- Email: [email protected]
Statistical optimization of operating conditions for supercritical carbon dioxide-based pretreatment of guayule bagasse
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Available online at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Statistical optimization of operating conditions for supercritical carbon dioxide-based pretreatment of guayule bagasse Narayanan Srinivasan, Lu-Kwang Ju* Department of Chemical and Biomolecular Engineering, The University of Akron, 200 E Buchtel Commons, Akron, OH 44325-3906, USA
article info
abstract
Article history:
A central composite design (CCD) was used to find the optimal temperature, pressure,
Received 1 June 2011
moisture and duration for the supercritical CO2-based pretreatment of guayule, a desert
Received in revised form
shrub for commercial production of hypoallergenic latex. The pretreatment involved:
10 April 2012
adding water to the biomass, raising system temperature, pressurizing with supercritical
Accepted 5 September 2012
CO2, holding the system for a period of time, and exploding the biomass by rapidly opening
Available online 6 October 2012
the vent valve. The pretreated biomass was then hydrolyzed at 30 C for 72 h in an enzyme solution with 5% (w/v) solid loading. The yields of released glucose and pentose were
Keywords:
determined and used as the CCD response variables. Statistical analysis of results led to the
Guayule
following recommended condition: 175 C, 26.2 MPa (3800 psi), 60% moisture and 30 min.
Supercritical CO2
The corresponding glucose and pentose yields were 56% and 61%. X-ray diffraction analysis
Biomass pretreatment
was done to get the crystallinity index of the bagasse before and after pretreatment.
Scanning electron microscope
Scanning electron microscopy was also used to examine the structural changes caused by
X-ray diffraction
pretreatment. These characterizations indicated that at milder conditions, the pretreat-
Back pressure
ment exposed cellulose and hemicellulose for subsequent enzyme attack; at harsher conditions, the pretreatment destroyed cellulose crystallinity and gave higher glucose yields. To isolate the explosion effect (caused by instantaneous pressure drop, DP) from the reaction (pressure) effect, a series of experiments was made with the bagasse pretreated at the same condition (27.6 MPa) but exploded against different back pressures. The explosion needed to be severe enough (with DP > 17 MPa) to give a sugar yield of over 50%. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Using lignocellulosic biomass as biorefinery feedstock for production of ethanol [1], butanol [2,3], and other chemicals [4,5] via fermentation typically requires the polymeric carbohydrates (cellulose and hemicellulose) to be broken down to small sugar molecules (e.g., glucose and xylose) prior to the fermentation. To enhance the saccharification, particularly if by the gentler enzyme hydrolysis, the biomass needs to be
pretreated to loosen the structure and increase the accessibility and susceptibility of cellulose and hemicellulose [6,7]. In our earlier study a supercritical CO2-based method was evaluated for its feasibility to pretreat waste guayule biomass (bagasse) [8]. Guayule (Parthenium argentatum Gray) [9,10] has attracted considerable interest due to its use in commercial production of hypoallergenic latex and high-quality, multipurpose resins [11e14]. However, latex and resins account for no more than 20% of the guayule dry weight. The rest of the
* Corresponding author. Tel.: þ1 330 972 7252; fax: þ1 330 972 5856. E-mail addresses: [email protected], [email protected] (L.-K. Ju). 0961-9534/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biombioe.2012.09.009
452
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Nomenclature CCD CrI D b D DNS ESEM FPU Iam I002 M
central composite design crystallinity index, as described in Eq. (1) duration of supercritical pretreatment, min dimensionless pretreatment duration in Eqs. (5) and (6), ¼(D 30)/10 dinitrosalicylic acid Environmental Scanning Electron Microscope filter paper unit, a measure of cellulase enzyme activity X-ray diffraction intensity at 2q ¼ 18.7 X-ray diffraction intensity at 2q ¼ 22.6 moisture, weight % of water in bagasse
plant material, already chopped, milled and extracted [15], is potentially a good source of waste biomass for biorefinery. The supercritical CO2-based pretreatment method used in our earlier study involved the following steps: adding a small amount of water to the bagasse loaded in the reactor, raising the system temperature, pressurizing using supercritical CO2, holding the system for a period of time, and exploding the bagasse by rapidly opening the valve to release the pressure [8]. The pretreated biomass was then subjected to enzyme hydrolysis at 30 C for 72 h and the yields of released glucose and total reducing sugars measured. The results confirmed the feasibility of the supercritical CO2-based method in preparing guayule bagasse for the subsequent enzymatic hydrolysis. Unlike the commonly used dilute acid pretreatment, the supercritical method did not incur appreciable loss/ hydrolysis of cellulose and hemicellulose and did not require neutralization or water wash of the pretreated biomass. Conceptually four operation factors may be important to the performance of the supercritical pretreatment. These factors are: temperature, pressure, moisture content of biomass, and the pretreatment (reaction) duration. In the previous feasibility study the pretreatment was only done at 6 conditions [8]. In this study the pretreatment was done at 26 conditions, chosen according to the factorial design, and the response surface analysis [16,17] was performed to determine the optimal pretreatment condition. In addition to the sugar yields from enzyme hydrolysis of pretreated bagasse, X-ray diffraction (XRD) analysis was used to assess the effect of pretreatment on the crystallinity of bagasse pretreated under different conditions, similar to the previous work of Zheng et al. [18]. Lowering cellulose crystallinity is known to enhance the performance of enzyme hydrolysis [19,20]. Scanning electron microscopy (SEM) was also used to examine the changes of bagasse structure caused by the pretreatment. Observing the plant material structures such as lignin sheath and cellulose and hemicellulose fibers with SEM had been reported in the literature [21]. Furthermore, the “explosion” (rapid pressure release) effect was shown important to the outcome of the supercritical pretreatment process in our previous study [8]. Hypothetically, larger amounts of CO2 would enter into the bagasse structure under the supercritical conditions, with the penetration being assisted by the water absorbed in bagasse [18,22]. When the vent
c M P b P DP SAS SEM T b T XRD
dimensionless moisture in Eqs. (5) and (6), ¼(M 60)/10 pressure, MPa dimensionless pressure in Eqs. (5) and (6), ¼( p 27.6)/3.45 pressure drop between the reaction pressure and the venting pressure, MPa Statistical Analysis System scanning electron microscopy temperature, C dimensionless temperature in Eqs. (5) and (6), ¼(T 165)/35 X-ray diffraction
valve is rapidly opened, CO2 expands and rushes out of the biomass, rupturing the bagasse structure and exposing the cellulose and hemicellulose for subsequent enzyme hydrolysis. Accordingly, the pretreatment effectiveness was expected to be affected by the magnitude of the pressure drop (DP) between the reaction pressure and the venting pressure employed. Nonetheless, the magnitude of DP needed for an adequate “explosion” effect was unknown. To study this latter factor in isolation, a set of experiments were done by using the same pressure in the supercritical reactor but different back pressures in the vent vessel (into which the reactor fluid was vented). In real production processes the CO2 collected in the vent vessel would be repressurized and used for the supercritical reactor. A lower DP (i.e., a higher P in the vent vessel) may be more economic in minimizing the energy required for repressurization. The objectives of this work were to optimize the pretreatment conditions in the reactor and the (back) pressure in the vent vessel, and to examine the effects of supercritical pretreatment on the crystallinity and structures of guayule bagasse (and their correlation with the outcome of subsequent enzyme hydrolysis).
2.
Materials and methods
2.1.
Materials
Guayule plant material was provided by Yulex Corporation (Carlsbad, CA). The harvested shrub was chopped, the leaf stream removed, and the branches wet milled for collection of latex as the primary product. The remaining biomass, referred to as “bagasse”, could then be extracted for resin and/or remaining rubber. The bagasse that had been extracted for resin and rubber by the supercritical extraction procedures [15] at Yulex Corporation was transported to our lab for use in this optimization study. The bagasse was in the form of rod-shaped dry powders of about 100e300 mm in diameter and 300e2000 mm in length. The bagasse sample had the following composition [8]: 21% (3%) cellulose, 13% (1%) hemicellulose, 54% (2%) acid insoluble materials, 0.7% (0.1%) acid soluble lignin, and 11% (3%) other (unaccounted for) materials. The enzyme solution (Spezyme CP) used in the study for enzyme hydrolysis of pretreated bagasse was obtained from
453
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Genencor. The enzyme solution was analyzed to have 32.3 FPU/mL of cellulase activity, 26.7 IU/mL of b-glucosidase activity, and 12.4 U/mL (¼208 nkat/mL) of xylanase activity.
Table 1 e Code values of variables e temperature, pressure, moisture, and pretreatment duration e used in the central composite design.
2.2.
Methods
Coding level
2.2.1.
Supercritical CO2-based pretreatment
Fifteen (15) grams of bagasse (moisture content predetermined, 4.4%e6.7%) were placed in a reactor (500 mL) along with water (the added amount calculated to give the investigated water content). The reactor wall was fully wrapped with heating coils containing circulating oil that had been preheated to the reaction temperature. The closed reactor was held at a constant pressure for a particular duration by pumping CO2 into the reactor using a syringe pump (Model 260D, ISCO Inc., Lincoln, NE). The reactor fluid was then vented rapidly, creating the fast pressure drop for the explosion effect. Thus pretreated solids were collected from the reactor for further evaluation and characterization.
2.2.2.
Enzymatic hydrolysis
The hydrolysis was done with 5 g (on the dry basis) pretreated bagasse in 100 mL citrate buffer (pH 4.8, 0.05 M sodium citrateecitric acid) containing 5 mg sodium azide and the following enzyme activities: 0.92 FPU/mL, 0.76 bglucosidase IU/mL and 0.35 xylanase U/mL (prepared from the Spezyme CP solution). The experiments were carried out at 50 C in 125-mL conical flasks shaken at 200 rpm (Model 4730 orbital shaker, Queue Systems, Inc., Parkersburg, WV). Samples were taken periodically from each flask for analyses of total reducing sugars and glucose.
2.2.3. Optimization of pretreatment temperature, pressure, moisture and duration Central composite design (CCD) was used to generate 26 treatment combinations, with temperature, pressure, moisture and duration of hydrolysis as independent variables and the yields of glucose and non-glucose reducing sugars (predominantly pentose) from enzymatic hydrolysis as the response variables in two separate statistical analyses. Table 1 shows the matrix of experimental design, with coded variables containing star points (a ¼ 2) and two replicates of the central point. The experimental data were analyzed by Statistical Analysis System (SAS, package version 9.1) and Minitab 5.1.
2.2.4.
Back pressure study
For the study of back pressure effect, all the bagasse samples were pretreated under the same operating conditions, i.e., 27.6 MPa (4000 psi), 165 C, 60% moisture and 30 min reaction time, but the supercritical fluid in the reactor was allowed to explode into a vessel maintained at different pressures using an automated back pressure regulator (Thar Technologies, Inc., Pittsburgh, PA). The bagasse pretreated was similarly subjected to enzyme hydrolysis for sugar yields and the XRD and SEM characterizations.
2.2.5.
Analytical techniques
Cellulose and hemicellulose contents in the bagasse samples were measured using the standard procedures prescribed by National Renewable Energy Laboratory (NREL; Golden, CO) of
Temperature, C
2 1 0 1 2
95 130 165 200 235
Pressure, MPa (psia) 20.7 24.1 27.6 31.0 34.5
(3000) (3500) (4000) (4500) (5000)
Moisture, %
Duration, min
40 50 60 70 80
10 20 30 40 50
Note: a The values given in parentheses are pressures in the unit of psi.
the US Department of Energy [23e26]. The contents of acid insoluble materials (including lignin) and acid soluble lignin, respectively, were also obtained from the above procedures. A Philips 3100 generator XRD powder diffractometer (Almelo, The Netherlands) fitted with Cu-Ka lamp at 40 kV and 35 mA was used to get the XRD pattern of the (untreated or pretreated) bagasse for a measure of the crystallinity index. All samples were scanned from 2q ¼ 2 e60 at a stepping rate of 0.02 /s. The crystallinity index (CrI) was estimated [18] as CrI ¼
I002 Iam 100 I002
(1)
where I002 and Iam are the diffraction intensities at 2q ¼ 22.6 and 18.7 , respectively. An FEI Quanta 200 Environmental Scanning Electron Microscope (ESEM, FEI Corporation, Brno, Czech Republic) at 25 kV and spot size 40 was used to examine the bagasse structures before and after different pretreatment conditions. A small amount of the sample was mounted on a stainless steel or aluminum stub (1 cm in diameter) with two sided adhesive tape. The sample was sputter coated with silver particles for 1 min and then viewed under vacuum. The total reducing sugar concentration was measured by the dinitrosalicylic acid (DNS) method, based on the color formation of DNS reagent when heated with reducing sugars [27]. The glucose concentration was measured by using the PGO (peroxidase and glucose oxidase) enzymes test kit (Sigma P7119). The glucose yield from guayule bagasse, going through both pretreatment and enzyme hydrolysis, was estimated by the following equations: Glucose yield ¼
Glucose obtained in hydrolysate Total glucose releasable from bagasse
Glucose in hydrolysate ¼Measured glucose concentration 0:1 ðLÞ Glucose releasable ¼ ð% cellulose in bagasseÞ ¼ 5ðgÞ 21%
180 162
180 162
(2) g L ð3Þ
(4)
The numbers given above corresponded to the use of 5 g biomass in 100 mL enzyme solution for the hydrolysis. It should be noted that the supercritical CO2-based pretreatment was predominantly physical and did not cause
454
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
detectable loss of solids. The value 180/162 was included to account for the addition of water as a result of hydrolysis. Similar equations were used to estimate the overall yield of reducing sugars and the yield of pentose. For the latter the pentose in hydrolyzate was estimated by subtracting the moles of glucose from moles of reducing sugars and then multiplying 150 (pentose molecular weight). The maximum pentose releasable was estimated by multiplying (% hemicellulose in bagasse) by the factor 150/132, which accounts for addition of water in the hemicellulose hydrolysis. The yields obtained with these equations were used as the response variables in the optimization of operating conditions for the supercritical pretreatment.
3.
Results
3.1.
Optimization of pretreatment process parameters
Enzyme hydrolysis of the pretreated bagasse showed the typical profiles of rapid initial production of reducing sugars and glucose, which approached the plateau sugar concentrations by about 24 h (data not shown, [8]). To ensure completion of the corresponding hydrolysis achievable for various samples, the sugar yields at 72 h were used as the response variables in the statistical CCD analysis. The 4 principal factors affecting the supercritical CO2 pretreatment process, i.e., temperature (T ), pressure (P), moisture (M ) and duration (D) of reaction, were analyzed for determination of the optimum conditions for maximal yields of glucose and pentose (non-glucose reducing sugars). All of the results obtained in the experimental design are summarized in Table 2. Table 3 shows the p values obtained from statistical analysis of the experimental data, for the factors with p < 0.075. (This 0.075 p value was chosen to allow inclusion of the P effects.) Minitab 5.1 statistical software was used to obtain the regression equations. The corresponding regression equations for glucose and pentose yields (%) are given as follows: b þ 0:68 P b þ 1:75 c b þ 0:32 P b Glucose yield ¼ 61:2 þ 2:89 T M þ 1:57 D
2
2
b þ 1:73 T bP b þ 1:06 T bc þ 0:83 D M 2 b þ 1:28 c MD r ¼ 0:92 (5)
2
b þ 0:71 P b þ 1:29 c b 0:47 P b Pentose yield ¼ 55:7 þ 1:79 T M þ 1:13 D 2 2 b 0:19 T bP b þ 0:44 T bc b r ¼ 0:87 0:09 D M þ 0:31 c MD (6) b T T0 =DT ¼ T 165=35Þ; Pð¼ b P 27:6=3:45Þ; Tð¼ c b D 30=10Þ Mð¼ M 60=10Þ;and Dð¼ in the above equations are the coded dimensionless parameters (Table 1), while the original operating conditions have the following units: T e C, P e MPa, M e wt% (water in bagasse), and D e min. The optimal conditions and the sugar yields at the optimal conditions obtained from the statistical analysis are given in Table 4. The two sets of conditions (one for maximal glucose yield, the other for maximal pentose yield)
did not differ very much. Accordingly, the following pretreatment conditions were recommended: T ¼ 175 C, P ¼ 26.2 MPa (3800 psi), M ¼ 60% and D ¼ 30 min. As shown by the p values (Table 3), the pressure effects b and P b2 terms having p values were less significant, with both P bP b term had smaller p close to or larger than 0.05. Only the T b 2 Þ and values. There were other significant nonlinear ð D bc b in addition to T b P) b effects (with interactive ( T M and c M D, p < 0.05). Nonetheless, it should be noted that there needs to be certain critical pressure to effect the pretreatment (by explosion and mild hydrolysis due to the acidity of CO2 in water) although the effect of pressure was less significant within the range studied in this work (i.e., 20.7e34.5 MPa). The optimal conditions given in Table 4 were not at the highest T or D studied, suggesting that too high temperature or too long reaction duration was not beneficial to sugar production (presumably due to degradation of sugars to other derivatives). Mok et al. [28] observed similar trends in dilute acid-catalyzed hydrolysis of cellulose. They attributed the lower glucose yield to the formation of certain oligomers that could no longer be hydrolyzed to glucose. It is also well known that extended hydrolysis could result in the formation of furfurals from pentose. (The furfurals could be inhibitory to the fermentation processes using hydrolyzate as the sugar source.) The lower temperature optimum for pentose yield (170 C) than that for glucose yield (179 C, Table 4) might be reflecting the higher susceptibility of pentose to degradation at high temperatures. While the CCD experiments allowed the identification of optimal operating conditions, the regression equations obtained were purely empirical. More mechanistic models and experimental designs are necessary for gaining more fundamental understanding of the pretreatment mechanisms.
3.2.
Crystallinity index
The guayule bagasse samples pretreated under different conditions were analyzed for X-ray diffraction patterns. Accordingly, the crystallinity indices (CrI) of these samples were calculated as described in the Analytical techniques section. The CrI values are plotted in Fig. 1 against the % glucose yields after both pretreatment and enzyme hydrolysis. For comparison the data point for enzyme hydrolysis of the bagasse sample that was not pretreated is also included in Fig. 1. The untreated bagasse had a CrI of 38.2 and the glucose yield from the untreated bagasse was just 26%. The CrI of some pretreated bagasse samples were higher than 38.2 while others were lower. For the samples with increased CrI, the pretreatment conditions might be strong enough to break down and/or remove the outer covering of lignin and pectin around the cellulose fibers but not harsh enough to significantly destroy the crystallinity of exposed cellulose. As the pretreatment conditions became increasingly harsher, the pretreatment broke down more of the crystalline cellulose apart from removing the lignin sheath and the CrI dropped below that of the untreated bagasse. All of the pretreatment conditions resulted in much higher glucose yields (>42%), as compared to the 26% glucose yield from the untreated sample. Even the milder pretreatments
455
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Table 2 e Glucose and pentose yields of the CCD experiments, showing the 26 pretreatment conditions. Temperature ( C)
Pressure (MPa)
Moisture (%)
Duration (min)
24.1 24.1 24.1 24.1 31.0 31.0 31.0 31.0 24.1 24.1 24.1 24.1 31.0 31.0 31.0 31.0 27.6 27.6 27.6 20.7 34.5 27.6 27.6 27.6 27.6 27.6
50 50 70 70 50 50 70 70 50 50 70 70 50 50 70 70 60 60 60 60 60 40 80 60 60 60
20 40 20 40 20 40 20 40 20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 10 50 30
130 130 130 130 130 130 130 130 200 200 200 200 200 200 200 200 95 165 235 165 165 165 165 165 165 165
Glucose yield (%) 48 51 51 47 48 48 47 52 55 49 53 57 46 55 54 49 42 56 46 52 45 53 51 44 54 54
1.6 1.6 1.4 0.6 2.3 0.5 2.7 3.4 0.6 2.1 1.5 0.4 0.7 1.9 1.2 4.3 3.0 1.6 2.1 4.8 4.3 2.7 1.4 4.6 1.9 0.7
Pentose yield (%) 54 56 57 52 54 53 53 58 59 55 57 60 52 54 52 55 47 61 50 57 51 60 56 49 62 61
0.5 1.4 2.1 0.3 2.2 2.3 2.8 2.2 1.4 0.8 1.6 2.2 1.9 2.0 2.6 5.2 0.9 0.9 3.7 3.7 0.9 0.6 2.2 1.0 1.0 2.9
Notes: The bagasse pretreated under each condition was subjected to enzyme hydrolysis in duplicate flasks. Periodical samples were taken from each flask for triplicate analyses of total sugars and glucose concentrations. The average glucose yield and pentose yield were reported with standard deviations (from the 2 3 analysis results) for the final samples taken at 72 h of enzyme hydrolysis.
(giving higher CrI) exposed more cellulose for the enzyme hydrolysis. As the harsher pretreatment caused more destruction of cellulose crystallinity, the CrI decreased and the glucose yields further increased (Fig. 1). This finding is consistent with the fact that the cellulase enzymes hydrolyze amorphous cellulose faster than crystalline cellulose [29]. On the other hand, there was no apparent correlation between the pentose yield and CrI (not shown). Tentatively, the following best-fit correlation was found between the glucose yield (%) and CrI for the supercritical CO2 pretreated guayule bagasse: CrI ¼ 77:6 0:8,ðGlucose yieldÞ
ðr2 Þ ¼ 0:84
(7)
Table 3 e The p-values of terms chosen in the regression analysis (with p < 0.075). Factors T P M D P2 D2 TP TM MD
p (glucose yield)
p (pentose yield)
0.000 0.047 0.000 0.000 0.049 0.018 0.000 0.015 0.005
0.000 0.061 0.011 0.019 0.055 0.034 0.021 0.007 0.023
3.3. SEM-revealed structural changes of bagasse due to pretreatment The bagasse samples before and after pretreatment were observed using SEM to probe for changes in the bagasse structure. The differences observed are shown in Fig. 2(a)e(d). From the SEM pictures of the raw bagasse in Fig. 2(a), it is clearly seen that there was a continuous sheath of lignin covering the cellulose fibers. The fiber structures can be seen underneath the lignin sheath. In the pretreated bagasse (Fig. 2(b)e(d)), the lignin sheath was broken and the exposed cellulose fibers are directly visible. The supercritical pretreatment was effective in causing physical damages to the bagasse and in removing the lignin sheath around the cellulose fibers. Also, some holes appeared in the fibers of pretreated bagasse. The size of the holes ranged roughly from 2 to 5 mm. The holes are marked in the figures with red circles. The holes were probably due to the rupture of the plant structure
Table 4 e Optimum pretreatment conditions for maximum glucose and pentose yields. Operating condition
Temperature ( C) Pressure (MPa) Moisture (%) Duration (min) Yield (%)
Glucose
Pentose
178.8 26.1 57.6 32.3 55.9
170.4 26.4 61.3 34.7 61.5
456
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
during the rapid explosion. Nonetheless, the extents of hole generation in the pretreated bagasse could not be effectively quantified and correlated with the sugar yields obtained under different pretreatment conditions. All of the pretreated bagasse (at different conditions) had holes in the plant structure when viewed under the SEM.
3.4.
Fig. 1 e Variation of crystallinity index (CrI) with % yield of glucose from enzyme hydrolysis of guayule bagasse untreated and pretreated at different conditions.
Effect of back pressure
The glucose yields obtained under different back pressures are shown in Fig. 3. All samples were pretreated at the same reaction condition, i.e., 27.6 MPa (4000 psi), 165 C, 60% moisture and 30 min reaction time, but were exploded against different back pressures. X-ray diffraction analysis was also done to obtain the CrI values of these bagasse samples, which values are shown in Fig. 3 too. The glucose yield decreased with increasing back pressure particularly when the back pressure was above 10.3 MPa. And, the trend of decreasing glucose yield corresponded well with the trend of increasing
Fig. 2 e SEM pictures: (a) untreated guayule bagasse; and bagasse pretreated at (b) 200 C, 24.1 MPa, 70% moisture, 20 min; (c) 165 C, 27.6 MPa, 60% moisture, 30 min; (d) 130 C, 24.1 MPa, 50% moisture, 40 min.
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
457
a
Fig. 3 e Effects of back pressure on glucose yield and crystallinity index of guayule bagasse pretreated by the supercritical CO2-based process.
CrI. Accordingly, certain magnitude of instantaneous pressure drop (DP), in this case, a drop from 27.6 to 10.3 MPa, was necessary for achieving a more optimal glucose yield, presumably by creating severe enough explosive damages to the biomass structure. There are two potentially important mechanisms involved in the supercritical CO2-based pretreatment; they are the physical explosion effect and the chemical hydrolysis effect (due to the acidity provided by the CO2 in water) [8,18,30]. Conceptually the explosion effect is determined more by the magnitude of instantaneous pressure drop DP (instead of the reaction pressure P), while the hydrolysis effect may be more influenced by the reaction P. In an attempt to differentiate the effects of DP and P, the variations of glucose and pentose yields with DP are shown in Fig. 4(a) and (b), respectively, for two types of pretreated samples (and the untreated bagasse). One type of samples was from the back pressure study. These samples were subjected to the same reaction/hydrolysis condition. Therefore, the different sugar yields at different DP could be attributed to the different explosion effects, to expose the cellulose and hemicellulose for enzyme hydrolysis and to lower the crystallinity of cellulose for easier hydrolysis. The other type of samples was pretreated at different reaction P (and the same T, M and D as in the back pressure study) but all were exploded to the atmosphere (i.e., DP and P were essentially the same, DP ¼ P 0.1 MPa). For these samples the different sugar yields could result from different explosion (DP) and/or hydrolysis (P) effects. As shown in Fig. 4(a) and (b), the glucose and pentose yields had similar dependency on DP, although the data were rather scattered. The sugar yields increased with DP up to about 21 MPa, and then plateaued at higher DP (seemingly with a slight decreasing trend at DP >about 27 MPa for glucose yield and about 24 MPa for pentose yield). As hemicellulose is more susceptible to dilute acid hydrolysis than cellulose is, the similar trends of glucose and pentose yields suggested that the main DP effect was to alter the extent of exposing the cellulose and hemicellulose for enzyme hydrolysis (instead of affecting the chemical hydrolysis occurring during the pretreatment).
b
Fig. 4 e Effects of pressure drop (DP) on (a) glucose yield and (b) pentose yield shown with the untreated bagasse (Raw) and two types of pretreated samples: BP e back pressure study (all pretreated at 27.6 MPa, 165 C, 60% moisture and 30 min reaction duration, and then exploded against different back pressures); ATMP e all samples exploded to the atmospheric pressure and pretreated at the same temperature, moisture and duration but different pressures (P).
4.
Conclusion
Results of the CCD experimental design indicated separate but close enough optimal supercritical CO2 pretreatment conditions for maximal glucose and pentose yields (i.e., 56% and 61%, respectively). Accordingly, the following pretreatment conditions were recommended: T ¼ 175 C, P ¼ 26.2 MPa (3800 psi), M ¼ 60% and D ¼ 30 min. The crystallinity indices estimated from the XRD analysis suggested that the main effect at milder pretreatment conditions was to expose the cellulose (and hemicellulose) for subsequent enzyme attack and that, with severer pretreatment conditions, cellulose crystallinity was also reduced to give correspondingly higher glucose yield after enzyme hydrolysis. The SEM results lent further support to the above framework of understanding, and
458
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
microscopic holes were visible on the fibers exposed by the pretreatment. The back pressure study results confirmed the importance of explosion effect on the pretreatment outcome and the explosion needed to be severe enough (with DP > 17 MPa, corresponding to the drop from a 27.6 MPa reaction pressure to a 10.3 MPa back pressure in this study) to give a sugar yield of over 50%.
Acknowledgment This work was supported by Yulex Corporation (Carlsbad, CA) and the U.S. Department of Agriculture under the Biomass Research and Development Initiative (grant number 68-3A757-610). Dr. J. Richard Elliott (Department of Chemical and Biomolecular Engineering, The University of Akron) allowed us to use freely the equipment available at the Ohio Supercritical Fluid Consortium. Dr. Katrina Cornish (Senior VP, R & D, Yulex Corporation) provided the bagasse samples and the information for supercritical extraction of resins and rubber. We are also grateful to Mr. Thomas P. Quick (Department of Geology and Environmental Science, The University of Akron) for his help with the X-ray diffraction and SEM studies.
references
[1] Sun Y, Cheng J. Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresour Technol 2002;83(1): 1e11. [2] Huang W-C, Ramey DE, Yang S- T. Continuous production of butanol by Clostridium acetobutylicum immobilized in a fibrous bed bioreactor. Appl Biochem Biotechnol 2004;113-116: 887e98. [3] Ezeji T, Qureshi N, Blaschek HP. Butanol production from agricultural residues: impact of degradation products on Clostridium beijerinckii growth and butanol fermentation. Biotechnol Bioeng 2007;97(6):1460e9. [4] Silva C, Roberto IC. Optimization of xylitol production by Candida guilliermondii FTI 20037 using response surface methodology. Process Biochem 2001;36(11):1119e24. [5] Matsumoto Ki, Kobayashi H, Ikeda K, Komanoya T, Fukuoka A, Taguchi S. Chemo-microbial conversion of cellulose into polyhydroxybutyrate through rutheniumcatalyzed hydrolysis of cellulose into glucose. Bioresour Technol 2011;102(3):3564e7. [6] Bhat M, Bhat S. Cellulose degrading enzymes and their potential industrial applications. Biotechnol Adv 1997; 15(3e4):583e620. [7] Chandra R, Bura R, Mabee W, Berlin A, Pan X, Saddler J. Substrate pretreatment: the key to effective enzymatic hydrolysis of lignocellulosics? Adv Biochem Eng Biotechnol 2007;108:67e93. [8] Srinivasan N, Ju L-K. Pretreatment of guayule biomass using supercritical carbon dioxide-based method. Bioresour Technol 2010;101(24):9785e91. [9] Cornish K. Hypoallergenic natural rubber products from Parthenum argentatum (gray) and other non-Hevea brasiliensis species. US Patent 5,717,050; 1998. [10] Schloman Jr WW, Wyzgoski F, McIntyre D, Cornish K, Siler DJ. Characterization and performance testing of guayule latex. Rubber Chem Technol 1996;69(2):215e22.
[11] Thames SF, Schuman TP, Reichel LW, Purvis WA, Poole PW. Guayule coproducts: emerging technology in industrial, marine, and peelable coatings. Int SAMPE Symp Exhibition 1996;41(2):1177e90. [12] Bultman JD, Gilbertson RL, Adaskaveg J, Amburgey TL, Parikh SV, Bailey CA. The efficacy of guayule resin as a pesticide. Bioresour Technol 1991;35(2):197e201. [13] Gutierrez C, Gonzalez-Coloma A, Hoffmann JJ. Antifeedant properties of natural products from Parthenium argentatum, P. argentatum, P. tomentosum (Asteraceae) and Castela emoryi (Simaroubeaceae) against Reticulitermes flavipes. Ind Crops Prod 1999;10(1):35e40. [14] Nakayama FS, Vinyard SH, Chow P, Bajwa DS, Youngquist JA, Muehl JH, et al. Guayule as a wood preservative. Ind Crops Prod 2001;14(2):105e11. [15] Cornish K, Martin JA, Marentis RT, Plamthottam S. Extraction and fractionation of biopolymers and resins from plant materials. US Patent 7,259,231; 2006. [16] Heck JX, Floˆres SH, Hertz PF, Ayub MAZ. Statistical optimization of thermo-tolerant xylanase activity from Amazon isolated Bacillus circulans on solid-state cultivation. Bioresour Technol 2006;97(15):1902e6. [17] Couto SR, Moldes D, Sanroma´n MSA. Optimum stability conditions of pH and temperature for ligninase and manganese-dependent peroxidase from Phanerochaete chrysosporium. Application to in vitro decolorization of poly R478 by MnP. World J Microbiol Biotechnol 2006;22(6):607e12. [18] Zheng Y, Lin HM, Wen J, Cao N, Yu X, Tsao GT. Supercritical carbon dioxide explosion as a pretreatment for cellulose hydrolysis. Biotechnol Lett 1995;17(8):845e50. [19] Yoshida M, Liu Y, Uchida S, Kawarada K, Ukagami Y, Ichinose H, et al. Effects of cellulose crystallinity, hemicellulose, and lignin on the enzymatic hydrolysis of Miscanthus sinensis to monosaccharides. Biosci Biotechnol Biochem 2008;72(3):805e10. [20] Laureano-Perez L, Teymouri F, Alizadeh H, Dale BE. Understanding factors that limit enzymatic hydrolysis of biomass. Appl Biochem Biotechnol 2005;124(1):1081e99. [21] Jones J, Semrau K. Wood hydrolysis for ethanol production? Previous experience and the economics of selected processes. Biomass (London) 1984;5(2):109e35. [22] Zheng Y, Lin H, Tsao G. Pretreatment for cellulose hydrolysis by carbon dioxide explosion. Biotechnol Prog 1998;14(6): 890e6. [23] Ehrman T. Standard method for determination of total solids in biomass. Laboratory analytical procedures 001. Golden, CO: National Renewable Energy Laboratory; 1994. [24] Ruiz R, Ehrman T. Determination of carbohydrates in biomass by high performance liquid chromatography. Laboratory analytical procedure 002. Golden, CO: National Renewable Energy Laboratory; 1996. [25] Templeton D, Ehrman T. Determination of acid-insoluble lignin in biomass. Laboratory analytical procedure 003. Golden, CO: National Renewable Energy Laboratory; 1995. [26] Ehrman T. Determination of acid-soluble lignin in biomass. Laboratory analytical procedure 004. Golden, CO: National Renewable Energy Laboratory; 1996. [27] Miller G. Use of dinitrosalicylic acid reagent for determination of reducing sugar. Anal Chem 1959;31(3):426e8. [28] Mok WS, Antal Jr MJ, Varhegyi G. Productive and parasitic pathways in dilute acid-catalyzed hydrolysis of cellulose. Ind Eng Chem Res 1992;31(1):94e100. [29] Teeri TT. Crystalline cellulose degradation: new insight into the function of cellobiohydrolases. Trends Biotechnol 1997; 15(5):160e7. [30] Kim K, Hong J. Supercritical CO2 pretreatment of lignocellulose enhances enzymatic cellulose hydrolysis. Bioresour Technol 2001;77(2):139e44.
Available online at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Statistical optimization of operating conditions for supercritical carbon dioxide-based pretreatment of guayule bagasse Narayanan Srinivasan, Lu-Kwang Ju* Department of Chemical and Biomolecular Engineering, The University of Akron, 200 E Buchtel Commons, Akron, OH 44325-3906, USA
article info
abstract
Article history:
A central composite design (CCD) was used to find the optimal temperature, pressure,
Received 1 June 2011
moisture and duration for the supercritical CO2-based pretreatment of guayule, a desert
Received in revised form
shrub for commercial production of hypoallergenic latex. The pretreatment involved:
10 April 2012
adding water to the biomass, raising system temperature, pressurizing with supercritical
Accepted 5 September 2012
CO2, holding the system for a period of time, and exploding the biomass by rapidly opening
Available online 6 October 2012
the vent valve. The pretreated biomass was then hydrolyzed at 30 C for 72 h in an enzyme solution with 5% (w/v) solid loading. The yields of released glucose and pentose were
Keywords:
determined and used as the CCD response variables. Statistical analysis of results led to the
Guayule
following recommended condition: 175 C, 26.2 MPa (3800 psi), 60% moisture and 30 min.
Supercritical CO2
The corresponding glucose and pentose yields were 56% and 61%. X-ray diffraction analysis
Biomass pretreatment
was done to get the crystallinity index of the bagasse before and after pretreatment.
Scanning electron microscope
Scanning electron microscopy was also used to examine the structural changes caused by
X-ray diffraction
pretreatment. These characterizations indicated that at milder conditions, the pretreat-
Back pressure
ment exposed cellulose and hemicellulose for subsequent enzyme attack; at harsher conditions, the pretreatment destroyed cellulose crystallinity and gave higher glucose yields. To isolate the explosion effect (caused by instantaneous pressure drop, DP) from the reaction (pressure) effect, a series of experiments was made with the bagasse pretreated at the same condition (27.6 MPa) but exploded against different back pressures. The explosion needed to be severe enough (with DP > 17 MPa) to give a sugar yield of over 50%. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Using lignocellulosic biomass as biorefinery feedstock for production of ethanol [1], butanol [2,3], and other chemicals [4,5] via fermentation typically requires the polymeric carbohydrates (cellulose and hemicellulose) to be broken down to small sugar molecules (e.g., glucose and xylose) prior to the fermentation. To enhance the saccharification, particularly if by the gentler enzyme hydrolysis, the biomass needs to be
pretreated to loosen the structure and increase the accessibility and susceptibility of cellulose and hemicellulose [6,7]. In our earlier study a supercritical CO2-based method was evaluated for its feasibility to pretreat waste guayule biomass (bagasse) [8]. Guayule (Parthenium argentatum Gray) [9,10] has attracted considerable interest due to its use in commercial production of hypoallergenic latex and high-quality, multipurpose resins [11e14]. However, latex and resins account for no more than 20% of the guayule dry weight. The rest of the
* Corresponding author. Tel.: þ1 330 972 7252; fax: þ1 330 972 5856. E-mail addresses: [email protected], [email protected] (L.-K. Ju). 0961-9534/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biombioe.2012.09.009
452
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Nomenclature CCD CrI D b D DNS ESEM FPU Iam I002 M
central composite design crystallinity index, as described in Eq. (1) duration of supercritical pretreatment, min dimensionless pretreatment duration in Eqs. (5) and (6), ¼(D 30)/10 dinitrosalicylic acid Environmental Scanning Electron Microscope filter paper unit, a measure of cellulase enzyme activity X-ray diffraction intensity at 2q ¼ 18.7 X-ray diffraction intensity at 2q ¼ 22.6 moisture, weight % of water in bagasse
plant material, already chopped, milled and extracted [15], is potentially a good source of waste biomass for biorefinery. The supercritical CO2-based pretreatment method used in our earlier study involved the following steps: adding a small amount of water to the bagasse loaded in the reactor, raising the system temperature, pressurizing using supercritical CO2, holding the system for a period of time, and exploding the bagasse by rapidly opening the valve to release the pressure [8]. The pretreated biomass was then subjected to enzyme hydrolysis at 30 C for 72 h and the yields of released glucose and total reducing sugars measured. The results confirmed the feasibility of the supercritical CO2-based method in preparing guayule bagasse for the subsequent enzymatic hydrolysis. Unlike the commonly used dilute acid pretreatment, the supercritical method did not incur appreciable loss/ hydrolysis of cellulose and hemicellulose and did not require neutralization or water wash of the pretreated biomass. Conceptually four operation factors may be important to the performance of the supercritical pretreatment. These factors are: temperature, pressure, moisture content of biomass, and the pretreatment (reaction) duration. In the previous feasibility study the pretreatment was only done at 6 conditions [8]. In this study the pretreatment was done at 26 conditions, chosen according to the factorial design, and the response surface analysis [16,17] was performed to determine the optimal pretreatment condition. In addition to the sugar yields from enzyme hydrolysis of pretreated bagasse, X-ray diffraction (XRD) analysis was used to assess the effect of pretreatment on the crystallinity of bagasse pretreated under different conditions, similar to the previous work of Zheng et al. [18]. Lowering cellulose crystallinity is known to enhance the performance of enzyme hydrolysis [19,20]. Scanning electron microscopy (SEM) was also used to examine the changes of bagasse structure caused by the pretreatment. Observing the plant material structures such as lignin sheath and cellulose and hemicellulose fibers with SEM had been reported in the literature [21]. Furthermore, the “explosion” (rapid pressure release) effect was shown important to the outcome of the supercritical pretreatment process in our previous study [8]. Hypothetically, larger amounts of CO2 would enter into the bagasse structure under the supercritical conditions, with the penetration being assisted by the water absorbed in bagasse [18,22]. When the vent
c M P b P DP SAS SEM T b T XRD
dimensionless moisture in Eqs. (5) and (6), ¼(M 60)/10 pressure, MPa dimensionless pressure in Eqs. (5) and (6), ¼( p 27.6)/3.45 pressure drop between the reaction pressure and the venting pressure, MPa Statistical Analysis System scanning electron microscopy temperature, C dimensionless temperature in Eqs. (5) and (6), ¼(T 165)/35 X-ray diffraction
valve is rapidly opened, CO2 expands and rushes out of the biomass, rupturing the bagasse structure and exposing the cellulose and hemicellulose for subsequent enzyme hydrolysis. Accordingly, the pretreatment effectiveness was expected to be affected by the magnitude of the pressure drop (DP) between the reaction pressure and the venting pressure employed. Nonetheless, the magnitude of DP needed for an adequate “explosion” effect was unknown. To study this latter factor in isolation, a set of experiments were done by using the same pressure in the supercritical reactor but different back pressures in the vent vessel (into which the reactor fluid was vented). In real production processes the CO2 collected in the vent vessel would be repressurized and used for the supercritical reactor. A lower DP (i.e., a higher P in the vent vessel) may be more economic in minimizing the energy required for repressurization. The objectives of this work were to optimize the pretreatment conditions in the reactor and the (back) pressure in the vent vessel, and to examine the effects of supercritical pretreatment on the crystallinity and structures of guayule bagasse (and their correlation with the outcome of subsequent enzyme hydrolysis).
2.
Materials and methods
2.1.
Materials
Guayule plant material was provided by Yulex Corporation (Carlsbad, CA). The harvested shrub was chopped, the leaf stream removed, and the branches wet milled for collection of latex as the primary product. The remaining biomass, referred to as “bagasse”, could then be extracted for resin and/or remaining rubber. The bagasse that had been extracted for resin and rubber by the supercritical extraction procedures [15] at Yulex Corporation was transported to our lab for use in this optimization study. The bagasse was in the form of rod-shaped dry powders of about 100e300 mm in diameter and 300e2000 mm in length. The bagasse sample had the following composition [8]: 21% (3%) cellulose, 13% (1%) hemicellulose, 54% (2%) acid insoluble materials, 0.7% (0.1%) acid soluble lignin, and 11% (3%) other (unaccounted for) materials. The enzyme solution (Spezyme CP) used in the study for enzyme hydrolysis of pretreated bagasse was obtained from
453
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Genencor. The enzyme solution was analyzed to have 32.3 FPU/mL of cellulase activity, 26.7 IU/mL of b-glucosidase activity, and 12.4 U/mL (¼208 nkat/mL) of xylanase activity.
Table 1 e Code values of variables e temperature, pressure, moisture, and pretreatment duration e used in the central composite design.
2.2.
Methods
Coding level
2.2.1.
Supercritical CO2-based pretreatment
Fifteen (15) grams of bagasse (moisture content predetermined, 4.4%e6.7%) were placed in a reactor (500 mL) along with water (the added amount calculated to give the investigated water content). The reactor wall was fully wrapped with heating coils containing circulating oil that had been preheated to the reaction temperature. The closed reactor was held at a constant pressure for a particular duration by pumping CO2 into the reactor using a syringe pump (Model 260D, ISCO Inc., Lincoln, NE). The reactor fluid was then vented rapidly, creating the fast pressure drop for the explosion effect. Thus pretreated solids were collected from the reactor for further evaluation and characterization.
2.2.2.
Enzymatic hydrolysis
The hydrolysis was done with 5 g (on the dry basis) pretreated bagasse in 100 mL citrate buffer (pH 4.8, 0.05 M sodium citrateecitric acid) containing 5 mg sodium azide and the following enzyme activities: 0.92 FPU/mL, 0.76 bglucosidase IU/mL and 0.35 xylanase U/mL (prepared from the Spezyme CP solution). The experiments were carried out at 50 C in 125-mL conical flasks shaken at 200 rpm (Model 4730 orbital shaker, Queue Systems, Inc., Parkersburg, WV). Samples were taken periodically from each flask for analyses of total reducing sugars and glucose.
2.2.3. Optimization of pretreatment temperature, pressure, moisture and duration Central composite design (CCD) was used to generate 26 treatment combinations, with temperature, pressure, moisture and duration of hydrolysis as independent variables and the yields of glucose and non-glucose reducing sugars (predominantly pentose) from enzymatic hydrolysis as the response variables in two separate statistical analyses. Table 1 shows the matrix of experimental design, with coded variables containing star points (a ¼ 2) and two replicates of the central point. The experimental data were analyzed by Statistical Analysis System (SAS, package version 9.1) and Minitab 5.1.
2.2.4.
Back pressure study
For the study of back pressure effect, all the bagasse samples were pretreated under the same operating conditions, i.e., 27.6 MPa (4000 psi), 165 C, 60% moisture and 30 min reaction time, but the supercritical fluid in the reactor was allowed to explode into a vessel maintained at different pressures using an automated back pressure regulator (Thar Technologies, Inc., Pittsburgh, PA). The bagasse pretreated was similarly subjected to enzyme hydrolysis for sugar yields and the XRD and SEM characterizations.
2.2.5.
Analytical techniques
Cellulose and hemicellulose contents in the bagasse samples were measured using the standard procedures prescribed by National Renewable Energy Laboratory (NREL; Golden, CO) of
Temperature, C
2 1 0 1 2
95 130 165 200 235
Pressure, MPa (psia) 20.7 24.1 27.6 31.0 34.5
(3000) (3500) (4000) (4500) (5000)
Moisture, %
Duration, min
40 50 60 70 80
10 20 30 40 50
Note: a The values given in parentheses are pressures in the unit of psi.
the US Department of Energy [23e26]. The contents of acid insoluble materials (including lignin) and acid soluble lignin, respectively, were also obtained from the above procedures. A Philips 3100 generator XRD powder diffractometer (Almelo, The Netherlands) fitted with Cu-Ka lamp at 40 kV and 35 mA was used to get the XRD pattern of the (untreated or pretreated) bagasse for a measure of the crystallinity index. All samples were scanned from 2q ¼ 2 e60 at a stepping rate of 0.02 /s. The crystallinity index (CrI) was estimated [18] as CrI ¼
I002 Iam 100 I002
(1)
where I002 and Iam are the diffraction intensities at 2q ¼ 22.6 and 18.7 , respectively. An FEI Quanta 200 Environmental Scanning Electron Microscope (ESEM, FEI Corporation, Brno, Czech Republic) at 25 kV and spot size 40 was used to examine the bagasse structures before and after different pretreatment conditions. A small amount of the sample was mounted on a stainless steel or aluminum stub (1 cm in diameter) with two sided adhesive tape. The sample was sputter coated with silver particles for 1 min and then viewed under vacuum. The total reducing sugar concentration was measured by the dinitrosalicylic acid (DNS) method, based on the color formation of DNS reagent when heated with reducing sugars [27]. The glucose concentration was measured by using the PGO (peroxidase and glucose oxidase) enzymes test kit (Sigma P7119). The glucose yield from guayule bagasse, going through both pretreatment and enzyme hydrolysis, was estimated by the following equations: Glucose yield ¼
Glucose obtained in hydrolysate Total glucose releasable from bagasse
Glucose in hydrolysate ¼Measured glucose concentration 0:1 ðLÞ Glucose releasable ¼ ð% cellulose in bagasseÞ ¼ 5ðgÞ 21%
180 162
180 162
(2) g L ð3Þ
(4)
The numbers given above corresponded to the use of 5 g biomass in 100 mL enzyme solution for the hydrolysis. It should be noted that the supercritical CO2-based pretreatment was predominantly physical and did not cause
454
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
detectable loss of solids. The value 180/162 was included to account for the addition of water as a result of hydrolysis. Similar equations were used to estimate the overall yield of reducing sugars and the yield of pentose. For the latter the pentose in hydrolyzate was estimated by subtracting the moles of glucose from moles of reducing sugars and then multiplying 150 (pentose molecular weight). The maximum pentose releasable was estimated by multiplying (% hemicellulose in bagasse) by the factor 150/132, which accounts for addition of water in the hemicellulose hydrolysis. The yields obtained with these equations were used as the response variables in the optimization of operating conditions for the supercritical pretreatment.
3.
Results
3.1.
Optimization of pretreatment process parameters
Enzyme hydrolysis of the pretreated bagasse showed the typical profiles of rapid initial production of reducing sugars and glucose, which approached the plateau sugar concentrations by about 24 h (data not shown, [8]). To ensure completion of the corresponding hydrolysis achievable for various samples, the sugar yields at 72 h were used as the response variables in the statistical CCD analysis. The 4 principal factors affecting the supercritical CO2 pretreatment process, i.e., temperature (T ), pressure (P), moisture (M ) and duration (D) of reaction, were analyzed for determination of the optimum conditions for maximal yields of glucose and pentose (non-glucose reducing sugars). All of the results obtained in the experimental design are summarized in Table 2. Table 3 shows the p values obtained from statistical analysis of the experimental data, for the factors with p < 0.075. (This 0.075 p value was chosen to allow inclusion of the P effects.) Minitab 5.1 statistical software was used to obtain the regression equations. The corresponding regression equations for glucose and pentose yields (%) are given as follows: b þ 0:68 P b þ 1:75 c b þ 0:32 P b Glucose yield ¼ 61:2 þ 2:89 T M þ 1:57 D
2
2
b þ 1:73 T bP b þ 1:06 T bc þ 0:83 D M 2 b þ 1:28 c MD r ¼ 0:92 (5)
2
b þ 0:71 P b þ 1:29 c b 0:47 P b Pentose yield ¼ 55:7 þ 1:79 T M þ 1:13 D 2 2 b 0:19 T bP b þ 0:44 T bc b r ¼ 0:87 0:09 D M þ 0:31 c MD (6) b T T0 =DT ¼ T 165=35Þ; Pð¼ b P 27:6=3:45Þ; Tð¼ c b D 30=10Þ Mð¼ M 60=10Þ;and Dð¼ in the above equations are the coded dimensionless parameters (Table 1), while the original operating conditions have the following units: T e C, P e MPa, M e wt% (water in bagasse), and D e min. The optimal conditions and the sugar yields at the optimal conditions obtained from the statistical analysis are given in Table 4. The two sets of conditions (one for maximal glucose yield, the other for maximal pentose yield)
did not differ very much. Accordingly, the following pretreatment conditions were recommended: T ¼ 175 C, P ¼ 26.2 MPa (3800 psi), M ¼ 60% and D ¼ 30 min. As shown by the p values (Table 3), the pressure effects b and P b2 terms having p values were less significant, with both P bP b term had smaller p close to or larger than 0.05. Only the T b 2 Þ and values. There were other significant nonlinear ð D bc b in addition to T b P) b effects (with interactive ( T M and c M D, p < 0.05). Nonetheless, it should be noted that there needs to be certain critical pressure to effect the pretreatment (by explosion and mild hydrolysis due to the acidity of CO2 in water) although the effect of pressure was less significant within the range studied in this work (i.e., 20.7e34.5 MPa). The optimal conditions given in Table 4 were not at the highest T or D studied, suggesting that too high temperature or too long reaction duration was not beneficial to sugar production (presumably due to degradation of sugars to other derivatives). Mok et al. [28] observed similar trends in dilute acid-catalyzed hydrolysis of cellulose. They attributed the lower glucose yield to the formation of certain oligomers that could no longer be hydrolyzed to glucose. It is also well known that extended hydrolysis could result in the formation of furfurals from pentose. (The furfurals could be inhibitory to the fermentation processes using hydrolyzate as the sugar source.) The lower temperature optimum for pentose yield (170 C) than that for glucose yield (179 C, Table 4) might be reflecting the higher susceptibility of pentose to degradation at high temperatures. While the CCD experiments allowed the identification of optimal operating conditions, the regression equations obtained were purely empirical. More mechanistic models and experimental designs are necessary for gaining more fundamental understanding of the pretreatment mechanisms.
3.2.
Crystallinity index
The guayule bagasse samples pretreated under different conditions were analyzed for X-ray diffraction patterns. Accordingly, the crystallinity indices (CrI) of these samples were calculated as described in the Analytical techniques section. The CrI values are plotted in Fig. 1 against the % glucose yields after both pretreatment and enzyme hydrolysis. For comparison the data point for enzyme hydrolysis of the bagasse sample that was not pretreated is also included in Fig. 1. The untreated bagasse had a CrI of 38.2 and the glucose yield from the untreated bagasse was just 26%. The CrI of some pretreated bagasse samples were higher than 38.2 while others were lower. For the samples with increased CrI, the pretreatment conditions might be strong enough to break down and/or remove the outer covering of lignin and pectin around the cellulose fibers but not harsh enough to significantly destroy the crystallinity of exposed cellulose. As the pretreatment conditions became increasingly harsher, the pretreatment broke down more of the crystalline cellulose apart from removing the lignin sheath and the CrI dropped below that of the untreated bagasse. All of the pretreatment conditions resulted in much higher glucose yields (>42%), as compared to the 26% glucose yield from the untreated sample. Even the milder pretreatments
455
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
Table 2 e Glucose and pentose yields of the CCD experiments, showing the 26 pretreatment conditions. Temperature ( C)
Pressure (MPa)
Moisture (%)
Duration (min)
24.1 24.1 24.1 24.1 31.0 31.0 31.0 31.0 24.1 24.1 24.1 24.1 31.0 31.0 31.0 31.0 27.6 27.6 27.6 20.7 34.5 27.6 27.6 27.6 27.6 27.6
50 50 70 70 50 50 70 70 50 50 70 70 50 50 70 70 60 60 60 60 60 40 80 60 60 60
20 40 20 40 20 40 20 40 20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 10 50 30
130 130 130 130 130 130 130 130 200 200 200 200 200 200 200 200 95 165 235 165 165 165 165 165 165 165
Glucose yield (%) 48 51 51 47 48 48 47 52 55 49 53 57 46 55 54 49 42 56 46 52 45 53 51 44 54 54
1.6 1.6 1.4 0.6 2.3 0.5 2.7 3.4 0.6 2.1 1.5 0.4 0.7 1.9 1.2 4.3 3.0 1.6 2.1 4.8 4.3 2.7 1.4 4.6 1.9 0.7
Pentose yield (%) 54 56 57 52 54 53 53 58 59 55 57 60 52 54 52 55 47 61 50 57 51 60 56 49 62 61
0.5 1.4 2.1 0.3 2.2 2.3 2.8 2.2 1.4 0.8 1.6 2.2 1.9 2.0 2.6 5.2 0.9 0.9 3.7 3.7 0.9 0.6 2.2 1.0 1.0 2.9
Notes: The bagasse pretreated under each condition was subjected to enzyme hydrolysis in duplicate flasks. Periodical samples were taken from each flask for triplicate analyses of total sugars and glucose concentrations. The average glucose yield and pentose yield were reported with standard deviations (from the 2 3 analysis results) for the final samples taken at 72 h of enzyme hydrolysis.
(giving higher CrI) exposed more cellulose for the enzyme hydrolysis. As the harsher pretreatment caused more destruction of cellulose crystallinity, the CrI decreased and the glucose yields further increased (Fig. 1). This finding is consistent with the fact that the cellulase enzymes hydrolyze amorphous cellulose faster than crystalline cellulose [29]. On the other hand, there was no apparent correlation between the pentose yield and CrI (not shown). Tentatively, the following best-fit correlation was found between the glucose yield (%) and CrI for the supercritical CO2 pretreated guayule bagasse: CrI ¼ 77:6 0:8,ðGlucose yieldÞ
ðr2 Þ ¼ 0:84
(7)
Table 3 e The p-values of terms chosen in the regression analysis (with p < 0.075). Factors T P M D P2 D2 TP TM MD
p (glucose yield)
p (pentose yield)
0.000 0.047 0.000 0.000 0.049 0.018 0.000 0.015 0.005
0.000 0.061 0.011 0.019 0.055 0.034 0.021 0.007 0.023
3.3. SEM-revealed structural changes of bagasse due to pretreatment The bagasse samples before and after pretreatment were observed using SEM to probe for changes in the bagasse structure. The differences observed are shown in Fig. 2(a)e(d). From the SEM pictures of the raw bagasse in Fig. 2(a), it is clearly seen that there was a continuous sheath of lignin covering the cellulose fibers. The fiber structures can be seen underneath the lignin sheath. In the pretreated bagasse (Fig. 2(b)e(d)), the lignin sheath was broken and the exposed cellulose fibers are directly visible. The supercritical pretreatment was effective in causing physical damages to the bagasse and in removing the lignin sheath around the cellulose fibers. Also, some holes appeared in the fibers of pretreated bagasse. The size of the holes ranged roughly from 2 to 5 mm. The holes are marked in the figures with red circles. The holes were probably due to the rupture of the plant structure
Table 4 e Optimum pretreatment conditions for maximum glucose and pentose yields. Operating condition
Temperature ( C) Pressure (MPa) Moisture (%) Duration (min) Yield (%)
Glucose
Pentose
178.8 26.1 57.6 32.3 55.9
170.4 26.4 61.3 34.7 61.5
456
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
during the rapid explosion. Nonetheless, the extents of hole generation in the pretreated bagasse could not be effectively quantified and correlated with the sugar yields obtained under different pretreatment conditions. All of the pretreated bagasse (at different conditions) had holes in the plant structure when viewed under the SEM.
3.4.
Fig. 1 e Variation of crystallinity index (CrI) with % yield of glucose from enzyme hydrolysis of guayule bagasse untreated and pretreated at different conditions.
Effect of back pressure
The glucose yields obtained under different back pressures are shown in Fig. 3. All samples were pretreated at the same reaction condition, i.e., 27.6 MPa (4000 psi), 165 C, 60% moisture and 30 min reaction time, but were exploded against different back pressures. X-ray diffraction analysis was also done to obtain the CrI values of these bagasse samples, which values are shown in Fig. 3 too. The glucose yield decreased with increasing back pressure particularly when the back pressure was above 10.3 MPa. And, the trend of decreasing glucose yield corresponded well with the trend of increasing
Fig. 2 e SEM pictures: (a) untreated guayule bagasse; and bagasse pretreated at (b) 200 C, 24.1 MPa, 70% moisture, 20 min; (c) 165 C, 27.6 MPa, 60% moisture, 30 min; (d) 130 C, 24.1 MPa, 50% moisture, 40 min.
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
457
a
Fig. 3 e Effects of back pressure on glucose yield and crystallinity index of guayule bagasse pretreated by the supercritical CO2-based process.
CrI. Accordingly, certain magnitude of instantaneous pressure drop (DP), in this case, a drop from 27.6 to 10.3 MPa, was necessary for achieving a more optimal glucose yield, presumably by creating severe enough explosive damages to the biomass structure. There are two potentially important mechanisms involved in the supercritical CO2-based pretreatment; they are the physical explosion effect and the chemical hydrolysis effect (due to the acidity provided by the CO2 in water) [8,18,30]. Conceptually the explosion effect is determined more by the magnitude of instantaneous pressure drop DP (instead of the reaction pressure P), while the hydrolysis effect may be more influenced by the reaction P. In an attempt to differentiate the effects of DP and P, the variations of glucose and pentose yields with DP are shown in Fig. 4(a) and (b), respectively, for two types of pretreated samples (and the untreated bagasse). One type of samples was from the back pressure study. These samples were subjected to the same reaction/hydrolysis condition. Therefore, the different sugar yields at different DP could be attributed to the different explosion effects, to expose the cellulose and hemicellulose for enzyme hydrolysis and to lower the crystallinity of cellulose for easier hydrolysis. The other type of samples was pretreated at different reaction P (and the same T, M and D as in the back pressure study) but all were exploded to the atmosphere (i.e., DP and P were essentially the same, DP ¼ P 0.1 MPa). For these samples the different sugar yields could result from different explosion (DP) and/or hydrolysis (P) effects. As shown in Fig. 4(a) and (b), the glucose and pentose yields had similar dependency on DP, although the data were rather scattered. The sugar yields increased with DP up to about 21 MPa, and then plateaued at higher DP (seemingly with a slight decreasing trend at DP >about 27 MPa for glucose yield and about 24 MPa for pentose yield). As hemicellulose is more susceptible to dilute acid hydrolysis than cellulose is, the similar trends of glucose and pentose yields suggested that the main DP effect was to alter the extent of exposing the cellulose and hemicellulose for enzyme hydrolysis (instead of affecting the chemical hydrolysis occurring during the pretreatment).
b
Fig. 4 e Effects of pressure drop (DP) on (a) glucose yield and (b) pentose yield shown with the untreated bagasse (Raw) and two types of pretreated samples: BP e back pressure study (all pretreated at 27.6 MPa, 165 C, 60% moisture and 30 min reaction duration, and then exploded against different back pressures); ATMP e all samples exploded to the atmospheric pressure and pretreated at the same temperature, moisture and duration but different pressures (P).
4.
Conclusion
Results of the CCD experimental design indicated separate but close enough optimal supercritical CO2 pretreatment conditions for maximal glucose and pentose yields (i.e., 56% and 61%, respectively). Accordingly, the following pretreatment conditions were recommended: T ¼ 175 C, P ¼ 26.2 MPa (3800 psi), M ¼ 60% and D ¼ 30 min. The crystallinity indices estimated from the XRD analysis suggested that the main effect at milder pretreatment conditions was to expose the cellulose (and hemicellulose) for subsequent enzyme attack and that, with severer pretreatment conditions, cellulose crystallinity was also reduced to give correspondingly higher glucose yield after enzyme hydrolysis. The SEM results lent further support to the above framework of understanding, and
458
b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 5 1 e4 5 8
microscopic holes were visible on the fibers exposed by the pretreatment. The back pressure study results confirmed the importance of explosion effect on the pretreatment outcome and the explosion needed to be severe enough (with DP > 17 MPa, corresponding to the drop from a 27.6 MPa reaction pressure to a 10.3 MPa back pressure in this study) to give a sugar yield of over 50%.
Acknowledgment This work was supported by Yulex Corporation (Carlsbad, CA) and the U.S. Department of Agriculture under the Biomass Research and Development Initiative (grant number 68-3A757-610). Dr. J. Richard Elliott (Department of Chemical and Biomolecular Engineering, The University of Akron) allowed us to use freely the equipment available at the Ohio Supercritical Fluid Consortium. Dr. Katrina Cornish (Senior VP, R & D, Yulex Corporation) provided the bagasse samples and the information for supercritical extraction of resins and rubber. We are also grateful to Mr. Thomas P. Quick (Department of Geology and Environmental Science, The University of Akron) for his help with the X-ray diffraction and SEM studies.
references
[1] Sun Y, Cheng J. Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresour Technol 2002;83(1): 1e11. [2] Huang W-C, Ramey DE, Yang S- T. Continuous production of butanol by Clostridium acetobutylicum immobilized in a fibrous bed bioreactor. Appl Biochem Biotechnol 2004;113-116: 887e98. [3] Ezeji T, Qureshi N, Blaschek HP. Butanol production from agricultural residues: impact of degradation products on Clostridium beijerinckii growth and butanol fermentation. Biotechnol Bioeng 2007;97(6):1460e9. [4] Silva C, Roberto IC. Optimization of xylitol production by Candida guilliermondii FTI 20037 using response surface methodology. Process Biochem 2001;36(11):1119e24. [5] Matsumoto Ki, Kobayashi H, Ikeda K, Komanoya T, Fukuoka A, Taguchi S. Chemo-microbial conversion of cellulose into polyhydroxybutyrate through rutheniumcatalyzed hydrolysis of cellulose into glucose. Bioresour Technol 2011;102(3):3564e7. [6] Bhat M, Bhat S. Cellulose degrading enzymes and their potential industrial applications. Biotechnol Adv 1997; 15(3e4):583e620. [7] Chandra R, Bura R, Mabee W, Berlin A, Pan X, Saddler J. Substrate pretreatment: the key to effective enzymatic hydrolysis of lignocellulosics? Adv Biochem Eng Biotechnol 2007;108:67e93. [8] Srinivasan N, Ju L-K. Pretreatment of guayule biomass using supercritical carbon dioxide-based method. Bioresour Technol 2010;101(24):9785e91. [9] Cornish K. Hypoallergenic natural rubber products from Parthenum argentatum (gray) and other non-Hevea brasiliensis species. US Patent 5,717,050; 1998. [10] Schloman Jr WW, Wyzgoski F, McIntyre D, Cornish K, Siler DJ. Characterization and performance testing of guayule latex. Rubber Chem Technol 1996;69(2):215e22.
[11] Thames SF, Schuman TP, Reichel LW, Purvis WA, Poole PW. Guayule coproducts: emerging technology in industrial, marine, and peelable coatings. Int SAMPE Symp Exhibition 1996;41(2):1177e90. [12] Bultman JD, Gilbertson RL, Adaskaveg J, Amburgey TL, Parikh SV, Bailey CA. The efficacy of guayule resin as a pesticide. Bioresour Technol 1991;35(2):197e201. [13] Gutierrez C, Gonzalez-Coloma A, Hoffmann JJ. Antifeedant properties of natural products from Parthenium argentatum, P. argentatum, P. tomentosum (Asteraceae) and Castela emoryi (Simaroubeaceae) against Reticulitermes flavipes. Ind Crops Prod 1999;10(1):35e40. [14] Nakayama FS, Vinyard SH, Chow P, Bajwa DS, Youngquist JA, Muehl JH, et al. Guayule as a wood preservative. Ind Crops Prod 2001;14(2):105e11. [15] Cornish K, Martin JA, Marentis RT, Plamthottam S. Extraction and fractionation of biopolymers and resins from plant materials. US Patent 7,259,231; 2006. [16] Heck JX, Floˆres SH, Hertz PF, Ayub MAZ. Statistical optimization of thermo-tolerant xylanase activity from Amazon isolated Bacillus circulans on solid-state cultivation. Bioresour Technol 2006;97(15):1902e6. [17] Couto SR, Moldes D, Sanroma´n MSA. Optimum stability conditions of pH and temperature for ligninase and manganese-dependent peroxidase from Phanerochaete chrysosporium. Application to in vitro decolorization of poly R478 by MnP. World J Microbiol Biotechnol 2006;22(6):607e12. [18] Zheng Y, Lin HM, Wen J, Cao N, Yu X, Tsao GT. Supercritical carbon dioxide explosion as a pretreatment for cellulose hydrolysis. Biotechnol Lett 1995;17(8):845e50. [19] Yoshida M, Liu Y, Uchida S, Kawarada K, Ukagami Y, Ichinose H, et al. Effects of cellulose crystallinity, hemicellulose, and lignin on the enzymatic hydrolysis of Miscanthus sinensis to monosaccharides. Biosci Biotechnol Biochem 2008;72(3):805e10. [20] Laureano-Perez L, Teymouri F, Alizadeh H, Dale BE. Understanding factors that limit enzymatic hydrolysis of biomass. Appl Biochem Biotechnol 2005;124(1):1081e99. [21] Jones J, Semrau K. Wood hydrolysis for ethanol production? Previous experience and the economics of selected processes. Biomass (London) 1984;5(2):109e35. [22] Zheng Y, Lin H, Tsao G. Pretreatment for cellulose hydrolysis by carbon dioxide explosion. Biotechnol Prog 1998;14(6): 890e6. [23] Ehrman T. Standard method for determination of total solids in biomass. Laboratory analytical procedures 001. Golden, CO: National Renewable Energy Laboratory; 1994. [24] Ruiz R, Ehrman T. Determination of carbohydrates in biomass by high performance liquid chromatography. Laboratory analytical procedure 002. Golden, CO: National Renewable Energy Laboratory; 1996. [25] Templeton D, Ehrman T. Determination of acid-insoluble lignin in biomass. Laboratory analytical procedure 003. Golden, CO: National Renewable Energy Laboratory; 1995. [26] Ehrman T. Determination of acid-soluble lignin in biomass. Laboratory analytical procedure 004. Golden, CO: National Renewable Energy Laboratory; 1996. [27] Miller G. Use of dinitrosalicylic acid reagent for determination of reducing sugar. Anal Chem 1959;31(3):426e8. [28] Mok WS, Antal Jr MJ, Varhegyi G. Productive and parasitic pathways in dilute acid-catalyzed hydrolysis of cellulose. Ind Eng Chem Res 1992;31(1):94e100. [29] Teeri TT. Crystalline cellulose degradation: new insight into the function of cellobiohydrolases. Trends Biotechnol 1997; 15(5):160e7. [30] Kim K, Hong J. Supercritical CO2 pretreatment of lignocellulose enhances enzymatic cellulose hydrolysis. Bioresour Technol 2001;77(2):139e44.