UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fermentable sugars from pyrolysis oil : extraction and hydrolysis of levoglucosan Bennett, Nicole Marie 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-0026a.pdf [ 8.51MB ]
Metadata
JSON: 831-1.0059037.json
JSON-LD: 831-1.0059037-ld.json
RDF/XML (Pretty): 831-1.0059037-rdf.xml
RDF/JSON: 831-1.0059037-rdf.json
Turtle: 831-1.0059037-turtle.txt
N-Triples: 831-1.0059037-rdf-ntriples.txt
Original Record: 831-1.0059037-source.json
Full Text
831-1.0059037-fulltext.txt
Citation
831-1.0059037.ris

Full Text

F E R M E N T A B L E SUGARS F R O M PYROLYSIS OIL: EXTRACTION A N D HYDROLYSIS OF L E V O G L U C O S A N by NICOLE MARIE BENNETT B.Sc.(ENG), The University of Guelph, 2003 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH C O L U M B I A December 2006 © Nicole Marie Bennett, 2006 A B S T R A C T Fermentable sugar obtained from lignocellulosic material exhibits great potential as a renewable feedstock for the production of bio-ethanol. One potentially viable source of fermentable sugars is pyrolysis oil, also commonly called bio-oil. Depending on the type of lignocellulosic material used and the operating conditions for the process of pyrolysis, bio-oil can contain up to 33 wt% (Li and Zhang, 2004) of 1,6-anhydro-P-D-glucopyranose (levoglucosan, LG), an anhydrosugar that can be readily hydrolyzed to glucose. This research investigated and improved the extraction of levoglucosan from bio-oil via phase separation, the acid-hydrolysis of the. levoglucosan once in the aqueous phase into glucose, and the subsequent fermentation of the glucose into ethanol using a Saccharomyces cerevisiae yeast strain, T2. Maximum levoglucosan extraction (7.8 wt% of the initial bio-oil) occurred when minimal amounts of water were used; optimally, at the point of phase separation. For the bio-oil used in this study, phase separation occurred when the total water in the solution equalled 41 wt% of the bio-oil. Extraction was improved when the temperature of the bio-oil was raised (34°C) and effective mixing was applied. Hydrolysis yields greater than 100% (based on levoglucosan) occurred at high temperatures (130°C), strong sulphuric acid concentrations (0.5M) and short reactions times (20 minutes). Temperature exhibited the most significant effect on the yield; yet a strong interactive effect between temperature and time due to sugar degradation at temperatures at or above 130°C prohibits the use of higher temperatures. The initial levoglucosan concentration in the aqueous phase also demonstrated a significant effect on the hydrolysis yield. As is, the hydrolysate was too toxic for yeast growth and fermentation due to the presence of inhibitory compounds such as organic acids and sugar degradation products. However, theoretical fermentation yields were made possible for hydrolysate fractions of ii up to 20% by using micro-aerophilic conditions and high yeast inoculums (1 g/L in vial). Fermentation rates of 0.40 g/L hr were observed under these conditions. iii T A B L E OF C O N T E N T S ABSTRACT ii T A B L E OF CONTENTS iv LIST OF T A B L E S viii LIST OF FIGURES . ". xi LIST OF ABBREVIATIONS A N D A C R O N Y M S xiv A C K N O W L E D G M E N T S xv CHAPTER I - INTRODUCTION 1 CHAPTER II - L ITERATURE REVIEW 2 2.1 Demand for bio-fuels 2 2.2 Lignocellulosic material 3 2.3 Pre-treatment of lignocellulosic material 4 2.3.1 Pyrolysis 6 2.4 Characteristics of levoglucosan 9 2.5 Extraction of levoglucosan from bio-oil 11 2.5.1 Factors affecting the extraction of levoglucosan from bio-oil 12 2.6 Hydrolysis of levoglucosan into glucose 13 2.6.1 Acid hydrolysis 14 2.6.1.1 Effects of temperature, time and acid concentration on hydrolysis of levoglucosan 16 2.7 Fermentation of lignocellulosic hydrolysates 17 2.7.1 Fermentation rates, productivity and yields in lignocellulosic hydrolysates 19 2.7.2 Fermentation inhibitors formed during biomass processing 20 2.7.2.1 Removal of inhibitors from lignocellulosic hydrolysates 22 CHAPTER III - R E S E A R C H OBJECTIVES 23 CHAPTER IV - M A T E R I A L S A N D METHODS 24 4.1 Characterization of VTT bio-oil 24 4.1.1 Determining the point of phase separation 25 4.2 The effect of water volume on the fractionation of components in bio-oil 25 4.2.1 Determination of total solids in the aqueous phase 27 4.2.2 Determination of sugar and organic acid content in the aqueous phase 28 4.3 Determining parameter effects of temperature, time and water volume on the extraction of L G into the aqueous phase 28 4.4 Acid-hydrolysis of levoglucosan into glucose 30 4.4.1 Initial investigation into levoglucosan hydrolysis 30 4.4.2 Determining the effects of hydrolysis parameters temperature, time and acid concentration on the conversion of levoglucosan into glucose 31 4.4.3 Determining the effects of initial levoglucosan concentration and acid concentration on hydrolysis yield 34 4.5 Fermentation of aqueous phase 35 4.5.1 Determining fermentability of the aqueous phase 36 4.5.2 Fermentation using micro-aerophilic conditions 38 4.5.3 Fermentation using high inoculation 40 4.6 Analytical 40 4.6.1 Sugar determination assay using HPLC analysis 40 4.6.1.1 Determination of sugar peaks 41 4.6.2 Sugar determination using GC analysis 42 4.6.3 Ethanol determination using GC analysis 44 4.6.4 Factorial experimental design 45 CHAPTER V - RESULTS A N D DISCUSSION 46 5.1 Determining the point of phase separation 46 5.2 Characterization of the aqueous phase in response to varying water volumes 48 5.2.1 Dissolved solids in aqueous phase 51 5.2.1.1 Levoglucosan and glucose in aqueous phase 54 5.2.1.2 Organic acids in aqueous phase 57 5.3 Determining parameter effects on levoglucosan extraction 59 5.4 Hydrolysis of levoglucosan into glucose 64 v 5.5 Determining the effect of time, temperature and acid concentration on levoglucosan hydrolysis into glucose 66 5.5.1 Hydrolysis conditions A 66 5.5.2 Hydrolysis conditions B 70 5.5.3 Hydrolysis conditions C 74 5.5.4 Application of combined severity model 78 5.6 Effects of initial levoglucosan concentration and acid concentration on hydrolysis yields 80 5.7 Fermentation of aqueous extracts 83 5.8 Fermentation techniques 86 5.8.1 Fermentation using micro-aerophilic conditions 86 5.8.2 Fermentation using high inoculation 90 CHAPTER VI - CONCLUSIONS 95 CHAPTER VII - RECOMMENDATIONS / FUTURE W O R K 97 REFERENCES 99 APPENDIX A - C H E M I C A L COMPOSITION OF BIO-OIL 106 APPENDIX B - E X P E R I M E N T A L D A T A 114 B . l Characterization of the aqueous phase in response to varying water volumes 114 B.2 Dissolved solids in aqueous phase 114 B.3 Levoglucosan and glucose in aqueous phase 115 B.4 Organic acids in aqueous phase 116 B.5 Determining parameter effects on levoglucosan extraction 118 B.6 Hydrolysis of levoglucosan into glucose 119 B.7 Determining the effect of time, temperature and acid concentration on levoglucosan hydrolysis into glucose 120 B.7.1 Hydrolysis conditions A 120 B.7.2 Hydrolysis conditions B 121 B.7.3 Hydrolysis conditions C 122 B.8 Application of combined severity model 123 vi B.9 Effects of initial L G concentration and acid concentration on hydrolysis yields 124 B.10 Fermentation of aqueous extracts 124 B . l l Fermentation using micro-aerophilic conditions 128 B. l2 Fermentation with micro-aerophilic and high inoculum 131 LIST OF T A B L E S Table 2.1: Typical composition of bio-oil 8 Table 4.1: Elemental composition and other properties of pine pyrolysis oil 24 Table 4.2: Volumetric ratios used to investigate extraction 26 Table 4.3: Temperature, time and water parameters used for extraction experiment 29 Table 4.4: Combination of parameters for surface response extraction experiment 29 Table 4.5: Hydrolysis conditions A 32 Table 4.6: Hydrolysis conditions B 32 Table 4.7: Hydrolysis conditions C 32 Table 4.8: Combination of parameters of time, temperature and acid concentration for hydrolysis experiments A and B 33 Table 4.9: Combination of parameters of time, temperature and acid concentration for hydrolysis experiment C 33 Table 4.10: Values of levoglucosan concentration and acid concentration assigned to parameter levels 34 Table 4.11: Combination of parameters for factorial designed hydrolysis experiment. ...35 Table 4.12: Fermentation flasks prepared using fractions of aqueous phase and distilled water ranging from 0 to 100% 37 Table 4.13: Fermentation flasks prepared using fractions of aqueous phase and distilled water ranging from 0 to 20% 38 Table 4.14: Fermentation vials containing incremental fractions of hydrolysate for micro-aerophilic fermentation experiment 39 Table 4.15: Sugar retention times for the HPLC 41 Table 5.1: Titration of water into bio-oil to determine the point of phase separation 46 Table 5.2: Acetic and formic acid concentrations for three different aqueous extracts. ...59 Table 5.3: Levoglucosan in the aqueous phase expressed as concentration (g/L) and as weight percent of bio-oil (wt%) in response to combinations of various time, temperature, and water values 60 Table 5.4: Concentrations of levoglucosan and glucose (g/L) after hydrolysis conditions A, also expressed as percent yield 67 viii Table 5.5: Final concentrations of levoglucosan and glucose (g/L) and hydrolysis yield (%) resulting from hydrolysis conditions B, using initial concentrations of 22.2 g/L and 2.7 g/L, respectively 71 Table 5.6: Variance and percent error between predicted and actual data for hydrolysis conditions B using Equation 5.4 74 Table 5.7: Final concentrations of levoglucosan and glucose as well as hydrolysis yield using combinations of hydrolysis conditions C ...75 Table 5.8: Hydrolysis yield as a result of various combinations of levoglucosan and acid concentrations 80 Table 5.9: Ethanol yields of flasks containing incremental fractions of aqueous extract in nutrient rich media 86 Table 5.10: Maximum yeast concentrations, ethanol yields, maximum ethanol concentrations, and productivity values in response to varying hydrolysate fractions in nutrient rich media under 90 Table 5.11: Yeast concentrations, ethanol yields, maximum ethanol concentrations, and productivity rates in response to varying fractions of hydrolysate and nutrient rich media under micro-aerophilic conditions and high inoculum 92 Table 6.1: Summary of optimal conditions for hydrolysis 96 Table A . l : Chemical composition of fresh VTT bio-oil as determined using GC/MS. ..107 Table B . l : Mass and volumes of organic phase and aqueous phase for experiment 5.2.1 114 Table B.2: Mass of crucibles before and after 105°C oven for 2 days using 3 mL sample of aqueous phase for experiment 5.2.1 114 Table B.3: Data used for GC calibration for experiment 5.2.1.1 115 Table B.4: Data for aqueous extracts 1 through 8 for L G and glucose determination in experiment 5.2.1.1 115 Table B.5: Titration data for organic acid calculations for experiment 5.2.1.2 116 Table B.6: Data used in model for determining organic acids for experiment 5.2.1.2....116 Table B.7: Excerpt from table of values used for organic acid model 117 Table B.8: Data for calibration curve for experiment 5.3 118 Table B.9: Data for samples 1 through 16 for experiment 5.3 118 ix Table B.10: Data for calibration curve for experiment 5.4 119 Table B . l 1: Data for samples 1 through 15 for experiment 5.4 119 Table B.12: Data for calibration curve for experiment 5.5.1 120 Table B . l3 : Samples 1 through 16 for hydrolysis conditions A 120 Table B.14: Data for calibration curve for experiment 5.5.2 121 Table B.15: Data for samples 1 through 16 for experiment 5.5.2 121 Table B. l6 : Data for calibration curve for experiment 5.5.3 122 Table B. l7 : Data for samples 1 though 17 for experiment 5.5.3 122 Table B.18: Data for combined severity model 123 Table B. l9 : L G and glucose concentrations for experiment 5.6 124 Table B.20: Data for calibration curve for experiment 5.7 124 Table B.21: Data for fermentation flasks for experiment 5.7 125 Table B.22: Data for calibration curve for flasks of 0% to 20% for experiment 5.7 126 Table B.23: Data for fermentation flasks of 0% to 20% strength for experiment 5.7 126 Table B.24: Data for calibration curve for experiment 5.8.1 128 Table B.25: GC data for fermentation flasks for experiment 5.8.1 128 Table B.26: Data for calibration curve for experiment 5.8.2 131 Table B.27: Data from fermentation vials for experiment 5.8.2 131 x LIST OF F I G U R E S Figure 2.1: Cellulose structure showing the repeating unit, cellobiose [OSU, 2004] 4 Figure 2.2: Pyrolysis as part of a bio-refinery concept. Arrows represent potential pathways 5 Figure 2.3: Structure of 1,6-anhydro-P-D-glucopyranose 9 Figure 2.4: Degradation products formed from glucose during dilute acid hydrolysis [Larsson et al., 1999] 21 Figure 4.1: Experimental set-up for micro-aerophilic conditions 39 Figure 4.2: Chromatograph of sugar standards using HPLC analysis 41 Figure 4.3: HPLC chromatograph illustrating the challenge in isolating levoglucosan peak 42 Figure 4.4: Calibration curve for sugar analysis by GC. Glucose (squares) and levoglucosan (diamonds) and their respective trend lines are shown ...... 43 Figure 4.5: Calibration curve used for ethanol analysis by GC 44 Figure 5.1: Bio-oil after phase separation 48 Figure 5.2: Mass transfer of bio-oil compounds into aqueous phase in response to varying water volumes 49 Figure 5.3: Mass of bio-oil compounds in the organic phase in response to various water volumes 50 Figure 5.4: Density of the aqueous fraction in response to varying water volumes 51 Figure 5.5: Concentration of dissolved solids in the aqueous phase in response to varying volumes of water 52 Figure 5.6: Weight percent of dissolved solids transferred into the aqueous phase in response to varying volumes of water used for phase separation 53 Figure 5.7: Concentration of levoglucosan (dotted line & open squares using the scale on the left) and glucose (solid line & filled diamonds using the scale on the right) in the aqueous phase in response to varying volumes of water used for phase separation. 54 Figure 5.8: Quantity of levoglucosan, expressed as a weight percentage of the bio-oil, transferred into the aqueous phase from the bio-oil in response to varying volume of water used for phase separation ; 55 xi Figure 5.9: Quantity of glucose, expressed as a weight percentage of the bio-oil, transferred into the aqueous phase in response to varying water volumes used for phase separation 56 Figure 5.10: Titration curves for three aqueous extracts prepared using different water to bio-oil volumetric ratios for extraction operation: squares =1:1 (water = 104 wt% of bio-oil), diamonds = 2:1 (water = 185 wt% of bio-oil), triangles = 4:1 (water = 359 wt% of bio-oil). A strong-base, weak-acid model using estimated concentrations of formic and acetic acid is shown as the solid lines 58 Figure 5.11: Correlation between actual and predicted amounts of levoglucosan transferred into the aqueous phase (15 data points) 61 Figure 5.12: Prediction profiles for the parameter effects of time, temperature and water on the amount of levoglucosan transferred into the aqueous phase .' 62 Figure 5.13: Glucose (open symbols) and L G (filled symbols) concentrations for aqueous extracts A (triangles), B (squares) and C (diamonds) using hydrolysis conditions of 120°C and 0.5M H 2 S0 4 over a period of 60 minutes 64 Figure 5.14: Proposed reaction of levoglucosan to glucose during hydrolysis 66 Figure 5.15: Correlation between actual and predicted amounts of levoglucosan hydrolyzed into glucose using hydrolysis conditions A for 15 data points (sample 11, the circled data point was not included in analysis) 68 Figure 5.16: Prediction profiles for hydrolysis conditions 68 Figure 5.17: Correlation between actual and predicted amounts of L G hydrolyzed into glucose using hydrolysis conditions B (16 data points) 72 Figure 5.18: Prediction profiles for hydrolysis conditions B. An optimal combination within the investigated range resulted in a yield of 85% 72 Figure 5.19: Correlation between actual and predicted amounts of L G hydrolyzed into glucose using hydrolysis conditions C (without sample 14, circled) 76 Figure 5.20: Prediction profiles for the effects of time, temperature and acid concentration on hydrolysis yield using hydrolysis conditions C 76 Figure 5.22: Glucose yield (%) as a function of combined severity using hydrolysis conditions A (diamonds), B (squares) and C (triangles) 79 xii Figure 5.23: Correlation between actual and predicted amounts of levoglucosan hydrolyzed into glucose using different initial levoglucosan and acid concentrations 81 Figure 5.24: Profile of hydrolysis yield in response to various initial L G concentrations and acid concentrations 81 Figure 5.25: Effect on ethanol production of incremental fractions of aqueous extract from pine pyrolysis oil ranging from 0 to 100% (volume of extract / total volume) 83 Figure 5.26: Effect on ethanol production of incremental fractions of aqueous extract from pine pyrolysis oil ranging from 0 to 20% (volume extract / total volume) 85 Figure 5.28: Metabolism of furfural into furfuryl alcohol 88 Figure 5.29: Ethanol production rates (g/L hr) in response to increasing fractions of hydrolyzed aqueous extract from pine pyrolysis oil ranging from 0 to 20% (v/v) and micro-aerophilic conditions 89 Figure 5.30: Ethanol concentrations for vials with varying fractions of hydrolyzed aqueous extract using micro-aerophilic conditions and a high yeast inoculum 91 Figure 5.31: Productivity of ethanol per gram of S. cerevisiae T2 yeast in response to incremental fraction of hydrolysate in the fermentation media 94 xiii LIST OF A B B R E V I A T I O N S A N D A C R O N Y M S CERC Clean Energy Research Center CS Combined severity d.w. Dry weight FID Flame ionization detector GC Gas chromatography Glu Glucose HPLC High pressure liquid chromatography L G Levoglucosan NSERC Natural Sciences and Engineering Research Council of Canada Ro Reaction ordinate TTL Transistor-transitor-logic UBC University of British Columbia VTT Technical Research Centre of Finland YPG Yeast peptone glucose xiv A C K N O W L E D G M E N T S Gratitude is expressed to Dr. Sheldon Duff for his patience, encouragement and support; specifically, for his consent to attend and present at a variety of academic conferences. This enhanced the overall experience and provided valuable skills that are not taught in the class room. Thank you to Dr. Steve Helle for his overall expertise and analytical assistance. Many thanks to my committee: Dr. Susan Nesbit, Dr. Dusko Posarac and Dr. Naoko Ellis for their patience and assistance. Appreciation is extended to UBC Department of Chemical and Biological Engineering, CERC and the Pulp and Paper Centre. Thank you to V T T for supplying the bio-oil and Tembec for the yeast. Lastly, a hearty cheers to my lab mates, chums, and biking community for entertaining me during the ride. This work was generously supported from research grants through NSERC and the BIOCAP Foundation of Canada. xv C H A P T E R I - I N T R O D U C T I O N This study focuses on the opportunity to utilize pyrolysis oil as a renewable, viable source of simple carbohydrates used for the production of bio-ethanol fuel. As the demand for ethanol production increases, it is important to encourage sustainable practices by developing diverse processes that can produce valuable resources from what is currently deemed as waste. The economic viability of obtaining fermentable sugars from pyrolysis oil is greater when the process is integrated as part of a bio-refinery concept where other value-added components can be co-produced from the raw biomass material. The process of pyrolysis has the potential to be a zero-waste system as the other two products, volatized organics (syngas) and solid carbon char, can be used for commercial applications and/or to supplement internal energy requirements. Research and development in the field of pyrolysis has led to vast improvements in obtaining high pyrolysis oil yields and, more recently, high concentrations of levoglucosan in this liquid product. Levoglucosan is one of the main carbohydrate compounds formed during the thermo-chemical degradation of cellulose and hemi-cellulose and is the compound of interest for this study. As an anhydrosugar, it can be converted to glucose using acid hydrolysis conditions. The glucose can then be fermented to ethanol using an adapted strain of Saccharomyces cerevisiae. To date, there has been little research specific to the fractionation and subsequent hydrolysis of levoglucosan from pyrolysis oil for the purpose of producing ethanol fuel. This study will investigate the fundamentals of this process, with an attempt to establish favourable conditions to improve extraction and hydrolysis efficiencies. Fermentation techniques previously applied to the fermentation of lignocellulosic hydrolysates will be applied to this process to verify the practicality of the concept. 1 CHAPTER II - LITERATURE REVIEW 2.1 Demand for bio-fuels At the current global consumption rate of crude oil, the end of easily accessible oil will occur within 35-70 years [McLaren, 2005; IEA 2004]. Moreover, rates of oil consumption are expected to increase as a result of recent and rapid economic growth in heavily populated countries, particularly in China and India. Irrespective of the supply situation, the concentration of atmospheric carbon dioxide associated with the use of petroleum and other fossil fuels, is climbing at an unprecedented rate and must be addressed. Risk analysis has proposed that the atmospheric concentration of carbon dioxide should not exceed 500 +/- 50 parts per million (ppm); thereafter, conditions may become irreversible [Pacala and Socolow, 2004]. The most recent reports estimate current concentrations to be at 380 ppm [Azapagic, 2006]. The integration of biomass-derived fuels can significantly and immediately address the concerns of carbon dioxide emissions and depleting fuel reserves. Governments of industrialized nations have recognized these concerns and have established biomass-derived fuel programs as part of their national energy strategies. The U.S. Department of Energy mandated that 30% of all petroleum fuel be replaced with bio-fuel by 2025 [US Department of Energy, 2006]. Similarly, a European Union directive has stated that all diesel and petroleum transportation fuel will have a biomass-derived fraction of 5.75% by 2010 [Klinke, 2004]. These policies will likely lead to dramatic growth in the production of bio-fuels over the next decade. Today, almost all of the ethanol production (90%) is derived from corn or sugar crops [Hamelinck et al., 2005]. The benefits of these energy crops (e-crops) are limited due to agricultural competition (food, feed, land) and environmental stresses associated with their production, such as the increased use of fertilizers and pesticides (which are petroleum derived). Many authors have identified these limitations [von Blottnitz and Curran, 2006; Kim and Dale, 2004; Sun and Cheng, 2002; Wyman, 1999]. 2 For a short term (transitional) period, these conventional production techniques are satisfactory. However, long-term transportation energy demands and other sustainability concerns can only be addressed by diversifying the feedstock to include lignocellulosic materials. Specifically, the most attractive feedstocks are likely to be wasted or underutilized renewable resources such as municipal solid waste, forestry and agricultural residues [Kim and Dale, 2004]. Life-cycle analyses (LCA) are important when developing a long-term bio-fuel production strategy. Tampier (2004) and Lave and coworkers (2000) have reviewed different production strategies using L C A . It has been agreed that lignocellulosic biomass offers a potential source for cheap and abundant fermentable sugars and low-cost ethanol production [Mosier et al., 2005; Badger, 2002; Sun and Cheng, 2002; Olsson and Hahn-Hagerdal, 1996]. With suitable processes, these carbohydrate rich materials can help achieve global long term, large-scale transportation fuel production goals [Hamelinck et al.,2005]. Ultimately, energy production from wasted or underutilized feedstock such as crop residues, grasses, and wood chips, is optimal as this can minimize new resource input and achieve the greatest "well-to-wheel"1 greenhouse gas reduction [IEA, 2003; Wyman, 1999]. If the production process is properly managed, the life-cycle can even be carbon-neutral [von Blottnita and Curran, 2006; Kim and Dale, 2004; Zaldivar et al., 2001]. 2.2 Lignocellulosic material It has been estimated that lignocellulosic material accounts for 50% of the world's biomass [Claasen et al., 1999] and is notably the most abundant polymer on earth [Zaldivar et al., 2001]. The major components of lignocellulose are: cellulose (-45% of dry weight), hemicellulose (-30% of dry weight), and lignin (-25% of dry weight) [Zaldivar et al., 2001]. Cellulose and hemicellulose are polysaccharides that can be hydrolyzed to sugars and eventually fermented into ethanol. Cellulose is a linear, high ' "Well-to-wheel" refers to the complete chain of fuel production, including feedstock production, transport to the refinery, conversion to final fuel, transport to refuelling stations, and final vehicle tailpipe emissions. 3 molecular weight polysaccharide of D-glucopyranose units linked by P(l,4)-glycosidic bonds (Figure 2.1). The dissacharide cellobiose, also seen in Figure 2.1, is the basic repeating unit of cellulose. Cellulose C H j O H H O H CK5OH M O H o \ j | / " H 0 ^ ° ^ ! j ^ H H ^ — C / O H O H C H j O M H OH C H j O H Cellobiose Figure 2.1: Cellulose structure showing the repeating unit, cellobiose [OSU, 2004]. Hemicellulose is a linear and branched heteropolymer containing sugar residues such as xylose, arabinose, mannose, galactose, and glucose. The hemicellulose structure is easier to hydrolyze than cellulose [Olsson and Hahn-Hagerdahl, 1996]. The third component, lignin, is found in plant cell walls to maintain strength and flexibility. Lignin is a large, complex, polyphenolic non-carbohydrate and therefore cannot be hydrolyzed to ethanol [Zaldiver et al., 2001]. The type of lignocellulosic material (i.e. the ratio of the cellulose, hemicellulose and lignin) is important for process design. Ideally, a feedstock of pure cellulose is beneficial for high yields of glucose. Even though proposed feedstocks such as forestry and agriculture residues have higher fractions of lignin they are still advantageous due to their low cost, abundance, and low level of required inputs (labour, land, water) when compared to traditionally used e-crops such as corn. Advances in bio-technology may offer genetically improved feedstock for the future. 2.3 Pre-treatment of lignocellulosic material For lignocellulosic materials to be converted to fermentable sugars, some form of pre-treatment is necessary to access the carbohydrates (as cellulose and hemicellulose) from the plant cell wall. Generally, one of the goals for pre-treatment is to reduce the size of the biomass particles to allow for greater accessibility to the polysaccharides. This may include effort to remove lignin and/or hemicellulose into separate streams. As an extra processing stage, pre-treatment would increase the cost of ethanol production, yet this can 4 be balanced by the improved recovery of sugar or other valuable components. In a review of current technologies, Mosier and coworkers (2005) discussed physical and chemical methods such as A F E X (ammonia fiber/freeze explosion), lime or acid addition, and steam explosion. In conclusion, Mosier stated the expense of the pre-treatments did not justify the improvements and more development is required. Moreover, an alternate direction was suggested as the "ultimate goal" of a pre-treatment process: an "efficient fractionation of lignocellulose into multiple streams, contain(ing) value-added compounds in concentrations that make purification, utilization and recovery economically feasible". Pyrolysis, a pre-treatment technology not reviewed by Mosier, has gained attention from industry, government and academia as an alternative to harnessing energy from biomass; both a pre-treatment in itself and as part of the bio-refinery concept [Holbein et al., 2005; VTT, 2006; Sensoz and Can, 2002]. In a bio-refinery, efficiencies and economics are maximized due to shared resources and co-generation of products. An example of possible process streams from pyrolysis is shown in Figure 2.2. biomass pyrolysis bio-oil char syngas extraction upgrade chemicals/products (i.e. oligosaccharides) combustion turbine engine heat power transport, fuel (i.e. ethanol) i i i Figure 2.2: Pyrolysis as part of a bio-refinery concept. Arrows represent potential pathways. 5 2.3.1 Pyrolysis Pyrolysis refers to the chemical decomposition of organic materials (biomass) by heating in the absence of oxygen [Wikipedia, 2006]. Three products are formed: a liquid fraction of organic chemicals, organic vapours, and a high carbon-content solid known as char. Although none of the pyrolysis oil's components are lipid (oil-like) in structure, it is commonly called "bio-oil" to suggest the energy carrying capacity of the liquefied biomass. Typical liquid, solid and gas yields are 60%, 25%, 15% (of original biomass weight), respectively, yet these are highly dependent on the feedstock and process conditions [Di Blasi et al., 1999]. The process of pyrolysis can potentially be a zero-waste system as the bio-oil and char have commercial applications and the energy rich gas (syngas) can be used to satisfy internal energy demands [Dynamotive, 2006; Badger, 2002]. Previous markets for pyrolysis products include low-volume, high-value specialty applications such as liquid smoke flavouring [Holbein et al., 2005]. Presently, with the increasing need for abundant sources of fermentable sugars, pyrolysis oil has been identified as an effective and strategic method to utilize and convert biomass into high-value end products [Li and Zhang, 2004; Czernik and Bridgewater, 2004; Holbein et al., 2005; Ozbay et al., 2006]. Pyrolysis processing conditions such as heating rate, final temperature, and residence time can be varied to achieve desired yields and compositions. Thorough reviews of operating conditions and their effect on product distribution have been written by Di Blasi and co-workers (1999), Bridgewater (2003), Ozbay and co-workers (2006) and Branca and co-workers (2003). As suggested by the referenced authors, slow or conventional pyrolysis is appropriate for maximum char yields. This involves heating biomass at low temperatures over a long period of time. To increase the syngas production, high temperatures with a long residence time are recommended. For fuel applications, high liquid yield is achieved using low vapour residence times and moderate temperatures. The actual retention times are highly dependant on the feedstock and operating temperature [Di Blasi et al., 1999] thus specifics are not mentioned here. 6 Research directed towards maximizing liquid yield gave rise to the Fast Pyrolysis technology as the most efficient method for biomass to liquid conversion [Oasmaa and Czernik, 1999; Bridgewater and Peacocke, 2000; Ozbay et al., 2006]. This involves an extremely rapid heating rate with vapour residence time of a few seconds or less in the reaction zone. Pyrolysis oil liquid yields as high as 70-80 wt%, based on starting biomass weight, have been achieved [Sipila et al.,1998]. Most fresh pyrolysis oils appear to be homogenous to the naked eye. A closer look using a microscope by Oasmaa and Czernik (1999) revealed black particles suspended in the liquid, identified as pyrolysis char, or derivatives thereof. Compositional characterizations of bio-oils have determined they are actually very heterogeneous; an extremely chemically complex mixture of water, water-soluble organic compounds and water-insolubles. The majority of these components are derived from the depolymerization arid fragmentation of the biomass during pyrolysis. Secondary reactions, such as condensation in the vapour or chemical decomposition during storage would lead to the final product distribution. Pyrolysis oil studies to date have primarily focused on qualitative characterization, identifying over 300 different compounds [Scholze and Meire, 2001; Sipila et al., 1998; Bridgewater, 1996], although the bulk of the bio-oil mass is comprised of a small fraction of these compounds. Branca and co-workers (2003) identified 40 compounds, representing 65 wt% of the oil. A table of some predominant compounds is included in Appendix A. A complete and quantitative analysis has proven to be difficult as the bio-oil contains high molecular weight compounds larger than gas-chromatograph detection ranges, as well as polar, non-volatile components only detectable by HPLC. Typical bio-oil composition (expressed as weight percentage) is shown in Table 2.1. 7 Table 2.1: Typical composition of bio-oil. Component Concentration (wt%) Water 20-25 Water insolubles (lignin) 25-30 Organic acids 5-12 Non-polar hydrocarbons 5-10 Anhydro sugars 5-10 Other oxygenated compounds 10-25 The single most abundant bio-oil component is water. Other major chemical groups identified in bio-oil include aldehydes, ketones, sugars, carboxylic acids and phenolics. Pyrolysis oils are corrosive to mild steel and aluminum and should be stored and transported in air-free stainless steel or plastics tanks at, or below, room temperature [Oasmaa and Czernik, 1999]. In contrast to petroleum fuels, bio-oils contain a large amount of oxygenated components. This, along with the high water fraction, challenges the direct use of bio-oil in internal combustion engines and blending with hydrocarbon fuels [Oasmaa and Czernik, 1999]. Other properties that limit bio-oil applications include viscosity, corrosiveness, high ash content, and instability [Sensoz and Can, 2002; Bridgewater, 2003]. However, crude bio-oils have successfully been used as a substitute for No. 6 fuel oil [Oasmaa and Czernik, 1999]. Czernik and Bridgewater (2004) also advocate for bio-oil use in diesel engines and gas turbine applications. Consequently, the opportunity to fractionate and recover specific value-added components rather than using bio-oil as a whole is potentially more feasible. For fuel applications, one compound of interest is levoglucosan (1,6-anhydro-P-D-glucopyranose, L G , C 6 H 1 0 O 5 , Figure 2.3), an anhydrosugar found in bio-oils at comparatively high concentrations and recognized as a precursor for glucose fermentation [Yu and Zhang, 2003; Feng et al., 2005]. 8 O H levogtucosarc Figure 2.3: Structure of 1,6-anhydro-P-D-glucopyranose. The successful recovery and use of L G from pyrolysis oil does not prevent the extraction and use of other pyrolysis-derived compounds, thereby increasing the overall value of bio-oil. This study will focus on the pathway of extracting levoglucosan from pyrolysis oil as a feedstock for transportation fuel (refer to Figure 2.2); following three unit operations of extraction, hydrolysis and fermentation. 2.4 Characteristics of levoglucosan Levoglucosan is one of the major carbohydrates found in lignocellulosic pyrolysis oils [Branca et al., 2003]: Branca and co-workers (2003) confirmed that levoglucosan is a product of the primary decomposition of depolymerized holocellulose (cellulose and hemicellulose). Defined as an anhydrosugar, a sugar from which one or more molecules of water have been removed resulting in the formation of an internal acetal structure, with the addition of a water molecule (hydrolysis) the compound can be converted into glucose. Based on the stoichiometric ratio, 1.0 g of levoglucosan can produce 1.11 g of glucose (Equation 2.1). C 6 H I 0 O 5 + H 2 0 = C 6 H 1 2 O 6 162g/mol+18g/mol= 180g/mol (2.1) Researchers have shown that L G obtained from pyrolysis oil can be a potentially viable feedstock for bio-ethanol production by identifying yeasts and bacteria that can successfully metabolize the sugars found in hydrolysates [Zhuang et al., 2000; Sun and Cheng, 2002; Y u and Zhang, 2003; L i and Zhang, 2004]. 9 L i and Zhang (2004) produced pyrolysis oils with maximum L G concentrations of 33% from waste cotton and 19% from waste newspaper. L i and Zhang also demonstrated that L G yield in the pyrolysis oil increases as the overall liquid fraction increases. Levoglucosan concentrations in biomass pyrolysis oils have been shown to be dependent on a number of factors, namely: feedstock characteristics, feedstock pre-treatment, and process operating conditions [Scott et al., 1995; Shafizadeh et al., 2004; Radlein, 1996]. Dobele and coworkers (2003) presented a promising pre-treatment using phosphoric acid, resulting in elevated levoglucosan production when compared to otherwise identical conditions. Similarly, improvements due to acid and water washes were reviewed by Branca and coworkers (2003). Since optimal process parameters are unique to the feedstock, desired outcome, etc., it would be beneficial to focus on a few viable feedstock streams, such as forestry and agricultural wastes. Taherzadeh and coworkers (1997) did experiments on various wood hydrolysates and showed pine to produce the maximum concentration of fermentable sugars. Pine wood is one of the main raw materials of the pulping industry in British Columbia [Larsson et al., 1999]. The addition of a pyrolysis unit to a pulp mill could lead to co-production of energy and products, potentially increasing the economic viability of the entire process. Historically, levoglucosan has been commercially processed from pyrolysis oils for oligosaccharide synthesis which can be used in pharmaceutical applications [Zhuang et al., 2000]. Recovery techniques used for this process result in a powder of high purity [Scott et al., 1995] and are therefore not appropriate for fuel applications as they are too complex, energy intensive, and costly. Levoglucosan does not have to be extensively purified to be converted to glucose [Yu and Zhang, 2003]. 10 2.5 Extraction of levoglucosan from bio-oil The water content in the bio-oil is dependent on processing conditions and the original moisture content of the feedstock. Siplia and coworkers (1998), Scholze and Meire (2001) and Oasmaa and Czernik (1999) all performed elemental analysis on bio-oil produced from various lignocellulosic materials and reported water contents ranging from 15-30 wt%. As is, water in the bio-oil is miscible with the oligomeric lignin-derived compounds because of the solubilizing effect of other polar hydrophilic compounds [Oasmaa and Czernik, 1999]. However, when the water fraction is increased and the solubilizing effect is exceeded, phase separation occurs, creating soluble and non-soluble fractions hereby referred to as the "aqueous" fraction and "non-aqueous" or "organic" fraction. Phase separation typically occurs when the water content in the bio-oil reaches approximately 30-45 wt% [BTG, 2006]. After phase separation, the aqueous phase contains 40-70 wt% of the total bio-oil [Li and Zhang, 2004]. This phase is comprised of polar, low-molecular weight compounds, mostly originating from the decomposition of carbohydrates, including an appreciable quantity of the oxygenated organic compounds [Oasmaa and Czernik, 1999]. The non-aqueous phase would therefore represent 30-60 wt% of the bio-oil. This phase is a dark, viscous, tar-like material, consisting primarily of lignin derivation. The lignin in the organic phase can be used for potential applications such as adhesive manufacturing and leather tanning [Suparno, 2005]. Anything not used can be anaerobically digested or burned to provide energy. Sipila and co-workers (1998) used water fractionation during their investigation of the chemical characterization of biomass-based pyrolysis oils. They observed that the order of addition (water to bio-oil, or bio-oil to water), as well as the speed at which the bio-oil is added to water affects the fractionation reaction. By slowly dripping the bio-oil into a large volume of water, homogenous powder-like droplets are formed. Yet, when adding bio-oil quickly, the water-insoluble fraction becomes sticky rather than powder-like. A similar observation was made by Scholze and Meier (2001). 11 Another pyrolysis derivative found at concentrations similar to L G is hydroxyacetaldehyde. Hydroxyacetaldehyde can be recovered for use as a food flavouring/browning agent [Holbein et al., 2005]. An opposing selectivity between levoglucosan and hydroxyacetaldehyde was observed by Branca and co-workers (2003) as they are both primary degradation components of cellulose. Previous research on bio-oil hydrolysis involved the application of acidic conditions to the entire bio-oil [Czernik and Bridgewater, 2004; Shafizadeh et al., 2004; Oasmaa and Czernik, 1999]; with no particular discussion of oligosaccharide separation or extraction. However, i f the volume of aqueous acid exceeded the volume required for phase separation, there would be little difference between a direction hydrolysis and a sequential extraction hydrolysis process. It was decided to conduct the extraction and hydrolysis as two separate stages to avoid the potential increase in inhibitor concentration that could result from lignin being exposed to the severe acidic conditions. Furthermore, once separated, the organic phase can be additionally processed into commercial products. Not all of the levoglucosan in the bio-oil is freely extracted into the aqueous phase. Two bio-oil characterization studies revealed a large proportion of levoglucosan amongst the lignin in the organic phase after fractionation [Li and Zhang, 2004; Sipila et al., 1998]. For Sipila and co-workers (1998), approximately one half of the total concentration of levoglucosan remained in the organic phase after adding bio-oil to water at a 1:10 ratio. 2.5.1 Factors affecting the extraction of levoglucosan from bio-oil Distillation of fermented broth is required as a final processing step in order to obtain 95% industrial grade ethanol. This is a very energy intensive and costly stage [Olsson and Hahn-Hagerdahl, 1996]. Therefore, a broth of high alcohol concentration, obtained from a hydrolysate with high sugar concentration, is desired. Theoretically, to obtain a 12 maximum concentration of levoglucosan in the aqueous phase, water and/or solvent addition should be minimized. Various organic solvents (methanol, pentane, toluene, ether) have been applied in previous studies to fractionate the bio-oil into functional groups for characterization purposes [Putun et al., 1999; Sipila et al., 1998] or to homogenize the bio-oil and reduce viscosity [Oasmaa and Czernik, 1999] for boiler fuel applications. Yet, to date, there has not been an examination of the effect of the type and/or volume of solvent addition on the relative levoglucosan concentration in the resultant aqueous phase. It is hypothesized that extraction conditions such as solvent volume, temperature, and time, may affect the recovery yields of desirable (LG) and undesirable (acids) compounds. This idea is similar to the research presented by Oasmaa and co-workers (2004) who investigated the effects of different alcohols on the amount of water-insolubles in the organic fraction. Scholze and Meire (2001) reported using a 1:10 ratio of bio-oil to water to remove lignin from the bio-oil. This ratio is not appropriate for fuel applications as the concentration of levoglucosan would be too dilute. Cold water (5°C) was recommended as it seemingly helped reduce the stickiness and allowed easier handling. 2.6 Hydrolysis of levoglucosan into glucose At the time of this writing, there is only one paper that reviewed the hydrolysis of levoglucosan in pyrolysis oil. Yu and Zhang (2003) diluted cellulosic pyrolysis oil four-fold with water (on a volume basis) and then added concentrated sulphuric acid. A maximum glucose concentration (17.4%; units not clear) was achieved using 0.2M sulphuric acid at 121 °C for 20 minutes. This combination of temperature and time replicates autoclaving operating conditions, satisfying two energy intensive steps at once. Yields greater than 100% based on levoglucosan were reported; yet the exact levoglucosan and glucose concentrations were not given. A limited range of values were investigated, leaving room for additional investigation and potential improvement. 13 Furthermore, it was not mentioned whether or not only the aqueous phase was hydrolyzed or the entire bio-oil. Due to the lack of research specific to levoglucosan hydrolysis, a similar concept that has been around for decades was reviewed: glucose from the acid hydrolysis of wood. Jerome Saeman (1945) is referenced as the first author of cellulose hydrolysis kinetics. Goldstein (1980) provides a review of some of the early ethanol production methods that use wood hydrolysis. Currently, with advancements in biotechnology, enzymatic hydrolysis has gained favour over acid hydrolysis as a promising method for the future. Yet, to date, enzymatic hydrolysis is more expensive than acid hydrolysis [Galbe and Zacchi, 2002]. The expense, the impending implications of the chemical composition of pyrolysis oil, and the fact that cellulases are specific to glycosidic bonds in cellulose and not the 1,6-anhydro bonds in levoglucosan, all support the decision to use acid hydrolysis instead of enzymatic hydrolysis for this project. 2.6.1 Acid hydrolysis Saeman (1945) reported cellulose hydrolysis to be a first-order, sequential reaction. Cellulose is first broken into its monomer glucose units, which are then degraded into other products. The Arrhenius equation, a standard equation used to quantify temperature effect on hydrolysis rate, was adapted to cellulose hydrolysis by including a factor for acid concentration. Referred to as the Saeman-Arrhenius equation, the acid concentration constant is raised to an empirically-determined exponent 'n' (Equation 2.2). K = [acidf A e ~ E / R T (2.2) where: E is the activation energy (kJ/mol) R is the universal gas constant (kJ/mol-K) T is temperature (K) 14 A is the pre-exponential constant (sec"1) [acid] is the acid concentration (mol/L) n is the empirically determined exponent of the [acid] term K is the kinetic rate constant (mol/L sec) The coefficients A , E and n would be unique for every circumstance. Determination of these parameters and the applicability of this equation to levoglucosan hydrolysis has not been published in literature. A range of the three parameters identified in the Saeman-Arrhenius equation (time, temperature and acid concentration) will be explored in this study. Acid hydrolysis can be divided into concentrated or dilute acid processes. Sulphuric acid is the most common type of acid used as it has demonstrated to be both effective and inexpensive when compared to other options [Badger, 2002]. Concentrated acid hydrolysis (30-70%; 3.75-11.5 M H 2 S0 4 ) results in a 90% yield of glucose from cellulose in combination with lower temperatures (40°C) [IEA, 2003; Hamelinck et al., 2005]. Challenges when using this level of acid concentration include corrosiveness (equipment cost), lack of cost effective techniques to recover the acid and the requirement to neutralize the sugar solution after hydrolysis [Hamelinck et al., 2005; Badger 2002]. Dilute-acid hydrolysis (<1%; 0.1 M H2SO4) operates in combination with higher temperatures, yet typically produces lower cellulose hydrolysis yields (50-70%) [Hamelinck et al., 2005]. The reported temperatures used for dilute-acid hydrolysis have varied amongst the references from 120°C to 240°C as they depend on the pre-processing conditions (availability of cellulose) [Lee et al., 1999; Parisi, 1989]. Another combination of parameters that has resulted in high hydrolysis efficiencies has been published by the U.S. Department of Energy (2006). It is a two-stage procedure: 0.7 M sulfuric acid, 190°C, for a 3-minute residence time, followed by 0.4 M sulfuric acid, 215°C, and again a 3-minute residence time. The first stage is operated under milder conditions to hydrolyze hemicellulose, while the second stage is optimized to hydrolyze 15 the more resistant cellulose fraction. These conditions derived yields up to 89% for mannose, 82% for galactose and 50% for glucose. One consequence of acid hydrolysis for both cellulose and levoglucosan is the formation of degradation products that can potentially inhibit microbial activity [Mosier et al., 2002; Palmqvist and Hahn-Hagerdal, 2000b]. In more severe conditions (higher temperatures and/or higher acid concentration) the production of these inhibitors increases [Hamelinck et al., 2005; Larsson et al., 1999]. Based on the greater accessibility of levoglucosan in the aqueous phase compared to the accessibility of cellulose bonds in lignocellulose, it is hypothesized that lower temperatures could efficiently hydrolysis levoglucosan, compared to the temperatures required for cellulose hydrolysis. If so, degradation products will be less of a concern. 2.6.1.1 Effects of temperature, time and acid concentration on hydrolysis of levoglucosan The three hydrolysis parameters identified in the Saeman-Arrhenius have been expressed as a single term called the combined severity (CS) factor, originally developed by Chum and co-workers (1990) to account for the combined effect of temperature, time and acid concentration. The determination of CS is a two step process. First, a term called the reaction ordinate (Ro) is found using the temperature and time: Ro = t•exp T - T , _ J : b 14.75 (2.3) where: t is the residence time (min) T r is the reaction temperature (°C) Tb is a reference temperature (100°C). 16 Then, the combined severity factor is determined by accounting for the pH of the reaction medium: CS = l o g R o - p H (2.4) Larsson and coworkers (1999) applied the combined severity factor to approximately 75 cellulose hydrolysis experiments. As the CS of the hydrolysis conditions increased, the yield of fermentable sugars increased to a maximum between CS 2.0-2.7 for mannose, and 3.0 -3.4 for glucose, above which it decreased. An attempt to apply this model to the hydrolysis conditions used for levoglucosan will be explored. 2.7 Fermentation of lignocellulosic hydrolysates To optimize the biomass-to-ethanol process, high fermentation yields are necessary. Fermentation in the context of ethanol production refers to the conversion of sugar to alcohol using, most commonly, yeast. Following the glycolytic pathway, this occurs during the metabolic breakdown of a nutrient molecule, such as glucose, under anaerobic conditions [Wikipedia, 2006]. When yeast ferments, it breaks down glucose molecules (C6H12O6) into exactly two molecules of ethanol (C2H6O) and two molecules of carbon dioxide (CO2). Theoretically, fermentation yields 0.51 g of ethanol for every 1.0 g of glucose. Maintanance energy requirements reduce this conversion rate to a maximum yield of approximately 0.45 g/g [Helle and Duff, 2004]. Improvements key to reducing fermentation costs are the development of engineered yeast or bacteria that can ferment both the five- and six-carbon sugars. Approximately 20% of the sugars found in lignocellulosics are pentoses [Wyman, 1999; Mosier et al., 2002]. Different fermentation organisms among bacteria, yeasts, and fungi (natural as well as recombinant) have been reviewed with an emphasis on their performance in lignocellulosic hydrolysates by Olsson and Hahn-Hagerdal (1996). Ongoing 17 developments of new fermentation strategies (i.e., batch culture, continuous culture, cell recycling, in situ ethanol removal, etc) are important for cost effective high yields. Many of these techniques are currently being explored [Lin and Tanaka, 2006]. Saccharomyces cerevisiae is the most common strain used for the fermentation of alcohol as it has a high selectivity for fermentation over new biomass growth [Roehr, 2000]. Its effectiveness in fermenting hexose sugars found in lignocellulose-derived hydrolysates was shown by Olsson and Hahn-Hagerdal (1996). S. cerevisiae can grow in pH ranges from 2.4 to 8.6 and temperatures between 28 to 35°C. At higher temperatures, ethanol production is promoted over growth [Roehr, 2000]. Although fermentation takes place under anaerobic conditions, oxygen is still required at low levels to maintain yeast metabolism and lipid biosynthesis [Helle and Duff, 2004]. The optimum level of oxygen for ethanol production is highly variable; depending on factors such as the type of cell suspension (immobilized or suspended), biomass concentration, stage of cell growth, etc. [Roehr, 2000]. If levels of aeration are too high, higher biomass yields and cell viability will occur at the expense of ethanol production. Semi-aerobic or micro-aerobic conditions can be used to satisfy the oxygen requirements, using a foam plug or rubber-vented stopper, respectively [Helle et al., 2003]. Micro-aerobic conditions have been found to produce more ethanol using glucose/xylose mixtures than otherwise identical semi-aerobic conditions [Helle et al., 2003]. Quantitatively, micro-aerobic and semi-aerobic oxygen uptake rates equate to aeration rates of 12 mmol (Vmol-Ohr and 26.8 mmol CVmol-Ohr, respectively [Sampaio et al., 2004]. This study does not use any oxygen detection equipment; rather, the oxygen availability is qualitatively classified as micro-aerobic using very limited gas exchange (needle in septa) and semi-aerobic using a foam plug. The term micro-aerobic generally refers to the environment; where as micro-aerophilic refers to the organism. However, both terms can be used interchangeably. 18 2.7.1 Fermentation rates, productivity and yields in lignocellulosic hydrolysates Specific productivity is the mass of ethanol produced per unit time per unit of cell mass. It is difficult to calculate in a batch culture as the cell mass changes over the course of the fermentation, therefore usually only volumetric productivity is reported. Total volumetric productivity is expressed in grams per litre per hour and is defined, for batch fermentation, as the final ethanol concentration divided by the total fermentation time. Generally, low ethanol productivity is observed for lignocellulose hydrolysates. Olsson and Hahn-Hagerdahl (1996) report a range of 0.11 to 0.37 g/L/hr for various S. cerevisiae strains during spent sulphite liquor fermentation. The highest productivity in this range was obtained using a very large cell mass concentration (75 g/L dry weight). Yu and Zhang (2003) report a rate of 0.59 g/L/hr for willow-derived hydrolysate. Klinke and coworkers (2004) provide a comprehensive review of ethanol production rates in various hydrolysates. Pine hydrolysate exhibited a yield of 39% (g EtOH/g glucose) after 24 hours using an inoculum of 10 g/L. The volumetric productivity was quite high at 2.87 g/L hr. These rates will be compared to results obtained from this study. The fermentation of other hexose and pentose sugars, such as mannose and arabinose, respectively, can lead to ethanol yields greater than 100%) when based on available glucose. Pine pyrolysis oil has been reported to have a higher fraction of mannose when compared to other pyrolysates as presented by Taherzadeh and coworkers (1997). Both high yield and high ethanol productivity are critical for economic feasibility [Olsson and Hahn-Hagerdal, 1996]. Particularly in toxic or inhibitory conditions, yeast strains that have adapted (mutated) to their environment can result in higher ethanol yields. Yu and Zhang (2003) recycled a strain of S. cerevisiae 12 times, which resulted in an average 6% higher ethanol yield. T2, the yeast used in this study, is an S. cerevisiae strain adapted to spent sulphite liquor fermentation and it has proven to be sufficiently robust to ferment hexose sugars in a highly inhibitory environment. Palmqvist and co-workers (1999) showed higher 19 fermentation rates when using the T2 yeast when compared to regular baker's yeast in lignocellulosic hydrolysates. It has also been shown to exhibit less lag, greater cell viability, and achieve greater fermentation rates than two other yeast strains in spent sulphite liquor [Helle et al., 2003]. 2.7.2 Fermentation inhibitors formed during biomass processing In order to produce fermentable sugars from lignocellulosic materials, some form of thermochemical treatment is required. During this treatment, compounds are generated which can inhibit downstream fermentation of the glucose [Olsson and Hahn-Hagerdal, 1996]. The type and concentration of inhibitors vary depending on the type of thermochemical treatment used and the severity of the processing conditions. Categories of inhibitors include weak organic acids, furans and phenolic compounds. Work that describes the mechanism and effect of the inhibitors is on-going [Palmqvist and Hahn-Hagerdal, 2000b; Mosier et al., 2002; Helle et al., 2003]. Acetic acid inhibition of ethanol fermentation has been well documented [Olsson and Hahn-Hagerdal, 1996; Taherzadeh et al., 1997; Palmqvist et al., 1999; Helle et al., 2003]. Acetic acid is a product of biomass pyrolysis and a co-product of fermentation. It is only inhibitory in an undissociated form when it can passively diffuse across the cell membrane and cause the internal pH to decrease [Palmqvist and Hahn-Hagerdal, 2000b]. An acid is undissociated when the pH of the solution is lower than its pKa (acetic acid's pKa is 4.7); therefore, maintaining a pH above 5 during fermentation will help reduce acetic acid inhibition. Formic acid has been found to have a similar inhibitory affects as acetic acid [Klinke et al., 2004]. Glucose decomposition results in inhibitory products such as hydroxymethyl furfural (HMF), levulinic acid and formic acid [Mosier, 2002; Zaldivar et al., 2001; Larsson et al., 1999, Figure 2.4]. Both furans and formic acid can also be produced from phenol monomers, which occur as a result of lignin degradation. 20 H—C - ~ 0 OH—C — H I H—C—OH i O H - C — H I H—C—OH Q H - C — H H 3HzP | J -CMJ1U<:OSC .l-hyJnwymrthylfutfitfa! levulinJc acid formic achi Figure 2.4: Degradation products formed from glucose during dilute acid hydrolysis [Larsson etal., 1999]. Furans in general have been documented to inhibit yeast growth and ethanol production rate, but do not significantly affect the ethanol yields in 5. cerevisiae [Palmqvist et al., 1999; Klinke et al., 2004]. It has been found that some inhibitory effects can be enhanced in the presence of other effects. Palmqvist and co-workers (2000) showed that 2-furfural and acetic acid exhibited synergistic inhibition of both growth and ethanol yield in S. cerevisiae. Other potentially inhibiting lignin-derived compounds include aromatic and polyaromatic components [Palmqvist and Hahn-Hagerdal, 2000a and 2000b]. These compounds are formed during pyrolysis, hydrolysis, and/or during storage over time, as a result of low pH conditions. By separating the lignin from the fermentable sugars through phase separation before acid hydrolysis, some of the risk of producing more lignin-derived compounds is avoided. Ethanol inhibition occurs when concentrations in the fermentation broth exceed 110 g/L [Kargupta et al., 1998]. It has been modelled as non-competitive product inhibition, affecting the yeast growth rate more than ethanol production rate [Brown et al., 1981]. The concentration of ethanol formed during this study does not exceed inhibitory levels. 21 2.7.2.1 Removal of inhibitors from lignocellulosic hydrolysates Inhibitors can be removed via precipitation, stripping, adsorption, or enzyme treatment. Research to overcome inhibition has been reviewed by Palmqvist and Hahn-Hagerdal (2000a and 2000b). A cost-benefit analysis would be required to evaluate if detoxification technology was economically feasible for each unique case of feedstock and processing conditions. An economic analysis by Palmqvist and Hahn-Hagerdal (2000a) reported an ethanol production cost increase of 22% when using the detoxification technique of over-liming. Rather than the removal of inhibitors, a number of fermentation techniques can be applied in an effort to improve ethanol production in inhibitory lignocellulosic hydrolysates. Some simple and effective methods include oxygen and pH control, as well as a high yeast concentration. The use of a large yeast inoculation has been shown to be a suitable way of increasing the volumetric productivity, as more energy is spent on ethanol production than cell growth [Palmqvist and Hahn-Hagerdal, 2000a]. Olsson and Hahn-Hagerdal (1996) exhibited a 2.5-fold increase in ethanol productivity from a ten-fold increase in cell mass. Cell immobilization can be used to achieve a high cell density and thus a higher ethanol productivity. Najafpour and co-workers (2004) reported a 10-fold increase in ethanol production using cell immobilization for batch fermentation. For continuous fermentation, cell recycling can be applied to maintain higher yields [Kargupta et al., 1998; Kishimoto et al., 1997]. 22 CHAPTER III - RESEARCH OBJECTIVES The overall goal of this research was to develop an alternative bio-ethanol production technology that utilizes pyrolysis oil as a renewable source of inexpensive carbohydrates. Specifically, the research objectives involve the investigation and improvement of three processing stages: I. Extraction of levoglucosan into the aqueous fraction using water Phase separation will be explored. Of interest will be the quantity of compounds, such as levoglucosan, glucose and organic acids, transferred into the aqueous phase. Conditions favourable to maximum levoglucosan extraction will be investigated by manipulating the process variables of temperature, time and volume of water. II. Acid hydrolysis of levoglucosan into glucose Once the levoglucosan is in the aqueous phase, sulphuric acid will be added to convert the anhydrosugar into glucose. Process conditions of temperature, time and acid concentration will be investigated to optimize the hydrolysis yield. JMP IN software and a combined severity model will be applied in an attempt to quantify the effects of the parameters. III. Fermentation of glucose into ethanol The toxicity of the hydrolyzed aqueous phase will be explored by subjecting an adapted yeast strain to fractions of nutrient rich media combined with the hydrolysate. Then, fermentation techniques will be applied to improve the fermentation. Ethanol productivity rates and maximum ethanol yields will be investigated and reported. 23 CHAPTER IV - MATERIALS AND METHODS 4.1 Characterization of VTT bio-oil The pyrolysis oil was prepared by VTT Processes2 from Scots Pine feedstock in August 2004. Properties of the bio-oil given by VTT are shown in Table 4.1. The bio-oil was stored at a low pH (pH -2.5, as received) in glass containers at 4°C with no light exposure for the duration of this study. The bio-oil was extremely sticky and very difficult to wash off glassware, therefore disposable pipettes are recommended. Based on the high viscosity of the bio-oil, a pipette pump, rather than the pipette bulb, allowed easier handling. A review of bio-oil toxicity [Diebold, 1997] revealed it is irritating to eyes, the respiratory system, and skin. Protective equipment and fume hood use are encouraged. Bio-oil has a distinctive, acrid, smoky smell due to the low molecular weight aldehydes and acids. Table 4.1: Elemental composition and other properties of pine pyrolysis oil. Component and Unit Value Test (if reported) Carbon in dry matter, m% 50.7 A S T M D 5373 Hydrogen in dry matter, m% 6.3 A S T M D 5373 Nitrogen in dry matter, m% 0.1 A S T M D 5373 Volatiles in dry matter, m% 83.8 DIN 51720 Ash in dry matter, m% 0.2 DIN 51719 H H V in dry matter, MJ/kg 20.63 DIN 51900 L H V in dry matter, MJ/kg 19.27 DIN 51900 Moisture as received, m% 7.0 DIN 51718 Water, wt % 21.1 Solids, wt % 0.008 Ash, wt % 0.03 Nitrogen, wt % 0.06 Carbon, wt % 43.5 2 VTT Processes. Biologinkuja 5, Espoo PL 1601, 02044 VTT, Finland 24 Hydrogen, wt % 7.09 Viscosity (20°C), cSt* 104 Viscosity (40°C), cSt 26 Viscosity (80°C), cSt 5 Density (15°C), g/mL 1.223 Flash point, °C 55 HHV, MJ/kg 17.72 L H V , MJ/kg 16.18 pH 2.5 * cSt = CentiStockes (cSt = CentiPoises (cp) / density) 4.1.1 Determining the point of phase separation A titration experiment was set up to observe the point of phase separation. The bio-oil and the experiment was conducted at room temperature. Small increments of water were added to 10 mL of bio-oil in a 50 mL beaker which allowed for a large surface contact area (all glassware used throughout this study is from the Pyrex line from Fisher Scientific, ON, Canada). Due to the small volume of water used for this study, the precipitates formed a sticky mass no matter the rate or the order in which it was added. Therefore, for easier handling and greater accuracy, water was always added to the bio-oil. Observations were recorded. After each incremental water addition, it was necessary to mix the solution by swirling the beaker by hand. Otherwise, the water remained on top of the bio-oil and did not diffuse into the solution. 4.2 The effect of water volume on the fractionation of components in bio-oil As previously described, pyrolysis operating conditions have been identified which result in high yields of L G in the pyrolysis oil [Brown et al., 2001]. Accordingly, it should be 25 ensured that a maximum amount of this levoglucosan is captured from the oil into the aqueous phase. The quantity and identity of the components transferred from the bio-oil into the aqueous phase are of interest for this study. A maximum concentration of levoglucosan is desired; however, not at the expense of inhibitory concentrations of other compounds. Theoretically, a minimal volume of solvent addition will result in maximum levoglucosan concentration. Yet, this phenomenon has not been scientifically reviewed. This experiment will contribute to the understanding of bio-oil fractionation with an attempt to improve the recovery of levoglucosan into the aqueous phase. Different volumes of bio-oil were dispensed into labelled, empty, 50 mL disposable centrifuge tubes using a disposable graduated pipette (all from Fisher Scientific, ON, Canada). Weights were recorded after each step. To these, different volumes of water were added to create different solvent-to-oil ratios, as shown in Table 4.2; henceforth referred to as Extract 1 through 8. Table 4.2: Volumetric ratios used to investigate extraction. Sample Bio-oil (mL) Water (mL) Total water (expressed as wt% of bio-oil) Ratio (bio-oil to water) Extract 1 5 25 430 1 : 5 Extract 2 10 40 348 1 : 4 Extract 3 10 30 266 1 : 3 Extract 4 15 30 185 1 : 2 Extract 5 15 20 130 1 : 1.33 Extract 6 25 25 103 1 : 1 Extract 7 20 10 62 1 : 0.5 Extract 8 20 5 41 1 : 0.25 Tubes were mixed vigorously on a vortex mixer (Thermolyne Type 37600, Iowa, USA) and left at room temperature overnight. The following morning, the tubes were centrifuged (Centrifuge C U 5000, Damon/IEC division) for 30 minutes at 3500 g. The aqueous layers were decanted and the mass and volumes were recorded and then used in the following analysis. 26 The unit most often used throughout this paper is weight percent (wt %). In context, the original water in the bio-oil equals 21.1 wt%, meaning, for 100 g of bio-oil 21.1 g of it is classified as water. Now, specific to the water calculation shown in Table 4.2, the total water volume includes both the original water (21.1 wt%) plus the water added. Using Extract 4 as an example: mass of bio-oil (g) =15 mL * 1.223 g/mL = 18.345 g original water (g) =21.1%* 18.345 g = 3.87g total water (wt%) = added water (g) + original water (g) / weight of bio-oil (g) (includes water) = (30 g + 3.87 g / 18.345 g) * 100% =185 wt% 4.2.1 Determination of total solids in the aqueous phase Three mL samples, in duplicate, of each extract were transferred into weighed and labelled ceramic crucibles using a 5000 uL pipette. The crucibles were placed in a 105°C microprocessor controlled oven (VWR, Oregon, USA) for 2 days and then weighed again to determine the concentration of total solids (Equation 4.1). total solids (g/mL) = (4.1) sample volume (mL) where: A = weight of crucible + dried residue (g) B = weight of empty crucible (g) 27 4.2.2 Determination of sugar and organic acid content in the aqueous phase A 10 mL sample of extract from three of the ratios (Extract 2, 4, and 6) was diluted with 20 mL distilled water and titrated with 1M NaOH (Fisher Scientific, ON, Canada). The pH was measured using a 71 OA pH probe (Orion Research, M A , USA) after each addition of NaOH. The data was entered into Excel using a weak acid strong base titration model adapted by Dr. Steve Helle. The model is further described in Appendix B. One mL aliquots from each extract were removed for glucose and levoglucosan analysis on the GC. The methods of preparation for GC analysis are described in Section 4.6. 4.3 Determining parameter effects of temperature, time and water volume on the extraction of L G into the aqueous phase Building on the results from previous experiments, the process parameters of temperature and time are now included in the fractionation operation as an attempt to improve the yields of levoglucosan extraction into the aqueous phase. There is little known about the rate of diffusion of some of the polar components in bio-oil into the aqueous fraction; for example, it is not know whether or not contact time can improve extraction efficiency. Although the phase separation is instantaneous, the conditions may be suitable for some of the smaller compounds to continue their transfer between phases. A factorial-design experiment was used to investigate the direct and interactive effects of the three experimental parameters on the desired outcome of levoglucosan concentration. Parameter values are assigned high (1), middle (0) and low (-1), as shown in Table 4.3. If every combination of the parameters was used, 27 samples (33) would be required. That is why a surface response design tool is used, reducing the experiment to 16 samples, but still covering enough axial and corner points to develop a valid predictive response model. 28 Table 4.3: Temperature, time and water parameters used for extraction experiment. Parameter Temperature Time Total water Level (°C) (min) (wt% of bio-oil) Low (-1) 4 0 62 Middle (0) 19 30 103 High(l ) 34 60 185 Equal volumes of bio-oil (10 mL) were placed in 16 labelled glass tubes: twelve combinations of extraction parameters and four middle points (Table 4.4). The temperature of the water was adjusted to reflect the reaction temperatures 34°C and 4°C, respectively. The water was added to room-temperature bio-oil to achieve the desired ratio. The tubes were capped, inverted by hand initially, and then mixed on a vortex shaker for 2 minutes. Samples were placed in the fridge (4°C), lab bench (19°C) or in the dry bath incubator (34°C) (Fisher scientific, ON, Canada) for the stipulated reaction time. A l l temperatures were verified before and during the experiment using thermometers. Table 4.4: Combination of parameters for surface response extraction experiment. Sample Temp Time Water 1 -1 -1 -1 2 -1 -1 1 3 1 0 0 4 -1 1 1 5 -1 0 0 6 1 -1 -1 7 0 0 0 8 -1 1 -1 9 1 1 -1 10 0 0 -1 11 0 1 0 12 0 -1 0 13 1 1 1 14 0 0 1 15 0 0 0 16 1 -1 1 When the reaction time was complete, samples were removed and the aqueous fraction was immediately decanted into clean 25 mL Erlenmeyer flasks. This occurred quite 29 easily as the aqueous phase flowed freely and the viscous organic phase generally stuck to the bottom of the test tubes. A 1 mL sample from each aqueous extract was used for GC analysis to determine L G and glucose concentrations. The GC data was entered into JMP IN software for the analysis of the parameter effects. See the analytical discussion, Section 4.6, for more on GC preparation and JMP IN analysis. 4.4 Acid-hydrolysis of levoglucosan into glucose Once optimal extraction parameters were established, the aqueous phase was subjected to acid hydrolysis conditions to convert the anhydrosugar into glucose. The path of investigation was similar to the extraction experiments. Initially, three aqueous extracts were hydrolyzed to provide a general understanding of the conversion response. Then, various acid concentration, temperature and time values were explored in a surface response experiment and JMP IN software. 4.4.1 Initial investigation into levoglucosan hydrolysis Three aliquots of three aqueous extracts were transferred to glass culture tubes, totalling nine samples (3x3). The extracts were prepared using the combination of bio-oil and water volumes presented as Extract 4, 6 and 7 in Table 4.3, representing total water volumes of 185 wt%, 103 wt% and 62 wt%, respectively. Five mL of the extracts were placed in clean 25 mL screw-top tubes. Five mL of 1.0 M sulphuric acid was added to attain an acid concentration of 0.5 M . Dilution of the aqueous extract was factored into the calculations. Controls were prepared using water instead of acid to witness the effect of temperature alone. A l l tubes were capped and placed in a 120°C dry block heater. At twenty minute increments, tubes were removed and immediately submerged in cold water. After samples were cooled, a 1 mL aliquot was taken for sugar determination by GC analysis. 30 4.4.2 Determining the effects of hydrolysis parameters temperature, time and acid concentration on the conversion of levoglucosan into glucose A quantitative analysis of the primary and secondary (interactive) effects of the hydrolysis parameters of time, temperature and acid concentration on the yield of glucose from levoglucosan was conducted using factorial design experiments. Due to potential secondary effects, it is possible that more than one combination of these parameters could derive a maximum value. Optimal conditions would vary considerably for different hydrolysates as different biomass materials exhibit different levels of buffering capacity [Xiang et al., 2003, Lee et al., 1999]. Based on the Arrhenius equation, an increase in temperature will increase the kinetic conversion rate constant. However, i f the exposure time is too long, the sequential reaction will continue and degradation of the sugars will occur. Using the quantitative empirical model, design parameters can be manipulated to reflect the available inputs and desirable output. In an attempt to improve upon the previous findings, three separate experiments were conducted using various ranges of hydrolysis parameters, referred to as hydrolysis conditions A, B, and C, respectively. These parameters, representing high (1), middle (0) and low (-1) values, are shown in Tables 4.5, 4.6, and 4.7. The combinations of parameters assigned to the samples are outlined in Tables 4.8 and 4.9, each including four middle points. Please note Table 4.9 has temperature and time in reverse positions than the previous table. This is only because of the order in which the parameters were entered into JMP IN. Results were analyzed for primary and secondary effects. Sugar concentrations produced for hydrolysis conditions A, B and C were determined using GC analysis. 31 Table 4.5: Hydrolysis conditions A. Parameter Level Temperature (°C) Time (min) Acid Concentration (M) Low (-1) 90 10 0.1 Middle (0) 105 . 25 0.3 High(l) 120 40 0.5 Table 4.6: Hydrolysis conditions B. Parameter Level Temperature (°C) Time (min) Acid Concentration (M) Low (-1) 90 10 0.2 Middle (0) 105 25 0.6 High(l ) 120 40 1.0 Table 4.7: Hydrolysis conditions C. Parameter Level Temperature (°C) Time (min) Acid Concentration (M) Low (-1) 110 20 0.50 Middle (0) 130 40 0.75 High(l) 150 60 1.0 Table 4.8: Combination of parameters of time, temperature and acid concentration for hydrolysis experiments A and B. Sample Temp Time Acid 1 -1 -1 -1 2 0 -1 0 3 1 -1 1 4 0 1 0 5 -1 1 -1 6 -1 0 0 7 0 0 1 8 0 0 -1 9 -1 1 1 10 1 1 1 11 0 0 0 12 0 0 0 13 1 -1 -1 14 1 1 -1 15 -1 -1 1 16 1 0 0 Table 4.9: Combination of parameters of time, temperature and acid concentration for hydrolysis experiment C. Sample Time Temp Acid 1 -1 -1 -1 2 1 -1 -1 3 0 -1 0 4 0 1 0 5 -1 0 0 6 0 0 0 7 1 -1 1 8 0 0 -1 9 -1 1 1 10 0 0 1 11 -1 1 -1 12 1 0 0 13 1 1 -1 14 -1 -1 1 15 1 1 1 16 0 0 0 33 4.4.3 Determining the effects of initial levoglucosan concentration and acid concentration on hydrolysis yield Throughout the investigation, it was observed that levoglucosan hydrolysis yields were quite variable even when using similar time, temperature and acid concentration conditions. The only noticeable difference between the experiments was the initial levoglucosan concentration. Based on these observations, a factorial design experiment was conducted to further explore this phenomenon. Theoretically, for the complete hydrolysis of a large concentration of levoglucosan, a large concentration of acid is required (molar ratio). This may lead to harsh hydrolysis conditions and catalyze the decomposition of sugars and other compounds. A full-factorial experiment was designed to examine the relationship and potential interactive effects between the initial levoglucosan concentration and the acid concentration on the final hydrolysis yield. First, an aqueous extract was prepared using room temperature conditions and a bio-oil to water ratio of 1:1. Three dilutions were subsequently prepared from this aqueous extract with the intent to achieve an even distribution of concentrations ranging from low (-1) to high (1) as shown in Table 4.10. Table 4.11 displays all the combination of a two-parameter, three-level experiment plus two middle points (23 + 2 = 10). A l l levoglucosan and glucose concentrations were determined using GC analysis. Table 4.10: Values of levoglucosan concentration and acid concentration assigned to parameter levels. Level Levoglucosan (g/L) Acid Concentration (M) (High) 1 123 0.75 (Middle) 0 47 0.45 (Low) -1 29 0.15 34 Table 4.11: Combination of parameters for factorial designed hydrolysis experiment. Sample L G Acid 1 -1 -1 2 1 -1 3 0 -1 4 0 0 5 1 0 6 1 1 7 0 0 8 -1 0 9 0 1 10 -1 1 For the small scale hydrolysis experiment, 2.5 mL of the various levoglucosan concentrations were placed in ten glass 10 mL tubes. Sulphuric acid was added to the tubes to achieve the final acid concentrations depicted in Table 4.10. Each dilution factor was recorded and included in calculations. The tubes were capped and placed in a dry-block heater at 110°C for 40 minutes. Tubes were removed and cooled in room temperature water. Two mL from each tube was used for sugar analysis using the HPLC (high pressure liquid chromatography). 4.5 Fermentation of aqueous phase Although fermentation is the final stage of the ethanol process, some of these experiments were conducted before the extraction and hydrolysis experiments to ensure the aqueous phase and its subsequent hydrolysate were fermentable. The yeast strain Saccharomyces cerevisiae type T2 was obtained from Tembec (Temiscaming, Quebec) and was maintained on yeast extract agar plates (1% yeast extract, 2% peptone, 2% glucose and 1.8% agar, all from Fisher Scientific, ON, Canada) for the duration of the study. A l l experiments were initiated with a starter culture to ensure uniform yeast concentration and characteristics for all vials and flasks. Starter cultures were prepared by aseptically transferring a loop full of yeast from the agar plates to a sterile 100 mL broth of rich Y P G 35 media (1% w/v yeast extract, 2% w/v peptone, and 2% w/v glucose) in a foam-plugged 250 mL Erlenmeyer flask. Glucose solutions (50% w/w) were separately autoclaved and added to the sterile YP broth just prior to inoculation. Starter cultures were incubated in an environmental shaker (Innova 4230 refrigerated incubator shaker, New Brunswick Scientific, NJ, USA) at 30°C, 150 rpm, overnight, or until desired yeast concentration was achieved. Yeast extract ensured a hearty nutrient supply was available for optimal growth and activity; however, large scale production would require cheaper sources of nutrients. Alternatives such as corn steep liquor or Fermaid K have been suggested by Helle and co-workers (2003). The starter culture was transferred into two 50 mL sterile centrifuge tubes and spun down at 3000 g for five minutes (Centrifuge C U 5000, Damon/IEC division). The supernatant was discarded. The yeast pellets were re-suspended using sterile distilled water to approximately 30 mL; enough to ensure at least 2 mL per flask. Yeast concentrations were determined using total suspended solids (TSS) assay as described by Equation 4.1. In order to achieve the higher yeast concentrations required in experiment 4.4.3, after the first growth stage and centrifuge step, each yeast pellet was transferred into other 100 mL rich Y P G flasks and allowed to grow for another 24 hours. Yeast was always transferred using sterile pipettes. It is known that high concentrations of acetic (4.8-5.5%) and formic acid (0.4-0.6%o) are present in pyrolysis oils [Branca et al., 2003]. As previously reported, organic acids are inhibitory to fermentation when undissociated; therefore maintaining the pH above the pKa of the acids (acetic = 4.8, formic = 3.74) is a technique used throughout the fermentation experiments. Moreover, the starting pH must account for the pH drop over the fermentation period. 4.5.1 Determining fermentability of the aqueous phase A l l of the experimental fermentation broth had the same rich Y P G concentration as the inoculum (1% yeast extract, 2% peptone, and 2% glucose). The flasks were prepared 36 using different fractions of aqueous extract and distilled water to bring the broth to its final volume. The aqueous extract was prepared using room temperature conditions and a bio-oil to water ratio of 1:1. The pH of the aqueous extract broth was adjusted using sodium hydroxide to a pH of 5.5 before inoculation. Twelve flasks were prepared, six in duplicate, to a final volume of 100 mL using incremental fractions of the two broths as outlined in Table 4.12. The flasks were autoclaved at 121 °C for 20 minutes. The 50% glucose solution, 50 g dextrose (Fisher Scientific, ON, Canada) with 50 mL distilled water, was autoclaved separately. Using aseptic techniques, glucose (2%) and yeast (2%) were added to each flask. Table 4.12: Fermentation flasks prepared using fractions of aqueous phase and distilled water ranging from 0 to 100%. Flask number Aqueous phase YPG broth (mL) Distilled water YPG broth (mL) 1,2 (control) 0 100 3,4 20 80 5,6 40 60 7,8 60 40 9,10 80 20 11,12 100 0 The fermentation was performed in 250 mL Erlenmeyer flasks using foam plugs to achieve a semi-aerobic environment. Flasks were incubated at 30°C in an environmental shaker at 150 rpm for the duration of the fermentation period, except for during sampling. To obtain a sample, the foam plug was removed and an aliquot was aseptically withdrawn and dispensed into a 2 mL centrifuge tube using a sterile long stem glass pipette and rubber bulb. The centrifuge tubes were frozen until further analysis. The preparation of standards and setting used for GC analysis are described in the analytical section (4.6). 37 After the initial analysis, a more refined range of hydrolysate fractions was investigated. The values are shown in Table 4.13. Table 4.13: Fermentation flasks prepared using fractions of aqueous phase and distilled water ranging from 0 to 20%. Flask number Aqueous phase YPG broth (mL) Distilled water YPG broth (mL) 1,2 (control) 0 100 3,4 2 98 5,6 5 95 7,8 10 90 9,10 15 85 11,12 20 80 4.5.2 Fermentation using micro-aerophilic conditions Modifications to the standard fermentation technique were necessary to overcome the level of toxicity. In addition, hydrolysate was now used instead of the unhydrolyzed aqueous extract pyrolysate to investigate the potential ethanol yields from hydrolyzed levoglucosan. To obtain micro-aerophilic conditions, serum vials were used (Wheaton serum vials, Fisher Scientific, ON, Canada). The vials were sealed with rubber septa, crimped with an aluminum cap, and vented by a 19-gauge needle. Twelve flasks were prepared, six in duplicate, using incremental fractions of distilled water Y P G broth and hydrolysate Y P G broth as outlined in Table 4.14. The hydrolysate was prepared in advance using 0.5 M sulphuric acid for 30 minutes at 110°C. The sugar concentration in the hydrolysate was determined using HPLC analysis and the ethanol concentrations were determined by GC. 38 Table 4.14: Fermentation vials containing incremental fractions of hydrolysate for micro-aerophilic fermentation experiment. Flask number Hydrolysate Y P G broth (mL) Distilled water Y P G broth (mL) 1,2 (control) 0 100 3,4 2 98 5,6 5 95 7,8 10 90 9,10 15 85 11,12 20 80 Vials were incubated at 30°C in an environmental shaker at 150 rpm for approximately 24 hours, during which they were periodically sampled. Samples were withdrawn using a 19-gauge needle with a 5 mL syringe. The needle and syringe remained in the vials during the fermentation period, other than being removed for the sampling itself. The vial set up is shown in Figure 4.1. The needle was flamed before being returned to the bottle via the same hole in the septa. Each sample was dispensed into a 2 mL centrifuge tube and frozen until further analysis. Figure 4.1: Experimental set-up for micro-aerophilic conditions. 39 4.5.3 Fermentation using high inoculation In a further effort to overcome inhibition, a higher yeast cell concentration was used. Serum vials were inoculated with 3 mL from a yeast solution of 35 g/L, resulting in a yeast concentration in the flasks of 1 g/L. Once again, twelve flasks were prepared, six in duplicate samples, using the same incremental fractions of both the hydrolysate and distilled water broths, represented in Table 4.13. Fermentation parameters (time, temperature, rpm) and sampling technique were identical to those described in Section 4.4.2. Ethanol concentrations were determined by GC analysis. 4.6 Analytical 4.6.1 Sugar determination assay using HPLC analysis Sugars, namely levoglucosan, mannose, galactose, arabinose, glucose and xylose, were analyzed on a Dionex HPLC system equipped with an AutoSampler AS50, Gradient Pump GP50 and Electrochemical Detector ED50A. The mobile phase was distilled and de-ioinized water (Milli-Q Ultra-pure water purification system, Millipore, M A , USA) and was maintained at a flow rate of 1.0 mL/min for 40 minutes, followed by a 0.025M sodium hydroxide wash for 20 minutes to release the retained molecules. A post column, pre-detector, injection of 0.025M sodium hydroxide was incorporated to increase the detection sensitivity (set at 1 mL/min using a TTL (transistor-transitor-logic) pump). The injection volume of the sample was 25 uL and the column temperature was a constant 35°C. Standard sugar solutions were prepared using powdered reagents all obtained from Fisher Scientific. The powders were mixed with de-ionized distilled water to create stock solutions of 5 g/L. Subsequently, 1 g/L, 0.5 g/L and 0.1 g/L solutions were prepared when required, via serial dilution. Fucose, a sugar not found in bio-oil, was used as an internal standard at a concentration of 1.0 g/L. A l l stock solutions were kept at 4°C when not in use. A typical chromatograph is shown as Figure 4.2. 40 OCT-5-05 #3 std3 lntAmp_1 <D W O O Z3 LL. , , , 1 , , , 1 r—, , 1 , , , 1 , , , 1 , . , 1 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Figure 4.2: Chromatograph of sugar standards using HPLC analysis. 4.6.1.1 Determination of sugar peaks Using the above parameters, sugar retention times were determined. The average retention times shown in Table 4.15 were obtained over multiple trials. Table 4.15: Sugar retention times for the HPLC. Sugar Retention time (minutes) Standard deviation (minutes) Levoglucosan 2.7 0.1 Fucose 6.2 0.4 Arabinose 15.3 0.5 Galactose 18.4 0.8 Glucose 21.8 0.8 Xylose 25.7 1.8 Mannose 28.9 1.1 1 41 Based on the chemical complexity of the bio-oil, the levoglucosan peak was not completely separated from other compounds, which challenged the accuracy of calculating the area under the peak. Therefore, to determine levoglucosan concentration, experiments were repeated using GC analysis. However, the glucose peaks were acceptable. Figure 4.3 illustrates the challenge in segregating the L G peak from other peaks. 2 Q 0 OCT-5-05 #4 [modified by BRITISH COLUMBIA] hC 750H 500H 250--200-c ro co o o _=3 D) O > 0.0 o O CO o ro cn o o Csj CM O 10.0 20.0 30.0 lntAmp_1 mm 40.0 50.0 60.0 Figure 4.3: HPLC chromatograph illustrating the challenge in isolating levoglucosan peak. 4.6.2 Sugar determination using G C analysis Sugar analysis was performed on the GC using a 1701 column (30 m, 0.25 mm ID, 1 um film, Agilent Technologies, CA, USA) on a Varian CD-3800 equipped with CP-8400 autosampler and FID (flame ionization detector). In order for the sugars to be detected using GC analysis, they were derivatized into per-acetylated aldononitrile compounds using the following technique. 42 One mL samples were transferred into clean culture tubes to which 0.1 mL ribose (internal standard), 0.2 g hydroxylamine hydrochloride and 2 mL 1-methylimidazole were added. The tubes were incubated at 80°C in a dry block heater for 10 minutes, after which they were immediately cooled in a water bath. Ten mL acetic anhydride was added for a 2 minute reaction, and the tubes were then cooled in a 15°C water bath. Addition of 5 mL of methylene chloride and 10 mL of water were each followed by mixing. This caused a separation of phases. The sugar derivatives are present in the organic phase (bottom phase). An aliquot from this phase was transferred into GC vials. For the GC analysis, temperatures were set at 270°C for the GC oven, 250°C for the injector and 285°C for the FID. The flow rate of helium (carrier gas) was held constant at 1 mL/minute and the split ratio was set at 50:1. The injection sample was 10 uL. Calibration curves were prepared using L G and glucose at three concentrations. Figure 4.4 shows a typical three-point calibration curve used for analysis. The peak ratio (y-axis) is an expression of the area under the curve for glucose (or levoglucosan) divided by the area under the internal standard peak of ribose. 45 -, 0 1 2 . 3 4 5 6 Peak Ratio Figure 4.4: Calibration curve for sugar analysis by GC. Glucose (squares) and levoglucosan (diamonds) and their respective trend lines are shown. 43 4.6.3 Ethanol determination using G C analysis The column used for ethanol analysis was an AT-Aquawax-DA 30 m column (0.32mm ID, 0.25 um film thickness, Alltech, CA, USA). The program maintained the GC oven at 50°C for 5 minutes, followed by a ramp up to 180°C at a rate of 15°C/minute. This temperature was held for 13.67 minutes. The column pressure was a steady 10 psi and the split ratio was turned off. The injector was set to 150°C and the sample size was the same as for the sugar analysis at 10 uL. Ethanol and butanol were used to make standard concentrations for a calibration curve, at 0.1 g/L, 0.5 g/L, 1.0 g/L and 5.0 g/L. A typical calibration curve is shown as Figure 4.5. 6 -i ! After sampling, the tubes were frozen. When ready for analysis, the tubes were thawed, and centrifuged at 13,000 g (Micro Centuer, Sanyo, IL, USA) for 5 min. A known volume (0.5 mL) was transferred into 1.5 mL Aligent GC vials. The centrifuge tubes with the yeast pellet were discarded. Samples were spiked with 0.1 mL of 10 g/L butanol and 44 brought to a final volume of 1.0 mL using Millipore pure water. A l l dilutions were taking into account during the calculations. 4.6.4 Factorial experimental design The software used for the design and analysis of factorial experiments was JMP IN (version 4.0.3). JMP IN allows a user to investigate multiple experimental parameters at the same time to explore the effects of parameter manipulation in addition to the potential interactive effects between the parameters. The Response Surface Method (RSM) is a design tool within JMP IN recommended for parameter optimization. R S M is an experimental technique that focuses on finding optimal values within a specified range of factors. For each experiment, the central composite design was used, encompassing both center and axial points. This is a more efficient method than the alternative of a full-factorial experiment, which investigates all possible combinations of the parameter levels requiring additional resources. The response (output) is modeled with a curved surface so that the maximum point of the surface (optimal response) can be found mathematically. From this a prediction equation is produced (Equation 4.1). The magnitude and direction of the primary effects or interactive effect between two parameters is described by the coefficient of the term. Individual equations are included within the results and discussion. R = a(x) + b(y) + c(z) + d(xx)+ e(xy) + f(yy) + g(zx) + h(zy) + i(zz) + (intercept) 4.1 where: R = result x, y, and z = coded values (between -1 and 1) for each parameter a through i = coefficients determined by JMP IN. 45 CHAPTER V - RESULTS AND DISCUSSION 5.1 Determining the point of phase separation Determining the volume of water required to induce phase separation was beneficial for further experimental design. This value would be considered the minimum volume used for the extraction experiment. As previously hypothesized, minimal solvent addition is preferred to achieve a maximum levoglucosan concentration in the aqueous solution. The observations (after mixing) are presented below in Table 5.1. The far left column is the volume of water added to 10 mL of room temperature bio-oil. The middle column expresses the total water (internal and additional) as a percentage of the original bio-oil weight. Table 5.1: Titration of water into bio-oil to determine the point of phase separation. Water added to bio-oil Observations (phase separation?) mL wt% 1.0 29 no 2.0 37 no 2.2 39 no 2.4 41 small conglomerates of thick brown globs suspended in solution 2.6 42 larger conglomerates forming, still suspended in solution 2.8 44 no change from 42 wt% 3.0 46 no change from 42 wt% 3.4 49 phases distinctively separated 3.8 52 no change from 49 wt% 4.2 55 no change from 49 wt% The colour of the bio-oil made it difficult to observe the phase separation precisely. Nevertheless, after the fourth increment, equivalent to a weight fraction of 41 wt%, it was evident that some of the dissolved material began to precipitate. This is equivalent to 41 g of water within a 100 g sample of bio-oil. A sample calculation is shown in Appendix B. With additional increments of water, the conglomerate of tar-like material grew until a distinct separation occurred at 49 wt%, after which the size of the organic phase did not 46 seem to change. The range of 41-49 wt% is on the higher end of other published values. According to BTG (2006), most pyrolysis oils separate when the total water is 30-45 wt% of the bio-oil. Oasmaa and Czernik (1999) reported phase separation values lower than this range for wood pyrolysates: 27, 23-25, and 31 wt% for birch, pine and poplar, respectively. The bio-oil density as given by VTT and periodically verified is 1.223 g/mL (+/- 0.1 g/mL). At the interface between the water and the bio-oil, a color change was observed. The dark brown bio-oil immediately changed to a yellow cloudy liquid. When mixed, prior to the point of phase separation, the solution would absorb the yellow color and return to a homogeneous brown liquid. After surpassing the ratio required for phase separation, mixing the solution instigated an immediate precipitation of the water-insoluble components. In this instance, the yellow color remained as the aqueous layer. The phenomenon behind the color reaction was not further investigated but has been described by other authors [Scholze and Meire, 2001]. A distinct separation of the aqueous and non-aqueous layers can be seen in Figure 5.1. Note the transparency of the aqueous layer. The aqueous phase becomes more transparent over time or with centrifugation as the suspended solids settle. The high viscosity of the organic phase is evident when the bottom layer in Figure 5.1 is compared to the tilt of the tube. A dark brown floating layer can be observed on the top of the aqueous layer. Sipila and co-workers (1998) determined similar floating layers in other pyrolysis oils to be derivatives of lignin, namely guaiacol and syringol, using GC-MS analysis. As the lignin structure in pine is predominately formed of guaiacyls [Larsson et al., 1999], one can assume these flakes to be derivatives of this material. 47 Figure 5.1: Bio-oil after phase separation. 5.2 Characterization of the aqueous phase in response to varying water volumes The mass transfer of components from the bio-oil into the aqueous phase and organic phase is affected by the amount of water used to induce phase separation, as illustrated in Figures 5.2 and 5.3 respectively. The first data point in both figures (41 wt%) is equivalent to the point of phase separation, as previously presented in Section 5.1, and is therefore determined as the lower boundary limit for the x-axis. Figure 5.2 shows the total mass of components transferred into the aqueous phase to decrease as the water used for phase separation increases. In other words, using less water is more effective at extracting water-soluble compounds from the bio-oil. This is an interesting and unexpected response. It was originally hypothesized that more water would assist in the fractionation of polar compounds into the aqueous phase. This was proven incorrect. Rather, a decreasing trend is evident. This result requires additional investigation to determine the mechanism behind this unusual phenomenon. 48 65 a c 3 p o a> 3 O O) v° i l l "o 58 -E -o ? a> a> o «-<<)•— «2 w -o E a> «-2 a ° ~ x £ o | '-5 «- 40 60 55 50 45 30 130 230 330 430 530 Total water in sample expressed as weight percent of bio-oil (wt %) Figure 5.2: Mass transfer of bio-oil compounds into aqueous phase in response to varying water volumes. Sipila and co-wo*rkers (1998) reported the fraction of water-soluble components to range from 60-80 wt%, for hardwood, pine and straw pyrolysis oil, respectively; no exact values were given. The volume of water used for phase separation and the original fraction of water in the pyrolysis oils were also not given. An opposing trend is shown for the mass transferred into the organic phase (Figure 5.3). Both data sets can be reasonably fit using a logarithmic trend-line, although the exact equations are not imperative for discussion. Combining the two data sets resulted in a mass closure within 2%. Raw data can be found in Appendix B. 49 50 -c 2 *- 25 30 130 230 330 430 530 Total water in sample expressed as weight percent of bio-oil (wt %) Figure 5.3: Mass of bio-oil compounds in the organic phase in response to various water volumes. The flux of a molecule at the interface between water and another immiscible solvent can be affected by its hydrophobicity. The more hydrophobic a molecule is, the less soluble it is in an aqueous phase. A molecule can change its hydrophobicity in response to a solution pH change. Yet, the pH for each extract was measured and found to be consistent at a pH of 2.0 (+/- 0.02). Rather, some other mechanism is affecting the quantity of compounds transferred into their respective phases. Although the reason is unknown, one can conclude the volume of water is an influential variable on the efficiency of extracting water soluble compounds from the bio-oil into the aqueous phase. Ultimately, an aqueous fraction with the highest concentration of levoglucosan as possible is desired. This can be achieved using minimum volumes. The identity of the compounds being transferred is unknown at this point. A volumetric analysis of each aqueous extract established that the density decreased with respect to an increase of water volume used for phase separation (Figure 5.4). This trend agrees with Figure 5.2: as fewer compounds are transferred into the aqueous phase, the 50 density is correspondingly lower. Initially, the first data point seems erroneous. However, as previously mentioned, a distinctive separation between the phases was not entirely evident. Therefore, the decanted sample of aqueous extract may have contained some small conglomerates of the organic phase. If the volume of water increased to infinity, the density of the aqueous fraction would flatten out at 1 g/mL, the density of pure water. ID 3 O 0) 3 IT n 14-o c V Q 1.8 1.6 1.4 1.2 1 0.8 100 200 300 400 500 600 Total water in sample expressed as weight percent of bio-oil (wt %) Figure 5.4: Density of the aqueous fraction in response to varying water volumes. 5.2.1 Dissolved solids in aqueous phase The density decrease with greater water volume is also validated by a corresponding decrease in dissolved solids transferred into the aqueous phase. Figure 5.5 displays the large change in dissolved solids concentration from 425 g/L to 45 g/L. As expected, the addition of water has a dilution effect on the dissolved solids. Yet one can observe there is another mechanism affecting the quantity of dissolved solids transferred into the aqueous phase other than dilution. If the concentration of dissolved solids was only affected by water, the dilution would be proportional to the amount of water added and the slope of the line would be constant. 51 _ 490 0 100 200 300 400 500 600 Total water in sample expressed as weight percent of bio-oil (wt %) Figure 5.5: Concentration of dissolved solids in the aqueous phase in response to varying volumes of water. To eliminate the effect of dilution, the dissolved solids can be expressed as a weight fraction relative to the bio-oil (Figure 5.6). 52 Figure 5.6: Weight percent of dissolved solids transferred into the aqueous phase in response to varying volumes of water used for'phase separation. Each data point represents the average of two samples. There is no clear relationship between the total water volume and amount of dissolved solids in the aqueous phase. One can conclude the fraction of dissolved solids transferred into the aqueous phase is not dependent on the volume of water used for phase separation. One can also conclude that these dissolved solids are a fairly large proportion of the bio-oil, approximately 20% by weight. This includes mono- and polysaccharides, heavy polar compounds such as lignin solids, and some polar compounds with moderate volatility (boiling points above 105°C) such as phenols and furans. When compared to Figure 5.2, the dissolved solids in the aqueous phase consist of approximately half of the total mass of water soluble compounds. The remaining mass that is not accounted for as dissolved solids would include volatile compounds with normal boiling points above 105°C (temperature of oven) such as low molecular mass carboxylic acids, aldehydes, alcohols, ketones and alkanes. The general flat line in 53 Figure 5.6 compared to the downward slope of Figure 5.2 infers the extraction of some compounds from the bio-oil into the aqueous phase are affected by water volume and others are not. Compounds of interest, such as levoglucosan, glucose and organic acids, are investigated further. 5.2.1.1 Levoglucosan and glucose in aqueous phase Dissolved solids incorporate all non-volatile compounds in the aqueous solution including lignin solids and low-molecular weight compounds such as sugars and acids. The final concentrations and potential partitioning effects due to various bio-oil to water ratios of these individual compounds is of interest in order to optimize the extraction recovery of the desired compounds. The concentrations of both glucose and levoglucosan, as determined by GC analysis, are shown in Figure 5.7. Figure 5.7: Concentration of levoglucosan (dotted line & open squares using the scale on the left) and glucose (solid line & filled diamonds using the scale on the right) in the aqueous phase in response to varying volumes of water used for phase separation. 54 As anticipated, the concentrations of levoglucosan and glucose are diluted when more water is used, ranging from 88 g/L to 5 g/L for levoglucosan and 10 g/L to 0.7 g/L for glucose. Exact values are given in Appendix B. Both compounds exhibit a similar curve; inferring the effect of water volume on the extraction of glucose and levoglucosan into the aqueous phase is the same for both compounds. The largest concentrations are achieved when water volume is at a minimum and the concentrations decrease non-linearly with the addition of more water. When the concentrations are expressed as a weight percentage of bio-oil, the effect of the water on extraction efficiency can be investigated (Figure 5.8 and Figure 5.9). re (0 re ^ re 5 .c ^ a = cn 9 5 .2 S 5 g-o Si s | in O. O O .C u) '55 0 0 100 200 300 400 500 600 Total water in sample expressed as weight percent of bio-oil (wt%) Figure 5.8: Quantity of levoglucosan, expressed as a weight percentage of the bio-oil, transferred into the aqueous phase from the bio-oil in response to varying volume of water used for phase separation. 55 .C 0.5 -O) 1 0.45 -ra 0.4 -w ra a> 1 0.35 -to re sz "5 i 0.3 -a 0 (0 !5 0.25 -3 O 0) o 0.2 -3 c CT re a> P, 0.15 -c 0) 0) a 0.1 -(0 o o 0.05 -3 O 0 -100 200 300 400 500 600 Total water in sample expressed as weight percent of bio-oil (wt%) Figure 5.9: Quantity of glucose, expressed as a weight percentage of the bio-oil, transferred into the aqueous phase in response to varying water volumes used for phase separation. Although the data in Figure 5.7 can be reasonably fit using a power relationship, Figures 5.8 and 5.9 provide more insight without the interference of dilution. When the amount of water used for phase separation exceeds the amount of bio-oil (water to bio-oil ratio greater than 1:1), the percent of levoglucosan and glucose extracted from the bio-oil into the aqueous phase is not greatly affected, holding at approximately 2.5 wt% and 0.3 wt%, respectively. When the water volume is less than the weight of the bio-oil (< 100 wt%), extraction efficiencies were higher. The relationships presented in Figures 5.7 through 5.9 support the conclusion that greater extraction efficiencies and maximum concentrations are achieved when the minimum amount of water is used. Therefore, the optimal amount of water used for the extraction stage would be approximately 41 wt%, the point of phase separation. Expressing the concentrations as a weight percentage of the bio-oil is an attempt to further understand the patterns behind the quantity and quality of compounds transferred 56 into the aqueous phase in response to different solvent volumes. Even i f the extraction efficiency did improve with more water, it would likely not be enough to overcome the dilution effect, therefore for process optimization, minimum water volumes would still be desirable. Nevertheless, with knowledge of extraction efficiency, the ratio of water and bio-oil could be optimized to achieve a desired final concentration. 5.2.1.2 Organic acids in aqueous phase The two organic acids selected for the model, acetic and formic, are quantitatively significant because of their inhibitory effects on yeast growth and fermentation and because they constitute 95% of the water-soluble acids found in softwood pyrolysis oils [Sipila et al., 1998]. Figure 5.10 displays the pH for three aqueous extracts in response to base addition; the actual pH is represented by the data points and a best fit model using a strong-base, weak-acid model (described in Appendix B) is shown as the solid line. 57 o -i > •• i i 1 — • — ' 0 0.2 0.4 0.6 0.8 1 mmol NaOH/mL sample Figure 5.10: Titration curves for three aqueous extracts prepared using different water to bio-oil volumetric ratios for extraction operation: squares =1:1 (water = 104 wt% of bio-oil), diamonds = 2:1 (water = 185 wt% of bio-oil), triangles = 4:1 (water = 359 wt% of bio-oil). A strong-base, weak-acid model using estimated concentrations of formic and acetic acid is shown as the solid lines. The titration model fit well for pH values from 2.5 (starting pH) to 6 using combinations of formic and acetic acid only (pKa's of 3.74 and 4.8, respectively). After pH 6, the actual data deviated from the models for all aqueous extracts. This could be explained by the presence of an unknown compound with a pKa of approximately 8.5. The model plateaus at pH 14, the maximum pH value. The concentrations used in the model that resulted in the good fit shown in Figure 5.10 are expressed as a weight percentage of the bio-oil in Table 5.2. As shown, the values are quite consistent: acetic acid comprised 2.8 wt% (+/- 0.1) and formic acid comprised 2.0 wt% (+/- 0.1); concluding the fraction of formic and acetic acid extracted into the aqueous phase is not affected for these three water and bio-oil combinations. 58 Table 5.2: Acetic and formic acid concentrations for three different aqueous extracts. Water added (mL) Bio-oil (g) Water in bio-oil (mL) Aqueous extract (mL) Total water • (wt%) Acetic acid Formic acid g/L wt % g/L wt% 25.0 28.8 6.0 41.5 : 93 19.0 2:7 15.0 2.2 30.0 16.6 3.5 39.5 184 12.0 2.8 8.0 1.9 40.1 11.4 2.4 46.5 355 7.0 2.9 5.0 2.0 The weight percent values coincide with other published values for organic acids found in wood pyrolysis oil [Helle et al., 2003; Taherzadeh et al., 1997]. No effect of water to bio-oil ratio is seen for these three data points. Although, referring to the trend observed for glucose and levoglucosan extractions in Figures 5.8 and 5.9, it is recommended to determine the organic acid concentration in aqueous extracts produced using water amounts less than 100 wt% to see if a similar jump in extraction effectiveness occurs. 5.3 Determining parameter effects on levoglucosan extraction In an attempt to further improve the extraction of levoglucosan into the aqueous phase, two processing parameters, temperature and time, were added to the previously investigated parameter of water volume in a factorial experiment. The results in Table 5.3 display the large range of levoglucosan concentrations found in the various aqueous extracts, from 11 g/L to 87 g/L. This is primarily due to the different volumes of water used (dilution). When the amount of levoglucosan transferred into the aqueous phase is expressed as a weight percent of the bio-oil, the dilution factor is removed and the effects of the parameters are more evident. 59 Table 5.3: Levoglucosan in the aqueous phase expressed as concentration (g/L) and as weight percent of bio-oil (wt%) in response to combinations of various time, temperature, and water values. Sample Number Time (min) Temp (°Q Total water (wt%) Levoglucosan (g/L) (wt%) 1 0 4 62 78.0 7.0 2 0 4 103 12.3 2.6 3 40 19 185 21.1 2.7 4 0 34 103 12.68 2.6 5 0 19 185 19.1 2.5 6 40 4 62 57.0 5.0 7 20 19 185 25.0 3.2 8 0 34 62 81.6 7.3 9 40 34 62 87.3 7.8 10 20 19 62 13.8 1.2 11 20 34 185 26.8 3.5 12 20 4 185 21.1 2.7 13 40 34 103 12.8 2.7 14 20 19 103 12.0 2.5 15 20 19 185 19.3 2.5 16 40 4 103 11.2 2.3 For unknown reasons, likely human error, sample vial 10 had a high deviation from the other results and was eliminated from analysis. The Grubb's test could not be used to determine outliers within this data set as the values do not fit a normal distribution curve as they are dependant on the conditions used. For example, although the Grubb's Z-value for sample 10 (1.19) was below the critical Z-value (2.59 at P = 0.05), meaning it was within the allowed deviation of the sample set, it can be viewed as erroneous when put into the context of the conditions (i.e. when compared to other values for total water of 62 wt%). With sample 10 omitted from analysis, the other 15 data points provided a correlation coefficient of 0.98 (Figure 5.11). 60 2 3 4 5 6 7 8 LG (wt%) Predicted P=0.0010 RSq=0.98 RMSE=0.4622 Figure 5.11: Correlation between actual and predicted amounts of levoglucosan transferred into the aqueous phase (15 data points). Prediction profiles (Figure 5.12) help to illustrate the parameter effects on the amount of levoglucosan transferred into the aqueous phase. If there are significant secondary (interactive) effects between the parameters, the profiles would be unique for each combination of parameters. If there are no secondary effects, the profiles would remain as is for every combination and the direct effects can be individually interpreted from the slope of the curves. JMP IN software allows the user to manipulate one parameter at a time by moving the vertical dashed bar and the resulting interaction effect can be visualized by a change in the other profiles. For the purposes of this project, the predictive profiles shown are a result of an optimal combination of parameters (solved mathematically by JMP IN) that produce a maximum weight percent of levoglucosan transferred (y-axis). The x-axis scale (-1 to 1) is the coded value for the parameters. The mathematical optimums for each parameter as well as the desired outcome are displayed in the middle of each axis. Refer to Table 4.4 for the full interpretation of the x-axis for Figure 5.12. 61 8.935 7.781102 1.4589 '—i i" i— i i i ' i i i—' 7 7 0.10409 ^ V 1 *~ T -1 "~ Time Temp Water Figure 5.12: Prediction profiles for the parameter effects of time, temperature and water on the amount of levoglucosan transferred into the aqueous phase. Within the given range of conditions, a maximum amount of levoglucosan, 7.8 wt%, was transferred using a combination of a medium time (0.10409), maximum temperature (+1), and minimum water (-1). This is interpreted as 22 minutes, 34°C, and a total water amount equalling 62 wt% of the bio-oil. Please note the significant digits for the numerical values presented in the JMP IN graphs cannot be adjusted. Where possible, the values are rounded to the first decimal point. The profile for the water parameter follows the same trend presented in Figure 5.8, although Figure 5.12 is determined from more data points for a smaller range and thus can be interpreted as more reliable for water values extending from 62 wt% to 185 wt%. The curve predicts that a maximum extraction efficiency of levoglucosan from the bio-oil into the aqueous phase is achieved when minimum water is used. This is limited by the boundary condition of phase separation, earlier reported at approximately 41 wt%. Based on the slope of the prediction profile, one can extend the range below -1 and hypothesize that the absolute maximum concentration of levoglucosan in the aqueous phase would be achieved at this boundary point of phase separation. Quantities presented in Figure 5.8 are similar to what is presented here: between 2.5 wt% and 3.0 wt% for the majority with 62 a large jump to higher extraction values (5.0-7.8 wt%) when the water volume was minimized. The effect of temperature is small in comparison to the effect of the water (f-ratios = 8 and 173 for temperature and water, respectively, with an f-critical value of 2.4 at P = 0.05). Nevertheless, an increase in temperature did increase the amount of levoglucosan transferred into the aqueous phase. Note the increase in the extraction efficiency from 5.02 wt% for sample 6 (4°C) to 7.81 for sample 9 (34°C) in otherwise similar conditions. It is probable that an increase in temperature allows the bio-oil to mix more readily, as it is less viscous, assisting the levoglucosan to more freely extract into the aqueous phase. Over the range examined, time had very little effect on the extraction of levoglucosan, as shown by the generally level curve in Figure 5.12. Mathematical models determined by JMP IN use the coded values for x, y, and z (between -1 and 1). Equation 5.1 and 5.2 present the prediction equations for high concentration and high weight percent, respectively. Only terms considered statistically significant (f-value is greater than f-critical for P = 0.05) are included in the model. L G (g/L) = 22.03 + 4.24^ -31 .76z + 3.62xy - 4.12zy + 21.86z2 (5.1) L G (wt%) = 2.83 + 0.42^ - 2.12z + 0.34xy - 0.34zy +1.8 l z 2 (5.2) where: x = coded value for time (minutes) y = coded value for temperature (°C) z = coded value for water (wt% of bio-oil) Interaction effects are only seen between the parameters of time and temperature as well as temperature and water. 63 5.4 Hydrolysis of levoglucosan into glucose Three aqueous extracts, using the same water values as Section 5.3 (62 wt%, 103 wt% and 184 wt%), henceforth referred to as Extract A , B, and C, were prepared using room temperature water and bio-oil and no reaction time (mix, shake, spin, decant). Optimal temperature and time conditions to achieve maximum levoglucosan concentrations in the aqueous phase, as determined in Section 5.3, were not applied. The aqueous extracts were then subjected to hydrolysis conditions of 120°C, 0.5M sulphuric acid, for 60 minutes. Samples were removed every twenty minutes. The resulting concentrations of L G and glucose, as determined by GC, are displayed in Figure 5.13. Figure 5.13: Glucose (open symbols) and L G (filled symbols) concentrations for aqueous extracts A (triangles), B (squares) and C (diamonds) using hydrolysis conditions of 120°C and 0.5 M H 2 S0 4 over a period of 60 minutes. 64 As expected, and confirmed with previous observation, levoglucosan concentrations are smaller when more water is used for phase separation, although the actual values are different than what was reported in Figure 5.7 using the same conditions (70 g/L, 22 g/L, and 17 g/L for Section 5.2.1.1 vs. 37 g/L, 22 g/L, and 8 g/L for this experiment). The initial concentration of levoglucosan appears to have an effect on the hydrolysis rate in spite of identical temperature and acid conditions; the greater the initial concentration, the faster the hydrolysis rate. With more experimentation, this could potentially be quantified as another parameter in the Saeman-Arrhenius equation (Equation 2.2). The levoglucosan in Extracts B and C was completely hydrolyzed between 35 and 40 minutes. Although there is no data point to support this statement, it is likely the levoglucosan in Extract A was completely hydrolyzed before 20 minutes. For all three aqueous extracts, the formation of new levoglucosan is apparent. In Extract C, the final L G concentration actually increases to a value greater than its final glucose concentration. The formation of new L G is likely a result of other cellulose compounds degrading under the hydrolysis conditions. This reaction has not been previously documented. The formation of new glucose follows a curve similar to the disappearance of levoglucosan for the first 20 minutes of the reaction. Thereafter, glucose concentrations in Extracts A and C continue to increase at a slower rate until a plateau at approximately 40 minutes. The glucose concentration in Extract B appears to increase at a steady rate until the final sample point. Final glucose concentrations for all three aqueous extracts result in maximum hydrolysis yields greater than 100% (170%., 154%o and 124% for Extracts A , B and C, respectively; based on levoglucosan consumption), inferring the contribution of other compounds to the formation of glucose, such as cellobiosan. These other oligosaccharides could potentially be the unknown peaks in the HPLC chromatograph, Figure 4.3. Furthermore, with respect to the different aqueous extracts, the source of additional glucose is greater 65 when less water is used for the phase separation. This trend (greater extraction using less water) follows the same response as glucose and levoglucosan. Although no glucose degradation can be seen in Figure 5.13, the rate of degradation could be off-set by the formation of glucose from other sources. Nevertheless, as no degradation is observed, conditions can be considered not too severe. Figure 5.14 shows a simplified model proposed for this overall reaction. Polysaccharides . Levoglucosan . Glucose Degradation compounds k, k 2 k 3 Figure 5.14: Proposed reaction of levoglucosan to glucose during hydrolysis. Glucose hydrolysis yields for the three controls (water was used instead of acid) after 60 minutes were 10%, 7% and 0%, based on initial levoglucosan concentrations, for Extracts A, B, and C, respectively. These small conversion yields can likely be attributed to the high temperature and general acidic conditions of the aqueous extract. 5.5 Determining the effect of time, temperature and acid concentration on levoglucosan hydrolysis into glucose Aqueous extracts used in the following hydrolysis experiments were prepared using water volumes equivalent to the weight of the bio-oil (100 wt%). 5.5.1 Hydrolysis conditions A Final glucose concentrations and yields for hydrolysis conditions " A " are shown in Table 5.4. The ranges selected for this initial investigation are guided by the optimal conditions found by Y u and Zhang (2003): 20 minutes, 121 °C and 0.2M sulphuric acid. A l l yields are based on absolute consumption (Glufinai-Gluinitiai / LGinit ia i -LGnnai) unless 66 identified otherwise. A large range of values resulted from the different parameter combinations. The last data point, labelled as sample 17, provides the L G and glucose concentrations before hydrolysis. Concentrations were measured by GC analysis. Table 5.4: Concentrations of levoglucosan and glucose (g/L) after hydrolysis conditions A, also expressed as percent yield. Sample Temp Time Acid L G Glucose Yield Number r o (min) (M) (g/L) (g/L) (%) 1 90 10 0.1 16.6 3.1 -7.8 2 105 10 0.3 10.5 8.3 19.9 3 120 10 0.5 1.2 22.4 61.2 4 105 40 0.3 3.6 19.8 56.9 5 90 40 0.1 12.9 5.6 7.4 6 90 25 0.3 8.8 7.7 15.5 7 105 25 0.5 0.7 24.6 67.2 8 105 25 0.1 10.3 8.3 19.4 9 90 40 0.5 4.0 16.3 44.8 10 120 40 0.5 0.9 24.6 67.7 11 105 25 0.3 1.3 23.2 64.0 12 105 25 0.3 2.7 14.1 34.9 13 120 10 0.1 10.8 8.4 20.4 14 120 40 0.1 1.9 23.6 66.4 15 90 10 0.5 12.4 5.4 6.3 16 120 25 0.3 1.1 25.0 69.4 17 0 0 0 30.9 4.2 A decrease in both levoglucosan and glucose resulted in the negative yield for sample 1. Due to the low severity of the conditions for sample 1, it is unlikely that degradation occurred; rather it could be experimental error. Nevertheless, sample 1 was still used in the statistical analysis. Sample 11, however, was not included for an obvious lack of fit due to experimental error. This time the data point was included in the figure to allow a visualization of the deviation (circled in Figure 5.15). Unfortunately, the multiple steps involved in derivitizing the sugars in the hydrolysates into compounds detectable by the GC create a large risk of error. The other 15 data points resulted in a reasonably good correlation between the predicted and actual data values (R = 0.91). 67 -25 0 25 50 75 100 Yield (%) Predicted P=0.0379 RSq=0.91 RMSE=13.567 j Figure 5.15: Correlation between actual and predicted amounts of levoglucosan hydrolyzed into glucose using hydrolysis conditions A for 15 data points (sample 11, the circled data point was not included in analysis). The effect of each process parameter on levoglucosan hydrolysis over the investigated ranges is shown in Figure 5.16. The optimal combination of parameters that resulted in the maximum yield of 81% is also shown. The coded values, 1.0, 0.92, and 1.0, represent 120°C, 39 minutes and 0.5M sulphuric acid, respectively (refer to Table 4.6). 111.9 g. 80.67408 -7.804 1 1 0.91622 ' 1 Temp Time Acid Figure 5.16: Prediction profiles for hydrolysis conditions A. 68 Complete hydrolysis conversions (glucose yields of 100%) were not achieved, contrary to greater than 100% yields recorded in Section 5.4, in spite of similar conditions. The only difference between the experiment in Section 5.4 and this experiment is the initial L G concentration. This experiment began with L G concentrations of 31 g/L; whereas the previous experiment started with lower concentrations ranging from 8.4-36.5 g/L. Once again, this suggests an effect of initial levoglucosan concentration on the reaction. Although it was believed an optimal combination of parameters would be found within the investigated ranges, based on Yu and Zhang (2003) as well as the first hydrolysis experiment (Section 5.4), these prediction profiles do not show any maximums; therefore conditions can be improved upon. The profile trends project that higher (more severe) conditions would lead to higher hydrolysis yields, especially for temperature and acid concentration. Due to potential interactive effects at higher conditions, one cannot confidently predict the results beyond the investigated range. The yield of glucose from levoglucosan is quantitatively modeled as Equations 5.3. Only statistically significant terms are included. Yield (%) = 42.38 + 21.89x +14.3 ly +14.15z (5.3) where: x = coded value (between -1 and 1) for temperature (°C) y = coded value (between -1 and 1) for time (minutes) z = coded value (between -1 and 1) for acid concentration (M) The model reveals there are no interaction effects. Statistically, temperature was the most significant; followed by an equal contribution of time and acid concentration (f-values were 26, 11 and 11, respectively). 69 5.5.2 Hydrolysis conditions B In an attempt to eliminate the challenge of interpreting results with too many manipulated variables, hydrolysis conditions B retained the same time and temperature ranges. It was hypothesized an optimum time would be found between 15 and 40 minutes with an increase in hydrolysis severity which would be achieved by an increase in acid concentration only. Section 5.5.1 demonstrated temperature has the greatest affect on hydrolysis yield. Potentially even a small increase in temperature could lead to high sugar degradation rates. The glucose and levoglucosan concentrations, as well as the hydrolysis yield, produced from combinations of hydrolysis conditions B are shown in Table 5.5, followed by the prediction profiles and model equation. The initial levoglucosan and glucose concentrations (before hydrolysis) are labelled as sample 17. Once again, a large range of values was obtained, producing hydrolysis yields from 0% to 78%. Yields are based on absolute consumption. 70 Table 5.5: Final concentrations of levoglucosan and glucose (g/L) and hydrolysis yield (%) resulting from hydrolysis conditions B, using initial concentrations of 22.2 g/L and 2.7 g/L, respectively. Sample Number Temp (°C) Time (min) Acid (M) L G (g/L) Glucose (g/L) Yield (%) 1 90 10 0.2 10.2 2.6 -0.7 2 105 10 0.6 7.0 7.2 29.7 3 120 10 1.0 0.6 18.8 74.8 4 105 40 0.6 1.2 18.1 73.6 5 90 40 0.2 8.3 6.0 24.1 6 90 25 0.6 6.5 6.6 25.3 7 105 25 1.0 0.7 17.9 70.6 8 105 25 0.2 4.1 11.0 46.1 9 90 40 1.0 3.3 12.2 50.5 10 120 40 1.0 0.4 19.8 78.5 11 105 25 0.6 1.7 15.8 64.1 12 105 25 • 0.6 1.5 16.0 ., 64.6 13 120 10 0.2 3.2 ' 12.5 52.0 14 120 40 0.2 0.5 19.6 78.2 15 90 10 1.0 7.7 4.0 9.5 16 120 25 0.6 0.4 19.1 75.6 17 0 0 0 22.2 2.7 A l l sixteen data points were used in the JMP IN analysis. Figure 5.17 shows the good correlation between the predictive model and the actual data (R 2 = 0.96). This is immediately followed by the prediction profiles in Figure 5.18, using an optimum combination of parameters. 71 -20 0 20 40 60 80 100 Yield (%) Predicted P=0.0019 RSq=0.96 RMSE=8.5354 Figure 5.17: Correlation between actual and predicted amounts of L G hydrolyzed into glucose using hydrolysis conditions B (16 data points). 100 2 CD •o 85.41906 CD E in c 8 o i i I I — i I I i i T 0.87984 " T 0.36122 T 1 Temp Time [Acid] Figure 5.18: Prediction profiles for hydrolysis conditions B. A n optimal combination within the investigated range resulted in a yield of 85%. With the range investigated, an optimal combination of parameters was determined to be 0.9, 0.4, and 1.0; interpreted as 118°C, 30 minutes and 1.0M sulphuric acid, respectively (Table 4.6). This set of conditions would produce a maximum hydrolysis yield of 85%. Even though the highest acid concentration in this experiment is double the highest 72 concentration for conditions A , still no optimum/maximum was found for this parameter. The increase in acid concentration did affect the temperature and time variables as both profiles are shown to taper off near the high end of the range. A sulphuric acid concentration of 1.0 M is considered concentrated; therefore an even greater concentration may be unreasonable due to the expense of the neutralization and acid recovery costs. Furthermore, it should be noted the increase in acid concentration from 0.5 M to 1.0 M only improved the hydrolysis yield by 5%. An economic analysis would be required to evaluate the viability of manipulating these variables. The predictive model is shown as Equation 5.4. Only significant terms are included. Yield (%) = 60.45 + 25.03* +\3.96y + 8.42z-8.06x 2 -4.49xy (5.4) where: x = coded value (between -1 and 1) for temperature (°C) y = coded value (between -1 and 1) for time (minutes) z = coded value (between -1 and 1) for acid concentration (M). Note the new interactive term between temperature and time, although at a much lower magnitude than the coefficient of the direct term. Similar to hydrolysis conditions A , all three parameters have a direct effect on the levoglucosan conversion yield. Temperature is the most statistically significant, with an f-ratio of 86. It is followed by time and acid concentration, with f-ratios of 27 and 10, respectively. An error analysis between the predicted and actual data is presented in Table 5.6. 73 Table 5.6: Variance and percent error between predicted and actual data for hydrolysis conditions B using Equation 5.4. Sample Temp (x) Time (y) Acid (z) Actual yield (%) Predicted yield (%) Absolute difference Percent error (%) 1 -1 -1 -1 -0.7 0.5 0.2 32.6 2 0 -1 0 29.7 46.5 16.8 56.5 3 1 -1 1 74.8 76.4 1.6 2.1 4 0 1 0 73.6 74.4 0.8 1.1 5 -1 1 -1 24.1 37.4 13.3 55.1 6 -1 0 0 25.3 27.4 2.0 8.0 7 0 0 1 70.6 68.9 1.7 2.4 8 0 0 -1 46.1 52.0 5.9 12.9 9 -1 1 1 50.5 54.2 3.8 7.5 10 1 1 1 78.5 95.3 16.8 21.5 11 0 0 0 64.1 60.5 3.6 5.6 12 0 0 0 64.6 60.5 4.1 6.4 13 1 -1 -1 52.0 . 59.5 7.6 14.5 14 1 1 -1 • 78.2 78.5 0.3 0.4 15 -1 -1 1 9.5 17.3 7.8 82.2 16 1 0 0 75.6 77.4 1.8 2.4 The greatest difference occurred for sample 10, the sample that was exposed to maximum values (1,1,1); however there does not appear to be any pattern between the error and the combinations of parameters. 5.5.3 Hydrolysis conditions C In an attempt to achieve full levoglucosan conversion (yield = 100%) as well as determine a sufficient acid concentration, the last factorial experiment for hydrolysis conditions focused on higher temperatures and a narrow acid concentration range from 0.5 to 1.0 M . In addition, the time range was extended to 60 minutes, not only to ensure a complete reaction but to potentially identify the point at which sugar degradation occurs. The results including the final glucose concentration and the hydrolysis yield are presented in Table 5.7. Yields are calculated as before; incorporating both initial and final concentrations of glucose and levoglucosan. One can see an improvement in hydrolysis yield ranging from 66% (sample 15: 1,1,1, apparent sugar degradation) to 220% 74 (sample 8: 0,0,-1). Original L G and glucose concentrations were determined as 9.83 g/L and 2.45 g/L, respectively. A l l data was measured using GC analysis. Table 5.7: Final concentrations of levoglucosan and glucose as well as hydrolysis yield using combinations of hydrolysis conditions C. Sample Time Temp Acid L G Glucose Yield number (min) (°Q (M) (g/L) (g/L) ( % ) 1 20 110 0.5 1.8 12.6 127 2 60 110 0.5 0.2 22.0 202 3 40 110 0.75 0 19.3 171 4 40 150 0.75 0.2 14.8 128 5 20 130 0.75 0 20.7 185 6 40 130 0.75 0 23.3 212 7 60 110 1.0 0 23.6 215 8 40 130 0.5 0 24.0 220 9 20 150 1.0 0.2 15.1 131 10 40 130 1.0 0.2 20.3 183 11 20 150 0.5 0 19.2 171 12 60 130 0.75 0 18.7 166 13 60 150 0.5 0.3 12.5 104 14 20 110 1.0 0 15.4 132 15 60 150 1.0 0 8.9 66 16 40 130 0.75 0 12.3 100 ~* The higher temperatures lead to values much higher than previously recorded; the majority of the samples exhibited hydrolysis yields greater than 100%. It was noticed that there is no correlation between the various hydrolysis experiments. For example, in Section 5.5.1 for hydrolysis conditions A , a combination of 120°C, 0.5M sulphuric acid and 40 minutes converted 67% of the levoglucosan. Extract C in the first hydrolysis investigation (Section 5.4), using the identical conditions, gave a hydrolysis yield of 124%. Very similar conditions in this trial, 130°C, 0.5 M , and 40 minutes, resulted in a yield of 220%. Once again, perhaps the initial levoglucosan concentration is the reason behind the different results. This is further investigated and discussed in Section 5.6. A l l but one data point was used for the statistical analysis. Sample 14 had a high standard deviation and is evidently erroneous when compared to the rest-of the data points (circled 75 in Figure 5.19). The other 15 data points were successfully used to produce a model with a valid correlation coefficient between the predicted and actual data (R = 0.97). Yield (%) Predicted P=0.0024 RSq=0.97 RMSE=12.741 Figure 5.19: Correlation between actual and predicted amounts of L G hydrolyzed into glucose using hydrolysis conditions C (without sample 14, circled). The prediction profile for each variable is shown in Figure 5.20 for a combination of conditions that produces a maximum yield. 240.2 6 216.1741 65.841 0.21686 ' -0.2341 ' -1 Time Temp Acid Figure 5.20: Prediction profiles for the effects of time, temperature and acid concentration on hydrolysis yield using hydrolysis conditions C. 76 An optimum combination of parameters within the investigated range was determined to be 0.2, -0.2 and -1 for time, temperature and acid concentration, respectively. This translates into 44 minutes, 125°C and 0.5 M sulphuric acid. This will give a yield of 216%, based on consumed levoglucosan. At these acid and time values, glucose degradation is modeled for temperatures higher than 125°C. The profiles shown in Figure 5.20 are unique to the selected combination of conditions as strong secondary (interactive) effects exist between the parameters. The strongest effect is apparent between temperature and time. For example, if temperature was set to high (temp = 1), the time curve would slope down. If temperature is set to low (temp = -1), the time curve would slope up. The interactive profiles are shown in Figure 5.21; i f the lines cross, the interactive effect is significant. 77 From the figure, one can see large interactive effects between temperature and time. A less significant effect between temperature and acid is also shown. This agrees with the magnitude of the terms included in the model (Equation 5.5). Yield (%) = 197.45 -24.87^ -9.67z -18.28x 2 -36.19xy - 4 4 . 2 3 / -12Mzy (5.5) where: x = coded value (between -1 and 1) for time (minutes) y = coded value (between -1 and 1) for temperature (°C) z = coded value (between -1 and 1) for acid concentration (M). 5.5.4 Application of combined severity model Even after investigating a multitude of variable combinations, an absolute optimum was not found. Therefore, in an attempt to help with future experimental designs the results were applied to the combined severity model. The combined severity factor was calculated for each set of hydrolysis conditions and plotted against the hydrolysis yield obtained using that combination. Figure 5.22 displays the results for hydrolysis conditions A, B and C. 78 Figure 5.22: G l u c o s e y i e l d (%) as a function o f c o m b i n e d severity us ing hydrolys is condit ions A (diamonds) , B (squares) and C (triangles). Results f r o m h y d r o l y s i s condit ions A and B effectively amalgamated together to produce a linear trend for c o m b i n e d severity values ranging f r o m 0.5 to 1.5. A b o v e CS values o f 1.5, the trend plateaus at a m a x i m u m hydrolys is y i e l d o f a p p r o x i m a t e l y 85%. Results f rom hydrolys is condi t ions C d i d not correlate w i t h the other data or w i t h i n itself. It is hypothesized the m o d e l is not applicable to condit ions that derive y i e l d s greater than 100%o as other, para l le l reactions are occurr ing. The extreme difference i n y i e l d s between the first two and the last experiment c o u l d once again point to the differences i n i n i t i a l levoglucosan concentrat ion. I f w o r k i n g w i t h lower i n i t i a l L G concentrations (<10 g/L), one can infer a CS va lue f r o m 1.2 to 2.5 w o u l d suff iciently convert a l l o f the L G into glucose. H y d r o l y s i s c o n d i t i o n s A and B began wi th an i n i t i a l L G concentrat ion o f 31 g/L and 22 g/L, respectively. F o r this range, CS values around 1.5 are o p t i m a l . N o t e for CS values higher than 3, the glucose y i e l d has decreased, in ferr ing these severe condit ions lead to glucose degradation. Larsson and coworkers (1999) reported opt imal glucose y i e l d s f r o m the h y d r o l y s i s o f dry w o o d us ing a c o m b i n e d severity factor o f 3. It is understood that condi t ions for dry w o o d 79 hydrolysis would have to be more severe than levoglucosan hydrolysis due to crystalline structure and protective lignin cover. 5.6 Effects of initial levoglucosan concentration and acid concentration on hydrolysis yields Based on the variability observed in previous experiments, this factorial design was conducted in an attempt to analyze the effect of initial levoglucosan concentration on hydrolysis yield. The concentration of levoglucosan is being used as a term to represent the general concentrations of other compounds in the aqueous phase that may be affecting the reaction. A second parameter, acid concentration, was included for interest purposes. It was assumed that at lower L G concentrations, less acid would be required for complete hydrolysis. Results for the ten samples are presented in Table 5.8. Temperature and time were maintained at 110°C and 40 minutes as guided by previous experimentation. To remove the obvious effect of initial L G concentration on the final glucose concentration, only the hydrolysis yield is presented. Table 5.8: Hydrolysis yield as a result of various combinations of levoglucosan and acid concentrations. Sample [LG] [Acid] Yield number (g/L) (M) (%) 1 29 0.15 30 2 123 0.15 36 3 47 0.15 47 4 47 0.45 49 5 123 0.45 38 6 123 0.75 36 7 47 0.45 49 8 29 0.45 33 9 47 0.75 55 10 29 0.75 34 A l l ten data points were used in JMP IN. The model correlated well with the actual data, as shown in Figure 5.23 (R 2 = 0.97). 80 yield% Predicted P=0.0031 RSq=0.97 RMSE=2.1149 Figure 5.23: Correlation between actual and predicted amounts of levoglucosan hydrolyzed into glucose using different initial levoglucosan and acid concentrations. As shown in Table 5.8, three out of the four top yields were produced using the medium levoglucosan concentration of 47 g/L. This result is evident by the concave curve (inverse-U) shown in the prediction profile (Figure 5.24). 56.54 g . 52.29543 28.037 Figure 5.24: Profile of hydrolysis yield in response to various initial L G concentrations and acid concentrations. 81 Within the ranges investigated, the optimal combination is determined as (0,1) for levoglucosan concentration (g/L) and acid concentration (M), respectively. This corresponds to 47 g/L and 0.75M and produces a hydrolysis yield of 52%. The mechanism behind this observation was not further investigated but seems to be a valuable pursuit based on the large improvement in yield (>20%) when conditions are optimized. The hydrolysis yields generally match previous experiments for temperatures around 110°C. Based on the temperature and time conditions, no optimum for acid concentration was found within the investigated range. The prediction profile forecasts that higher acid concentrations would lead to higher hydrolysis yields. Contrary to the initial hypothesis, no interactive effect between L G and acid concentration was found; therefore one does not have to adjust the acid concentration in response to the selected levoglucosan concentration. Hence, there is no xy term in the mathematical model (Equation 5.6). Yield (%) = 50.22 + 2.26* + 2.0\y - 15.67x2 (5.6) where: x = coded value (between -1 and 1) for temperature y = coded value (between -1 and l).for acid concentration. Both levoglucosan and acid concentration had a direct and significant effect on the yield, as shown by Equation 5.6. The profiles and equation predict optimal hydrolysis yields occur when levoglucosan concentrations are approximately 50 g/L. Concentrations that are higher or lower than this reduce the hydrolysis yield. Unfortunately, it is challenging to compare the results to previous experiments based on the variety of temperature, acid concentration and time combinations. The results agree with Section 5.4 where the hydrolysis yields improve with increasing initial L G concentrations from 8 to 37 g/L. However, the results contradict the trend observed from hydrolysis experiments A, B, and C where the hydrolysis yields increase with a decreasing concentration from 10 to 31 g/L. Although the optimum levoglucosan concentrations will likely change with different temperature and time conditions, what can be confidently concluded from this experiment 82 is that the initial L G concentration (or, a compound correlated to the L G concentration) has a significant effect on the hydrolysis yield. It is recommended that this experiment be investigated further. 5.7 Fermentation of aqueous extracts The fermentation experiments were chronologically conducted first to ensure the project could be further pursued based on the level of toxicity. The pH of the aqueous extract was always adjusted to 5.5 before inoculation as the original pH of the extract (2.5) would be too acidic for the yeast to survive. Results shown are the average of two flasks. The aqueous extracts were prepared using a water volume equivalent to 100 wt% of the bio-oil. As shown in Figure 5.25, little to no ethanol was produced, even with the smallest addition of aqueous extract, confirming an extremely high level of toxicity. 0 5 10 15 20 25 30 Time (h) - • - 2 0 % * 40% •••€!60% - K - 8 0 % - • - 1 0 0 % -•-control Figure 5.25: Effect on ethanol production of incremental fractions of aqueous extract from pine pyrolysis oil ranging from 0 to 100% (volume of extract / total volume). 83 After the first sample (time = 3 hours), a small concentration of ethanol is observed. It is interesting to note the flask with media prepared from 100% pyrolysate produced the highest ethanol concentration by the end of the sampling period (1 g/L). Therefore, one can assume fermentation is occurring; however the yields are completely impractical and evidently extremely inhibited by components within the bio-oil. Compounds interfering with yeast metabolism include acetic acid, formic acid and a high concentration of other dissolved solids [Palmqvist and Hahn-Hagerdal, 2000a]. It has been reported that acetic acid will completely inhibit growth of S. cerevisiae i f the concentration of undissociated acid exceeds 5.0 g/L [Taherzadeh et al., 1997]. Helle and coworkers (2003) observed that pronounced effects on fermentation occur at smaller concentrations (1.5 g/L). For formic acid, Olsson and Hahn-Hagerdal (1996) report growth inhibition at concentrations of 2.7 g/L. Other extractives such as terpenes, alcohols, lignin derivatives and aromatic compounds such as tannins result in additional toxicity [Olsson and Hahn-Hagerdal, 1996]. These compounds likely contribute to the yellowish-brown color of the aqueous extract. The second fermentation run focused on fractions of less than 20% v/v to narrow the range of investigation. The results, shown in Figure 5.26, exemplify exactly how toxic the conditions are: the addition of as little as 5% aqueous extract to the nutrient rich media completely inhibits fermentation. 84 Figure 5.26: Effect on ethanol production of incremental fractions of aqueous extract from pine pyrolysis oil ranging from 0 to 20% (volume extract / total volume). Aside from the control, only the smallest fraction (2%) obtained near theoretical yields. After 22 hours, the final ethanol concentration for the 10% flask reached the same value as in the 5% flask, inferring a more prominent lag with higher fractions of aqueous extract, yet no effect on final ethanol yield. This phenomenon is more obvious in other experiments and is discussed in Section 5.8.1 in response to Figure 5.27. Based on available glucose, yields were found to be greater than a theoretical value of 51% (g/g) (Table 5.9). For the flask containing 2% aqueous extract, the additional ethanol concentration could be from alternate fermentable sugars, such as mannose. It is not known why the yield is greater than theoretical for the control. 85 Table 5.9: Ethanol yields of flasks containing incremental fractions of aqueous extract in nutrient rich media. Aqueous Yield Extract (%) (% g/g) 0 63 2 57 5 4 10 1 15 0 20 0 5.8 Fermentation techniques To overcome inhibitory conditions, a number of modifications to the standard fermentation methodology were applied. For these experiments, the extracts were hydrolyzed; therefore higher ethanol concentrations should be obtained. However, the hydrolysis conditions will have formed new inhibitory compounds such as furfurals that will hinder yeast fermentation. The hydrolyzed pyrolysis extract is referred to as hydrolysate. In all experiments, the pH of the hydrolysate was initially adjusted to 5.5. Over the course of fermentation, it decreased to as low as 4.7. The change in pH is due to the side-production of organic acids during fermentation. At these pH values, inhibition due to acetic acid alone is not likely. 5.8.1 Fermentation using micro-aerophilic conditions Previous work from Helle and Duff (2004) recorded improved fermentation rates in toxic hydrolysates when using micro-aerophilic conditions. This was achieved by restricting air exchange to a small needle in a rubber septa. Glycolysis under anaerobic conditions produces less energy per glucose molecule than complete aerobic oxidation, increasing the flux through the pathway, pushing the reaction faster and allocating resources to anaerobic fermentation rather than cellular growth [Wikipedia, 2006]. Figure 5.27 shows an improvement in the final ethanol yield after 26 hours for the vial with 5% hydrolysate. 86 Vials with fractions greater than 5% hydrolysate were still too inhibitory for ethanol production. 10 15 Time (hours) 20 25 30 -•-2.5% * 5% 10% 15% 20% control Figure 5.27: Ethanol production using fractions of hydrolyzed aqueous extract from pine pyrolysis oil ranging from 0 to 20% (v/v) under micro-aerophilic conditions. Overall, Figure 5.27 displays a different fermentation profile than Figure 5.26. Firstly, there are two additional data points in Figure 5.27 which more accurately describe the fermentation during the exponential growth phase, versus the smooth best-fit curve in Figure 5.26. A second reason for the difference is that this response is from micro-aerophilic conditions, compared to semi-aerobic conditions for the previous figure. A rather significant lag phase can be seen for all vials. It is evident that the lag phase is greater for the vials with a greater hydrolysate fraction. Helle and Duff (2004) reported a pronounced lag phase while fermenting softwood hydrolysates while furfural is being metabolized into furfuryl alcohol by S. cerevisiae (Figure 5.28). In this experiment, once the initial lag phase was overcome, ethanol was produced at exponential rates for the 87 control, 2.5% and 5% vials. This agrees with Olsson and Hahn-Hagerdal (1996) who report that the inhibitory effects disappear once furfural is metabolized. 0 Figure 5.28: Metabolism of furfural into furfuryl alcohol. In Figure 5.27, the 2.5% vial surpassed the control at 22 hours. At 26 hours, the concentration of ethanol in the 5% vial was higher than both the 2.5% vial and the control. With higher fortifications, higher final ethanol concentrations are expected; nevertheless the rate of increase in fermentation is noteworthy. To investigate this further, productivity (the rate of ethanol production in g/L per hour) is presented in Figure 5.29. Note that each data point is determined by dividing the ethanol concentration at the point of sampling by the sampling time. 88 0.6 5 10 15 20 25 30 Time (h) -a -2.5% 5% - e - 10% HK -15% - 9 - 2 0 % -^-control | Figure 5.29: Ethanol production rates (g/L hr) in response to increasing fractions of hydrolyzed aqueous. extract from pine pyrolysis oil ranging from 0 to 20% (v/v) and micro-aerophilic conditions. One can see that productivity gradually improves for the 2.5% and 5% vials over time. Therefore, with an extended fermentation time, vials with greater hydrolysate fractions can overcome their inhibition and reach ethanol productivities equivalent to the control. One can infer the yeast is adapting to the conditions, or the concentrations of dominant inhibitory compounds are decreasing (metabolized). The final yields and productivity of ethanol after 26 hours for the 2.5% and 5% vials were close to the concentrations achieved from the control. A summary of the data is presented in Table 5.10. The vials that are fortified with hydrolysate fractions should have greater final ethanol concentrations due to the sugars present in the hydrolysate. The colour of the medium prevented optical density biomass determination; therefore yeast growth analysis is limited during the fermentation period. Rather, yeast concentrations were determined afterwards using suspended solid analysis. This data is 89 also presented in Table 5.10 and is the average of two flasks. The initial yeast concentration was the same for each flask, approximately 0.02 g/L. Table 5.10: Maximum yeast concentrations, ethanol yields, maximum ethanol concentrations, and productivity values in response to varying hydrolysate fractions in nutrient rich media under micro-aerophilic conditions. Hydrolysate (%) Yield (%) Maximum ethanol concentration (g/L) Maximum yeast concentration (g/L) Productivity at maximum EtOH concentration (g/L hr) 0 55 10.9 2.8 0.43 2.5 54 10.8 2.3 0.42 5 56 11.4 1.7 0.45 10 6 1.3 1.2 0.06 15 5 1.1 0.1 0.05 20 1 0.3 0.1 0.01 Little to no yeast growth occurred for the vials containing fractions of hydrolysate of 15% or greater. For fractions less than 15%, a linear decrease in yeast growth as a function of hydrolysate fraction is observed; inferring the inhibition of 5". cerevisiae growth is proportional to the concentration of compounds found in the hydrolysate. 5.8.2 Fermentation using high inoculation If the compound causing the pronounced lag phase was indeed furfural, it has been reported that a large inoculum can overcome this inhibition [Helle and Duff, 2004]. An inoculum concentration 50-times higher than previous experiments was used, equating to 1 g/L in each vial. A small side investigation using alternative fermentation techniques of over-liming and additional nutrient for a 10%) hydrolysate fraction was conducted simultaneously and the results are included in the graphs below. Over-liming, by bringing the pH of the substrate to 11 before hydrolysis, is reported to remove volatile and non-volatile inhibitors such as furans and phenols [Larsson et al., 1999]. The additional nutrients will determine if there is any inhibition due to nutrient limitation. Results from this fermentation are shown in Figure 5.30. Lines connecting the data points were omitted as it made it difficult to see the individual data points. 90 16 # X • + • • • i r » -•k • o • o •k Q • • 4 o X X 10 15 Time (hours) 20 25 30 • Control • 5% A 10% o 15% x 20% • nutrients + overlime Figure 5.30: Ethanol concentrations for vials with varying fractions of hydrolyzed aqueous extract using micro-aerophilic conditions and a high yeast inoculum. By using the higher concentration of yeast to inoculate, a significant reduction in lag phase is evident. Furthermore, the combination of micro-aerophilic conditions with a high inoculum successfully resulted in the fermentation of fortifications up to 20%; a much greater improvement over previous trials. Overall, higher production rates and theoretical fermentation yields were achieved for all of the hydrolysate fractions. The over-liming and additional nutrient samples did not show an improvement; therefore no further investigation was undertaken with regards to these or other detoxification techniques. One can observe in Figure 5.30 that the initial rate of ethanol production decreases proportionately to an increase in hydrolysate fraction above 5%. Yet, at the end of the sampling period, the 15% and 20% vials surpassed all other vials; as expected once the inhibitory levels of furfural were metabolized into furfuryl alcohols. 91 The ethanol concentrations from vials with 15% and 20% hydrolysate fractions overlay one another for the last two data points. Both vials exhibit a large increase in ethanol concentration between the 11th and 12th hour. This is observed again for the last two data points (between 23.5 and 26.5 hrs). The latter of the two jumps occurs after 24 hours; potentially inferring yeast adaptation has allowed the yeast to metabolize the sugars more effectively. The maximum ethanol concentration for the control peaked at 12 hours. A l l other vials were at their highest ethanol concentration for the last sampling point with a trend inferring ethanol production would continue. For future work, it is recommended to extend the period of fermentation. A summary of the results (average of two flasks) is shown in Table 5.11. Also, as fractions up to 20% were successfully fermented, it is recommended for future work to examine higher fortifications. Table 5.11: Yeast concentrations, ethanol yields, maximum ethanol concentrations, and productivity rates in response to varying fractions of hydrolysate and nutrient rich media under micro-aerophilic conditions and high inoculum. Hydrolysate (%) Yield (%) Maximum ethanol concentration (g/L) Maximum yeast concentration (g/L) Productivity at max EtOH concentration (g/L hr) 0 42 8.4 3.4 0.7 5 54 12.7 2.8 0.5 10 46 12.2 1.8 0.5 15 46 13.6 1.1 0.5 20 44 14.6 0.9 0.6 10 (extra nutrients) 24 6.3 1.9 0.3 10 (over-liming) 26 6.8 2.3 0.2 When the productivity is calculated at each sample interval, the rate decreases over time for the vials containing 0% (control), 5%, and 10% fractions. The 15% and 20% vials held a steady rate until approximately 24 hours, at which point the productivity surpassed the other vials. Steady production rates are more beneficial for continuous fermentation design. The production values in this experiment are similar to published values. Klinke 92 and coworkers (2004) reported 0.39 (g/L)/hr for pine wood hydrolysate fermentation using a 10 g/L dry weight (d.w.) inoculum. After 24 hours, yeast concentrations in the vials containing 15% and 20% hydrolysate did not notably change from the initial values; inferring conditions were too inhibitory for biomass growth. A linear correlation between the yeast growth and hydrolysate fraction infers the inhibitory effect is proportional to the concentration of compounds in the hydrolysate. Many different compounds have been documented to inhibit yeast growth. A comprehensive summary is presented in Klinke and coworkers (2004). Both alternative fermentation techniques (over-liming and extra nutrients) showed a small improvement in cell growth compared to the other 10% flask, but as previously discussed, did not improve fermentation rates. The final yeast concentrations in Table 5.11 were applied to calculate the overall productivity for each hydrolysate fortification level, with units of grams of ethanol produced per gram of yeast per hour (g EtOH/ g yeast per hr). As shown in Figure.5.31, the 15% fraction was the most efficient. The 10% and 20% vials exhibit the trade-off involving a higher inhibitor concentration that accompanies the higher sugar concentration. 93 (A « X o LU •3 u 3 T3 O 0.450 0.400 0.350 0.300 0.250 0.200 0.150 0.100 0.050 0.000 T f||if|p flilfffe y H •H I • • j HU flllli -— B T iiiiii 111 H p ilillPlil 11 I I Illi Fraction of Hydrolysate(%) Figure 5.31: Productivity of ethanol per gram of S. cerevisiae T2 yeast in response to incremental fraction of hydrolysate in the fermentation media. 94 CHAPTER VI - CONCLUSIONS This study, examined an alternative process for the production of bio-ethanol using fermentable sugars and predecessors of fermentable sugars, mainly levoglucosan, derived from pyrolysis oil. A series of unit operations was investigated at the bench-scale level, including: the extraction of levoglucosan from pyrolysis oil, the hydrolysis of levoglucosan into glucose using sulphuric acid, and the fermentation of the hydrolysate into ethanol using an adapted strain of S. cerevisiae. The complexity of the pyrolysis oil, the lack of previous research, and the breadth of the project challenged the ability to optimize each reaction. Nonetheless, optimal parameters would be highly specific to each application (upstream and downstream processing). This study can provide valuable preliminary information to tailor and improve upon. Overall, the concept of using levoglucosan from pyrolysis oil to produce ethanol fuel proved to be technically viable using simple and inexpensive techniques. I. Extraction Maximum values of levoglucosan (7.8 wt%) from the VTT pine pyrolysis oil are extracted into the aqueous fraction when minimal values of water are used. This is limited by the boundary condition of phase separation, which occurs when the volume of total water (internal and added) equals approximately 40% of the weight of the bio-oil. The final concentration of levoglucosan in the aqueous phase is directly governed by the volume of added solvent; yet a sufficient reaction time (20 minutes), effective mixing, and slighted heated pyrolysis oil (34°C, to reduce viscosity) is recommended for optimal extraction yields. Along with levoglucosan, a significant concentration of dissolved solids (approximately 20 wt% of the initial bio-oil) are extracted into the aqueous fraction, including 2.8 wf% of acetic acid and 2.0 wt%> of formic acid. II. Hydrolysis The most significant parameter that affects the acid hydrolysis of levoglucosan to glucose is the reaction temperature. As well, the initial concentration of levoglucosan (or a 95 compound related to) in the aqueous phase confounded the hydrolysis yield. Of the ranges investigated, the optimal combinations of parameters are presented in Table 6.1. Table 6.1: Summary of optimal conditions for hydrolysis. Section Initial L G concentration (g/L) Time (minutes) Temperature (°C) Acid concentration (M) Hydroylsis yield (%) 5.4 37 60 120 0.5 170 5.5.1 31 39 120 0.5 81 5.5.2 22 30 118 1.0 85 5.5.3 9 44 125 0.5 216 A large alternative source of glucose is present in the aqueous phase as it is possible to obtain greater than theoretical hydrolysis yields. Expressed as a combined severity factor, combinations of parameters equating to a CS value of 1.5 are favourable. III. Fermentation As is, the aqueous fraction of the pyrolysis oil is too toxic for ethanol fermentation. However, when high concentrations of inoculum (1 g/L in flask) and micro-aerophilic conditions are applied, nutrient rich media with fractions as high as 20% hydrolysate can successfully produce theoretical ethanol yields (0.46 g EtOH/ g glucose) using the S. cerevisiae T2 yeast. The growth of yeast was found to be proportional to the concentration of hydrolysate; however, it does not affect the fermentation. Large lag phases (~4 hours) are apparent; likely due to the presence of furfural. After 24 hours, the highest overall ethanol production rates were achieved for the 15% hydrolysate fraction at approximately 0.4 g EtOH/ g yeast per hour. 96 C H A P T E R VII - R E C O M M E N D A T I O N S / F U T U R E W O R K I. Extraction The amount of levoglucosan found in the aqueous phase was expressed as a weight percent of the bio-oil rather than a yield as the total concentration of levoglucosan in the bio-oil was not known. It would be valuable to quantify the total (initial) concentration in the bio-oil or in the organic phase as to evaluate the actual efficiency of extracting the levoglucosan. If a large concentration of levoglucosan is still present in the organic phase, an attempt to extract the residual could be explored. Levoglucosan in the organic phase can be determined using the GC/MS procedure outlined in Sipila and coworkers (1998). Aqueous solvents other than water can be investigated to study the effect on levoglucosan extraction. By using sulphuric acid to induce phase separation, one could potentially combine phase separation and hydrolysis. It is recommended to determine the quantity of organic acids transferred into the aqueous phase for extracts prepared with water volumes less than 100 wt% of the bio-oil. This would identify the concentrations of acetic and formic acid that accompany maximum levoglucosan concentrations. Lastly, it would be worthwhile to investigate i f pH has an effect on the amount of levoglucosan transferred into the bio-oil as well as the precipitation of undesirable compounds. II. Hydrolysis The determination of rate constants for levoglucosan hydrolysis at different temperatures and acid concentrations would improve the ability to design appropriate reactions. A greater range of variables should be explored, including higher acid concentrations. 97 A more in-depth investigation regarding the effect of initial levoglucosan on hydrolysis yield would be very valuable as it was shown to have a significant effect on the yield. This may affect the desired outcome for the extraction step. In addition, future hydrolysis experiments should work with one large stock solution of aqueous extract to ensure the concentration of levoglucosan and other compounds in the aqueous phase are not affecting the experiment. III. Fermentation A yeast strain adapted to the conditions found in hydrolyzed aqueous fractions of pyrolysis oil could improve the fermentation ability of the yeast in the otherwise toxic environment. This could be accomplished initially by repeated cycling of the yeast. As well, expanding on this operation would include fermenting greater fractions of the hydrolysate and for longer time periods of time. IV. Analytical Almost all of the experiments were re-run using GC analysis once the challenge in determining the area under the levoglucosan peak when using HPLC analysis was identified. It is recommended to re-evaluate the HPLC method to see i f the peak separation can be improved as the method for GC analysis includes many procedural steps which increase the chance of error. It was observed that a greater deviation in the predicted concentrations from the actual values occurred at higher concentrations (>50 g/L) in the standard calibration curves as well as the 'predicted versus actual' figures provided by JMP IN. It would be valuable to put a greater effort in determining an effective and reliable analytical method for bio-oil analysis. 98 R E F E R E N C E S Azapagic, A. , May 3, 2006. Presentation at the official opening of the Department of Chemical and Biological Engineering. Sustainable Energy and the Environment. UBC, Vancouver. Badger, P.C., 2002. Ethanol from cellulose: A general review. In Trends in new crops and new uses. 17-21. Janick, J., Whipkey, A. (eds). Alexandria, V A : ASHS Press. Branca, C , Giudicianni, P., Di Blasi, C. 2003. GC/MS characterization of liquids generated from low-temperature pyrolysis of wood. Industrial & Engineering Chemistry Research. 42: 3190-3202. Bridgewater, A . V . , 1996. Production of high grade fuels and chemical from catalytic pyrolysis of biomass. Catalysis Today. 29: 285-295. Bridgewater, A .V . , 2003. Renewable fuels and chemicals by thermal processing of biomass. Chemical Engineering Journal. 91: 87-102. Bridgewater, A . V . , Peacocke, G.V.C., 2000. Fast pyrolysis process for biomass. Renewable and Sustainable Energy Reviews. 4: 1-73. Brown, S.W., Oliver, S.G., Harrison, D.E.F., Righelato, R.C., 1981. Ethanol inhibition of yeast growth and fermentation: Differences in the magnitude and complexity of the effect. European Journal of Applied Microbiology & Biotechnology. 11: 151-155. BTG Biomass Technology Group. Bio-oil applications, http://www.btgworld.com/ technologies/bio-oil-applications.html (last accessed August 10, 2006). Chum, H.L., Johnson, D.K., Black, S.K., and Overend, R.P., 1990. Pretreatment-catalyst effects of the combined severity parameter. Applied Biochemistry & Biotechnology. 24/25: 1-14. Classen, P.A., Sijtsma L. , Stams, A.J . , De Vries, S.S., Weusthuis R.A., 1999. Utilization of biomass for the supply of energy carriers. Applied Microbiology & Biotechnology. 52: 741-755. Czernik, S., Bridgewater, A . V . , 2004. Overview of applications of biomass fast pyrolysis oil. Energy and Fuels. 18: 590-598. 99 Di Blasi, C , Signorelli, G., Di Russo, C., Rea, G., 1999. Product distribution from pyrolysis of wood and agricultural residues. Industrial Engineering Chemistry Research. 38: 2216-2224. Diebold, J.P., 1997. Review of the Toxicity of Biomass Pyrolysis Liquids Formed at Low Temperatures. N R E L Report No. TP-430-22739. Dobele, G., Dizhbite, T., Rossinskaja, G., Telysheva. G., Meier, D., Radtke, S., Faix., O., 2003. Pre-treatment of biomass with phosphoric acid prior to fast pyrolysis: A promising method for obtaining 1,6 anhydrosaccharides in high yields. Journal of Analytical Applied Pyrolysis: 68-69, 197-211. Dynamotive. www.dynamotive.com (last accessed Aug 1, 2006). Feng, W., van der Kooi, H.J., de Swaan Arons, J., 2005. Application of the SAFT equation of state to biomass fast pyrolysis liquid. Chemical Engineering Science, 60: 617-624. Galbe, M . , Zacchi, G., 2002. A review of the production of ethanol from softwood. Applied Microbiology and Biotechnology. 59: 618-628. Goldstein, I.S., 1980. Hydrolysis of wood. Tappi. 63: 141-143. Hamelinck, C.N., van Hooijdonk, G., Faaij, A.P.C., 2005. Ethanol from lignocellulosic biomass: techno-economic performance in short-, middle- and long-term. Biomass and Bioenergy. 28: 384-410. Helle, S., Cameron, D., Lam, J., White, B.; Duff, S., 2003. Effect of inhibitory compounds found in biomass hydrolysates on growth and xylose fermentation by a genetically engineered strain of S. cerevisiae. Enzyme and Microbial Technology. 33: 786-792. Helle, S., Duff, S.J.B., 2004. Supplementing spent sulfite pulping liquor with a lignocellulosic hydrolysate to increase pentose/hexose co-fermentation efficiency and ethanol yield. Final report. NRCan. Holbein, B.E., Stephen, J.D., Layzell, D.B., 2005. Canadian Pyrolysis Initiative: New Directions in Biorefining (Working Paper). BIOCAP Canadian Foundation: Ontario. International Energy Agency (IEA), 2004. Bio-fuels for Transport. An International Perspective. OECD/IEA, Paris. 100 Kargupta, K., Datta, S., Sanyal, S.K., 1998. Analysis of the performance of a continuous membrane bioreactor with cell recycling during ethanol fermentation. Biochemical Engineering Journal. 1: 31-37. Kim, S., Dale, B.E., 2004. Global potential bio-ethanol production from wasted crops and crop residues. Biomass & Bioenergy. 26: 361-375. Kishimoto, M . , Nitta, Y. , Kamoshita, Y. , Suzuki, T., Suga, K. , 1997. Ethanol production in an immobilized cell reactor coupled with the recycling of effluent from the bottom of a distillation column. Journal of Fermentation and Bioengineering. 84: 449-454. Klinke, H.B., Thomsen, A .B . , Ahring, B.K., 2004. Inhibition of ethanol-producing yeast and bacteria by degradation products produced during pre-treatment of biomass. Applied Microbiology & Biotechnology. 66: 10-26. Larsson, S., Palmqvist, E., Hahn-Hagerdal, B., Tengborg, C., Stenberg, K., Zacchi, G., Nilvebrant, N.O., 1999. The generation of fermentation inhibitors during dilute acid hydrolysis of softwood. Enzyme and Microbial Technology. 24: 151-159. Lave, L. Maclean, H. , Hendrickson, C , Lankey, R., 2000. Life-cycle analysis of alternative automobile fuel/propulsion technologies. Environmental Science and Technology. 34: 3598-3605. Lee, Y . Y . , Iyer, P., Torget, R.W., 1999. Dilute-acid hydrolysis of lignocellulosic biomass. Advances in Biochemical Engineering/Biotechnology. 65: 94-114. L i , L. , Zhang, H. , 2004. Preparing levoglucosan derived from waste material by pyrolysis. Energy Sources. 26: 1053-1059. Lin, Y. , Tanaka, S., 2006. Ethanol fermentation from biomass resources: current state and prospects. Applied Microbiology and Biotechnology. 69: 627-642. McLaren, J.S., 2005. Crop biotechnology provides an opportunity to develop a sustainable future. Trends in Biotechnology. 23: 339-342. Mosier, N.S., Ladisch, C M . , Ladisch, M.R., 2002. Characterization of acid catalytic domains for cellulose hydrolysis and glucose degradation. Biotechnology & Bioengineering. 79: 610-618. 101 Mosier, N.S., Wyman, C , Dale, B., Elander, R., Lee, Y . Y . , Holtzapple, M . , Ladisch, M . , 2005. Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresource Technology. 96: 673-686. Najafpour, G., Younesi, H. , Ku Syahidah Ku Ismail, 2004. Ethanol fermentation in an immobilized cell reactor using Saccharomyces cerevisiae. Bioresource Technology. 92: 251-260. Oasmaa, A. , Czernik, S., 1999. Fuel oil quality of biomass pyrolysis oils - state of the art for the end users. Energy and Fuels. 13: 914-921. Oasmaa, A. , Kuoppala, E., Selin, J-F., Gust, S., Solantausta, Y. , 2004. Fast pyrolysis of forestry residue and pine. 4. Improvement of the product quality by solvent addition. Energy & Fuels. 18: 1578-1583. Olsson, L., Hahn-Hagerdal, B., 1996. Fermentation of lignocellulosic hydrolysates for ethanol production. Applied Microbiology. 18: 312-331. Oregon State University, 2004. Wood Science and Engineering. http://woodscience.oregonstate.edu/research.php (last accessed Sept 8, 2006) Ozbay, N . , Putun, A.E . , Putun, E., 2006. Bio-oil production from rapid pyrolysis of cottonseed cake: product yields and compositions. International Journal of Energy Research. 30: 501-510. Pacala, S., Socolow, R., 2004. Stabilization wedges: solving the climate problem for the next 50 years with current technology. Science. 305: 968-972. Palmqvist, E., Almeida, J.S., Hahn-Hagerdal, B., 1999. Influence of furfural on anaerobic glycolytic kinetics of S. cerevisiae in batch culture. Biotechnology & Bioengineering. 62: 447-454. Palmqvist, E., Hahn-Hagerdal, B., 2000a. Fermentation of lignocellulosic hydrolysates. I: inhibition and detoxification. Bioresource Technology. 74: 17-24. Palmqvist, E., Hahn-Hagerdal, B., 2000b. Fermentation of lignocellulosic hydrolysates. II: inhibitors and mechanisms of inhibition. Bioresource Tech. 74: 25-33. Palmqvist, E., Grage, H. , Meinander, N.Q., Hahn-Hagerdal, B., 2000. Main and interaction effects of aceitic acid, furfural and p-hydroxybenzoic acid on growth and ethanol productivity of yeasts. Biotechnology and Bioengineering. 63: 46-55. 102 Parisi, F., 1989. Advances in lignocellulosic hydrolysis and utilization of hydrolysates. ' Advances in Biochemical Engineering and Biotechnology. 38: 1-35. Putun, A.E. , Ozcan, A. , Putun E., 1999. Pyrolysis of hazelnet shells in a fixed-bed tubular reactor: yields and structural analysis of bio-oil. Journal of Analytical and Applied Pyrolysis. 52: 33-49. Radlein, D., 1996. Fast pyrolysis for the production of chemicals. In Bio-oil production & utilization, 113-123. Bridgewater, A .V. , Hogan, E.N. (eds). Newbury, UK: CPL Press. Roehr, M . , 2000. The biotechnology of ethanol: classical and future applications. Wiley-V C H : Germany. Saeman, J.F., 1945. Kinetics of wood saccharificafion. Hydrolysis of cellulose and decomposition of sugars in dilute acid at high temperatures. Industrial and Engineering Chemistry. 37: 43-52. Sampaio FC, Torre P, Passos F M , Perego P, Passos FJ, Converti A. , 2004. Xylose metabolism in Debaryomyces hansenii UFV-170. Effect of the specific oxygen uptake rate. Biotechnology Progress. 20: 1641-1650. Scholze, B., Meire, D. 2001. Characterization of the water-insoluble fraction from pyrolysis oil (pyrolytic lignin). Part I. PY - GC/MS, FTIR, and functional groups. Analytical & Applied Pyrolysis. 60: 41-54. Scott, D.S., Radlein, P.J., Majerski, P., 1995. Process for the production of anhydrosugars from lignin and cellulose containing biomass by pyrolysis. US Patent, 5,395,455. Sensoz, S., Can, M . , 2002. Pyrolysis of pine {Pinus Brutia Ten.) chips: 2. Structural analysis of bio-oil. Energy Sources. 24: 357-364. Shafizadeh, F., Furneaux, R.I I.. Cochran, T.G., Scholl, J.P., Sakai, Y . , 2004. Production of levoglucosan and glucose from pyrolysis of cellulosic materials. Applied Polymer Science. 23: 3525-3539. Sipila, K. , Kuoppala, E., Fagernas, L. , Oasmaa, A. , 1998. Characterization of biomass-based flash pyrolysis oils. Biomass and Bioenergy. 14: 103-113. 103 Smith, M.T., Cameron, D.R., Duff, S.J.B. 1997. Comparison of industrial yeast strains for fermentation of spent sulphite liquor fortified with wood hydrolysate. Industrial Micro-technology & Biotechnology. 18: 18-21. Suparo, O., Covington, A.D. , Phillips, P.S., Evans, C.S., 2005. An innovative new application for waste phenolic compounds: Use of Kraft lignin and naphthols in leather tanning. Resources, Conservation and Recycling. 45: 114-127. Sun Y. , Cheng, J., 2002. Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresource Technology. 83: 1-11. Taherzadeh, M . , Eklund, R., Gustafsson, L. , Niklasson, C , Liden, G., 1997. Characterization and fermentation of dilute-acid hydrolyzates from wood. Industrial & Engineering Chemistry Research. 36: 4659-4665. Tampier, M . , 2004. Best uses of biomass. Refocus. 5: 22-25. U.S. Department of Energy, 2006. Biomass program: dilute acid hydrolysis. http://wwwl.eere.energy.gov/biomass/dilute_acid.html (last accessed Aug 25, 2006). von Blottnitz, H. , Curran, M.A. , 2006. A review of assessments conducted on bio-ethanol as a transportation fuel from a net energy, greenhouse gas, and environmental life cycle perspective. Journal of Cleaner Production. Article in press. Available on-line at http://www.sciencedirect.com/science/journal/09596526. VTT Technical Research Centre of Finland, 2006. www.vtt.fi (last accessed December 1,2006). Wikipedia, 2006. www.wikipedia.com (last accessed Aug 15, 2006). Wyman, C.E., 1999. Biomass ethanol: technological progress, opportunities and commercial challenges. Annual Review: Energy and Environment. 24: 189-226. Xiang, Q., Kim, J.S., Lee, Y . Y . , 2003. A comprehensive kinetic model for dilute-acid hydrolysis of cellulose. Applied Biochemistry and Biotechnology. 105: 337-352. Yu, Z., Zhang, H. , 2003. Pretreatments of cellulose pyrolysate for ethanol production by Saccharomyces cerevisiae, Pichia sp. YZ-1 and Zymomonas mobilis. Biomass & Bioenergy. 24: 257-262. 104 Zaldivar, K. , Nielsen, J., Olsson, L., 2001. Fuel ethanol production from lignocellulose: a challenge for metabolic engineering and process integration. Applied Microbiology & Biotechnology. 56: 17-34. Zhuang, X . L . , Zhang, H.X. , Tang, J.J., 2000. Levoglucosan kinase involved in citric acid fermentation by Aspergillus niger CBX-209 using levoglucosan as sole carbon and energy source. Biomass and Bioenergy: 21: 53-60. 105 APPENDIX A - CHEMICAL COMPOSITION OF BIO-OIL Table A . l identifies over 66 components of bio-oil from pine using GC/MS analysis. Many functional groups are represented: hydroxyl carbonyl, carboxylic, methoxyl, and phenolic. 106 Table A . l : Chemical composition of fresh VTT bio-oil as determined using GC/MS. Tabic t Chemical Compost)ions of the fresh bitM>i'j from pine by CC/MS "No, 'Compound Rclentt wi lime Molecular formula IMofecutar structure 1 Propanoic tcid ' 2.633 cimoz 0 II H O — c— C K 2 — ca 3 2 Paraldehyde 2,689 C6H1203 3 Hc.xanr. 2.2,5-trinmhyt-.V562 05HI2© Ma j c - CK 2— C H _ ~ CHMfe 2 4 Ithanechioic acid. S-| I -(mahyithiojpropyll ester 4.125 C6H120S2 1 Acs — CH — E t S Cyofbpcrttanol, 2-m«hyi"» trans-4.163 C6HS2 0 C.i r ox "j 4(!H).pyridifia*ic 4.896 C5H5 N 6 0 ^ 7 5,9-D«d*«dkr»*2-one. 6,1 CM i methyl-, ( E ) -5 SOI C14H240 8 OxitanemeChanol, acetate 5.S43 C5 H8 03 0 CH 2 — O A c 9 3(2H)-Pyraiio«e, ciihydro-2-roelhyl-6.633 CS H 8 02 o r 0 107 10 . 2-Pentcncnitrile. 5-hydroxy-, (E)-6.772 C5 H7 N 0 NC •t'l J 2(3HWcuranooe 7.025 C4 H4.02 13 Nonanol 7.1! '6' _ _ _ _ _ Me™ (Gtt-2) 7~^m: , • ... " • • p i — : G H : ' . - ' • ' 14 '; 1 ;,2 -Cyc [operfiawed Fine C 5 H 6 0 2 o : 15 7-Piirancarbox a 1 dehydc, >methy.l-S..444 Cf> M6 02 Me 0 V ^ - CHO • 4- Meihyl i :SH -furait-2-• one • S 865 • C S H 6 G 2 ; • Met . i 17 Phenol '9.148' C6 H 6 G • IS Cyclohexanooe, 4-hydrexy-9.482 C 6 H 1 0 O 2 1 .XT" 9.596 C6 H10-O4 -• ,:2o; : IPipertdiiiev-' ..dimethyl- • 9 mi crms-u Me ^ " s U e 108 2i> •: V "22 •2-Cyc'lopeiilen-1 -one, 2--hydr6xy--3-mcthy!-: l .Mter^enedipl , 4.4 -•rfuabis-' 10.43B it) Sh> • of> m 02-1 L H N I U S 6H QM HO OH OH : .23 ^•Meiliyl-SFI-forari-a-•bhc" • 10,866 C5 116 0 2 24 ' Pentaridlc acid; ;2-' Ci«etli.o«ymefhyl:}-4-; ' Q X O -11.191 " C ? H : 1 2 0 4 . ' 25 /Phenol, 2-methyl- 1.1.362 C7 H'S'Q " \ fiutanoic acid, 2-cthyl-. 2-propenyl ester ' 11.419 C9H-i'6"02 27 cln k- 2-mc(hvi-3-ircrlivlciH'-11.488 c a n t i c 2. ,)»D_ccn»2-at JI.72G-: ;-29 2,5«Norboman«diol- IV. 860' '••C7MI2Q2' •• HO '• . ^ " ^ . . '•' .30 ' Phenol,. 4«methyi-! 11.983 C7 IHg0 xr HO '31 Phenol,.2-rtiet.lio.xy- 52,316 C 7 H 8 0 2 a OH 109 %2 C y c l o p e n ' j i i i l I ? d u i t c t h s 1-mcthvlelhenvh , ( l k . 2 R , l S ) -12.941 C J O H l i S O M e CH 2 M s ..!. • ••: : 1 . ^ K \ R s / • - M B - • • ^ao , . \ ••/. Piopanu'ic a*. i J , 2- \ \ OH 1 * C t M i 2 ( > 0 2 / O - -• : j f :" Me — {CH z) 8—O—C— Pr-1 methyl . m n \ , 1 •34 13.1 SO- • C9H20O2Si •35 '. 2- C J K I oppnten-1 -one, 3- oriiyl-2-Hydroxy-1 0 2 ? ' : C7 H10 02 O H 36 Phenol, 2,3-dimetiiy!- •14.161- CS H I 0 O OH a Mo 3 7 P h e n o l , 2- incthoxy-4- 1 5 . 0 1 2 " C S H 1 0 O 2 •• m e i h y l -• My \ ;:.:Rtien0l,:,2-metliG'xy-4-' - i i i e t h y l -. 1.5.426 : W W H > Q 2 CMS • . •>*» . S 39 ' 1 , 2 - B c n z a K : d i o l . 15 778-• C 6 W 6 0 2 •OH . . 2 - P r o p ^ n o t r a n d , 7-im -nhyk ctlivl e s t e r ; .15..9R9 ' Oft1il :0O2; r' 41 •2*Fu'rancarb'oxaldchydc, 5-(h ydrox vrnetb y] )-16,416 OS H6 0 3 OHC ° \ _ - C E 2 " - O H 110 42 2- Fu ra nca rbox iii de h yd e 5-{iiydrosymerh>'?)-1 16.60? ) C 6 M 03 OHC 0 V _ C K 2 - O H •^•'•'••M^Ws/ " 43 • 2,3- Dimethoxyioluene 16.824 C9 HI2 02 [ X X 44 i ,2-Benzined iol, 3-rncihyl- •' 17.536 C 7 H 8 0 2 OK HO Ms : 45 I'henol, 4--thyi-2-methosy-17,924 C9 1-112 02 OMe E t '46 . 1,2-Bcnxeneditil, 4-mciivyl- • IS.380 C7 H8 02 OH Me 47 2»Methoxy-4-• vinyl phenol 18.928 C 9 H.I 0 02 OMQ H 0 I CH —= CK ^ 48 Eitgcnol 20.124 C 1 0 H 12 0 2 C H 2 ~ C B = C H j | •49 Bfhanonc, l-(2-hyrffo.xy-6-me'tboxyphcnyl)-20L24R C 9 H.I 0 0 3 OH 6r OHi) 50 Phertol, 2-mcthoxy-4-jropyl-5 20.396 C10 H 14 0 2 • O M e 111 51 O.apW c acid, he pty j; csie r- ' 20; 76 2 G17 H34 02 Ka - { C H 2 ) 6 — O — C — ( C H 2 i B~«8 • 5'2 4- Ethyl catechol 20.92! C 8 H*Q G 2 H O OH 53 •J 4 H \drusy-2-IIKIIJ i » \bcnza1dchvde 21.237 I C 8 H 8 03 54 | Phenol, 2-meth6xy-•(l.-propenyl)-CIO H :!'2 0 2 Q M S CM = CH — : Me '55 Phenol, 2-met hoxy-4-(i-propenyl)-22.569 C I O H12 02 H O C H = C H — Me 56 j Phenol 5-I[|2-{4-•hifdrox-yphcii>:S)ethyJ]a mirisjmeihyi]- .^ methoxy-22.81 C 1 6 H 4 9 N 03 r I 57 I 5-Hepteiv'3-yn-2-ol, 6-merhyl-5-£1~' methylethyl}-23.44-2 e n H I S o OH CMS 2 ' .1 I! . • Me — CH'— 0 = C — C —. P r - i 58 Ethanone, l-(4-hydroxy-3-•methoxypheny!' 23.509 09111003 Ac 59 I (+ -)-2-Phenethanamiiie, met.hyl-N-vani'Hyl 24.606 C17H21N02 112 • Phenol, 4-{3-hydroxy-1 -peo'pc n y 1)- 2- meth ox y~ 25,698 CIO H l 2 0 3 ft HO ' j O M e hi i-Ethoxy-4-t n e i Ho x yb c n z a 1 tfch yd e 25. f: CI O H 12 0 3 ' O E t C H O 62 M:elhyik-'(2- hydoxy-3-e{hoxy-ben/.yl)c!hcr '2-7.469 CI OH 1403 63 P he n o 1, 4- (3 : Si yd ro s y-l -propenyl)-2-rnethoxy-' 28.0-15 CIO 1112 0 3 ^ s C H • = C H — CH 2 - OH . 0 H O ] ' ' 6 4 / J ; 2-Propcnai. 3-(4-hy*iroxy-2-mcchox-ypheriyl}-29.407 C K ) H10 0 3 XX J C H - = C H " " C H O • ;. ; B •enzeiic bid a no ic acid, 2,3-dimeilioxy-\ M , 239 C I 2 H16 0 4 CMS M«0 . X. ^ ( C H 2 ) 3 - C 0 2 H u 66 3H-lni idaxo[ i ,2-• aj»fid61d[3,2-c]qu incline, 2,9-dshydro-9«inethy!-46,278 CIS H i.SN3 C n C x It J M O * ~ * 113 APPENDIX B - EXPERIMENTAL DATA B.l Characterization of the aqueous phase in response to varying water volumes Table B . l : Mass and volumes of organic phase and aqueous phase for experiment 5.2.1. Sample Number Total water (wt% of bio-oil) Actual bio-oil mass (g) Actual water added (g) Aqueous fraction (g) Aqueous fraction (mL) Organic fraction (g) 1 500 5.16 24.73 27.47 24.5 2.43 2 359 11.80 39.84 46.41 39.5 5.21 3 266 12.15 29.77 36.34 32.0 5.59 4 185 18.22 29.79 40.10 31.5 7.88 5 131 17.99 19.86 30.17 25.0 7.66 6 104 30.02 24.78 42.40 32.0 12.37 7 62 24.39 10.04 24.69 18.0 9.64 8 41 24.54 4.80 20.07 11.0 8.98 Example calculations using sample 1 Total water (wt% of bio-oil) = (24.73 + (5.16*0.21)) / (5.16) * 100 = 500 Aqueous phase (wt% of bio-oil) = ((27.47 - 24.73) / 5.16) * 100 = 53.10 Organic phase (wt% of bio-oil) = 2.43 / 5.16 * 100 = 47.09 Density of aqueous phase (g/mL) = 27.47 / 24.5 * 100 = 1.12 B.2 Dissolved solids in aqueous phase Table B.2: Mass of crucibles before and after 105°C oven for 2 days using 3 mL sample of aqueous phase for experiment 5.2.1. Trial 1 Trial 2 Sample Crucible Crucible weight Crucible After 48 number weight (g) after 48 hrs (g) weight (g) hrs (g) 1 109.86 110 108.99 109.12 2 113.57 113.72 29.01 29.22 3 114.48 114.68 113.61 113.78 4 114.12 114.42 28.83 29.18 5 30.81 31.2 114.43 114.9 6 27.41 27.96 28.01 28.63 7 31.1 32.05 113.89 114.99 8 31.17 32.46 26.09 27.35 114 Example calculations using sample 1 - trial 1 Dissolved solids (g/L) - (110 - 109.86) / 0.003 = 46.67 Dissolved solids (wt% of bio-oil) = (46.67 * (24.5 1 1000)) / 5.16 * 100 = 21.37 B.3 Levoglucosan and glucose in aqueous phase Table B.3: Data used for GC calibration for experiment 5.2.1.1. Sample L G Ratio Glucose Ratio number (g/L) (LG/ribose) (g/L) (glu/ribose) 1 10.2 1.627 5.9 0.939 2 23.4 3.326 10.6 1.661 3 43.4 5.624 21 2.622 la 10.2 1.671 5.9 0.880 2a 23.4 3.341 10.6 1.599 3a 43.4 6.036 21 2.631 Slope for L G = 7.2738 Slope for glucose = 7.4928 Table B.4: Data for aqueous extracts 1 through 8 for L G and glucose determination in experiment 5.2.1.1. Sample Ratio (LG/ribose) Ratio (glu/ribose) 1 0.644 0.093 2 1.074 0.128 3 1.355 0.157 4 1.878 0.207 5 2.546 0.257 6 2.962 0.307 7 9.594 0.684 8 12.035 1.323 Example calculation for sample 1 Concentration of L G (g/L) = 0.644 * 7.2737 - 4.68 Amount of L G transferred in aqueous phase (wt% of bio-oil) = ((4.68 * (24.5 / 1000)) / 5.16) * 100 115 B.4 Organic acids in aqueous phase Table B.5: Titration data for organic acid calculations for experiment 5.2.1.2. Vol NaOH added (mL) p H -sample 1 mmol NaOH/mL sample (1) Vol NaOH added (mL) p H -sample 2 mmol NaOH/mL sample (2) Vol NaOH added (mL) p H -sample 3 mmol NaOH/mL sample (3) 0 2.8 0 0 2.5 0 0 2.6 0 0.6 3.24 0.06 0.5 3.15 0.05 0.6 3.55 0.155 1 3.5 0.1 1 3.6 0.1 1 4.03 0.19 1.4 3.7 0.14 1.5 4 0.15 1.4 4.5 0.223 1.6 3.8 0.16 2 4.3 0.2 1.7 4.8 0.27 1.8 3.9 0.18 3.6 5.7 0.36 2 5.3 0.317 2.3 4.1 0.23 4 7.3 0.4 2.4 6.5 0.35 2.7 4.2 0.27 4.6 9.6 0.46 2.7 9.3 0.387 3.2 4.5 0.32 5.1 9.9 0.51 3 9.7 0.424 3.8 4.6 0.38 5.5 10 0.55 3.4 10.1 0.466 4.3 4.8 0.43 3.6 10.3 0.517 4.7 5 0.47 5.3 5.2 0.53 5.7 5.5 0.57 6.2 6.1 0.62 6.9 7.7 0.69 7.2 8.3 0.72 7.8 8.9 0.78 Table B.6: Data used in model for determining organic acids for experiment 5.2.1.2. Model Acetic acid Formic acid Vol sample (before dilution) (mL) 10 10 [NaOH] (mol/L) 1 1 pKa 4.8 3.74 M W 60 46 K w 1.00E-14 1.00E-14 Sample 1 (g/L) 19 15 concentration (mol/L) 0.316667 0.326087 Sample 2 (g/L) 12 8 concentration (mol/L) 0.2 0.173913 Sample 3 (g/L) 7 5 concentration (mol/L) 0.116667 0.108696 Example calculation for sample 1 - first sample point mmol NaOH / mL sample = volume NaOH(mL)* {[NaOH](mol/L) /vol sample (mL)} 116 = 0 * (1 / 10 ) = 0 Model: enter in values for concentration (shaded cells) to fit model lines to data points Table B.7: Excerpt from table of values used for organic acid model. PH [H+] alpha (HAc) alpha (Ac) Vb mmol NaOH/mL sample 0 1 1.00E+00 1.58E-05 -5.00 -0.499967827 0.05 0.8912509 1.00E+00 1.78E-05 -4.71 -0.471211262 0.1 0.7943282 1.00E+00 2.00E-05 -4.43 -0.442643222 0.15 0.7079458 1.00E+00 2.24E-05 -4.14 -0.414448108 0.2 0.6309573 1.00E+00 2.51 E-05 -3.87 -0.386800657 0.25 0.5623413 1.00E+00 2.82E-05 -3.60 -0.35986177 0.3 0.5011872 1.00E+00 3.16E-05 -3.34 -0.333775066 0.35 0.4466836 1.00E+00 3.55E-05 -3.09 -0.308664294 0.4 0.3981072 1.00E+00 3.98E-05 -2.85 -0.284631672 0.45 0.3548134 1.00E+00 4.47E-05 -2.62 -0.261757151 0.5 0.3162278 1.00E+00 5.01 E-05 -2.40 -0.240098537 0.55 0.2818383 1.00E+00 5.62E-05 -2.20 -0.219692363 0.6 0.2511886 1.00E+00 6.31 E-05 -2.01 -0.200555374 0.65 0.2238721 1.00E+00 7.08E-05 -1.83 -0.182686456 0.7 0.1995262 1.00E+00 7.94E-05 -1.66 -0.166068862 0.75 0.1778279 1.00E+00 8.91 E-05 -1.51 -0.150672584 0.8 0.1584893 1.00E+00 1.00E-04 -1.36 -0.13645675 0.85 0.1412538 1.00E+00 1.12E-04 -1.23 -0.123371938 0.9 0.1258925 1.00E+00 1.26E-04 -1.11 -0.111362334 0.95 0.1122018 1.00E+00 1.41 E-04 -1.00 -0.100367687 1 0.1 1.00E+00 1.58E-04 -0.90 -0.090325015 Shaded areas called from values entered into Table B.6 Vb = vol sample (mL)* {(alpha(Ac-) * Ac (mol/L) + alpha(For) * For (mbl/LJ) - [H+] + [OH-]} / (NaOH (mol/L) + [H+] - [OH-]) mmol NaOH / mL sample = Vb * {[NaOH] (mol/L) / vol sample (mL)} Acetic acid concentration (mol/L) = sample (g/L) / M W (g/mol) = 19 / 60 = 0.316 117 B.5 Determining parameter effects on levoglucosan extraction Table B.8: Data for calibration curve for experiment 5.3. Sample L G (g/L) Glucose (g/L) Ratio of area under peaks (LG / ribose) Ratio of area under peaks (glu / ribose) 1 40.7 24.2 5.3537789 4.2698223 2 22.5 12 2.925695 2.1887459 3 11.6 6.5 1.6022943 1.1425768 lb 40.7 24.2 5.3250998 4.252591 2b 22.5 12 2.8159845 2.3696112 3b 11.6 6.5 1.6001802 1.1882848 Slope for L G = 7.6429 Slope for glucose = 5.5865 Table B.9: Data for samples 1 through 16 for experiment 5.3. Samples Ratio area under peaks (LG / ribose) Ratio area under peaks (glu / ribose) 1 10.199 0.761 2 1.614 0.211 3 2.763 0.331 4 1.646 0.198 5 2.493 0.305 6 7.336 0.608 7 3.271 0.307 8 10.667 0.935 9 11.417 0.911 10 1.800 0.255 11 3.509 0.390 12 2.754 0.279 13 1.675 0.198 14 1.573 0.198 15 2.521 0.275 16 1.465 0.207 Example calculations for sample 1 (weight of 10 mL bio-oil sample = 12.33 g) L G (g/L) = 10.99 * 7.6429 = 77.97 Aqueous phase (mL) = water added + % transferred (using Equation from Figure 5. = 5 + [-3.419*ln(62) + 74.422] = 11.03 L G in aqueous phase (g) = 77.97 * 11.03/1000 = 0.86017 L G in aqueous phase (wt%) = 0.86017 / 12.33 = 6.98 B.6 Hydrolysis of levoglucosan into glucose Table B.10: Data for calibration curve for experiment 5.4. Samples L G (g/L) Glucose (g/L) Ratio of area under peaks (LG / ribose) Ratio of area under peaks (glu / ribose) 1 13.5 9.8 X* X 2 43.2 41.1 0.394358 0.435158 3 74.3 87.3 0.770428 1.018872 la 13.5 9.8 0.129902 0.124704 2a 43.2 41.1 0.414974 0.46468 3a 74.3 87.3 0.943469 1.340387 * x = no data Slope for L G = 0.011 Slope for glucose = 0.013 Table B. 11: Data for samples 1 through 15 for experiment 5.4. Sample number Labels Ratio of area under (LG/ref) Ratio of area under (glu/ref) 1 E X T R A C T A 0.092885 0 2 E X T R A C T B 0.237075 0 3 E X T R A C T C 0.402599 0.028228 4 t* = 20 A 0.009572 0.046809 5 t=40 A 0 0.064655 6 t=60 A 0.065516 0.068102 7 NO ACID A 0.033788 0 8 t =20 B 0.024781 0.094371 9 t =40 B 0 0.151832 10 t=60 B 0.01132 0.195093 11 NO ACID B 0.072484 0.009757 12 t=20 C 0.025051 0.297551 13 t=40 C 0.02941 0.359946 14 t=60 C 0.022624 0.364234 15 NO ACID C 0.018875 0.02546 * t = time Example calculations for sample 1 L G concentration (g/L) = 0.0925885 / 0.011 = 8.4405 Note: dilution factor of 2 applies to samples 4 though 15 B.7 Determining the effect of time, temperature and acid concentration on levoglucosan hydrolysis into glucose B.7.1 Hydrolysis conditions A Table B.12: Data for calibration curve for experiment 5.5.1. Sample L G (g/L) Glu (g/L) Ratio of area under peaks (LG / ribose) Ratio of area under peaks (glu / ribose) 1 10.9 8.1 1.611732 1.3618231 2 20 14.2 2.7198344 2.3055726 3 43.5 22.8 5.6536928 3.2038845 la 10.9 8.1 1.6279254 1.3125651 2a 20 14.2 2.9012405 2.3732281 3a 43.5 22.8 6.34523 3.5370841 Slope for L G = 0.1389 Slope for glucose = 0.1536 Table B . l3 : Samples 1 through 16 for hydrolysis conditions A. Sample Ratio of area under number peaks (sugar / ribose) L G Glu 1 1.155886 0.240238 2 0.725457 0.638899 3 0.08274 1.723558 4 0.249952 1.520448 5 0.892432 0.42823 6 0.609093 0.589714 7 0.049951 1.885309 8 0.711559 0.633638 9 0.2736612 1.254759 10 0.06126 1.887407 11 0.087774 1.784109 12 0.184964 1.082529 13 0.748033 0.641924 14 0.128433 1.808829 15 0.859925 0.4159393 16 0.072691 1.919658 no hyd. 1.074066 0.162936 Example calculation for sample 1 L G concentration (g/L) = 1.155891 / 0.1389 * 2 = 16.64 Note: all samples subject to dilution factor of two. B.7.2 Hydrolysis conditions B Table B.14: Data for calibration curve for experiment 5.5.2. Sample number L G (g/L) Glucose (g/L) Ratio of area under peaks (LG / ribose) Ratio of area under peaks (glu / ribose) 1 10.9 8.1 1.736985 1.342025 2 20 14.2 2.960898 2.134042 3 43.5 22.8 6.619045 3.153298 la 10.9 8.1 1.716695 1.179272 2a 20 14.2 2.864144 1.885463 3a 43.5 22.8 6.385912 3.06893 Slope for L G = 0.1493 Slope for glucose = 0.1394 Table B.15: Data for samples 1 through 16 for experiment 5.5.2. Sample number Ratio of peaks (su area under gar / ribose) L G Glu 1 0.759104 0.179666 2 0.525141 0.499456 3 0.045939 1.3099 4 0.090899 1.261734 5 0.61637 0.41975 6 0.487255 0.462147 7 0.047703 1.245519 8 0.305962 0.7669 9 0.245437 0.850352 10 0.030915 1.376443 11 0.122946 1.102652 12 0.109073 1.118335 13 0.237494 0.87413 14 0.038057 1.366697 15 0.573714 0.281885 16 0.032025 1.331951 no hyd. 1.656048 0.185753 121 L G concentration (g/L) = 0.759104 / 0.1493 * 2 = 10.17 Note: all samples subject to dilution factor of two B.7.3 Hydrolysis conditions C Table B.16: Data for calibration curve for experiment 5.5.3. Sample number L G (g/L) Glucose (g/L) Ratio of area under peaks (LG / ribose) Ratio of area under peaks (glu / ribose) 1 13.5 9.8 0.521389 0.477096 2 43.2 41.1 1.904066 2.117855 3 74.3 . 87.3 3.266653 4.300824 la 13.5 9.8 0.155182 0.057761 2a 43.2 41.1 0.883383 0.931455 3a 74.3 87.3 1.508509 1.807465 L G and Glu used for samples 1 though 16 LG(a) and Glu(a) used for sample 17 Slope for L G = 0.0497 Slope for Glu = 0.0439 Slope for LG(a) = 0.0201 Slope for Glu(a) = 0.0209 Table B. l7 : Data for samples 1 though 17 for experiment 5.5.3. Sample Ratio of area under Ratio of area under number peaks (LG / ribose) peaks (Glu / ribose) 1 0.045775 0.276613 2 0.004439 0.481644 3 0 0.422875 4 0.003996 0.324923 5 0 0.453487 6 0 0.511664 7 0 0.518384 8 0 0.527271 9 0.0029 0.332151 10 0.002635 0.44438 11 0 0.421492 12 0 0.410715 13 0.003936 0.274919 14 0 0.338184 15 0 0.195736 16 0 0.26965 17 0.094813 0.02632 122 Example calculation for sample 1 L G concentration = 0.045775 / 0.0497 * 2 = 1.84 Sample 17: L G concentration = 0.02632 / 0.0193 * 2 = 9.83 Note: all sample subject to dilution factor of two B.8 Application of combined severity model Table B. l8 : Data for combined severity model. Hydrolysis conditions ID A + B A B C C Sample number Ro CS CS Ro CS 1 5.076484 -0.29444 0.006593 39.39734 1.294437 2 14.0352 0.62434 0.92537 593.2185 2.472185 3 38.80377 1.287844 1.588874 152.8765 2.059402 4 56.14078 1.2264 1.52743 458.6296 2.536523 5 20.30594 0.784744 1.085774 78.79469 1.771558 6 12.69121 0.580624 0.580624 305.7531 2.360432 7 35.08799 1.244128 1.545158 593.2185 2.773215 8 35.08799 0.545158 0.846188 305.7531 2.184341 9 20.30594 1.006593 1.307623 118.192 2.072588 10 155.2151 1.889904 2.190934 305.7531 2.485371 11 35.08799 1.02228 1.32331 118.192 1.771558 12 35.08799 1.02228 1.32331 1186.437 2.949306 13 38.80377 0.588874 0.889904 1779.655 2.949306 14 155.2151 1.190934 1.491964 39.39734 1.595467 15 5.076484 0.404533 0.705563 1779.655 3.250336 16 97.00941 1.463935 1.764965 305.7531 2.360432 Example calculation for sample 1 - Hydrolysis conditions A Ro = time * (exp(temp - 100) / 14.75) = 10 * (exp(90 - 100) / 14.75)) = 5.076484 CS =log(Ro)-pH - log(5.076484) - 1 = -0.29444 B.9 Effects of initial L G concentration and acid concentration on hydrolysis yields Table B. l9 : L G and glucose concentrations for experiment 5.6. Sample number L G before Glu before dilution dilution factor factor High 1.2325 0.0184 Middle 0.9416 0.0174 Low 0.5883 0.0055 1 0.2393 0.0889 2 0.3539 0.2212 3 0.2695 0.2228 4 0.4542 0.2321 5 0.1923 0.2363 6 0.3062 0.2248 7 0.3431 0.2305 8 0.1435 0.0969 9 0.3543 0.259 10 0.1302 0.1001 Note: "High", samples 2, 5, 6 are subject to dilution factor of 200 A l l others subject to dilution factor of 100 B.10 Fermentation of aqueous extracts B.10.1 Toxicity study using aqueous extracts ranging from 0 to 100% strength Table B.20: Data for calibration curve for experiment 5.7. Sample Ratio EtOH number (Et/But) (g/L) 1 0.011932 0.1 2 0.060018 0.5 3 0.143458 1 l a 0.014392 0.1 2a 0.057481 0.5 3a 0.132292 1 Avg slope for EtOH: 7.4563 124 Table B.21: Data for fermentation flasks for experiment 5.7 Time sample Peaks Ratio (h) E t O H But Et/But 3 1 22128 39437 0.1122 2 3389 42574 0.0159 . 3 1048 34626 0.0061 4 1536 30438 0.0101 5 5712 33107 0.0345 6 4405 31702 0.0278 7 5006 36076 0.0278 8 9632 37540 0.0513 9 10641 30319 0.0702 10 5658 28204 0.0401 11 7869 20969 0.0751 12 6213 37062 0.0335 8.5 1 92559 31562 0.5865 2 72772 28861 0.5043 3 506 25905 0.0039 4 2996 28560 0.0210 5 2235 49689 0.0090 6 1652 20062 0.0165 7 4749 49834 0.0191 8 3737 21581 0.0346 9 6961 46733 0.0298 10 4395 55213 0.0159 11 11429 58659 0.0390 12 11690 23291 0.1004 24 1 47012 40591 0.2316 2 38566 26291 0.2934 3 802 30644 0.0052 4 903 40283 0.0045 5 3646 42743 0.0171 6 3482 50003 0.0139 7 3992 33053 0.0242 8 4387 40088 0.0219 9 7132 36639 0.0389 10 6815 33441 0.0408 11 9094 51219 0.0355 12 10388 52952 0.0392 27.5 1 34591 31469 0.2198 2 52513 42578 0.2467 3 2860 67953 0.0084 4 2594 54693 0.0095 5 3595 52017 0.0138 6 4952 43613 0.0227 7 8656 49880 0.0347 8 7771 44793 0.0347 9 7684 36933 0.0416 10 7765 118689 0.0131 11 6696 41242 0.0325 12 5927 33926 0.0349 Example calculation for time = 3 hours, sample 1 Concentration (g/L) = ratio * slope * dilution factor = 0.1122 * 7.4563* 4 = 3.3486 Note: all sample subject to dilution factor of 4 B.10.2 Fermentation of aqueous extracts ranging from 0 to 20% strength for experiment 5.7 Table B.22: Data for calibration curve for flasks of 0% to 20% for experiment 5.7. Sample number Ratio EtOH (Et/But) (g/L) 1 0.031246 0.1 2 0.067538 0.5 3 0.128872 1 4 0.635524 5 la 0.031867 0.1 2a 0.068319 0.5 3a 0.129508 1 4a 0.643709 5 Avg slope for EtOH = 7.86 Table B.23: Data for fermentation flasks of 0% to 20% strength for experiment 5.7. Time sample Peaks (h) EtOH But 1 1 20307 171364 2 14476 200941 3 11657 237853 4 4572 98051 5 12817 215678 6 2840 94136 7 4312 190544 8 5166 244265 9 4515 263868 126 10 4064 200070 11 2574 183292 3 1 53939 155513 2 62305 166607 3 43030 223911 4 24363 137703 5 7883 187252 6 9319 214149 7 4539 181942 8 9769 171419 9 3135 204949 10 3640 211565 11 2745 201839 5 1 457014 X 2 423160 X 3 368570 X 4 300960 X 5 163143 X 6 11126 128566 7 216191 X 8 210962 X 9 278422 X 10 282091 X 11 243616 X 9 1 519858 120055 2 238485 117983 3 185118 121492 4 375 175 5 13475 126993 6 31492 X 7 4863 168438 8 3791 175492 9 2641 181680 10 2917 173468 11 2311 148464 21 1 88035 152477 2 240659 135130 3 165090 115433 4 70707 125310 5 17010 133032 6 9497 121466 7 6237 131968 8 8330 152217 9 2147 159937 10 11295 150023 11 1899 154295 Note: all samples subject to dilution factor of 4 B. l l Fermentation using micro-aerophilic conditions Table B.24: Data for calibration curve for experiment 5.8.1. Sample Ratio EtOH number (Et/But) (g/L) 1 0.016401 0.1 2 0.065887 0.5 3 0.127683 1 4 0.63679 5 Slope for EtOH = 7.8473 Table B.25: GC data for fermentation flasks for experiment 5.8.1. Time Sample Peaks (h) number EtOH But 1 1 325 114508 2 818 174725 3 1776 64215 4 2788 131170 5 797 157853 6 957 160653 7 854 126281 8 775 110866 9 2901 122880 10 274 146058 11 1237 102854 2.25 1 3610 81441 2 2556 88306 3 2234 83824 4 3103 131102 5 890 98192 6 2354. 109369 7 694 114472 8 1504 176235 9 3677 171378 10 7514 304086 11 1673 142156 4.25 1 13820 171607 2 24861 261388 3 8560 162642 4 10003 157089 5 2927 123878 6 3249 107554 7 1207 86866 8 1667 196845 9 2143 140147 10 4212 187057 11 5545 198025 5.75 1 12310 103983 2 • 26524 174273 3 27589 258816 4 11005 151286 5 3473 124719 6 5495 145426 7 793 129316 8 591 114592 9 2920 110020 10 350 62384 11 2698 112518 7.25 1 49433 108401 2 50033 99120 3 12527 109776 4 30364 123775 5 11422 109730 6 11367 109160 7 1172 111456 8 2067 105279 9 1742 79789 10 2420 99750 11 433 93248 8.75 1 55972 71207 2 56897 73942 3 57935 100423 4 52729 131905 5 6765 70659 6 12221 74458 7 442 22715 8 1066 61308 9 264 47832 10 125 25361 11 245 82039 11.75 1 59296 74492 2 74516 71134 3 45508 76297 4 46018 77939 5 29395 101268 6 26388 93345 7 1203 17244 8 909 87496 9 1466 82789 10 979 82956 11 765 77658 22 1 115278 78939 2 123492 79296 3 145029 89197 4 140245 92321 5 83937 75988 6 59245 43250 7 4768 19638 8 5346 32633 9 1905 9366 10 . 1794 13048 11 3424 84792 25.5 1 126598 73258 2 133655 76131 3 127479 77927 4 144232 78906 5 125894 80351 6 140687 67633 7 12371 106631 8 1286 81872 9 1182 77791 10 551 92628 11 1342 78424 Note: all sample subject to dilution factor of 4 B.12 Fermentation with micro-aerophilic and high inoculum Table B.26: Data for calibration curve for experiment 5.8.2. Sample number Peaks Ratio EtOH EtOH But (Et/But) (g/L) 1 15034 219361 0.013707 0.1 2 72310 212357 0.068102 0.5 3 158907 235478 0.134965 1 4 722540 205426 0.703455 5 la 33221 233171 0.028495 0.1 2a 68385 204984 0.066722 0.5 3a 147868 226786 0.130403 1 4a 645253 201272 0.641175 5 Ave slope from calibration curve for EtOH = 7.4512 Table B.27: Data from fermentation vials for experiment 5.8.2. Time Sample number Peaks (h) EtOH But 2 1 90721 215082 2 83787 202713 3 81735 216047 4 90894 205882 5 41000 196024 6 55693 217504 7 18817 216854 8 17325 215465 9 16877 206055 10 8374 222343 11 36812 187324 12 59344 324983 4 1 143206 213510 2 122202 199838 3 81656 156680 4 135247 212847 5 91977 208581 6 100466 221920 7 49682 211386 8 4761 294784 9 30844 197401 10 28528 219432 11 71226 203932 12 99906 216139 6 1 186826 213166 2 175248 212554 3 162724 209519 4 210044 208974 5 140423 213263 6 141231 213263 7 92481 282746 8 81341 206247 9 46690 198394 10 33441 196878 11 123175 204279 12 141775 210458 163854 8 1 173575 2 192718 217789 3 205544 206839 4 232404 203819 5 169034 198943 6 165494 214756 7 71465 204404 8 92396 205900 9 61022 218694 10 51087 223414 11 132251 204576 12 174758 215682 219991 210684 10.25 1 310695 2 270997 3 288583 217555 4 283053 219548 5 138479 141124 6 232462 212460 7 130427 214207 8 110558 223518 9 68342 200116 10 71259 223998 11 189996 208089 12 247126 221886 12 1 302581 212516 2 245761 178940 214369 3 302288 4 324430 221292 • 5 374055 227683 6 370691 239361 7 402909 258355 8 452799 283892 9 294310 230499 10 301158 222819 11 164904 272743 12 148303 235503 23.5 1 X X 2 X X 3 278165 225067 142112 4 312514 5 407503 219190 6 349550 182511 7 442739 224373 8 455874 230934 9 269559 237404 10 462413 234246 11 190736 230886 12 271157 259225 26.5 1 125870 248284 2 110363 243827 3 X X 4 481566 227332 5 471571 245077 6 474920 220631 7 547966 253842 8 541284 225952 9 539694 217242 10 542425 228755 11 243812 216094 12 341430 360375 Note: all samples subject to dilution factor of 4 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0059037/manifest

Comment

Related Items