Laboratory-scale method for enzymatic saccharification of lignocellulosic biomass at high-solids loadings
© Roche et al. 2009
Received: 23 July 2009
Accepted: 04 November 2009
Published: 04 November 2009
Skip to main content
© Roche et al. 2009
Received: 23 July 2009
Accepted: 04 November 2009
Published: 04 November 2009
Screening new lignocellulosic biomass pretreatments and advanced enzyme systems at process relevant conditions is a key factor in the development of economically viable lignocellulosic ethanol. Shake flasks, the reaction vessel commonly used for screening enzymatic saccharifications of cellulosic biomass, do not provide adequate mixing at high-solids concentrations when shaking is not supplemented with hand mixing.
We identified roller bottle reactors (RBRs) as laboratory-scale reaction vessels that can provide adequate mixing for enzymatic saccharifications at high-solids biomass loadings without any additional hand mixing. Using the RBRs, we developed a method for screening both pretreated biomass and enzyme systems at process-relevant conditions. RBRs were shown to be scalable between 125 mL and 2 L. Results from enzymatic saccharifications of five biomass pretreatments of different severities and two enzyme preparations suggest that this system will work well for a variety of biomass substrates and enzyme systems. A study of intermittent mixing regimes suggests that mass transfer limitations of enzymatic saccharifications at high-solids loadings are significant but can be mitigated with a relatively low amount of mixing input.
Effective initial mixing to promote good enzyme distribution and continued, but not necessarily continuous, mixing is necessary in order to facilitate high biomass conversion rates. The simplicity and robustness of the bench-scale RBR system, combined with its ability to accommodate numerous reaction vessels, will be useful in screening new biomass pretreatments and advanced enzyme systems at high-solids loadings.
As the demand for non-petroleum based fuels continues to grow, more emphasis will be placed on producing a cost-competitive liquid transportation biofuel such as ethanol. One clean and renewable domestic energy source that can feasibly displace a significant fraction of petroleum usage in the USA is ethanol produced from lignocellulosic biomass [1–3]. Although large-scale ethanol production is not a new concept, converting lignocellulosic biomass to ethanol is not a trivial matter. Significant challenges lie with hydrolysis of biomass into fermentable sugars. Advanced conversion technologies must be developed to allow for the efficient conversion of lignocellulosic biomass to ethanol [4, 5].
Research indicates that a chemical pretreatment followed by enzymatic hydrolysis increases the overall saccharification efficiency [6, 7]. A promising approach to improving process economics involves increasing biomass concentration in both pretreatment and enzymatic hydrolysis. Higher starting biomass substrate concentrations lead to higher product concentrations throughout the production process. This will result in reduced capital and production costs associated with the reduction of equipment size and energy usage for heating, cooling and mixing. While a few high-solids enzymatic hydrolysis studies have been conducted, including experiments with rotating horizontal reactor vessels, most have only been at scales ≥ 1 L [4, 8–12]. To study the synergisms between pretreatment and enzymatic hydrolysis at process relevant conditions, it is necessary to be able to effectively screen both pretreated biomass and enzyme preparations at high-solids loadings (≥ 15% insoluble solids) in small-scale reactors.
Typically, shake flasks (SFs) are used for screening both pretreated biomass for enzyme digestibility and enzyme preparations for efficiency and effectiveness in digesting biomass at the National Renewable Energy Laboratory (NREL) and elsewhere. However, this is usually performed at a low insoluble solids loading (< 10%). At high insoluble solids loadings (≥ 15%), mixing modes that require bulk fluidity, such as conventional stirring or shaking, become ineffective [10, 11]. Mass transfer limitations that occur with ineffective mixing, such as poor enzyme distribution and localized hydrolysis product build-up, confound saccharification screening results and must, therefore, be mitigated.
In this study, we compared small-scale enzymatic saccharification vessels with three different mixing mechanisms: shaking, gravitational tumbling and hand stirring. This comparison assessed the reaction systems for their efficiency and repeatability in converting biomass at high-solids loadings, where biomass conversion was the measure of effectiveness of enzymatic saccharification. When biomass and enzyme are effectively mixed, yield is similar, regardless of reactor system. The roller bottle reactor (RBR) system was well mixed in every instance of continuous rolling. Therefore, our results showed that a high-solids enzymatic saccharification method requiring the least user intervention (that is the RBR) would be the most efficient system for this work. Using this method, bench and floor scale enzymatic saccharifications verified the scalability of the reactor system. A few different intermittent mixing modes were also studied. In addition, we present a general equation for calculating conversion in high-solids systems and an extension of that equation to include significant, strategically chosen biomass hydrolysis products. This method, including the general conversion equation, can be applied to screen a wide variety of pretreated biomass, as well as enzyme preparations at high-solids loadings.
Pretreatment conditions and yields.
1.6 to 2.7†
1.6 to 2.2†
3 to 5 min‡
2 to 4 min‡
0.8 to 1.9
0.8 to 1.3
-0.2 to 0
-0.5 to -0.2
Monomeric xylan Yield
Total xylan yield
Three small-scale reaction systems with different mixing modes (shaking, gravitational tumbling and hand stirring) were evaluated for their efficiency and consistency in converting high-solids loadings of biomass. For mixing via shaking, experiments were performed in 250 mL wide-mouth Pyrex bottles (Thermo Fisher Scientific, Inc, MA, USA). A rotary shaking incubator (New Brunswick Scientific, NJ, USA) provided mixing at 130 rpm and maintained the temperature of the SFs. The SFs were not homogenized or hand-mixed before sampling unless specified in the mixing-mode description. For gravitational tumbling and hand stirring, experiments were performed in wide-mouth polypropylene bottles of two sizes: 125 mL and 250 mL (Thermo Fisher Scientific, Inc, MA, USA). The reaction vessels that were mixed via gravitational tumbling, denoted as RBRs, rotated horizontally at 2 rpm for 250 mL bottles and 4 rpm for 125 mL bottles on a three deck roller apparatus for mini bottles (Wheaton Industries Inc, NJ, USA). Previous work showed that mixing speed does not affect the biomass conversion in the range of 2-20 rpm . Temperature control was achieved by housing the roller apparatus in a general purpose incubator (Model 1545, VWR International, LLC, PA, USA). The hand-mixed reaction vessels (HMR) were stirred with a sterilized spatula for 30 s at each prescribed mixing time and were incubated standing vertical in a stationary incubator.
Inside a laminar flow hood, autoclaved or ethanol sterilized SFs, RBRs and HMRs were loaded with a gram mass that was equal to half of the volume capacity of the vessel in millilitres (for example, 125 g into a 250 mL vessel). Of that mass, reactors were charged with 15%, 20% or 30% insoluble solids and 50 mM (pH 4.8) sterile filtered citrate buffer. Tetracycline (10 μg/mL) was added to inhibit microbial contamination. Two enzyme preparations were used for our studies: spezyme CP [Lot No. 301-05021-011] and GC220 [Lot No. 4900759448] (Genencor-Danisco, NY, USA). For spezyme CP and GC220, respectively, the total protein was assayed at 134 mg/mL and 202 mg/mL (BCA assay, Pierce Biotechnology, Inc, IL, USA) and the specific activities were determined to be 0.49 FPU/mg protein and 0.60 FPU/mg protein using the NREL laboratory analytical procedure "Measurement of cellulase activities" . The enzyme preparations were sterile filtered prior to loading into the reaction vessels. The enzyme was loaded at 5, 10, 15, 20, or 30 mg protein/g cellulose. The cellulose content of each pretreated material is summarized in Table 1.
Reactors loaded with PCS, the buffer and the antimicrobial agent were brought to 48°C, while mixing, before adding the enzyme. The enzyme was distributed on the surface of the PCS slurry across the length of the vessel. The enzyme was not mechanically mixed into the RBR or the SF unless specified. For 250 mL vessels, enzymatic saccharification slurry samples (~1 mL for 15% and 20% initial insoluble solids loadings and ~2 mL for 30% insoluble solids loadings) were taken every 4 hours for the first 8 hours and then once a day thereafter for a total of 7 days. Saccharifications reactions performed in 125 mL vessels were sampled at 8 hours, 1 day, 2 days, 4 days and 7 days. Reactor sampling was performed under sterile conditions in a laminar flow hood using standard aseptic techniques. The samples were not homogenized by hand mixing prior to sampling for the RBRs or for the SFs unless specified. Although our avoidance of hand mixing is unconventional, it was necessary in order to prevent mixing in addition to the primary mixing mechanism of study.
Enzymatic hydrolysis slurry samples were centrifuged in 0.45 μm nylon membrane microcentrifuge filters (No. ODM45C35, Pall Corp, MI, USA) at 12,500 rpm for 5 min. The filtered liquid was diluted 1:5 with deionized water in high-performance liquid chromatography (HPLC) vials for subsequent analysis. Samples were run on an Agilent 1100 series HPLC with a Shodex SP0810 Sugar Column (Kawasaki, Japan) run at 85°C. Deionized water pumped at 0.6 mL/min was the eluent.
The density of the end-point hydrolyzate slurry liquid fraction (ρ l ) was measured on a density meter (DMA5000, Anton Paar, VA, USA). Values for mass fraction insoluble solids (f is ) of the slurry at the end of the reaction were measured using a direct calculation method . Briefly, the mass fraction total solids (soluble and insoluble) of the hydrolysis slurry and the mass fraction soluble solids in the separated liquid were determined using HR83 halogen moisture analysers, and the f is of the hydrolysis slurry was calculated from these measures with mass balance relationships.
where r j, iis the molecular weight (MW) ratio of the polysaccharide j to its respective mono-or oligo-saccharide i (for example, r G, cbis two glucan units to cellobiose [324.32/342.34]), f i is the mass fraction of component i as a part of the total slurry [g i/g slurry], x j is the mass fraction of j in the insoluble solids [g j/g insoluble solids], N is the number of hydrolyzed components considered, M is the number of insoluble solids hydrolyzable components considered, Δ denotes a change from the initial conditions and the subscripts is and 0 refer to insoluble solids and initial condition, respectively.
where m is the density additive amount for concentrations of glucose, cellobiose and xylose, which was determined as 0.456, and ρ l,0 is the initial density of the liquid fraction of the saccharification slurry. Xylose is included in the f is and ρ l estimations because the xylan conversion to xylose contributes significantly to the decrease in f is and increase in liquid density .
where t = t (ν, 95) is the 95% t-value corresponding to ν = N - 1 degrees of freedom, S x is the standard deviation, and N is the number of values. The uncertainty values calculated in this way were used for the error bars in the figures.
The results show that, with mixing using only the shaking action of the rotating platform, SFs give lower conversions for every insoluble solids loading tested with respect to the corresponding insoluble solids loading in RBRs. Results similar to the RBRs have been observed in SFs when the enzyme is well mixed by hand initially and before each sample, even up to 30% insoluble solids (see section Intermittent Mixing below). At 30% initial insoluble solids loadings, RBRs converted approximately 2.5 times as much of the potential sugar as the SFs. Not only do the SFs give lower yields but they also show greater variability, particularly with the higher solids loadings. The greater variability can probably be attributed to the lack of consistency of the shaking only mechanism among high-solids SFs, coupled with sampling of the heterogeneous mixture within each individual SF. RBRs enable a higher conversion of biomass at all insoluble solids loadings tested here than the SFs using the shaking mechanism alone. This is likely due to a more effective mixing by horizontal tumbling and, therefore, fewer mass transfer limitations and localized product build-up. By visual observation, the 30% initial insoluble solids RBRs began to liquefy within 4 hours, whereas it took the unstirred SFs nearly 6 days to begin to liquefy.
In order to examine minimum mixing requirements for adequate mass transfer, we looked at several intermittent mixing modes: hand mix at time zero (HMR t0); hand mix at time zero; and then hand mix once per day (HMR t0, 1pd), shake flask with hand mix at time zero and then at each sampling (SF; HMR t0, sample), roll for 4 hours after time zero and then roll for 1 hour per day (RBR 4 hr, 1hrpd), roll for 4 hours after time zero and then roll for 1 hour per 2 days (RBR 4 hr, 1hrp2d), and roll for 24 hours after time zero and then shake for remainder (RBR 24 hr, SF). This set of experiments was run in quadruplicate in 125 mL HMRs and RBRs. For these experiments, all chemical conditions were held constant at 20% initial insoluble solids of PCS 1 and 20 mg protein/g cellulose GC220 enzyme loading. The hand-mixed SF experiments (SF; HMR t0, sample) were run separately in 250 ml wide-mouth Pyrex bottles filled with 100 g of biomass.
The data indicate conversion trends for the specific mixing modes. The rate of enzymatic hydrolysis is higher for the conditions that are mixed well early in the experiment, experiencing better conversions than conditions with less initial mixing. For conditions that received at least intermittent mixing throughout the 7 days of enzymatic saccharification, we observed a higher continued rate of conversion compared to saccharifications with no further mixing beyond the initial mixing. The experiment that was hand-stirred only initially exhibits the first conversion trend of a high initial rate of conversion, while the RBRs that were mixed 4 hourly initially, and intermittently thereafter, follow the second conversion trend of continued higher conversion later in the reaction. The intermittent mixing conditions, HMR that was mixed at time zero and once per day thereafter and RBR that rolled for 24 hours and shaken for the remainder, exhibited both of these behaviours. The hand-mixed SF experiment was not sampled at 8 hours and, therefore, it is not possible to determine the initial rate.
The hand-mixed SF mixing mode provided statistically equivalent cellulose conversions as the continuous roller-bottle mixing mode, both for 20% initial insoluble solids (Figure 6) and 30% initial insoluble solids (data not shown). Although not always documented in experimental procedures, it is customary to hand-mix the biomass after enzyme addition to and before sampling from SFs when experiments with high-solids loadings are performed. It is important to note that, in the hand-mixed SF method, hand mixing is a more prominent mixing mode than shaking for the high-solids conditions early in the saccharification reaction. However, hand mixing is a somewhat ill-defined mixing mode, in that its effectiveness will depend on how frequently the hand mixing is performed, and hand mixing would be difficult to replicate at pilot and manufacturing scales. Conversely, the continuous roller-bottle mixing mode is well defined and there is precedent for mixing by rotation at larger scales (for example, cement mixers).
Horizontally rotating bottles for the enzymatic saccharification of biomass has proved to be a more significantly consistent reactor system than SFs at high-solids loadings. The gravitational tumbling achieved in the RBRs provided sufficient mixing throughout the entire reaction vessel, thus mitigating mass transfer limitations that may confound enzymatic saccharification results. Using small-scale, continuously rotating bottle reactors, we have developed a method for screening pretreated biomass for enzyme digestibility and enzyme preparations for effectiveness in digesting biomass at process relevant, high-solids conditions. The reactor system and method was shown to be scalable between bench (125 mL) and floor (2 L) scales at 20% initial insoluble solids loadings. As a part of this enzymatic saccharification method, we generalized and expanded previously developed yield calculations that account for the multiphase nature of the high-solids enzymatic saccharification slurry. The results support our hypothesis that this reactor system will work well for a variety of biomass types and pretreatment severities at high-solids conditions. The simplicity of this reactor system and its ability to accommodate numerous reaction vessels will be useful in screening new biomass pretreatments and advanced enzyme systems at high-solids loadings.
Intermittent mixing regimes for the RBR and hand-stirred systems were explored to determine whether adequate conversion can still be obtained with substantially less total mixing, thus reducing energy costs. Effective initial mixing to promote good enzyme distribution and continued, but not necessarily continuous, mixing is required to facilitate high conversion rates.
hand-mixed reaction vessel
high-performance liquid chromatography
National Renewable Energy Laboratory
pretreated corn stover
roller bottle reactor
This work was funded by the US Department of Energy through the office of the Biomass Program. We would like to thank Dan Schell and Nick Nagle for providing the pretreated biomass used in this study and Jeff Wolfe for performing the biomass analysis. We would also like to thank Nancy Dowe-Farmer for providing the hand-mixed SF experimental data.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.