NO20170031A1 - Electron beam and steam explosion pretreatments of biomass for production in a low energy biorefinery - Google Patents

Abstract

Various combinations of pretreatment techniques and reaction conditions were evaluated to produce high enzyme reactivity biomass. Pretreatment methods such as electron beam irradiation (EB), steam explosion (SE) were used alone or by combination with sequential steps of electron beam irradiation and steam explosion. Acid catalyzed steam explosion (S02SE) was tested to improve treatment efficiency. In addition, an organic solvent extraction known as 'clean fractionation (CF)' was used to enhance the effectiveness of the pretreatment steps, and to recover extractives that could potentially provide value. The combination of techniques demonstrated to have synergistic effects that could improve fermentable sugar yield, and reduce formation of byproducts typically caused by the use of severe treatment conditions. The combination of steam explosion and low dose radiation make these findings an invention.

Description

Electron beam and steam explosion pretreatments of biomass for production in a low energy biorefinery

Inventors: Carlos Aizpurua, North Carolina State University (NCSU), USA

Walter Chappas, Chappas Inc., USA

Timo Leskinen, North Carolina State University (NCSU), USA

Stephen Kelley, North Carolina State University (NCSU), USA

Svein Jarle Horn, Norwegian University of Life Science (NMBU), Norway Martin Haudstvedt, Abadjom Consulting AS, Oslo, Norway

Assignee: Abadjom Consulting AS, Oslo, Norway

ABSTRACT

Various combinations of pretreatment techniques and reaction conditions were evaluated to produce biomass with high enzyme reactivity. Pretreatment methods such as electron beam irradiation (EB), steam explosion (SE) were used alone or by combination with sequential steps of electron beam irradiation and steam explosion. Acid catalyzed steam explosion (SO2SE) was tested to improve treatment efficiency. In addition, an organic solvent extraction known as ‘clean fractionation (CF)’ was used to enhance the effectiveness of the pretreatment steps, and to recover extractives that could potentially provide value. The combination of techniques demonstrated to have synergistic effects that could improve fermentable sugar yield, and reduce formation of byproducts that are typically caused by use of severe treatment conditions.

The combination of steam explosion and low dose radiation, make these findings an invention.

ABBREVIATIONS AND TERMINOLOGY

Carbohydrate = Components composed of hexose and pentose type sugars, from monomeric to polymeric molecular weights

DCM = dichloromethane

DM = Dry matter

EB = Ebeam = Electron beam

Extractives = Common term used for hydrophobic components in native wood, generally referred as resin in the case of softwoods.

Extractables = In context of this report, the term is used for the materials that were dissolved from wood during solvent extraction. Include part of the carbohydrate and lignin components in addition to the native extractives.

Glucan = Glucose type carbohydrates

HMF = 5-hydroxymethyl furfural

EtOAc = Ethyl acetate

EtOH = Ethanol

MeOH = Methanol

MIBK = Methyl isobutyl ketone

Monosaccharide = Single hexose or pentose sugar

Polysaccharide = Polymer composed of hexose or pentose type sugars

Pulp = pretreated wood material, with part of the components removed by dissolution

Saccharide = Sugar

SE = Steam explosion

SO2= Sulfur dioxide

SO2SE = Sulfur dioxide catalyzed steam explosion

Substrate = pretreated wood material subjected to hydrolysis or other reactions

BACKGROUND

Biorefinery processes have become a topic of interest in the field of energy, chemical and material production. Aim of these processes is the effective utilization of renewable biomass in a manner that provides acceptable financial returns, low energy demand and overall environmental benefits. Creating an economically viable Biorefinery requires minimizing the high capital costs, the ability to use low cost variable biomass feedstocks, and the complete conversion of the biomass into products or process energy. For example, the conversion of wood feedstock into carbohydrate-based ethanol and polymeric lignin precursor in a single process has been widely investigated. (Kamm, Gruber & Kamm 2000, Zhang 2008) A number of alternatives for ethanol production from wood have been studied, and in all cases a series of processing steps must be optimized in order to reach a commercially viable ethanol yield with low energy consumption. One of the most important unit operations in terms of capital costs and energy demand, is the pretreatment step, where the wood is structurally and compositionally altered to increase its reactivity to enzymatic depolymerization in the downstream hydrolysis and fermentation steps. In addition, the formation of bi-products that commercially acceptable may be critical to the over all profitability.

Wood is composed of long fiber like cells that are responsible for its physical strength. Cell wall of the fibers is a composite of three types of polymeric components that are cellulose, lignin and hemicelluloses. Cellulose is a linear polymer consisting of glucose units joined together by (1→4)-glycosidic linkages. Lignin is a highly branched heterogeneous polymer that is built of phenylpropanoid units, linked with various types of ether and carbon-carbon bonds. Hemicelluloses are branched heteropolysaccharides, consisting mainly of six-carbon sugars glucose and mannose in softwoods, and five-carbon sugars xylose in hardwoods. Total mass balance of these components inside wood is dependent on species but can be roughly estimated to be 42 to 45 wt% cellulose, 20 to 29 wt% lignin, and 25 to 30 wt% of hemicelluloses. Wood also contains 2 to 5 wt% complex, extractive compounds made up of e.g. fatty acids, resin acids, and sterols, depending on the species.

The main polymer components are not deposited evenly inside cell wall, as the wall can be divided to distinct layers according to changes in composition. This complex structure has naturally developed to resist mechanical stress from forces of nature or degradation by natural microorganisms like fungi. The effective hydrolysis of the highly orientated, crystalline cellulose domains are particularly challenging for the cost effective production of fermentable sugars.

Enzyme catalyzed depolymerization (hydrolysis) has been found to be environmentally friendly and potentially cost effective method for converting wood polysaccharides into monomeric sugars, which can then be fermented to ethanol or other products. Typically, the enzymes that are used in such process are isolated from fungi and cultivated for industrial uses. These cellulolytic enzymes are typically composed of three types based on their role in hydrolysis of carbohydrate chains: Endocellulase cleaves long chains to produce smaller polymers with increasing amount of free chain ends, exocellulase attacks to free chain ends cleaving off disaccharides like cellobiose, and finally cellobiase produces monomeric glucose from disaccharides. The plant cell wall is naturally very recalcitrant towards activity of these cellulolytic enzymes and for this reason pretreatment steps are needed to disrupt the native carbohydrate ultrastructure.

Steam explosion has been widely recognized as potential pretreatment method for increasing enzyme activity on wood (Hamelinck, Hooijdonk & Faaij 2005, Zhu & Pan 2010). Steam explosion is also at the heart of two recent biorefinery commercialization efforts (Chemtex (http://www.chemtex.com/), DONG (http://www.dongenergy.com/en/innovation/utilising/pages/biorefining.aspx). Steam explosion acts to depolymerize the hemicellulose and amorphous cellulose, and disrupt the crystalline cellulose, while increasing surface area.

Depolymerization of polysaccharides, and decrystallization of the cellulose, are both required for effective enzymatic processing, but there are a number of technologies that can initiate these two processes. Thus, radiation induced degradation has also been studied as a wood pretreatment technology (Ardica, Calderaro & Cappadona 1985, Khan, Labrie & McKeown 1986, Karthika, Arun & Rekha 2012). Irradiation alone has shown to be marginally efficient at increasing enzyme activity. However, the combination of EB irradiation and SE pretreatments could be effective for decrystallization and depolymerization of polysaccharides, and since these two processes operate on very different mechanisms. Thus, a combination of EB and SE offers a unique opportunity to ‘tune’ each of these two different processes to increase their overall effectiveness and lower their costs.

In order to determine if and when a proposed Biorefinery process scheme is economically feasible it is possible to use engineering process modeling tools to track the mass and energy flows, and overall costs of a proposed process. The engineering process modeling tools can be used to design and optimize a complex integrated Biorefinery, and to then study the effects of operating scale and estimate economic impacts of different process options.

SUMMARY OF THE INVENTION

High severity SE can cause significant compositional changes in treated wood that make the wood more amenable to enzymatic hydrolysis, but simultaneously leads to high sugar losses and a poor economic return. Thus, moderate steam explosion severity was found to the best suitable option for high enzyme reactivity and yield sugar yield. Long retention times at low temperatures were somewhat effective, but could be enhanced by the preliminary EB treatment. EB alone is not efficient enough as a pretreatment, but will cause synergistic effects when combined with SE. The figure show where the two pretreatment activities fit into the total process.

Figure 1 The production process from wood to ethanol

Pretreatment tests with Birch wood were especially successful as enzymatic hydrolysis of above 90 % of all remaining polysaccharide components in pretreated wood was achieved. Pine wood showed lower response to the EB and SE pretreatments with lower hemicellulose dissolution, and lower cellulose sugar release during enzymatic hydrolysis.

Pretreatments are most dramatically affecting to the molecular weights of hemicellulose component in wood cell wall. Hemicelluloses are converted into highly hydrolysable water soluble form, and removal of hemicelluloses into the water phase is likely to increase accessibility of enzymes to the remaining cellulose crystalline fibrils. The lignin moieties in wood were found to go through changes during the pretreatments, with very different resulting lignin distribution in SE treated Birch and Pine. The Pine wood showed significant recalcitrance towards enzymatic hydrolysis, regardless of successful removal of hemicelluloses, and this phenomenon may be related to the lignin character and crosslinked structure. Yield of enzymatically hydrolyzed sugars remained in many cases at lower level than the amount of theoretically highly accessible water soluble hemicelluloses. This indicates to recalcitrance mechanisms that are caused by non-specific adsorption or inhibiting compounds rather than low accessibility into the cell wall.

SO2impregnation prior SE was observed to increase enzymatic activity on Pine significantly, but still not to the level that was achieved with birch substrates using non-catalyzed steam explosion. Major drawback of the SO2treatments was excessive loss of material during the pretreatment which lead to an overall lower ethanol yield. Positive effects of SO2could be observed as greatly increased amount of lignin that could be isolated in the solvent extraction process. In summary, the SO2impregnation was the only effective treatment to achieve even modest enzyme activity with Pine, and did have reasonable yields of soluble lignin as side product. With optimization, this method could be a solution for treatment of recalcitrant Pine wood.

THE BENEFITS OF THE INVENTION

Enzymatic hydrolysis was found to be efficient for the pretreated Birch substrates. With optimization of processing EB and SE treatments, up to 98 % of theoretical sugar in the solid substrate were released. Enzyme loadings of 5 and 10 FPU/g were tested, and results suggest less than 10 FPU/g loadings are more effective for the hydrolysis. Pine wood showed significant recalcitrance towards enzymatic hydrolysis. The combination of EB and SE reached only around 20 % of the potential sugar yield, although significantly better sugar yields were obtained after EB and SO2SE, where close to 90 % of the polysaccharides could be hydrolyzed.

Engineering process simulation was used to determine for different scenarios if the process and current technologies under analysis are suitable for an economically feasible process. Combination of two pretreatments techniques (EB and SE), production of a side product (clean fractionation and equipment needed for recovery and upgrading the side product), and different ethanol dehydration steps (common distillation or membranes) where evaluated with an intensive and meticulous economic analysis. As a basis for the evaluation, the 2011 NREL report in combination with laboratory data were used in order to collect all the information needed to perform the process simulation. Aspen Plus and WinGems were chosen as the simulation tools.

Although many of the combinations could be of interest, the focus was put on Birch and combination of low dose EB and SE.

The figure shows how the analysis were done.

Figure 2 Generic overview of the work flow from the pretreatments to the economic analysis.

A total of four cases for two different pathways of ethanol production are evaluated. Note that due to the differences in the base cases two different feedstockS were used (birch and corn stover), two pretreatment techniques, EB/SE vs dilute acid, and two different methods for ethanol dehydration (selective membranes and typical distillation/molecular sieve) are analyzed for the same plant size.

Case I II III IV

p p

Raw Material Corn Stover Corn Stover Hardwood Birch Hardwood Birch

Table 1 Basis for evaluation

The most critical component is the price for raw material. A large amount of corn stover in US South East could be available to $ 59 per BDMT. Birch price was set to $ 71 per BDMT. The results were:

Figure 3 shows the cost breakdown for the four different cases.

The cash cost gives the price that is needed to give investors 10 % IRR.

For Norway, the costs might be significantly changed, but the overall picture is clear. The invention can be compared to the best available solutions per the framework from NREL.

Detailed description of invention

In the experiments, various pretreatment strategies were used to increase enzyme activity on wood, and consequently optimize the yield of monomeric, fermentable sugars for subsequent production of bioethanol. The pretreatment work was conducted by a partnership including Chappas, Inc., the Norwegian University of Life Sciences (NMBU), North Carolina State University (NCSU), and Bio Oil AS. In combination, these partners are skilled in the individual pretreatment steps, including EB (Chappas), SE (NMBU), producing homogenous particles, and woody biomass extractions and enzyme hydrolysis, and engineering process modeling (NCSU). With these combined skills, the team had the ability to conduct experimental process development work, and the detailed chemical and enzyme analysis needed to evaluate the different process options. The experimental work also provides data for the engineering process modeling work that can analyze the potential a large scale commercial process.

To obtain the basic input data for the engineering process models, the team conducted a series of analytical measurements, including compositional and structural changes of the solid residue following pretreatment, measuring the enzyme activity on these solid residues, analysis of by-product formation, and characterization of potential lignin side product from solvent extraction.

Note that fermentation of monosaccharides into ethanol was not part of this experimental program since this is a relatively well-developed technology.

Table 2 Description of the project stages and work done to complete the experiments

After pretreatments with EB and SE, the analytical work was focused on four specific areas:

1. Composition of the solid substrate was determined in order to estimate yield losses during pretreatment and the amount of available polysaccharide components remaining in the solid substrate.

2. Enzymatic activity on the pretreated solid substrates was tested to evaluate effectiveness of pretreatments for creating a reactive wood substrates for the enzymatic hydrolysis process.

3. Quantity of soluble components (extractives) was measured and their composition was analyzed in order to track formation of low molecular weight by-products, and to estimate losses of fermentable sugars during the solvent extraction.

4. Evaluation of the physical changes in wood, especially in terms of degree of polymerization and crystallinity to understand the synergistic effects of the pretreatments, and how to optimize the reaction conditions.

Enzymatic hydrolysis was found to be efficient for the pretreated Birch substrates. With optimization of processing EB and SE treatments, up to 98 % of theoretical sugar in the solid substrate were released. Enzyme loadings of 5 and 10 FPU/g were tested, and results suggest less than 10 FPU/g loadings are more effective for the hydrolysis. Pine wood showed significant recalcitrance towards enzymatic hydrolysis. The combination of EB and SE reached only around 20 % of the potential sugar yield, although significantly better sugar yields were obtained after EB and SO2SE, where close to 90 % of the polysaccharides could be hydrolyzed. However, the initial solids yield for the EB and SO2SE pretreatment was low.

A second goal of the analytical work was to understand the effects of the soluble materials generated by the EB and SE processes. These soluble materials could be grouped into three categories, 1) the original wood extractives, 2) lignin fragments and 3) sugar fragments and dehydration products. As expected the mass yield of the soluble products increased as the severity of the EB and SE treatments increased. The relatively low molecular weight lignin oligomers could be isolated with moderate yield and purity, depending on the pretreatment. Chemical and physical properties of lignin side product need to be characterized further to determine best way for utilization in valuable manner. Overall, the production of side-product lignin was found higher in severe pretreatments where ethanol yield is reduced due excessive loss of polysaccharide components. The water-soluble sugar fragments and dehydration products have potential to be used for production of co-products (e.g. furfurals), as varying portions of hemicelluloses can be isolated from water streams involved to organic solvent extraction, or by direct water extractions of the pretreated substrate. The chemical composition of the watersoluble fractionation was also screened. The value of these fractions are heavily dependent on a detailed analysis of the isolation and recovery processes, and their commercial value.

Experimental design and analytical methods

Various combinations of EB and SE pretreatment techniques and conditions were evaluated. Pretreatment methods were used alone or by combination with sequential steps of EB and SE. In addition, a solvent extraction known as ‘clean fractionation’ was used to enhance the effectiveness of the pretreatment steps, and to recover extractives that could potentially provide value. The combination of techniques was expected to have synergistic effects that could improve fermentable sugar yield and reduce formation of byproducts that are typically caused by use of severe treatment conditions.

Pretreatments experiments were conducted in groups, where the conditions for the next group of pretreatment were selected according to the results from the prior work. The pretreatment conditions could be independently optimized for either the hardwood and softwood species, which due to their chemical differences responded in different ways.

Work with the pretreated samples was focused on three main categories: 1) compositional and structural changes in solid residues, 2) enzyme activity on the solid residues substrates 3) analyses of the co-products and their properties. All the obtained information was then used to populate a process model that was created on the basis of models published by National Renewable Energy Laboratory (NREL).

Materials

The wood feedstocks used in this project were Pine and Birch, provided by Bio Oil AS. The used materials were sawdust preparations with particle size below several millimeters. Wood was used without drying, with moisture content determined on weight basis. The EB pretreatments were conducted at Chappas Inc., US and the samples were then sent to NMBU, Norway, for SE treatments. After shipping the pretreated samples to NCSU they were stored in cold room for several weeks as received with high moisture content, and frozen for longer periods.

Several model materials were used during the project. These materials were majorly purchased from commercial supplier such as Fisher Sci. and Sigma-Aldrich. The used cellulose models for hydrolysis experiments were Whatman CF1 cotton cellulose, avicel microcrystalline cellulose, and fully bleached softwood kraft pulp. Ball milling was done using planetary laboratory mill when applied. Lignin models were either Indulin kraft lignin or EMAL lignin isolated from Pine sawdust at NCSU. Model compounds for irradiation studies were analytical grade ethylguaiacol and 1-methylglucose.

Pretreatment methods

In a pretreatment process of wood the native cell structures need to be altered in fiber level to increase fluid and mass transfer, and in molecular level to decrease crystallinity of cellulose and depolymerize/dissolve part the main components surrounding the cellulose fibrils. Used conditions need to be balanced to obtain highly active substrate, but simultaneously maintain good yield and prevent excess conversion of polysaccharide components into secondary non-fermentable compounds.

An innovative component of this project is the combination EB and SE, and this new combined pretreatment demands extensive optimization work to identify preferred operating conditions. The EB has been used for pretreatment of biomass, and is expected to effectively decrease cellulose crystallinity and assist depolymerization of lignin-hemicellulose matrix. Anyhow, the expected synergistic effects when combined to SE is unknown, and demands in exploration of the wide range of operating conditions.

ELECTRON BEAM (EB) IRRADIATION

Irradiation of wood with high energy electrons will have two effects. The first is the production of radical species directly from wood polymers, and the second is the production of hydroxyl radicals from the water present in wood. Based on the concentration and energy of the radicals they can lead to depolymerization of wood polymers, or coupling reactions. A decrease in cellulose crystallinity has been reported to follow from irradiation treatment (Iller et al.2002, Karthika, Arun & Rekha 2012). Radicals with lifetimes of months have been observed in wood lignin and low molecular weight phenolics.

STEAM EXPLOSION (SE)

Action of SE is based on the combined use of high temperature, moisture and pressure (typically from 2 to 15 minutes at 160 to 220 ºC) that will cause the rapid depolymerization of especially hemicellulose carbohydrates also known as ‘autohydrolysis’. The autohydrolysis can be accelerated by organic acids generated from the initial cleavage of acetyl groups, common in wood, to form acetic acid. At the higher reaction temperatures and longer times the autohydrolysis can also generate furan side products, which are detrimental to the downstream enzyme depolymerization and fermentation steps. During the autohydrolysis treatments, the pressure is suddenly released and the expansion of water vapor will cause secondary physical ‘explosion’ of the wood fibers resulting in a decrease in cellulose crystallinity and increase in surface area.

The SE process has been studied intensively, and the treatment severity can be adjusted by controlling temperature and retention time of the pressurized step. Generally SE conditions need to be adjusted according to the used feedstock materials, and is known to be less effective for hardwoods compared to softwoods.

The SE reaction temperature and retention time can be converted into one parameter to account for the interaction between the two parameters. The severity factor (Ro) was developed by Boussaid et al.

199, where Rois the severity factor, t retention time, and T temperature.

Ro=te<(T-100)/14.75)>

STEAM EXPLOSION CATALYZED WITH SULFUR DIOXIDE (SO2SE)

Acid catalysts have reported to increase the efficiency of steam explosion treatment significantly in terms of enzyme activity on residual solid wood (Mackie et al. 1985, Clark & Mackie 1987). Potential acidic catalysts are for example sulfur dioxide or sulfuric acid. Sulfur dioxide has the potential advantage of limiting the degradation of the lignin fraction, and increasing its solubility, while limiting degradation of the carbohydrate streams.

CLEAN FRACTIONATION (CF)

Clean fractionation is a specific subset of the more general organosolv processes, introduced by Bozell et al. 1995, and is intended to divide the main wood components into separate fractions using acid catalyzed, high temperature ‘pulping’ in ternary mixture of methyl isobutyl ketone (MIBK), ethanol, and water. In this process cellulose is remained as solid pulp, lignin and hemicellulose are transformed into solution phase where they can be easily isolated by phase separation of the solvent system into organic and aqueous phases. This process has the advantage of recovering a very pure steam of relatively uncondensed lignin, and a moderate molecular hemicellulose stream.

Applied pretreatment conditions and their optimization

Due to the limited access to the EB system our approach to the combined pretreatment reactions was divided into five groups of experiments, where conditions have been systematically modified. The conditions that were used in the pretreatments are summarized in tables 3 - 6.

GROUP 1 - HIGH SEVERITY EB AND SE

Single step pretreatment by SE was first tested with Birch and Pine woods using conditions that roughly correspond moderate to high treatment severity, based on prior work in the literature. Changes in the wood materials was significant, as the particle size of the samples reduced drastically. The resulting substrates were more of a thick slurry than a solid particle. Compositional analysis revealed essentially a complete loss of hemicellulose sugars.

Table 3 Group 1 – High severity EB and SE pretreatments

Prior to SE samples were irradiated with dosages of 10kGy, 100kGy, and 1000kGy. The resulting materials where recovered in very low yields and these samples were not analyzed by NCSU, but several analyses were performed at NMBU that are included in this report.

GROUP 2 - LOW TO MODERATE EB AND SE COMBINATIONS

A second group of Birch and Pine woods was subjected to EB and SE, with lower radiation dosages followed by lower severity SE compared to group 1. Samples were also exposed to the SE conditions without any EB treatment. This allowed for a direct comparison to the effects of EB treatment. Response of Birch and Pine woods to the pretreatments showed difference between the two species.

Table 4 Group 2 - low severity SE with and without preliminary EB treatments.

GROUP 3 - OPTIMIZED EB AND SE CONDITIONS

After relatively successful pretreatments results seen in group 2, the combination of EB and SE was further refined in group 3. Different severities were used for Birch and Pine to compensate the greater recalcitrance of the Pine. The data showed that temperature might be a more important variable than prolonged low temperature treatment, but decreases in SE severity can be compensated by increases in EB irradiation.

Table 5 Group 3 – further optimized EB and SE conditions, with specific optimization in relation to wood species

GROUP 4 - EB FOLLOWED BY SULFUR DIOXIDE CATALYZED SE

An alternative approach that has been used previously for overcoming Pine recalcitrance is to subject the samples to sulfur dioxide vapor prior to SE (SO2SE). While the SO2SE treatments produced major changes, they led to very low yields of solid substrate, and were economically unattractive.

Table 6 Group 4 - EB and SO2SE pretreatments of Pine

GROUP 5 – EB COMBINED WITH CLEAN FRACTIONATION

Due to the high expenses of multi-step pretreatment and subsequent solvent extraction, it was tested if electron beam alone combined with a clean fractionation process would offer accessible substrates for enzymes, with option of high lignin co-product yields. The clean fractionation treatments were done at University of Tennessee. The yields of solid substrates were considerably lower than after EB-SE pretreatments in set 2 or 3. Enzyme activity on birch substrates was high, but Pine did show significant recalcitrance like in other tested pretreatments. EB treatment prior to the clean fractionation treatments did not show any significant changes in substrate yields or enzymatic activity.

Table 7 Group 5 - Clean fractionation pretreatments with and without prior EB treatment.

*Substrate yields are calculated as follows m(residue, dry basis)/m(starting material, air dried), so moisture present in the starting material causes lower calculated yield for residue.

Analysis of pretreated wood

COMPOSITIONAL ANALYSIS

Compositional analysis was performed for all the samples following the same protocol. Compositional analysis was performed in order to determine the effects of pretreatment on the biomass composition.

Klason lignin and acid soluble lignin were measured following the Tappi procedure. Moisture content in wood was determined before the extraction. Wood sample was extracted 24h with Benzene-Ethanol 2:1 mixture. The extract was evaporated to dry in fume hood and further in vacuum oven at 40 ºC. Extractive content was calculated based on weight loss between dry mass of starting wood and dry mass of extracted wood. Samples from extracted wood were weighted accurately close to 100 mg (dry basis) and mixed with 1.5 ml of 72 % sulfuric acid. Samples were let to hydrolyze at room temperature with occasional stirring for 2 hours. The resulting black mixtures were transferred to 100 ml bottles using 56 ml of water. Diluted samples were further hydrolyzed in autoclave at 120 ºC for 1.5 hours. At this point of analysis samples had clear water phase and brown colored residue at the bottom of the bottles. The residue was filtered to pre-weighted sinters and dried in 105 ºC oven for 24 h. Sinters were cooled in dessicator for 20 min and weighted. Amount of acid insoluble lignin was calculated from weight gain. Filtrate was diluted to volume of 100 ml and 1 ml samples were taken and diluted to 50 ml. UV-absorbance of this solution was recorded at 205 nm wavelength, and amount of acid soluble lignin was calculated using Lambert-Beer equation with extinction coefficient of 110 l/g*cm. Another 20 ml portion was taken from filtrate and neutralized using approx.400 mg of calcium carbonate. After 2 hours the pH was tested to be in range between 5 and 6. Clear liquid was withdrawn from above the formed Ca2SO4 precipitation, filtered with 45 µm syringe filter and analyzed with HPLC. Concentrations of each sugar compound were calculated based on calibration with according standard solutions.

ENZYME HYDROLYSIS

Procedure followed at NCSU

The enzyme hydrolysis was performed at very high doses, 40 FPU/g, to determine the ultimate potential reactivity. The hydrolysis was also performed at relative low doses, 5 and 10 FPU/g, to test the biomass response with economically viable enzyme dose rates.

Moisture content of samples was determined and amounts corresponding to 1 g of dry mass were weighted accurately to 50 ml sample tubes. Solution of C-TEC II enzymes (concentration 1:100 in buffer solution) was measured to correspond 5, 10 or 40 FPU/g of substrate and added to sample tube. Needed volume of 50 mM sodium acetate buffer was added to gain 5 % consistency of wood. Samples were mixed shortly in vortex mixer and placed to shaker bath at 50 ºC for 72 h. During the incubation period, the samples were taken out twice per day and mixed shortly in vortex mixer. The residence time for the hydrolysis was 92 hours for the first set of samples and 72 hours for the subsequent sets of samples. After incubation residual solids were filtered to pre-weighted sinters, dried in 105 ºC oven for 24 h, and weighted after 20 min cooling in dessicator. Filtrates were diluted to 50 ml volume, filtered with 45 µm syringe filter and analyzed with HPLC. Concentrations of each sugar compound were calculated based on calibration with according standard solutions.

For experiments where water soluble components were hydrolyzed separately from the solids, the isolation of the water soluble material was performed during 4 h incubation in 2/3 volume of the used buffer solution, that was pressed out of the solids before the hydrolysis. The soluble materials and solids were then hydrolyzed by same protocol after addition of the remaining buffer and enzyme solutions (total 10 FPU dosage of enzymes was 1/6 for the soluble and 5/6 for solid material).

Procedure followed at NMBU

1.5 g DM of the sample, 15 ml 0.1 M Succinate buffer and 14.8 ml H2O were mixed in a 50 ml test tube. The solution was heated to 50°C before 0.2 ml Enzyme (Cellic CTech2 (Novozymes)) was added. The tubes were placed in a shaking incubator at 50°C. After 0, 4 and 24 hours a sample of 1 ml was taken out of the test tube. This sample was centrifuged at 14,000 rpm for 2 minutes and 0.7 ml of the supernatant was collected for HPLC analysis. The samples from sample sets 4 that had been treated with SO2were quite acidic and pH was adjusted with 1M NaOH prior to enzyme hydrolysis.

An additional experiment with extra addition of GH61 (oxidative enzyme) and Gallic Acid (GA; electron donor) was also performed. In this test 0.3 ml 0.1M GA, 0.489 ml GH61 and 0.198 ml Cellic CTech2 were added, keeping the total enzyme load at the same concentration as in the standard assay.

CHARACTERIZATION OF PRETREATMENT EFFECTS IN WOOD STRUCTURE

Molecular weight analysis by size-exclusion chromatography

Molecular weight distribution (MWD) analyses were carried out using a methodology specifically developed for the needs of this project. The protocol uses two different derivatization methods to preferentially measure the carbohydrate and lignin components.

Benzoylation for determination of carbohydrate MWD

For dissolution of wood prior reaction 1 g of ionic liquid was first weighted to 8 ml screw cap vial with magnetic stirring bar. Then 200 µl of NMI-Pyr 3:1 co-solvent mixture was added and solvent was homogenized with vortex mixing. Vacuum oven dried wood sawdust (particle size 0.85 – 0.25 mm) was added in 10 mg portion and mixed well using vortex mixer. Then sample was placed to 60 ºC oil bath, and let to dissolve during 18 h using stirring speed of 200 rpm. After the needed incubation time (typically 66 h), viscous solution still containing some solids was taken out of the bath and cooled for few minutes, and then 112 µl of benzoyl chloride was added. The reagent was mixed with the wood solution shortly (~5 s) in a vortex mixer, and then sample was left to react for 4 h at room temperature using slow magnetic stirring. Reaction was stopped by 4 ml addition of 75 % ethanol, and vortexing for one minute. Then the precipitated mixture was transferred to a centrifuge tube, and solvent volume was added to 20 ml. Solvent was removed using centrifuge, and resulting solids were washed twice with 20 ml of ethanol by shaking and centrifuging the solvent. Solid product was left to dry under low vacuum overnight, and further dried in a room temperature vacuum oven.

Acetobromination for determination of lignin MWD

Dried 10 mg sample was weighted to 8 ml screw cap vial and dispersed to 2 ml of glacial acetic acid. Reaction was started by adding 218 µl of acetyl bromide. Sample was protected from light and left to stir at room temperature with 300 rpm mixing. Typically 26 h time was used, if not specified for the single experiment. After the wanted reaction time, the dissolved sample was transferred to 50 ml round bottom flask, and solvent was evaporated using a high vacuum rotary evaporator. When all the solvent had evaporated, the sample was dissolved in 30 ml of DMF for analysis.

Isolated lignin samples were acetobrominated using the same reagents and solvent, but with only 2h reaction time due to rapid solubilization of the samples. Evaporation and sampling for GPC was conducted as described below.

Gel Permeation Chromatography (GPC) for measuring MWD

The benzoylated and acetobrominated samples were dissolved directly after derivatization and solvent removal in 20ml or 30ml of DMF, depending on the solubility. Samples were left to dissolve for one h for benzoylated samples and 15 min for acetobrominated samples prior filtration through 0.45 μm PTFE filter. Injection volume of 50 μl was used with manual injection. The GPC system consisted of an Agilent G1312A pump connected to Waters HT6E and HT2 styragel columns with a Waters 484 UV-absorbance detector. Detection wavelengths were 275 and 285 nm for the benzoylated and acetobrominated samples, respectively. Mobile phase of dimethylformamide (DMF) consisting 0.05 M lithium bromide (LiBr) was used. Calibration was done using either narrow polystyrene standards ranging from 1,860,000 to 820 Da, or benzoylated dextran and saccharides from 1,360,000 to 180 Da. Empower software was used for controlling the system operation and determination of specific MW parameters.

PRETREATMENT IRRADIATION EXPERIMENTS WITH MODEL COMPOUNDS

Model compounds (ethylguaiacol & 1-methylglucose) were weighted in 8 ml vials (2 g of sample), either as pure compounds, or dissolved in 5 ml of water or 50 % ethanol-water solution. Irradiations were conducted by Chappas Inc. using gamma radiation from cobalt-60 source. The three tested irradiation dosages were 0.1 kGy, 1.0 kGy, and 10 kGy, with the irradiation times of 1 min, 10 min, and 100 min respectively.

Polymeric cellulose (avicel) and lignin (indulin) were mixed in equal portions of 5 g in a 20 ml vial, and 1 ml of water was added to the mixture before sealing the vial. Mixture was homogenized by intensive vortexing. Same irradiation conditions were used than for low Mw models.

X-RAY DIFFRACTION

The crystallinity (CrI) of the pretreated birch samples was determined using a powder X-ray diffractometer. The operating conditions were an angular range 9 to 41 using the 2Ɵ method with a step size of 0.05 with 5 seconds of hold time. The analyzed samples were those treated with 25, 75, and 100 kGy plus the control sample (untreated birch). The CrI was calculated by the following equation; using the intensities of crystalline region at 2Ɵ = 22.5-22.5 and the amorphous region at 2Ɵ = 18 respectively.

Where I: Intensity

PARTICLE SIZE REDUCTION

Particle size distribution was determined by sieving method. Extracted and air dried wood material was placed in a series of sieves with openings of 0.85 & 0.25 mm. Sieves were shaken in mechanical beater/shaker apparatus for 6 minutes. Three fractions of particles were collected: Ones that remained on top of the first sieve (>0.85 mm), ones passed the first sieve (0.85 – 0.25 mm), and ones that passed both sieves (< 0.25 mm). The fractions were weighted, and corresponding weight fractions were calculated on basis of m% from total collected material.

SOLVENT EXTRACTIONS

Solvent extractions were carried out using a Soxhlet extractor apparatus. Extractions were done similarly for all types of the used solvents/solvent mixtures, with few exceptions for water extraction. The dried wood was placed to a cellulose extraction thimble (typically approx.10 g DM), glass fiber thimbles and moist samples were used in case of water extraction. Approximately 200 – 300 ml of solvent was loaded to the reservoir, and apparatus was placed of heater place. Temperature of the heater was adjusted so that solvent circulated with approx.1 drop per second rate. Extraction was continued 24 h. After extraction, the solvent was collected, and samples was dried under air flow and stored.

ISOLATION AND CHARACTERIZATION OF LIGNIN SIDE PRODUCT

Lignin was isolated from pretreated wood using the so called “clean fractionation” process, which utilizes solvent system composed of methyl isobutyl ketone (MIBK), ethanol, and water. The extraction was done in most of the cases using a soxhlet extractor, as described in the solvent extractions section. After the extraction, the liquid containing sample and the solvent was collected from solvent reservoir and place to separation funnel. When excess water is added to the solvent after extraction, components separate into two phases; one rich in MIBK or the other in water. Lignin and soluble carbohydrates (hemicellulose fractions) are partitioned into MIBK and water phases, respectively. Dissolved materials were recovered by evaporation of the solvent using rotary evaporator, and drying under high vacuum. Separation of the products between phases was measured by weight basis, and compositions of isolated products were analyzed by compositional analysis and proton NMR (<1>H-NMR). The MWD of recovered lignin fraction was determined by GPC.

Any lignin-rich fractions either from clean fractionation or other solvent extractions such as DCM could be purified using gradual precipitation technique. The whole material was typically dissolved in acetone, and then hexanes were added to the solution on very slow stepwise manner. The precipitating products were collected by centrifugation. The dissolution and precipitation steps were repeated 3 times for each solvent composition to ensure complete fractionation. The last fraction that could not be precipitated, was recovered by evaporation of the solvent, with composition majorly of hexanes.

BY-PRODUCT AND WATER SOLUBLE COMPONENT ANALYSIS

Extractive components that were isolated with benzene-ethanol extraction were analyzed by<1>H-NMR as follows. Samples were dried in vacuum oven at 40 ºC for 24 h prior sample preparation.10 mg of solid samples were weighted to small vial and 1 ml of DMSO-d6 was added. Dissolution was let to happen at room temperature using vortex mixing if needed. Samples were then transferred to NMR tubes. Spectra were recorded with 300 MHz spectrometer using 428 scans. Spectra were phased and baseline corrected and integrated with Spinworks software. Residual DMSO-d6 solvent signal was used as integration standard and received integral values were normalized using excel. Normalized proton integral ratios were converted to approximate weight ratios using average proton densities in model structures that correspond to compounds of interest.

Water soluble carbohydrates and lignin were quantified in similar manner than in compositional analysis of solid material, only exception was that hydrolysis in sulfuric acid was done only with dilute acid solution. Moisture content of samples was determined and amounts corresponding to 10 g of dry mass were extracted with 400 ml of water for 24 h. After extraction the total volume was diluted to 500 ml. Acidity of the sample extracts was determined by pH value using table top pH-meter. Next 56 ml samples were taken from extracts and 1.5 ml of 72 % sulfuric acid was added. Samples were then hydrolyzed in autoclave at 120 ºC for 1.5 hours. After hydrolysis, the following steps to measure acid insoluble lignin, acid soluble lignin and sugar composition were similar to compositional analysis.

Furfural and acid type by-products are water soluble and can be quantified after water extraction of pretreated wood. Relative acid formation was estimated by measuring pH of the water filtrates. Furfurals could be quantified by their UV-absorbance after the filtrate had undergone acid hydrolysis, using method published by Martinez et al. Wavelength used to record absorbance (A) of furfurals was 284 nm, and another wavelength 320 nm was used to subtract the effects of phenolic components from the results. Calculation could be done using following equation:

A284- A320= 0.127*total furan in mg/L 0.056

ENZYMATIC HYDROLYSIS OF CELLULOSE IN PRESENCE OF EXTRACTABLE COMPONENTS

Cellulose models were used in fibrous state (CF1 cellulose, bleached softwood pulp), or after mild ball milling of 10 min using planetary mill. Added materials were either lignin obtained from clean fractionation pretreatment, lignin isolated by MIBK phase after solvent extraction, DCM extractives, or fractionated parts of the aforementioned materials. The details about the used additive are reported with the results. The extractives or lignin were added to cellulose as a solution, typically in a plastic centrifuge tube, or as a solid dispersion in aqueous buffer. Organic solvent such as MIBK-EtOH-water and acetone were used to dissolve the added materials before addition to cellulose. The deposition of the additives was aided either by refluxing 2 hours in the solvent, using sonication of 8 minutes, or simply letting the solution to rest at room temperature or in freezer. The applied procedures are reported with corresponding results. After the deposition period, the solvent was evaporated with rotary evaporator, and the material in a tube was left to mild vacuum overnight. The hydrolysis was conducted following the NCSU protocol, as described in enzymatic hydrolysis section.

Process modeling

The overall process is shown schematically in Figure 3. Key elements from this process include the E-Beam irradiation, steam explosion, organic solvent extraction, and the recovery of the solvent. The pretreatment steps studied in Work Package I are highlighted in blue in Figure 3. The fermentation step, highlighted in green in Figure 3, is not part of this experimental work plan, and were calculated from the DOE NREL process models, which are in turn based on extensive experimental work. Thus, experimental data from the NREL 2011 report was used for conversion, and process simulation purposes. The ethanol recovery block, highlighted in yellow, can either be a standard distillation process or a low energy membrane system which was part of the broader project. The flows of sugars are combined to maximize the production of ethanol. The organic solvent extraction allows for recovery of an ‘extractives’ fraction. Finally, the biomass residues left over from hydrolysis and fermentation will be lignin-rich, but also include residual sugars and proteins from hydrolysis and fermentation processes. In the process model the residue stream is treated as a source of process heat.

It is important to note that the solvent extraction steps allow for recovery of a significant fraction of the starting biomass (10-40% on a dry basis), but also introduces significant complexity with the solvent recovery and recovery loops (highlighted in red).

Figure 3 Scheme of the unit processes from biomass to ethanol

The goal of populating a complete process model in Aspen Plus was to obtain mass and energy balance that allows for the comparison of the costs for configurations of the biorefinery process. The overall process model is based on a combination of two extensively studied Aspen Plus process models developed by the US DOE NREL. The first one is “Process design and economics for biochemical conversion of lignocellulosic biomass to ethanol” and the second is the ‘Clean Fractionation’ pulping process that is designed to completely separate biomass into its respective components with minimal contamination.

These well-known processes, in particular the “bioethanol model” have been extensively verified by DOE and engineering companies who have helped design the commercial biorefineries that are coming on stream today. When populating a process simulation, it is necessary to provide background information on specific steps to allow the model to converge and provide useful information. Also, these DOE NREL models provide data for preforming the economic analysis and serve as the basis for comparison.

In a selected set of modeling analyses these two processes combined to evaluate the integration of the clean fractionation into the biorefinery.

Characterization of pretreatment effects on composition of wood substrates

The chemical composition of the biomass before and after the pretreatment steps was tracked to follow the effects of pretreatment, and to maximize the yield and reactivity of the biomass. The analysis is performed in two stages; effect on the amount of soluble material removed during solvent extraction and the changes on the ratios of the polymeric components remaining in the solid residues after extraction.

The applied pretreatments were found to greatly affect the composition and structure of the wood through dissolution, depolymerization, and degradation of the main components; cellulose, hemicelluloses, and lignin. The composition of the starting materials is presented in table 8, and comparison to the results of the compositional analyses presented in tables 9 – 16 reveals the great effects of the pretreatments on the wood composition. In general, the treatments providing low contents of acid insoluble and acid soluble lignin in the solid substrate are ideal for the following hydrolysis treatments.

Table 8 Compositional analysis for the untreated material Carbohydrates % Lignin %

GROUP 1 - HIGH SEVERITY EB AND SE

High severity EB removed hemicellulose sugars very effectively, and especially for Birch, the resulting solids were solely composed of cellulose and lignin. A majority of hemicellulose were counted into mass balance as extractives. Lignin content in Pine did not show similar relative increase, and part of the hemicelluloses had remained in the solid substrate. Amount of material that was dissolved during benzene-ethanol extraction was very high for birch up to 39 %, while lower for Pine with only up to 22 % extractive content. Based extractive compositions determined by<1>H-NMR, large part of the soluble material is likely hemicellulose originated carbohydrates and small lignin fragments. Furfural formation (see section 7) was estimated at NMBU for the irradiated samples belonging to set 1 samples. Due to high severity of the SE, and especially due to the high irradiation dosages, the by-product formation is higher than presented for set 2 samples, but still remains in the level of several mass percentages based on the starting material.

Table 9 Compositional analysis of Birch and Pine substrates after high severity steam explosion

GROUP 2 - LOW TO MODERATE EB AND SE COMBINATIONS

Following the EB and SE pretreatments these samples were subjected to MIBK-ethanol-water extraction prior to enzymatic hydrolysis. This set of experiments show the “base case” for the enzyme hydrolysis, and also allow for the recovery of extractives. A more detailed analysis of the extractives can be found from sections 7 and 9. After the extraction the solids are analyzed in order to determine the changes in the material composition as shown in Table 4.3.

Table 10 Compositional analysis for MIBK extracted Birch based on pulp

Table 11 Mass balance for MIBK extracted Birch

When temperature is 170 ºC it can be observed on table 10 that there are no appreciable changes on the amount of xylan. When it comes about galactan, mannan and arabinan we can say that those sugars are totally hydrolyzed and/or degraded to furfural during the pretreatment process.

Based on the table 10, we can see that radiation is an important factor for the increasing in the soluble material removed during the extraction. Later, the effects on the amount and quality of the extractives will be discussed.

Lignin or solid residue after hydrolysis can be separated as acid soluble (A.S.R) and insoluble (A.I.R). The changes produced as results of the pretreatment can be observed on the figure 4. On this figure where we can see that as the severity of the pretreatment is increased, the amount of soluble residues decrease and those who are not soluble increase. Explanation to this can be efficient hydrolysis of hemicelluloses at high temperatures which increases relative acid insoluble lignin content, and simultaneous solubilization of low molecular weight lignins which usually are detected as acid soluble.

Figure 4 Change in glucan content in Birch as a function of steam explosion temperature and retention time. With and without radiation (left and right respectively)

Table 12 Compositional analysis for MIBK extracted Pine based on pulp

Table 13 Mass balance for MIBK extracted Pine

OPTIMIZED IRRADIATION AND STEAM EXPLOSION CONDITIONS (3<RD>SET OF SAMPLES)

The aim of the third set of pretreatment was to find optimum conditions of temperature, retention time and radiation dose based on the results of the second set pretreatments. It is notable that irradiation doses for birch samples are lower than what were used for Pine in this treatment set.

Prior compositional analysis benzene-ethanol extraction was performed, and the amount of soluble material has an increment while severity of steam explosion increase, as observed with pretreatment set 2. The results of the compositional analysis of the pretreated biomass are summarized in the following table 14. It was observed that excessive loss of xylan can be avoided by using short retention times. Used temperature setting of 185 ºC was applied for steam explosion, and showed good potential in preserving hydrolysable xylan in the solid residue, while being efficient in increasing the activity of the substrate.

Table 14 Compositional analysis for birch based on pulp

SULFUR DIOXIDE CATALYZED STEAM EXPLOSION (SO2SE) WITH IRRADIATION (4<TH>SET OF SAMPLES)

A SO2SE pretreatment was applied to Pine to overcome the recalcitrance that was observed with pretreatment sets 1 to 3.

All samples obtained from the SO2SE pretreatment (table 15) showed a large fraction of benzeneethanol extraction. The extractives were roughly double to Pine samples subjected to similar pretreatements but without SO2impregnation. Both the SE time and EB dose increased the amount of glucose and increased the amount of acid insoluble lignin, but neither effects was as great as the SO2SE pretreatment .

Lignin content in the residues recovered from samples subjected to SO2treatment increased significantly relative to native Pine wood. In mildest treatment (sample P-18) the lignin content is only moderately higher, but the most severe conditions (sample P-28) have resulted lignin content to be more than half of the substrate, up to 58 %. This seems logical when we take into consideration the difference in yield losses between the mildest and most severe treatments. In all treatment conditions especially xylan type hemicelluloses and most of galactoglucomannans have been removed from wood. The high lignin contents in benzene-ethanol extracted solid samples seem to follow from increasing cellulose conversion into extractable materials. From mild to severe conditions the cellulose content is decreasing, while relatively lignin content and amount of extractives are increased simultaneously.

Yield losses for mild SO2catalyzed treatments are coming from loss of hemicelluloses in form of furfural by-products or losses in water soluble form during the sample processing. Increasing the treatment severity will lead to increased degradation of cellulose.

Table 15 Compositional analysis of samples from Sulfur dioxide catalyzed pretreatment set 4.

Treatment conditions can be found from table 15 (below).

ELECTRON BEAM IRRADIATION COMBINED TO CLEAN FRACTIONATION (5<TH>SET OF SAMPLES)

Irradiation prior to clean fractionation treatment was tested, with aim to increase the lignin separation form the solid substrate into the soluble streams. This should have increased the cellulose content in the solid substrates, which is not seen in the table 16. Irradiation may have caused increased depolymerization of cellulose into water soluble fragments, which actually lowers the content of cellulose in the solid substrate. In case of Pine samples, the irradiation didn’t show any significant effects. Overall the severity of clean fractionation treatment seems to cause more intensive degradation and component solubilization than SE treatments, and thus mild degradation caused by irradiation doesn’t seem to have the same significance for the clean fractionation as it does for SE. For less severe clean fraction conditions, the effects of EB would likely be more obvious.

Table 16 Compositional analysis of Birch and Pine pulps after clean fractionation pretreatments.

Structural alteration of wood during the pretreatments

MOLECULAR WEIGHT ANALYSIS BY SIZE-EXCLUSION CHROMATOGRAPHY

Depolymerization of the wood polymers is an essential part of an effective pretreatment, and increasing the enzymatic hydrolysis of wood polysaccharides to allow for their conversion to fermentable monomeric sugars.

Molecular weight (MW) analysis of the SE Birch and Pine showed a decrease in the MW of the hemicellulose fraction, and simultaneously increasing MW of lignin (Figure 5). These trends are consistent with acid catalyzed condensation reactions at the high temperature used for the SE. The SE conditions used in this work did not cause a significant change in the MW of cellulose. This suggests that the cellulose structure may not be in key role in the differences that were observed between the wood species. Most importantly the analyses revealed a substantial difference between Birch and Pine lignin, in terms of their MW before and after pretreatment. Overall the Birch lignin showed monomodal molecular weight distribution (MWD), whereas both native and pretreated Pines samples showed bimodal MWD and a significantly higher MW. The MW in the pretreated substrates were higher than in the native wood, especially for the Pine. This behavior may help explain the recalcitrance of Pine, as higher MW lignin structures are more likely to prevent disintegration of the cellulose during the enzymatic hydrolysis.

Figure 5 Molecular weight analysis of steam exploded (200ºC, 6 min) Pine and Birch woods.

IRRADIATION EXPERIMENTS WITH MODEL COMPOUNDS

Degradation of wood components under the irradiation was studied by subjecting two model compounds, similar to repeating units of cellulose and lignin, to gamma irradiation. The experiments were done under air or nitrogen atmospheres, for neat compounds, or alternative solutions on water and water-EtOH. Polymeric cellulose and lignin were irradiated as mixtures containing 10 m% moisture. The applied irradiation dosages were 0.1, 1.0, and 10 kGy.

Results from the GC-MS analysis of the irradiated Ethylguaiacol (lignin model) showed no traces of any degradation products (figure 6), and the signal from starting material was giving very similar intensities than in control sample. The results could be confirmed with<1>H-NMR analysis, which as well showed no signals from degradation products. Similar NMR analysis gave same negative effects for the Methylglucose (cellulose model) samples.

Possible depolymerization effects of gamma irradiation on polymeric cellulose (avicel) and lignin (indulin) were studied by method where irradiated samples were benzoylated and analyzed by sizeexclusion chromatography. As a result the molecular weight distributions did not show changes that could be correlated with the applied irradiation dosages (figure 7), and the slight observed differences may have been caused by variation in the benzoylation reaction and subsequent washing step.

Figure 6 GC-MS analysis of several irradiated samples (left), compared to the non-irradiated control (right).

Figure 7 Molecular weight profile of benzoylated 1:1 cellulose-lignin mixtures after varying dosages of irradiation. Highest molecular weights in the figure are located at the low retention volumes, and decreasing towards the right end of the chromatogram.

Reason for the negligible effects of the gamma irradiation to the model compounds is unclear and experimental errors are the likely cause of these results. The similar irradiation dosages were used, but with electron beam type irradiation, to pretreat the wood samples, which showed substantial effects of the irradiation step.

DETERMINATION OF CRYSTALLINITY BY X-RAY DIFFRACTION

Crystallinity is one of the major contributors to slow enzyme hydrolysis on pure cellulose and native wood. This has been commonly associated with the inaccessibility of the bulk cellulose to the hydrolysis enzymes. Pretreatments are commonly increase surface area and depolymerize the carbohydrates, but effects in relative crystalline proportion of wood may not be evident due to the removal of the amorphous hemicelluloses.

Changes in the cellulose crystallinity of irradiated wood under some of the pretreatment conditions was studied by X-ray diffraction. The measurements can offer quantitative information about the relative proportion of crystalline and amorphous domains of cellulose, and can monitor changes with different EB treatments. Somewhat surprisingly the EB treatment between 25 kGy to 100 kGy only had a subtle effect on the overall crystallinity of cellulose as shown in Figure 8.

Figure 8 X ray diffraction normalized intensity of irradiated samples (0, 25, 75, and 100 kGy) as a function of diffraction angle (2 theta).

For the untreated sample, a crystallinity index of 53.6% was obtained, and when the irradiation dose was increased from 25 to 100 kGy the CrI goes from 55.5% up to 59.4%. When EB and SE are applied as two-step pretreatment, the enzymatic hydrolysis can reach up to 72% conversion of the total carbohydrate content in original birch wood, showing a significant improvement compared to SE alone. The hypothesis of the effect in the CrI increment is that it may have an effect on the total solid yield after SE. When samples are treated only with SE the solid yields are around 80% of the samples and when the samples are treated first with EB, yields can reach up to 96 % of the total mass. Other analyses need to be done in order to validate this hypothesis.

PARTICLE SIZE REDUCTION

Changes in the particle size distribution of the EB and SE pretreatment residues (i.e. particle size reduction) was determined after DCM extraction of the residues. The results are shown in figure 9. EB alone had no change on the particle size distribution, while SE significantly decreased the particle size distribution. When samples were subjected to EB prior the SE, the particle size distribution was even more significant. These observations indicate to synergistic effects between the two pretreatment methods, what it comes to breaking down the physical structures of the wood fibers, and changing their morphology. The samples subjected to SO2-SE pretreatment showed the greatest reduction in the particle size distribution.

Figure 9 The particle size distributions of DCM extracted pretreated substrates, shown as the weight fractions of the three size fractions that were separated by sieving.

Effects of the particle size to the enzymatic activity on the substrate were found to generally follow the trend where particle size reduction could notably increase the sugar release from the substrate, but as an exception this was found to be the opposite in case of highly digestible steam exploded Birch substrate.

Enzymatic hydrolysis of pretreated wood

Some of the pretreatment conditions were found to be too mild or severe to be effect, however a set of EB/SE conditions were identified that significantly enhanced the enzymatic hydrolysis of the woody biomass. There were major differences between the wood species. The hardwood (Birch) consistently showed higher hydrolysis rates and efficiency than softwood (Pine).

Enzymatic hydrolysis efficiency is strongly related to the amount of enzymes that is used per mass unit of the substrate. In this project the used enzyme loadings were chosen between an industrially reasonable loading of 5 FPU/g, and higher loading of 10 FPU/g that is more suitable for analytical work. Loading of 40 FPU/g was tested only with Pine to study mechanisms behind low enzyme activity on the softwood substrate, and expectedly resulted higher sugar release suggesting that irreversible binding or soluble inhibitors may also play a role in the enzyme reactivity.

Generally, the increase from 5 to 10 FPU loading was found to improve sugar yield significantly, but for very highly active birch substrates or the non-responsive pine substrates the lower 5 FPU loading reached nearly the same hydrolysis yield then 10 FPU loading. These results suggest that 5 FPU is not a limiting factor if the woody substrate has been successfully converted to highly active form.

The Pine sample had low reactivity for 5 and 10 FPU. Interestingly, when extremely high 40 FPU loading was tested with moderately active pine substrates (set 3), the increase in enzyme loading clearly improved sugar yields. Addition of excess enzymes, and following significant yield improvement suggests a type of inhibition where non-specific binding to lignin or formed by-products prevent the action of added enzymes.

Enzymatic hydrolysis tests showed clear improvement as the SE temperature was increased between 170ºC and 200ºC. Birch samples responded to the treatment with varying degree. At lower temperature the enzymatic hydrolysis was inefficient but higher temperature offered accessible substrates with short 2 min and longer 6 min retentions times. Effects of EB could be seen most clearly with long retention time at 170ºC or short retention at 200ºC, in which conditions the SE alone did not reach high activity, but combination with EB increased the effectiveness of the treatment up to 1.5 to 3 fold.

Overall, the Pine samples responded poorly. While both the SE and EB could improve reactivity, the overall sugar conversion at reasonable enzyme dosage was commonly less than half that of Birch.

HIGH SEVERITY STEAM EXPLOSION (1<ST>SET OF EXPERIMENTS)

BioOil had sent samples of fresh Birch (B), Pine (P) and Spruce (S) to NC State University for pretreatment by 3 different radiation levels (10-100-1000).300 g (wet weight) of each these samples was returned to Norway and pretreated by Steam explosion (10 minutes, 210 C). Subsequently these double pretreated samples were enzymatically hydrolyzed.

The steam exploded samples were hydrolyzed by Cellic CTech2 (Novozymes) at 50 g/l DM. Figure 6.2 shows the amount of released sugars after 24 h of hydrolysis at 50 °C.

As expected Hardwood cellulose is easier to hydrolyse and Birch shows the highest amount of sugar release, but the highest radiation levels are negative. The release of glucose from the softwoods pine and spruce were rather similar but clearly lower than for birch.

It is known that harsh pretreatments may degrade C6 sugars (such as glucose and mannose) to hydroxy-methyl furfural (HMF) and C5 sugars to furfural. UMB therefore analyzed these compounds in the hydrolysates.

As expected the release of furfural was highest for the birch samples which contain high levels of xylan (C5 polymer). For all three wood types the release of both furfural and HMF increased by radiation doses. Especially going from 100 to 1000 in radiation led to a sharp increase in furfural production.

Birch shows the highest amount of sugar release, but the effects resulting from the highest radiation levels (1000 kGy) are negative. The release of glucose from the softwoods pine and spruce were rather similar but clearly lower than for birch. The highest radiation levels seem to for lead to a lot of degradation of C6 sugars in the softwood samples. Thus, for the following experiments, the radiation levels where limited to 100 kGy.

LOW SEVERITY STEAM EXPLOSION AND COMBINATION WITH ELECTRON BEAM IRRADIATION (2<ND>SET OF EXPERIMENTS)

The results from the enzymatic hydrolysis of the samples from set 2 provided significant differences sugar yields for Birch samples that would be important for commercial large scale operations. Still these results left a significant amount of sugar unhydrolyzed, and this sugar would commonly end as a waste residues fraction following distillation.

Figure 11 shows the release of soluble sugars after enzymatic hydrolysis for samples pretreated by both irradiation and steam explosion. The sugars in Birch are clearly most accessible, followed by Pine and Spruce, respectively. Figure 12 and 13 shows the initial release of sugars over the first 24 h of hydrolysis. Increasing the SE temperature from 170 to 200 °C, and residence time from 2 to 6 minutes, increases the sugar release in Birch.

Figure 10 Released sugars after enzymatic hydrolysis of materials from irradiation combined to severe SE treatment. (Norwegian University of Life Sciences (NMBU))

Figure 11 Released sugars after enzymatic hydrolysis.50 g/l DM, hydrolysis at 50°C for 24 h.

(Norwegian University of Life Sciences (NMBU))

Figure 12 Glucose release from thepretreated samples as function of hydrolysis time. (Norwegian University of Life Sciences (NMBU))

Figure 13 Enzymatic release of mannose (from P and S) and xylose (from B). (Norwegian University of Life Sciences (NMBU))

Figures 14 & 15 show the sugar release from solvent extracted Birch and Pine substrates, with two different enzyme loadings. The effect of treatment temperature is substantial for Birch, but almost negligible for Pine. Part of the very low sugar yield may be explained by the extraction of the most reactive carbohydrates fragments during the solvent extraction. Simultaneously, soluble inhibitors would have been removed by solvent extraction, these results indicate to presence of a solid inhibitor/enzyme binder such as lignin.

Since in a commercial plant the introduction of a organic solvent system, and the associated capital and power required for solvent recovery, will require large investment and increase in operating cost, the potential for a water extraction system was studied as alternative way to remove possible soluble inhibitors to improve conversion of the Birch substrate. Separate hydrolysis of hemicellulose and cellulose offers option for process optimization, and is used by the plants designed by both Chemtex/Beta renewables and Dong Energy. In this work water extraction would allow for further processing of the hemicellulose sugars into value-added co-products.

To measure the effects of water extraction, one experiment was conducted with separated hydrolysis of the solid and water soluble streams. Results in the Figure 16 revealed that this approach did not bring positive effects compared to hydrolysis of the whole mixtures, but showed good potential for production of xylose type sugars in a separate stream, without penalties in terms of total sugar yield.

Series of hydrolysis experiments were done to address the differences between Birch and Pine. For these hydrolysis, the materials were extracted with DCM to remove the low molecular compounds that could cause inhibition, and resulting materials were sieved to homogeneous particle sizes. With this approach the results can be attributed hydrolysis without any particle size effects. Results in figure 17 indicate that chemical deactivation of enzymes in solution was not taking place, but particle size related kinetics seem to affect especially with Pine samples, yet the differences between the species remained large.

Figure 14 Sugar recovery based on the substrate for MIBK extracted Birch samples

5 FPU 10 FPU

1

5 FPU 10 FPU

Figure 15 Sugar recovery based on the substrate for MIBK extracted Pine samples

Figure 16 Sugar release from steam exploded Birch (200ºC; 6min), where hydrolysis was done separately for water soluble and water insoluble materials (separated).

Figure 17 Sugar release from DCM extracted substrates after enzymatic hydrolysis (Done with homogenized particle sizes to remove particle size effects between two different starting wood preparations).

OPTIMIZED EB AND SE CONDITIONS FOR DIFFERENT WOOD SPECIES (3<RD>SET OF EXPERIMENTS)

For the third set of pretreatments the conditions can be divided for those used for Birch and Pine, where the latter represent higher severity. Conditions for Birch could be optimized for high sugar yields, but less success was archived with Pine samples, even though promising effects were seen.

Birch samples

Overall the Birch responded well to the optimized conditions. Applied enzyme dosages between 5 and 10 FPU showed greatest differences for the conditions where the activity did not reach the highest measured levels, but for highly active substrates the industrially more viable 5 FPU loading offered considerably high yields.

185 ºC SE treatment was used for Birch and showed good performance with longer reaction times, but was less effective for short reaction times. This shows how the autocatalytic reactions require certain severity to become effective for the pretreatments (measured to be around 3 in ROscale), and this triggers significant increase in the reactivity. As seen with the earlier pretreatments, the EB could improve the yields especially for the samples where the SE conditions were close to autohydrolysis activation energy.

Several of the samples from set 3 were extracted with MIBK and DCM solvent extractions. Enzymatic activity remains high in the birch substrates after MIBK-ethanol-water extraction. All tested Birch substrates had around 50 % conversions into simple sugars. These values are slightly lower than those that has determined for same substrates directly after the pretreatment. Explanation is likely that extraction step is removing some of the most accessible carbohydrates, and lowering the overall carbohydrate content. Further work is required to see if the extracted carbohydrates can be returned to enzyme hydrolysis process train after they are removed from the MIBK or DCM organic phase. MIBK-ethanol-water extraction before enzymatic hydrolysis might increase the total sugar release by reduced inhibition compared to hydrolysis directly after pretreatment, since large proportion of lignin is removed to the isolated MIBK phase. The low molecular weight extractives do not cause inhibitory effects based on the results shown with hydrolysis of DCM extracted set 2 samples and model experiments presented in section 10. Main motive to study the MIBK extraction with these materials was the possible integration of lignin co-product production into the ethanol process, which requires the extraction unit.

Figure 18 Enzymatic hydrolysis efficiency results based on the birch substrate (25 kGy) 80%

70%

60%

50%

50 kGy 185 C 4 min 40%

50 kGy 200 C 4 min 30%

20%

10%

0%

5 10

Enzyme Dose (FPU/g of substrate)

Figure 19 Enzymatic hydrolysis efficiency results based on the birch substrate (50 kGy)

0%

0%

0%

0%

0%

75 kGy 185 C 2 min 0%

75 kGy 185 C 4 min 0%

75 kGy 200 C 2 min 0%

75 kGy 200 C 6 min 0%

0%

0%

5 10

Enzyme Dose (FPU/g of substrate)

Figure 20 Enzymatic hydrolysis efficiency results based on the birch substrate (75 kGy)

Figure 21 Release of glucose after 0, 4 and 24 hours of hydrolysis (June 2012). (Norwegian University of Life Sciences (NMBU))

Table 17 Enzyme activity on solid residues from MIBK-EtOH-water extracted Birch

Pine samples

P15-P17 were EB treated with 100 kGy, and SE at 200°C for 2, 4 and 6 minutes. The highest yield of glucose is obtained for P17 the most severe treatment (6 min). P23-P25 were irradiated at 200 kGy and glucose yield is similar for all three samples. Thus, by increasing the EB dose from 100 to 200 Gy, the sugar yield can be increased or the residence time in the SE unit can be reduced.

Different enzymes were also tested for these samples, which are high in mannose sugars. An additional experiment with extra addition of GH61 (oxidative enzyme) and Gallic Acid (GA; electron donor) was also performed. In this test 0.3 ml 0.1M GA, 0.489 ml GH61 and 0.198 ml Cellic CTech2 were added, keeping the total enzyme load at the same concentration as in the standard assay. The results are shown in figure 23. These results indicate that extra GH61 and GA did not increase the yield of glucose.

Figure 22 Release of glucose after 0, 4 and 24 hours of hydrolysis (June 2012). (Norwegian University of Life Sciences (NMBU))

Figure 23 Release of glucose after 0, 4 and 24 hours of hydrolysis (December 2012).

Figure 24 Efficiency of the enzymatic hydrolysis of Pine using low and high enzyme loadings, measured in basis of weight loss

A select group of Pine samples that were subjected to high severity treatments (P15, P25, here MP15, MP25) in addition to corresponding SO2impregnated samples (P18, P28, here MP18, MP28) were extracted with MIBK-EtOH-water at NCSU and sent to enzymatic hydrolysis to NMBU. Figure 6.16 shows yield of glucose for the extracted samples which are consistently low.

Figure 25 Release of glucose after 0, 4 and 24 hours of hydrolysis (December 2012). (Norwegian University of Life Sciences (NMBU))

SULFUR DIOXIDE (SO2-SE) CATALYZED STEAM EXPLOSION WITH IRRADIATION (4<TH>SET OF EXPERIMENTS)

The (SO2-SE) impregnation was tested as an option for recalcitrate Pine samples. Based on enzymatic hydrolysis using 10 FPU enzyme dosage the enzyme reactivity of SO2-SE pine increased compared to non-catalyzed samples. Anyhow, the efficiency of sugar hydrolysis does not reach the limit of sugar availability in the substrate that is roughly estimated to be between 50 to 60 % based on amount of measured sugars in compositional analysis and sugars in extractives.

Sugar release values indicate that best conditions are 100 kGy EB with severe SE step, or 200 kGy EB with low severity SE. Severe conditions lead to increasing lignin content in substrate which is showing as low sugar release.

As a result of the most severe treatment conditions, the fibrous structure of wood was nearly completely destroyed, resulting very low water retention capability for these samples. The collected samples were heterogeneous slurries with separate water and solid materials, whereas for low severity samples the solid materials could retain most of the water. Due to packing problems during the logistics of the samples, part of the water phases were lost, which could in term explain the loss of water soluble sugars from the high severity samples, and explain the discrepancies that can be seen for especially P-25 samples between the Figures 26 & 27. Another possible source of error in these particular experiments was the difficulty to adjust the pH in the samples treated with SO2since due to acidic residues from the pretreatment. Sample 26 has low yield of glucose and this might be because the pH was not optimal for the enzymes.

Figure 26 Sugar release from SO2impregnated Pine samples after 72h hydrolysis.

Figure 27 Release of glucose after 0, 4 and 24 hours of hydrolysis (December 2012). (Norwegian University of Life Sciences (NMBU))

ELECTRON BEAM IRRADIATION COMBINED TO CLEAN FRACTIONATION (5<TH>SET OF EXPERIMENTS)

Based on the four tested samples, the electron beam irradiation doesn’t have any enhancing effects on the clean fractionation type pretreatment process. Contrary to the expected result, the sugar release from irradiated Birch decreased compared to the control sample.

The recalcitrance of Pine substrate remains through clean fractionation, even nearly all of the hemicelluloses are being removed from the substrate. This strongly indicates to lignin related recalcitrance. Only the increased accessibility following from hemicellulose removal is not enough to increase enzyme activity on pretreated wood substrate.

Table 18 Enzymatic hydrolysis of clean fractionation pretreated Birch and Pine pulps.

EVALUATION OF THE SYNERGISTIC EFFECTS OF EB AND SE FOR OPTIMAL ENZYME ACTIVITY ON BIRCH

Enzymatic treatments of Birch resulted nearly complete hydrolysis of all polysaccharides after several combinations of pretreatment conditions. Figure 18 presents four of the most promising treatment conditions for comparison. As we can observe the three factors, irradiation dosage, temperature, and retention time are not constant for the successful treatments, but work more as a complex combination. The results from Birch treatment sets 2 and 3 were summarized using severity factor as measure of SE severity. This way both of the two factors affecting to SE severity could be plotted against the used EB dosage. From figure 19 we can see how increase in SE severity is the controlling factor for sugar release, and the most visible increase in sugar release as function of steam explosion severity can be observed around log Rovalues of 2.8-3.2. Graphs in figure 20 suggest that supplementing effect of EB is also strongest around this same log Roregion. Surprising is the largest measured sugar release with 75 kGy EB , when compared to the 100 kGy dosage.

The optimal conditions for the pretreatment would be those that offer close to maximum sugar release with minimal severity of the two pretreatment steps. From this summary we could conclude that the conditions around 75 kGy irradiation followed by steam explosion of approximately log Ro~ 3 severity may offer the most energy efficient pre-treatment conditions.

Figure 28 Based on these three conditions of radiation (25, 50 and 75 kGy), the best EB-SE combinations for improving enzymatic hydrolysis efficiency were chosen in one plot for comparison.

Figure 29 Summary of enzymatic sugar release using 10 FPU load from all analyzed birch samples as a function of steam explosion severity factor (log Ro) and the electron beam irradiation dosage.

Figure 30 Evaluation of optimal irradiation dosage and steam explosion conditions, in terms of the synergistic effects. This was done by determining the relative increase caused by increasing irradiation dosage, at different severities of the steam explosion.

Quantification and analysis of hydrophobic extractives, water soluble carbohydrates, and by-products

HYDROPHOBIC MATERIALS EXTRACTED WITH ORGANIC SOLVENT EXTRACTION (EXTRACTIVES)

As a part of compositional analysis, extractives components were removed from solid pretreated substrates by soxhlet extraction. Benzene-Ethanol mixture 2:1 was chosen as a solvent to be used in the standard compositional analysis procedure. Acetone and dichloromethane (DCM) were also tested for more selective isolation of small molecular weight hydrophobic extractives. Results with acetone were similar with benzene-ethanol, but DCM offered higher selectivity that can be seen as lower extraction yields.

MIBK-Ethanol-water extraction was studied as it is the solvent system that is used in clean fractionation process , and could possibly be used to isolate lignin prior enzymatic hydrolysis unit process

Based on figures 31 – 34 it is clear that EB is an important factor that increases the soluble materials removed during extraction. For non-irradiated SE samples (figure 31) we can see that temperature does not have a big impact but reaction time does. However, for samples treated with 100 kGy of EB there is a significant temperature effect increasing the extractable material up to 35 wt% of the sample. This could be explained by radiation degrading or deconstructing the matrix making easier for the steam to penetrate wood changing or modifying the structure of lignin, hemicelluloses and cellulose. The increment of the extracted material indicates the dissociation of the lignocellulosic complex. The same behavior can be observed on the figures 32 & 33 which include extraction data for Birch samples from three different extraction systems. The DCM extraction was showing the same trends for Birch and Pine, but the yields of extracted materials were significantly lower. This is due to the low polarity of the DCM and the relatively polar nature of the degraded wood fragments.

Figures 35 & 36 show NMR analysis about composition of MIBK-EtOH-water extracts from SE Birch, in addition to the Benzene-EtOH extracts of SO2treated Pine. Both sets of samples show the large carbohydrates portion in the extractables, which corresponds to the overall sugar yield losses following extraction.

Figure 31 Effect of electron beam irradiation on the amount of soluble material extracted from Pine (set 2) using MIBK-Ethanol-water. (Above) Pine treated with steam explosion (Below) Pine treated with combination of electron beam and steam explosion

Figure 32 Amount of Benzene-ethanol extractives as a function of treatment severity. Samples are from treatment set 3.

Figure 33 Comparison between three tested solvent systems in terms of the quantity of isolated material from pretreated wood.

Figure 34 Amount of DCM extractives as a function of treatment severity. Samples are from selected pretreated Birch and Pine substrates.

Figure 35 Composition of MIBK-EtOH-water extraction of steam exploded Birch samples from pretreatment set 2. Increasing amount of sugars are depolymerized and turned more soluble with increasing steam explosion severity.

Figure 36 Composition of Benzene-Ethanol extractives of two SO2treated Pine samples from pretreatment set 4

FURFURALS AND ACIDIC BY-PRODUCTS

Samples from pretreatment set 2 were analyzed to quantitate the acidic by-products. The analyses were done from the samples that were subjected to water extraction. From measured pH-values of water extracts (figure 37), there are clear differences between samples treated at 170ºC and 200ºC, with the higher SE temperature producing a lower pH. The basic values, above pH 7.0, which were measured for some samples is very surprising since even the relatively mild pretreatments should produce acetic acid and an acidic pH. This might be due to a calibration error, but the trend matches the expectation. Acid formation was noticeable at 200ºC, which is in good agreement with the expected de-acetylation as a function of treatment temperature. While the same trend is observed from Pine the pH did not drop below 4 except for the most severe conditions. This indicates that acetic acid formation, and the following beneficial depolymerization of carbohydrates is more efficient in case of birch.

Figure 37 Acidity of water extracts from selected pretreatment set 2 samples.

It is known that harsh pretreatments may degrade C6 sugars (such as glucose and mannose) to hydroxymethyl furfural (HMF) and C5 sugars to furfural. These compounds were analyzed in the hydrolysates of the most severe pretreatment conditions of sample set 1 (Figure 38). As expected the release of furfural was highest for the birch samples which contain high levels of xylan (C5 polymer). For all three wood types the release of both furfural and HMF increased by radiation doses. Especially going to very high EB doses, 100 to 1000 kGy, led to a sharp increase in HMF production. Thus, the highest radiation levels seem to lead to extensive degradation of C6 sugars in the softwood samples.

Figure 38 Concentration of HMF and furfural in 50 g/l hydrolysates. (Norwegian University of Life Sciences (NMBU))

The furfural type by-product formation for sample set 2 was not very significant, as it is shown in the figures 39 & 40. Conversion into furfural was lower than 2 % in all cases compared to dry mass of the solid substrates, and even under 1 % with pine.

The condensation tank of the SE unit contained no traces of furfural at 170 °C treatments. Low amounts was found after 200 °C treatments (0.03 g for B, 0.04 g for S and 0.07 g for P).

Figure 39. Sugar degradation products formed in SE. Measured at 50 g/l DM. (Norwegian University of Life Sciences (NMBU))

Figure 40 UV determined furfural HMF concentration as function of treatment conditions determined from birch and pine

Co-products from water extraction

EXTRACTION OF WATER SOLUBLE CARBOHYDRATES AND LIGNIN

Water extraction is one practical option that could be integrated to the bioethanol biorefinery for the purposes of co-product production. Samples from pretreatment set 2 were analyzed to quantitate components that turn water soluble during the pretreatments. Figure 41 shows how the concentration of water-soluble carbohydrates increase as the treatment severity increases. Also, the concentration of water-soluble lignin derived phenolic fragments increase with increasing severity.

Figure 42 Conversion of carbohydrate polymers in wood into water soluble low molecular weight compounds during the EB-SE treatment set 2.

Sugar composition of both Birch and Pine water extracts corresponds well with natural composition of specific hemicelluloses for these species. Tested pretreatment conditions are not causing significant conversion of cellulose into water soluble glucose oligomers, which indicates to low effects to the well orientated cellulose fibrils.

Lignin co-products by solvent extraction

One aim of this project was to develop biorefinery options that include more than just ethanol, but also other products suitable as feedstock for chemical industry. In addition to the natural extractive compounds there are two other options for chemicals, polysaccharides and aromatics. Isolation of these main components was tested with solvent systems that are presented in the section 7. Main goal of the work conducted in this section was to find out could higher MW lignin could be isolated, and could the extracted sugars be returned into the hydrolysis/fermentation stream to avoid yield losses to final ethanol yields.

APPLIED EXTRACTION PROCESSES AND SELECTION OF SAMPLES

Two different variations of MIBK-Ethanol-water were used in this project. The main difference between them is that clean fractionation conditions are much more severe and combined with acid catalyst in manner that the method alone can work as an alternative pretreatment, whereas Soxhlet extraction imitates more a washing type process that has been included to the modeled process introduced earlear.

Isolation of lignin was different between two methods (table 19). In clean fractionation the extracted solvent is evaporated entirely and residue is washed with water to isolate lignin as an insoluble residue. After soxhlet extraction the ternary solvent mixture was phase separated by water addition and then water and MIBK phases were separately evaporated to dry, and the residues were collected without further purification.

Samples for the soxhlet extraction (table 21) were selected based on high expected sugar yields, based on the enzymatic hydrolysis. This makes them the most attractive for the process

Table 19 Conditions that were used in two different solvent extraction methods to isolate lignin

Table 20 Samples that were treated with clean fractionation system. Control samples are non-treated starting materials. Treated samples have gone through electron beam irradiation prior CF-treatment.

Table 21 List of samples with varying pretreatment conditions that were chosen for lignin isolation using soxhlet extraction system.

RECOVERY OF LIGNIN RICH PRODUCT AFTER CLEAN FRACTIONATION EXPERIMENTS

This addition of water soluble reagents prevents analytical quantification of water soluble residues, nevertheless in the actual process the water soluble stream can be returned to the hydrolysis/fermentation. Lignin could be isolated when MIBK-EtOH-water is first evaporated, and residue is washed with excess water. With clean fractionation the lignin recovery reached higher yields compared to soxhlet type extraction (table 9.4). There was no significant effects to resulting yields by combination with electron beam irradiation.

Table 22 Mass balances of lignin materials that were isolated by clean fractionation. Water soluble materials were not isolated or quantified due to water soluble inorganic chemicals added during the treatment.

COMPONENT RECOVERY AFTER SOXHLET EXTRACTION AND SEPARATION OF MIBK AND WATER PHASES

As the treatment conditions for birch samples were relatively similar, the resulting mass balances for materials that were dissolved during the extraction are very similar for all four samples. Materials that were isolated from water phase after the extraction corresponded to 11.3 to 13.4 % of starting materials, and yields of MIBK isolated materials between 6.0 and 6.8 %. For SO2catalyzed pine increase in treatment severity clearly affected to amounts of extractable materials, and overall yields of extractables were up to 59 % of starting material. Majority of soluble materials were carbohydrates in all cases, but increasing treatment severity seemed to increase amount of dissolved lignin in relation to carbohydrates. Yields of all recovered materials after phase separation are listed in table 23.

Table 23 Mass balances of materials that were isolated by solvent extraction and separated to water and MIBK fractions by phase separation.

When water was used to separate single phase extraction solvent containing MIBK-ethanol-water (62:27:11) into two separated phases of MIBK and water, there was a precipitation of dark brown solid taking place. The yields of precipitated solids were not significant in case of birch, but for pine samples yield reaches up to 6.3 %. Solids were not fully soluble in DMSO, so NMR could not be used to analyze the composition. Small portion of solids that could be dissolved and analyzed by NMR, showed signals in aromatic region indicating to significant lignin content.

CHARACTERIZATION OF PRODUCTS FROM MIBK AND WATER STREAMS AFTER SOXHLET TYPE EXTRACTION

Quality of the products from solvent extraction and phase separation was evaluated mostly in terms of lignin content. Most of the analyses were done to the isolated lignin rich material from the MIBK stream, but several to study the composition of the sugar rich water stream. Compositional analysis by acid hydrolysis, and<1>H-NMR based analysis showed that lignin fractions of around 90 m% acid contents based on acid hydrolysis could be isolated without any further purification steps, but NMR analysis suggests that these values may be overestimations due to presence of extractive compounds that are not acid soluble. The residual impurities were determined to be sugars, and especially the hydrophilic extractive components. Details of the analyses are shown in tables 24 – 26.

Acetobromination procedure (Asikkala et al.) is a suitable method for selectively examining the molecular weights of lignin components. Lignin is modified chemically by reaction with acetyl bromide, and the ensuing products become soluble into organic solvents like THF. The carbohydrates that are present in the sample are not visible in size-exclusion chromatograms due to their inability to absorb with UV light at 280 nm that is used for detection.

Molecular weight analysis revealed that lignin isolated from MIBK phase is composed of large fractions of oligomeric structures having in average 3 to 4 phenylpropanoid units. Lignins from SO2treated Pine are composed mainly of oligomer sized molecules like the lignins from Birch. Exception is that the severe SO2-catalyzed treatment seems to solubilize also some larger molecular weight lignin polymers. This is expected when we look at the yields of isolated lignins in table 23.

The solids that were precipitating during phase separation were also examined using acetobromination, but these samples were not completely soluble during the derivatization reaction. Anyhow, parts of these samples could be analyzed and showed significantly higher molecular weights than the lignin samples isolated from MIBK phases. It might be possible that the solvating characteristics of both layers in phase separated extraction system were too weak to keep high molecular weight lignin in a dissolved state. As such the larger lignin polymers precipitated during the phase separation process. Another explanation is that possible crosslinking reactions within lignin took place within the soxhlet extraction system when the extracted solution was heated for solvent circulation. Insolubility of the solid material in either DMSO or acetic acid would indicate to crosslinking reactions.

Table 24 Composition of MIBK residues of extracted birch samples. n.d = not determined; trace = signal hardly observed; low = low intensity signal; high = intense signal estimated to correspond major part of the sample

Table 25 Composition of MIBK residues of extracted Pine samples. n.d = not determined; trace = signal hardly observed; low = low intensity signal; high = intense signal estimated to correspond major part of the sample

Table 26 Relative composition of MIBK-ethanol-water extracted products, and benzene-ethanol extractives determined by<1>H-NMR.

Figure 43 Yields of MIBK-ethanol-water soxhlet extracted and phase separated components based on solid substrate (determined by<1>H-NMR).

Figure 44 Molecular weight distributions of Birch and Pine lignin isolated by soxhlet extraction and phase separation

Table 27 Molecular weights of lignin materials isolated by soxhlet extraction and phase separation.

Main peak = Molecular weight region where signal is strongest; Mn= Molecular weight of average polymer having highest population in the sample; Mw= Molecular weight of average polymer responsible for majority samples mass; PDI = Measure for width of molecular weight distribution (Mw/ Mn)

PURIFICATION OF LIGNIN RICH EXTRACTED MATERIALS BY SUBSEQUENT ORGANIC SOLVENT FRACTIONATION

Fractional precipitation was tested as a method to purify the lignin rich materials that were obtained from solvent extractions. This was especially targeted for isolation of the hydrophobic compounds from lignin. By primary dissolution in acetone, and subsequent reduction of the solvent polarity by hexanes, it’s possible to precipitate lignin polymers in order of reducing molecular weight, and eventually have fractions of solid lignin and dissolved extractives that can be isolated by solvent evaporation (figure 45). The preliminary results showed potential to be used as purification method for lignin co-products, but large scale operations were not evaluated.

Figure 45 Separation of the DCM extractives from steam exploded Birch (200C, 6min) into purified fractions according to solubility in series of solvents.

ANALYSIS OF HYDROPHOBIC COMPONENTS OF THE EXTRACTIVES

Amount of hydrophobic small molecular weight compounds in the total MIBK-EtOH-water extract stream was analyzed for one pretreated birch (B-24) sample. In this preliminary analysis the MIBK-EtOH-water extractives were added to weaker organic solvents and soluble products were analyzed by gas chromatography. Whole separation scheme is presented in appendices 18 - 21 As expected the total yield of the hydrophobic extractives was small.

Identified compounds are those that were identified with over 90 % accuracy, and they present only a part from all of detected compounds. Phenolic compounds were not detected which may be due to low volatility, or solubility in the analyzed solvent fractions. Due to the low yield and complexity of the hydrophobic extractives, these compounds are not a promising alternative to be isolated as side products of the bioethanol process and low oxygen content components could be more viable to be used in energy production by combustion.

Table 28 Amount of hydrophobic extractives and identified compounds in pretreated substrate.

Extractives are isolated MIBK-EtOH-water soluble material with subsequent extractions with pentane and ethyl acetate.

Yield enhancing action of the extracted compounds on the enzymatic hydrolysis of cellulose

As the hydrophobic components (extractives, parts of the lignin) are removed during the solvent extraction, action of the removal of these components to the enzymatic hydrolysis of cellulose was studied. Based on the existing literature around the topic, at least part of these removed components were expected to act as enzyme inhibitors (Ximenes et al.). This would reduce the yield of glucose from the samples. In the large scale process it may not be feasible to perform extensive solvent extraction, and this may cause yield losses due to inhibitory effects of the extractives that remain on the substrate.

To test the extractable component effects on the hydrolysis, model studies were done using pure cellulose with added hydrophobic DCM extractives (unfractionated and fractionated, see section 9) or lignin from clean fractionation. To imitate the spread deposition of the components on the native wood fibers, the added extractables were deposited from acetone or MIBK-EtOH-water solutions. Alternatively the extractables were dispersed in the aqueous buffer solutions.

The results obtained from the performed experiments were somewhat surprising, as the there was no evidence of significant inhibitory effects from hydrophobic extractives or low molecular weight lignin. Actually, positive yield enhancing effects of nearly all of the added components were observed when deposition was made from organic solvent under cold temperatures (figure 46). Deposition from ambient temperature solution or dispersion in aqueous buffer resulted no positive effects (figures 47 & 48). Part of the positive effects were caused by deposition treatments (figure 10.4), especially with the aid of refluxing or sonication, but also true additive originated effects were obvious

The dependence of the positive additive effects observed for the extractives under the enumerated deposition conditions in cold temperature and organic solvent unfortunately seems to rule out viable large scale application, unless the conditions for water deposition are found. Yet there is significance of these results for the overall process since it could be confirmed that the extractable compounds do not inhibit enzyme activity at the examined concentrations and won’t need to be removed from the pretreated wood for optimal sugar yields. The inhibitory effects towards the fermenting yeast that converts the sugars to ethanol may still become an issue, but the fermentation step of the biorefinery process was not within the scope of this project.

Figure 46 Deposition of fractionated MIBK lignin on ball-milled CF1 cellulose from acetone solution at cold temperature, and resulting glucose release after 72 h hydrolysis.

Figure 47 Deposition of unfractionated DCM extractives on CF1 cellulose as dispersion in aqueous buffer, and resulting glucose release after 72 h hydrolysis.

Figure 48 Deposition of unfractionated DCM extractives on bleached pulp from acetone solution at ambient temperature, and resulting glucose release after 72 h hydrolysis

Figure 49 Deposition of clean fractionation lignin on CF1 cellulose from MIBK-EtOH-water solution, using various dispersion methods before deposition at cold temperature, and resulting glucose release after 72 h hydrolysis.

REFERENCES AND COMPARING TO PRIOR ART

References

Ardica, S., Calderaro, E. & Cappadona, C.1985, "Radiation pretreatments of cellulose materials for the enhancement of enzymatic hydrolysis—II. Wood chips, paper, grain straw, hay, kapok", Radiation Physics and Chemistry (1977), vol.26, no.6, pp.701-704.

Asikkala, J., Tamminen, T. & Argyropoulos, D.S.2012, "Accurate and Reproducible Determination of Lignin Molar Mass by Acetobromination", Journal of Agricultural and Food Chemistry, vol.60, no.

36, pp.8968-8973.

Boussaid, A., Robinson J., Cai, Y-J., Gregg, D. J., Saddler, J. N.1999, "Fermentability of the Hemicellulose

–Derived Sugars from Steam-Exploded Softwood (Douglas Fir)", Biotechnology and Bioengineering, vol.64, no.3

Bozell, J. J.; Black, S. K.; Myers, M.1995, “Clean Fractionation of Lignocellulosics: A New Process for Preparation of Alternative Feedstocks for the Chemical Industry”, Proceedings of the 8<th>International Symposium on Wood and Pulping Chemistry, 6-9 June 1995, Helsinki, Finland, pp.

697-704

Clark, T.A. & Mackie, K.L.1987, "Steam Explosion of the Softwood Pinus Radiata with Sulphur Dioxide Addition. I. Process Optimisation", Journal of Wood Chemistry and Technology, vol.7, no.3, pp.

373-403.

Hamelinck, C.N., Hooijdonk, G.v. & Faaij, A.P.2005, "Ethanol from lignocellulosic biomass: technoeconomic performance in short-, middle- and long-term", Biomass and Bioenergy, vol.28, no.4, pp.384-410.

Iller, E., Kukiełka, A., Stupińska, H. & Mikołajczyk, W.2002, "Electron-beam stimulation of the reactivity of cellulose pulps for production of derivatives", Radiation Physics and Chemistry, vol.

63, no.3–6, pp.253-257.

Kamm, B., Gruber, P.R. & Kamm, M.2000, "Biorefineries ? Industrial Processes and Products" in Ullmann's Encyclopedia of Industrial Chemistry Wiley-VCH Verlag GmbH & Co. KGaA, .

Karthika, K., Arun, A.B. & Rekha, P.D.2012, "Enzymatic hydrolysis and characterization of lignocellulosic biomass exposed to electron beam irradiation", Carbohydrate Polymers, vol.90, no.2, pp.1038-1045.

Khan, A.W., Labrie, J.P. & McKeown, J.1986, "Effect of electron-beam irradiation pretreatment on the enzymatic hydrolysis of softwood", Biotechnology and Bioengineering, vol.28, no.9, pp.1449-1453.

Mackie, K.L., Brownell, H.H., West, K.L. & Saddler, J.N.1985, "Effect of Sulphur Dioxide and Sulphuric Acid on Steam Explosion of Aspenwood", Journal of Wood Chemistry and Technology, vol.5, no.

3, pp.405-425.

Martinez, A., Rodriguez, M.E., York, S.W., Preston, J.F. & Ingram, L.O.2000, "Use of UV Absorbance To Monitor Furans in Dilute Acid Hydrolysates of Biomass", Biotechnology progress, vol.16, no.4, pp.637-641.

Zhang, Y.-.P.2008, "Reviving the carbohydrate economy via multi-product lignocellulose biorefineries", Journal of industrial microbiology & biotechnology, vol.35, no.5, pp.367-375.

Zhu, J.Y. & Pan, X.J.2010, "Woody biomass pretreatment for cellulosic ethanol production: Technology and energy consumption evaluation", Bioresource technology, vol.101, no.13, pp.4992-5002.

Ximenes, E., Kim, Y., Mosier, N., Dien, B., Ladisch, M.2010 “Inhibition of cellulases by phenols”, Enzyme and Microbial Technology, vol.46, pp.170–176

“Effect of Temperature and Retention Time of Steam Exploded Hardwood and Softwood” 243<rd>ACS National Meeting & Exposition, San Diego, USA, March 2012.

“Production of Wood-Based Fuels and Chemicals from a Steam Explosion-Solvent Extraction Biorefinery” Technoport 2012-Sharing Posibilities, Trondheim, Norway, April 2012.

Leskinen, T.; Kelley, S.; Argyropoulos, D. S.; “Development of a facile methodology for the molecular weight determination of pretreated lignocellulosic materials”, 64th Southeast Regional Meeting of the American Chemical Society, Raleigh, NC, United States, November 14-17, 2012.

“Simulation of Ethanol Production Processes Based on Steam Explosion-Solvent Extraction Wood” 7<th>Annual NC State Graduate Student Research Symposium, Raleigh, USA, March, 2013.

“Combination of Steam Explosion and Organic Solvent Extraction for Improving Enzymatic Hydrolysis of Birch and Pine” 244<th>ACS National Meeting & Exposition, New Orleans, USA, April, 2013.

Patents

EB and SE has been part of many patents. The following list has been investigated. Some of the patents are variations of other patents. For several of the patents, the investigations have been limited to a translated list of claims.

Publication number Priority Date Title

US20100297705A1 2010-11-25 Processing biomass

CN102459562A 2012-05-16 Processing biomass

US20120107920A1 2012-05-03 Method of producing sugar solution and saccharification device US7932065B2 2011-04-26 Processing biomass

US20100124583A1 2010-05-20 Processing biomass

US20100087687A1 2010-04-08 Processing biomass

US7682811B2 2010-03-23 Systems and methods for producing biofuels and related materials

US20090035826A1 2009-02-05 Method for the production of alcohol from a pretreated lignocellulosic feedstock

US20090286295A1 2009-11-19 Processing biomass

US20110111456A1 2011-05-12 Processing biomass

US20110287498A1 2011-11-24 Processing biomass

US20120100577A1 2012-04-26 Processing biomass

US20100200806A1 2010-08-12 Saccharifying biomass

JP2008535664A 2008-09-04 Biomass processing in order to obtain the fermentable sugars US20090017503A1 2009-01-15 Method and Apparatus for Saccharide Precipitation From Pretreated Lignocellulosic Materials

US20100203495A1 2010-08-12 Processing biomass

EP2006354A2 2008-12-24 Novel method for production liquid fuel from biomass US20120315675A1 2012-12-13 Processing biomass

JP2006081483A 2006-03-30 Biomass ethanol using waste mushroom bed of mushroom as raw material

US20120315683A1 2012-12-13 Ethanol production from lignocellulosic biomass with recovery of combustible fuel materials

WO2011159915A1 2011-12-22 Fermentation of biomass

CN102216435A 2011-10-12 Processing biomass

US20130143290A1 2013-06-06 System for the treatment of biomass to facilitate the production of ethanol

US20140011248A1 2014-01-09 Conversion of biomass

JP2011139686A 2011-07-21 Continuous method for producing ethanol by parallel saccharifying fermentation reaction

US20140011258A1 2014-01-09 Processing biomass

US20130052682A1 2013-02-28 Processing biomass

US20150342224A1 2015-12-03 Processing biomas

The most active inventor has been Marshall Medoff. Many of his patents include radiation in the range of 10 Mrad to 50 or 200 Mrad. We seek patent for low dose radiation with an upper limit of 100 kGy, which is equal to 10 Mrad.

The main conclusion from this evaluation of the mentioned process with low dose radiation followed by steam explosion is that it is patentable.

As said, the combination of radiation and steam explosion is already patented. However, the combination of low dose radiation followed by steam explosion, are not found in any of the screened patents. Accordingly, it is an invention.

Claims (11)

CLAIMS What is claimed is:

1. A method comprising converting a treated biomass feedstock to a product selected from the group consisting of alcohols, sugars, hydrocarbons, and mixture thereof, utilizing an enzyme and/or microorganism, wherein the treated biomass feedstock has been prepared by irradiating a biomass feedstock with an electron beam at a dose up to 100 kGy (10 Mrad) of radiation, and thereafter treating the same biomass feedstock with steam explosion with temperature about or less than 220⁰ C so as to render the said biomass feedstock more susceptible to chemical, enzymatic and/or microorganism attack than the biomass feedstock prior to electron beam treatment, the biomass feedstock having been physically treated prior to electron beam irradiation by cutting, grinding and milling to make the particles homogenous.

2. The method of claim 1, wherein the sequence change to steam explosion before radiation by electron beam.

3. The method of claim 1, wherein the particles are in-homogenous.

4. The method of claim 1, wherein the biomass feedstock comprises a cellulosic or lignocellulosic material.

5. A method comprising irradiating a biomass feedstock with an electron beam at a dose up to 100 kGy of radiation, and thereafter treating the said biomass feedstock with steam explosion with temperature about or less than 220⁰ C so as to render the said biomass feedstock more susceptible to chemical, enzymatic and/or microorganism attack than the biomass feedstock prior to electron beam treatment, the biomass feedstock having been physically treated prior to electron beam irradiation by cutting, grinding and milling to make the particles homogenous; and converting the said biomass to a product selected from the group consisting of alcohols, sugars, hydrocarbons, and mixture thereof , utilizing an enzyme and/or microorganism.

6. The method of claim 5, wherein the sequence change to steam explosion before radiation by electron beam.

7. The method of claim 5, wherein the particles are in-homogenous.

8. The method of claim 5, wherein the biomass feedstock comprises of cellulosic or lignocellulosic material.

9. The method of claim1 and claim 5 where the biomass feedback is treated with SO2prior to steam explosion.

10. The method of any one of the above claims, wherein the biomass feedstock is selected from the group consisting of wood, wood waste, particle board, sawdust, paper, paper products, paper waste, agricultural waste, sewage, silage, grasses, rice hulls, bagasse, cotton, jute, hemp, flax, bamboo, sisal, abaca, straw, com cobs, com stover, switchgrass, alfalfa, hay, rice hulls, coconut hair, cotton, synthetic celluloses, seaweed, algae, and mixtures thereof.

11. The method of any one of the above claims, wherein the biomass feedstock after pretreatment will be used for other purposes than described in the claims above, for example food or animal feed.