Department of biology and biological engineering CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2018 Influence of inhibitors on the hydrolysis of spruce residues For the production of bioethanol Master’s thesis in biotechnology DAVID SUNDBERG Influence of Inhibitors on the Hydrolysis of Spruce Residues For the Production of Bioethanol DAVID SUNDBERG Department of biology and biological engineering CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2018 i Influence of Inhibitors on Hydrolysis of Spruce Residues For the Production of Bioethanol DAVID SUNDBERG © DAVID SUNDBERG, 2018 Supervisor: David Benjamin Nickel, Industrial Biotechnology Examiner: Carl Johan Franzén, Industrial Biotechnology Master’s Thesis 2018: October Department of biology and biological engineering Chalmers University of Technology SE-412 96 Göteborg Sweden Telephone + 46 (0)31-772 1000 ii iii Influence of Inhibitors on Hydrolysis of Spruce Residues For the Production of Bioethanol DAVID SUNDBERG Department of biology and biological engineering Chalmers University of Technology ABSTRACT Sustainable fuel ethanol may be produced from lignocellulosic materials such as wood and straw. The use of forestry residues may allow for bioethanol to be produced in a sustainable way. However, very little systematic work has been performed on the pretreatment, hydrolysis and fermentation of forest residues. Utilizing forest residues would increase the value of forest products, contribute to the development of a circular economy and combat climate change by diminishing the need for fossil fuels in the transportation sector. In this project, the inhibitory effect on the enzymes responsible for glucose release from Norway spruce (Picea abies) residues from branches, tips and needles during enzymatic hydrolysis have been investigated. Inhibitors native to the material and produced during the pretreatment increases the process cost and diminishes yields for bioethanol production. An anaerobic parallel shake flask system for performing the hydrolysis has been developed and assembled. The material has been both chemically and physically characterized, and method development for analysis of pretreated and hydrolyzed material using high performance liquid chromatography and ion chromatography has been done. Method development using a Kinetex F5 Core-shell LC column, Rezex™ RPM- Monosaccharide Pb+2 (8%) column, and a Rezex™ ROA-Organic Acid H+ (8%), 150 x 7.8 mm LC column has laid a foundation for future method development that might result in large time savings for the analysis of both pretreated and hydrolyzed spruce residue material. The holocellulosic fractions of the material were consistent with the literature. It was concluded that mild pretreatment conditions were likely to produce the highest amount of releasable sugars. Nevertheless, the materials were quite recalcitrant, and only about 30% of the theoretical glucose yield on cellulose was achieved. Despite some indications that increased ethanol and acetate concentrations in the hydrolysate decreases the hydrolysis rate and yield respectively it is likely that increasing inhibitor concentrations within realistic spans do not affect productivity and yield significantly. Keywords: Bioethanol, Inhibition, Hydrolysis, Picea abies iv Contents ABSTRACT ........................................................................................................................... iii List of abbreviations ................................................................................................................... 1 Introduction ................................................................................................................................ 3 Background ............................................................................................................................ 3 Aims and scope ...................................................................................................................... 6 Methods & materials .................................................................................................................. 9 Chemicals ............................................................................................................................... 9 Standards and stocks .......................................................................................................... 9 Analysis methods ................................................................................................................... 9 Water insoluble solids determination ................................................................................. 9 High-performance liquid chromatography ....................................................................... 11 Ion chromatography ......................................................................................................... 12 Determination of compositional & physical characteristics of spruce residues ................... 12 Liquid fraction density determination .............................................................................. 14 Enzymatic assay for activity determination ......................................................................... 15 Hydrolysis method development .......................................................................................... 15 Hydrolysis of spruce residues .............................................................................................. 18 Hydrolysis experiments .................................................................................................... 18 Results ...................................................................................................................................... 23 Method development of hydrolysis experiments ................................................................. 23 Method development of HPLC columns .............................................................................. 23 Kinetex F5 column ........................................................................................................... 24 Rezex Pb column .............................................................................................................. 26 Rezex ROA column ......................................................................................................... 29 Composition of pretreated spruce residue slurry .................................................................. 29 Liquid fraction .................................................................................................................. 30 Solid fraction .................................................................................................................... 32 Hydrolysis result .................................................................................................................. 33 WIS content in hydrolysate .............................................................................................. 34 Microbial growth in hydrolysate flasks ............................................................................ 34 Composition of the liquid fraction from hydrolysates ..................................................... 34 Glucose productivity rates and yields .............................................................................. 36 v Statistical analysis ............................................................................................................ 39 Enzymatic activity and enzyme assay calibration curves .................................................... 39 Discussion ................................................................................................................................ 41 The hydrolysis system .......................................................................................................... 41 Effectiveness of analysis methods and analysis method development ................................ 41 Effects of pretreatment parameters on slurry composition .................................................. 43 Effect of inhibitors on hydrolysis rates and yields ............................................................... 44 Blind spots ............................................................................................................................ 45 Conclusions and outlook .......................................................................................................... 47 Acknowledgments .................................................................................................................... 47 Sources ..................................................................................................................................... 49 Appendix .................................................................................................................................. 51 Standards and stocks Table .................................................................................................. 51 Materials Table ..................................................................................................................... 53 HPLC column method development experiments ............................................................... 53 Sugar recovery standards for total carbohydrate analysis .................................................... 57 1 List of abbreviations Table 1 – Abbreviations used in this thesis. Abbreviation Word/Term ACN Acetonitrile ARGM Arabinose, rhamnose, galactose and mannose FPU Filter Paper Units HMF 5-(hydroxymethyl)-furfural HPLC High-performance liquid chromatography IC Ion chromatography PDA Photo diode array RI Refractive index SSCF Simultaneous saccharification and co-fermentation UV/VIS Ultraviolet-Visible Spectroscopy WIS Water insoluable solids YPD Yeast peptone dextrose RISE Research Institute of Sweden NADH Nicotinamide adenine dinucleotide 2 3 Introduction In this section, the scientific background and theory will be covered. The aim and scope of the project will also be detailed. Background Anthropogenic climate change has over the last 70 years caused major changes to the planet’s climate: Increased atmospheric and ocean temperatures, melting ice caps and sea level rise which threaten many unique ecosystems, the security of coastal regions and diminishing agricultural yields for our most important crops [1, 2]. The cause for climate change is largely attributed to human emissions of greenhouse gases into the atmosphere from the burning of fossil fuels [2]. Researchers and policy-makers are working to develop sustainable alternatives to replace the use of fossil fuels [2]. Biofuels are one such group of alternatives which could be used in the transportation sector. Examples of these include bioethanol, biodiesel and biogas [3-5]. Biodiesel dominated biofuels globally in 2010, followed by bioethanol with a 17% volumetric share of global biofuel production, corresponding to 86 billion liters or 2800 ktoe [3-5]. Today, bioethanol is produced from agricultural products and residues. As shown by Wang et al. and Borrion et al., greenhouse gas emissions can be significantly reduced by using bioethanol compared to conventional gasoline [6, 7]. Since it utilizes existing refueling infrastructure and technology, bioethanol is an appealing alternative fuel. Bioethanol can be produced from different sugar sources and depending on the source the ethanol is classified as either first or second generation biofuel. First generation biofuels are produced from sources commonly rich in starch such as sugar canes, rice, corn and potatoes [8-10]. Policy makers fear however that using food crops for biofuel production might inflate food prices, although the science on the matter does not show concrete evidence for this proposition [11-14]. There is therefore a growing interest in second generation bioethanol which instead uses wood, wood residues and other agricultural residues such as wheat straw as substrates [10, 15]. Wood and wood residues are lignocellulosic materials, which dry weight consist mostly of the cell wall, which in turn consists of up to 75% polysaccharides [16]. Lignocellulosic materials thus hold a large potential for being utilized as a feedstock for fermentative bioethanol production. Cellulose is one of the most common polymers in lignocellulosic material together with the amorphous polymer hemicellulose and the polyaromatic lignin [10, 17]. Cellulose is a sturdy, crystalline glucose polymer while hemicellulose consists of a heterogeneous mixture of monomer sugars, usually with a backbone of xylan or galactoglucomannan with side chains of galactose, arabinose, rhamnose, mannose and other sugars [10, 17]. There are however important differences between lignocellulosic materials. In softwood the concentration of xylose for example is lower than in hardwood [18]. Both hardwood and softwood lignin contains guaiacyl and syringyl units, but softwood contains fewer syringyl units [18, 19]. This makes lignin in softwood more stable under acid conditions compared to hardwood. [18]. 4 Table 2 – Mass of sugars and polysaccharides in different parts of spruce wood. Values are listed in mg/g dried wood or bark. Cellulose Hemicellulose Glucose Xylose Mannose Sapwood[20] 466 195 35.0 57.3 94.2 Heartwood[20] 440 195 30.6 59.8 86.2 Inner bark[21] 225 266 75 25 13 Outer bark[21] 107 228 46 45 22 Stem wood[22] 420 273 Bark[22] 266 92 Branches[22] 290 300 Needles[22] 282 254 The polysaccharide content of different parts of the softwood Norway spruce, Picea abies, has been investigated in previous studies [20-24]. Willför et al. characterized stem wood, both heart wood and sapwood for non-cellulosic polysaccharides, cellulose, and water soluble saccharides [20]. They identified glucose, mannose, xylose, galactose, rhamnose, arabinose, glucuronic acid, galacturonic acid and 4-O-methyl-glucuronic acid [20]. Mannose was the most common monomer sugar in hemicellulose followed by xylose [20]. Krogell et al. investigated the composition of inner and outer bark in Norway spruce [21]. The polysaccharides identified were the same as in the study by Willför et. al [20, 21]. Glucose was the overwhelmingly most common monomer in inner bark while being in a slight majority in outer bark followed by xylose in both cases [21]. Cellulose made up 22.5 mass percent (m/m = m%) and 10.7m% for inner and outer bark respectively, but older sources have found values even lower [21, 23, 24]. Räisänen et al. have documented the composition of stem wood, bark, branches and needles in Norway spruce [22]. According to the data, cellulose content is highest in stem wood and lower in branches, needles, and bark respectively [22]. The hemicellulose concentration is however highest in branches and lower in stem wood, needles, and bark respectively [22]. A comparison of the data from Willför et.al, Krogel et. al and Räisänen et al. can be found in Table 2. The amount of polysaccharides differs significantly between different parts of the tree but also between studies. Räisänen et al. reported higher hemicellulose content in stem wood compared with Willför et al. and higher cellulose content but much lower hemicellulose content in bark than Krogell et al. [20-22]. Cellulose is dominating in stem wood while hemicellulose is more common in branches and needles [20, 22]. It is important to understand the composition and characteristics of the specific material to be able to develop an efficient process for the production of bioethanol. However, knowing the composition and characteristics of the untreated material is not sufficient. The specifics of the process design for bioethanol production will change the material throughout. This means that it is also of great importance to understand the effects of each unit operation on the material. The composition of a mixture of Norway spruce residues has yet to be investigated. An overview of a possible production process can be seen in Figure 1. The lignocellulosic material is first pretreated. Steam explosion with sulfuric acid or sulfur dioxide, also known as 5 acid catalyzed steam explosion, is one of the most widely used physico-chemical pretreatments for lignocellulosic materials [25]. The lignin is separated out and the remaining slurry is enzymatically hydrolyzed to produce monosaccharides. The sugars are then fermented to produce CO2 and ethanol, where the latter is distilled to be used as a fuel for transportation. Figure 1 – Overview of a hypothetical biorefinery process of lignocellulosic material for the main production of bioethanol. Secondary production streams of lignin could be used to produce natural binders and adhesives while the CO2 produced during fermentation could be used as supercritical CO2 and be used in chemical extraction plants. The pretreatment process for second generation substrates has been investigated to improve effectiveness and efficiency. Talebnia et al. has found that sugar yields of up to 99.6% and ethanol yields of up to 99.0% compared to their respective theoretical maximal value are achievable in wheat straw [26]. But it was also concluded that there is no optimal pretreatment method as of now [26]. Alvira et al. have identified low temperatures, the avoidance of milling and grinding before pretreatment, using as small pretreatment reactors as justifiable, and having low moisture content as important parameters for an energy-efficient pretreatment [25]. The pretreatment should also be designed so that lignin recovery is possible and that a minimal amount of solid bio-waste is produced [25]. Furthermore, the process should produce monosaccharides or short oligosaccharides in high concentrations, but avoid sugar or lignin degradation into inhibitors [25]. This is important to ensure both high product yields and high survivability of the fermenting yeast [25]. The pretreatment should also be tailored to generate sugar profiles that are compatible with the fermentative capabilities of the yeast, which is important with respect to pentoses such as xylose and arabinose [25]. The pentoses pose a problem as non-engineered S. cerevisiae cannot ferment them [27]. However, genetically engineered strains are able to ferment both pentoses and hexoses, called co- fermentation, using both ideal feedstocks and realistic lignocellulosic materials [11, 27, 28]. During acid catalyzed steam explosion the material is heated to high temperatures and pressurized for up to a few minutes followed by rapidly depressurized of the reactor [25]. This causes rapid auto-hydrolysis of the hemicellulose and disintegration of the material [25]. The structural removal of the hemicellulose increases cellulose accessibility for the enzymes and increases pore size [25]. Small pore sizes has been determined to be the main factor for lower hydrolysis rates because small pores trap the enzymes [25]. The steam explosion also removes 6 and redistributes some of the lignin [10]. Because the hemicellulose and lignin are decomposed into smaller pieces they can be more easily separated into a liquid fraction [10]. The crystalline cellulose however is sturdier and some of it will remain solid until subject to the enzymatic treatment where it is hydrolyzed into glucose [10]. Steam explosion has several advantages compared to other pretreatments such as using less hazardous chemicals, its feasibility to be industrially implemented and the ability to use larger wood chip sizes, thus eliminating excessive grinding and milling [25]. Steam explosion has been proven useful for poplar, olive residues, herbaceous residues and wheat straw and it has been demonstrated to be one of the most effective strategies for softwoods [25]. A drawback is the production of degradation products [25]. These consist of aromatic furan derivatives such as furfural and HMF which are formed from incomplete sugar degradation. HMF is derived from the degradation of hexoses while furfural is derived from pentoses. Furthermore, weak organic acids are produced such as acetic acid, formic acid, levulinic acid as well as lignin derivatives [10, 25]. Acetic acid is derived from the acetyl groups of the hemicellulose while formic and levulinic acid are further degradation products of HMF [25]. Several aspects of the material have been investigated for their effect on the hydrolysis and the following fermentation into bioethanol [25, 26, 29-31]. Increasing degree of cellulose crystallinity will decrease the rate of hydrolysis [29]. The lignin will provide a physical barrier as well as act as a site for enzyme to bind unspecifically [25]. The waxy outside of tree bark also hinders enzyme to reach the cellulose and must be addressed in the pretreatment [25]. The differences in structure of the cellulose and hemicellulose require complex enzymatic mixtures which makes the choice of enzymes important [26]. The presence of pretreatment and hydrolysis products as well as degradation products such as aromatics, organic acids and lignin can contribute to lowering the effective activity of any enzymes added or might work toxic against the fermenter, which is usually S. cerevisiae [26, 30, 31]. Some of the inhibitors are indigenous in the material, while some are produced during pretreatment of the material [25, 30]. Second generation bioethanol is currently not commercially viable. This is not due to a lack of substrate availability, which are cheap and readily available, but rather in part because of the technical barriers in the pretreatment and hydrolysis process described previously to lower the costs of bioethanol production to make it commercially viable several parts of the process needs to be focused on [10]. These include: Lowering the energy requirements of the pretreatment, improving the cellulose and hemicellulose conversion rate into monomeric carbohydrates, develop industrial scale processes with the ability to co-ferment both hexoses and pentoses, manage to extract and process the lignin into other useful chemical products, lowering enzyme costs, and minimizing the influence of inhibitors [10]. Aims and scope This thesis is intended to contribute to a fundamental understanding of inhibitory action during hydrolysis to enable production of bioethanol from Norway spruce residues using S. cerevisiae as the fermenter. More specifically, this project has aimed to generate knowledge 7 about the composition of tips, needles and branches from Norway spruce as well as inhibitor profiles for the same material pretreated using acid catalyzed steam explosion. A lab-scale hydrolysis process has been developed to study inhibitory action on hydrolysis. Furthermore, analysis method development for the detection of inhibitors and hydrolysis products has been investigated. The objective was to develop a faster analysis method for hydrolysate samples. Several research questions have been attempted to be answered along with the insights aspired to be reached and mentioned in the above paragraph. Are there any indications that effects of the inhibitors can be grouped for modelling purposes and if so, what indications? Do the results call for any special considerations for a future simultaneous saccharification and co- fermentation (SSCF) process, and if so, how? What are important blind spots in this project which needs to be addressed or acknowledged when moving forward? To address these questions, the glucose production rate during hydrolysis was explored under different inhibition loading scenarios. 8 9 Methods & materials This chapter contains the chemicals and materials used as well as the methods applied to perform the different experiments. All dilutions in this report are reported in the form 1:x, where x is the fractional concentration of the stock concentration listed in Table 11 in the appendix. As an example, a 1:3 solution has one third the concentration of the stock concentration. Chemicals All chemicals used are listed in Table 3. All solid chemicals used were anhydrous unless otherwise noted. The spruce residues were obtained from the Biorefinery Demo Plant in Örnsköldsvik, Sweden. Standards and stocks All standards and stocks prepared and used extensively throughout this project are listen in Table 11 the appendix. All standards and stocks were weighed into 50 mL falcon tubes and diluted to 45 mL with MilliQ water. They were stored frozen at -20 °C when not needed. Analysis methods Water insoluble solids determination Water insoluble solid (WIS) determination was performed on the pretreated material as well as on the endpoint samples of hydrolysate slurry. Samples were washed with MilliQ water and centrifuged consecutively using the Avanti J-26S XP Beckman coulter centrifuge at 8000 rpm for 20 min at 4 °C until the glucose concentration in the supernatant was below 50 mg/L. The supernatant was removed between each centrifugation. The glucose concentration was measured using MQuant glucose test strips from Merck Millipore. Weighed aluminum dishes were dried at 105 °C in a BINDER drying and heating chamber overnight. 2 g of wet solid hydrolysate from each sample was placed in the aluminum dish and dried at 105 °C overnight. The dried material was weighed and frozen to be used for two- step acid hydrolysis, as previously described, in the future. The WIS was calculated by dividing the product between the mass of the washed wet solids and the mass of the dried solids with the product between the original mass of the wet slurry and the mass of the washed wet slurry added to the aluminum dish. 10 Table 3 – List of all chemicals used during this project. Chemical name CAS-number Manufacturer Sulfuric Acid 7664-93-9 EMSURE® Merck Millipore Calcium carbonate 471-34-1 EMSURE® Merck Millipore D-Glucose 50-99-7 Formedium™ D-Xylose 58-86-6 Sigma Aldrich D-Galactose 59-23-4 Sigma Aldrich L-Arabinose 5328-37-0 Biochemica, Applichem D-Mannose 3458-28-4 Sigma Aldrich Rhamnose 3615-41-6 Sigma Aldrich Sodium acetate 127-09-3 Sigma Aldrich Trisodium citrate dihydrate 6132-04-3 Sigma Aldrich Trans-ferulic acid 537-98-4 Sigma Aldrich Ferulic acid 1135-24-6 Apin Chemicals Ltd. Levulinic acid 123-76-2 Sigma Aldrich Ethanol 64-17-5 BDH Chemicals, VWR Glycerol 56-81-5 BDH Chemicals, VWR 5-(hydroxymethyl)furfural (HMF) 67-47-0 Sigma Aldrich Furfural 98-01-1 Sigma Aldrich Vanillin 121-33-5 Sigma Aldrich Formic acid 64-18-6 EMSURE® Merck Millipore 3,5-dinitrosalicylic acid 609-99-4 Sigma Aldrich Potassium sodium tartrate tetrahydrate 304-59-6 Sigma Aldrich Phenol 108-95-2 Merck, VWR international Sodium metabisulfite 7681-57-4 Sigma Aldrich Sodium hydroxide 1310-73-2 BDH Chemicals, VWR Citric acid monohydrate 77-92-9 Sigma Aldrich Hydrochloric acid 7647-01-0 EMSURE® Merck Millipore Cellic CTec2 50-99-7 Sigma Aldrich Nitrogen gas 7727-37-9 N/A Phosphoric acid 7664-38-2 Sigma Aldrich Acetonitrile 75-05-8 Sigma Aldrich Methanol 67-56-1 BDH Chemicals, VWR Sorbic acid 110-44-1 Sigma Aldrich 11 High-performance liquid chromatography High-performance liquid chromatography (HPLC) was used to quantify acids, aromatics as well as glucose, ethanol and glycerol in pretreated material and in hydrolysate slurry. The analytes were separated and detected on a Jasco Extreme system running ChromNav CFR Ver.2.01.7 build 6, LC-4000 with a BS-400-1 Bottle stand, an AS-4150 RHPLC Autosampler, a PU-4180 RHPLC Pump, a CO-4061 Column oven, an UV-4075 UV/VIS detector, a RI- 4030 refractive index (RI) detector and a LC-NetII/ADC interface box using a Rezex™ ROA- Organic Acid H+ (8%), 150 x 7.8 mm LC column running each sample for 45 min with an injection volume of 5 µL 5 mM H2SO4 at 80 °C with a mobile phase speed of 0.8 mL min -1 . Detection was made with the UV/VIS detector at 210 nm and the RI detector. HPLC standards used with the pretreated material were glucose (25 g L -1 ), galactose (25 g L -1 ), mannose (25 g L -1 ), xylose (25 g L -1 ), ethanol (25 g L -1 ), glycerol (5 g L -1 ), 5- (hydroxymethyl)furfural (HMF) (10 g L -1 ), furfural (10 g L -1 ), vanillin (2.5 g L -1 ), levulinic acid (2.5 g L -1 ), acetate (12.5 g L -1 ), citrate (15 g L -1 ), formic acid (5 g L -1 ), ferulic acid (0.5 g L -1 ), and trans-ferulic acid (0.5 g L -1 ). The standards were run in a binary dilution series from 1:1 to 1:64 dilutions and were made with MilliQ water. HPLC standards used with the hydrolysate include the above mentioned standards with the same dilutions except vanillin, ferulic acid and trans-ferulic acid which were not used. HPLC method development was conducted to analyze inhibitor samples faster and more precise. As opposed to the characterization of the material, the method development was carried out using the above mentioned column and HPLC system, although not together. A Kinetex F5 Core-shell LC column, a Rezex™ RPM-Monosaccharide Pb+2 (8%) column and another HPLC system with the same hardware and software as mentioned in the previous paragraph but with an MD-4010 Photo diode array (PDA) detector instead of the UV/VIS detector and RI detector was also used. The two HPLC systems will here forth be referred to as the isocratic HPLC and the gradient HPLC respectively as the latter was set up to run gradients and mixed mobile phases while the former was not. The three columns used will hereby be referenced as the Rezex ROA, Kinetex F5 and Rezex Pb columns. Standards used during the HPLC method development were the same as mentioned above, with the exception of vanillin, ferulic and trans-ferulic acid which were not used arabinose (25 g L -1 ), rhamnose (25 g L -1 ) and sorbic acid (1.5 g L -1 ) which were used. Dilutions used are specified in Table 13-15 in the appendix and were made with MilliQ water. Sixteen experiments were conducted using the Kinetex F5 column on the gradient HPLC that are detailed in Table 13 in the appendix. The first experiments were based on the protocol for detecting food additives available in the technical information for the column [32]. Isocratic and gradient methods were performed using 0.1v% phosphoric acid or MilliQ water together with acetonitrile (ACN). A near optimum ratio (±10%) between the two liquid phases was determined as well as near optimum detection wavelengths (±10 nm). A hydrolysis sample was run and peak detection was applied. Calibration curves for HMF and Furfural were done. The objective was to develop a faster analysis method for hydrolysate samples. 12 Twelve experiments were conducted using the Rezex Pb column on the isocratic HPLC that are detailed in Table 14 in appendix for the detection of monosaccharides and alcohols (ethanol and glycerol). First experiments were based on the method developed by McGinley [33]. Different flow rates, temperatures and mobile phases were tested. Mobile phases tested were MilliQ water, 0.4v% MeOH diluted in MilliQ water, 4v% MeOH diluted in MilliQ water and 4v% ACN diluted in MilliQ water. Calibration series for glucose, xylose, mannose, rhamnose, arabinose, galactose, ethanol, glycerol and the new sugar mix were made and run. The objective was to find a replacement analysis method for the use of ion chromatography (IC). Five experiments were conducted using the Rezex ROA column on the gradient HPLC that are detailed in Table 15 in appendix. The method was based on the same method used on isocratic HPLC. Sugar, acid and aromatic mixes were run as well as one hydrolysate sample, the acid single standards and HMF and furfural. The objective was to develop a method on a second HPLC system to allow for parallel analysis and thereby shorten analysis duration. Ion chromatography IC was performed using a Dionex ICS-3000 Reagent-Free™ IC system running Chromeleon™ Chromatography Data System for the characterization of the pretreated slurry. IC standards used with the pretreated material were sorbitol (100 mg L -1 ), mannitol (100 mg L -1 ), arabinose (100 mg L -1 ), rhamnose (100 mg L -1 ), galactose (100 mg L -1 ), glucose (100 mg L -1 ), xylose (100 mg L -1 ), mannose (100 mg L -1 ) and fructose (9 mg L -1 ). Dilutions of IC standards were made with 10 mg L -1 fructose solution, which served as an internal standard. The liquid fraction samples were run together with a binary dilution series of the sorbitol, mannitol, arabinose, rhamnose, galactose, glucose, and xylose standards at dilutions 1:1 to 1:128 and the mannose standard at dilutions 1:1 to 1:16 (see Table 11 in the appendix). The solid fraction samples were run together with a binary dilution series of the sorbitol, mannitol, arabinose, rhamnose, galactose, glucose, and xylose standards at dilutions 1:1 to 1:32 and the mannose standard at dilutions 1:1 to 1:16 (see Table 11 in the appendix). The recovery standards used with the solid fraction were mannose (340 mg L -1 ), galactose (173 mg L -1 ), glucose (1384 mg L -1 ), xylose (208 mg L -1 ), and arabinose (92 mg L -1 ) and were run at 1:50 dilution. Determination of compositional & physical characteristics of spruce residues Prior to this project, Norway spruce residues from branches, tips and needles had been pretreated using 1 L kg -1 4m% sulfuric acid and steam explosion in accordance with a design of experiments plan. The pretreatment experiments were performed by Emma Johansson at the Research Institute of Sweden (RISE) Processum, Örnsköldsvik. This has resulted in thirteen different batches of pretreated slurry that were frozen. See Table 4 which presents an overview of these batches. The temperature and residence time in the reactor were design variables in a central composite design. 13 Table 4 – Overview of the thirteen different batches of pretreatment material. The rows ‘Temperature’ to ‘Severity factor’ describe the parameters included in the experimental design of pretreatments performed on the spruce residues. The severity factor is a quantity that relates biomatrix opening to pH, temperature and holding time of the pretreatment [34]. The WIS listed is the mass percentage of water insoluble solids in the material after pretreatment. Material ID MAT.IB.100 MAT.IB.101 MAT.IB.102 MAT.IB.103 MAT.IB.104 MAT.IB.105 Temperature (°C) 204 214 211 214 204 204 Time (min) 20 20 4.4 7 7 7 Severity factor 4.36 4.66 3.91 4.20 3.91 3.91 WIS (m%) 14.49 13.87 15.10 13.11 17.17 16.91 Material ID MAT.IB.106 MAT.IB.107 MAT.IB.108 MAT.IB.109 MAT.IB.110 MAT.IB.111 Temperature (°C) 209 209 209 209 200 206 Time (min) 13.5 13.5 13.5 13.5 13.5 13.5 Severity factor 4.34 4.34 4.34 4.34 4.07 4.25 WIS (m%) 14.18 13.77 15.73 15.89 18.68 17.55 Material ID MAT.IB.112 Temperature (°C) 213 Time (min) 22.6 Severity factor 4.68 WIS (m%) 12.51 The pretreated slurry was separated by centrifugation at 9000 rpm for 25 min using an Avanti J-26S XP Beckman coulter centrifuge into two fractions, a liquid fraction and a solid fraction. The liquid fraction was filtered from any particulates while the wet solids were washed and dried in accordance with the methodology for WIS determination. Excess dried solids were stored at -20 °C until further analysis. The measured WIS for each pretreatment batch can be seen in Table 4. In this project, triplicate sample of the liquid fraction were prepared and analyzed using IC to identify the sugars, and HPLC to identify the inhibitors. The solid fraction was hydrolyzed and the fraction of polysaccharides was calculated according to A. Sluiter, et al. by two-step sulfuric acid hydrolysis [35]. The generated sugar monomers were measured in tri- or hexaplicates using IC. For the liquid fraction analysis on the IC, each sample was diluted 1:10 into a 10 mg L -1 fructose solution. Proteins were removed by heating each sample to 100°C for 1h on a heating block to precipitate them and filtered out with a 25 mm w/ 0.2 µm PTFE membrane syringe filter prior to analysis. The samples were further diluted to 1:500 into the 10 mg L -1 fructose solution. The samples were run on the IC together with the standards according to the analysis method section. For the liquid fraction analysis on the HPLC, each sample was diluted 1:3 with MilliQ water. The samples were then filtered through 0.2 µm nylon filters. The samples were run on the HPLC together with the standards according to the analysis method section. The solids were powdered using a Qiagen TissueLyser II and weighed into samples á 60 mg and dried overnight at 105 °C. 600 µL of 12 M sulfuric acid was added to each sample. The 14 samples were put in a 30 °C water bath under stirring at 150 rpm. During incubation, every 20 minute the samples were vortexed briefly. After 60 min, 16.8 mL of MilliQ water was added. The samples were autoclaved at 121 °C for 1h together with the recovery standards, which were processed to quantify the sugar loss throughout the two-step hydrolysis. After cooling, solid CaCO3 was then added to neutralize the acid. The samples were diluted to 1:50 using 50 mg L -1 fructose and centrifuged at 5000 rpm for 5 min at 4 °C. They were then filtered through 25 mm w/ 0.2 µm PTFE membrane syringe filters. Lastly, the samples and the sugar recovery standards were heated for 10 min at 90 °C before being analyzed together with the standards according to the analysis method section. The amount of cellulose and hemicellulose present in the dried, pretreated slurry was then determined using the results from the IC analysis of the solid fraction. Below the equation used for calculating the polymeric sugars from the monomer data is presented: 𝐶𝑝𝑜𝑙𝑦𝑚𝑒𝑟,𝑗 = 1 𝑛 ∑ 𝐶𝐷𝑎𝑡𝑎,𝑖,𝑗 ∗ 𝑅𝑗 ∗ 𝐴𝐶𝑜𝑟𝑟,𝑗 ∗𝑛 𝑖=1 𝑉 𝑚𝑖,𝑗 (1) Where 𝐶𝑝𝑜𝑙𝑦𝑚𝑒𝑟,𝑗 is the concentration of polymeric sugar j expressed as a mass fraction (mass polymeric sugar per mass solids in pretreated slurry), n is the number of replicates for each pretreatment sample set, 𝐶𝐷𝑎𝑡𝑎,𝑖,𝑗 is the concentration of the monomeric sugar as determined by the IC analysis, 𝑅𝑗 is the recovery rate for sugar j, 𝐴𝐶𝑜𝑟𝑟,𝑗 is the anhydrous correction term for sugar j which is listed in Table 5, V is the total volume of the hydrolysis reaction, i.e. 17.346 mL, and 𝑚𝑖,𝑗 is the weighed mass of water insoluble solids used in the two-step hydrolysis process for each replicate i and each sugar j. Table 5 – The anhydrous correction terms for each of the monosaccharides used during characterization of the solid fraction. The term is an empirical fraction between the mass of the polymeric sugar per sugar unit and the mass of the monomer sugar. Sugar Anhydrous Correction term Arabinose 0.88 Galactose 0.9 Glucose 0.9 Xylose 0.88 Mannose 0.9 Liquid fraction density determination Pretreated slurry was thawed and adjusted to pH 5.00 using 2 M NaOH. Two 50 mL falcon tubes containing 25.96 g and 25.94 g of pH-adjusted slurry were prepared. 1122 µL of 1 M citrate buffer and 1353 µL of Cellic CTec2 (LOT # SLBS6227) diluted 1:10 in MilliQ water was added to each sample. The samples were then centrifuged at 5000 rpm for 20 min at 30 °C. The supernatant was then transferred to a new set of falcon tubes and centrifuged again under the same conditions. The volume and mass of the liquid of two replicates was then measured and the average density calculated. 15 Enzymatic assay for activity determination An enzymatic assay was performed to determine the activity of the Cellic CTec2 enzyme mixture (Lot # SLBS6227) on filter paper. The protocol used was adapted from Xiao et al. and Adney et.al [36, 37]. The assay measured glucose by detecting the reducing sugars through a color reaction caused by di-nitrosalicylic acid. Samples were added to a Microtest Plate 96 Well,F plate and the absorption at 540 nm was measured using both a Spectro star nano BMG Labtech plate reader and a BMG FLUOstar OMEGA Microplate Reader after the color reaction. The activity was then calculated in terms of Filter Paper Units (FPU) per mL enzyme solution in accordance with the aforementioned protocols. Hydrolysis method development A method for conducting parallelized hydrolysis experiments under anaerobic conditions was developed. The system consisted of ten conical shake flasks containing the hydrolysate mixture inside a shake incubator with nitrogen gas bubbled through the hydrolysate. Tubing for gas was taped and zip-tied to the lab walls and ceiling. A needle valve and a manometer were added to the gas supply tubing to regulate and measure the gas flow. An Acro 50 w/ 0.2 nm PTFE membrane air filter ensured sterile nitrogen supply. Before reaching the shake flasks, the sterile nitrogen was bubbled through a Schott bottle filled with MilliQ water to moisturize the nitrogen and prevent excessive evaporation of the hydrolysate liquid during hydrolysis. The gas was spread to the shake flasks using a 1-to-10 branching tube made using Y-connectors. To ensure similar flows in all tubes, water was first pumped through 1-to-5 branching tubes with the help of pressurized air. The relative volumes of water from each opening over a fixed amount of time were observed. Four manometers were then connected to separate ends of the branching tube with the fifth opening clamped. This was done because only four extra manometers were available. Using the Schott bottle and pressurized air, humidified air was passed through the system and the relative flow was noted. Measuring the humidified air flow rate was repeated until every opening on the branching tube had been clamped once. The four manometers were then removed. This allowed estimations to be made about the difference in flow rate between the different tubes. A three-way-valve was added before the Schott bottle and the air filter to avoid nitrogen- enrichment in the shaker or in the room. Tubing was connected to the third opening leading off to a ventilation arm. Thereby, the nitrogen flowed directly to ventilation during sampling. Furthermore, the exhaust gas from the shake flasks was combined into one tube leading to the ventilation as well. Three holes were drilled through each of ten rubber stoppers that had been temporarily solidified by liquid nitrogen: A larger center one and two smaller ones on either side of the larger hole. A 21cm long metal tube with an outer diameter of 6.05 mm and an inner diameter of 3.73 mm was tightly fitted through the center hole of each rubber stopper while two 10 cm 16 long metal tubes with an outer diameter of 3.3 mm and an inner diameter of 1.75 mm were tightly fitted through the smaller holes of each rubber stopper. The system was tested to ensure that gas could pass through the slurry under realistic conditions. A 500 mL conical shake flask with 125g prepared slurry was made. The prepared slurry was made by adding 90.67 g of pretreated slurry from MAT.lB.108 so that the final WIS concentration would reach 10m%, 31.1 mL of 0.48 M NaOH to adjust the pH to 5.00 and 417 µL Cellic CTec 2 solution (VCSI0003) with a concentration of 150 FPU/mL to reach 5 FPU/gWIS to the shake flask. MilliQ water was added to reach the final mass of 125 g. A rubber stopper was added and then connected to the system. 6.67 vvm of nitrogen gas was supplied to the shake flask and the shake incubator was set to 30 °C and 180 rpm for 3 hours. As sampling from the central metal tube was impossible under realistic conditions due to the high heterogeneity of the slurry, the system was changed to serve gas only. Sampling was performed manually from the shake flask by lifting the stopper. The shaking speed was set to 120 rpm. The method was refined during the hydrolysis experiments. The gas flow was lowered from 6.67 vvm to 3.33 vvm during the second set of hydrolysis experiments. The combinatory setup between the branching tube, each shake flask and the debranching tube went through a couple of revisions. Tubes and branches were labelled from the second experiment on and arranged as shown in Table 6 and Figure 2. A schematic overview of the system can be seen in Figure 3. Table 6 – Arrangement description of the connections for the branching tubes (a-j) with the shake flasks (1-10) and the debranching tubes (A-J) for the second to fourth set of hydrolysis experiments. Set 2 Set 3 Set 4 a1A a3A a5A b2B b5B b6B c3C c6C c4C d4D d8D d7D e5E e1E e3E f6F f4F f8F g7G g7G g1G h8H h10H h2H i9I i2I i9I j10J j9J J10J 17 Figure 2 – Connections of the branching tubes (a-j) with the shake flasks (1-10) and the debranching tubes (A-J) for the third and fourth set of hydrolysis experiments seen from above. The green tree diagram depicts the branching tube while the red tree diagram depicts the debranching tube. “In” denotes where the gas inlet to the incubator and “Out” denotes the outlet of the gas flows from the incubator. Circles symbolize the shake flasks, squares the stoppers, gray squares in the background the four sticky pads in the incubator and the remaining lines tubing. The numbers represent the flask ID and sampling order. Figure 3 – Schematic overview of the hydrolysis experimental setup. Nitrogen gas travelled from the wall through the needle valve, manometer and air filter into the water-filled Schott bottle where it got moisturized. The gas then travelled either into the incubator and shake flasks following the green path or directly to the ventilation arm following the red path, depending on the setting of the 3-way valve. The ventilation arm also vented the gas from the shake flasks. 18 Hydrolysis of spruce residues Four sets of hydrolysis experiments were carried out in shake flasks at 30 °C under sparging with nitrogen gas through the hydrolysate and a mixing speed of 120 rpm for 72 h. Sampling was conducted at 0, 3, 6, 9, 24, 25, 26, 33, 48 57, and 72 h after hydrolysis start. The hydrolysis experiments were carried out using the pretreated slurry of material MAT.lB.107 as it was one of the center points in the experimental design of the pretreatment. After the hydrolysis the hydrolysate was tested for microbial contamination by plating 100 µL of hydrolysate diluted 1:10 in sterile MilliQ water on yeast peptone dextrose (YPD) agar plates. The plates were incubated for 48 hours at 35 °C in a shaking incubator and colony forming units were counted. 10 grams of hydrolysate from each control sample after 72 hours of hydrolysis was used for WIS determination. After washing the hydrolysate, the collected supernatant with trace amounts of WIS was then consecutively centrifuged and the supernatant was discarded to estimate the error of the WIS measurement. Hydrolysis experiments The total weight of each experiment was 125 g. The experiments contained 10m% of WIS, they had an enzymatic activity of 8.5 FPU/gWIS and a citrate buffer concentration of 43.16 mmol/g at an initial pH of 5.0. The remainder was MilliQ water sterilized by autoclavation and inhibitor. Six control experiments were conducted without the addition of any inhibitor. The influence of ten potential inhibitors were tested in duplicates: Glycerol (10 g/L), formic acid (1 g/L), HMF (2 g/L), furfural (0.83 g/L), levulinic acid (2 g/L), ethanol (70 g/L), acetate (1.6 g/L), glucose (20 g/L), xylose (5 g/L) and a mixture of arabinose, rhamnose, galactose and mannose (ARGM) (in total 12.5 g/L). The ARGM solution was split 10.28m%, 6.86m%, 24.49m% and 58.37m% between arabinose, rhamnose, galactose and mannose respectively. These percentages were calculated from the characterization of the material to be near identical to the slurry. The inhibitor concentrations listed above indicate the target initial concentration in the liquid fraction of the hydrolysate. For HMF, furfural, levulinic acid, acetate and formic acid; the above mentioned concentrations were derived by taking the difference between the concentrations of inhibitors in MAT.lB.107 in the liquid fraction and the concentrations of inhibitors in the batch of pretreated slurry with the highest concentration of the specific inhibitor in the liquid fraction. These values were then increased by 10-15% depending on the compound, to account for uncertainties in the composition measurements. For the sugars, the concentration increase was derived by calculating the theoretical maximum amount of releasable sugar monomers and the concentration was then increased by 5%. The concentration increase of ethanol was set due to the project requirement that the finished process, including both hydrolysis and fermentation must work under ethanol concentrations of 70 g/L. Sampling was conducted by removing 3-4 mL of slurry using 5 mL serological pipettes with cut tips. The sample was weighed in 15 mL falcon tubes and centrifuged at 5000 rpm at 4 °C for 12 min. The liquid fraction was filtered through 25 mm w/ 0.2 µm PTFE membrane syringe filters. The solid and liquid fractions were then frozen at -20 °C until analysis. The 19 liquid fraction was analyzed using the HPLC and samples were diluted 1:3. Calibration curves derived during characterization was used to integrate the peaks and obtain the concentration of each hydrolysis sample. Remainders from each hydrolysis set were collected and frozen at -20 °C. From the first and second set, 10 g from each sample was collected and stored, and from the remaining sets, all remaining slurry was collected and stored. For deriving the glucose produced and the productivity the three assumptions below were made. The third assumption was made because time restrictions prohibited the measurement of the WIS for every sample and without that parameter the glucose production and productivity for the hydrolysis could not be calculated (See Table 7). 1. Each sample taken from the shake flask is representative of the entire flask. 2. The density of the liquid fraction does not change during the hydrolysis. 3. The WIS remains at 10m% throughout the hydrolysis. To calculate the amount of glucose produced during each of the sampling times nine quantities were used which are listed in Table 7. 20 Table 7 – List of quantities used to calculate the mass of glucose produced during hydrolysis. WIS denotes the percentage of water insoluble solids and is assumed to be equal to 0.1. ρliq is the density of the liquid fraction and is assumed to be constant. i denotes the i'th sample in chronological order of any one specific hydrolysis experiment. Quantity Initial value (i = 0) Sequential value (1 ≤ i ≤ 10) Description Cglc According to data According to data Concentration of glucose in the liquid fraction at each sampling. (g L-1) mtot 125g 𝑚𝑡𝑜𝑡,𝑖−1 − 𝑚𝑟𝑒𝑚,𝑖−1 Mass of slurry in the shake flask prior to sampling. (g) mrem According to data According to data Mass of slurry removed during each sampling. (g) mg Lc 𝐶𝑔𝑙𝑐,0 ∗ 𝑚𝑡𝑜𝑡,0 ∗ (1 − 𝑊𝐼𝑆) 𝜌𝑙𝑖𝑞 𝐶𝑔𝑙𝑐,𝑖 ∗ 𝑚𝑡𝑜𝑡,𝑖 ∗ (1 − 𝑊𝐼𝑆) 𝜌𝑙𝑖𝑞 Mass of glucose in the shake flask prior to sampling. (g) mg Lc,rem 𝑚𝑟𝑒𝑚,0 𝑚𝑡𝑜𝑡,0 ⁄ ∗ 𝑚𝑔𝑙𝑐,0 𝑚𝑟𝑒𝑚,𝑖 𝑚𝑡𝑜𝑡,𝑖 ⁄ ∗ 𝑚𝑔𝑙𝑐,𝑖 Mass of glucose removed during each sampling. (g) mg Lc,prod 0g 𝑚𝑔𝑙𝑐,𝑖 − 𝑚𝑔𝑙𝑐,0 + ∑ 𝑚𝑔𝑙𝑐,𝑟𝑒𝑚,𝑛 𝑖−1 𝑛=1 Mass of glucose produced after each sampling. (g) Productivity expressed as mass glucose produced per hour and mass WIS was calculated using equation 2. 𝑃𝑡𝑗,𝑡𝑖,𝑡𝑘 = 1 𝑊𝐼𝑆 (( 𝑡𝑗−𝑡𝑖 𝑡𝑘−𝑡𝑖 ) 𝑚𝑔𝑙𝑐,𝑝𝑟𝑜𝑑,𝑘−𝑚𝑔𝑙𝑐,𝑝𝑟𝑜𝑑,𝑗 𝑡𝑘−𝑡𝑗 + 𝑚𝑔𝑙𝑐,𝑝𝑟𝑜𝑑,𝑗−𝑚𝑔𝑙𝑐,𝑝𝑟𝑜𝑑,𝑖 𝑡𝑗−𝑡𝑖 ( 𝑡𝑘−𝑡𝑗 𝑡𝑘−𝑡𝑖 )) (2) 𝑃𝑡𝑗,𝑡𝑖,𝑡𝑘 is the estimated productivity in the j th time point using the i th and k th time point where i