Kinetics of Formic Acid-Catalyzed Cellulose Hydrol
Kinetics of Formic Acid-Catalyzed Cellulose Hydrol
Kinetics of Formic Acid-Catalyzed Cellulose Hydrol
net/publication/274931667
CITATIONS READS
7 718
3 authors, including:
Some of the authors of this publication are also working on these related projects:
PREBIO - High performance pretreatment of lignocellulosic biomasses for sustainable biorefinery value chains View project
Value adding of forest biomass – Raw materials for polymers from side products of biorefining View project
All content following this page was uploaded by Laura Kupiainen on 10 December 2015.
Keywords: Dilute sulfuric acid; Cellulose hydrolysis; Kinetic modelling; Reaction kinetics
Contact information: Department of Process and Environmental Engineering, University of Oulu, P.O.
BOX 4300, FI-90014 University of Oulu, Finland; *Corresponding author: juha.ahola@oulu.fi
INTRODUCTION
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2645
PEER-REVIEWED ARTICLE bioresources.com
dilute sulfuric acid (Karimi et al. 2006), and enzymes (Taherzadeh and Karimi 2007)
have been studied for cellulose hydrolysis; the last two examples are the most studied
catalysts. In supercritical conditions, reaction times are of a few seconds, which hinders
the control of processing solid raw material. Under milder conditions, e.g., enzymatic
hydrolysis, a pretreatment step, long reaction times, and a series of expensive and even
enzymes derived from genetically modified organisms are required (Huan et al. 2011).
Acid-catalyzed cellulose hydrolysis has the advantage of simultaneously
producing glucose, hydroxymethylfurfural (HMF), formic acid (FA), and levulinic acid
(LA), which are versatile value-added chemicals. These chemicals are formed through
serial reactions, which are illustrated in the simplified scheme shown in Fig. 1. The
drawback is that the reaction conditions are harsh and the selectivity for glucose
production is limited due to side reactions (Mok et al. 1992), transformations in cellulose
(Bouchard et al. 1989), and decomposition reactions of glucose (Saeman 1945). The
glucose yield is about 50% to 60% in a plug flow reactor (McParland et al. 1982;
Thompson and Grethlein 1979) and 70% in a semi-batch reactor (Mok et al. 1992). The
ideal reactor configuration is a countercurrent reactor with a theoretical glucose yield of
90% (Greenwald et al. 1983), but this reactor design is challenging to realize in practice.
However, Kim et al. (2001) and Gurgel et al. (2012) have obtained significant glucose
yields in a bed-shrinking flow-through reactor and in a batch reactor, respectively. These
interesting results were achieved under extremely low acid conditions, i.e., 0.07% to
0.28% H2SO4, although kinetic studies of cellulose hydrolysis have shown that glucose
yield is enhanced under short reaction times by increasing temperature and acid
concentration (Fagan et al. 1971; Malester et al. 1992; Saeman 1945). Therefore, the in-
depth studies of reaction conditions that are atypical for acid-catalyzed cellulose
hydrolysis would be valuable.
In the present study, formic acid is used as a cellulose hydrolysis catalyst under
conditions comparable to extremely low sulfuric acid conditions. Formic acid, the
strongest monocarboxylic acid, is formed during cellulose hydrolysis via glucose
decomposition reactions. Unlike sulfuric acid, formic acid is a volatile compound that can
be separated and recovered by thermal operations. Formic acid is an effective solvent for
delignifying and fractionating biomass (Dapia et al. 2002; Zhang et al. 2010). Previously,
formic acid has been studied as a pretreatment agent prior to enzymatic hydrolysis with
(Sindhu et al. 2010) or without (Marzialetti et al. 2011) a mineral acid catalyst. It has also
been used for cellulose hydrolysis in the presence of HCl at low temperature, i.e., 55 to
65 °C (Sun et al. 2007), or in the absence of a mineral acid catalyst at high temperature,
i.e., 230 to 270 °C (Asaoka and Funazukuri 2011). However, a systematic kinetic study
of cellulose hydrolysis in formic acid is lacking under reaction conditions corresponding
to dilute sulfuric acid conditions at high temperature.
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2646
PEER-REVIEWED ARTICLE bioresources.com
In this study, the aim was to develop a kinetic model for cellulose hydrolysis and
glucose formation in formic acid. Kinetic studies based on dilute sulfuric acid (Fagan et
al. 1971; Girisuta et al. 2007; Gurgel et al. 2012; McParland et al. 1982) typically rely on
the empirical power law model developed by Saeman (1945). Instead, we used a strategy
in which the rate constant equation is based on specific acid catalysis theory and the
hydrogen ion concentration is evaluated at the reaction temperature. Formation of HMF
and levulinic acid, as well as humins by-products, was also incorporated into the kinetic
model.
EXPERIMENTAL
Materials
Microcrystalline cellulose was purchased from Acros Organics. The average
particle size of the cellulose was 90 µm. The crystallinity index of cellulose was in the
range 0.72 to 0.75 (Kupiainen et al. 2012a). The calibration chemicals for HPLC were
sourced from Merck and Alfa Aesar. The chemicals were used as supplied. Formic acid
catalyst solutions of 5% to 20% (w/w) were prepared using purified water from a Milli-Q
system.
Experimental Procedure
Cellulose hydrolysis experiments were conducted in batch reactors made of
zirconium tubing. The inner volume of the reactor was about 40 mL. A PTFE-coated
thermo element sensor was inserted into the reactor through a zirconium cap to enable the
temperature measurement inside the reactor during the experiments. The average heating-
up time was 2.9 min to a temperature of 200 °C. The temperature was dropped to below
100 °C in 0.5 min after the experiment using a cold water bath.
For the experiment, 3 g of cellulose was weighed and 30 mL of a catalyst solution
was pipetted into the reactor. A preheating oven set at 440 to 470 °C and a fluidized sand
bath (SBL-2D, Techne) set at the desired reaction temperature were used for rapid and
precise temperature control. The experiment began when the reactor was placed in the
preheating oven. After reaching the reaction temperature, the reactor was transferred into
the sand bath. The experiment was stopped by quenching the reactor in a cold water bath.
After the experiment, solids were separated from the liquid by filtration using a filter
paper (Whatman 1). A sample was taken from the filtrate for HPLC analysis, and the
cake was washed with about 80 mL of deionized water, dried overnight at 105 °C, and
weighed.
HPLC Analysis
Concentrations of cellobiose, glucose, 5-hydroxymethyl-2-furfural, levulinic acid
and furfural were analyzed by high-performance liquid chromatography (HPLC) using an
ICSep ICE-Coregel 87H3 column (Transgenomic) and a refractive index detector
(Agilent Technologies). A diode array detector (Agilent Technologies) with a wavelength
of 280 nm was also used to detect smaller concentrations of HMF and furfural. The
column was operated at 60 °C. The mobile phase was 5 mM H2SO4 with a flow rate of
0.8 mL/min.
A small peak was perceived after glucose in every sample. To elucidate the peak,
individual xylose and fructose samples were analyzed. Their retention times were similar
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2647
PEER-REVIEWED ARTICLE bioresources.com
to the peak after glucose. Because furfural, a xylose decomposition product, is also
present in every sample of cellulose hydrolysis, it is likely that the peak after glucose is
xylose from a xylan impurity of the microcrystalline cellulose. Fructose, an isomerization
product of glucose, is known to decompose quickly into HMF.
pH Measurements
The pH was measured using a SenTix 81 pH electrode (WTW) connected to an
inoLab pH 720 meter (WTW). The pH meter was calibrated at three points (1.679, 4.006,
and 6.865 pH).
Yield Calculation
For the molar cellulose concentration (Cc), cellulose is treated as glucan
(anhydroglucose), i.e., MWglucan unit = MWg – MWH2O = 180.2 g/mol – 18.0 g/mol = 162.2
g/mol. In this way, cellulose is converted to glucose equivalents, which represents the
maximum theoretical glucose from cellulose. The cellulose concentration as glucose
equivalents is calculated as follows:
where mc is the mass of cellulose or the mass of weighed cake from experiments, V is the
liquid volume and MW is the molar mass. The glucose yield (Yg) is defined in this study
per initial cellulose concentration as glucose equivalents:
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2648
PEER-REVIEWED ARTICLE bioresources.com
The increasing temperature enhanced the glucose and HMF yields, whereas the
decreasing temperature enhanced the LA yield (vide infra Fig. 8). The increasing formic
acid concentration also slightly enhanced the glucose and HMF yields. The results of
cellulose hydrolysis in formic acid are in accordance with the findings by Girisuta et al.
(2007) for sulfuric acid.
Fig. 2. First-order kinetics of cellulose hydrolysis at 200 °C in 10% HCOOH. Symbols: (■) 100 g/L
(◊) 39 g/L (○) 10 g/L
Fig. 3. The effect of initial cellulose concentration on glucose yield in formic acid. Symbols: (■)
100 g/L (○) 10 g/L
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2649
PEER-REVIEWED ARTICLE bioresources.com
The effect of initial cellulose concentration on glucose yield has been considered
in a few studies. For example, there was no correlation found between an initial slurry
concentration (5% to 13.5%) and the glucose yield at 200 to 240 °C in the study by
Thompson and Grethlein (1979). Contrary to that, glucose yields of this study were
affected by the initial cellulose concentration. Figure 3 shows that glucose yield increased
significantly with the decreasing initial cellulose concentration at the same conversion
level. However, it was earlier found that glucose decomposes to an intermediate
compound in formic acid with first-order kinetics (Kupiainen et al. 2011), and that the
overall cellulose hydrolysis reaction in formic acid also follows first-order kinetics
(Kupiainen et al. 2012a). This indicates that glucose yield cannot be affected by the
initial cellulose concentration unless there is an additional side-reaction from cellulose.
The results imply that a simplified reaction scheme of two consecutive reactions is an
insufficient description of cellulose hydrolysis.
Kinetic Model
The kinetics of cellulose hydrolysis is classically modeled using two
consecutive first-order reactions (Fagan et al. 1971; Franzidis et al. 1983; Malester et
al. 1998; McParland et al. 1982; Ranganathan et al. 1985; Saeman 1945; Thompson and
Grethlein 1979). The first reaction is for cellulose hydrolysis, and the second one is
for glucose decomposition. Mok et al. (1992) experimentally found soluble non-
glucose products from cellulose. Girisuta et al. (2007) included a side-reaction from
cellulose to unknown by-products in their kinetic model.
In this study, a reaction scheme shown in Fig. 4 is used for cellulose
hydrolysis and glucose decomposition. In the scheme, Y is soluble (non-glucose) by-
products from cellulose hydrolysis and X is insoluble by-products (so-called humins)
from glucose decomposition reactions. For glucose decomposition, the reaction
scheme has been taken from our previous study with its kinetic parameters
(Kupiainen et al. 2011).
ki kH 2O kH CH ,T e
R T Tmean
. (3)
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2650
PEER-REVIEWED ARTICLE bioresources.com
The hydrogen ion concentration, CH+,T, is calculated here at the reaction
temperature due to the characteristics of weak organic acids (Kupiainen et al. 2010). This
was done by estimating the temperature dependence of the dissociation constant, Ka,FA,
based on the literature (Kim et al. 1996). First, the initial acid concentration, CHCOOH,0,
was calculated based on the pH measured at room temperature (CH+ = 10–pH,25°C) by
solving a system of nonlinear equations (Eq. 4) in Matlab 7.5.0. The initial acid
concentration was then used to calculate the hydrogen ion concentration at the reaction
temperature (CH+,T). The equations are:
CH CHCOO K a , FA CHCOOH 0
CHCOOH ,0
CHCOOH CHCOO 0 (4)
CH CHCOO 0
dC j
ji Ri (5)
dt i 1
The non-isothermalicity, i.e., the temperature data measured with respect to time, was
incorporated in the model through Eq. 3.
Estimated Parameters
A total of 51 experiments with an initial cellulose concentration of 100 g/L were
used in the parameter estimation. It was assumed that glucose decomposition is similar in
the presence and absence of cellulose.
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2651
PEER-REVIEWED ARTICLE bioresources.com
The model for independent glucose decomposition was developed earlier by the
authors (Kupiainen et al. 2011), and thus the kinetic parameters were estimated solely for
two reactions of cellulose hydrolysis.
The best estimates of the parameters for cellulose hydrolysis are presented in
Table 1. The R2 value was 99.4%. As presented in Fig. 5, the sensitivity analysis shows
that the parameters are well-defined. Furthermore, the parity plot in Fig. 6 indicates that
the model is in good agreement with the data for both cellulose conversion and glucose
yield.
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2652
PEER-REVIEWED ARTICLE bioresources.com
reactions, interfere with the cellulose gravimetric analysis, which is a common way to
determine unreacted cellulose (Gurgel et al. 2012; Mok et al. 1992). The kinetic model
was then used to study the data points. Figure 9 shows how cellulose data points are
corrected by subtracting the amount of humins estimated by the model from the
experimental cellulose values. It can be seen from Fig. 9 that within short reaction times,
original cellulose data points equal to the corrected ones. This indicates that there are no
humins formed at the beginning of the reaction, as expected, but the formation of humins
had a clear effect on the determination of cellulose conversion at the longest reaction time
points at 200 °C (20% HCOOH) and at 220 °C (10% to 20% HCOOH). These suspicious
values at the longest reaction times were excluded from the data set when the kinetic
parameters were estimated.
Fig. 7. Cellulose hydrolysis in formic acid (symbols) and the model prediction (lines); Symbols: (*)
5%, (○) 10%, (□) 20% HCOOH
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2653
PEER-REVIEWED ARTICLE bioresources.com
hand, reaction step 2 is more sensitive to temperature than reaction steps 1 and 3 (Ea,2
>Ea,1>Ea,3), which enhances the glucose yield at higher temperatures. However, the
amount of material lost in the side reaction is still significant based on the model.
Fig. 8. Product formation in formic acid (symbols) and the model prediction (lines); Symbols: (o)
Glucose (□) HMF (◊) LA
Fig. 9. Cellulose hydrolysis with experimental data points corrected by the model estimate of
forming solids; Symbols: (∆) Original cellulose, (▼) Cellulose – solids
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2654
PEER-REVIEWED ARTICLE bioresources.com
DISCUSSION
The model was used to predict the maximum glucose yield from hydrolysis of
microcrystalline cellulose in 20% formic acid. Figure 10 shows the effect of temperature
and time on glucose yield. The model predicts that the maximum glucose yield, 24%
(mol/mol), is achieved in 1 min at 240 °C, which is outside the studied temperature
range. However, the predicted condition area with formic acid corresponds to the
optimum condition area with sulfuric acid (Fagan et al. 1971; Gurgel et al. 2012;
McParland et al. 1982).
Fig. 10. Contour plot prediction for glucose yield (%) from microcrystalline cellulose in 20%
HCOOH.
Figure 11 shows the contour plot for the yield of soluble non-glucose compounds
from cellulose. The model predicts a significant yield of soluble by-products even under
reaction conditions appropriate for glucose production. Under these conditions, the yield
of soluble by-products from cellulose was higher than the yield of insoluble by-products
from glucose decomposition. Based on the model presented in this study, the main reason
for the low glucose yield is the side reaction from microcrystalline cellulose to non-
glucose products, and not the glucose decomposition reactions. In addition, the cellulose
hydrolysis reaction has to proceed almost to completion before the rate of glucose
decomposition exceeds the rate of glucose formation from the cellulose. These findings
are consistent with Mok et al. (1992) concerning non-hydrolyzable oligomers, which
were found in substantial amounts from cellulose hydrolysis in water. In dilute hydrogen
ion concentrations used in this study, a significant side reaction is possible from cellulose
to non-glucose products.
Fig. 11. Contour plot prediction for the yield of soluble by-products (%) from microcrystalline
cellulose in 20% HCOOH
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2655
PEER-REVIEWED ARTICLE bioresources.com
Because the kinetic parameters from independent glucose decomposition
experiments can be successfully applied in the modeling of the whole cellulose
hydrolysis system, it can be concluded that there are neither major interactions between
cellulose and glucose nor its decomposition products. This gives advantages for those
modeling cellulosic systems. Experiments can be performed only for different cellulosic
raw materials instead of an extensive series of both hydrolysis and glucose decomposition
experiments. The kinetic parameters have to be fitted to the hydrolysis part of the model,
while the kinetic parameters for glucose decomposition are already available in the
scientific literature. It has already been shown that the kinetic model for formic-acid-
catalyzed glucose decomposition is applicable to predict glucose decomposition in 0.09%
to 0.50% sulfuric acid at 180 to 220 °C (Kupiainen 2012). However, lignin and metal
impurities contained in lignocellulosic raw materials may have an influence on glucose
reactions.
Despite the low glucose yield afforded in this study, there is a potential to
intensify cellulose hydrolysis in formic acid. For example, we found glucose yield to be
two-fold higher from organosolv pulp than from microcrystalline cellulose (Kupiainen et
al. 2012b). In addition, Gurgel et al. (2012) achieved glucose yields (defined as in this
study) of up to 48% from cellulosic bagasse pulp under extremely low sulfuric acid
conditions, which are similar to formic acid conditions used in this study. However, in
order to maximize the glucose yield, a challenging optimization task has to be solved
under reaction conditions of high temperature and a short reaction time. With reliable
kinetics, it is possible to evaluate these complicated phenomena taking place in the
reaction system and to design an appropriate reactor for the production of chemicals.
CONCLUSIONS
1. A kinetic model with a rate constant expression derived from specific acid catalysis
theory was developed for the formic acid-catalyzed hydrolysis of microcrystalline
cellulose. In addition, hydrogen ion concentrations were calculated at the reaction
temperature. The model showed good agreement with experimental data.
2. A previously established glucose decomposition model was successfully incorporated
into the cellulose hydrolysis model, indicating that there are no interactions between
glucose decomposition and cellulose hydrolysis.
3. Based on the model, substantial amounts of raw material are lost in side reactions
from cellulose to by-products in a batch reactor under reaction conditions similar to
those for very dilute sulfuric acid. Glucose decomposition reactions have a minor
effect on the glucose selectivity.
4. The model also has an ability to explain the formation of hydroxymethylfurfural and
levulinic acid from glucose.
ACKNOWLEDGMENTS
The authors would like to thank the Emil Aaltonen Foundation and Oulu
University Apuraharahasto for their financial support.
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2656
PEER-REVIEWED ARTICLE bioresources.com
REFERENCES CITED
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2657
PEER-REVIEWED ARTICLE bioresources.com
51(8), 3295-3300.
Kupiainen, L., Ahola, J., and Tanskanen J. (2012b). “Hydrolysis of organosolv wheat
pulp in formic acid at high temperature for glucose production,” Bioresour. Technol.
116(7), 29-35.
Lee, J., Xu, Y., and Huber, G. W. (2013). “High-throughput screening of monometallic
catalysts for aqueous-phase hydrogenation of biomass-derived oxygenates,” Appl.
Catal. B-Environ. 140-141(8-9), 98-107.
Ma, Y., Ji, W., Zhu, X., Tian, L., and Wan, X. (2012). “Effect of extremely low AlCl3 on
hydrolysis of cellulose in high temperature liquid water,” Biomass Bioenergy 39(4),
106-111.
Malester, I., Green, M., and Shelef, G. (1992). “Kinetics of dilute acid hydrolysis of
cellulose originating from municipal solid wastes,” Ind. Eng. Chem. Res. 31(8), 1998-
2003.
Marzialetti, T., Miller, S. J., Jones, C. W., and Agarwal, P. K. (2011). “Switchgrass
pretreatment and hydrolysis using low concentrations of formic acid,” J. Chem.
Technol. Biotechnol. 86(5), 706-713.
McParland, J. J., Grethlein, H. E., and Converse, A. O. (1982). “Kinetics of acid
hydrolysis of corn stover,” Sol. Energy 28(1), 55-63.
Mok, W., Antal, M., and Varhegyi, G. (1992). “Productive and parasitic pathways in
dilute acid-catalyzed hydrolysis of cellulose,” Ind. Eng. Chem. Res. 31, 94-100.
Ranganathan, S., Macdonald, D., and Bakhshi, N. (1985). “Kinetic study of wheat straw
hydrolysis using sulphuric acid,” Can. J. Chem. Eng. 6(1), 840-844.
Saeman, J. F. (1945). Kinetics of wood saccharification. “Hydrolysis of cellulose and
decomposition of sugars in dilute acid at high temperature,” Ind. Eng. Chem. 37(1),
43-52.
Sasaki, M., Kabyemela, B., Malaluan, R., Hirose, S., Takeda, N., Adschiri, T., and Arai,
K. (1998). “Cellulose hydrolysis in subcritical and supercritical water,” J. Supercrit.
Fluid. 13(1-3), 261-268.
Sindhu, R., Binod, P., Satyanagalakshmi, K., Janu, K. U., Sajna, K. V., Kurien, N.,
Sukumaran, R. K., and Pandey, A. (2010). “Formic acid as a potential pretreatment
agent for the conversion of sugarcane bagasse to bioethanol,” Appl. Biochem.
Biotechnol. 162(8), 2313-2323.
Sun, Y., Lin, L., Pang, C., Deng, H., Peng, H., Li, J., He, B., and Liu, S. (2007).
“Hydrolysis of cotton fiber cellulose in formic acid,” Energ. Fuel 21(4), 2386-2389.
Taherzadeh, M., and Karimi, K. (2007). “Enzyme-based hydrolysis processes for ethanol
from lignocellulosic materials: A review,” BioResources 2(4), 707-738.
Thompson, D. R., and Grethlein, H. E. (1979). “Design and evaluation of a plug flow
reactor for acid hydrolysis of cellulose,” Ind. Eng. Chem. Prod. R. D. 18(3), 166-169.
Tullo, A. H. (2010). “Catalyzing biobased chemicals,” Chem. Eng. News 88(38), 15-17.
Werpy, T., and Petersen, G. (ed.) (2004). “Top value added chemicals from biomass.
Volume I – Results of screening for potential candidates from sugars and synthesis
gas,” National Renewable Energy Laboratory, Golden, CO.
Zhang, M., Qi, W., Liu, R., Su, R., Wu, S., and He, Z. (2010). “Fractionating
lignocellulose by formic acid: Characterization of major components,” Biomass
Bioenergy 34(4), 525-532.
Article submitted: January 3, 2014; Peer review completed: March 16, 2014; Revised
version received and accepted: March 18, 2014; Published: March 25, 2014.
Kupiainen et al. (2014). “Formic acid treatment,” BioResources 9(2), 2645-2658. 2658
View publication stats