Furfural production from xylose + glucose feedings and simultaneous N 2-stripping

Article abstract of DOI:10.1039/c2gc36092f

The current furfural manufacturing process is based on homogeneous catalysts as well as steam as the stripping agent. Novel xylose-dehydration research studies include heterogeneous catalysts with high acidity and tailored selectivity. This work aims to evaluate the effect of additional glucose with the xylose feeding during simultaneous N₂-stripping of furfural catalyzed by ion-exchange resins. Given the low batch performance of Amberlyst 70, the N₂-stripping data showed high furfural yields and selectivity in the condensate stream. The different continuous feeding configurations showed that xylose/glucose ratios similar to the real pentosan-rich biomass could be fed achieving furfural yields of 75% at 200 °C. Moreover, the proposed study serves as a preliminary study to achieve high xylose conversion and relatively low glucose dehydration rates, showing its potential as a possible future process for the upgrading of carbohydrates to furan-based fuel additives.

Full text of DOI:10.1039/c2gc36092f

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Green Chemistry
Cite this: Green Chem., 2012, 14, 3132
www.rsc.org/greenchem
PAPER
Furfural production from xylose + glucose feedings and simultaneous
N -stripping
2
I. Agirrezabal-Telleria,* J. Requies, M. B. Güemez and P. L. Arias
Received 14th July 2012, Accepted 20th August 2012
DOI: 10.1039/c2gc36092f
The current furfural manufacturing process is based on homogeneous catalysts as well as steam as the
stripping agent. Novel xylose-dehydration research studies include heterogeneous catalysts with high
acidity and tailored selectivity. This work aims to evaluate the effect of additional glucose to the xylose
feeding during simultaneous N -stripping of furfural catalyzed by ion-exchange resins. Given the low
2
batch performance of Amberlyst 70, the N -stripping data showed high furfural yields and selectivity in
2
the condensate stream. The different continuous feeding configurations showed that xylose/glucose ratios
similar to the real pentosan-rich biomass could be fed achieving furfural yields of 75% at 200 °C.
Moreover, the proposed study serves as a preliminary study to achieve high xylose conversion and
relatively low glucose dehydration rates, showing its potential as a possible future process for the
upgrading of carbohydrates to furan-based fuel additives.
1
. Introduction
environmental and economical issues related to FUR production
show that green chemicals should be based not only on renew-
able biomass feedstock, but also on green production processes.
Current novel research studies are mainly focused on using
heterogeneous catalysts with adequate surface acidity and
textural properties in order to increase the final FUR yield.
Furfural (FUR) is a versatile building-block produced from
pentosan-rich biomass. It is considered as a selective solvent for
organic compounds and as a precursor to produce new gener-
1
ation P-series fuels or furan resins. It is currently produced from
9,10
11
the hydrolysis of the hemicellulosic fraction of the biomass.
During the initial stage of FUR production, the xylans generate
pentose carbohydrates, which are further cyclodehydrated to
FUR. Commercially, FUR is produced using sulfuric acid as a
homogeneous catalyst as well as a high steam to product ratio in
Modified mesoporous supports,
zeolitic structures or differ-
12
ent oxides are just a few examples of the broad range of hetero-
geneous acid catalysts tested during xylose dehydration. These
catalysts showed high xylose conversion and FUR selectivity
13–15
values, as well as high stability in repetitive runs.
All of
2
order to strip the FUR and avoid its further degradation.
these studies were based on using xylose (X) as the single model
molecule fed to the experimental reactor.
However, the enhanced secondary-reactions in the acid-catalyzed
reaction medium lead to maximum yields of 50%. Moreover,
these conditions imply several drawbacks: high vapour product
dilution leading to expensive purification stages, safety issues
and environmental problems due to toxic waste effluents. For
these reasons, the improvement of appropriate chemical technol-
ogy remains of great interest for the growth of furan-based
industry.
However, it has to be noted that the current production process
still makes use of raw biomass containing the whole cellulosic +
5
hemicellulosic fraction. When an acid catalyst is set in contact
with the raw biomass, the main polymers are hydrolyzed to all
types of sugar monomers, leading to parallel simultaneous reac-
16
tions. The different studies on pentosan-rich biomass showed
17,18
high FUR yields.
However, the sole interference that C6
Several process-optimization studies, such as rapid FUR
sugars, such as glucose (G), might have on the final FUR yield
was never reported. Indeed, this might be one of the most impor-
tant limitations when evaluating the feasibility of the reported
systems at an industrial scale.
3
4
quenching at high reaction-temperatures or microwave-heating,
showed considerable improvement to the current manufacturing
yield. However, most of the current FUR manufacturing is still
based on low-temperature steam-stripping processes. Different
reports have been also published using novel solvents such as
5
In a recent publication, the use of nitrogen as the stripping
agent during the production of FUR from xylose catalyzed by
6
7
8
19
dimethylsulfoxide, ionic-liquids or supercritical CO . Even if
promising results were achieved using these systems, several
2
an ion-exchange sulfonic resin (A70) was reported. This work
showed remarkable results concerning high FUR yield values as
well as high product purity in the stripped stream. Moreover, the
developed method showed that the catalyst could be readily sep-
arated and the nitrogen recycled with few post-treatment steps.
For these reasons, this work aims to study the effect of
glucose on the production of FUR from xylose (using the same
Department of Chemical and Environmental Engineering, Engineering
School of the University of the Basque Country (EHU-UPV), Alameda
Urquijo s/n, 48013 Bilbao, Spain. E-mail: iker.aguirrezabal@ehu.es;
Fax: +34 946014179; Tel: +34 617912295
3132 | Green Chem., 2012, 14, 3132–3140
This journal is © The Royal Society of Chemistry 2012

View Article Online
carbohydrate ratio as in the real biomass feeds) combined with
simultaneous N -stripping. In general terms, the main goal of
2
this work is to develop a green FUR production process, by com-
bining water as a solvent, a heterogeneous catalyst (A70), renew-
able sugars (X + G) and recyclable nitrogen. The study will be
divided into three different N -stripping systems: semi-continu-
2
ous, continuous and continuous + draining. Moreover, the effect
of the temperature on the final xylose conversion and FUR selec-
tivity values will be also evaluated. Finally, the experimental
data will be modelled based on the kinetic data gathered for
X + G feedings.
2. Materials and methods
2
.1 Kinetic and N -stripping tests
2
The xylose dehydration tests using X and X + G as feeds were
carried out under two operating systems, using an A70/X ratio
0
of 0.6:
1
. Batchwise:
. N
W: using water as a solvent
WT: using water–toluene systems (1 : 1 v/v)
-stripping (water): SC: under semi-continuous operation
C: continuous X or X + G feeding and FUR
stripping
2
Scheme 1 The N -stripping set-up used during this study.
2
2
using a Gilson 305 pump. The stripped streams were continu-
ously weighted and the reactor volume was set constant by
adjusting the stripping rate with respect to the inlet flow. During
the CD tests, the same procedure as in the C tests was used,
except that a continuous liquid flow set at 2 mL min− was
drained from the reactor in order to operate at a nearly steady-
state regime.
CD: continuous X or X + G feeding, FUR
stripping and liquid draining from the reactor
1
Xylose dehydration batch tests were performed in a 200 mL
temperature-controlled stain-steel reactor stirred at 500 rpm
Autoclave Engineers). The system was pressurized at 20 bar in
(
order to keep the reaction medium in the liquid-phase. Once the
reaction solvent was heated up until the temperature of study, the
corresponding X solution was pumped with a Waters 515 pump
2
.2 Product analysis
−1
to obtain an X of 30 g L . Liquid samples were taken at
Reagents and products in the reactor and in the condensate were
0
specific time intervals using a manual needle valve.
quantified using an HPLC module ICS-3000 from Dionex
The N -stripping tests were carried out in a 2 L reactor
Autoclave Engineers), with controlled electric-heating and
coupled to an AS40 autosampler. X and G were quantified using
2
a CarboPac PA20 3 × 150 mm column, at 30 °C and 0.5 mL min−
1
,
(
stirred at 500 rpm (Scheme 1). In a typical SC experiment oper-
ating at 175 °C and 8 bar, the reactor was first loaded with the
corresponding amount of A70 and heated up to 190 °C with
deionized water (75% of the total initial reactor volume). The
rest 25% of the total reactor volume (containing fixed compo-
sitions of X) was fed from a nitrogen-pressurized vessel
using 8 mM of NaOH as the mobile phase. Detection
was performed using an electrochemical cell, with integrated
amperometry and the Standard Carbohydrate Quad method.
FUR and HMF were quantified using a Kinetex C18-XB 150 ×
4.6 mm column from Phenomenex at 40 °C and coupled to a
UV-2070 Plus detector from Jasco at 280 nm wavelength. The
−
1
−1
(
to reach an X of 30 g L ). This set-up allowed the X solution
mobile phase was pumped at 1 mL min and consisted of
0
to be held at room temperature until the desired temperature was
reached in the reactor and minimize the initial X degradation.
0.01 M H
Conversion of X (XCONV), G (GCONV), selectivity of FUR
(FUR ) and HMF (HMF ), FUR yield (FUR ) and HMF yield
(HMF
2 4
SO and 10% v/v acetonitrile aqueous solution.
During the N -stripping tests, mass-flow controlled nitrogen was
S
S
Y
2
−1
bubbled into the liquid bottom at 150 mL min (STP). This gas
flow stripped the water–furfural vapour stream. The vapour flow
was later fed to a condenser (cooled by the Peltier effect at
Y
) were calculated using the equations below:
mol of xylose at time t
XCONVð%Þ ¼ 1 ꢀ
1
0 °C), where gas and liquid streams were separated. The con-
initial mol of xylose
densate was continuously weighted. Automatic control valves
were used to regulate the reactor-pressure and the liquid level in
the condenser.
mol of FUR at time t
FUR ð%Þ ¼
S
mol of converted xylose
The C tests were performed as follows: the reactor was loaded
with water and the corresponding A70 load. For an experiment
at 175 °C, the reactor was pressurized at 8 bar, heated up and
FURY ð%Þ ¼ XCONV ꢁ FURS
−1
nitrogen bubbled at 150 mL min . The reactor stripping-rate
was stabilized by stripping 0.3 L of the reactor volume. At this
moment, the corresponding X or X + G feedings were pumped
mol of glucose at time t
GCONVð%Þ ¼ 1 ꢀ
initial mol of glucose
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3132–3140 | 3133

View Article Online
21
mol of HMF at time t
neglected). However, the presence of FUR and X under acid-
HMFSð%Þ ¼
catalyzed conditions led to strong FUR drop by the end of the
tests. G dehydration tests were also carried out under the same
mol of converted glucose
Y
reaction conditions (Table 1). However, the final HMF values
Y
HMFY ð%Þ ¼ GCONV ꢁ HMFS
(
∼3%) were very low in comparison to the FUR values (20%).
Y
This is a clear indication that glucose showed lower dehydration
reactivity as compared to its isomer, fructose. Heterogeneous cat-
alysts showing high HMFY have been extensively reported. It
has to be noted that final GCONV values reached up to 59%,
which clearly indicated that high amounts of humins and coke
deposits were also produced.
Secondary-products were identified by GC-MS using a
20
DB-FFAP column from Agilent, helium as a carrier gas at
1
1
mL min− and an injection volume of 1 μL. Catalyst acid-
capacity was measured using the Cation Salt Splitting Capacity
method (MTM 0230) provided by Dow Chemical. Firstly, sulfo-
nic acid sites were activated with 1 N HCl. Later the protonated
For X + G tests, an X /G ratio of 0.66 was established. This
0
0
+
groups were substituted by Na (with 1 N NaNO ) and leached
was based on the average biomass composition determined for
3
22
in the HNO form. In order to know the acid capacity, the filtrate
corncobs or bagasse, the most common raw biomass in the
FUR production. Comparing the plots in Fig. 1, the presence of
G under batch conditions increased the initial xylose conversion
due to enhanced secondary reactions, even if at 4 h no XCONV
significant differences were observed (Table 1). Final GCONV
were slightly increased with respect to pure G tests. The most
remarkable differences during X + G feedings were given for the
3
was titrated with 0.1 N NaOH.
3. Results and discussion
3
.1 Batch kinetic study catalyzed by A70
The efficiency of A70 as a catalyst during X dehydration and
FUR
reactions derived from the interaction between FUR
high G (see Section 3.2).
On the other hand, the low FUR
considerably improved using toluene as the extracting agent
(Fig. 1a and 1b). The final FUR values did not reach over 46%
Y
. This can be mainly attributed to the enhanced secondary
1
9
simultaneous SC tests was previously reported. As observed in
Fig. 1a, the Brönsted acid-sites present on the A70 surface
convert X at a high rate (almost complete conversion at 4 h), but
they also enhance the formation of secondary products via con-
densation and furfural resinification. The batchwise tests per-
formed using X as the only feed at 175 °C showed a FURY
maximum at 15% of XCONV (the calculated Weisz–Prater
modulus is below 0.3, so the mass-transfer limitation can be
R
and the
R
values obtained in W were
Y
Y
for X feed and 39% for X + G feed. It was previously reported
that FUR resinification effects in toluene using A70 as a catalyst
19
were negligible, so low FUR values are mainly achieved due
Y
R R T R R
Fig. 1 Evolution of the experimental (dots) and modelled (lines) values of X , FUR , FUR , G and HMF during the kinetic tests at 175 °C using
A70 with X (a) and X + G (b) feedings.
3134 | Green Chem., 2012, 14, 3132–3140
This journal is © The Royal Society of Chemistry 2012

View Article Online
Table 1 Conversion and yield values of the batchwise and N
75 °C
2
-stripping tests at 175 and 200 °C using X and X + G feedings
1
200 °C
X
X
X + G
X
X
X + G
X g L−
1
FUR
X
CONV
FUR
G
HMF
FUR
X
FUR
G
CONV
HMF
Y
CONV
Y
Y
CONV
Y
CONV
Y
CONV
Y
b
Batchwise
W
30 93
20
46
55
79
75
64
66
95
93
96
70
72
60
55
17
39
52
78
62
62
42
68
68
52
32
41
31
32
3
6
4
10
6
12
6
99
99
98
74
76
78
75
25
62
68
74
72
69
70
99
99
97
74
75
86
79
21
40
73
78
75
70
66
70
68
67
57
62
63
59
16
22
20
20
21
15
20
WT 30 94
SC
C
a,b
N
2
-Stripping
30 97
10 80
2
3
0
0
77
77
CD
30 69
a
b
Pressure conditions fixed according to the reaction-temperature. S.D. of batch and N
2
-stripping tests, 3 and 5%, respectively.
to side-reactions. Moreover, the presence of toluene or another
suitable co-solvent in commercial processes would add several
environmental issues related to its use, as well as extra separation
costs.
The kinetic constants for X and X + G feedings were iterated
until reaching the tolerance range specified as sum of least-
squares. The values estimated by the model were in most cases
in good agreement with the experimental data. The kinetic con-
stants derived from these equations are summarized in Table 2.
As observed for X feeding, the xylose dehydration reaction is
The kinetic data obtained from the tests performed at 200 °C
showed higher XCONV and FUR values, directly attributed to
Y
enhanced activity and reduced side-reactions, respectively. When
ruled by one dehydration constant (k′ ) and two side-reactions
1
toluene is used as an extractant, FUR could be increased up to
(k′ and k′ ). Similar kinetic constants were also applied for G
Y
2
3
6
2%. The G dehydration activity was also increased, showing
tests. Comparing the X and G systems, it is clear that the conden-
HMF around 20%. However, HMF production from G using
A70 was lower than its state-of-the-art. It seems that the sole
sation reactions for G (k′ ) are considerably increased as com-
Y
5
1
pared to k′ , leading to strong HMF drops. On the other hand,
2
Y
use of toluene shows lower extraction than other reported bi-
phasic systems.
the presence of X + G adds an extra kinetic constant (k′6)
23
derived from the reactions between G and FUR (eqn (6)),
showing a value of 0.00023 L min−
1
g
−1
gA70 at 175 °C. The
−1
kinetic activity associated with this constant might be the main
contributor for the FURY drop with respect to the X feeding.
The kinetic constants at 200 °C followed the same trend as
shown in Table 1, where mainly dehydration rates were consider-
ably increased.
3
.2 Modelling of the kinetic data
The batch kinetic data were modelled using the kinetic equations
1
9
based on the main reactions during X dehydration. Glucose
reactions were assimilated to the X dehydration:
equations for the X kinetic model
During the tests in WT another FUR extraction constant (KT)
was added (eqn (3)). It represented the diffusion of FUR through
the water–toluene interphase. The diffusion of FUR in the
water–toluene phases occurred in both directions, so the FURT/
FURR partition coefficient (R : 2.6) was also included. The K
dXR
¼
ꢀk′1XR ꢀ k′2XRFURR ðWÞ
ð1Þ
ð2Þ
dt
I
T
−1
−1
interphase showed a value of 3.7 min gA70
.
dFURR
¼
k′1XR ꢀ k′2XRFURR ꢀ k′3FURR ðWÞ
dt
3
.3 Simultaneous N
2
-stripping
dFURR
¼
k′1XR ꢀ k′2XRFURR ꢀ k′3FURR ꢀ KTFURR
dt
19
3
.3.1 Semi-continuous (SC). In a previous publication, the
FURT
effect of N -stripping during xylose dehydration catalyzed by
A70 was reported. For X of 1 wt%, FUR values of 70% were
measured, whilst even furfural–water phase separation was
2
þ KT
ðWTÞ
ð3Þ
R1
0
Y
Additional equations for the X + G kinetic model
achieved at X of 10 wt% or higher. During this work, an X /G
0
0
0
dGR
ratio of 0.66 and an X of 3 wt% were used. The SC tests aimed
to evaluate the effect of the temperature and the addition of G to
the X feeding.
0
¼
ꢀk′4GR ꢀ k′5GRHMFR ꢀ k′6GRFURR ðWÞ
ð4Þ
ð5Þ
dt
dHMFR
During the SC tests at 175 °C, the reactor volume decreased at
¼
k′4GR ꢀ k′5GRHMFR ꢀ k′3HMFR ðWÞ
−
1
dt
an average rate of 7 mL min . This rate was fairly constant
under different reaction-temperature and pressure conditions.
Tests were finished when the reactor volume reached 10% of the
initial volume. Fig. 2a represents the typical profile of X and
FUR in the reactor, as well as the concentration of FUR in the
stripped stream. During the initial stage of the process, the high
dFURR
¼
k′1XR ꢀ k′2XRFURR ꢀ k′3FURR
k′6GRFURR ðWÞ
This journal is © The Royal Society of Chemistry 2012
dt
ꢀ
ð6Þ
Green Chem., 2012, 14, 3132–3140 | 3135

View Article Online
Table 2 Kinetic constants for the W tests at 175 and 200 °C for X and X + G feedings
75 °C
k′
1
200 °C
k′
Feeding
1
k′
2
k′
4
k′
5
k′
6
1
k′
2
k′
4
k′
5
k′
6
a
Batchwise
W
X
X + G
0.00044
0.00047
0.00021
0.00024
—
0.00016
—
0.0012
—
0.00023
0.011
0.013
0.00027
0.00029
—
0.0051
—
0.0032
—
0.0020
a
−1
−1
at 175 °C and 0.005 min−1 : min−1
−1
k′
3
fixed for FUR and HMF resinification reactions: 0.00012 min
g
A70
1 4
gA70 at 200 °C.Units ⇒ k′ and k′
−
1
−1 −1
−1
A70
.
g
A70 , k′
2
, k′
5
and k′
6
: L min
g
g
R R C R R
Fig. 2 Evolution of the experimental (dot) and modelled (line) values of X , FUR and FUR , G and HMF during the SC tests at 175 °C using
A70 with X (a) and X + G (b) feedings.
XR yielded high FURR and FUR . At this point, the process
The use of mixed X + G feedings reduced the final FUR at
Y
C
showed a FUR of 60% at 175 °C, whilst the overall FUR was
175 °C (see Table 1). As mentioned in Section 3.1, the rates of
FUR degradation were enhanced when extra G was added. Even
if FUR is partially removed from the medium, the final FURY
did not reach over 52%. As observed in Fig. 2a, the modelled
data were also in good agreement with the experimental values.
Y
Y
reduced to 55% by the end of the tests (Table 1). XCONV rates
were slightly higher during SC when compared to W tests. This
might be attributed to faster kinetics when the reactor volume
was reduced. On the other hand, the use of nitrogen as the strip-
ping agent for G feedings showed very low final HMF values at
The reported WT system using A70 reached a final FUR of
Y
Y
1
75 °C. Even if the water–FUR system does not represent the
39%, which enforced the idea that nitrogen acts as a better
extracting agent than toluene at 175 °C.
real liquid–liquid–vapour equilibrium case (the biggest vapour
molar fraction in the stripped phase is nitrogen), it is remarkable
that the water–FUR equilibrium shows a vapour phase richer in
FUR than the liquid phase. On the other hand, and even if the
HMF boiling point (115 °C) is lower than FUR (160 °C), the
water–HMF equilibrium shows an inverse effect (this experi-
mental result has been validated through phase equilibrium cal-
The SC tests carried out at 200 °C showed considerable differ-
ences with respect to 175 °C, possibly due to the so-called
“entropy-effect” at high reaction temperature, leading to slower
side-reactions. In this case, the addition of G barely affected the
final FUR , and it even increased the HMF of the final product.
Y
Y
It has to be noted that the main FUR contribution was based on
Y
culations). This resulted in a [HMF] /[HMF]R ratio of 0.1,
the stripped stream and the HMF in the remaining liquid in the
C
Y
whilst [FUR] /[FUR] was ∼6–8.
reactor.
C
R
3136 | Green Chem., 2012, 14, 3132–3140
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 3 (a) Reactor sample when using X + G feeding in SC tests at 175 °C. (b) Stripped stream sample when using X + G feeding in SC tests at
1
75 °C.
One of the most important results reported dealt with the fact
oligomers or humins, which are sometimes undetectable by the
GC-MS technique. On the other hand, the GC-MS spectrum
from the condensate stream (Fig. 3b) showed high FURS
(derived from the area percentage), reaching maximum values of
that the stripped stream showed a FUR very close to 100%
S
1
9
when using xylose as the only feed. In this case, the GC-MS
analysis was also checked for X + G feedings. Fig. 3a shows the
chromatogram of the reactor sample, where FURS is around
97% FUR . One of the biggest issues related to the FUR com-
S
3
0%. Side-products such as acetaldehyde (3.9 min), formic acid
mercial processes deals with the costs of final product separation.
(6.6), acetic acid (10.3) or 1–3 dihydroxypropanone (12.7) were
In comparison to steam, N -stripping shows the advantage of
2
derived from the simultaneous condensation-reactions. It has to
be noted that most by-products in the reactor tend to form
lower product dilution, as well as relatively easy nitrogen recy-
cling with very little post-treatment. The high product-purity
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3132–3140 | 3137

View Article Online
achieved under SC conditions suggests that future applications
not able to catalyze the xylose dehydration reaction at the same
of simultaneous N -stripping would considerably reduce the
rate by the end of the tests. This affected the FUR , and thus
2
R
final separation costs.
lower FUR stripping rates were achieved. GCONV at 175 °C was
set at around 35% for all concentrations (Table 1) and similar
3
.3.2 Modelling of the SC experimental data. The SC experi-
increasing G
the continuous process at 175 °C showed high FUR
whilst the difference between X and X + G feedings was mostly
R
trends as in X feeding were observed. In general,
mental data were also modelled following the same equations as
in Section 3.2. The SC tests using the X model were based on
eqn (1), (7) and (8). When using X + G feedings, the equations
for XR (eqn (1)), GR (eqn (4)), HMFR (eqn (5)) and FURC
eqn (8)) were not altered, whilst the expression for the FURR
was substituted by eqn (9).
Y
(∼65%),
= 30 g L− . Moreover, the GC-MS analysis of
1
enhanced at X
0
the stripped stream showed that similar FUR
were also achieved (96% purity).
as in the SC tests
S
(
On the other hand, the C tests at 200 °C using 30 g L− of
X showed a different profile (Fig. 4b and 4d). Catalyst deactiv-
ation was here compensated by the exponential kinetic activity
induced by a higher temperature, showing constant XR and
FURR profiles during the tests. XCONV rates were increased, so
lower X and higher FUR were achieved. The influence of G
1
Additional equations for FUR stripping (X model)
dFURR
FURR
¼
k′1XR ꢀ k′2XRFURR ꢀ k′3FURR ꢀ KS
dt
VR
FURRðdVR=dtÞ
R
R
ꢀ
ð7Þ
ð8Þ
−1
VR
affected mostly the FUR for 30 g L feeding (Table 1). The G
Y
conversion was maintained constant during the reactions, even if
it showed a slight decrease after 4 h of reaction. Final GCONV
values were set at around 60%. As observed in Fig. 3d, the low
HMF stripping extraction efficiency, together with the high
dFURC
FURR FURCðdVC=dtÞ
¼
KS
ꢀ
dt
VC
VC
where VR (reactor volume in L) and VC (condensate volume
in L).
Additional equations for FUR stripping (X + G model)
GCONV, led to an increase of HMF all along the C tests. Even if
R
similar FUR extraction efficiency as at 175 °C was achieved,
minimized secondary reactions increased the final FUR up to
Y
7
5%. On the other hand, the tests carried out at lower X concen-
dFURR
¼
k′1XR ꢀ k′2XRFURR ꢀ k′3FURR
FURR FURRðdVR=dtÞ
−1
trations (10 and 20 g L ) showed that even FUR values of
Y
dt
7
8% could be achieved at 200 °C.
ꢀ
K′6FURRGR ꢀ KS
ꢀ
ð9Þ
It has to be noted that the main goal of the FUR manufactur-
VR
VR
ing process should be focused on increasing the XCONV and the
The kinetic constants derived from the batch W tests (Table 2)
were applied in these equations. As observed in Fig. 2, the esti-
mations derived from the model were in good agreement with
FUR as well as reducing the G conversion for further proces-
Y
sing. Taking into account the low GCONV compared to XCONV, a
partial liquid draining configuration shows interesting potentials.
This suggests that disadvantages related to the accumulation of
by-products and residues might be reduced. A process based on
the experimental data. Moreover, the N -stripping diffusion con-
2
−1
stant (K ) showed a similar value (0.049 ± 0.002 L min ) to the
S
one reported before. There were no significant differences
a partial liquid draining and simultaneous N -stripping is indi-
2
between the K values at 175 and 200 °C using X and X + G
cated in Scheme 2: the pentosan-rich biomass is first hydrolyzed,
FUR stripped and the drain separated into residual X and
G. FUR can be upgraded to methyltetrahydrofuran (MTHF) by
S
feedings, meaning that the addition of glucose generated no
additional resistance to the diffusion of FUR in the liquid–
vapour interphase. Thus, the main condition responsible for
24
the hydrogenation of 2-methylfuran. On the other hand, the
residual G could be further upgraded to HMF and other added-
value products (disubstituted furan derivatives), such as 2,5-
dimethylfuran (DMF), which can replace key petroleum-derived
achieving higher FUR at 200 °C than 175 °C was the mini-
Y
mized secondary reactions, leading to higher FURR and thus
higher FURC.
23
building-blocks.
3
.3.3 Under continuous feeding and simultaneous N -strip-
2
ping (C). The main FUR manufacturing process is still based on
the SC steam-stripping system. The advantages of using a con-
tinuous system are mainly time-related, since semi-continuous
production is limited by the biomass charge/discharge intervals.
During the C tests, X and X + G feedings were continuously
3.3.4 Under continuous feeding, simultaneous N -stripping
2
and partial liquid draining (CD). The goal of the CD tests was
to remove part of the liquid in the reactor (mainly the remaining
glucose) and evaluate its influence on the FUR production from
−1
xylose. In order to keep the same stripping-rate (7 mL min ) as
−1
−1
−1
pumped at 7 mL min using different xylose concentrations
10, 20 and 30 wt%), as shown in Table 1. The G feeding con-
centration was calculated according to the 0.66 X /G pumped
in the C tests, 9 mL min of 30 g L of X were fed and
2 mL min− of reactor liquid drained.
1
(
The evolution of the concentration profiles was similar to
the C tests, just that XCONV and GCONV values were slightly
lower (Table 1). A partial reactor draining reduced the retention
time of the sugars, decreasing the XCONV from 77% (C) to
69% (CD) for X feedings operating at 175 °C and from 60 to
55% for X + G feedings, respectively. The XCONV reduction
was reasoned by the enhanced A70 deactivation when
0
0
ratio. The product-stripping rate was fixed in order to operate
under the same reactor volume conditions. Fig. 4a and 4b show
the evolution of XR and FURR using an X feeding at 175 and
2
00 °C, respectively.
The interference of G during the C tests was negligible com-
pared to X feedings. But it affected the GCONV, mainly due to
enhanced A70 deactivation (40%). The surface acid-sites were
−1
the feeding flow was increased from 7 to 9 mL min with
3138 | Green Chem., 2012, 14, 3132–3140
This journal is © The Royal Society of Chemistry 2012

View Article Online
R R C R R
Fig. 4 Evolution of the experimental data of X , FUR and FUR , G and HMF
during the continuous feeding of 30 g L− of X catalyzed by
1
A70. (a): X feeding at 175 °C. (b): X feeding at 200 °C. (c): X + G feeding at 175 °C. (d): X + G feeding at 200 °C.
Scheme 2 Possible future process for the production of FUR and other derivatives from pentosan-rich biomass.
X + G. This effect was less obvious at CD tests at 200 °C.
with C tests (70% for X and 66% for X + G). The GCONV was
On the other hand, FUR at 175 °C showed the same trend,
barely affected operating at high temperature, suggesting that
40% of non-reacted G could be further upgraded. The HMF
Y
whilst FUR at 200 °C were comparable to the ones obtained
Y
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3132–3140 | 3139

View Article Online
generated could also be continuously drained, achieving an
KS
KT
Stripping constant for liquid–vapour mass transfer
(L min−
1
)
HMF of 20%.
Y
Comparing the C and CD tests, the inconvenience of operating
under C conditions is related to the accumulation of by-products,
whilst through partial draining an optimum reactant concen-
tration and dehydration rates could be fixed. It is true that C tests
at 200 °C and low X0 showed the highest FURY values.
Mass-transfer constant in WT systems (Units)
Acknowledgements
This work was supported by funds from the Spanish Ministerio
de Economía y Competitividad ENE2009-12743-C04-04 and
from the Gobierno Vasco (Programa de Formación de Personal
Investigador del Departamento de Educación, Universidades
e Investigación). The authors gratefully acknowledge Dow
Chemical for kindly supplying the Amberlyst 70 catalyst.
However, the feasibility of simultaneous N -stripping and X + G
2
feedings should be evaluated using CD type configurations at X
0
−1
similar around 30 g L , which could achieve a FUR of 66% at
Y
7
9% XCONV and high product purity in the condensate stream, as
well as relatively low GCONV
.
4. Conclusions
Furfural production has been studied using Amberlyst 70 as a
heterogeneous acid-catalyst under different operating configur-
References
ations. Batchwise conditions showed low FUR due to enhanced
Y
1
R. Karinen, K. Vilonen and M. Niemelä, ChemSusChem, 2011, 4, 1002–
016.
side-reactions in the presence of glucose and HMF. However,
nitrogen turned out to be an effective stripping agent. FURY
values of 75% were achieved under continuous feeding con-
ditions. Taking into account that glucose was fed in relatively
1
2 K. J. Zeitsch, First Ed. in Sugar Series, Elsevier, The Netherlands, 2000,
vol. 13, 1.
3
D. R. Arnold and J. L. Buzzard, Proc. South African Chem. Eng. Cong.,
003, 3–5.
4 R. Weingarten, J. Cho, W. C. J. Conner and G. W. Huber, Green Chem.,
010, 12, 1423–1429.
2
higher flow rates than xylose, the proposed N -stripping
2
2
configurations are promising. Moreover, high furfural selectivity
5
A. S. Mamman, J. M. Lee, Y. C. Kim, I. T. Hwang, N. J. Park,
was achieved in the condensate stream. The FUR values were
Y
Y. K. Hwang, J. S. Chang and J. S. Hwang, Biofuels, Bioprod. Biorefin.,
higher using nitrogen than toluene, which enforces the idea that
2
008, 2, 438–454.
6 A. S. Dias, M. Pillinger and A. A. Valente, J. Catal., 2005, 229,
14–423.
N -stripping shows more industrial feasibility with respect to
2
4
other biphasic water/solvent systems. Taking into account the
7
S. Lima, P. Neves, M. M. Antunes, M. Pillinger, N. Ignatyev and
low kinetic performance of A70, further N -stripping studies
2
A. A. Valente, Appl. Catal., A, 2009, 363, 93–99.
should be focused on using hydrothermally stable and selective
heterogeneous catalysts. The method developed during this work
showed promising results as a starting point to produce furan-
based fuel additives from pentosan-rich biomass.
8 Y. C. Kim and H. S. Lee, J. Ind. Eng. Chem. (Seoul, Repub. Korea),
2001, 7, 424–429.
9
I. Agirrezabal-Telleria, J. Requies, M. B. Güemez and P. L. Arias, Appl.
Catal., B, 2011, 115–116, 178.
10 A. S. Dias, M. Pillinger and A. A. Valente, Microporous Mesoporous
Mater., 2006, 94, 214–225.
1
1 C. Moreau, R. Durand, D. Peyron, J. Duhamet and P. Rivalier, Ind. Crop.
Prod., 1998, 7, 95–99.
2 R. Weingarten, G. A. Tompsett, W. C. Conner and G. W. Huber, J. Catal.,
2011, 279, 174–182.
Nomenclature
1
1
1
−
1
X0
XR
GR
Initial xylose concentration in the reactor (g L )
Xylose concentration in the reactor (g L )
Glucose concentration in the reactor (g L
−
1
3 E. Lam, E. Majid, A. C. W. Leung, J. H. Chong, K. A. Mahmoud and
J. H. T. Luong, ChemSusChem, 2011, 4, 535–541.
4 A. S. Dias, S. Lima, M. Pillinger and A. A. Valente, Catal. Lett., 2007,
114, 151–160.
−
1
)
XCONV Xylose conversion (%)
GCONV Glucose conversion (%)
15 S. Lima, M. M. Antunes, A. Fernandes, M. Pillinger, M. F. Ribeiro and
A. A. Valente, Appl. Catal., A, 2010, 388, 141–148.
A70
FURR
Amberlyst 70
FUR concentration in the water-phase in the reactor
(
16 R. O’Neill, M. N. Ahmad, L. Vanoye and F. Aiouache, Ind. Eng. Chem.
Res., 2009, 48, 4300–4306.
−1
g L )
FURT FUR concentration in the toluene phase (g L
HMFR HMF concentration in the water-phase in the reactor
17 H. D. Mansilla, J. Baeza, S. Urzua, G. Maturana, J. Villasenor and
−
1
N. Duran, Bioresour. Technol., 1998, 66, 189–193.
8 A. Chareonlimkun, V. Champreda, A. Shotipruk and N. Laosiripojana,
Bioresour. Technol., 2010, 101, 4179–4186.
9 I. Agirrezabal-Telleria, A. Larreategui, J. Requies, M. B. Güemez and
P. L. Arias, Bioresour. Technol., 2011, 102, 7478–7485.
)
1
−
1
(
g L )
1
FURC FUR concentration in the condensate stripped stream
−
1
20 G. W. Huber, J. N. Cheda, C. J. Barrett and J. A. Dumesic, Science,
005, 308, 1446–1450.
1 M. A. Vannice, Kinetics of Catalytic Reactions, Springer, 2005, p. 65.
(
g L )
2
HMFC HMF concentration in the condensate stripped stream
2
−
1
(
g L )
22 J. W. Lee, J. Y. Zhu, D. Scordia and T. W. Jeffries, Appl. Biochem.
Biotechnol., 2011, 165, 814–822.
2
FURY FUR yield (%)
HMFY HMF yield (%)
3 Y. Román-Leshkov, J. N. Chheda and J. A. Dumesic, Science, 2006, 312,
933–1937.
1
k′
Kinetic constants with respect to the A70 catalyst mass
load
2
4 J. P. Lange, E. van der Heide, B. van Buijtenen and R. Price, Chem-
SusChem, 2012, 5, 150–166.
3140 | Green Chem., 2012, 14, 3132–3140
This journal is © The Royal Society of Chemistry 2012