- Email: [email protected]
Kinetic characterization for dilute sulfuric acid hydrolysis of timber varieties and switchgrass
Available online at www.sciencedirect.com
Bioresource Technology 99 (2008) 3855–3863
Kinetic characterization for dilute sulfuric acid hydrolysis of timber varieties and switchgrass Shu Chiang Yat a, Alan Berger b, David R. Shonnard a
a,*
Biochemical Engineering Laboratory, Department of Chemical Engineering, Michigan Technological University, Houghton, MI 49931, United States b Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA 18602, United States Received 1 February 2007; received in revised form 18 June 2007; accepted 18 June 2007 Available online 29 September 2007
Abstract Hydrolysis of four timber species (aspen, balsam fir, basswood, and red maple) and switchgrass was studied using dilute sulfuric acid at 50 g dry biomass/L under similar conditions previously described as acid pretreatment. The primary goal was to obtain detailed kinetic data of xylose formation and degradation from a match between a first order reaction model and the experimental data at various final reactor temperatures (160–190 C), sulfuric acid concentrations (0.25–1.0% w/v), and particle sizes (28–10/20 mesh) in a glass-lined 1 L well-mixed batch reactor. Reaction rates for the generation of xylose from hemicellulose and the generation of furfural from xylose were strongly dependent on both temperature and acid concentration. However, no effect was observed for the particle sizes studied. Oligomer sugars, representing incomplete products of hydrolysis, were observed early in the reaction period for all sugars (xylose, glucose, arabinose, mannose, and galactose), but were reduced to low concentrations at later times (higher hemicellulose conversions). Maximum yields for xylose ranged from 70% (balsam) to 94% (switchgrass), for glucose from 10.6% to 13.6%, and for other minor sugars from 8.6% to 58.9%. Xylose formation activation energies and the pre-exponential factors for the timber species and switchgrass were in a range of 49–180 kJ/mol and from 7.5 · 104 to 2.6 · 1020 min1, respectively. In addition, for xylose degradation, the activation energies and the pre-exponential factors ranged from 130 to 170 kJ/mol and from 6.8 · 1013 to 3.7 · 1017 min1, respectively. There was a near linear dependence on acid concentration observed for xylose degradation. Our results suggest that mixtures of biomass species may be processed together and still achieve high yields for all species. 2007 Elsevier Ltd. All rights reserved. Keywords: Dilute acid; Hydrolysis; Switchgrass; Timber
1. Introduction With only 4.5% of the world’s population, the United States consumes about 25% of global energy and produces roughly 25% of the planet’s CO2 emissions (US DOE, 2005). The United States consumed an average of 20.6 million barrels of petroleum per day during the first nine months of 2005, the same amount year-over-year as in 2004 (US DOE, 2005). The average gasoline price for the third quarter of 2005 was $2.56 per gallon, up $0.67 per
*
Corresponding author. Tel.: +1 906 487 3468; fax: +1 906 487 3213. E-mail address: [email protected] (D.R. Shonnard).
0960-8524/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2007.06.046
gallon from the third quarter of 2004. Gasoline prices are affected by natural disasters such as hurricanes as shown in the aftermath of Katrina and Rita in 2005. Conversion of abundant cellulosic biomass to ethanol as a transportation fuel presents an important opportunity to improve energy security, reduce the trade deficit, reduce greenhouse gas emission, and improve price stability (Wyman, 1999). Improvements in pretreatment, such as dilute acid hydrolysis, will aid in preparing feedstocks for enzymatic hydrolysis and fermentation without generating significant concentrations of fermentation inhibitors. The goals of the pretreatment are to decompose the polymeric components of the wood and form monomer sugars, enhance enzymatic conversion of the cellulose
3856
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
fraction, and, hopefully, obtain a higher ethanol yield. Numerous pretreatment approaches have been investigated at many national laboratories, universities, and industrial locations over the past 25 years. In the past, it has been difficult to compare the performance and economics of these various approaches due to differences in feedstocks, chemical analysis methods, and data reporting methodologies. Recently, a group of pretreatment researchers across North America have begun to collaborate to investigate different pretreatment approaches on a common basis to allow meaningful comparison. These researchers have formed the Biomass Refining Consortium for Applied Fundamentals and Innovation (CAFI) to advance the efficacy and knowledge base of pretreatment technologies (Wyman et al., 2005a). Pretreatment methods are either physical or chemical, but some methods incorporate both effects. Physical pretreatment methods, such as comminution (mechanical reduction in biomass particulate size) and steam explosion, can be classified into mechanical and non-mechanical pretreatment. Physical forces used in mechanical pretreatments can subdivide lignocellulosic material into fine particles which are substantially susceptible to acid or enzymatic hydrolysis. Non-mechanical physical pretreatments cause decomposition of lignocellulosics by exposing them to harsh external forces other than mechanical forces. Chemical pretreatments have been used extensively for removal of lignin surrounding cellulose and for destroying its crystalline structure. Traditionally, the paper industry has utilized pulping processes for delignification of cellulosic materials to produce high strength, long fiber paper products. It has been considered, however, that these processes are quite severe and expensive to be used for pretreatment of lignocellulosics for production of ethanol. Biorefineries will be designed to process specific feedstocks, such as corn stover, switchgrass or other woody energy crops and forest resources. In forest regions, the feedstock will most likely be a mixture of timber species grown locally, possibly mixed with herbaceous biomass, rather than a monoculture feed. Flexible pretreatment process technologies that can handle biomass with variable compositions of cellulose, hemicellulose, lignin, and ash will be the most efficient. In this project, the kinetic characterization of several timber species was studied using diluted sulfuric acid hydrolysis. The goals of this research are (1) to measure the concentrations of fermentable monomer and oligomer sugars plus their non-fermentable byproducts; (2) to determine the kinetic parameters for xylose formation and degradation from a match between a reaction model and experimental data; and (3) interpret the single timber species results as to potential mixture effects on pretreatment performance. Kinetic parameters obtained from this study can in the future be used in reactor models of various configurations to identify optimum conditions for biomass conversion to fermentable sugars.
2. Methods 2.1. Raw material preparation Each tree species was harvested from forest lands in the vicinity of Michigan Technological University (Houghton, MI), cut into 15 inch lengths and debarked. The debarked wood logs were then cut into flakes using a rotary drum blade and then dried overnight in an oven at 100 C. Then, the wood flakes were hammer milled into fine particles and screened into fractions (10–30 mesh). The switchgrass sample was provided by National Renewable Energy Laboratory (NREL) and was also screened into fractions without further size reduction. The total carbohydrate, lignin and ash analysis of each species was studied based on the Laboratory Analytical Procedure (LAP #002) of ‘‘Determination of Structural Carbohydrates and Lignin in Biomass’’ (NREL, 2006). 2.2. Dilute sulfuric acid hydrolysis The hydrolysis experiments were carried out using a 1 L high pressure/high temperature Parr reactor model 4571 (Moline, IL USA). This stainless-steel reactor was equipped with a glass liner, a heat exchanger, a 6 anglebladed impeller, and temperature controller (Parr controller model 4842). Each dry biomass sample (25 g) was mixed with 500 mL of 0.25–1.0% w/w sulfuric acid and was heated to the desired maximum temperature (160, 175, and 190 C) inside the closed reactor. To control maximum reactor temperature, compressed air was used as the heat removal fluid in the cooling circuit and was controlled manually to regulate the temperature to within ± 1 C. The reactor was closed and placed in the reactor oven to initiate heating. The stirrer speed was adjusted to 50 rpm and the reactor temperature was measured using a thermocouple probe connected to the control unit. Temperature was recorded every 5 min until the maximum was reached (over approximately a 70 min heat up period) and thereafter every 8 min. Samples (5 ml) were collected (after purging the sample dip tube of 8 ml solution) at 100 C, 135 C, and thereafter at 3 min intervals until reaching the maximum temperature, Tmax (160, 175, and 190 C). Tmax was maintained for 32 min and four 5-mL samples were collected at 8 min intervals. Each collected samples was cooled in an ice bath to stop the reaction. The cooled samples were filtered using a 0.2 lm nitrocellulose membrane to remove suspended solids. Filtered samples were then separated into two vials: one for monomeric sugars analysis and one for total sugars analysis. The monomeric sugar content was analyzed using HPLC after the samples were neutralized using 6 N NaOH to pH 5 to 6 (measured using litmus paper). The total sugar analysis was used to identify the amount of oligomers in each sample obtained during the course of the hydrolysis reaction (NREL, 2005). One milliliter of
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
each filtered sample was dispensed into an amber crimptopped HPLC vial. The samples were acidified to 4% acid concentration using 96% H2SO4, crimped tightly, and then autoclaved for one hour at 121 C to hydrolyze oligomers to monomer sugars. After completion of the autoclave cycle, each sample was subsequently neutralized to pH 5 to 6 using 6N NaOH. Because a precipitate formed after the neutralizing step the samples were placed into a centrifuge (MARATHON 21 K, Fisher Scientific (Pittsburgh, PA, USA)) at 8000 rpm for 5 min. After completion of the centrifuge cycle, the clear solution was carefully and slowly drawn out from the vials (to avoid reintroduction of solids) and transferred to new HPLC vials. 2.3. HPLC analysis Sugars and pretreatment byproducts were measured using an Agilent Series 1100 HPLC (Santa Clara, CA, USA). Separation was accomplished using a Bio-Rad Aminex HPX-87P column with deashing cartridge (BioRad) and quantified with a refractive index detector for sugars and a diode-array detector for fufural and acetic acid. 2.4. Kinetic model development The kinetics of hemicellulose hydrolysis was first modeled by Mehlberg and Tsao (1979) in the following reaction: ↓
↓
↑
Oligomers → Xylose → Degradation Products
Hemicellulose
ð1Þ
→ Xylan II →
The concentrations of xylan I, xylan II and oligomers are very difficult to determine and since the reaction to produce xylan I and xylan II is very rapid at higher temperatures above 160 C (Saeman, 1945; Grohmann et al., 1986; Bienkowski et al., 1984; Ladisch, 1989), the model can be simplified to the following: k1
dH ¼ k 1 H dt dX H ¼ k1 k2X dt 0:88
k2
Hemicellulose ðH Þ! XyloseðX Þ ! Degradation Products ðDÞ ð2Þ
Two models were developed in this project. The first model includes the first step in conversion of hemicellulose to xylose. The second model involves the process of xylose degradation. 2.4.1. Hemicellulose hydrolysis and xylose model Eqs. (3) and (4) are the mass balance equations for H and X in a well-stirred batch reactor under the assumption of constant reactor liquid volume and first order reaction for each species:
ð3Þ ð4Þ
where k1 and k2 are first order reaction rate constants for xylose formation and xylose degradation, respectively. The reaction rate constant, k1, is assumed to have Arrhenius-type temperature dependence: E1 k 1 ¼ A1 exp ð5Þ RT 1 where E1 = activation energy (kJ/mol), R = 8.3143 · 103 (kJ/molK), T1 = temperature (K), A1 = pre-exponential factor (min1). The pre-exponential factors for hemicellulose hydrolysis reactions are assumed to be dependent upon acid concentration: A1 ¼ A1o C m1
ð6Þ
where A1o = pre-exponential factor for hemicellulose hydrolysis (min1), C = acid concentration (% w), m1 = acid concentration exponent for the rate constant k1 (dimensionless). As the reaction temperature changed over time, the reaction rate constant model changed as a function of time as shown in Eq. (7). E1 m1 k 1 ¼ A1o C exp ð7Þ RT 1 ðtÞ Therefore, a numerical solution of Eq. (3) at each new time step (i + 1) is required: H iþ1 ¼ H i k 1i H i Dt
→ Xylan I → ↑
3857
where
E1 k 1i ¼ Aio C m1 exp RT 1 ðtÞ
ð8Þ ð9Þ
Hence, the xylose concentration obtained from the solution of Eq. (3) is X iþ1 ¼
H max H iþ1 0:88
ð10Þ
where Hmax = the maximum concentration of xylan that can be produced in each species (g xylose oligomer/L) as determined by the total carbohydrate analysis. In Eq. (10) it is assumed that negligible furfural is generated from xylose during the heat up period (period of time for application of the model), which is an approximation that will be shown to be a valid assumption. The factor of 0.88 is the ratio of molecular weight of xylan to xylose (132/150). Eqs. (8)–(10) were integrated using Microsoft Excel and the Trapezoidal Rule of integration with a time step (Dt) of 0.01 min., which was determined to be the optimum time step for the minimization or numerical error (Yat, 2006). The kinetic parameters, A1 and E1, were determined using Excel Solver function with initial guesses. The sum of squared errors between experimental data and model predictions were minimized by improving the values of the A1 and E1 using Eqs. (8)–(10).
3858
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
2.4.2. Xylose degradation model Eq. (4) was used to model the formation of xylose degradation products. The degradation products formed during the maximum temperature period. It was assumed that all of the reactive hemicellulose (xylan) was hydrolyzed completely to xylose (H = 0) prior to xylose degradation. Therefore, Eq. (4) becomes dX ¼ k 2 X dt
a relatively small number of representative graphs of concentration versus time are shown. 3.1. Biomass composition The composition of the five feedstocks is shown in Table 1 with a comparison of results obtained in this study to the literatures value for the same biomass species. The major components of these substrates were xylan, glucan, and lignin. Measured glucan and xylan in this study were generally similar to the literature values, with the exception of switchgrass whose glucan value is higher in this study. Lignin content measured in this study for all species is generally higher than in the literature studies, due to the generation of tars which become associated with the lignin fraction. The minor sugars of galactan, araban, and mannan measured in this study were comparable to the literature values.
ð11Þ
and the solution for the initial condition of t = 0, X = Xmax (where Xmax is measured) is given by Eq. (12). ln
X X max
¼ k 2 t
ð12Þ
By plotting the experimental data according to the form of Eq. (12), k2 was determined as the slope of the linear regression fit to the data. Eq. (13) is used to determine the pre-exponential factor (A2) and activation energy (E2) of xylose degradation: ln k 2 ¼ ln A2
E2 1 R T2
ð13Þ
3.2. Monosaccharides and oligosaccharides formation
where T2 is the value of maximum temperature achieved in several experiments. The value of k2 determined at each maximum temperature as per Eq. (12) was used to determine the kinetic parameters in Eq. (13) using linear regression analysis. To obtain the acid concentration exponent (m2) for the xylose degradation model, Eq. (14) is used: E2 ln k 2 ¼ m2 ln C þ ln A2 exp ð14Þ RT 2
The xylan contents from structural carbohydrate of various biomass species used in this project was found to be range from 6% to 19% (dry basis) using the procedure provided by NREL LAP #002 (NREL, 2006). Thus, the maximum possible xylose concentration from these hydrolysis experiments was 3–9 g/L, depending on biomass type. A typical plot of the time-temperature profile using the Parr reactor is shown in Fig. 1. The reactor temperature increased almost linearly after 40 min from the start of the heating period until the Tmax was reached at about 70 min. for Tmax = 175 C. Over this time period most of the hemicellulose was hydrolyzed to sugars, as shown in subsequent figures, and a linear trendline was fit to the temperature data for use in kinetic modeling. By way of comparison to the literature methods, in most of the studies reported, the acid was injected after the sample was heated to the desired temperature (Lloyd and Wyman, 2005; Esteghlalian et al., 1997; Garrote et al., 2001), or the acidified solution was heated quickly from
By plotting the natural log of calculated k2 from Eq. (12) at each concentration experiment as a function of natural log of acid concentration, the slope of the linear regression is equal to m2. 3. Results and discussion Due to the large number of pretreatment experiments conducted for each biomass species (8 experiments), only Table 1 Compositional analysis of raw biomass samples (percent by weight) Species
% Biomass composition
References
Glucan
Xylan
Galactan
Araban
Mannan
Lignin
Aspen Balsam Red Maple
57.3 46.8 46.6
16.0 4.8 17.3
0.8 1.0 0.6
0.4 0.5 0.5
2.3 12.4 3.5
16.3 29.4 24.0
Brooks et al. (1978)
Switchgrass
32.2
20.3
–
3.7
0.4
19.5
Esteghlalian et al. (1997)
Aspen Balsam Basswood Red Maple Switchgrass
52.43 47.09 43.99 43.18 47.72
14.60 6.23 15.31 17.69 19.06
3.52 5.45 3.41 5.71 4.18
2.41 5.41 3.49 4.13 8.11
5.32 11.49 2.91 5.37 6.30
26.69 36.04 28.44 36.49 26.04
This study
7
Monomer
6
Oligomer
5 4 3 2 1
0
20
40
60
80
100
0
120
50
Time (min)
60
70
80
90
100
110
Time (min)
Fig. 1. Basswood acid hydrolysis time-temperature profile at 0.5% H2SO4, Tmax = 175 C, 20–10 mesh.
room temperature to Tmax by plunging the reactor into a hot sand bath. Monosaccharide data for biomass hydrolysis at 0.5% sulfuric acid, 175 C, and 20–10 mesh particle size is displayed in Fig. 2. The figure shows the time-concentration data of each sugar plus furfural, one of the main byproducts of dilute acid hydrolysis (along with insoluble tars, which were not measured in this work). The concentrations of xylose and other sugars were small prior to achieving 135 C in the reactor. At temperatures above 135 C, the xylose concentration increased rapidly until Tmax was reached. Afterward, the concentration of xylose decreased while the temperature was maintained constant. Similar trends in the concentrations of the hemicellulose minor sugars were also observed. After Tmax was reached, significant amounts of glucose and furfural were formed. Furfural is the degradation product produced from xylose and glucose was formed from the hydrolysis of the cellulose fraction. The highest concentration of xylose oligomer appeared early in the pretreatment process as shown in Fig. 3, prior to the production of significant concentrations of monosaccharide. Oligomer formation may be beneficial during hydrolysis since monomer sugars can be formed subsequently even at conditions different from the reactor. A similar trend was observed for other sugar oligomers and for other biomass species. After reaching Tmax the concentra-
Fig. 3. Aspen acid hydrolysis at 0.5% H2SO4, Tmax = 175 C, 20–10 mesh (xylose oligomer & monomer profile).
tion of xylose oligomer decreased as it formed monomer; however, furfural was produced simultaneously. At the time in the reaction when maximum xylose concentration was achieved, oligomer concentration already had decreased to a small amount, relative to maximum oligomer concentration. This maximum total xylose concentration occurred before substantial furfural had accumulated. Thus, careful control of reactor temperature and residence time is needed to optimize monomer concentration and minimize non-fermentable byproduct formation. 3.3. Maximum temperature effects Maximum temperature was one of the three process variables investigated. The temperature of the reactor was ramped from room temperature to the Tmax (160, 175, and 190 C). A typical plot of the effect of different Tmax on dilute acid hydrolysis of biomass is shown in Fig. 4. The concentrations of xylose observed during the temperature ramp to Tmax are nearly identical, indicating the good reproducibility of the temperature ramp in these experiments. For the experiment at a maximum temperature of 160 C, the maximum concentration of xylose was delayed and its magnitude was lower than for higher Tmax experiments. In addition, the rate of xylose degradation was small relative to the higher temperature experiments. 3
8
6
Xylose Galactose
5
Arabinose
Xylose Concentration (g/L)
Glucose
7
Concentration (g/L)
3859
8
200 180 160 140 120 100 80 60 40 20 0
Xylose Concentration (g/L)
Temperature (ºC)
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
Mannose
4
Furfural
3 2 1
160 C
2.5
175 C 190 C
2 1.5 1 0.5 0
0 45
55
65
75
85
95
105
Time (min) Fig. 2. Basswood acid hydrolysis at 0.5% H2SO4, Tmax = 175 C, 20– 10 mesh (monomers content).
40
50
60
70
80
90
100
110
Time (min) Fig. 4. Balsam acid hydrolysis at 0.5% H2SO4, 20–10 mesh (xylose profile for different Tmax).
3860
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
In the Tmax = 190 C experiment, xylose degraded rapidly within 10 min to near zero concentration. 3.4. Acid concentration effects Another process variable that affects hemicellulose hydrolysis and xylose formation/degradation during pretreatment is the acid concentration of the reaction medium. A representative plot is displayed in Fig. 5 on the effect of acid loading on xylose concentration for all of the biomass species tested. As the concentration of acid increased, the rate of xylose formation and degradation increase. For example, at 0.25% sulfuric acid, the rate of xylose formation was relatively slow and more time was required to achieve maximum xylose concentration. At the maximum acid concentration of 1.0%, the rate of xylose degradation after achieving maximum temperature of 175 C was the highest for these experiments. 3.5. Xylose formation and degradation kinetic parameters
Xylose Concentration (g/L)
The xylose formation model was fitted to the experimental data in order to determine the kinetic parameters of xylose formation. A match between the model and the experimental data which is representative of all experiments is shown in Fig. 6. As discussed before, to get the 9 8 7 6 5 4 3 2 1 0 40
50
60
70
80
90
100
110
Time (min) 0.25% w/w Sulfuric Acid
0.5% w/w Sulfuric Acid
1.0% w/w Sulfuric Acid
Fig. 5. Red maple acid hydrolysis at Tmax = 175 C, 20–10 mesh (xylose profile for various sulfuric acid concentrations).
8 Predicted Model
7
Experimental Data
Concentration (g/L)
6 5 4
best fit the sum of squared error between the theoretical model and the experimental data was minimized by varying the pre-exponential factor and the activation energy. The kinetic parameters of woody biomass pretreatment during xylose formation are summarized in Table 2 at different experimental conditions. The pre-exponential factors for all species and reactor conditions ranged from 7.53 · 104 min1 to 2.63 · 1020 min1 and the activation energies varied from 49 kJ/mol to 179 kJ/mol. The xylose degradation kinetics was calculated using the equations obtained from Section 2.4.2. Best fit of Eqs. (12)–(14) to degradation kinetic data that is representative of all experiments is shown in Figs. 7–9. The kinetic parameters and the acid concentration exponents are summarized in Table 3. Kinetic parameters obtained from this research are compared in Table 4 to other studies for related biomass species even though the conduct of our experiments was different from the literature studies, as mentioned earlier. The values of A, E, and m are in general comparable to the literature values. The range of A1 and E1 values reported in this work are a result of conducting experiments under different temperature, acid concentration, and particle size conditions. 3.6. Overall sugars yield The average sugar yield of each woody species obtained from the eight experiment runs for each species is found in Fig. 10 (Yat, 2006). Yield was calculated at the point in the reaction when maximum xylose concentration occurred. All yields reported were normalized to the total potential sugars in the original untreated feedstock obtained from the total carbohydrate analysis in Table 1. The overall xylose yield was the major interest in this project since this sugar is the dominant monomer in hemicellulose. Average xylose yield for each species varied from about 70% for balsam to 92% for switchgrass. No clear trend could be seen for xylose yield under different process conditions (Tmax, C) (Yat, 2006). The glucose yield varied from only 11% to 13% for all species, showing that dilute acid is ineffective in completely hydrolyzing cellulose, although at longer times in the reactor at Tmax glucose continued to accumulate, as did degradation byproducts. Yields of the minor sugars were generally below 50% and often less than 20%. One notable exception was a high yield for mannose from balsam, which has a high percentage of mannose in the hemicellulose. 3.7. Discussion of results
3 2 1 0 45
55
65
75
85
95
105
Time (min) Fig. 6. Aspen acid hydrolysis at 0.5% H2SO4, Tmax = 160 C, 20–10 mesh (comparison of xylose formation model to xylose experimental data).
Recently, most pretreatment research has focused on comparing methods, such as dilute acid, AFEX, steam explosion and acid-alkali pretreatment, usually with single biomass species (Wyman et al., 2005b). The results reported here were on a single method applied to multiple woody biomass species and switchgrass in separate experiments. Acid hydrolysis of different biomass species are
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
3861
Table 2 Kinetic parameters during xylose formation at various reactor conditions Acid concentration (% w)
Particle size
Kinetic parameters
Temperature (C) 160
Aspen 0.25
20–10
0.50
20–10 >28
1.00
20–10
Balsam 0.25
20–10
0.50
20–10 >28
1.00
20–10
Basswood 0.25
20–10
0.50
20–10 >28
1.00
20–10
Red Maple 0.25
20–10
0.50
20–10 >28
1.00
20–10
Switchgrass 0.25
20–10
0.50
20–10 >28
1.00
20–10
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
illustrated by the data presented in this paper, and the hydrolysis of hemicellulose to xylose is likely a factor in the pretreatment of biomass materials when acid is present. Our results on single biomass species provide insights as to how mixtures of biomass species might respond to dilute acid hydrolysis. Our results suggest that mixtures of biomass species may be processed together and still achieve high yields for all species. The only exception to this conclusion might be balsam whose average yield was lower
140.70 2.80 · 1016
89.65 8.35 · 109
154.36 7.58 · 1017
145.76 1.11 · 1017
115.11 2.48 · 1013
175
175
190
151.85 2.65 · 1017 134.70 5.47 · 1015 120.23 4.85 · 1013 97.18 1.53 · 1011
136.10 7.55 · 1015 116.85 1.90 · 1013
148.60 1.88 · 1017
151.52 2.78 · 1017 67.83 1.54 · 107 58.12 1.00 · 106 48.72 7.53 · 104
71.74 4.64 · 107 74.40 1.25 · 108
67.37 1.18 · 107
179.13 2.63 · 1020 126.27 2.42 · 1014 139.08 6.63 · 1015 102.67 4.46 · 1011
134.44 2.58 · 1015 130.94 7.55 · 1014
117.36 1.94 · 1013
149.45 1.40 · 1017 104.07 6.42 · 1011 111.53 5.56 · 1012 98.31 2.03 · 1011
88.65 5.77 · 109 96.41 5.86 · 1010
88.16 6.63 · 109
167.48 1.03 · 1019 70.75 8.56 · 107 101.29 3.72 · 1011 78.24 1.21 · 109
74.36 1.87 · 108 111.07 6.90 · 1012
65.94 1.89 · 107
than the other species tested. The kinetic parameters obtained in this study (except for balsam) were similar and the reactor residence times for reaching maximum xylose generation were nearly identical. However, mixture effects (interactions that might affect species reactivities) must be investigated to confirm these initial observations. The data measured in this study confirm that a kinetic model including irreversible first order reactions for xylose production and degradation is appropriate for this system.
3862
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863 Table 3 Kinetic parameters of timber varieties plus switchgrass during xylose degradation
0 -0.5
ln (X/Xm)
-1 -1.5 -2 -2.5
y = -0.1400x - 0.0133
-3
2
R = 0.9975
-3.5 -4 0
5
10
15
20
25
30
Time (min) Fig. 7. Red maple acid hydrolysis at 0.5% H2SO4, Tmax = 190 C, 20– 10 mesh (xylose degradation model). 0 -0.5 -1
ln (k2)
-1.5
y = -15593x + 31.855
-2 -2.5
R2 = 0.9721
-3 -3.5 -4 -4.5 -5 0.00215
0.0022
0.00225
0.0023
0.00235
1/T (K-1) Fig. 8. Red maple acid hydrolysis kinetic parameters determination of xylose degradation using Eq. (13). 0 -0.5 -1
ln (k2)
-1.5 -2
y = 1.023x - 2.2257 2
R = 0.9435
-2.5 -3 -3.5 -4 -1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
ln (C) Fig. 9. Red maple acid hydrolysis acid concentration exponent determination of xylose degradation using Eq. (14).
Aspen Balsam Basswood Red Maple Switchgrass
Pre-exponential factor, A2 (min1)
Acid concentration exponent, m2
Activation energy, E2 (kJ/ mol)
6.51 · 1016 7.59 · 1015 2.52 · 1013 6.83 · 1013 3.73 · 1017
1.0 0.9 1.2 1.0 1.4
155.36 147.56 126.89 129.64 165.59
The concentrations of monomer sugars in these experiments were dilute (less than 1%). However, the observation of irreversible reaction kinetics may not apply to higher concentrations anticipated in commercial applications, as shown before for acid hydrolysis for glucose production at higher substrate concentrations (Bienkowski et al., 1987). The derived kinetic model in this paper was employed to predict xylose concentration as a function of time throughout the experiment. The model was fitted to the experiment data in order to determine the relevant kinetic parameters. After the kinetic data was obtained, it is possible to use these parameters to predict pretreatment hydrolysis performance for various reactor configurations. Modeling reactor performance will allow for the determination of optimum reactor conditions to maximize production of fermentable sugars and to minimize degradation of sugars to non-fermentable and inhibitory byproducts. Oligomer kinetics was not included in the model of xylose production, as indicated in the reaction mechanism shown in Eq. (2). The kinetic parameters were obtained based on xylose concentration alone, by essentially lumping both hemicellulose and oligomers together as a single component of biomass in the model. Thus, the model does not discriminate between xylose hydrolyzed directly from hemicellulose or xylose from xylan oligomer. The frequency of sampling for obtaining oligomer was too small and the concentrations of oligomer were not large enough to justify including oligomer kinetics in the kinetic model. Refinements in this kinetic model might consider the inclusion of oligomer kinetics more completely in the model.
Table 4 Kinetic parameters of biomass dilute acid hydrolysis Materials
Xylose formation 1
Wheat Straw Switchgrass Poplar Corn Stover Paper Birch Southern Red Oak Aspen Balsam Basswood Red Maple Switchgrass
Xylose degradation 1
References
A1 (min )
m1
E1 (kJ/mol)
A2 (min )
m2
E2 (kJ/mol)
2.25 · 1020 1.9 · 1021 3.3 · 1021 6.7 · 1016 2.67 · 1016 1.04 · 1014 1.53 · 1011–2.65 · 1017 7.53 · 104–2.78 · 1017 4.46 · 1011–2.63 · 1020 5.77 · 109–1.40 · 1017 1.89 · 107–1.03 · 1019
1.55 0.4 0.4 1.5 1.0 1.54 1.75 1.75 1.75 1.75 1.75
167.0 169.0 176.7 129.8 126.6 120.1 97.18–151.85 48.72–151.52 102.67–179.13 88.65–149.45 65.94–167.48
1.52 · 1015 3.8 · 1010 8.5 · 1010 3.7 · 1010 – – 6.51 · 1016 7.59 · 1015 2.52 · 1013 6.83 · 1013 3.73 · 1017
2.0 1.45 0.55 0.5 – – 1.0 0.9 1.2 1.0 1.4
141.0 99.5 102.0 98.4 – – 155.36 147.56 126.89 129.64 165.59
Ranganathan et al. (1985) Esteghlalian et al. (1997)
Maloney et al. (1985) Kim and Lee (1987) In this project
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863 100.0
Glucose
90.0
Xylose
80.0
Galactose Arabinose
% Yield
70.0
Mannose
60.0 50.0 40.0 30.0 20.0 10.0 0.0 Aspen
Balsam
Basswood
Red Maple
Switchgrass
Biomass Species Fig. 10. Overall average sugars yield of biomass species.
During recent years, increasing interest has been shown in the utilization of biomass as a renewable resource. Timber species used in this project are available in nearly every forest covered county of the Upper Midwest region. This woody biomass may be an important feedstock used to produce bioethanol in the future. Future studies with these timber species should focus on measuring mixture effects on pretreatment yields, on refining kinetic models for several reactor configurations, and on integrated studies that include subsequent conversion steps. Acknowledgements The authors wish to acknowledge support from NSF MUSES program (BES-0524872) for Shu Chiang Yat and NSF REU program (EEC 0453174) for Alan Berger. Donation of switchgrass samples was from Dr. Jim McMillan at the National Renewable Energy Laboratory. References Bienkowski, P.R., Ladisch, M.R., Voloch, M., Tsao, G.T., 1984. Acid hydrolysis of pretreated lignicellulose from corn residue. Biotechnol. Bioeng. Symp. 14, 511–523. Bienkowski, P.R., Ladisch, M.R., Narayan, R., Tsao, G.T., Eckert, R., 1987. ;Correlation of glucose (dextrose) degradation at 90–190 C in 0.4–20% acid’’. Chem. Eng. Commun. 51, 179–192. Brooks, R.E., Bellamy, W.D., Su, T.M. 1978. Bioconversion of Plant Biomass to Ethanol. Annual Report and Revised Research Plan. US Department of Energy, pp. 78.
3863
Esteghlalian, A., Hashimoto, A.G., Fenske, J.J., Penner, M.H., 1997. Modeling and optimization of the dilute-sulfuric-acid pretreatment of corn stover, poplar and switchgrass. Bioresour. Technol. 59, 129– 136. Garrote, G., Dominguez, H., Parajo, J.C., 2001. Generation of xylose solutions from Eucalyptus globulus wood by autohydrolysis-posthydrolysis processes: posthydrolysis kinetics. Bioresour. Technol. 79, 155–164. Grohmann, K., Torget, R., Himmel, M., 1986. Dilute acid pretreatment of biomass at high solids concentrations. Biotech. Bioeng. Symp. 17, 135– 151. Kim, S.B., Lee, Y.Y., 1987. Kinetics in acid-catalyzed hydrolysis of hardwood hemicellulose. Biotechnol. Bioeng. Symp. 17, 71–84. Ladisch, M., 1989. Hydrolysis. In: Kitani, O., Hall, C.W. (Eds.), Biomass Handbook. Gordon and Breach Science Publishers, New York, NY USA, pp. 434–451. Lloyd, T.A., Wyman, C.E., 2005. Combined sugar yields for dilute sulfuric acid pretreatment of corn stover followed by enzymatic hydrolysis of the remaining solids. Bioresour. Technol. 96, 1967–1977. Maloney, M.T., Chapman, T.W., Baker, A.J., 1985. Dilute-acid hydrolysis of paper birch: kinetics studies of xylan and acetyl group hydrolysis. Biotechnol. Bioengineering 27, 355–361. Mehlberg, R., Tsao, G.T. 1979. Low Liquid Hemicellulose Hydrolysis of Hydrochloric Acid. presented at 178th ACS National Meeting, Washington, DC, September. NREL Biomass Program. 2005. Determination of Sugars, Byproducts and Degradation Products in Liquid Fraction Process Samples. http:// devafdc.nrel.gov/pdfs/9462.pdf. LAP #013. NREL Biomass Program. 2006. Determination of Structural Carbohydrates and Lignin in Biomass. http://devafdc.nrel.gov/pdfs/9572.pdf. LAP #002. Ranganathan, S., MacDonald, D.G., Bakhshi, N.N., 1985. Kinetic studies of wheat straw hydrolysis using sulfuric acid. Can. J. Chem. Eng. 63, 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. US Department of Energy. 2005. United States Country Analysis Brief. http://www.eia.doe.gov/cabs/Usa/Oil.html. Accessed February 2006. Wyman, C.E., 1999. Biomass ethanol: technical progress, opportunities, and commercial challenges. Ann. Rev. Energ. Environ. 24, 189– 226. Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y., 2005a. Coordinated development of leading biomass pretreatment technologies. Bioresour. Technol. 96, 1959–1966. Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y., 2005b. Comparative sugar recovery data from laboratory scale application of leading pretreatment technologies to corn stover. Bioresour. Technol. 96, 2026–2032. Yat, S.C. 2006. Kinetic Characterization for Pretreatment of Timber Varieties and Switchgrass using Diluted Acid Hydrolysis. M.Sc. Thesis. Michigan Technological University.
Bioresource Technology 99 (2008) 3855–3863
Kinetic characterization for dilute sulfuric acid hydrolysis of timber varieties and switchgrass Shu Chiang Yat a, Alan Berger b, David R. Shonnard a
a,*
Biochemical Engineering Laboratory, Department of Chemical Engineering, Michigan Technological University, Houghton, MI 49931, United States b Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA 18602, United States Received 1 February 2007; received in revised form 18 June 2007; accepted 18 June 2007 Available online 29 September 2007
Abstract Hydrolysis of four timber species (aspen, balsam fir, basswood, and red maple) and switchgrass was studied using dilute sulfuric acid at 50 g dry biomass/L under similar conditions previously described as acid pretreatment. The primary goal was to obtain detailed kinetic data of xylose formation and degradation from a match between a first order reaction model and the experimental data at various final reactor temperatures (160–190 C), sulfuric acid concentrations (0.25–1.0% w/v), and particle sizes (28–10/20 mesh) in a glass-lined 1 L well-mixed batch reactor. Reaction rates for the generation of xylose from hemicellulose and the generation of furfural from xylose were strongly dependent on both temperature and acid concentration. However, no effect was observed for the particle sizes studied. Oligomer sugars, representing incomplete products of hydrolysis, were observed early in the reaction period for all sugars (xylose, glucose, arabinose, mannose, and galactose), but were reduced to low concentrations at later times (higher hemicellulose conversions). Maximum yields for xylose ranged from 70% (balsam) to 94% (switchgrass), for glucose from 10.6% to 13.6%, and for other minor sugars from 8.6% to 58.9%. Xylose formation activation energies and the pre-exponential factors for the timber species and switchgrass were in a range of 49–180 kJ/mol and from 7.5 · 104 to 2.6 · 1020 min1, respectively. In addition, for xylose degradation, the activation energies and the pre-exponential factors ranged from 130 to 170 kJ/mol and from 6.8 · 1013 to 3.7 · 1017 min1, respectively. There was a near linear dependence on acid concentration observed for xylose degradation. Our results suggest that mixtures of biomass species may be processed together and still achieve high yields for all species. 2007 Elsevier Ltd. All rights reserved. Keywords: Dilute acid; Hydrolysis; Switchgrass; Timber
1. Introduction With only 4.5% of the world’s population, the United States consumes about 25% of global energy and produces roughly 25% of the planet’s CO2 emissions (US DOE, 2005). The United States consumed an average of 20.6 million barrels of petroleum per day during the first nine months of 2005, the same amount year-over-year as in 2004 (US DOE, 2005). The average gasoline price for the third quarter of 2005 was $2.56 per gallon, up $0.67 per
*
Corresponding author. Tel.: +1 906 487 3468; fax: +1 906 487 3213. E-mail address: [email protected] (D.R. Shonnard).
0960-8524/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2007.06.046
gallon from the third quarter of 2004. Gasoline prices are affected by natural disasters such as hurricanes as shown in the aftermath of Katrina and Rita in 2005. Conversion of abundant cellulosic biomass to ethanol as a transportation fuel presents an important opportunity to improve energy security, reduce the trade deficit, reduce greenhouse gas emission, and improve price stability (Wyman, 1999). Improvements in pretreatment, such as dilute acid hydrolysis, will aid in preparing feedstocks for enzymatic hydrolysis and fermentation without generating significant concentrations of fermentation inhibitors. The goals of the pretreatment are to decompose the polymeric components of the wood and form monomer sugars, enhance enzymatic conversion of the cellulose
3856
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
fraction, and, hopefully, obtain a higher ethanol yield. Numerous pretreatment approaches have been investigated at many national laboratories, universities, and industrial locations over the past 25 years. In the past, it has been difficult to compare the performance and economics of these various approaches due to differences in feedstocks, chemical analysis methods, and data reporting methodologies. Recently, a group of pretreatment researchers across North America have begun to collaborate to investigate different pretreatment approaches on a common basis to allow meaningful comparison. These researchers have formed the Biomass Refining Consortium for Applied Fundamentals and Innovation (CAFI) to advance the efficacy and knowledge base of pretreatment technologies (Wyman et al., 2005a). Pretreatment methods are either physical or chemical, but some methods incorporate both effects. Physical pretreatment methods, such as comminution (mechanical reduction in biomass particulate size) and steam explosion, can be classified into mechanical and non-mechanical pretreatment. Physical forces used in mechanical pretreatments can subdivide lignocellulosic material into fine particles which are substantially susceptible to acid or enzymatic hydrolysis. Non-mechanical physical pretreatments cause decomposition of lignocellulosics by exposing them to harsh external forces other than mechanical forces. Chemical pretreatments have been used extensively for removal of lignin surrounding cellulose and for destroying its crystalline structure. Traditionally, the paper industry has utilized pulping processes for delignification of cellulosic materials to produce high strength, long fiber paper products. It has been considered, however, that these processes are quite severe and expensive to be used for pretreatment of lignocellulosics for production of ethanol. Biorefineries will be designed to process specific feedstocks, such as corn stover, switchgrass or other woody energy crops and forest resources. In forest regions, the feedstock will most likely be a mixture of timber species grown locally, possibly mixed with herbaceous biomass, rather than a monoculture feed. Flexible pretreatment process technologies that can handle biomass with variable compositions of cellulose, hemicellulose, lignin, and ash will be the most efficient. In this project, the kinetic characterization of several timber species was studied using diluted sulfuric acid hydrolysis. The goals of this research are (1) to measure the concentrations of fermentable monomer and oligomer sugars plus their non-fermentable byproducts; (2) to determine the kinetic parameters for xylose formation and degradation from a match between a reaction model and experimental data; and (3) interpret the single timber species results as to potential mixture effects on pretreatment performance. Kinetic parameters obtained from this study can in the future be used in reactor models of various configurations to identify optimum conditions for biomass conversion to fermentable sugars.
2. Methods 2.1. Raw material preparation Each tree species was harvested from forest lands in the vicinity of Michigan Technological University (Houghton, MI), cut into 15 inch lengths and debarked. The debarked wood logs were then cut into flakes using a rotary drum blade and then dried overnight in an oven at 100 C. Then, the wood flakes were hammer milled into fine particles and screened into fractions (10–30 mesh). The switchgrass sample was provided by National Renewable Energy Laboratory (NREL) and was also screened into fractions without further size reduction. The total carbohydrate, lignin and ash analysis of each species was studied based on the Laboratory Analytical Procedure (LAP #002) of ‘‘Determination of Structural Carbohydrates and Lignin in Biomass’’ (NREL, 2006). 2.2. Dilute sulfuric acid hydrolysis The hydrolysis experiments were carried out using a 1 L high pressure/high temperature Parr reactor model 4571 (Moline, IL USA). This stainless-steel reactor was equipped with a glass liner, a heat exchanger, a 6 anglebladed impeller, and temperature controller (Parr controller model 4842). Each dry biomass sample (25 g) was mixed with 500 mL of 0.25–1.0% w/w sulfuric acid and was heated to the desired maximum temperature (160, 175, and 190 C) inside the closed reactor. To control maximum reactor temperature, compressed air was used as the heat removal fluid in the cooling circuit and was controlled manually to regulate the temperature to within ± 1 C. The reactor was closed and placed in the reactor oven to initiate heating. The stirrer speed was adjusted to 50 rpm and the reactor temperature was measured using a thermocouple probe connected to the control unit. Temperature was recorded every 5 min until the maximum was reached (over approximately a 70 min heat up period) and thereafter every 8 min. Samples (5 ml) were collected (after purging the sample dip tube of 8 ml solution) at 100 C, 135 C, and thereafter at 3 min intervals until reaching the maximum temperature, Tmax (160, 175, and 190 C). Tmax was maintained for 32 min and four 5-mL samples were collected at 8 min intervals. Each collected samples was cooled in an ice bath to stop the reaction. The cooled samples were filtered using a 0.2 lm nitrocellulose membrane to remove suspended solids. Filtered samples were then separated into two vials: one for monomeric sugars analysis and one for total sugars analysis. The monomeric sugar content was analyzed using HPLC after the samples were neutralized using 6 N NaOH to pH 5 to 6 (measured using litmus paper). The total sugar analysis was used to identify the amount of oligomers in each sample obtained during the course of the hydrolysis reaction (NREL, 2005). One milliliter of
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
each filtered sample was dispensed into an amber crimptopped HPLC vial. The samples were acidified to 4% acid concentration using 96% H2SO4, crimped tightly, and then autoclaved for one hour at 121 C to hydrolyze oligomers to monomer sugars. After completion of the autoclave cycle, each sample was subsequently neutralized to pH 5 to 6 using 6N NaOH. Because a precipitate formed after the neutralizing step the samples were placed into a centrifuge (MARATHON 21 K, Fisher Scientific (Pittsburgh, PA, USA)) at 8000 rpm for 5 min. After completion of the centrifuge cycle, the clear solution was carefully and slowly drawn out from the vials (to avoid reintroduction of solids) and transferred to new HPLC vials. 2.3. HPLC analysis Sugars and pretreatment byproducts were measured using an Agilent Series 1100 HPLC (Santa Clara, CA, USA). Separation was accomplished using a Bio-Rad Aminex HPX-87P column with deashing cartridge (BioRad) and quantified with a refractive index detector for sugars and a diode-array detector for fufural and acetic acid. 2.4. Kinetic model development The kinetics of hemicellulose hydrolysis was first modeled by Mehlberg and Tsao (1979) in the following reaction: ↓
↓
↑
Oligomers → Xylose → Degradation Products
Hemicellulose
ð1Þ
→ Xylan II →
The concentrations of xylan I, xylan II and oligomers are very difficult to determine and since the reaction to produce xylan I and xylan II is very rapid at higher temperatures above 160 C (Saeman, 1945; Grohmann et al., 1986; Bienkowski et al., 1984; Ladisch, 1989), the model can be simplified to the following: k1
dH ¼ k 1 H dt dX H ¼ k1 k2X dt 0:88
k2
Hemicellulose ðH Þ! XyloseðX Þ ! Degradation Products ðDÞ ð2Þ
Two models were developed in this project. The first model includes the first step in conversion of hemicellulose to xylose. The second model involves the process of xylose degradation. 2.4.1. Hemicellulose hydrolysis and xylose model Eqs. (3) and (4) are the mass balance equations for H and X in a well-stirred batch reactor under the assumption of constant reactor liquid volume and first order reaction for each species:
ð3Þ ð4Þ
where k1 and k2 are first order reaction rate constants for xylose formation and xylose degradation, respectively. The reaction rate constant, k1, is assumed to have Arrhenius-type temperature dependence: E1 k 1 ¼ A1 exp ð5Þ RT 1 where E1 = activation energy (kJ/mol), R = 8.3143 · 103 (kJ/molK), T1 = temperature (K), A1 = pre-exponential factor (min1). The pre-exponential factors for hemicellulose hydrolysis reactions are assumed to be dependent upon acid concentration: A1 ¼ A1o C m1
ð6Þ
where A1o = pre-exponential factor for hemicellulose hydrolysis (min1), C = acid concentration (% w), m1 = acid concentration exponent for the rate constant k1 (dimensionless). As the reaction temperature changed over time, the reaction rate constant model changed as a function of time as shown in Eq. (7). E1 m1 k 1 ¼ A1o C exp ð7Þ RT 1 ðtÞ Therefore, a numerical solution of Eq. (3) at each new time step (i + 1) is required: H iþ1 ¼ H i k 1i H i Dt
→ Xylan I → ↑
3857
where
E1 k 1i ¼ Aio C m1 exp RT 1 ðtÞ
ð8Þ ð9Þ
Hence, the xylose concentration obtained from the solution of Eq. (3) is X iþ1 ¼
H max H iþ1 0:88
ð10Þ
where Hmax = the maximum concentration of xylan that can be produced in each species (g xylose oligomer/L) as determined by the total carbohydrate analysis. In Eq. (10) it is assumed that negligible furfural is generated from xylose during the heat up period (period of time for application of the model), which is an approximation that will be shown to be a valid assumption. The factor of 0.88 is the ratio of molecular weight of xylan to xylose (132/150). Eqs. (8)–(10) were integrated using Microsoft Excel and the Trapezoidal Rule of integration with a time step (Dt) of 0.01 min., which was determined to be the optimum time step for the minimization or numerical error (Yat, 2006). The kinetic parameters, A1 and E1, were determined using Excel Solver function with initial guesses. The sum of squared errors between experimental data and model predictions were minimized by improving the values of the A1 and E1 using Eqs. (8)–(10).
3858
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
2.4.2. Xylose degradation model Eq. (4) was used to model the formation of xylose degradation products. The degradation products formed during the maximum temperature period. It was assumed that all of the reactive hemicellulose (xylan) was hydrolyzed completely to xylose (H = 0) prior to xylose degradation. Therefore, Eq. (4) becomes dX ¼ k 2 X dt
a relatively small number of representative graphs of concentration versus time are shown. 3.1. Biomass composition The composition of the five feedstocks is shown in Table 1 with a comparison of results obtained in this study to the literatures value for the same biomass species. The major components of these substrates were xylan, glucan, and lignin. Measured glucan and xylan in this study were generally similar to the literature values, with the exception of switchgrass whose glucan value is higher in this study. Lignin content measured in this study for all species is generally higher than in the literature studies, due to the generation of tars which become associated with the lignin fraction. The minor sugars of galactan, araban, and mannan measured in this study were comparable to the literature values.
ð11Þ
and the solution for the initial condition of t = 0, X = Xmax (where Xmax is measured) is given by Eq. (12). ln
X X max
¼ k 2 t
ð12Þ
By plotting the experimental data according to the form of Eq. (12), k2 was determined as the slope of the linear regression fit to the data. Eq. (13) is used to determine the pre-exponential factor (A2) and activation energy (E2) of xylose degradation: ln k 2 ¼ ln A2
E2 1 R T2
ð13Þ
3.2. Monosaccharides and oligosaccharides formation
where T2 is the value of maximum temperature achieved in several experiments. The value of k2 determined at each maximum temperature as per Eq. (12) was used to determine the kinetic parameters in Eq. (13) using linear regression analysis. To obtain the acid concentration exponent (m2) for the xylose degradation model, Eq. (14) is used: E2 ln k 2 ¼ m2 ln C þ ln A2 exp ð14Þ RT 2
The xylan contents from structural carbohydrate of various biomass species used in this project was found to be range from 6% to 19% (dry basis) using the procedure provided by NREL LAP #002 (NREL, 2006). Thus, the maximum possible xylose concentration from these hydrolysis experiments was 3–9 g/L, depending on biomass type. A typical plot of the time-temperature profile using the Parr reactor is shown in Fig. 1. The reactor temperature increased almost linearly after 40 min from the start of the heating period until the Tmax was reached at about 70 min. for Tmax = 175 C. Over this time period most of the hemicellulose was hydrolyzed to sugars, as shown in subsequent figures, and a linear trendline was fit to the temperature data for use in kinetic modeling. By way of comparison to the literature methods, in most of the studies reported, the acid was injected after the sample was heated to the desired temperature (Lloyd and Wyman, 2005; Esteghlalian et al., 1997; Garrote et al., 2001), or the acidified solution was heated quickly from
By plotting the natural log of calculated k2 from Eq. (12) at each concentration experiment as a function of natural log of acid concentration, the slope of the linear regression is equal to m2. 3. Results and discussion Due to the large number of pretreatment experiments conducted for each biomass species (8 experiments), only Table 1 Compositional analysis of raw biomass samples (percent by weight) Species
% Biomass composition
References
Glucan
Xylan
Galactan
Araban
Mannan
Lignin
Aspen Balsam Red Maple
57.3 46.8 46.6
16.0 4.8 17.3
0.8 1.0 0.6
0.4 0.5 0.5
2.3 12.4 3.5
16.3 29.4 24.0
Brooks et al. (1978)
Switchgrass
32.2
20.3
–
3.7
0.4
19.5
Esteghlalian et al. (1997)
Aspen Balsam Basswood Red Maple Switchgrass
52.43 47.09 43.99 43.18 47.72
14.60 6.23 15.31 17.69 19.06
3.52 5.45 3.41 5.71 4.18
2.41 5.41 3.49 4.13 8.11
5.32 11.49 2.91 5.37 6.30
26.69 36.04 28.44 36.49 26.04
This study
7
Monomer
6
Oligomer
5 4 3 2 1
0
20
40
60
80
100
0
120
50
Time (min)
60
70
80
90
100
110
Time (min)
Fig. 1. Basswood acid hydrolysis time-temperature profile at 0.5% H2SO4, Tmax = 175 C, 20–10 mesh.
room temperature to Tmax by plunging the reactor into a hot sand bath. Monosaccharide data for biomass hydrolysis at 0.5% sulfuric acid, 175 C, and 20–10 mesh particle size is displayed in Fig. 2. The figure shows the time-concentration data of each sugar plus furfural, one of the main byproducts of dilute acid hydrolysis (along with insoluble tars, which were not measured in this work). The concentrations of xylose and other sugars were small prior to achieving 135 C in the reactor. At temperatures above 135 C, the xylose concentration increased rapidly until Tmax was reached. Afterward, the concentration of xylose decreased while the temperature was maintained constant. Similar trends in the concentrations of the hemicellulose minor sugars were also observed. After Tmax was reached, significant amounts of glucose and furfural were formed. Furfural is the degradation product produced from xylose and glucose was formed from the hydrolysis of the cellulose fraction. The highest concentration of xylose oligomer appeared early in the pretreatment process as shown in Fig. 3, prior to the production of significant concentrations of monosaccharide. Oligomer formation may be beneficial during hydrolysis since monomer sugars can be formed subsequently even at conditions different from the reactor. A similar trend was observed for other sugar oligomers and for other biomass species. After reaching Tmax the concentra-
Fig. 3. Aspen acid hydrolysis at 0.5% H2SO4, Tmax = 175 C, 20–10 mesh (xylose oligomer & monomer profile).
tion of xylose oligomer decreased as it formed monomer; however, furfural was produced simultaneously. At the time in the reaction when maximum xylose concentration was achieved, oligomer concentration already had decreased to a small amount, relative to maximum oligomer concentration. This maximum total xylose concentration occurred before substantial furfural had accumulated. Thus, careful control of reactor temperature and residence time is needed to optimize monomer concentration and minimize non-fermentable byproduct formation. 3.3. Maximum temperature effects Maximum temperature was one of the three process variables investigated. The temperature of the reactor was ramped from room temperature to the Tmax (160, 175, and 190 C). A typical plot of the effect of different Tmax on dilute acid hydrolysis of biomass is shown in Fig. 4. The concentrations of xylose observed during the temperature ramp to Tmax are nearly identical, indicating the good reproducibility of the temperature ramp in these experiments. For the experiment at a maximum temperature of 160 C, the maximum concentration of xylose was delayed and its magnitude was lower than for higher Tmax experiments. In addition, the rate of xylose degradation was small relative to the higher temperature experiments. 3
8
6
Xylose Galactose
5
Arabinose
Xylose Concentration (g/L)
Glucose
7
Concentration (g/L)
3859
8
200 180 160 140 120 100 80 60 40 20 0
Xylose Concentration (g/L)
Temperature (ºC)
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
Mannose
4
Furfural
3 2 1
160 C
2.5
175 C 190 C
2 1.5 1 0.5 0
0 45
55
65
75
85
95
105
Time (min) Fig. 2. Basswood acid hydrolysis at 0.5% H2SO4, Tmax = 175 C, 20– 10 mesh (monomers content).
40
50
60
70
80
90
100
110
Time (min) Fig. 4. Balsam acid hydrolysis at 0.5% H2SO4, 20–10 mesh (xylose profile for different Tmax).
3860
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
In the Tmax = 190 C experiment, xylose degraded rapidly within 10 min to near zero concentration. 3.4. Acid concentration effects Another process variable that affects hemicellulose hydrolysis and xylose formation/degradation during pretreatment is the acid concentration of the reaction medium. A representative plot is displayed in Fig. 5 on the effect of acid loading on xylose concentration for all of the biomass species tested. As the concentration of acid increased, the rate of xylose formation and degradation increase. For example, at 0.25% sulfuric acid, the rate of xylose formation was relatively slow and more time was required to achieve maximum xylose concentration. At the maximum acid concentration of 1.0%, the rate of xylose degradation after achieving maximum temperature of 175 C was the highest for these experiments. 3.5. Xylose formation and degradation kinetic parameters
Xylose Concentration (g/L)
The xylose formation model was fitted to the experimental data in order to determine the kinetic parameters of xylose formation. A match between the model and the experimental data which is representative of all experiments is shown in Fig. 6. As discussed before, to get the 9 8 7 6 5 4 3 2 1 0 40
50
60
70
80
90
100
110
Time (min) 0.25% w/w Sulfuric Acid
0.5% w/w Sulfuric Acid
1.0% w/w Sulfuric Acid
Fig. 5. Red maple acid hydrolysis at Tmax = 175 C, 20–10 mesh (xylose profile for various sulfuric acid concentrations).
8 Predicted Model
7
Experimental Data
Concentration (g/L)
6 5 4
best fit the sum of squared error between the theoretical model and the experimental data was minimized by varying the pre-exponential factor and the activation energy. The kinetic parameters of woody biomass pretreatment during xylose formation are summarized in Table 2 at different experimental conditions. The pre-exponential factors for all species and reactor conditions ranged from 7.53 · 104 min1 to 2.63 · 1020 min1 and the activation energies varied from 49 kJ/mol to 179 kJ/mol. The xylose degradation kinetics was calculated using the equations obtained from Section 2.4.2. Best fit of Eqs. (12)–(14) to degradation kinetic data that is representative of all experiments is shown in Figs. 7–9. The kinetic parameters and the acid concentration exponents are summarized in Table 3. Kinetic parameters obtained from this research are compared in Table 4 to other studies for related biomass species even though the conduct of our experiments was different from the literature studies, as mentioned earlier. The values of A, E, and m are in general comparable to the literature values. The range of A1 and E1 values reported in this work are a result of conducting experiments under different temperature, acid concentration, and particle size conditions. 3.6. Overall sugars yield The average sugar yield of each woody species obtained from the eight experiment runs for each species is found in Fig. 10 (Yat, 2006). Yield was calculated at the point in the reaction when maximum xylose concentration occurred. All yields reported were normalized to the total potential sugars in the original untreated feedstock obtained from the total carbohydrate analysis in Table 1. The overall xylose yield was the major interest in this project since this sugar is the dominant monomer in hemicellulose. Average xylose yield for each species varied from about 70% for balsam to 92% for switchgrass. No clear trend could be seen for xylose yield under different process conditions (Tmax, C) (Yat, 2006). The glucose yield varied from only 11% to 13% for all species, showing that dilute acid is ineffective in completely hydrolyzing cellulose, although at longer times in the reactor at Tmax glucose continued to accumulate, as did degradation byproducts. Yields of the minor sugars were generally below 50% and often less than 20%. One notable exception was a high yield for mannose from balsam, which has a high percentage of mannose in the hemicellulose. 3.7. Discussion of results
3 2 1 0 45
55
65
75
85
95
105
Time (min) Fig. 6. Aspen acid hydrolysis at 0.5% H2SO4, Tmax = 160 C, 20–10 mesh (comparison of xylose formation model to xylose experimental data).
Recently, most pretreatment research has focused on comparing methods, such as dilute acid, AFEX, steam explosion and acid-alkali pretreatment, usually with single biomass species (Wyman et al., 2005b). The results reported here were on a single method applied to multiple woody biomass species and switchgrass in separate experiments. Acid hydrolysis of different biomass species are
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863
3861
Table 2 Kinetic parameters during xylose formation at various reactor conditions Acid concentration (% w)
Particle size
Kinetic parameters
Temperature (C) 160
Aspen 0.25
20–10
0.50
20–10 >28
1.00
20–10
Balsam 0.25
20–10
0.50
20–10 >28
1.00
20–10
Basswood 0.25
20–10
0.50
20–10 >28
1.00
20–10
Red Maple 0.25
20–10
0.50
20–10 >28
1.00
20–10
Switchgrass 0.25
20–10
0.50
20–10 >28
1.00
20–10
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1) E1 (kJ/mol) A1 (min1)
illustrated by the data presented in this paper, and the hydrolysis of hemicellulose to xylose is likely a factor in the pretreatment of biomass materials when acid is present. Our results on single biomass species provide insights as to how mixtures of biomass species might respond to dilute acid hydrolysis. Our results suggest that mixtures of biomass species may be processed together and still achieve high yields for all species. The only exception to this conclusion might be balsam whose average yield was lower
140.70 2.80 · 1016
89.65 8.35 · 109
154.36 7.58 · 1017
145.76 1.11 · 1017
115.11 2.48 · 1013
175
175
190
151.85 2.65 · 1017 134.70 5.47 · 1015 120.23 4.85 · 1013 97.18 1.53 · 1011
136.10 7.55 · 1015 116.85 1.90 · 1013
148.60 1.88 · 1017
151.52 2.78 · 1017 67.83 1.54 · 107 58.12 1.00 · 106 48.72 7.53 · 104
71.74 4.64 · 107 74.40 1.25 · 108
67.37 1.18 · 107
179.13 2.63 · 1020 126.27 2.42 · 1014 139.08 6.63 · 1015 102.67 4.46 · 1011
134.44 2.58 · 1015 130.94 7.55 · 1014
117.36 1.94 · 1013
149.45 1.40 · 1017 104.07 6.42 · 1011 111.53 5.56 · 1012 98.31 2.03 · 1011
88.65 5.77 · 109 96.41 5.86 · 1010
88.16 6.63 · 109
167.48 1.03 · 1019 70.75 8.56 · 107 101.29 3.72 · 1011 78.24 1.21 · 109
74.36 1.87 · 108 111.07 6.90 · 1012
65.94 1.89 · 107
than the other species tested. The kinetic parameters obtained in this study (except for balsam) were similar and the reactor residence times for reaching maximum xylose generation were nearly identical. However, mixture effects (interactions that might affect species reactivities) must be investigated to confirm these initial observations. The data measured in this study confirm that a kinetic model including irreversible first order reactions for xylose production and degradation is appropriate for this system.
3862
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863 Table 3 Kinetic parameters of timber varieties plus switchgrass during xylose degradation
0 -0.5
ln (X/Xm)
-1 -1.5 -2 -2.5
y = -0.1400x - 0.0133
-3
2
R = 0.9975
-3.5 -4 0
5
10
15
20
25
30
Time (min) Fig. 7. Red maple acid hydrolysis at 0.5% H2SO4, Tmax = 190 C, 20– 10 mesh (xylose degradation model). 0 -0.5 -1
ln (k2)
-1.5
y = -15593x + 31.855
-2 -2.5
R2 = 0.9721
-3 -3.5 -4 -4.5 -5 0.00215
0.0022
0.00225
0.0023
0.00235
1/T (K-1) Fig. 8. Red maple acid hydrolysis kinetic parameters determination of xylose degradation using Eq. (13). 0 -0.5 -1
ln (k2)
-1.5 -2
y = 1.023x - 2.2257 2
R = 0.9435
-2.5 -3 -3.5 -4 -1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
ln (C) Fig. 9. Red maple acid hydrolysis acid concentration exponent determination of xylose degradation using Eq. (14).
Aspen Balsam Basswood Red Maple Switchgrass
Pre-exponential factor, A2 (min1)
Acid concentration exponent, m2
Activation energy, E2 (kJ/ mol)
6.51 · 1016 7.59 · 1015 2.52 · 1013 6.83 · 1013 3.73 · 1017
1.0 0.9 1.2 1.0 1.4
155.36 147.56 126.89 129.64 165.59
The concentrations of monomer sugars in these experiments were dilute (less than 1%). However, the observation of irreversible reaction kinetics may not apply to higher concentrations anticipated in commercial applications, as shown before for acid hydrolysis for glucose production at higher substrate concentrations (Bienkowski et al., 1987). The derived kinetic model in this paper was employed to predict xylose concentration as a function of time throughout the experiment. The model was fitted to the experiment data in order to determine the relevant kinetic parameters. After the kinetic data was obtained, it is possible to use these parameters to predict pretreatment hydrolysis performance for various reactor configurations. Modeling reactor performance will allow for the determination of optimum reactor conditions to maximize production of fermentable sugars and to minimize degradation of sugars to non-fermentable and inhibitory byproducts. Oligomer kinetics was not included in the model of xylose production, as indicated in the reaction mechanism shown in Eq. (2). The kinetic parameters were obtained based on xylose concentration alone, by essentially lumping both hemicellulose and oligomers together as a single component of biomass in the model. Thus, the model does not discriminate between xylose hydrolyzed directly from hemicellulose or xylose from xylan oligomer. The frequency of sampling for obtaining oligomer was too small and the concentrations of oligomer were not large enough to justify including oligomer kinetics in the kinetic model. Refinements in this kinetic model might consider the inclusion of oligomer kinetics more completely in the model.
Table 4 Kinetic parameters of biomass dilute acid hydrolysis Materials
Xylose formation 1
Wheat Straw Switchgrass Poplar Corn Stover Paper Birch Southern Red Oak Aspen Balsam Basswood Red Maple Switchgrass
Xylose degradation 1
References
A1 (min )
m1
E1 (kJ/mol)
A2 (min )
m2
E2 (kJ/mol)
2.25 · 1020 1.9 · 1021 3.3 · 1021 6.7 · 1016 2.67 · 1016 1.04 · 1014 1.53 · 1011–2.65 · 1017 7.53 · 104–2.78 · 1017 4.46 · 1011–2.63 · 1020 5.77 · 109–1.40 · 1017 1.89 · 107–1.03 · 1019
1.55 0.4 0.4 1.5 1.0 1.54 1.75 1.75 1.75 1.75 1.75
167.0 169.0 176.7 129.8 126.6 120.1 97.18–151.85 48.72–151.52 102.67–179.13 88.65–149.45 65.94–167.48
1.52 · 1015 3.8 · 1010 8.5 · 1010 3.7 · 1010 – – 6.51 · 1016 7.59 · 1015 2.52 · 1013 6.83 · 1013 3.73 · 1017
2.0 1.45 0.55 0.5 – – 1.0 0.9 1.2 1.0 1.4
141.0 99.5 102.0 98.4 – – 155.36 147.56 126.89 129.64 165.59
Ranganathan et al. (1985) Esteghlalian et al. (1997)
Maloney et al. (1985) Kim and Lee (1987) In this project
S.C. Yat et al. / Bioresource Technology 99 (2008) 3855–3863 100.0
Glucose
90.0
Xylose
80.0
Galactose Arabinose
% Yield
70.0
Mannose
60.0 50.0 40.0 30.0 20.0 10.0 0.0 Aspen
Balsam
Basswood
Red Maple
Switchgrass
Biomass Species Fig. 10. Overall average sugars yield of biomass species.
During recent years, increasing interest has been shown in the utilization of biomass as a renewable resource. Timber species used in this project are available in nearly every forest covered county of the Upper Midwest region. This woody biomass may be an important feedstock used to produce bioethanol in the future. Future studies with these timber species should focus on measuring mixture effects on pretreatment yields, on refining kinetic models for several reactor configurations, and on integrated studies that include subsequent conversion steps. Acknowledgements The authors wish to acknowledge support from NSF MUSES program (BES-0524872) for Shu Chiang Yat and NSF REU program (EEC 0453174) for Alan Berger. Donation of switchgrass samples was from Dr. Jim McMillan at the National Renewable Energy Laboratory. References Bienkowski, P.R., Ladisch, M.R., Voloch, M., Tsao, G.T., 1984. Acid hydrolysis of pretreated lignicellulose from corn residue. Biotechnol. Bioeng. Symp. 14, 511–523. Bienkowski, P.R., Ladisch, M.R., Narayan, R., Tsao, G.T., Eckert, R., 1987. ;Correlation of glucose (dextrose) degradation at 90–190 C in 0.4–20% acid’’. Chem. Eng. Commun. 51, 179–192. Brooks, R.E., Bellamy, W.D., Su, T.M. 1978. Bioconversion of Plant Biomass to Ethanol. Annual Report and Revised Research Plan. US Department of Energy, pp. 78.
3863
Esteghlalian, A., Hashimoto, A.G., Fenske, J.J., Penner, M.H., 1997. Modeling and optimization of the dilute-sulfuric-acid pretreatment of corn stover, poplar and switchgrass. Bioresour. Technol. 59, 129– 136. Garrote, G., Dominguez, H., Parajo, J.C., 2001. Generation of xylose solutions from Eucalyptus globulus wood by autohydrolysis-posthydrolysis processes: posthydrolysis kinetics. Bioresour. Technol. 79, 155–164. Grohmann, K., Torget, R., Himmel, M., 1986. Dilute acid pretreatment of biomass at high solids concentrations. Biotech. Bioeng. Symp. 17, 135– 151. Kim, S.B., Lee, Y.Y., 1987. Kinetics in acid-catalyzed hydrolysis of hardwood hemicellulose. Biotechnol. Bioeng. Symp. 17, 71–84. Ladisch, M., 1989. Hydrolysis. In: Kitani, O., Hall, C.W. (Eds.), Biomass Handbook. Gordon and Breach Science Publishers, New York, NY USA, pp. 434–451. Lloyd, T.A., Wyman, C.E., 2005. Combined sugar yields for dilute sulfuric acid pretreatment of corn stover followed by enzymatic hydrolysis of the remaining solids. Bioresour. Technol. 96, 1967–1977. Maloney, M.T., Chapman, T.W., Baker, A.J., 1985. Dilute-acid hydrolysis of paper birch: kinetics studies of xylan and acetyl group hydrolysis. Biotechnol. Bioengineering 27, 355–361. Mehlberg, R., Tsao, G.T. 1979. Low Liquid Hemicellulose Hydrolysis of Hydrochloric Acid. presented at 178th ACS National Meeting, Washington, DC, September. NREL Biomass Program. 2005. Determination of Sugars, Byproducts and Degradation Products in Liquid Fraction Process Samples. http:// devafdc.nrel.gov/pdfs/9462.pdf. LAP #013. NREL Biomass Program. 2006. Determination of Structural Carbohydrates and Lignin in Biomass. http://devafdc.nrel.gov/pdfs/9572.pdf. LAP #002. Ranganathan, S., MacDonald, D.G., Bakhshi, N.N., 1985. Kinetic studies of wheat straw hydrolysis using sulfuric acid. Can. J. Chem. Eng. 63, 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. US Department of Energy. 2005. United States Country Analysis Brief. http://www.eia.doe.gov/cabs/Usa/Oil.html. Accessed February 2006. Wyman, C.E., 1999. Biomass ethanol: technical progress, opportunities, and commercial challenges. Ann. Rev. Energ. Environ. 24, 189– 226. Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y., 2005a. Coordinated development of leading biomass pretreatment technologies. Bioresour. Technol. 96, 1959–1966. Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y., 2005b. Comparative sugar recovery data from laboratory scale application of leading pretreatment technologies to corn stover. Bioresour. Technol. 96, 2026–2032. Yat, S.C. 2006. Kinetic Characterization for Pretreatment of Timber Varieties and Switchgrass using Diluted Acid Hydrolysis. M.Sc. Thesis. Michigan Technological University.