- Email: [email protected]
Optimizing thermomechanical pretreatment conditions to enhance enzymatic hydrolysis of wheat straw by response surface methodology
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Available at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Optimizing thermomechanical pretreatment conditions to enhance enzymatic hydrolysis of wheat straw by response surface methodology Z. Maache-Rezzoug a, G. Pierre b, A. Nouviaire a, T. Maugard b, S.A. Rezzoug a,* a
LEPTIAB, Universite´ de La Rochelle - Poˆles Sciences et Technologie, Baˆtiment Marie Curie, Avenue Michel Cre´peau, 17042 La Rochelle, France UMR 6250 CNRS-ULR, LIENSs, Universite´ de La Rochelle, Poˆles Sciences et Technologie, Baˆtiment Marie Curie, Avenue Michel Cre´peau, 17042 La Rochelle, France b
article info
abstract
Article history:
Wheat straw was pretreated with a thermomechanical process developed in our laboratory
Received 1 March 2010
to improve the enzymatic hydrolysis extent of potentially fermentable sugars. This process
Received in revised form
involves subjecting the lignocellulosic biomass for a short time to saturated steam pres-
5 April 2011
sure, followed by an instantaneous decompression to vacuum at 5 kPa. Increasing of the
Accepted 15 April 2011
heat induced by saturated steam result in intensive vapour formation in the capillary
Available online 11 May 2011
porous structure of the plant material and the subsequent release of the pressure to vacuum allows fixing the expanded structure. Response surface methodology (RSM) based
Keywords:
on central composite design was used to optimize three independent variables of the
Thermomechanical pretreatment
pretreatment process: processing pressure (300e700 kPa), initial moisture contents of
Wheat straw
wheat straw (10e40%) and processing time (3e62 min). The process was optimised for
Enzymatic hydrolysis
hydrolysis yield and initial hydrolysis rate obtained by enzymatic hydrolysis on the pre-
Response surface methodology
treated solids by Celluclast (1.5 L). The analysis of variance (ANOVA) revealed that, among
Hydrolysis yield
the process variables, processing pressure and processing time have the most significant
Initial hydrolysis rate
effect on the hydrolysis yield and on initial rate of hydrolysis whereas initial moisture content observed significantly lower effect on the two responses. The predicted hydrolysis yield and in a lesser extent the predicted initial rate of hydrolysis agreed satisfactorily with the experimental values with R2 of 96% and 86% respectively. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Currently, corn is the primary raw material for ethanol production in the United States [1] and sugarcane in Brazil [2]. However, the corn and sugarcane to ethanol industry draws its feedstock from food and is quite mature with little possibility of process improvement. Biofuels from food sources are known as first-generation biofuels. Lignocellulosic biomass is an interesting and necessary enlargement of the biomass used for the production of renewable biofuels. It can be used as
a potential substrate to produce ethanol which is considered as one of the most important renewable fuels contributing to the reduction of negative environmental impacts generated by the worldwide utilisation of fossil fuels [3]. These oils are known as second-generation biofuels. Lignocellulosic materials to be considered for ethanol production include wood, crops from annual plants, agricultural residue and waste paper [4]. However, there are a lot of challenges and obstacles such as cost, technology and environmental issues that need to be overcome. Considerable research efforts have been made to
* Corresponding author. Tel.: þ33 5 46 45 86 15; fax: þ33 5 46 45 82 41. E-mail address: [email protected] (S.A. Rezzoug). 0961-9534/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.biombioe.2011.04.012
3130
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
improve conversion yields of lignocellulosic materials by the insertion of pretreatment step before to enzymatic hydrolysis. The purposes of pretreatment is increasing of porosity as well as removing the lignin to make lignocellulosic materials more accessible to enzymes by destroying the cell structure and break down the various physical and chemical barriers [5,6]. These pretreatments can be physical, physicochemical, chemical or biological [7] and can be combined between them, with the objective to obtain a high yield from enzymatic hydrolysis, to generate small quantities of co-products and inhibitors of fermentation while reducing costs. The purely physical processes of pretreatment can be summarized by an intense mechanical grinding in order to increase the accessible surfaces to enzymes, thus supporting the further hydrolysis and reducing the polymerisation degree of cellulose. This kind of pretreatment can be interesting if the granulometry is very low, which limits its interest and its use due to the energy cost of grinding ( < 2 mm), [8]. Steam explosion pretreatment consists to expose the biomass up to high pressure (1500e5000 kPa) and temperature (180e250 C) in presence of steam during a determined time, up to 90 min [9], followed by a rapid reduction in pressure, in order to breakdown the lignocellulosic structure. This technology was implemented at industrial scale with batch processes (Stake and Iotech) [10]. The treatment leads to a partial self-hydrolysis of hemicelluloses, depolymerisation of lignin and a destructuration of cellulose, largely dependent on treatment temperature. However, the yields of enzymatic hydrolysis are about 50% and some inhibiting compounds of the fermentation appeared following the partial pyrolysis of cellulose during the steam explosion. Steam explosion is often combined to acidification with H2SO4 or SO2 [1,11], to improve the yield of hydrolysis (90% of potential glucose and nearly 100% of sugars resulting from hemicelluloses) at moderate temperatures (150e200 C against 250 C in absence of catalyst) during 15e20 min, thus minimizing the formation of sugar degradation products. According to the type of biomass and for the same processing pretreatment conditions, steam explosion with acid catalysis (H2SO4) exhibited better results in terms of final sugar yield than other techniques in which catalysts as NH3 are used such as AFEX process (Ammonia Fiber EXplosion) developed in USA [12]. Among other advantages of steam explosion, it was noted [10] a low energy consumption, small quantities of generated undesirable products, a simplicity of implementation in batch, low quantities of chemical reagents and a good adaptation to the big particles size such as shavings. Steam explosion is thus considered to be one of the most promising methods to make biomass more accessible to enzymes. In order to obtain higher yields of enzymatic hydrolysis than those described in the literature and limiting the production of inhibitors of alcoholic fermentation, a thermomechanical pretreatment termed ’’D.I.C.’’ process (in french: De´tente Instantane´e Controˆle´e) was performed on wheat straw. This physical process is close to steam explosion technology [13,14]. The first difference is the moderate processing pressure for the proposed pretreatment. 0.7 MPa in this study compared to at least 5 MPa for steam explosion, as reported in the literature [15,16]. The difference is that the D.I.C treatment comprises two additional steps: instauration of initial vacuum before injection of the saturated steam and explosion towards vacuum.
The first step allows to reduce the resistance of air and thus to facilitate the diffusion of steam into the product. Consequently, the time necessary to reach the steam equilibrium temperature is reduced. The second step consists to an abrupt decompression which carries out towards vacuum (5 kPa) instead of atmospheric pressure like it is the case with the steam explosion process [17]. Due to the instantaneous character of this transformation as well as the adiabatic nature of the transition of steam inside the product, the water vaporisation induces a fast cooling inhibiting the possible further thermal reactions and thus contributing to reduction of undesired degrading products. The temperature is quickly stabilized at a balance temperature of the considered final pressure, limiting the reactions of degradations. This study aims to assess the effect three processing parameters of D.I.C pretreatment: processing steam pressure, initial moisture content of wheat straw and processing time on two responses: the hydrolysis yield and initial rate of hydrolysis, through response surface methodology (RSM). RSM is defined as a statistical method using quantitative data from an appropriate experimental design to determine and simultaneously solve multivariate equations. The main advantage of RSM is the reduced number of experimental trials needed to evaluate multiple parameters and their interactions.
2.
Materials and methods
2.1.
Plant material
Wheat straw used in this study grown in Charente-Maritime region in France and was ground in Gindomix (GM 200) Retsch crusher at 7500 RPM during 50 s and then sieved to obtain two particle sizes: 600e1000 mm and 50e600 mm. These two sizes were firstly studied to determine the effect of particle size. The third size was obtained by calibrating in sieve to obtain a fraction ranging between 1000 and 8000 mm. Due to the low mechanical effect this fraction was qualified as uncrushed straw and was used for the optimisation by response surface methodology. The raw moisture content, measured at 105 C, with infrared balance (Sartorius MA 30) was about 10.6%. As one of variables tested in this study, the moisture content was obtained by spraying distilled water on wheat straw to reach a fixed value. For moisture equilibration, the samples were maintained for 24 h in sealed bags at 4 C prior to pretreatment processing.
2.2. Experimental set-up for the thermomechanical processing The experimental set-up (Fig. 1), largely described in different previous works [18e20] is composed of three main elements: The processing stainless steel vessel (2) where the samples were placed and treated. The vacuum system which consists mainly from a vacuum tank (4) with a volume (1600 l), 130 fold greater than the processing vessel (12 l), and 4 kW vacuum pump (5) (Hibon, France). The initial vacuum pressure of the vacuum container was maintained at 5 kPa in all experiments.
3131
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Fig. 1 e Schematic of apparatus for pretreatment by D.I.C. processing 1. Boiler, 2. Pretreatment vessel, 3. Valve, 4. Vacuum tank, 5. Vacuum pump, 6. Extract container. All parts and valves of the apparatus are made in stainless steel.
A pneumatic valve (3) that separate the processing vessel from the vacuum tank. It can be opened in less than 0.2 s; this ensures a rapid decompression within the reactor.
2.3. Protocol of wheat straw thermomechanical pretreatment 50 g of Wheat straw is firstly placed in the D.I.C vessel (Figs. 1 and 2) which is maintained under a vacuum (w 5 kPa) through its connection to a vacuum container (Figs. 1e4). The vacuum allows a better diffusion of the heating fluid through the plant and consequently heat transfer between the steam and wheat straw is improved and the time to reach the desired processing pressure (or processing temperature) is shortened. After closing the electropneumatic valve (Figs. 1e3) which connects the pretreatment vessel to the vacuum container, an atmosphere of saturated steam pressure (between 300 and 700 kPa in this study) is created within the pretreatment vessel. After a processing time at fixed processing pressure, the thermal treatment
is followed by a rapid decompression resulting in a rapid drop in pressure. The equilibrium pressure, after decompression, depends on the operating pressure: the higher the operating pressure, the higher the equilibrium pressure. The vapour pressure of the bubbles created in wheat straw by autovaporisation expands the material and the extent of this expansion depends on the rheological properties of the product at the initial moisture content and temperature. The difference between the high pressure level and the vacuum determines the amount of steam generated by flash vaporisation and thus the intensity of the mechanical constraints responsible of the microalveolation phenomenon.
2.4.
Enzymes
Celluclast-1.5L, the enzyme concentrate used for cellulose hydrolysis, was a commercial Trichoderma reesei cellulase preparation contains endo-glucanases, exo-glucanases, cellobiohydrolases and b-glucosidases. This preparation was a brownish
18
g of glucose/100 g d.m
16 14 12 50-600 µm
10
600-1000 µm
8
1000-8000 µm
6 4 2 0 0
200
400
600
800
1000
1200
hydrolysis time (min) Fig. 2 e Kinetics of glucose production versus wheat straw particle size for unpretreated samples.
3132
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
35
g of glucose/100 g d.m
30 25 20
50-600 µm/15 min,7 bar 600-1000 µm/15 min,7 bar
15
1000-8000 µm/15 min,7 bar 1000-8000 µm/15 min,5 bar
10
1000-8000 µm/15 min,3 bar
5 0 0
200
400
600
800
1000
1200
Hydrolysis time (min)
Fig. 3 e Kinetics of glucose for samples pretreated by D.I.C thermomechanical process at preliminary processing pressures. The pretreatment was performed at residual moisture content and 15 min as processing time.
liquid with a density of approx. 1.20 g/ml and contained 191 mg protein/ml (calculated by the Smith assay). The cellulasic activity of concentrate was 96 FPU/ml. One unit of FPU is defined as the enzyme amount which releases 1 mmol of glucose equivalents from Whatman n 1 filter paper in 1 min. Optimum conditions of activity were between 4.5 and 6 for pH and 50 Ce60 C for temperature.
2.5.
Determination of the chemical composition of straw
Dilute acid and dilute alkali solution have been used to separate hemicellulose, lignin and cellulose from the straw. In a first time, the straw material was submitted to a reaction with 3% (v/v) H2SO4 solution at 120 C for 1 h [21]. In these conditions, only the hemicellulose was hydrolysed. The resulting solid material (cellulignin) was separated by centrifugation and the content was estimated by heating the weighed dry matter at 50 C. The supernatant (hemicellulose) was also dried, weighed and analyzed by HPLC. In a second time, the cellulignin fraction was treated with 1% (w/v) sodium hydroxide solution in a solid/liquid ratio of 1 g/10 g of cellulignin at 120 C for 90 min, conditions for an efficient lignin hydrolysis [22]. The residual solid material (cellulose pulp) was separated by centrifugation, washed with water to
remove the residual alkali, dried at 50 C and weighed. The supernatant (lignin) content was estimated by heating the weighed dry matter.
2.6.
Cellulose hydrolysis
Celluclast-1.5 L (480 FPU/L) was added to 50 mM citratephosphate buffer (pH ¼ 4.6) and then mixed to the substrate (10 g/l). The experiments were carried out in 100 ml Erlenmeyer flasks containing 10 ml total reaction volume (buffer enzyme mixture). The flasks were sealed and incubated in a rotary shaker at 600 rpm at 50 C during 20 h. To follow the hydrolysis, a flask was withdrawn at different times and the liquid phase (hydrolysate) immediately heated for 5 min on a boiling water bath to precipitate the proteins and prevent further hydrolysis. The mixture was then centrifuged at 14000 rpm for 2 min to remove solids. The cellulose hydrolysis yield of samples was determined by 3.5 dinitrosalicyclic acid method (DNS method).
2.7.
HPLC analysis
After enzymatic and acid hydrolysis of straw, the concentrations of glucose, xylose and arabinose in the hydrolysates were
Fig. 4 e Contours plots showing the simultaneous effects of processing pressure and processing time at the central point of initial moisture content of wheat straw: 25%.
3133
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
determined using a High Performance Liquid Chromatography (HPLC) system from Agilent (1100 LC and differential refractometer, Waters model 410), with a Transgenomic ICE-ORH-801 column (Ionic exclusion, 300 6.5 mm, Interchim) eluted with sulphuric acid/water (0.1% v/v) at room temperature and at a flow rate of 0.6 ml/min.
2.8.
Scanning electron microscopy
A Philips-FEI Quanta 200 ESEM/FEG Scanning Electron Microscopy operated at 20 kV, with a detector of secondary electrons Everhardt-Thornley, was used to image the control and pretreated samples. To improve the quality of the SEM images, a high vacuum was achieved.
2.9.
Experimental design
A response surface methodology was employed for optimizing the operating conditions of the D.I.C pretreatment process to give high hydrolysis yield and high initial hydrolysis rate. These two responses are assumed to be affected by three independent variables, xi (processing pressure x1, initial moisture content of wheat straw x2 and processing time x3). It is also assumed that the responses h, which was experimentally measured, defined the system. h ¼ f ðx1 ; x2 ; x3 Þ
(1)
Second degree polynomial equation were assumed to approximate the true function: h ¼ b0 þ
3 X i¼1
bi xi þ
3 X i¼1
bii x2i þ
2 3 X X
bij xi xj
(2)
i¼1 j¼iþ1
where b0, bi, bii and bij are regression coefficients, and xi are the coded variables linearly related to xi. The coding of xi into xi is expressed by the following equation: 2 xi xi (3) di where xi ¼ actual value in original units; xi ¼ mean of high and low levels of xi; and di ¼ difference between the low and high levels of xi. A central composite rotatable design [23] was used. For the three variables, the design yielded 19 experiments with eight (23) factorial points, six axial points to form a central composite design and five central points for replications. The range and the centre point were chosen after preliminary trials (Table 1). The 19 experiments were run in random order to minimize the effects of unexpected variability in observed responses due to extraneous factors. Response surfaces as represented by Fig. 4 were drawn by using the analysis design procedure of Statgraphics Plus for Windows software (5.1 version) [24,25] xi ¼
Table 1 e Coded levels for independent variables used in developing experimental data and the temperature corresponding to the different processing pressure. Coded level a
Results and discussion
3.1.
Effect of particle size on glucose production
The chemical composition of the raw material is given in Table 2. As indicated by Ballesteros et al. [26] for steam
0
1
þa
Steam pressure (kPa) 200 300 450 600 700 158.8 165 (Corresponding temperature C) 120 133.5 148 Moisture content (%) 10 16 25 34 40 Processing time (min) 3 15 32.5 50 62 ffiffiffiffi p 4 a (axial distance) ¼ N, N is the number of experiments of orthogonal design, i.e of the factorial design. In this case a ¼ 1.6818.
explosion pretreatment, the most important variables are processing time, temperature and chip size. The authors argued that when larger chips are used, heat transfer problem overcook the exterior of the lignocellulosic structure and incompletely autohydrolyse the interior. Consequently, prior to steam explosion, particle size must be reduced, which requires a significant amount of energy. In this study, the raw material was milled to three different particle size (1000e8000; 600e1000; 50e600 mm). As expected from Fig. 2, it is clear that the particle size have a strong effect on glucose production. For the largest chips, the maximum glucose production was about 10 g of glucose/100 g d m while for the smallest ones the maximum glucose production was about 15 g of glucose/ 100 g d m. According to Hendriks and Zeeman [27], the reduction in particle size leads to an increase in available specific surface and a reduction of the degree of polymerisation (DP). These factors increase the total hydrolysis yield of the lignocellulose. To avoid the energetic cost of grinding and to preserve hemicellulose fraction, the experimental design was performed on uncrushed wheat straw (1000e8000 mm). The objective being to find the optimum of glucose yield, close to that obtained with grinding. From Fig. 3, it can be seen that for uncrushed wheat straw, a yield of 17 g of glucose/100 g dm can be obtained with a pretreatment performed at 700 kPa and 15 min as processing time.
3.2.
Models fitting
The complete design matrix together with the values of both the experimental and predicted responses is given in Table 3. Central composite design was used to develop correlation between the thermomechanical pretreatment variables to
Table 2 e Chemical composition of wheat straw. Components a
3.
1
Cellulose Hemicelluloseb Ligninb Ashesc pH
Composition of wheat straw (g/100 g d m) 31 43 22 4 6.73
1 3 5 1 0.2
a determined by acid and enzymatic hydrolysis. b determined by acid hydrolysis. c ashes may include components as proteins and lipids.
3134
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Table 3 e Experimental data and yield of produced glucose with different combinations of processing pressure (x1), initial moisture content of wheat straw (x2) and processing time (x3) used in the randomized central composite design. Run
Responses Independent variables x1
1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 ea 10 þa 11 0 12 0 13 0 14 0 15 0 16 0 17 0 18 0 19 0 Mean absolute error for
x
2
x3
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 ea 0 þa 0 0 ea 0 þa 0 0 0 0 0 0 0 0 0 0 replications
Hydrolysis yield (g of glucose/100 g d m)a Experimental
Predicted
Deviation (%)
32.85 31.79 11.45 11.51 13.17 11.21 11.83 8.09 9.01 28.36 12.87 11.21 7.45 19.24 14.13 13.19 14.21 15.14 16.16 1.13%
30.42 31.19 11.62 12.06 11.18 9.61 10.99 9.08 10.35 29.03 12.57 13.53 6.23 22.48 14.49 14.49 14.49 14.49 14.49
2.43 0.60 0.17 0.55 1.99 1.60 0.84 0.99 1.34 0.67 0.30 2.32 1.22 3.24 0.36 1.30 0.28 0.65 1.67
Initial rate of hydrolysis (g/l/h) Experimental
Predicted
Deviation (%)
1.0105 1.1290 0.1297 0.2005 0.0639 0.0224 0.2173 0.2281 0.0418 0.8147 0.2431 0.2184 0.1013 0.6691 0.1074 0.1074 0.1074 0.1074 0.1074
0.1665 0.0560 0.0388 0.1091 0.1381 0.0263 0.1270 0.1375 0.0489 0.0897 0.0276 0.0684 0.2593 0.3001 0.0168 0.0170 0.0173 0.0050 0.0164
1.1770 1.1850 0.0909 0.0914 0.2020 0.0902 0.0903 0.0906 0.0907 0.7250 0.2707 0.1500 0.3606 0.3690 0.0906 0.0904 0.0901 0.1024 0.0910 0.0053 g/l/h
a 100% of the theoretical glucose production as about 32 g of glucose/100 g d m.
the fermentable sugar hydrolysis yield and initial hydrolysis rate. The hydrolysis yield was found to range between 7.45 and32.85% whereas the initial rate ranged from 0.0901 to 1.185 g/l/h. Runs 15e19 at the centre point were used to determine the experimental error. According to the sequential model, sum of squares, the models were selected based on the highest polynomials order and were not aliased. For both responses of hydrolysis yield and initial hydrolysis rate, the quadratic model was selected, as suggested by used software. The final empirical models in terms of coded factors are given by Eq. (4) where h is one of the two considered response: h ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3
the experimental and predicted hydrolysis yield from the model displayed by Table 4. In fact, from Table 3, it can be clearly seen that the deviations between predicted and experimental results are higher for initial hydrolysis rate showing that the predicted values for hydrolysis yield would be more accurate and closer to their actual values. The possible reason for the higher error of estimation given by equation corresponding to the initial hydrolysis rate could be that there were other parameters affecting this response, other than the three variables (processing pressure, initial moisture content of wheat straw and processing time) being studied in this work as opening velocity of the valve that separated the vacuum tank and the pretreatment vessel.
(4)
where x1, x2 and x3 are the processing pressure, initial moisture content of wheat straw and processing time, respectively. The coefficients bi, are displayed in Table 4. Positive sign in front of the terms indicates synergetic effect, whereas negative sign indicates antagonistic effect. The quality of the models developed was evaluated based on the correlation coefficient R2, and also the standard deviation values. The R2 for the two obtained equations were found to be 0.961 and 0.863 respectively for hydrolysis yield and initial hydrolysis rate. This indicates that 96.1% and 86.3% of the total variation in hydrolysis yield and initial hydrolysis rate were attributed to the experimental variables studied. The R2 of 0.863 was considered as moderate to validate the fit, which might be lead to large variation in initial hydrolysis rate predicted from the model displayed by Table 4. However, the R2 of 0.961 was considered relatively high as the value was close to unity, indicating that there was a good agreement between
Table 4 e Regression coefficients for the response surface and p-values. Regression Hydrolysis P-value Initial rate P-value coefficients yield of hydrolysis of the models b0 b1 b2 b3 b11 b12 b13 b22 b23 b33
14.499 11.109 0.570 9.659 3.672 1.174 9.302 1.02 0.164 0.101
e 0.0001 0.4019 0.0002 0.0038 0.2143 0.0003 0.1685 0.8466 0.8768
0.1074 0.4595 0.0146 0.3375 0.2268 0.0300 0.5171 0.0871 0.0261 0.1964
e <0.0001 0.5462 0.0003 0.0005 0.3602 0.0001 0.0172 0.4109 0.0009
3135
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Further investigations need to be carried out in order to verify this. The mean absolute error corresponding to the two studied responses are 1.13% and 0.0053 g/l, respectively for the hydrolysis yield and initial hydrolysis rate, indicating a good reproducibility of experiments. The adequacy of the two models was further justified through analysis of variance (ANOVA). The ANOVA for the quadratic models for the two responses is listed in Table 5. The Fisher’s test (F-test) carried out on experimental data make it possible to estimate the statistical significance of the proposed model. The F-test value of the models being 32.54 and 8.70 respectively for hydrolysis yield and for initial hydrolysis rate, with a low probability value ( p < 0.01), we can conclude that they were statistically significant at 99.9% confidence level. It should be noted that p-value indicates the statistical significance of each parameter. It is based on hypothesis that a parameter is not significant, thus the more this probability is close to zero, the more effect is significant. Thus, based on 95% confidence level, for the hydrolysis yield, x1, x3,x21 and x1x3were statistically significant model terms whereas x2,x22 ,x23 , x1x2 and x2x3 were all insignificant to the response. For the initial hydrolysis rate, x1, x3,x21 , x1x3 and x23 were significant model terms. From these statistical results, it was shown that the above models were adequate to predict the hydrolysis yield and the initial hydrolysis rate within the range of the studied variables.
3.3. Effects of processing variables on hydrolysis yield and initial hydrolysis rate Based on the p-values (Table 4), the processing pressure of wheat straw pretreatment (x1) showed the lowest values for either two responses of 0.0001, indicating that it had the most significant effect on the hydrolysis yield and on initial hydrolysis rate, compared to initial moisture content of wheat straw (x2) and processing time (x3). The effect of processing
time could be also considered as significant with low p-values of 0.0002 and 0.0003 respectively for hydrolysis yield and initial hydrolysis rate while initial moisture content of wheat straw was not a critical variable with a high p-value. This surprising result is attributed to the fact that the moisture contents considered in the experimental design correspond to initial values before submitting wheat straw to pretreatment. The water contents that should be taken into account are those during the pretreatment, because these values are those that really take part in physicochemical reactions. During the beginning of pretreatment the principal heating source results from the transfer of latent heat of steam condensation, the material reaches the equilibrium temperature. After this period, there is no heat transfer and the exchange is dominated by the absorption of condensed steam, resulting in an increase of moisture content. The amount of absorbed water reaches equilibrium after a certain time depending on pressure level [28]. Zarguili et al. [29] showed an increase in the moisture content during the pretreatment of starch when the pressure level and the processing time increase. Besides, the quadratic effect of processing pressurex21 , was relatively significant for the two studied responses as well. For initial hydrolysis rate the quadratic effects of processing time and in a lesser extent the quadratic effect of initial moisture content were also significant. Fig. 4 shows the three-dimensional response surfaces which was constructed to show the interaction effects of two pretreatment variables (processing pressure and processing time) on the two studied responses. For these plots, the initial moisture content of wheat straw was fixed at central level (25%). As can be seen from these figures, for low processing times (w15 min), the hydrolysis yield as well as the initial hydrolysis rate are closely constant when the processing pressure increase from 300 to 600 kPa while for a high processing time (w50 min) one can observe an increasing in the two responses: from 11 to 32 g of glucose/100 g d m for hydrolysis yield which represents almost
Table 5 e Analysis of variance for the fit of experimental data to response surface models.
Hydrolysis yield
Initial hydrolysis rate
a p-value<0.01. b p-value<0.05.
Source
Degrees of freedom
Sum of square
Mean square
F-ratio
Model Linear Quadratic Interactions Residual Lack of fit Pure error
9 3 3 3 12 5 7
966.53 741.01 49.64 175.86 39.58 34.51 5.07
107.39 247.00 16.55 58.62 3.30 6.90 0.72
32.54a 74.84a 5.02b 17.76b e 5.44 e
R2
0.96
Model Linear Quadratic Interactions Residual Lack of fit Pure error R2
9 3 3 3 12 5 7 0.86
1.9819 1.1110 0.3331 0.5379 0.3039 0.2972 0.0067
0.2202 0.3702 0.1111 0.1793 0.0253 0.0594 0.0009
8.7035a 14.6324a 4.3913 b 7.0869 b e 0.0210
3136
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
100% of the theoretical cellulose and from 0.003 to 0.98 g/l/h for initial hydrolysis rate. Thus, in the range of studied variables, a low processing time or a low processing pressure are insufficient to break down the lignocellusic structure and subsequent hydrolysis of the hemicellulose fraction which lead to a high sugar recovery. The results obtained agreed with works of Balat et al. [30] which reported that in uncatalysed steam explosion processes as it is the case in this study, hemicelluloses is thought to be hydrolysed by acetic and other acids released during steam explosion pretreatment causing an increasing in global hydrolysis. For enzymatic hydrolysis of sunflower stalks Ruiz et al. [31] also argued that the hydrolysis yield increased with the temperature (or steam pressure). In this study, the optimal values of hydrolysis yield and initial hydrolysis rate were reached for high processing pressure and high processing time.
3.4. Effect of D.I.C thermomechanical pretreatment on the composition and microstructure of wheat straw To evaluate the effect of the proposed thermomechanical pretreatment on the composition of wheat straw, processing pressure was fixed at 700 kPa and the processing time was varied from 20 to 40 min. From Table 6, it can be seen that the cellulose fraction in the pretreated wheat straw increased from 34 to 37 g of glucose/100 g d m when the processing time increased from 20 to 40 min. These results are in agreement with those of Kabel et al. [32] which indicated that the cellulose level in the solid residues after pretreatment increased with the pretreatment severity while in the same time the xylan fraction decreased. From Table 6, we can see that the hemicellulosic fractions remained closely constant and this is in agreement with the review of Mosier et al. [6] who argued that an effective pretreatment preserves the pentose fractions. Moreover, the lignin fraction decreased from 22% in raw material to 16 and 17% for the pretreated samples. In a study on hydrothermal pretreatment of wheat straw through IBUS process, Petersen et al. [33] pointed out some degradation and a possible dissolution of lignin fraction. Another important parameter is the acidity of the system. According to Ruiz et al. [31] who studied the steam explosion
of sunflower stalks, at higher pretreatment temperature lower final pH values were obtained, ranging from 4.1 to 5.2 for a severity factor R0 in the range 3.1e4.5. As described by Overend and Chornet [34]: R0 ¼ t exp ðT 100=14:75Þ; where t is the residence time in minutes and T is the pretreatment temperature in C. Another study [35] reported on pH of extracts obtained after steam explosion of sugarcane bagasse which varied in the range 3.2e4.7, for experiments with severity factor between 3.7 and 4.3 while in our study the severity factor for the experiments reported in Table 6 was 3.21 and 3.51 for the samples pretreated for 20 min and 40 min for which almost a neutral values of pH were obtained: 5.92 and 5.66, respectively. According to Sun [36] a low pH values (high acidity) is an indication of sugar degradation and a formation of inhibitor components which is a negative aspect for further fermentation and resulting in a lower yield in ethanol. The author indicated that in a number of pretreatments of lignocellulosic materials although disrupting the plant cell walls and improves enzymatic access to the cellulose, the by-products as acetic acid, formic acid or furfural and its derivatives inhibit the in-vitro activity of several enzymes. In our study a weak decreasing of acidity from 5.92 to 5.66 may indicate a low sugar degradation and formation of inhibitors. The lignocellulosic complex is made up of matrix of cellulose and lignin bound by hemicellulose chains. During the pretreatment, this matrix should be broken in order to reduce the crystallinity of cellulose and increase the fraction of amorphous phase, the most suitable form for enzymatic attack. The cellulose hydrolysis is also affected by the porosity, generally expressed by the accessible surface area in the lignocellulosic materials [36]. The use of SEM permits a successful visualisation of morphological and structural changes of wheat straw cellulose fibres after the proposed thermomechanical pretreatment. These structural changes can be observed by comparing Fig. 5a and b. The SEM image of raw wheat straw prior to pretreatment shows a large part of relatively compact zones, arranged tightly, with a mean circumference of ellipsoidal alveoles of about 58.2 mm. After the thermomechanical pretreatment, the structure is more expanded with a large alveoles, some defibrillation and dispersion is visible in the microstructure with a high mean circumference alveoles of about 92 mm.
Table 6 e Chemical composition of wheat straw for two thermomechanical pretreated samples at 700 kPa and two processing times: 20 and 40 min. Composition of wheat straw (g/100 g d m) 700 kPa, 20 min 700 kPa, 40 min R0 (severity factor) Cellulosea Hemicelluloseb Ligninb pH Hydrolysis yield (%) Initial hydrolysis rate (g/l/h)
3.21 34 49 17 5.92 0.04 19.2 0.385
a determined by acid and enzymatic hydrolysis. b determined by acid hydrolysis.
3.51 37 47 16 5.66 0.02 32.1 0.769
Fig. 5 e Microstructure of wheat straw (1000e8000 mm) unpretreated (a) and pretreated (b) during 15 min at 700 kPa (magnification X 1000).
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
3.5.
Conclusion
From this work on a new pretreatment method to produce fermentable sugars from wheat straw an encouraging results were obtained. We concluded that hydrolysis yield and initial hydrolysis rate of wheat straw after pretreatment by the proposed thermomechanical pretreatment were more influenced by processing pressure and processing time than by initial moisture content. The significant variables exerted a positive linear effect on hydrolysis yield and initial rate of hydrolysis. For the two responses, the empirical models gave R2 of 96 and 86% and p-values less than 0.01, which implied a good agreement between the predicted and actual values. The optimal conditions to obtain the highest yield are showed in Fig. 4: 600 kPa for the processing pressure and 50 min for processing time. Initial moisture content being not a critical variable, we consider that the optimal value is its residual value (w10%). In this optimisation the vacuum degree was fixed at 50 kPa. This factor will be investigated in a next research. The second part of this study showed that, at residual moisture content, a high hydrolysis yield and initial hydrolysis rate can be achieved. 700 kPa, which can be considered as a moderate processing pressure, compared to the literature, and 40 min as processing time, are sufficient to hydrolyse 100% of the theoretical cellulose present in raw material. In these conditions, the predicted value of hydrolysis yield (33 g of glucose/100 g d m) is close to that observed (32.1 g of glucose/100 g d m). In addition, as shown by Fig. 5, the proposed pretreatment enhance the accessibility of enzymes in the substrate due to the flash expansion provoked by the sudden release of steam pressure. Using this pretreatment method, a reproducible hydroloysis yield can be achieved without adding any solvent or acid, representing economic and ecologic advantages.
Acknowledgements The authors acknowledge the financial support of Region Poitou-Charentes and OSEO Innovation through A0809006T project.
references
[1] Silverstein RA, Chen Y, Sharma-Shivappa RR, Boyette MD, Osborne J. A comparison of chemical pre-treatment methods for improving saccharification of cotton stalks. Bioresour Technol 2007;98:3000e11. [2] Solomon BD, Barnes JR, Kathleen E, Halvorsen KE. Grain and cellulosic ethanol: history, economics, and energy policy. Biomass Bioenerg 2007;31:416e25. ´ J. Fuel ethanol production: process [3] Cardona CA, Sa´nchez O design trends and integration opportunities. Bioresour Technol 2007;98:2415e57. [4] Tabka MG, Herpoe¨l-Gimbert I, Monod F, Asther M, Sigoillot JC. Enzymatic saccharification of wheat straw for bioethanol production by a combined cellulose xylanase and ferulolyl esterase treatment. Enzyme Microb Tech 2006;39: 897e902.
3137
[5] Cara C, Ruiz E, Ballesteros M, Manzanares P, Jose´ Negro M. Castro. Production of fuel ethanol from steam-explosion pretreated olive tree pruning. Fuel 2008;87:692e700. [6] Mosier N, Wyman C, Dale B, Elander R, Lee YY, Holtzapple M, et al. Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresour Technol 2005;96:673e86. [7] Hu G, Heitmann JA, Rojas OJ. Feedstock pre-treatment strategies for producing ethanol from wood, bark and forest residues. BioResources 2008;3:270e94. [8] Ogier J-C, Ballerini D, Leygue J-P, Rigal L, Pourquie´ J. Production d’e´thanol a` partir de biomasse lignocellulosique. Oil Gas Sci Technol 1999;54:67e94. [9] Sassner P, Ma˚rtensson C-G, Galbe M, Zacchi G. Steam pretreatment of H2SO4-impregnated Salix for the production of bioethanol. Bioresour Technol 2008;99:137e45. [10] Ballerini D, Alazard-Toux N. Les Biocarburants: e´tat des Lieux, Perspectives et Enjeux du De´veloppement. France: Technip ed, ISBN 2710808692; 2006. 348 pages. [11] Rodriguez G, Rodriguez R, Jime´nez A, Guille´n R, FernandezBolan˜os J. Effect of steam treatment of alperujo on the composition, enzymatic saccharification, and in vitro digestibility of alperujo. J. Agric. Food. Chem 2007;55:136e42. [12] Balan V, Dale B, Bals B. Separation of proteins from grasses integrated with ammonia fiber explosion (AFEX) pretreatment and cellulose hydrolysis. US Patent WO/2008/ 020901. [13] Zhang L, Li D, Wan L, Wang T, Zhang L, Chen XD, et al. Effect of steam explosion on biodegradation of lignin in wheat straw. Bioresour Technol 2008;99:8512e5. [14] Viola E, Cardinale M, Santarcangelo R, Villone A, Zimbardi F. Ethanol from eel grass via steam explosion and enzymatic hydrolysis. Biomass Bioenerg 2008;32:613e8. [15] Talebnia F, Karakashev D, Angelidaki I. Production of bioethanol from wheat straw: an overview on pretreatment, hydrolysis and fermentation. Bioresour Technol 2010;101: 4744e53. [16] Karimi K, Kheradmandinia S, Taherzadeh MJ. Conversion of rice straw to sugars by dilute-acid hydrolysis. Biomass Bioenerg 2006;30:247e53. [17] Rezzoug S-A, Maache-Rezzoug Z, Sannier F, Allaf K. A thermomechanical preprocessing for pectin isolation from orange peel. Optimisation by response surface methodology. Int J Food Eng 2008;4(1). Article 10. [18] Rezzoug S-A, Maache-Rezzoug Z, Mazoyer J, Jeannin M, Allaf K. Effect of instantaneous controlled pressure drop proccess on hydration capacity of scleroglucan. Optimisation of operating conditions by response surface methodology. Carbohydr Polym 2000;42:73e84. [19] Nouviaire A, Lancien R, Maache-Rezzoug Z. Influence of hydrothermal treatment on rheological and cooking characteristics of fresh egg pasta. J Cereal Sci 2008;47:283e91. [20] Maache-Rezzoug Z, Zarguili I, Loisel C, Queveau D, Bule´on A. Structural modifications and thermal transitions of standard maize starch after D.I.C. hydrothermal treatment. Carbohydr Polym 2008;74:802e12. [21] Wen Z, Liao W, Chen S. Hydrolysis of animal manure lignocellulosics for reducing sugar production. Bioresour Technol 2004;91:31e9. [22] Mussatto SI, Roberto IC. Acid hydrolysis and fermentation of brewer’s spent grain to produce xylitol. J Sci Food Agric 2005; 85:2453e60. [23] Benoist D, Tourbier Y, Germain-Tourbier S. Plan d’expe´riences: Construction et Analyse. Paris: Lavoisier Tec & Doc; 1994. Chap 5, pp. 208e389. [24] Statgraphics Plus for Windows 5.1. Experimental design. Manugistics, Inc.; 1994. [25] Statgraphics Plus for Windows 5.1. User Manual. Manugistics, Inc.; 1994.
3138
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
[26] Ballesteros I, Oliva JM, Navarro AA, Gonzalez A, Carrasco J, Ballesteros M. Effect of chip size on steam explosion pretreatment of softwood. Appl Biochem Biotech 2000;84-86: 99e110. [27] Hendriks ATWM, Zeeman G. Pretreatments to enhance the digestibility of lignocellulosic biomass. Bioresour Technol 2009;100:10e8. [28] Bahrani SA, Loisel C, Monteau J-Y, Rezzoug S-A, MaacheRezzoug Z. Estimation of effective moisture diffusivity in starchy materials following hydrothermal treatments. To be published in “Defect and Diffusion Forum”. ISSN 1021e0386 (2011). [29] Zarguili I, Maache-Rezzoug Z, Loisel C, Doublier J-L. A mathematical model to describe the change of moisture distribution in maize starch during DIC hydrothermal treatment. Int J Food Sci Tech 2009;44:10e7. [30] Balat M, Balat H, Oz C. Progress in bioethanol processing. Prog Energ Comb Sci 2008;34:551e73. [31] Ruiz E, Cara C, Manzanares P, Ballesteros M, Castro E. Evaluation of steam explosion pretreatment for enzymatic
[32]
[33]
[34]
[35]
[36]
hydrolysis of sunflower stalks. Enzyme Microb Technol 2008; 42:160e6. Kabel MA, Bos G, Zeevalking J, Voragen AGJ, Schols HA. Effect of pretretment severity on xylan solubility and enzymatic breakdown of the remaining cellulose from wheat straw. Bioresour Technol 2007;98:2034e42. Petersen MØ, Larsen J, Thomsen MH. Optimization of hydrothermal pretretment of wheat straw for production of bioethanol at low water consumptioin without adition of chemicals. BiomassBioenerg 2009;33:834e40. Overend RP, Chornet E. Fractionation of lignocellulosics by steam aqueous pretreatments. Philos Trans R Soc Lond A 1987;96(5):862e70. Kaar WE, Gutierrez CV, Kinishita CM. Steam explosioin of sugarcane bagasse as pretreatment for conversion to ethanol. Biomass Bioenerg 1998;14:277e87. Sun RC. Detoxification and separation of lignocellulosic biomass prior to fermentation for bioethanol production by removal of lignin and hemicellulose. BioResources 2008;4(2):452e5.
Available at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Optimizing thermomechanical pretreatment conditions to enhance enzymatic hydrolysis of wheat straw by response surface methodology Z. Maache-Rezzoug a, G. Pierre b, A. Nouviaire a, T. Maugard b, S.A. Rezzoug a,* a
LEPTIAB, Universite´ de La Rochelle - Poˆles Sciences et Technologie, Baˆtiment Marie Curie, Avenue Michel Cre´peau, 17042 La Rochelle, France UMR 6250 CNRS-ULR, LIENSs, Universite´ de La Rochelle, Poˆles Sciences et Technologie, Baˆtiment Marie Curie, Avenue Michel Cre´peau, 17042 La Rochelle, France b
article info
abstract
Article history:
Wheat straw was pretreated with a thermomechanical process developed in our laboratory
Received 1 March 2010
to improve the enzymatic hydrolysis extent of potentially fermentable sugars. This process
Received in revised form
involves subjecting the lignocellulosic biomass for a short time to saturated steam pres-
5 April 2011
sure, followed by an instantaneous decompression to vacuum at 5 kPa. Increasing of the
Accepted 15 April 2011
heat induced by saturated steam result in intensive vapour formation in the capillary
Available online 11 May 2011
porous structure of the plant material and the subsequent release of the pressure to vacuum allows fixing the expanded structure. Response surface methodology (RSM) based
Keywords:
on central composite design was used to optimize three independent variables of the
Thermomechanical pretreatment
pretreatment process: processing pressure (300e700 kPa), initial moisture contents of
Wheat straw
wheat straw (10e40%) and processing time (3e62 min). The process was optimised for
Enzymatic hydrolysis
hydrolysis yield and initial hydrolysis rate obtained by enzymatic hydrolysis on the pre-
Response surface methodology
treated solids by Celluclast (1.5 L). The analysis of variance (ANOVA) revealed that, among
Hydrolysis yield
the process variables, processing pressure and processing time have the most significant
Initial hydrolysis rate
effect on the hydrolysis yield and on initial rate of hydrolysis whereas initial moisture content observed significantly lower effect on the two responses. The predicted hydrolysis yield and in a lesser extent the predicted initial rate of hydrolysis agreed satisfactorily with the experimental values with R2 of 96% and 86% respectively. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Currently, corn is the primary raw material for ethanol production in the United States [1] and sugarcane in Brazil [2]. However, the corn and sugarcane to ethanol industry draws its feedstock from food and is quite mature with little possibility of process improvement. Biofuels from food sources are known as first-generation biofuels. Lignocellulosic biomass is an interesting and necessary enlargement of the biomass used for the production of renewable biofuels. It can be used as
a potential substrate to produce ethanol which is considered as one of the most important renewable fuels contributing to the reduction of negative environmental impacts generated by the worldwide utilisation of fossil fuels [3]. These oils are known as second-generation biofuels. Lignocellulosic materials to be considered for ethanol production include wood, crops from annual plants, agricultural residue and waste paper [4]. However, there are a lot of challenges and obstacles such as cost, technology and environmental issues that need to be overcome. Considerable research efforts have been made to
* Corresponding author. Tel.: þ33 5 46 45 86 15; fax: þ33 5 46 45 82 41. E-mail address: [email protected] (S.A. Rezzoug). 0961-9534/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.biombioe.2011.04.012
3130
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
improve conversion yields of lignocellulosic materials by the insertion of pretreatment step before to enzymatic hydrolysis. The purposes of pretreatment is increasing of porosity as well as removing the lignin to make lignocellulosic materials more accessible to enzymes by destroying the cell structure and break down the various physical and chemical barriers [5,6]. These pretreatments can be physical, physicochemical, chemical or biological [7] and can be combined between them, with the objective to obtain a high yield from enzymatic hydrolysis, to generate small quantities of co-products and inhibitors of fermentation while reducing costs. The purely physical processes of pretreatment can be summarized by an intense mechanical grinding in order to increase the accessible surfaces to enzymes, thus supporting the further hydrolysis and reducing the polymerisation degree of cellulose. This kind of pretreatment can be interesting if the granulometry is very low, which limits its interest and its use due to the energy cost of grinding ( < 2 mm), [8]. Steam explosion pretreatment consists to expose the biomass up to high pressure (1500e5000 kPa) and temperature (180e250 C) in presence of steam during a determined time, up to 90 min [9], followed by a rapid reduction in pressure, in order to breakdown the lignocellulosic structure. This technology was implemented at industrial scale with batch processes (Stake and Iotech) [10]. The treatment leads to a partial self-hydrolysis of hemicelluloses, depolymerisation of lignin and a destructuration of cellulose, largely dependent on treatment temperature. However, the yields of enzymatic hydrolysis are about 50% and some inhibiting compounds of the fermentation appeared following the partial pyrolysis of cellulose during the steam explosion. Steam explosion is often combined to acidification with H2SO4 or SO2 [1,11], to improve the yield of hydrolysis (90% of potential glucose and nearly 100% of sugars resulting from hemicelluloses) at moderate temperatures (150e200 C against 250 C in absence of catalyst) during 15e20 min, thus minimizing the formation of sugar degradation products. According to the type of biomass and for the same processing pretreatment conditions, steam explosion with acid catalysis (H2SO4) exhibited better results in terms of final sugar yield than other techniques in which catalysts as NH3 are used such as AFEX process (Ammonia Fiber EXplosion) developed in USA [12]. Among other advantages of steam explosion, it was noted [10] a low energy consumption, small quantities of generated undesirable products, a simplicity of implementation in batch, low quantities of chemical reagents and a good adaptation to the big particles size such as shavings. Steam explosion is thus considered to be one of the most promising methods to make biomass more accessible to enzymes. In order to obtain higher yields of enzymatic hydrolysis than those described in the literature and limiting the production of inhibitors of alcoholic fermentation, a thermomechanical pretreatment termed ’’D.I.C.’’ process (in french: De´tente Instantane´e Controˆle´e) was performed on wheat straw. This physical process is close to steam explosion technology [13,14]. The first difference is the moderate processing pressure for the proposed pretreatment. 0.7 MPa in this study compared to at least 5 MPa for steam explosion, as reported in the literature [15,16]. The difference is that the D.I.C treatment comprises two additional steps: instauration of initial vacuum before injection of the saturated steam and explosion towards vacuum.
The first step allows to reduce the resistance of air and thus to facilitate the diffusion of steam into the product. Consequently, the time necessary to reach the steam equilibrium temperature is reduced. The second step consists to an abrupt decompression which carries out towards vacuum (5 kPa) instead of atmospheric pressure like it is the case with the steam explosion process [17]. Due to the instantaneous character of this transformation as well as the adiabatic nature of the transition of steam inside the product, the water vaporisation induces a fast cooling inhibiting the possible further thermal reactions and thus contributing to reduction of undesired degrading products. The temperature is quickly stabilized at a balance temperature of the considered final pressure, limiting the reactions of degradations. This study aims to assess the effect three processing parameters of D.I.C pretreatment: processing steam pressure, initial moisture content of wheat straw and processing time on two responses: the hydrolysis yield and initial rate of hydrolysis, through response surface methodology (RSM). RSM is defined as a statistical method using quantitative data from an appropriate experimental design to determine and simultaneously solve multivariate equations. The main advantage of RSM is the reduced number of experimental trials needed to evaluate multiple parameters and their interactions.
2.
Materials and methods
2.1.
Plant material
Wheat straw used in this study grown in Charente-Maritime region in France and was ground in Gindomix (GM 200) Retsch crusher at 7500 RPM during 50 s and then sieved to obtain two particle sizes: 600e1000 mm and 50e600 mm. These two sizes were firstly studied to determine the effect of particle size. The third size was obtained by calibrating in sieve to obtain a fraction ranging between 1000 and 8000 mm. Due to the low mechanical effect this fraction was qualified as uncrushed straw and was used for the optimisation by response surface methodology. The raw moisture content, measured at 105 C, with infrared balance (Sartorius MA 30) was about 10.6%. As one of variables tested in this study, the moisture content was obtained by spraying distilled water on wheat straw to reach a fixed value. For moisture equilibration, the samples were maintained for 24 h in sealed bags at 4 C prior to pretreatment processing.
2.2. Experimental set-up for the thermomechanical processing The experimental set-up (Fig. 1), largely described in different previous works [18e20] is composed of three main elements: The processing stainless steel vessel (2) where the samples were placed and treated. The vacuum system which consists mainly from a vacuum tank (4) with a volume (1600 l), 130 fold greater than the processing vessel (12 l), and 4 kW vacuum pump (5) (Hibon, France). The initial vacuum pressure of the vacuum container was maintained at 5 kPa in all experiments.
3131
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Fig. 1 e Schematic of apparatus for pretreatment by D.I.C. processing 1. Boiler, 2. Pretreatment vessel, 3. Valve, 4. Vacuum tank, 5. Vacuum pump, 6. Extract container. All parts and valves of the apparatus are made in stainless steel.
A pneumatic valve (3) that separate the processing vessel from the vacuum tank. It can be opened in less than 0.2 s; this ensures a rapid decompression within the reactor.
2.3. Protocol of wheat straw thermomechanical pretreatment 50 g of Wheat straw is firstly placed in the D.I.C vessel (Figs. 1 and 2) which is maintained under a vacuum (w 5 kPa) through its connection to a vacuum container (Figs. 1e4). The vacuum allows a better diffusion of the heating fluid through the plant and consequently heat transfer between the steam and wheat straw is improved and the time to reach the desired processing pressure (or processing temperature) is shortened. After closing the electropneumatic valve (Figs. 1e3) which connects the pretreatment vessel to the vacuum container, an atmosphere of saturated steam pressure (between 300 and 700 kPa in this study) is created within the pretreatment vessel. After a processing time at fixed processing pressure, the thermal treatment
is followed by a rapid decompression resulting in a rapid drop in pressure. The equilibrium pressure, after decompression, depends on the operating pressure: the higher the operating pressure, the higher the equilibrium pressure. The vapour pressure of the bubbles created in wheat straw by autovaporisation expands the material and the extent of this expansion depends on the rheological properties of the product at the initial moisture content and temperature. The difference between the high pressure level and the vacuum determines the amount of steam generated by flash vaporisation and thus the intensity of the mechanical constraints responsible of the microalveolation phenomenon.
2.4.
Enzymes
Celluclast-1.5L, the enzyme concentrate used for cellulose hydrolysis, was a commercial Trichoderma reesei cellulase preparation contains endo-glucanases, exo-glucanases, cellobiohydrolases and b-glucosidases. This preparation was a brownish
18
g of glucose/100 g d.m
16 14 12 50-600 µm
10
600-1000 µm
8
1000-8000 µm
6 4 2 0 0
200
400
600
800
1000
1200
hydrolysis time (min) Fig. 2 e Kinetics of glucose production versus wheat straw particle size for unpretreated samples.
3132
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
35
g of glucose/100 g d.m
30 25 20
50-600 µm/15 min,7 bar 600-1000 µm/15 min,7 bar
15
1000-8000 µm/15 min,7 bar 1000-8000 µm/15 min,5 bar
10
1000-8000 µm/15 min,3 bar
5 0 0
200
400
600
800
1000
1200
Hydrolysis time (min)
Fig. 3 e Kinetics of glucose for samples pretreated by D.I.C thermomechanical process at preliminary processing pressures. The pretreatment was performed at residual moisture content and 15 min as processing time.
liquid with a density of approx. 1.20 g/ml and contained 191 mg protein/ml (calculated by the Smith assay). The cellulasic activity of concentrate was 96 FPU/ml. One unit of FPU is defined as the enzyme amount which releases 1 mmol of glucose equivalents from Whatman n 1 filter paper in 1 min. Optimum conditions of activity were between 4.5 and 6 for pH and 50 Ce60 C for temperature.
2.5.
Determination of the chemical composition of straw
Dilute acid and dilute alkali solution have been used to separate hemicellulose, lignin and cellulose from the straw. In a first time, the straw material was submitted to a reaction with 3% (v/v) H2SO4 solution at 120 C for 1 h [21]. In these conditions, only the hemicellulose was hydrolysed. The resulting solid material (cellulignin) was separated by centrifugation and the content was estimated by heating the weighed dry matter at 50 C. The supernatant (hemicellulose) was also dried, weighed and analyzed by HPLC. In a second time, the cellulignin fraction was treated with 1% (w/v) sodium hydroxide solution in a solid/liquid ratio of 1 g/10 g of cellulignin at 120 C for 90 min, conditions for an efficient lignin hydrolysis [22]. The residual solid material (cellulose pulp) was separated by centrifugation, washed with water to
remove the residual alkali, dried at 50 C and weighed. The supernatant (lignin) content was estimated by heating the weighed dry matter.
2.6.
Cellulose hydrolysis
Celluclast-1.5 L (480 FPU/L) was added to 50 mM citratephosphate buffer (pH ¼ 4.6) and then mixed to the substrate (10 g/l). The experiments were carried out in 100 ml Erlenmeyer flasks containing 10 ml total reaction volume (buffer enzyme mixture). The flasks were sealed and incubated in a rotary shaker at 600 rpm at 50 C during 20 h. To follow the hydrolysis, a flask was withdrawn at different times and the liquid phase (hydrolysate) immediately heated for 5 min on a boiling water bath to precipitate the proteins and prevent further hydrolysis. The mixture was then centrifuged at 14000 rpm for 2 min to remove solids. The cellulose hydrolysis yield of samples was determined by 3.5 dinitrosalicyclic acid method (DNS method).
2.7.
HPLC analysis
After enzymatic and acid hydrolysis of straw, the concentrations of glucose, xylose and arabinose in the hydrolysates were
Fig. 4 e Contours plots showing the simultaneous effects of processing pressure and processing time at the central point of initial moisture content of wheat straw: 25%.
3133
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
determined using a High Performance Liquid Chromatography (HPLC) system from Agilent (1100 LC and differential refractometer, Waters model 410), with a Transgenomic ICE-ORH-801 column (Ionic exclusion, 300 6.5 mm, Interchim) eluted with sulphuric acid/water (0.1% v/v) at room temperature and at a flow rate of 0.6 ml/min.
2.8.
Scanning electron microscopy
A Philips-FEI Quanta 200 ESEM/FEG Scanning Electron Microscopy operated at 20 kV, with a detector of secondary electrons Everhardt-Thornley, was used to image the control and pretreated samples. To improve the quality of the SEM images, a high vacuum was achieved.
2.9.
Experimental design
A response surface methodology was employed for optimizing the operating conditions of the D.I.C pretreatment process to give high hydrolysis yield and high initial hydrolysis rate. These two responses are assumed to be affected by three independent variables, xi (processing pressure x1, initial moisture content of wheat straw x2 and processing time x3). It is also assumed that the responses h, which was experimentally measured, defined the system. h ¼ f ðx1 ; x2 ; x3 Þ
(1)
Second degree polynomial equation were assumed to approximate the true function: h ¼ b0 þ
3 X i¼1
bi xi þ
3 X i¼1
bii x2i þ
2 3 X X
bij xi xj
(2)
i¼1 j¼iþ1
where b0, bi, bii and bij are regression coefficients, and xi are the coded variables linearly related to xi. The coding of xi into xi is expressed by the following equation: 2 xi xi (3) di where xi ¼ actual value in original units; xi ¼ mean of high and low levels of xi; and di ¼ difference between the low and high levels of xi. A central composite rotatable design [23] was used. For the three variables, the design yielded 19 experiments with eight (23) factorial points, six axial points to form a central composite design and five central points for replications. The range and the centre point were chosen after preliminary trials (Table 1). The 19 experiments were run in random order to minimize the effects of unexpected variability in observed responses due to extraneous factors. Response surfaces as represented by Fig. 4 were drawn by using the analysis design procedure of Statgraphics Plus for Windows software (5.1 version) [24,25] xi ¼
Table 1 e Coded levels for independent variables used in developing experimental data and the temperature corresponding to the different processing pressure. Coded level a
Results and discussion
3.1.
Effect of particle size on glucose production
The chemical composition of the raw material is given in Table 2. As indicated by Ballesteros et al. [26] for steam
0
1
þa
Steam pressure (kPa) 200 300 450 600 700 158.8 165 (Corresponding temperature C) 120 133.5 148 Moisture content (%) 10 16 25 34 40 Processing time (min) 3 15 32.5 50 62 ffiffiffiffi p 4 a (axial distance) ¼ N, N is the number of experiments of orthogonal design, i.e of the factorial design. In this case a ¼ 1.6818.
explosion pretreatment, the most important variables are processing time, temperature and chip size. The authors argued that when larger chips are used, heat transfer problem overcook the exterior of the lignocellulosic structure and incompletely autohydrolyse the interior. Consequently, prior to steam explosion, particle size must be reduced, which requires a significant amount of energy. In this study, the raw material was milled to three different particle size (1000e8000; 600e1000; 50e600 mm). As expected from Fig. 2, it is clear that the particle size have a strong effect on glucose production. For the largest chips, the maximum glucose production was about 10 g of glucose/100 g d m while for the smallest ones the maximum glucose production was about 15 g of glucose/ 100 g d m. According to Hendriks and Zeeman [27], the reduction in particle size leads to an increase in available specific surface and a reduction of the degree of polymerisation (DP). These factors increase the total hydrolysis yield of the lignocellulose. To avoid the energetic cost of grinding and to preserve hemicellulose fraction, the experimental design was performed on uncrushed wheat straw (1000e8000 mm). The objective being to find the optimum of glucose yield, close to that obtained with grinding. From Fig. 3, it can be seen that for uncrushed wheat straw, a yield of 17 g of glucose/100 g dm can be obtained with a pretreatment performed at 700 kPa and 15 min as processing time.
3.2.
Models fitting
The complete design matrix together with the values of both the experimental and predicted responses is given in Table 3. Central composite design was used to develop correlation between the thermomechanical pretreatment variables to
Table 2 e Chemical composition of wheat straw. Components a
3.
1
Cellulose Hemicelluloseb Ligninb Ashesc pH
Composition of wheat straw (g/100 g d m) 31 43 22 4 6.73
1 3 5 1 0.2
a determined by acid and enzymatic hydrolysis. b determined by acid hydrolysis. c ashes may include components as proteins and lipids.
3134
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Table 3 e Experimental data and yield of produced glucose with different combinations of processing pressure (x1), initial moisture content of wheat straw (x2) and processing time (x3) used in the randomized central composite design. Run
Responses Independent variables x1
1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 ea 10 þa 11 0 12 0 13 0 14 0 15 0 16 0 17 0 18 0 19 0 Mean absolute error for
x
2
x3
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 ea 0 þa 0 0 ea 0 þa 0 0 0 0 0 0 0 0 0 0 replications
Hydrolysis yield (g of glucose/100 g d m)a Experimental
Predicted
Deviation (%)
32.85 31.79 11.45 11.51 13.17 11.21 11.83 8.09 9.01 28.36 12.87 11.21 7.45 19.24 14.13 13.19 14.21 15.14 16.16 1.13%
30.42 31.19 11.62 12.06 11.18 9.61 10.99 9.08 10.35 29.03 12.57 13.53 6.23 22.48 14.49 14.49 14.49 14.49 14.49
2.43 0.60 0.17 0.55 1.99 1.60 0.84 0.99 1.34 0.67 0.30 2.32 1.22 3.24 0.36 1.30 0.28 0.65 1.67
Initial rate of hydrolysis (g/l/h) Experimental
Predicted
Deviation (%)
1.0105 1.1290 0.1297 0.2005 0.0639 0.0224 0.2173 0.2281 0.0418 0.8147 0.2431 0.2184 0.1013 0.6691 0.1074 0.1074 0.1074 0.1074 0.1074
0.1665 0.0560 0.0388 0.1091 0.1381 0.0263 0.1270 0.1375 0.0489 0.0897 0.0276 0.0684 0.2593 0.3001 0.0168 0.0170 0.0173 0.0050 0.0164
1.1770 1.1850 0.0909 0.0914 0.2020 0.0902 0.0903 0.0906 0.0907 0.7250 0.2707 0.1500 0.3606 0.3690 0.0906 0.0904 0.0901 0.1024 0.0910 0.0053 g/l/h
a 100% of the theoretical glucose production as about 32 g of glucose/100 g d m.
the fermentable sugar hydrolysis yield and initial hydrolysis rate. The hydrolysis yield was found to range between 7.45 and32.85% whereas the initial rate ranged from 0.0901 to 1.185 g/l/h. Runs 15e19 at the centre point were used to determine the experimental error. According to the sequential model, sum of squares, the models were selected based on the highest polynomials order and were not aliased. For both responses of hydrolysis yield and initial hydrolysis rate, the quadratic model was selected, as suggested by used software. The final empirical models in terms of coded factors are given by Eq. (4) where h is one of the two considered response: h ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3
the experimental and predicted hydrolysis yield from the model displayed by Table 4. In fact, from Table 3, it can be clearly seen that the deviations between predicted and experimental results are higher for initial hydrolysis rate showing that the predicted values for hydrolysis yield would be more accurate and closer to their actual values. The possible reason for the higher error of estimation given by equation corresponding to the initial hydrolysis rate could be that there were other parameters affecting this response, other than the three variables (processing pressure, initial moisture content of wheat straw and processing time) being studied in this work as opening velocity of the valve that separated the vacuum tank and the pretreatment vessel.
(4)
where x1, x2 and x3 are the processing pressure, initial moisture content of wheat straw and processing time, respectively. The coefficients bi, are displayed in Table 4. Positive sign in front of the terms indicates synergetic effect, whereas negative sign indicates antagonistic effect. The quality of the models developed was evaluated based on the correlation coefficient R2, and also the standard deviation values. The R2 for the two obtained equations were found to be 0.961 and 0.863 respectively for hydrolysis yield and initial hydrolysis rate. This indicates that 96.1% and 86.3% of the total variation in hydrolysis yield and initial hydrolysis rate were attributed to the experimental variables studied. The R2 of 0.863 was considered as moderate to validate the fit, which might be lead to large variation in initial hydrolysis rate predicted from the model displayed by Table 4. However, the R2 of 0.961 was considered relatively high as the value was close to unity, indicating that there was a good agreement between
Table 4 e Regression coefficients for the response surface and p-values. Regression Hydrolysis P-value Initial rate P-value coefficients yield of hydrolysis of the models b0 b1 b2 b3 b11 b12 b13 b22 b23 b33
14.499 11.109 0.570 9.659 3.672 1.174 9.302 1.02 0.164 0.101
e 0.0001 0.4019 0.0002 0.0038 0.2143 0.0003 0.1685 0.8466 0.8768
0.1074 0.4595 0.0146 0.3375 0.2268 0.0300 0.5171 0.0871 0.0261 0.1964
e <0.0001 0.5462 0.0003 0.0005 0.3602 0.0001 0.0172 0.4109 0.0009
3135
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
Further investigations need to be carried out in order to verify this. The mean absolute error corresponding to the two studied responses are 1.13% and 0.0053 g/l, respectively for the hydrolysis yield and initial hydrolysis rate, indicating a good reproducibility of experiments. The adequacy of the two models was further justified through analysis of variance (ANOVA). The ANOVA for the quadratic models for the two responses is listed in Table 5. The Fisher’s test (F-test) carried out on experimental data make it possible to estimate the statistical significance of the proposed model. The F-test value of the models being 32.54 and 8.70 respectively for hydrolysis yield and for initial hydrolysis rate, with a low probability value ( p < 0.01), we can conclude that they were statistically significant at 99.9% confidence level. It should be noted that p-value indicates the statistical significance of each parameter. It is based on hypothesis that a parameter is not significant, thus the more this probability is close to zero, the more effect is significant. Thus, based on 95% confidence level, for the hydrolysis yield, x1, x3,x21 and x1x3were statistically significant model terms whereas x2,x22 ,x23 , x1x2 and x2x3 were all insignificant to the response. For the initial hydrolysis rate, x1, x3,x21 , x1x3 and x23 were significant model terms. From these statistical results, it was shown that the above models were adequate to predict the hydrolysis yield and the initial hydrolysis rate within the range of the studied variables.
3.3. Effects of processing variables on hydrolysis yield and initial hydrolysis rate Based on the p-values (Table 4), the processing pressure of wheat straw pretreatment (x1) showed the lowest values for either two responses of 0.0001, indicating that it had the most significant effect on the hydrolysis yield and on initial hydrolysis rate, compared to initial moisture content of wheat straw (x2) and processing time (x3). The effect of processing
time could be also considered as significant with low p-values of 0.0002 and 0.0003 respectively for hydrolysis yield and initial hydrolysis rate while initial moisture content of wheat straw was not a critical variable with a high p-value. This surprising result is attributed to the fact that the moisture contents considered in the experimental design correspond to initial values before submitting wheat straw to pretreatment. The water contents that should be taken into account are those during the pretreatment, because these values are those that really take part in physicochemical reactions. During the beginning of pretreatment the principal heating source results from the transfer of latent heat of steam condensation, the material reaches the equilibrium temperature. After this period, there is no heat transfer and the exchange is dominated by the absorption of condensed steam, resulting in an increase of moisture content. The amount of absorbed water reaches equilibrium after a certain time depending on pressure level [28]. Zarguili et al. [29] showed an increase in the moisture content during the pretreatment of starch when the pressure level and the processing time increase. Besides, the quadratic effect of processing pressurex21 , was relatively significant for the two studied responses as well. For initial hydrolysis rate the quadratic effects of processing time and in a lesser extent the quadratic effect of initial moisture content were also significant. Fig. 4 shows the three-dimensional response surfaces which was constructed to show the interaction effects of two pretreatment variables (processing pressure and processing time) on the two studied responses. For these plots, the initial moisture content of wheat straw was fixed at central level (25%). As can be seen from these figures, for low processing times (w15 min), the hydrolysis yield as well as the initial hydrolysis rate are closely constant when the processing pressure increase from 300 to 600 kPa while for a high processing time (w50 min) one can observe an increasing in the two responses: from 11 to 32 g of glucose/100 g d m for hydrolysis yield which represents almost
Table 5 e Analysis of variance for the fit of experimental data to response surface models.
Hydrolysis yield
Initial hydrolysis rate
a p-value<0.01. b p-value<0.05.
Source
Degrees of freedom
Sum of square
Mean square
F-ratio
Model Linear Quadratic Interactions Residual Lack of fit Pure error
9 3 3 3 12 5 7
966.53 741.01 49.64 175.86 39.58 34.51 5.07
107.39 247.00 16.55 58.62 3.30 6.90 0.72
32.54a 74.84a 5.02b 17.76b e 5.44 e
R2
0.96
Model Linear Quadratic Interactions Residual Lack of fit Pure error R2
9 3 3 3 12 5 7 0.86
1.9819 1.1110 0.3331 0.5379 0.3039 0.2972 0.0067
0.2202 0.3702 0.1111 0.1793 0.0253 0.0594 0.0009
8.7035a 14.6324a 4.3913 b 7.0869 b e 0.0210
3136
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
100% of the theoretical cellulose and from 0.003 to 0.98 g/l/h for initial hydrolysis rate. Thus, in the range of studied variables, a low processing time or a low processing pressure are insufficient to break down the lignocellusic structure and subsequent hydrolysis of the hemicellulose fraction which lead to a high sugar recovery. The results obtained agreed with works of Balat et al. [30] which reported that in uncatalysed steam explosion processes as it is the case in this study, hemicelluloses is thought to be hydrolysed by acetic and other acids released during steam explosion pretreatment causing an increasing in global hydrolysis. For enzymatic hydrolysis of sunflower stalks Ruiz et al. [31] also argued that the hydrolysis yield increased with the temperature (or steam pressure). In this study, the optimal values of hydrolysis yield and initial hydrolysis rate were reached for high processing pressure and high processing time.
3.4. Effect of D.I.C thermomechanical pretreatment on the composition and microstructure of wheat straw To evaluate the effect of the proposed thermomechanical pretreatment on the composition of wheat straw, processing pressure was fixed at 700 kPa and the processing time was varied from 20 to 40 min. From Table 6, it can be seen that the cellulose fraction in the pretreated wheat straw increased from 34 to 37 g of glucose/100 g d m when the processing time increased from 20 to 40 min. These results are in agreement with those of Kabel et al. [32] which indicated that the cellulose level in the solid residues after pretreatment increased with the pretreatment severity while in the same time the xylan fraction decreased. From Table 6, we can see that the hemicellulosic fractions remained closely constant and this is in agreement with the review of Mosier et al. [6] who argued that an effective pretreatment preserves the pentose fractions. Moreover, the lignin fraction decreased from 22% in raw material to 16 and 17% for the pretreated samples. In a study on hydrothermal pretreatment of wheat straw through IBUS process, Petersen et al. [33] pointed out some degradation and a possible dissolution of lignin fraction. Another important parameter is the acidity of the system. According to Ruiz et al. [31] who studied the steam explosion
of sunflower stalks, at higher pretreatment temperature lower final pH values were obtained, ranging from 4.1 to 5.2 for a severity factor R0 in the range 3.1e4.5. As described by Overend and Chornet [34]: R0 ¼ t exp ðT 100=14:75Þ; where t is the residence time in minutes and T is the pretreatment temperature in C. Another study [35] reported on pH of extracts obtained after steam explosion of sugarcane bagasse which varied in the range 3.2e4.7, for experiments with severity factor between 3.7 and 4.3 while in our study the severity factor for the experiments reported in Table 6 was 3.21 and 3.51 for the samples pretreated for 20 min and 40 min for which almost a neutral values of pH were obtained: 5.92 and 5.66, respectively. According to Sun [36] a low pH values (high acidity) is an indication of sugar degradation and a formation of inhibitor components which is a negative aspect for further fermentation and resulting in a lower yield in ethanol. The author indicated that in a number of pretreatments of lignocellulosic materials although disrupting the plant cell walls and improves enzymatic access to the cellulose, the by-products as acetic acid, formic acid or furfural and its derivatives inhibit the in-vitro activity of several enzymes. In our study a weak decreasing of acidity from 5.92 to 5.66 may indicate a low sugar degradation and formation of inhibitors. The lignocellulosic complex is made up of matrix of cellulose and lignin bound by hemicellulose chains. During the pretreatment, this matrix should be broken in order to reduce the crystallinity of cellulose and increase the fraction of amorphous phase, the most suitable form for enzymatic attack. The cellulose hydrolysis is also affected by the porosity, generally expressed by the accessible surface area in the lignocellulosic materials [36]. The use of SEM permits a successful visualisation of morphological and structural changes of wheat straw cellulose fibres after the proposed thermomechanical pretreatment. These structural changes can be observed by comparing Fig. 5a and b. The SEM image of raw wheat straw prior to pretreatment shows a large part of relatively compact zones, arranged tightly, with a mean circumference of ellipsoidal alveoles of about 58.2 mm. After the thermomechanical pretreatment, the structure is more expanded with a large alveoles, some defibrillation and dispersion is visible in the microstructure with a high mean circumference alveoles of about 92 mm.
Table 6 e Chemical composition of wheat straw for two thermomechanical pretreated samples at 700 kPa and two processing times: 20 and 40 min. Composition of wheat straw (g/100 g d m) 700 kPa, 20 min 700 kPa, 40 min R0 (severity factor) Cellulosea Hemicelluloseb Ligninb pH Hydrolysis yield (%) Initial hydrolysis rate (g/l/h)
3.21 34 49 17 5.92 0.04 19.2 0.385
a determined by acid and enzymatic hydrolysis. b determined by acid hydrolysis.
3.51 37 47 16 5.66 0.02 32.1 0.769
Fig. 5 e Microstructure of wheat straw (1000e8000 mm) unpretreated (a) and pretreated (b) during 15 min at 700 kPa (magnification X 1000).
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
3.5.
Conclusion
From this work on a new pretreatment method to produce fermentable sugars from wheat straw an encouraging results were obtained. We concluded that hydrolysis yield and initial hydrolysis rate of wheat straw after pretreatment by the proposed thermomechanical pretreatment were more influenced by processing pressure and processing time than by initial moisture content. The significant variables exerted a positive linear effect on hydrolysis yield and initial rate of hydrolysis. For the two responses, the empirical models gave R2 of 96 and 86% and p-values less than 0.01, which implied a good agreement between the predicted and actual values. The optimal conditions to obtain the highest yield are showed in Fig. 4: 600 kPa for the processing pressure and 50 min for processing time. Initial moisture content being not a critical variable, we consider that the optimal value is its residual value (w10%). In this optimisation the vacuum degree was fixed at 50 kPa. This factor will be investigated in a next research. The second part of this study showed that, at residual moisture content, a high hydrolysis yield and initial hydrolysis rate can be achieved. 700 kPa, which can be considered as a moderate processing pressure, compared to the literature, and 40 min as processing time, are sufficient to hydrolyse 100% of the theoretical cellulose present in raw material. In these conditions, the predicted value of hydrolysis yield (33 g of glucose/100 g d m) is close to that observed (32.1 g of glucose/100 g d m). In addition, as shown by Fig. 5, the proposed pretreatment enhance the accessibility of enzymes in the substrate due to the flash expansion provoked by the sudden release of steam pressure. Using this pretreatment method, a reproducible hydroloysis yield can be achieved without adding any solvent or acid, representing economic and ecologic advantages.
Acknowledgements The authors acknowledge the financial support of Region Poitou-Charentes and OSEO Innovation through A0809006T project.
references
[1] Silverstein RA, Chen Y, Sharma-Shivappa RR, Boyette MD, Osborne J. A comparison of chemical pre-treatment methods for improving saccharification of cotton stalks. Bioresour Technol 2007;98:3000e11. [2] Solomon BD, Barnes JR, Kathleen E, Halvorsen KE. Grain and cellulosic ethanol: history, economics, and energy policy. Biomass Bioenerg 2007;31:416e25. ´ J. Fuel ethanol production: process [3] Cardona CA, Sa´nchez O design trends and integration opportunities. Bioresour Technol 2007;98:2415e57. [4] Tabka MG, Herpoe¨l-Gimbert I, Monod F, Asther M, Sigoillot JC. Enzymatic saccharification of wheat straw for bioethanol production by a combined cellulose xylanase and ferulolyl esterase treatment. Enzyme Microb Tech 2006;39: 897e902.
3137
[5] Cara C, Ruiz E, Ballesteros M, Manzanares P, Jose´ Negro M. Castro. Production of fuel ethanol from steam-explosion pretreated olive tree pruning. Fuel 2008;87:692e700. [6] Mosier N, Wyman C, Dale B, Elander R, Lee YY, Holtzapple M, et al. Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresour Technol 2005;96:673e86. [7] Hu G, Heitmann JA, Rojas OJ. Feedstock pre-treatment strategies for producing ethanol from wood, bark and forest residues. BioResources 2008;3:270e94. [8] Ogier J-C, Ballerini D, Leygue J-P, Rigal L, Pourquie´ J. Production d’e´thanol a` partir de biomasse lignocellulosique. Oil Gas Sci Technol 1999;54:67e94. [9] Sassner P, Ma˚rtensson C-G, Galbe M, Zacchi G. Steam pretreatment of H2SO4-impregnated Salix for the production of bioethanol. Bioresour Technol 2008;99:137e45. [10] Ballerini D, Alazard-Toux N. Les Biocarburants: e´tat des Lieux, Perspectives et Enjeux du De´veloppement. France: Technip ed, ISBN 2710808692; 2006. 348 pages. [11] Rodriguez G, Rodriguez R, Jime´nez A, Guille´n R, FernandezBolan˜os J. Effect of steam treatment of alperujo on the composition, enzymatic saccharification, and in vitro digestibility of alperujo. J. Agric. Food. Chem 2007;55:136e42. [12] Balan V, Dale B, Bals B. Separation of proteins from grasses integrated with ammonia fiber explosion (AFEX) pretreatment and cellulose hydrolysis. US Patent WO/2008/ 020901. [13] Zhang L, Li D, Wan L, Wang T, Zhang L, Chen XD, et al. Effect of steam explosion on biodegradation of lignin in wheat straw. Bioresour Technol 2008;99:8512e5. [14] Viola E, Cardinale M, Santarcangelo R, Villone A, Zimbardi F. Ethanol from eel grass via steam explosion and enzymatic hydrolysis. Biomass Bioenerg 2008;32:613e8. [15] Talebnia F, Karakashev D, Angelidaki I. Production of bioethanol from wheat straw: an overview on pretreatment, hydrolysis and fermentation. Bioresour Technol 2010;101: 4744e53. [16] Karimi K, Kheradmandinia S, Taherzadeh MJ. Conversion of rice straw to sugars by dilute-acid hydrolysis. Biomass Bioenerg 2006;30:247e53. [17] Rezzoug S-A, Maache-Rezzoug Z, Sannier F, Allaf K. A thermomechanical preprocessing for pectin isolation from orange peel. Optimisation by response surface methodology. Int J Food Eng 2008;4(1). Article 10. [18] Rezzoug S-A, Maache-Rezzoug Z, Mazoyer J, Jeannin M, Allaf K. Effect of instantaneous controlled pressure drop proccess on hydration capacity of scleroglucan. Optimisation of operating conditions by response surface methodology. Carbohydr Polym 2000;42:73e84. [19] Nouviaire A, Lancien R, Maache-Rezzoug Z. Influence of hydrothermal treatment on rheological and cooking characteristics of fresh egg pasta. J Cereal Sci 2008;47:283e91. [20] Maache-Rezzoug Z, Zarguili I, Loisel C, Queveau D, Bule´on A. Structural modifications and thermal transitions of standard maize starch after D.I.C. hydrothermal treatment. Carbohydr Polym 2008;74:802e12. [21] Wen Z, Liao W, Chen S. Hydrolysis of animal manure lignocellulosics for reducing sugar production. Bioresour Technol 2004;91:31e9. [22] Mussatto SI, Roberto IC. Acid hydrolysis and fermentation of brewer’s spent grain to produce xylitol. J Sci Food Agric 2005; 85:2453e60. [23] Benoist D, Tourbier Y, Germain-Tourbier S. Plan d’expe´riences: Construction et Analyse. Paris: Lavoisier Tec & Doc; 1994. Chap 5, pp. 208e389. [24] Statgraphics Plus for Windows 5.1. Experimental design. Manugistics, Inc.; 1994. [25] Statgraphics Plus for Windows 5.1. User Manual. Manugistics, Inc.; 1994.
3138
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 3 1 2 9 e3 1 3 8
[26] Ballesteros I, Oliva JM, Navarro AA, Gonzalez A, Carrasco J, Ballesteros M. Effect of chip size on steam explosion pretreatment of softwood. Appl Biochem Biotech 2000;84-86: 99e110. [27] Hendriks ATWM, Zeeman G. Pretreatments to enhance the digestibility of lignocellulosic biomass. Bioresour Technol 2009;100:10e8. [28] Bahrani SA, Loisel C, Monteau J-Y, Rezzoug S-A, MaacheRezzoug Z. Estimation of effective moisture diffusivity in starchy materials following hydrothermal treatments. To be published in “Defect and Diffusion Forum”. ISSN 1021e0386 (2011). [29] Zarguili I, Maache-Rezzoug Z, Loisel C, Doublier J-L. A mathematical model to describe the change of moisture distribution in maize starch during DIC hydrothermal treatment. Int J Food Sci Tech 2009;44:10e7. [30] Balat M, Balat H, Oz C. Progress in bioethanol processing. Prog Energ Comb Sci 2008;34:551e73. [31] Ruiz E, Cara C, Manzanares P, Ballesteros M, Castro E. Evaluation of steam explosion pretreatment for enzymatic
[32]
[33]
[34]
[35]
[36]
hydrolysis of sunflower stalks. Enzyme Microb Technol 2008; 42:160e6. Kabel MA, Bos G, Zeevalking J, Voragen AGJ, Schols HA. Effect of pretretment severity on xylan solubility and enzymatic breakdown of the remaining cellulose from wheat straw. Bioresour Technol 2007;98:2034e42. Petersen MØ, Larsen J, Thomsen MH. Optimization of hydrothermal pretretment of wheat straw for production of bioethanol at low water consumptioin without adition of chemicals. BiomassBioenerg 2009;33:834e40. Overend RP, Chornet E. Fractionation of lignocellulosics by steam aqueous pretreatments. Philos Trans R Soc Lond A 1987;96(5):862e70. Kaar WE, Gutierrez CV, Kinishita CM. Steam explosioin of sugarcane bagasse as pretreatment for conversion to ethanol. Biomass Bioenerg 1998;14:277e87. Sun RC. Detoxification and separation of lignocellulosic biomass prior to fermentation for bioethanol production by removal of lignin and hemicellulose. BioResources 2008;4(2):452e5.