Maximizing biorefinery sugar yields using corn stover pellets: impact of soaking in aqueous ammonia pretreatment temperature and high-solids loading

Increasing the solid loading rate of the substrate in pretreatment and enzymatic hydrolysis provides many attractive benefits for biorefinery production. However, by introducing a higher concentration of solids, yields are typically reduced to levels that offset the savings. Previous studies have shown that preprocessing methods such as pelleting can improve downstream conversion; however, the literature has not explored how the thermomechanical breakdown associated with pelleting may influence downstream reaction mechanisms or help recover the reductions typically observed in high solids systems. This research aimed to maximize yields by analyzing the synergy of pelleting, pretreatment temperature, reactor design, and feeding strategies on the high-solid soaking in aqueous ammonia pretreatment and hydrolysis of corn stover. Pelleting the feedstock allowed glucose yields above 90%, even when pretreatment loadings were increased to 40% (w/v). Temperature was found to be a critical parameter in pretreatment conversion as the severity impacted biomass structure and overall yields. In the subsequent high solid hydrolysis (30% w/v), a gravimetric reactor design and fed-batch feeding were used to improved solubilization. Final sugar concentrations above 140 g/L were achieved in 72 h. These results suggest that the downstream impact of pelleting in a high-solids system can make pelleting an innovative and cost-effective option for biorefineries.

The demand for carbon-based chemicals has rapidly increased in recent years due to a diverse market economy. Using lignocellulosic sources for these chemicals compared to their petroleum-based counterparts is an attractive alternative due to their sustainability and carbon neutrality (Gnansounou and Dauriat, 2010). However, the current biorefinery production techniques have significant technological and financial drawbacks that must be addressed before industrial-scale production can occur (Gnansounou and Dauriat, 2010). One specific strategy that asserts an attractive, economical solution to biorefinery viability is to increase the solid loading rate of the substrate (SL) within the pretreatment (PT) and enzymatic hydrolysis (EH) steps for sugar production. By increasing the initial biomass concentrations, higher sugar concentrations can be produced at faster rates. In addition, the capital and production cost requirements can be reduced as smaller reactors require less water and energy to meet the production goals (Roche et al., 2009; Modenbach and Nokes, 2012; Geng et al., 2015).

Ammonia-based pretreatment methods are particularly of interest for a high-solid model. One such method, soaking in aqueous ammonia, has potential due to its high lignin solubility, hemicellulose retention, efficient chemical recovery, and mild processing conditions (Hodge et al., 2008; Kim et al., 2016; Raj and Krishnan, 2019; Zhao et al., 2020). This method of pretreatment is especially promising for preprocessed feedstocks such as corn stover pellets, as preprocessing may enhance conversion in mild conditions (Nahar and Pryor, 2017). However, the research on understanding the synergy of pelleting and soaking in aqueous ammonia pretreatment is limited, regardless of the solid loadings.

Following most pretreatment methods, one critical aspect that hinders sugar production in enzymatic hydrolysis at high-solid loadings is the viscosity of the slurry. At concentrations above 15% (w/w), the particles’ shear force and lubricity are impacted greatly, leading to poor mixing and yields with conventional methods (Jørgensen et al., 2007; Kristensen et al., 2009). One approach that has been researched to circumvent this is to improve the mixing method. Gravimetric rolling is a particularly attractive mixing method as it requires significantly less energy compared to conventional impeller setups (Jørgensen et al., 2007; Roche et al., 2009).

Another approach that has been researched to reduce viscosity and enable uniform conversion in the hydrolysis slurry is to add solids and/or enzymes in a fed-batch system (Larsen et al., 2008; Kristensen et al., 2009; Roberts et al., 2011; Geng et al., 2015). In due course, by lowering the initial solid loading rate, the slurry gets time to liquefy, and any diffusion or mixing limitations can be effectively minimized or avoided (Modenbach and Nokes, 2012). The concern with this process is the additional time required to convert the substrate. Therefore, it is critical to identify the optimal loading rate of solids and analyze the shortest retention time required to produce maximum sugar yields.

Preprocessing techniques, such as pelleting, offer unique benefits for biorefinery production. The most direct advantage is the reduction in transportation costs. According to Tumuluru (2023), the bulk density of baled corn stover ranges from 110 to 200 kg/m 3, whereas pellets have a bulk density of approximately 575–675 kg/m 3. This substantial increase in density—about three to five times greater—makes pellets easier to transport to biorefineries. Pradhan et al. (2021) and Tumuluru et al. (2024) indicated that higher bulk density leads to lower transportation costs. A 2017 case study on feedstock supply system design published by the Idaho National Laboratory indicated that transporting pellets by truck costs about 50% less than transporting other forms of biomass, such as bales or ground biomass. Additionally, using pellets instead of ground biomass can help biorefineries address storage, feeding, and conversion challenges. Tumuluru (2016) found in her studies high moisture pelleting of woody biomass that when the benefits for both upstream and downstream processes using pellets are quantified in monetary terms, the costs associated with pelleting can often be offset by the savings achieved.

In addition to improving the transportation efficiencies, pelleting offers several other advantages, including enhanced handling and conveyance throughout the supply chain, better particle size distribution and density, and improved structural component fractionation due to the compression and extrusion of the biomass particles in the pellet die. Many researchers have indicated several key benefits of pelleting biomass (Tumuluru, 2016; Gong et al., 2021; Chen et al., 2025) a) improved digestibility, where high pressure and feedstock moisture help break apart cross linking structures such as lignin, making cellulose more accessible; b) reduced pretreatment severity, as densified biomass often allows biorefineries to lower pretreatment temperatures and residence times while still achieving higher sugar yields; and c) improved logistics, because pelleting significantly increases bulk density and enhances storage, handling, and feeding characteristics. Literature review has indicated that there is lack of data regarding high solid loadings of corn stover pellets to maximize sugar yields.

This study aimed to investigate the unexplored synergy of promising methods to convert raw and pelleted corn stover under high-solids loadings. The pretreatment parameters included in this study were 10%, 20%, and 30% (w/v) solid loadings and temperatures of 60 °C, 90 °C, and 120 °C for loose and pelleted stover. Compositional analysis, Fourier Transform Infrared (FTIR) absorption, and Scanning Electron Microscopic (SEM) imaging were collected to compare the structure of the constituents that occurred during the pelleting and pretreatment process. Following pretreatment analysis, 90 °C pretreated stover was subjected to a 72-h high-solid hydrolysis at 10% (w/v), 20% (w/v), and 30% (w/v) loadings using shake flasks and a gravimetric rolling system to determine the optimal mechanical mixing method. Using the gravimetric rolling system, the 90 °C pretreated stover was again subjected to high-solid hydrolysis at 10% (w/v), 20% (w/v), and 30% (w/v) loadings for 72 h using a fed-batch approach. In this process, solids and a combination of solids and enzymes were incrementally added to the hydrolysis slurry at various increments before 24 h.

Corn stover feedstock and pellet production

Corn stover bales were procured from Golden Harvest O6-N80 (Poweshiek County, Malcom City, Iowa) with a wide range of moisture contents (10%–33%, w. b.) and shipped to Idaho National Laboratory for further preprocessing studies. The low-moisture corn stover bales were further processed in a hammer mill and pilot-scale pellet mill to produce pellets. For this study, the corn stover bales of low moisture, about 10% (w.b.), were ground in a HG200 stage-1 grinder (Vermeer Corporation-Agriculture; Pella, Iowa, United States) fitted with a 3-in (76.2 mm) screen. The stage-1 ground corn stover was further size reduced in a E−4424-TF hammer mill (Bliss Industries; Ponca City, OK, United States) fitted with a ¼-in (6.35 mm) screen. Finally, the ground corn stover was stored in bulk containers and reconditioned to the desired moisture levels (14%, w.b.) for the pelleting tests in a 35A-75-80HP pilot-scale ring die pellet mill (Bliss Industries; Ponca City, Oklahoma, United States) fitted with a 6 mm die and die compression (L/D) ratio of 7 (Tumuluru, 2023). The corn stover pellets produced had a bulk density of 630 kg/m 3, unit density of 1210 kg/m 3, and durability of 98%–99% (Tumuluru, 2023). Tumuluru (2023) described in detail the methods that were used to measure the physical properties such as unit, and bulk density and durability of the corn stover pellets. The author used ASABE standards for measuring these properties (Tumuluru et al., 2010; Thoreson et al., 2012). These pellets and milled corn stover were shipped to North Dakota State University for further pretreatment and enzymatic hydrolysis testing.

Biomass moisture content determination

The moisture content of the ground and pelleted corn stover was determined using a FD115-UL convection oven (Binder; Germany) and the National Renewable Energy Laboratory (NREL) protocol (Sluiter et al., 2016).

Pretreatment

Corn stover was pretreated by soaking each sample in AX1303 aqueous ammonia (Milipore Sigma; Burlington, MA, United States). Pretreatment was conducted in a 1L, series 4600 Parr pressure vessel (Parr Instrument Company; Moline, IL, United States). The reactor was heated externally with heating coils in an insulated unit and a flat pressure gasket on the reactor was secured with compression bolts to prevent venting of ammonia from the system at higher temperatures. For each study, 100 g (dry weight) of the loose or pelleted material was added to the reactor, followed by the proper volume of ammonia to achieve the target solid loadings outlined in Section 2.8.1. Each batch was stirred initially, and there was no additional stirring or agitation during pretreatment to prevent undesired differences in heating times or slurry distribution. The temperature during the pretreatment process was monitored using the thermometer on the reactor, with the retention start time recorded as the time at which the slurry reached the target temperature. The temperature ramp of the slurry from ambient for each test case took between 10–20 min. Concerning previously optimized low-severity experimentation results, the pretreatment retention time was constant at 4 h, and the ammonia concentration was 18% (w/w) (Nahar and Pryor, 2017; Pandey et al., 2019).

Following the pretreatment, the biomass was washed with 7–8 L of DI water and centrifuged at 2500 g in 500 mL containers before vacuum filtration through Whatman #41 (20–25 um pore size) filter paper. After the pretreated material was washed, a small portion (∼25 g wet basis) was separated to air dry, grind, and sieve for compositional, SEM, and FTIR analysis. The remaining substrate was separated into large trays and left to air dry to an approximate moisture content of 40% (wet basis) to allow for sufficient citrate buffer addition in a subsequent high-solid enzymatic hydrolysis. The dried samples were then tested for moisture and ash content in triplicate, transferred into large Ziploc bags, and stored at 4 °C for subsequent hydrolysis experiments. Moisture content was used to calculate the solid remaining after pretreatment.

Enzymes

Each set of pretreated material was subjected to enzymatic hydrolysis with a combination of different enzymes. The three different types of enzymes NS50013 (cellulase), Cellic HTec (xylanase), and Novozyme 188 (β-glucosidase) were collected from Novozymes (Franklinton, NC, United States). Cellulase was added to hydrolyze cellulose, xylanase was added to hydrolyze xylose, and β-glucosidase was added to hydrolyze cellobiose, which was an inhibitory end-product of cellulose hydrolysis.

Before conducting enzymatic hydrolysis, the activity of each enzyme solution was calculated. The cellulase activity of NS50013 and the β-glucosidase activity of Novozyme 188 were found to be 60 filter paper units (FPU)/mL and 476 cellobiose units (CBU)/mL, respectively, as determined by (Ghose, 1987). Using the procedure outlined by Bailey et al. (1992), the xylanase activity of Cellic HTec was determined to be 10,600 xylanase units (XU)/mL.

Enzymatic hydrolysis

Enzymatic hydrolysis of the pretreated solids was conducted in triplicate to determine glucose and xylose yields (% theoretical) in each experiment. Based on the target solid loadings outlined in Sections 2.8.1, 2.8.2.1, 2.8.2.2, each vessel was loaded with pretreated biomass and then mixed with 50 mM sodium citrate buffer (pH 4.8) and 0.04% sodium azide to prevent microbial contamination. This buffer mixture was selected as previous studies (Nahar and Pryor, 2017; Nahar and Pryor, 2014) showed this concentration of buffer and pH was suitable for the enzymes used in this study. Mixing rates varied and are listed in Section 2.8.2.1. Before mixing, cellulase, β-glucosidase, and hemicellulase enzymes were also added at 25 FPU/glucan, 25 CPU/glucan, and 500 XU/glucan loading rates to each mixture. These enzyme loadings were based on our previous published work (Pryor et al., 2012; Pandey et al., 2019). All samples were incubated at 50 °C for 72 h. 1 mL aliquots were taken from the liquid portion of the hydrolysis assay in 24 h intervals and when solids/enzymes were added in fed-batch sequences. Aliquots were centrifuged at 13,226 × g for 5 min (Galaxy 16 Micro-centrifuge, VWR International; Bristol, CT, United States). The supernatant was filtered through a 0.2-μm nylon filter (Pall Corporation; West Chester, PA, United States), divided into 20 μL samples, and stored in a freezer at −20 °C until sugar analysis. Sugar contents were analyzed via HPLC, and saccharification yield was calculated as:

Note: Hydration factors for glucose and xylose were 0.9 and 0.88, respectively.

High-performance liquid chromatography (HPLC) analysis

Enzymatic hydrolysis samples were analyzed using an HPLC (Waters Corporation; Milford, MA, United States) to determine the concentration of glucose, xylose, and cellobiose. A Bio-Rad Aminex HPX-87P column (Bio-Rad Laboratories; Hercules, CA, United States) and a refractive index (RI) detector (Model 2414, Waters Corporation; Milford, MA, United States) were used for sugar identification using HPLC analysis. Each 20 μL hydrolysis sample was subjected to a mobile phase of 18-mΩ nanopure water at a 0.6 mL/min flow rate. A 4-point standard curve of glucose, xylose, and cellobiose was generated for each run using standards of known concentration. These standard curves were used to quantify the unknown sugar concentrations in each hydrolysis sample.

Characterization of nontreated and pretreated solids

Chemical composition

Compositional analysis was conducted in triplicate using standard NREL protocol (Sluiter et al., 2016). Compositional analysis was conducted before and after the pretreatment to determine the original and pretreated material’s glucan, xylan, and lignin content. This was done to make precise conversion calculations in subsequent hydrolysis experiments.

Physical characterization

Nontreated, pretreated, loose, and pelleted corn stover were air-dried for 5 days and then ground using a kitchen coffee grinder. To assess the chemical and physical impact of the pretreatment, ground samples were sieved using less than 180 μm and collected for further characterization, such as FTIR and SEM microscopy.

Fourier Transform Infrared (FTIR) analysis

FTIR spectroscopic analysis was carried out to identify changes in the functional groups induced by the pretreatment concerning pelleting, solid loading, and temperature. This analysis was conducted on the ground loose and pelleted corn stover samples. A Nicolet 6700 Fourier Transform Infrared Spectrometer (ThermoFisher Scientific; Madison, WI, United States) with a universal Attenuated Total Reflectance (ATR) was used to measure the infrared spectrum. For each sample, a spectrum from 4000 to 400 cm −1 was collected at 2 cm −1 resolution.

Scanning electron microscopy (SEM) analysis

Physical changes resulting from the pretreatment and the pelleting were observed for all 90 °C and 120 °C samples using SEM analysis. Images were taken at various magnifications ranging from ×500 to 7,500X using a JEOL JSM-6490 LV scanning electron microscope (JEOL United States, Inc.; Peabody, MA, United States) located at NDSU’s Electron Microscopy Center in Fargo, ND.

Experimental design and statistical analysis

Testing solid loadings for pretreatment

Soaking in aqueous ammonia pretreatment of loose and pelleted corn stover was conducted in triplicate. Solid loadings and temperatures were varied as shown in Table 1 using a factorial approach. Loose corn stover was pretreated at only 20% (w/v) solid loadings, as the material burned significantly at higher temperatures and solid loadings, leading to poor pretreatment results.

Glucose and xylose conversion were measured in a low solid enzymatic hydrolysis (1% glucan w/v) using the pretreated material to determine the pretreatment efficacy. Each hydrolysis sample was tested in triplicate.

Testing solid loadings for enzymatic hydrolysis

Reactor designs

The effect of laboratory-scale reactor design in high-solid hydrolysis was investigated by comparing three setups with variable loadings of 10%, 20%, and 30% (w/v). At 90 °C, 20% (w/v) pretreated pelleted corn stover was utilized for this experiment as conditions provided maximum sugar yields from the first experiment.

Firstly, a standard shake flask system was used in a MaxQ 7000 agitating water bath shaker (Thermo Scientific; Dubuque, IA, United States) set at 200 rpm. 13 mm diameter stainless steel mixing balls (ten) were added to each 150 mL vessel to facilitate material solubilization. Total slurry volume was held consistent at 50 mL, leading to a mixing ball to slurry ratio of approximately 1:4. Next, a gravimetric rolling system was used to mix the material in a large 550D Isotemp incubator (Thermo Fisher Scientific; Waltham, MA, United States). Round 100 mL and 200 mL Pyrex bottles were rolled at 80 rpm using a laboratory scale bottle roller (Thermo Fisher Scientific; Waltham, MA, United States). Cap inserts were 3-D printed and attached to prevent the material from sticking in the inlet during the rolling process. In one setup, 13 mm stainless steel mixing balls (ten) were added to each vessel to facilitate material solubilization. In the other setup, stainless steel inserts were attached inside half of the bottles along with the 13 mm stainless-steel mixing balls (ten) to try and enhance the tumbling of the material.

Feeding strategies

Three different feeding strategies were applied to the hydrolysis setup: i) a batch method (B) with all solids and enzymes added initially, ii) a fed-batch system with the incremental addition of solids (S), and iii) a fed-batch system with the incremental addition of both solids and enzymes (S.E). For the fed-batch methods, the material was added before 24 hours, so conversion would not be significantly reduced in 72-hour hydrolysis. Additions were made in 5% (w/v) solid intervals at these times so the material could rapidly solubilize. The specific addition times for each solid loading are summarized in Table 2.

This setup used two sets of pretreated materials: 90 °C samples at 20% (w/v) and 40% (w/v) pretreatment solids. The 20% (w/v) pretreated material was selected as it resulted in the highest glucose and xylose yields during the first experiment, and the 40% (w/v) pretreated material was selected as it was the highest solid content tested and produced high glucose and xylose yields in low glucan hydrolysis.

Statistical analysis

Data obtained from HPLC analysis was used to determine glucose and xylose yield percentages for each sample. Statistical analysis of the yield results was performed using SAS® (SAS Institute Inc.; Cary, NC, United States) to verify if the results were significantly different (p < 0.05) from each other. In the pretreatment experiment, a multiple mean comparison test was used to determine if there were any interactions between biomass form (loose and pellets), pretreatment temperature, and solid loadings. In the reactor design experiment, a multiple mean comparison test was used to determine if the designs produced significantly different results under different solid loadings in enzymatic hydrolysis. In the feeding strategy experiment, a multiple mean comparison test was used to determine if the feeding strategies produced significantly different results under different solid loadings in enzymatic hydrolysis.

Effects of soaking in aqueous ammonia pretreatment parameters on corn stover saccharification

Effects of pretreatment parameters on material composition

The composition of nontreated and ammonia pretreated loose corn stover (CS) and pelleted corn stover (PS) are shown in Figures 1a,b. Figure 1a displays the material composition of both nontreated and pretreated samples. Figure 1b displays the calculated solid remaining and delignification values for each sample following pretreatment. Results showed that pelleting slightly reduced the xylan and lignin content of the corn stover but did not have a statistically significant impact on the glucan content. These results were consistent with previous research that suggested pelleting before pretreatment slightly reduces the hemicellulose content but did significantly change the cell wall composition of the biomass (Kumar et al., 2012; Nahar and Pryor, 2014).

Results also revealed that soaking in aqueous ammonia pretreatment increased the available glucan and xylan content while removing a statistically significant amount of lignin in most cases. This suggests that lignin was effectively removed while retaining the hexose and pentose sugars, as confirmed in previous research (Kim et al., 2011; Prabakar et al., 2018; Raj and Krishnan, 2019). Comparing the different solid loadings and temperatures showed that as temperature increased, the delignification of loose stover decreased significantly. This decrease is attributed to the material burning on the sides of the reactor, which led to poor pretreatment results at higher temperatures. Based on these poor results, only the loose stover pretreated at 60 °C was used in the next experiment. Compared to the loose stover, the delignification of the pelleted stover was similar between the different solid loadings and was the highest following the 90 °C pretreatment. This delignification data suggested that higher severity pretreatment was more effective for the pelleted corn stover.

One unexpected result was the low delignification of the pelleted stover after 120 °C pretreatment. While previous research suggested increasing temperature would result in higher yields, the statistically significant drop in delignification during this study between 90 °C and 120 °C pretreatments suggests a significant drop in pretreatment efficacy. This finding is discussed in detail in Section 3.1.3.2.

Ammonia recovery and recycling are critical factors for the commercial deployment of ammonia-based pretreatment technologies. In this study, pretreatment was conducted in a sealed batch reactor, where the ammonia concentration (18% w/w) remained constant throughout the process. Because the objective was to isolate the effects of pretreatment conditions on biomass deconstruction, we did not perform mass-balance measurements of ammonia loss, recovery, or recycling. Previous studies (Balan et al., 2025; Campbell et al., 2013) have demonstrated that ammonia can be efficiently recovered (>95%) through depressurization and condensation in commercial AFEX configurations. Future work should incorporate ammonia recovery analysis to enable a complete techno-economic and environmental assessment of the SAA process.

Effects of pretreatment parameters on saccharification efficiency

Enzymatic hydrolysis of loose corn stover and pelleted corn stover without any pretreatment was carried out to see the effect of pelleting alone on hydrolysis yields. Glucose and xylose yields (% of theoretical) for pelleted corn stover increased from 21% to 41% and from 9% to 25%, respectively, compared to loose corn stover without any pretreatment (Figures 2a,b). The results are consistent with prior work showing increased hydrolysis yields from pelleted biomass without pretreatment, including pelleted switchgrass (Rijal et al., 2012), and pelleted corn stover (Nahar and Pryor, 2017; Nahar and Pryor, 2014).

Low solid hydrolysis of the pretreated loose and pelleted corn stover was also carried out to see the effects of pelleting, pretreatment temperature, and pretreatment solid loading on hydrolysis yields. The quantified glucose and xylose yields of the subsequent low solid enzymatic hydrolysis are shown in Figures 2a,b, respectively. These figures represent the maximum possible yields as low solid loadings (1% glucan) were used, and sugar conversion plateaued between 24 h and 48 h.

The experimental results revealed that the pelleting significantly improved glucose and xylose yields, ranging from 25% to 45%, depending on temperature. These statistically significant improvements can be attributed to the distribution of ammonia in the pretreatment slurry. While the loose stover absorbed all the ammonia during the initial mixing of the slurry, the dense pellet mixture had a significant amount of ammonia in the reaction vessel, leading to better heat transfer and less scorching of the material.

Pretreatment temperature was also found to be a statistically significant contributor to the yield changes. The conversion decreased linearly for the loose corn stover as the temperature increased. However, the pelleted stover pretreated at 90 °C achieved slightly higher yields than 60 °C and significantly higher than 120 °C. These results suggested an upper-temperature threshold for soaking in aqueous ammonia pretreatment of loose and pelleted corn stover.

The solid loading changes for the pelleted corn stover resulted in different yield changes depending on the temperature. For 60 °C and 120 °C temperatures, the result showed that increasing the solid loading by 10% (w/v) led to statistically significantly different yields with drops of ∼5% - 15%. However, for the samples pretreated at 90 °C, increasing the solid loading from 20% to 40% (w/v) did not significantly impact the yields. Ultimately, maximum glucose yields of pelleted stover at solid loadings of 20%, 30%, and 40% (w/v) were above 90%. Maximum xylose yields were slightly lower but were above 85% for each solid loading. These results suggest that increasing pretreatment solid loadings of pelleted stover above those tested in this study may be possible.

Effects of pretreatment parameters on material characterization

FTIR spectral analysis of original and pretreated corn stover

Fourier transform Infrared spectrum analysis of the nontreated and ammonia soaked loose and pelleted corn stover is shown in Figure 3. The absorbance spectrum in Figure 3a showed a reduced intensity at 1730 cm -1, a band attributed to C=O bonds. This result was consistent with previous research suggesting pelleting disrupts ester groups in lignin and hemicellulose (Li et al., 2014; Xu et al., 2015; Nahar and Pryor, 2017).

FTIR profiles of the pelleted stover recorded at different temperatures also showed significant changes (Figure 3b). Following the 120 °C pretreatment, the relative absorbance change of the 1510 cm -1 band is considerably higher than the samples pretreated at 60 °C and 90 °C. Previous research has associated this band with the C=C stretching of lignin (Xu et al., 2015). This was surprising as higher severity ammonia pretreatments typically remove lignin (Nahar and Pryor, 2017).

FTIR profiles of the pretreated pelleted stover at different solid loadings were not significantly different in comparison to the pelletization and temperature changes (Figure 3c). This result suggested that the structure of lignin and hemicellulose are similar, regardless of the solid loading. However, the relative intensity of certain bands increased slightly as solid loading increased. Noteworthy changes include an increase of lignin and hemicellulose linkages (C-O-C) at 1462, 1425, and 1320 cm -1. The increase in these linkages suggested that the pretreatment was less effective, as removing linkages between cellulose, hemicellulose, and lignin was critical for enzyme accessibility (Faulon et al., 1994).

SEM imaging of original and pretreated corn stover

SEM images for nontreated and pretreated loose and pelleted corn stover at 20% (w/v) solid loadings are shown in Figure 4. Figures 4a,b exhibited minimal structural changes induced by the pretreatment of loose stover at 90 °C, as the corn stover fibers remained unadulterated. The 90 °C pretreated pelleted material displayed fiber peeling and exposure, correlating to the restructuring of hemicellulose and lignin. (Figures 4c,d). These distortions increased surface area and improved accessibility for enzymes and sugar yields in subsequent enzymatic hydrolysis tests (Kang et al., 2011). Comparing images of the untreated loose and pelleted corn stover showed pelleting disrupted the fibrils and increased the surface area before pretreatment (Figures 4a,c). This disruption can be attributed to the high shear force and heat developed during pelleting and was a contributor to the improved sugar yields observed in hydrolysis (Theerarattananoon et al., 2012; Nahar and Pryor, 2017). The SEM images were intended only as qualitative indicators of surface disruption. The coarse features reflect the high solids pretreatment conditions, which limit higher magnification imaging without introducing artifacts.

The pelleted corn stover following pretreatment at 120 °C under high magnification revealed the formation of spherical structures on the cellular surface (Figure 4e). While these features resemble similar morphologies reported under high-severity acid pretreatment (Selig et al., 2007; Hu et al., 2012), alkaline conditions used here are not known to promote pseudo-lignin formation. The surface deposits could be associated with unexpectedly low lignin removal and hydrolysis yields as it is detrimental to proper conversion (Pu et al., 2013). Further investigation into using complementary analytical techniques is needed to determine the origin and chemical nature of the surface deposits in alkaline pretreatment. This result suggested surface deposits formation at temperatures as low as 120 °C, and caution should be taken in increasing ammonia temperatures above 90 °C.

Effects of reactor design on high-solids enzymatic hydrolysis

Glucose and xylose yields following 72 h hydrolysis at 10%, 20%, and 30% (w/v) solid loadings are shown in Figures 5a,b, respectively. At low solid loadings (10% w/v), glucose yields above 87% and xylose yields above 81% were maintained with no statistical difference between the mixing strategies. However, in response to the increased solid loading, mixing with the gravimetric roller without inserts produced significantly higher sugar yields than the shake flask setup. At high solid loadings (20% and 30% w/v), rolling bottles without inserts produced maximum glucose yields of 71% and 50%, respectively and the yield was ∼25% higher than the shake flask setup. Compared to the shake flasks, this system displayed reduced variability even at high solids, implying homogeneous mixing. Observation of the rheology of the material during hydrolysis provided key insight into these results. While the material stuck together on the bottom of the shake flasks and stuck to the inserts in the roller reactor setup, the material became pourable around 60 h for the rolling reactor setup without inserts at 30% solid loadings (w/v), as the material was able to tumble and break apart on itself freely. This solubilization is critical in achieving maximum hydrolysis yields and is typical of high-energy, pilot-scale setups (Jørgensen et al., 2007; Roche et al., 2009). In this study, we relied on reproducible qualitative measurements to gain an initial understanding of the underlying mechanism. However, further research is needed to investigate how factors such as mixing dynamics and particle size influence the rheological behavior, which will be essential for fully elucidating the mechanism responsible for the improved performance of corn stover pellets observed in this work.

Our results suggest that a bench-scale gravimetric rolling system can effectively mix a high-solid hydrolysis slurry, but longer hydrolysis times (>72 h) are likely required for maximum conversion in a fed-batch approach. As time is a limiting factor in biorefinery economics, fed-batch methods were investigated to determine if quicker solubilization could improve the reduction of yields without reactor modification.

Effects of feeding strategies on high-solids enzymatic hydrolysis

Enzymatic hydrolysis conversion (72 h) for the three feeding methods is shown in Figure 6 as a function of time for one pretreatment condition (40% w/v, 90 °C). At low solid loadings (10% w/v), the batch method converted the most rapidly and resulted in yields slightly higher than the fed-batch methods. This was likely due to the initial solubility of the slurry, as the material was solubilized at the beginning of hydrolysis (<1 h) and allowed the enzymes to diffuse effectively. In comparison, the fed-batch methods performed significantly better under higher solid loadings (20% and 30% w/v) as solubilization was enabled earlier in hydrolysis due to the reduced solid-to-liquid ratio. As a part of this, it was found that the incremental addition of enzymes had no statistical effect on 72 h yields. This suggests that end-product inhibition played a minimal role, possibly due to the mixing system that was used (Kristensen et al., 2009; Xue et al., 2012). Additionally, it indicates that utilizing a high-energy impeller system could increase yields even further if earlier solubilization could occur. One additional observation from Figure 6 is that there are no data points at 24 h or 48 h for the 30% batch loading. These points were omitted as aliquots could not be taken due to a lack of liquid in the slurry.

Final glucose and xylose concentrations for the three different feeding methods are shown in Figure 7. Figure 7a exhibits hydrolysis results following the most effective pretreatment (20% w/v solid loading), while Figure 7b exhibits hydrolysis results following pretreatment with the highest solid loading (40% w/v). As shown, the 30% (w/v) solids fed-batch hydrolysis resulted in glucose concentrations of 112 and 102 g/L using 20% and 40% (w/v) pretreated material, respectively. These pretreatment and hydrolysis conditions also resulted in maximum glucose yields of 74% and 67% and total sugar concentrations of 161 g/L and 143 g/L. With high sugar concentrations, economical distillation of ethanol would likely occur following a subsequent fermentation or simultaneous saccharification and fermentation setup as ethanol concentrations above 4% (w/w) are desired (Jørgensen et al., 2007; Hodge et al., 2008).

High-solid processing is an advantageous strategy to improve biorefinery viability. This study investigated the impact of pelleting, pretreatment temperature, reactor design, and feeding strategies on sugar yields with high solid loadings conditions. Pelleting of corn stover allowed ammonia pretreatment loadings up to 40% (w/v) while still producing glucose yields above 90%. Gravimetric rolling and fed-batch methods improved high-solid hydrolysis solubilization and produced total sugar concentrations above 140 g/L. This study indicated that pelleting corn stover can increase sugar yields by up to 45%; the increased yields can help to offset the pelleting cost and make pelleting an innovative and cost-effective option for biorefineries.