              Feedstock Considerations and Impacts on Biorefining
                                 Andy Aden, PE
                                     NREL
                                 Dec. 10, 2009
                                       
Introduction
Biomass is ubiquitous.  It can be found in nearly every part of the globe except from the harshest of climates.  This makes it an important renewable resource of interest to a large number of stakeholders internationally.  However, not all biomass is created equal.  There are a wide variety of lignocellulosic materials that could potentially be utilized for biorefining into liquid fuels and other products.  
Biomass can be grouped and classified in many ways.  Near-term biomass feedstocks can be byproducts from existing processes, such as the stalks, leaves and other plant matter left over after harvesting (i.e. corn stover, wheat straw, etc).  Forest residues from thinning operations or from existing lumber processing are also a potential feedstock, as is bagasse, the plant fiber left after extracting sugar from sugar cane.  Dedicated biomass crops have also been envisioned for biorefining purposes.  These would include fast-growing tree plantations (poplar), energy cane, and grasses (switchgrass, miscanthus).  
Significant variability exists for all biomass feedstocks of interest to refiners.  Even within a single type of biomass (i.e. corn stover), significant compositional variation has been seen.  The reasons for this variation are due to a wide variety of genetic and environmental factors, including soil conditions, precipitation, harvesting time, harvesting method, and plant genetics.  This variation in composition can have a very large impact in processability and in the overall economics of biorefining.  Therefore, the objective of this report is to outline the key biomass feedstock considerations and the potential impact they may have on overall yield and processability within the biorefinery.  Much of this information comes from decades of study of the structure and composition of biomass resources, as well as hands-on pilot plant experience with conversion of these materials to liquid fuels and other products at NREL.
Background
Biomass feedstocks considered here represent a mixture of those identified as being of potentially high-volume in the United States.  These include wheat straw, corn stover, forest residues and thinnings, switchgrass, and plantation trees such as poplar.  Other feedstocks will be discussed but will not be primary foci, including bagasse, sorghum, energy cane, rice straw, and municipal solid waste (MSW).  
While there are many sources of biomass, there are also many envisioned methods of converting that biomass to fuels.  To focus the discussion, three representative conversion technologies will be considered here:  one biochemical (using enzymes and fermentative organisms) and two thermochemical (gasification coupled with mixed alcohol and/or Fischer-Tropsch synthesis).  These are well documented in two NREL design reports[.].  The Fischer-Tropsch (FT) design was documented in an earlier report to EPA from NREL.  Other processes will not be discussed here, including Coskata's syngas fermentation, Bluefire Ethanol's concentrated acid process, and certain pyrolysis processes.
Many methods of feedstock compositional analysis have been developed over the years.  Quantifying the components of biomass and biomass intermediates remains an active area of research, aimed at producing more accurate and faster methods of analysis.  When considering a specific biomass resource for refining, it is always best to conduct this analysis on representative materials because of the large variability described above.  However, several collections of existing compositional data exist[,].
MSW can be considered a potential feedstock for biomass conversion processes, both biochemical and thermochemical.  In either process, the organic fraction of MSW is targeted.  However, there are many challenges and considerations associated with this particular feedstock.  These largely relate to the potential quantity and quality of biomass from MSW.  Maintaining a consistent quality MSW feedstock is extremely difficult.  MSW is also not easily defined.  In many parts of the country, "trash" is not segregated in any fashion and can be extremely variable in contents.  It would have to be pre-sorted to better isolate a biomass fraction and this would add cost.  However, in certain parts of the country, this pre-sorting infrastructure is in place, such as California.  Sometimes called a MRF or MERF (Materials Recovery Facility), this facility separates co-mingled trash into separate fractions of aluminum, paper, cardboard, etc.  Each of these can then be sold to recycle markets overseas and domestic.  Sometimes, curbside segregation can result in materials such as "urban greenwaste" (food/wood/yard trimmings) that could more easily be utilized as a biomass feedsource.  However this is not widely practiced in the US which makes it difficult to visual urban greenwaste as a high-impact biomass feedstock.  
It's also important to keep in mind that current markets do exist for these MSW materials.  Cardboard and paper can be sold overseas for over $100/ton, which makes using such materials for biofuels cost prohibitive.  Urban greenwaste is utilized for commercial mulch and organic composting.  In the past, several biomass companies have developed business plans that require negative cost feedstocks.  In other words, they count on tipping fees remaining in place for companies to dispose of waste.  In essence the biomass company gets paid to take the feedstock instead of paying for feedstock.  However, the problem with this strategy is that the value of these materials could quickly rise with significant demand present, which could cause feedstock costs to rise.  
Discussion
The general topics of comparison for the various feedstocks include:
   * Type of analysis conducted (carbohydrate vs. ultimate)
   * Energy content (heating value of biomass, HHV, LHV)
   * Moisture, water content
   * Physical structure of biomass
   * Cellulose
   * Hemicellulose
   * Lignin
   * Acetyl groups
   * Other organics
   * Ash and other inorganics
   Type of analysis conducted.  The type of analysis conducted is largely determined by the process under consideration.  For a biochemical process, a carbohydrate analysis is more informative because of the known stoichiometric reactions within a biochemical process for carbohydrate conversion.  This is typically accomplished using published laboratory analytical procedures and equipment such as high performance liquid chromatography (HPLC).  In a biochemical process, only the carbohydrate fractions of the biomass are converted to ethanol.  The remaining fractions (lignin, ash, etc) are either converted to heat and power, or remain relatively unconverted.  For a thermochemical process, an elemental (also known as "ultimate") analysis is more informative.  This type of analysis has been used to analyze other solid resources such as coal for many years.  It simply quantifies how much of the organic elements are present within a sample (Carbon, Hydrogen, Oxygen, Nitrogen, Sulfur, Chlorine, etc).  Other inorganic compounds are found in the Ash fraction, which can further be broken down into fractions of potassium, sodium, etc.  The C,H, and O in particular represent the potential available materials for synthesizing mixed alcohols downstream.
   Energy content.  The energy content of biomass is often expressed in terms of a higher or lower heating value.  In thermodynamic literature, these are often referred to as a heat of combustion; the heat released from the biomass when it's combusted and the chemical bonds are broken.  The higher heating value (HHV) assumes that all the water is in the liquid phase at the end of combustion, while the LHV assumes that the water is in the vapor phase.  In this fashion, the HHV includes the heat that can be obtained by condensing the water vapor produced by combustion.  Biomass feedstocks with higher energy contents have more energy available from which to create liquid fuels.  Woods, for example, often have higher heating values compared to grasses like stover.  It is important to calculate heating values on a moisture-free or dry basis because the presence of water in a biomass feedstock will reduce the gross heating value.
   Moisture, water content.  The presence of moisture in a feedstock can have a significant impact on the biorefining process, especially if it is a thermochemical process.  Processing feedstocks with higher moisture content in a thermochemical process will produce lower overall efficiencies than drier feedstocks.  This is due to the energy lost in vaporizing this moisture.  Wetter feedstocks will also have lower efficiencies in the gasification step as well.  In a biochemical process, the moisture content of the biomass has less of an impact on the overall efficiency.  Because these processes are mostly aqueous-phase systems to begin with, water is often added in order to slurry the biomass and make it more "flowable".  In the design report process as designed, water is added until the biomass slurry entering pretreatment is approximately 30% total solids, or approximately 70% moisture.  Therefore, starting with a feedstock that is 15% moisture will require the addition of more water than a feedstock that is 50% moisture.  Technoeconomic analysis has shown that maximizing the solids concentrations through the biochemical processes helps to reduce cost because of economies of scale.  However, this is limited by a pretreatment's ability to disrupt the biomass matrix sufficiently.  Certain pretreatment reactors and feeding systems can also have limitations on the solids concentration they can utilize.  Woods are expected to arrive at a biorefinery with moisture concentrations near 50%, while feedstocks that have been field dried are expected to be closer to 15% moisture.  The moisture content can be affected by time of biomass collection, and other processing steps such as pelletization, drying, torrefaction, etc.
   Physical structure of biomass.  Biomass can be analyzed and classified at many different levels.  For example, biomass often contains different anatomical fractions that can be visualized without the aid of any microscopic tools.  Fractions of a corn plant, for example, include stalks, husks, cobs, nodes, etc.  As one dives deeper into the biomass, different classification systems come in to play.  Within the anatomical fractions, different cell types exist.  Ultimately, the biomass polymers of most interest (cellulose, hemicelluloses, and lignin) are found within the plant's cell walls.  
   Nature has made these biomass materials resistant to degradation so as to guard against environmental elements and biological attack.  The three most common compounds are cellulose, hemicellulose, and lignin.  These compounds are interwoven physically with each other and also chemically bonded to one another.  Cellulose within plant cell walls comes in the form of microfibrils.  Hemicellulose is a general group of other polysaccharides and organic acids that form a sheath around the cellulose.  They also form covalently bonds and ester linkages with the lignin present.  Different biomass resources will contain different versions of how these compounds are physically arranged.  These differences can lead to varying degrees of "convertability".  For example, wheat straw has been shown to be more amenable to biochemical conversion (at lower severity conditions) than corn stover.  This is part of the reason why it's a targeted feedstock.  Corn stover and switchgrass are less amenable to conversion and require more severe conditions to effectively pretreat.  Feedstocks that have already been pre-processed, such as bagasse or corn fiber, can also be more amenable to conversion because part of the matrix has already been disrupted through the initial processing steps.  
   Cellulose.  Cellulose is a long polymer chain of glucose monomer units.  It is very similar in this fashion to starch.  The difference between starch and cellulose is how the glucose units are bonded.  Starch is bonded through alpha linkages, which are much easier to break than the beta linkages found in cellulose.  In this fashion, nature uses these compounds differently, using starch as an energy storage system, and cellulose as a structural function.  Together, cellulose and hemicellulose make up the total potential fraction of the biomass that can be directly converted to ethanol and other biofuels in a biochemical process.  Large variability in cellulose concentration is seen in biomass feedstocks.  Some niche biomass feedstocks have been seen to have almost 75% cellulose on a dry-weight basis, however this is not typical.  In general, cellulose concentrations for most feedstocks range between 30% and 45%.  Woods (both hardwood and softwood) tend to have higher cellulose concentrations than grasses and agricultural residues and therefore, have higher theoretical yield potentials of biofuel.  
   Hemicellulose.  Hemicellulose is a general definition that captures additional biomass carbohydrates (xylan, galactan, arabinan, mannan) and organic acids (uronic acid, ferulic acid).  Hemicellulose also contains the acetyl functional groups that become acetic acid during dilute acid pretreatment.  Hemicellulose is generally thought of as "easier to convert" than cellulose, which is why processes are often designed to use cheaper chemicals to hydrolyze the hemicellulose during pretreatment as opposed to expensive enzymes.  Hemicellulase enzyme systems are under development but are not as developed as the cellulase systems.  Commercial hemicellulase preps are currently used in pulping and other industries.  However, they will contain much different functional activities than cellulase preps, such as xylanase, ferulic-acid-esterase, etc.  In a biochemical process, if the hemicellulose is not hydrolyzed chemically, the burden becomes placed on the enzymatic system.  This is seen with some of the alternative pretreatments such as AFEX, for example.  Under this pretreatment, the ammonia helps to swell fibers and disrupt the biomass matrix, however, does very little to hydrolyze or otherwise chemically convert the hemicellulose.  Concentrations of hemicellulose can range dramatically between biomass feedstocks (typically 15-30% dry weight of biomass).  The relative concentrations of hemicellulosic sugars can also vary.  For example, softwoods typically contain higher concentrations of mannan than hardwoods or grasses.  
   Lignin.  Lignin is the substance within a cell wall that gives a plant structural rigidity.  It is a large molecular weight compound that contains aromatic rings and several functional groups (methoxy, etc.).  The empirical formula for lignin will vary between biomass feedstocks.  In other words, the lignin in hardwoods is different than lignin found in softwoods, which is different from lignin found in grasses and straws.  Lignin can be depolymerized through base-catalyzed reactions.  It can also be extracted from biomass using various organic solvents.  Several methods of quantifying lignin within biomass exist, including Klason lignin, and other such determinations.  In a thermochemical process, the lignin, hemicellulose, and cellulose are broken down to produce a mixture of gases (CO, H2, CO2, CH4, etc), tars (organics with molecular weight greater than or equal to benzene), solid char, and ash.  In a dilute acid biochemical process, the lignin carries through the process relatively unchanged.  Some of the lignin is solubilized if temperatures in pretreatment exceed the glass transition temperature of lignin (~135C).  However, the lignin will recondense into solid form once the temperature is reduced.  Lignin is known to inhibit enzymatic hydrolysis reactions however this interaction between cellulose, lignin, and enzymes is complex and the subject of much research.  Lignin is known to play a role in "non-specific binding" which means that the cellulase enzymes will bind to the lignin instead of the cellulose.  This causes enzyme to try to act on a substrate that it wasn't meant for, which equates ultimately to the need for higher enzyme loadings.  However, digestibility experiments have also shown that completely extracting the lignin does not necessarily create more digestible cellulose.  The theory here is that when lignin and hemicellulose are completely extracted, the cellulose microfibrils may hydrogen bond to one another and become even more recalcitrant than before.  But aside from that, feedstocks with less lignin are generally more amenable to enzymatic conversion, and feedstocks with higher lignin are more recalcitrant, requiring higher enzyme doses or have yields that are much lower.  In general, woods have higher lignin fractions than grasses and straws.  Lignin composition of some woody biomass has been seen in the high 20's percent on a dry weight basis.
   Acetyl groups.  The amount of acetylation of biomass depends on the specific feedstock.  The acetyl groups are part of the hemicelluloses and are easily cleaved during dilute acid hydrolysis.  In this low pH environment, they become acetic acid, which is a well-known fermentation inhibitor.  For some ammonia-based alternative pretreatments the acetyl groups convert to acetimide, which helps to reduce the amount of acetic acid a fermenting organism will see.  During gasification, the specific fate of these compounds are unknown, but they in some fashion become part of the syngas, char, and tars.  In general higher acetylation is seen with hardwoods than grasses or straws.  For biomass that is more highly acetylated (>4% dry wt), additional processing is required to reduce the acetic acid levels to a more manageable level (<10 g/L).  This can include membrane separation, ion exchange, etc.  This can add cost to a process.
   Other organics.  Other organic compounds that can be present in biomass include acids (uronic acids), proteins, "extractives", other sugars (fructose, sucrose), pectins, and other unidentified "soluble solids".  Methods for quantifying these compounds continue to undergo development.  The fate of many of these compounds within a biochemical process are largely unknown.  In process models these are largely assumed to be combusted with lignin residue to provide heat and power, or can end up in wastewater treatment.  Helping to identify the additional organic compounds present within these biomass fractions can help to close material balances to a greater degree.
   Ash and other inorganics.  Ash is comprised of the inorganic materials and minerals entering the process.  They can either be "structural inorganics" which are metals such as potassium, sodium, sulfur, etc. found in the biomass itself, or they can be non-structural inorganics.  The structural inorganics help provide certain functionalities within the living plant just as they do for human bodies.  Non-structural inorganics are sources of material that come from the environment when the biomass is collected.  For example, when corn stover is collected off of the field, a certain amount of soil and dirt will come with the stover.  When the biomass is analyzed, this will show up in the analysis as ash.  This is part of why grasses and straws typically have higher ash contents than woods.  Ash is effectively inert material.  Therefore, it is desirable to have low-ash biomass for biorefining because you don't want to pay a lot of money for biomass fractions that can't be converted into anything useful.  
   In a biochemical process, the ash is largely inert and will end up as dry ash coming from the boiler.  However, if ash fractions (potassium, etc) that go to wastewater treatment are too high, they can become problematic.  They can impact process equipment by eroding it if present in high enough concentration.  Rice straw is a primary example.  Rice straw is notorious for having high concentrations of silica.  This has been shown to erode process equipment at those concentrations.  This can increase processing costs and is unfavorable.  In a thermochemical process, the ash will come out of the gasifier as a solid dry ash for lower temperature systems.  In high temperature gasification systems, the ash will leave as a molten form called "slag".  This will cool into a glassy looking substance that has a number of potential uses, including being added to road base, cement and other structural foundation material.  
   Sulfur is one inorganic that is of particular concern to thermochemical processes.  The thermochemical mixed alcohol process as designed uses a nickel-based tar reforming catalyst to reform tars into additional syngas yield.  Sulfur is a known poison to nickel-based catalysts.  As such, experiments have shown catalyst to deactivate more quickly in the presence of sulfur and lose full reforming potential when regenerated.  This means that certain active sites have been poisoned irreversibly.  Woods typically have low sulfur levels, around 10 parts per million (ppm).  However agricultural residues and energy crops (corn stover, switchgrass) can have sulfur levels as high as 500 ppm, which can greatly impact the process.  In a gasifier's reducing atmosphere, sulfur will typically come in the form of H2S.  Sulfur can also impact fuel synthesis catalysts.  In this particular mixed alcohol process, a Cobalt-moly-sulfide based catalyst is used to convert syngas into mixed alcohols.  This particular catalyst actually requires the presence of small amounts of sulfur to remain sulfided.  However, alternative catalysts may run the risk of being poisoned by sulfur, which would require additional cost due to more stringent acid gas scrubbing.
   
   Summary
   Given all of the information described thus far, a qualitative comparison of process yield, efficiency, etc. for the chosen feedstocks was completed.  This is summarized in Table 1.  There are many considerations that can impact these parameters.  Many experimental lessons have been learned throughout the years for comparing feedstocks to one another.  Those learnings were incorporated into this comparison as much as possible.
   In a biochemical process, corn stover and switchgrass have shown to behave similarly.  They have similar composition, are anatomically similar, and therefore at this juncture would be expected to produce similar yields in such a process.  While wheat straw also maintains a similar composition to corn stover and switchgrass, experiments have shown it to be more amenable to conversion at lower severity.  Poplar and other hardwoods, however, are expected to perform differently in a biochemical process.  The total carbohydrate potential, and therefore the theoretical ethanol yield is larger due to these woods having more cellulose.  The heating value of the woods is also greater.  The difference in moisture in a biochemical process (15% vs. 50%) has little impact on the efficiency or yield.  Because of the much greater lignin content, the electric coproduct credit for a biochemical process is also anticipated to be greater for woods than for stover or switchgrass.  However, this increased lignin can potentially have negative impacts on certain sections of the process, particularly the enzymatic hydrolysis.  For the woods, the yields would likely be achieved through certain process improvements that help to overcome these negative impacts. These improvements may cost more money, but cost was not a factor in this particular analysis.  Another area for potentially increased cost comes from the higher levels of acetylation in hardwoods compared to stover and grasses.  Additional unit operations might be needed to reduce the acetic acid levels prior to fermentation to avoid inhibition of the cells.
   For a thermochemical process, woods make an excellent feedstock.  They have high heating values compared to stover and switchgrass, and they are low in sulfur (which can poison the tar reforming catalyst).  One of their drawbacks is their higher moisture content.  This tends to lower the overall efficiency of these processes, however, higher yields from woods are still observed compared to stover and switchgrass.  Forest residues have a very large compositional variability depending on what forests they come from (hardwood, softwood), how they are harvested, etc.  For the purposes of this study, they were assumed to be similar to poplar, however, detailed compositional analysis would be required to verify this for a specific feedstock or biorefinery project.  The yields for the FT process lower than the mixed alcohols process, but the overall energy efficiencies of both processes are approximately the same.
   For the data summarized in Table 1, a few cells were noted with "?".  These are the overall efficiencies for biochemical conversion of poplar for 2010, 2015, and 2022.  Researchers at NREL don't currently have a good understanding of how the overall efficiency might differ between woods and grasses (or residues) in a biochemical process.  As such, NREL's current recommendation is to assume that the efficiencies are similar for both feedstock classes until this is better understood through future research.  
   
   
   Table 1  -  Biomass feedstock comparison between biochemical and thermochemical processes

   
