Chloride ions enhance furfural formation from d-xylose in dilute aqueous acidic solutions

Article abstract of DOI:10.1039/b927424c

Furfural production through traditional processes is accompanied by acidic waste stream production and high energy consumption. Modern furfural production process concepts will have to consider environmental concerns and energy requirements besides economics, moreover will have to be integrated within widened biorefinery concepts. In this paper, some particular aspects of the chemistry of d-xylose reaction to furfural are addressed, with the aim to clarify the reaction mechanism leading to furfural and to define new green catalytic pathways for its production. Specifically, reducing the use of mineral acids is addressed by the introduction of alternative catalysts. In this sense, chloride salts were tested in dilute acidic solutions at temperatures between 170 and 200 °C. Results indicate that the Cl^- ions promote the formation of the 1,2-enediol from the acyclic form of xylose, and thus the subsequent acid catalyzed dehydration to furfural. For this reason the presence of Cl^- ions led to significant improvements with respect to the H₂SO₄ case. The addition of NaCl to a 50 mM HCl aqueous solution gave 90% selectivity to furfural. Among the salts tested FeCl ₃ showed very interesting preliminary results, producing exceptionally high xylose reaction rates.

Full text of DOI:10.1039/b927424c

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/greenchem | Green Chemistry
Chloride ions enhance furfural formation from D-xylose in dilute aqueous
acidic solutions
Gianluca Marcotullio* and Wiebren De Jong
Received 4th January 2010, Accepted 30th July 2010
DOI: 10.1039/b927424c
Furfural production through traditional processes is accompanied by acidic waste stream
production and high energy consumption. Modern furfural production process concepts will have
to consider environmental concerns and energy requirements besides economics, moreover will
have to be integrated within widened biorefinery concepts. In this paper, some particular aspects
of the chemistry of D-xylose reaction to furfural are addressed, with the aim to clarify the reaction
mechanism leading to furfural and to define new green catalytic pathways for its production.
Specifically, reducing the use of mineral acids is addressed by the introduction of alternative
catalysts. In this sense, chloride salts were tested in dilute acidic solutions at temperatures between
170 and 200 ◦C. Results indicate that the Cl- ions promote the formation of the 1,2-enediol from
the acyclic form of xylose, and thus the subsequent acid catalyzed dehydration to furfural. For this
reason the presence of Cl- ions led to significant improvements with respect to the H2SO4 case.
The addition of NaCl to a 50 mM HCl aqueous solution gave 90% selectivity to furfural. Among
the salts tested FeCl3 showed very interesting preliminary results, producing exceptionally high
xylose reaction rates.
Evidently, such reaction conditions are prohibitive for any
Introduction
industrial scale application due to the high chemical utilization.
Furfural is a readily accessible chemical from lignocellulosic
Nevertheless, a deeper understanding of the reaction mechanism
biomass, being produced industrially since 1921. It has enor-
taking place under those conditions is necessary in order to
mous potential as a starting material, both in the chemical
approach the same yields under significantly milder conditions.
industry and for liquid fuel production. On the other hand,
The acid choice was simple, since it is known from literature that
environmental concerns combined with energy and waste
different acids catalyze differently the dehydration of pentoses
streams disposal costs, are among the main reasons why furfural
into furfural,2 and among strong acids HCl is reported to
production in EU and USA has been strongly hampered
be the preferable choice.1,2 Such observation suggests that the
in the last few decades. The severity required in order to
mechanism of furfural formation goes beyond the specific acid
obtain good furfural yields, both in terms of mineral acids
catalysis. Also very important is the role played by NaCl,
concentration and process temperatures, is not practical with
which besides the salting-out effect enhancing the separation
current environmental constraints and energy costs. Novel green
of furfural during distillation, was reported to increase sugars
process concepts are needed in order to minimize the carbon
reaction rate. Such effect was thought to be the combined result
footprint and waste streams related to furfural production. In
of the higher boiling temperature and of the increased activity
this way, the great potential of furfural as a biomass derived
coefficient of the acid.2,4
intermediate in the chemical and fuel industries could be
More recently it was shown that the addition of KCl, NaCl,
unlocked to the disadvantage of oil derivatives.1 To achieve this,
CaCl2, MgCl2 or FeCl3 to pure water increased the reaction
rate of both xylose and xylotriose at 180 ◦C. This effect was
a deeper comprehension of the chemistry of pentoses is needed,
as well as the establishment of new catalytic pathways for their
more pronounced for xylotriose than for xylose.5 Similar results
conversion into furfural.
were reported in recent works when treating corn stover with
Furfural can be gained relatively easily from nearly all kinds
diluted inorganic salts.6,7 A peculiar catalytic effect of FeCl3
of lignocellulosic biomass. For this reason, in the past, various
on hemicellulose hydrolysis was also evidenced in these works,
standard methods based on furfural production were established
and FeCl3 in water was shown to be significantly more effective
for the estimation of pentosans content of plant material.2
than a strong acid solution of the same pH. In particular it was
Among others, one standard method first described in 19383
shown7 that Fe3+ and Cl- ions are both responsible for this effect.
is commonly accepted,1 where 100% of the initial pentosans
Gravitis et al.8 reported the metal cations to catalyze the reaction
can be recovered as furfural when the raw material is distilled
of biomass derived carbohydrates to furfural proportionally to
atmospherically in a 12 wt% HCl solution saturated with NaCl.
their ionization potential, mentioning an increasing effectiveness
respectively for K+, Na+, Ca2+, Mg2+, Fe3+.
Furthermore metal salts, and especially metal chlorides, have
been shown to have significant effects in microwave aided
hydrolysis of chitosan,9 the hydrolysis of ethers and amines in
Delft University of Technology Process and Energy Leeghwaterstraat, 44
Delft, 2628CA, Netherlands. E-mail: g.marcotullio@tudelft.nl;
Tel: 0031626407152
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1739–1746 | 1739
©

View Article Online
near and supercritical water,10 but also on hemicellulose and
cellulose hydrolysis both in solid state11,12 and in dilute aqueous
solution.5,6,7 Some metal chlorides have also been shown to
affect the dehydration of glucose and fructose to HMF in ionic
liquids.13
This paper addresses the use of different chloride salts
for furfural production, evaluating the role of the different
electrolytes in solution in terms of xylose rate of reaction and
furfural selectivity and yield, with the aim of exploring new
catalytic pathways leading to high furfural yields with a reduced
mineral acid usage.
evident and surprisingly higher furfural yields are achieved,
see Fig. 1. Thus NaCl is confirmed4 to affect xylose reaction
rate already at relatively low concentrations in acidic solutions,
although, contrary to what was believed before,2,4 this effect can
neither be ascribed to the increased activity of the acid, nor to
the increased boiling temperature or the salting-out effect. From
the Debye-Hu¨ckel theory the ions activity coefficient g can be
related to the ionic strength I in dilute solutions according to
the equation log10g i = -Azi2I1/2, in which the expression z is
the ion charge and A is a constant dependent on temperature,
density and dielectric constant of the solvent.16 Thus, under
these conditions of dilution, increasing the salt concentration
leads to a decrease of the ions activities, and consequently to
a decreased acid activity. Furthermore, the reaction is carried
out at constant and controlled temperature in liquid solution
without any separation by steam stripping, thus boiling point
raise or salting-out effect do not play any role in this case.
In a previous work17 where H2SO4 was used as catalyst for
the production of furfural from xylose in the same range of
conditions used in this work, increasing the temperature led to
negligible variations in terms of furfural selectivity and yield.
Contrary to this result, in Fig. 2 it is shown how a significant
increase of furfural yield can be obtained raising the temperature
from 170 to 190 ◦C when dilute HCl is used instead of H2SO4.
Results and discussion
Fig. 1 shows the results of xylose dehydration in dilute aqueous
acid solutions with the addition of different amounts of NaCl.
The reaction rate of xylose in two equimolar solutions of HCl
and H2SO4 at 200 ◦C does not show a significant deviation,
nevertheless furfural yield is significantly higher in the former
case with respect to the latter. The second dissociation constant
of H2SO4 is known to decrease markedly with temperature,14,15
to such an extent that H2SO4 can be considered with very
good approximation to be monoprotic at 200 ◦C (pKa2
ꢀ
4), and consequently the H+ concentration of two equimolar
solutions of HCl and H2SO4 can be assumed to be equal at those
conditions. The different results in terms of furfural yield can
thus be ascribed to the presence of different anions in solution,
in particular Cl- having a more beneficial effect than HSO4 .
-
Fig. 2 Effect of temperature on xylose conversion and maximum
furfural yield in 150 mM HCl. Xylose (dotted line) and furfural
concentration (solid line) as from eqn (2) after fitting to the experimental
results.
Fig. 1 Effect of NaCl addition on xylose conversion and furfural yield.
Xylose (dotted line) and furfural concentration (solid line) as from eqn
(2) after fitting to the experimental results.
Taking all these results into account it can rather be postulated
that the ionic species in solution other than H+ induce a change
in the reaction mechanism, and Cl- ions seem to be the main
responsible for this effect.
In the presence of increasing amounts of NaCl in dilute aque-
ous HCl a clear increase in xylose reaction rate is immediately
1740 | Green Chem., 2010, 12, 1739–1746
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Different salts, namely KCl, NaCl, CaCl2 and FeCl3, were
also tested in 50 mM aqueous HCl, keeping a constant Cl-
molarity in every solution. Results showed very similar reaction
behavior for the three salts KCl, NaCl and CaCl2, both in terms
of xylose reaction rate and furfural yield. For these three salts
the concentration of Cl- in solution seems to be the main feature
for the kinetics of xylose reaction, the different cations seemingly
playing only a minor role, see Fig. 3.
electrolytes but rather on the total ion concentration, whereas,
by carefully considering the results shown in Fig. 1, 2 and 3, it
becomes evident that the nature of the ions is very important
even in very dilute solutions. Furthermore when comparing the
results for CaCl2 with those for KCl and NaCl we can notice that,
even if the total ions concentration in solution is 25% lower in
the first case if compared to the other two, the results are very
similar to each other. Thus a direct contribution of the ions to the
chemistry of the reaction, rather than a change of the solvent-
reactant interactions, can be considered to play the major role.
Xylose reaction in two equimolar solutions of KCl and KBr
was also studied at the same conditions of temperature and
acidity, see Fig. 3. The reaction rate was about 15% slower
in KBr with respect to KCl, even though faster than in the
absence of any added salt, while the furfural maximum yield
was comparable in both cases, see also Table 1. This last result
denotes the addition of metal halides in general to aqueous
acidic solutions to potentially have surprisingly beneficial effect
in terms of xylose conversion into furfural, with Cl- showing
to be somewhat more effective than Br-. The reasons for this
difference could include the nucleophilicity of the ions, as well
as their size and charge dispersion. Nevertheless the results
presented here are not enough to draw a clear conclusion. In
this sense further work is needed which could unequivocally
clarify the effects brought about when different ions of the series
I, Br, Cl and F are present in solution.
Reaction rate of D-xylose to furfural
Different reaction kinetic mechanisms leading from xylose to
furfural have been proposed in literature, from simple ones
to more complicated,1,2,20,21,22,23 but almost all of them can be
expressed in a simplified way, as depicted in Fig. 5. As such a
simple reaction scheme accurately fits the experimental results,
Fig. 3 Effect of different chloride salts on xylose conversion and
furfural yield. Xylose (dotted line) and furfural concentration (solid
line) as from eqn (2) after fitting to the experimental results.
A significantly different behavior was observed when adding
FeCl3, where the xylose reacted so fast that no sugar could be
detected already after a residence time of 124 s, thus, even if the
exact value could not be derived, the reaction rate appeared to
be more than 10 times larger than in the other cases. Hence the
Fe3+ ion itself can be considered to have a relevant catalytic effect
on the reaction of xylose. On the other hand, furfural maximum
yield was lower when adding FeCl3 compared to the other cases,
which accounted for 62% of the initial xylose, in agreement with
what was shown earlier when treating corn stover under very
similar conditions.7
The addition of electrolytes to the water can generally induce
significant changes by attracting or orienting its molecules, or
change its ordered short range structure and in general its
interaction with the solute. Similar effects are responsible for
the vapor pressure increase of aqueous electrolytes solutions, for
salting-out or salting-in effects,18 or colloids precipitation as first
studied by Franz Hofmeister more than a hundred years ago.19
In the same way these effects could be regarded as responsible for
the reported results in terms of xylose reaction rates and furfural
yield. Nevertheless, such effects, especially considering dilute
solutions, are normally not so dependent on the nature of the
Fig. 4 ln(kX /[H+]) plot against temperature.
Green Chem., 2010, 12, 1739–1746 | 1741
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Table 1 Condensed results from kinetic experiments
Acid/
mM Salt
Salt/ Total [Cl-]/
wt% mM
Selectivitya Maximum
Entry Acid
Temp./◦C kX /10-4[s-1] k1/10-4[s-1] k2/10-4[s-1] k3/10-4[s-1] [%]
Yieldb [%]
1
2
3
4
5
6
7
8
HCl
HCl
HCl
H2SO4
HCl
HCl
HCl
HCl
HCl
HCl
HCl
H2SO4
HCl
HCl
HCl
150
150
150
50
50
50
—
—
—
—
—
KCl
—
—
—
—
—
2.5
150
150
150
—
170
180
190
200
200
200
200
200
200
200
200
200
210
200
200
200
170
185
170
185
200
18.5
42.0
95.0
46.2
46.7
67.3
64.5
57.0
63.0
>600
74.4
78.7
175.0
119.1
196.8
66.5
11.7
39.5
9.6
13.68
31.5
77.9
32.6
37.3
54.9
52.84
46.7
51.0
>400
63.1
65.1
144.4
107.4
159.4
42.2
9.1
4.83
10.5
17.1
13.6
9.4
12.4
11.7
10.3
12.0
>200
11.3
13.6
30.6
11.6
37.4
24.3
2.6
1.5
2.5
5.3
2.0
2.2
1.7
2.2
2.2
2.0
8.0
2.2
2.0
4.0
3.4
4.3
0.2
0.6
1.1
0.4
0.9
5.1
73.9
75.0
82.0
70.6
79.9
81.6
81.9
82.0
81.0
65.7
84.8
82.7
82.5
90.2
81.0
63.5
77.7
82.2
74.5
71.2
87.7
59.2
62.7
69.1
61.4
68.7
74.1
72.8
72.0
72.3
62.0
76.1
75.3
75.5
81.3
74.4
62.1
65.7
74.3
65.6
63.7
79.2
50
390
390
390c
50
NaCl 2.0
KBr 4.1
50
50
50
50
50
50
9
CaCl2 2.5d 390
10
11
12
13
14
15
16
17
18
19
20
21
FeCl3 3.1e
NaCl 3.5
NaCl 3.5
NaCl 3.5
NaCl 5.0
390
650
600
390
906
50
50
NaCl 10.0 1762
NaCl 10.0 1712
NaCl 5.0
NaCl 5.0
—
—
Formic Acid 220
HCl
HCl
HCl
HCl
HCl
50
50
50
50
906
906
50
50
956
32.4
7.1
21.0
161.4
7.0
2.4
8.5
22.7
—
—
29.5
184.1
100
NaCl 5.0
a Calculated as k1/kX . b Maximum furfural yield calculated from (3). c Considered as [Cl-] + [Br-]. d CaCl2 dihydrate. e FeCl3 hexahydrate.
furfural can also be analytically derived as a function of the
three main kinetic parameters as:
k
3
⎡
⎤
⎥
⎥
⎥
⎥
k
−k −k
⎢
⎛
⎞
⎟
⎟
⎟
⎟
⎠
3
1 2
CF max
CX0
k1
k1 + k2
k3
⎜
⎢
(3)
=
⎜
⎜
⎢
⎜
k1 + k2
⎝
⎢
⎢
⎥
⎣
⎦
Fig. 5 D-Xylose simplified scheme of reaction.
The eqn (2) were fitted to the experimental results deriving
the kinetic parameters k1,k2,k3 by least square minimization.
The model showed very good agreement with the experimental
measurements, and the results are displayed in Table 1.
The three kinetic parameters show a similar dependence on
acid concentration, as it was also noticed earlier,17 thus no
significant improvements in terms of maximum furfural yield
canbe achievedincreasing acidconcentration within therangeof
interest for this work. Contrary to this, as it can be noticed from
Table 1, chloride salts generally tend to have a direct influence
on the first kinetic parameter k1 being much less important for
the others, leading to an increase of both xylose reaction rate
and furfural maximum yield CFmax/CX0 . From these results it
is clear that a new reaction mechanism for xylose dehydration,
different from the simple acid catalysed one, is followed in the
presence of chloride salts.
it can by used to describe the reaction. As the initial sugar
concentration is kept constant for every experiment, and rather
low (35 mM), second order reactions between furfural and
intermediates can be ruled out; the effect of increasing initial
sugar concentration is presently being studied and it will be
discussed elsewhere. Such approximation allows for describing
the reaction rates by a relatively simple analytical expression:
dC
X
dt
⎧
⎪
r
=
= −(k1 + k2 )CX
= k1CX −k1bCi
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
D−Xylose
dC
i
r
=
(1)
Intermediate
dt
dC
F
r
=
= k1bCi −k3CF
dt
Furfural
⎪
⎩
Reaction intermediate(s) have never been clearly identified
in literature, nevertheless, judging from experimental results,
their concentration can be assumed to be low and not varying
significantly with time. Hence the steady-state approximation
can be introduced for the intermediate resulting in dCi/dt =
Mechanism of furfural formation from pentoses
~
~
1
k1CX - k1bCi 0 and then dC /dt k C - k C . Introducing this
=
=
F
X
3
F
Contradictory theories exist in literature concerning the mech-
anism of furfural formation from pentoses. On the one hand,
many authors consider the reaction to proceed via the acyclic
form of the pentoses, first through a 1,2-enediol formation
and subsequent three dehydration steps yielding furfural.24,25
On the other hand, other authors believe the reaction to take
place starting from the pyranose form of the pentoses,22 which
recently was postulated to take place by the action of H+
on a specific position of the pyranose ring, leading to a first
assumption makes the integration easier and the concentration
of xylose and furfural can be written as:
+k )t
CX t = C 0 e−(k
1
2
⎧
⎪
( )
t = C
X
⎪
⎪
⎨
⎪
(2)
k
e−(k )t −e−k
+k
t
3
1
1
2
C
F ( )
⎪
(
)
(
)
X
k
−k −k
0
3
1 2
⎪
⎩
where the initial furfural concentration CF0 = 0 and CX0 is
the initial xylose concentration. The maximum molar yield of
1742 | Green Chem., 2010, 12, 1739–1746
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
dehydration and structure rearrangement yielding the furanose
ring and subsequent dehydration to furfural.20,26 It has been
demonstrated that after reacting xylose-1-14C in 12% aqueous
HCl the “aldehydic” carbon of the pentose at C-1 position
was found nearly completely at the aldehyde group of the
final product, 2-furaldehyde-a-14C.27 Such result is perfectly
aligned with the former theory, whereas it cannot be explained
according to the mechanism proposed in the latter. Furthermore,
in analogy to the reaction of xylose to furfural, an isomerization
reaction of the D-glucose to D-fructose has been shown to take
place in acidic conditions at high temperature, proving this to
be the first step in the reaction leading from D-glucose to 5-
hydroxymethyl furfural (HMF).28 Besides, ketopentoses, which
present a significantly higher proportion of acyclic form in
water solution if compared to aldopentoses,29,30 react much
faster in water yielding furfural proportionally to the acidity
of the solution.24 From all these evidences provided by different
authors throughout the years, the former theory appears to be
strongly supported, whereas the latter fails to comply with some
of the earlier works. For these reasons, in this work, the authors
accept the first theory.
A strong indication was also provided for the 1,2-enediol
intermediate to be irreversibly formed from the aldopentoses
in acidic conditions, showing on the other hand a partial
equilibration with the keto form.24 Such indication is in agree-
ment with the general aldoses reluctance to isomerization in
acidic conditions,28 and thus, since enolization reactions are
normally reversible, the 1,2-enediol formation from xylose in
acidic conditions can be considered to be the rate limiting step
in the formation of furfural. In accordance with this, HMF is
known to be formed much faster from fructose than from glucose
in acidic conditions, and fructose has often been measured in the
reaction from glucose to HMF, indicating aldose isomerization
takes place as an intermediate step.28,31
previous works. In such mechanism the presence of Cl- favors
the formation of the 1,2-enediol 2, which can equilibrate with
both the aldo 1 and keto 3 form of the sugar, and reacts to form
furfural 4 in presence of an acid.
Accordingly, it is evident from the experimental results that,
with increasing chloride salts concentration, xylose reacts signif-
icantly faster to furfural, while no significant effect is observed
on the side reactions, resulting in a remarkable selectivity
improvement. When the salt concentration becomes higher than
a certain threshold with respect to the H+ molar concentration
[H+], the increasing selectivity trend to furfural is inverted and a
small drop is observed, see Fig. 1. This can be due to the larger
concentration of the intermediate(s) which are prone to undergo
loss reactions, including possible condensation reactions with
furfural. In the simplified kinetic model proposed here this effect
results in an increase of k2, as it is observed for 10 wt% NaCl,
see Table 1 entry 15. As mentioned before, 2 and 3 are more
reactive than 1, also in absence of acids,24,31 yielding furfural
proportionally to the acidity. Accordingly, the faster xylose
reaction and the poor furfural yields obtained using a weak
acid in presence of 10 wt% NaCl (Table 1, entry 16) can be
explained due to the relatively high pH in presence of a relatively
large concentration of chloride salts. In addition, comparing the
results when using different strong acids like HCl and H2SO4
led to different results as discussed before (Fig. 1), whereas in
the same conditions, when a larger concentration of chloride
salt is also present, results are very similar (Table 1; entries 11–
12), meaning the H+ source does not significantly influence the
reaction when a larger concentration of Cl- is also present.
Hence, even if the reaction of xylose can be catalyzed
by chloride salts by favoring the 1,2-enediol formation, the
presence of a strong acid is necessary in order to favor the
selectivity to furfural. This conclusion is in agreement with what
was previously shown for ketopentoses, where higher acidity
influenced the selectivity to furfural not influencing so much the
sugars rate of reaction.24
Effect of the Cl- ion on furfural kinetics of formation
The mechanism in Fig. 6 helps also to explain the Arrhenius
plot depicted in Fig. 4. In this plot the natural logarithm of the
total xylose reaction rate kinetic parameter kX = k1 + k2 divided
by [H+] was plotted against the inverse of the temperature, in
order to show the rate of reaction of xylose excluding the effect
of acid concentration.
The bottom line in Fig. 4 shows the H2SO4 catalysed reaction,
i.e. the well known H+ catalyzed route, where formation of
2 through reaction (I) is the rate limiting step, Fig. 6. The
linear relationship between kX and [H+] can clearly be observed,
accordingly the points for different H2SO4 concentrations are
condensed on one straight line in Fig. 4. These results are also
in very good agreement with the known kinetics provided in a
previous work2 for the same range of acid concentration and
temperature.
Based on the results obtained in this work, it seems reasonable
to postulate that one particular step among the multi-step
reaction from xylose to furfural to be catalysed by both H+
and Cl-, this particular step being the formation of the 1,2-
enediol intermediate. Thus the mechanism depicted in Fig. 6
can be drawn, which appears to be in agreement with the
results obtained in this work and with what was observed in
When a relatively large Cl- concentration is present, kX /[H+]
values are larger, and accordingly they shift to higher values
on the Arrhenius plot in Fig. 4. Such shift can be explained
with a switch in the rate limiting step, namely from reaction (I),
which is bypassed due to the Cl- catalyzed formation of 2, to the
subsequent H+ catalyzed dehydration reaction (III). The 50 mM
HCl curved line is particularly revealing the transition: at the
higher temperatures it tends to the H+ catalysed mechanism (I);
Fig. 6 Mechanism of furfural formation from xylose.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1739–1746 | 1743
©

View Article Online
lowering the temperature the Cl- catalyzed reaction (II) becomes
faster, having seemingly a lower activation energy, leading to a
switch to reaction (III) becoming therefore limiting step.
When reaction (III) is the rate limiting step, the kinetic pa-
rameter kX should be directly proportional to [H+], as previously
shown, and indirectly to [Cl-] through 2 formation. Contrary to
this, no linearity is evident between kX and [Cl-] or [H+], the data
showing a marked less than proportional relationship. It can be
analytically demonstrated, and it is also qualitatively intuitive,
that such relationship indicates reaction (II) to be reversible,
contrary to (I), being the equilibrium between 1 and 2 only partly
attained. This means that the rates of reaction (II) and (III) are
comparable, thus concentration of 2 varies accordingly. Due to
this, an increased acidity at fixed [Cl-] would cause reaction (III)
to be faster, at the same time lowering to some extent the mean
concentration of 2, and resulting in a less than proportional
observed increase of xylose reaction rate. In the same way,
an increase of [Cl-] at fixed [H+] would tend to increase the
average concentration of 2, resulting again in only a less than
proportional observed increase of furfural rate of formation
because of the combined effect of reaction (III) and (II). In this
way the 150mM HCl straight line intermediate position in Fig. 4
can also be explained.
For these conditions the competing side reactions also become
more complex, involving 1, 2 and possibly 3, very reactive
ketopentoses at lower acid concentrations as mentioned before.
The peculiar Cl- behavior in this reaction is particularly
interesting by itself. As it has been shown, there is good reason
to believe Cl- ions to act as a catalyst in the enolization reaction
of xylose, which is normally catalyzed in basic environment
even at lower temperatures. Alternatively the Cl- ions could
be considered to favor the presence of the 1,2-enediol with
respect to both the aldose and ketose by a stabilizing effect,
thus favoring the subsequent dehydration reactions starting from
such form. Anyway, based on the results presented in this work,
the 1,2-enediol promoting effect induced by the Cl- remains
an hypothesis consistent with the reaction kinetics of xylose to
furfural, although no ultimate evidence is provided in support of
this theory. This hypothesis, as well as a more general extension
to other halides like I, Br and F, is to be verified with further
investigation.
As far as the peculiar effect of FeCl3 is concerned, not
enough data are available to point out how Fe3+ could enter
the mechanism proposed in Fig. 6. From the data available, and
considering the acidic character of Fe3+, it can be assumed to
catalyze reaction (I) or (III) in a similar fashion like H+ but
much more vigorously, although a different mechanism cannot
be excluded. In terms of selectivity to furfural the data here
reported are also not deviating very much from the specific H+
catalyzed reaction, being just moderately lower. The catalytic
strength of Fe3+ in water, or its hydrated forms, is indeed of great
interest and worth more investigation in the future.
Contrary to the dehydration reactions yielding furfural from
pentoses and HMF from hexoses, which require strong acids, al-
doses isomerization is normally favored by basic conditions,28,31
and proved to proceed via the 1,2-enediol formation.32 More-
over, in the past, it was also observed that furfural formation
from pentoses was not a simple dehydration reaction, since only
poor furfural yields were obtained when xylose was reacted with
dehydrating catalysts like pure ZnCl2 or phosphorous pentoxide,
whereas, when an aqueous solution of ZnCl2 was employed,
furfural yield approached that obtained using strong acids
solutions.33 Thus a discrepancy in terms of “catalytic require-
ments” becomes evident in the process of furfural formation
from pentoses, presenting in turn an enolization followed by
three dehydration reactions, being the two reactions normally
favored by different catalysts.
Thus, when thinking about industrial furfural production, it
should be kept in mind that the high severity normally required
in terms of acid concentration and temperature is mainly needed
to overcome the first enolization step as it is normally not favored
by acid, while acid is indeed needed for the following dehydration
steps. Thus, in order to achieve better furfural yields at milder
conditions and in a more environmental friendly manner, the
combination of two different catalysts, having in turn basic and
acidic character, would probably be the key. A recent remarkable
work34 is based on a similar observation where the combined use
of acid and basic solid catalysts in a one-pot reaction allowed for
exceptional HMF yields from both glucose and fructose under
relatively mild conditions.
In the present work it has been shown that such double
catalytic effect can seemingly be achieved by the addition of
metal chlorides to aqueous acidic solutions, allowing for high
furfural yields and relatively high xylose reaction rates. In terms
of furfural yield maximization an optimum for H+ and Cl-
concentrations exists, also in relation with the loss reactions,
involving a relatively large Cl- excess and a relatively low acid
concentration. Noteworthy seawater typical composition with
the addition of a small amount of strong acid could be a suitable
solvent for furfural production.
Conclusions
Chloride salts were shown to enhance the reaction of xylose
to furfural in aqueous acidic solution at temperatures between
◦
170 and 200 C. Experimental results indicated that the Cl-
ions promote the formation of the 1,2-enediol from the acyclic
form of the aldose, which undergoes subsequent acid catalyzed
dehydration reactions to furfural.
The effect caused by the chloride salts addition to aqueous
acid solutions was studied under various conditions. Significant
improvements were observed with respect to the H2SO4 -based
case, furfural yield and selectivity respectively increasing by 18%
and 28%, selectivity in particular attaining 90%. At the same
time a 4-fold xylose reaction rate increase was possible by 1.7 M
NaCl addition keeping the acid concentration as low as 50mM
(0.18 wt% HCl).
Among the salts tested FeCl3 showed very interesting prelimi-
nary results, producing exceptionally high xylose reaction rates.
Such result can be explained by the combined effect of the Cl-
ions and the acidic character of the Fe3+, even though more
investigation is needed in order to clarify this aspect.
Experimental
The experimental setup consists of a coiled tube reactor
immersed in a thermostatic oil bath. The reactor is made of
titanium grade 2 (Merinox, Alblasserdam, NL) to ensure high
1744 | Green Chem., 2010, 12, 1739–1746
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
resistance to corrosion and minimal catalytic effects. The tube
reactor has an external diameter of 3.2 mm, internal diameter
of 1.7 mm, and a length of 4.40 m. The reactants are fed to
the reactor by an HPLC pump, while the reaction temperature
is precisely controlled and kept constant by the thermostatic oil
bath. The reactants solution enters the reactor at the set pressure
and is rapidly brought to the desired temperature by means of
an electrical pre-heater situated upstream the oil bath. The pre-
heater consists of an aluminium block, thermally insulated from
the environment, which encloses the tube reactor for a length of
approximately 15 cm and is equipped with an electrical heating
module and relative electronic control. Downstream the reactor
is a cooling double pipe heat exchanger fed with tap water, a
filter and a back pressure regulator. The reaction products are
collected and analyzed as such. Reaction pressure is set at 60
bar, significantly larger than the water vapor pressure at the set
temperatures, to ensure complete liquid operation. Such reactor
resembles with very good approximation an ideal plug flow
reactor and it allows liquid phase reaction kinetics studies to
analytically derived from (2), being selectivity S = (k1/k1 + k2)
and furfural yield expressed as from eqn (3).
List of abbreviations
HMF
wt
5-hydroxymethyl Furfural
Weight
pKa2
Second acid dissociation constant
Ionic strength
Ion charge
I
z
g
Ions activity coefficient
[H+], [Cl-] H+, Cl- molarity [mol/litre]
mM
M
millimolar [mmol/litre]
molarity [mol/litre]
HPLC
High Performance Liquid Chromatography
Acknowledgements
This study has been done in the framework of the Marie
Curie EST project INECSE (MEST-CT- 2005-021018) and the
European FP6 Integrated Project Biosynergy (038994-SES6),
which are gratefully acknowledged. The authors would like to
thank Michel van den Brink for the precious support in all
laboratory related issues, and Miguel Tavares Cardoso for all
the fruitful talks.
◦
be performed under conditions up to 250 C and 250 bar and
to precisely control residence times between 100 to 5000 s. A
thorough description of the reactor and its main characteristics
is available elsewhere.35
Materials and methods
Xylose reagent grade (Sigma-Aldrich, ≥99%) was used as model
compound in the experiments, and for HPLC calibration.
Furfural reagent grade (Sigma-Aldrich, 99%) was used for
HPLC calibration after distillation under reduced pressure (50
mbar) for further purification. Pure formic acid, HCl 40%wt
water solution, H2SO4 ≥ 97.5% and all inorganic salts (NaCl,
KCl, CaCl2·2H2O, FeCl3·6H2O) were purchased from Sigma-
Aldrich. The acid concentration varied between 50 and 100 mM,
except for formic acid which concentration was 220 mM. Xylose
concentration for every experiment was 35 mM. Acidic aqueous
solutions of xylose were prepared with different amounts of
inorganic salts, and fed to the tube reactor after stirring to ensure
a complete dissolution of the solids.
References
1 K. J. Zeitsch, The chemistry and technology of furfural and its many
by-products, Sugar Series, Elsevier, 2000vol. 13.
2 A. P. Dunlop and F. N. Peters, The Furans, ACS Monograph Series,
Reinhold Publishing Corporation, New York, 1953.
3 E. E. Hughes and S. F. Acree, Journal of Research of the National
Bureau of Standards, 1938, 21, 327–336.
4 E. I. Fulmer, L. M. Christensen, R. M. Hixon and R. L. Foster,
J. Phys. Chem., 1936, 40.
5 C. Liu and C. E. Wyman, Carbohydr. Res., 2006, 341, 2550–2556.
6 L. Liu,, et al., Bioresour. Technol., 2009, 100, 5853–5858.
7 L. Liu,, et al., Bioresour. Technol., 2009, 100, 5865–5871.
8 J. Gravitis, N. Vedernikov, J. Zandersons and A. Kokorevics, ACS
Symp. Ser., 2001, 784, 110–122.
9 R. Xing,, et al., Carbohydr. Res., 2005, 340, 2150–2153.
10 L. A. Torry, R. Kaminsky, M. T. Klein and M. R. Klotz, J. Supercrit.
Fluids, 1992, 5, 163–168.
11 N. Shimada, H. Kawamoto and S. Saka, Carbohydr. Res., 2007, 342,
1373–1377.
HPLC analysis
Analysis of the reaction products was carried out by means
of an HPLC apparatus equipped with a Resex ROA-Organic
acid column, 8% cross linked H+, 300 ¥ 7.80 mm, (Phenomenex
Inc., Torrance, CA, USA). A Marathon XT auto-sampler (Sep-
arations, Ambacht, NL) was used to enhance reproducibility.
Xylose was quantified by means of a Refractive Index detector
(Varian Model 350), whereas for furfural both Refractive Index
detector and UV detector (Varian Model 310 Pro Star) were
used. 10 mM sulfuric acid solution in demineralized water was
used as the eluent at a flow rate of 0.8 mL min-1 with a column
temperature of 80 ◦C and a run time of 33 min.
12 A. S. Amarasekara and C. C. Ebede, Bioresour. Technol., 2009, 100,
5301–5304.
13 H. Zhao, J. E. Holladay, H. Brown and Z. C. Zhang, Science, 2007,
316, 1597–1600.
14 A. G. Dickinson, D. J. Wesolowsky, D. A. Palmer and R. E. Mesmer,
J. Phys. Chem., 1990, 94, 7978–7985.
15 W. L. Marshall and E. V. Jones, J. Phys. Chem., 1966, 70, 4928–4040.
16 W. J. Moore, Physical Chemistry, Longman Group Ltd., 1974, fifth
edn.
17 G. Marcotullio, H. J. Heidweiller and W. De Jong, Reaction kinetic
assessment for selective production of furfural from C5 sugars
contained in biomass, in Proceedings of 16th European Biomass
Conference and Exhibition, Valencia, Spain 2008.
18 J. Prausnitz and J. Targovnik, Chem. Eng. Data Ser., 1958, 3, 234–239.
19 W. Kunz, J. Henle and B. Ninham, Curr. Opin. Colloid Interface Sci.,
2004, 9, 19–37.
20 L. J. Antal Jr, T. Leesomboon, W. S. Mok and G. N. Richards,
Carbohydr. Res., 1991, 217, 71–85.
21 A. P. Dunlop, Ind. Eng. Chem., 1948, 40, 204–209.
22 E. R. Garrett and B. H. Dvorchik, J. Pharm. Sci., 1969, 58, 813–820.
23 D. F. Root, J. F. Saeman, J. F. Harris and W. K. Neill, Forest Prod. J.,
1959, 9, 158–165.
Experimental data processing
Experimental results were fitted to eqn (2) and the kinetic
parameters k1,k2,k3 derived by least square minimization using
the Solver tool embedded in the commercial software Microsoft
ExcelTM (2003). Selectivity to furfural and yield could also be
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1739–1746 | 1745
©

View Article Online
24 T. Ahmad, L. Kenne, K. Olsson and O. Theander, Carbohydr. Res.,
1995, 276, 309–320.
25 M. S. Feather, D. W. Harris and S. B. Nichols, J. Org. Chem., 1972,
37, 1606–1608.
26 R. M. Nimlos, X. Quian, M. Davis, M. E. Himmel and D. K.
Johnson, J. Phys. Chem. A, 2006, 110, 11824–11838.
27 W. A. Bonner and M. R. Roth, J. Am. Chem. Soc., 1959, 81.
28 D. W. Harris and M. S. Feather, Carbohydr. Res., 1973, 30, 359–365.
29 J. Wu, A. S. Serriani and T. Vuorinen, Carbohydr. Res., 1990, 206,
1–12.
30 K. N. Drew, J. Zajicek, G. Bondo, B. Bose and A. S. Serriani,
Carbohydr. Res., 1998, 307, 199–209.
31 M. Watanabe,, et al., Carbohydr. Res., 2005, 340, 1925–1930.
32 S. J. Eitelman and D. Horton, Carbohydr. Res., 2006, 341, 2658–
2668.
33 C. D. Hurd and L. L. Isenhour, J. Am. Chem. Soc., 1932, 54, 317–330.
34 A. Takagaki, M. Ohara, S. Nishimura and K. Ebitani, Chem.
Commun., 2009, 6276–6278.
35 G. Marcotullio, M. A. T. Cardoso, W. De Jong and A. H. M.
Verkooijen, Int. J. Chem. React. Eng., 2009, 7, A67.
1746 | Green Chem., 2010, 12, 1739–1746
This journal is
The Royal Society of Chemistry 2010
©