al., Optimal strategy for clean and efficient biomass combustion dation mechanisms, J. Therm. Anal. Calorim. 92 (3) (2008) 795e799.
based on ash deposition tendency and kinetic analysis, J. Clean. Prod. (2020) [47] X. Lv, et al., Low-temperature torrefaction of phyllostachys heterocycla cv.
271. pubescens: effect of two torrefaction procedures on the composition of bio-oil
[41] Q. Niu, et al., Exploring catalytic pyrolysis of Palm Shell over HZSM-5 by gas obtained, ACS Sustain. Chem. Eng. 5 (6) (2017) 4869e4878.
Chromatography/mass spectrometry and photoionization mass spectrometry,

563