Monitoring of the cellulosic ethanol fermentation process by near-infrared spectroscopy
Graphical abstract
Introduction
The first commercial-scale plants for the production of second-generation (2G) or cellulosic ethanol have already begun operating worldwide, accompanied by new challenges associated with large industrial-scale processes involving complex lignocellulosic materials. One difficulty that still needs to be overcome concerns the requirement to work with high solids loadings in order to make the process economically feasible, and the consequent increased levels of inhibitors (Martins et al., 2015, Schneiderman et al., 2015). Carbohydrate and lignin degradation reactions during the biomass pretreatment step lead to the formation of inhibitors that can negatively affect the performance of enzymes and yeasts (Mosier et al., 2005). Therefore, techniques able to facilitate the monitoring of reactions and control of the enzymatic hydrolysis and fermentation process are highly desirable in the biofuels industrial sector. Quantification of glucose consumption and ethanol formation during the fermentation step of the 2G ethanol production process is crucial for bioprocess optimization and the achievement of high efficiency, leading to overall cost reductions. However, a unique challenge for monitoring the fermentation broth derived from a lignocellulosic source, such as sugarcane bagasse, is related to the presence of such plant cell-wall degradation compounds which includes soluble lignin, phenolics, sugar degradation products (furfural, hydroxymethyl furfural), organic acids, as well as monomeric and oligomeric sugars (Kim et al., 2011, Mosier et al., 2005). Such a diverse and complex mix of compounds imposes technical limitations to the development of analytical methodologies capable of discriminating and quantifying specific compounds in lignocellulosic derived fermentation broth (Ewanick et al., 2013). In addition to that, there are drawbacks associated with the conventional analytical methods used to characterize the compounds present in this highly complex medium. Techniques such as high performance liquid chromatography (HPLC) usually require off-line manual sampling and involve laborious sample preparation procedures that can be costly, time-consuming, and present a contamination risk (Ewanick et al., 2013). Conventional chemical analytical methods do not enable the acquisition of real-time information on the factors that affect the performance of the fermentation process. Hence, the development of techniques and methodologies able to directly obtain information during lignocellulosic biomass fermentation, in a fast, accurate, and reproducible manner, would offer new ways of effectively monitoring and controlling large industrial-scale bioprocesses.
Near-infrared (NIR) spectroscopy with multivariate calibration represents an alternative to the conventional chemical techniques. This methodology can be used to rapidly and non-destructively determine multiple compounds in a reaction medium in a single measurement, quantifying multi-constituent mixtures with minimal or no sample preparation (Blanco and Villarroya, 2002). However, NIR spectroscopy requires a calibration step to establish the mathematical model employed to give the concentrations of specific components of the sample (Lourenço et al., 2012). Principal component analysis (PCA), principal component regression (PCR), and partial least squares (PLS) regression are examples of chemometric methods frequently used in multiple regression spectrometric calibration. These chemometric methods are widely used for spectral data analysis because the models obtained can be easily interpreted and the required information can be efficiently retrieved. The PLS methodology is commonly applied in the monitoring of complex bioprocesses and has been used with spectrometric data for various bioprocess analyses (Milligan et al., 2014, Ribeiro et al., 2008, Rodríguez-Zúñiga et al., 2014, Sampaio et al., 2014). Studies have shown that the pretreatment of NIR spectra can be applied to remove noise and facilitate the quantification of specific compounds in multicomponent mixtures, including alcoholic fermentation broths (Scarff et al., 2006). For instance, there are various reports describing the use of PLS-NIR methodology to monitor alcoholic fermentation from a purified source of glucose (Blanco et al., 2004, Blanco et al., 2006, Finn et al., 2006) as well as from a starch source of fermentable sugar (Hao et al., 2012, Liebmann et al., 2009). The applicability of an external optical-fiber NIR instrument to monitor the alcoholic fermentation process has also been previously reported (Cavinato et al., 1990). Furthermore, the PLS-NIR technique is able to discriminate similar biological compounds (such as glucose and xylose) present in fermentation reactions (Monrroy et al., 2015, Morita et al., 2014). Although NIR has been successfully applied in a broad spectrum of different bioprocesses, it has not yet been fully investigated for monitoring the progress of the fermentation process step in 2G ethanol production which uses lignocellulosic sources of biomass as feedstock. Given the highly complex nature of the medium derived from a lignocellulosic source it is expected that the mathematical treatment of NIR spectra may present unique challenges in order to obtain reliable models.
The aim of the study was to develop a NIR spectroscopy technique to monitor the progress of glucose consumption and ethanol production during the alcoholic fermentation reaction of the 2G ethanol process. For this purpose, lignocellulosic residues (bagasse, straw, and tops) from four commercial sugarcane varieties were used for 2G ethanol production, following the sequential steps of dilute acid pretreatment, enzymatic hydrolysis, and alcoholic fermentation by Saccharomyces cerevisiae. The mathematical treatment of the NIR spectra was performed using PLS regression and the number of latent variables was defined based on the results of leave-one-out cross-validation (LOOCV).
Section snippets
Fermentation broth samples
The materials used were three untreated residue fractions (bagasse, straw, and tops) from four commercial varieties of sugarcane (SP79-1011, RB867515, SP81-3250, and RB92579), kindly supplied by Sumaúma Mill (Marechal Deodoro, GO, Brazil). The preparation, pretreatment, enzymatic hydrolysis, and alcoholic fermentation steps applied to the lignocellulosic sugarcane biomasses were carried out according to the methodology described by Pereira et al. (2015). In brief, the pretreatment step was
Removal of outliers
The Leverage test was used to identify possible outliers in the data set, since the presence of outliers could lead to inaccuracy in predicting the concentration values (Drennen et al., 1991). As a result, five outliers were removed from the fermentation broth sample set. After application of the Leverage test, the data consisted of 103 fermentation broth samples. The training set consisted of 83 experimental data (44 samples from sugarcane bagasse, 21 samples from straw and 18 samples from
Conclusions
NIR spectroscopy combined with PLS regression was able to monitor both glucose and ethanol during 2G ethanol production. The PLS-NIR model was based on different varieties of sugarcane, and could be applied to bagasse, straw and tops of sugarcane residues. The SMA, NWWL, and first derivative methods constituted the best model (M4) for spectrum preprocessing. Furthermore, the calibration data set was effective for prediction of the concentrations of the fermentation components, with RMSEP lower
Acknowledgements
The authors thank the Brazilian agencies FAPESP, CAPES, and CNPq for financial support, and the staff of Embrapa Instrumentation for technical assistance.
References (30)
- et al.
Critical review of chemometric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy
Trac-Trend. Anal. Chem.
(2010) - et al.
NIR spectroscopy: a rapid-response analytical tool
Trac-Trend. Anal. Chem.
(2002) - et al.
Monitoring alcoholic fermentation by joint use of soft and hard modelling methods
Anal. Chim. Acta
(2006) - et al.
Prediction of the ethanol yield of dry-grind maize grain using near infrared spectroscopy
Biosyst. Eng.
(2012) - et al.
Soluble inhibitors/deactivators of cellulase enzymes from lignocellulosic biomass
Enzyme Microb. Technol.
(2011) - et al.
Determination of glucose and ethanol in bioethanol production by near infrared spectroscopy and chemometrics
Anal. Chim. Acta
(2009) - et al.
Effects of the pretreatment method on high solids enzymatic hydrolysis and ethanol fermentation of the cellulosic fraction of sugarcane bagasse
Bioresour. Technol.
(2015) - et al.
Features of promising technologies for pretreatment of lignocellulosic biomass
Bioresour. Technol.
(2005) - et al.
Multivariate calibration methods applied to the monitoring of the enzymatic synthesis of amipicilin
Chemom. Intell. Lab.
(2008) - et al.
In situ near infrared spectroscopy monitoring of cyprosin production by recombinant Saccharomyces cerevisiae strains
J. Biotechnol.
(2014)
Quantifying second generation ethanol inhibition: design of experiments approach and kinetic model development
Bioresour. Technol.
Multivariate calibration leverages and spectral F-ratios via the filter factor representation
J. Chemom.
Analytical monitoring of alcoholic fermentation using NIR spectroscopy
Biotechnol. Bioeng.
Noninvasive method for monitoring ethanol in fermentation processes using fiberoptic near-infrared spectroscopy
Anal. Chem.
Near-infrared reflectance spectroscopy-principal components regression analyses of soil properties
Soil Sci. Soc. Am. J.
Cited by (35)
Towards achieving online prediction of starch in postharvest sweet potato [Ipomoea batatas (L.) Lam] by NIR combined with linear algorithm
2023, Journal of Food Composition and AnalysisElectronic nose signals-based deep learning models to realize high-precision monitoring of simultaneous saccharification and fermentation of cassava
2022, Microchemical JournalCitation Excerpt :So, how to quickly and effectively monitor the changes of key parameters or odor information in the process of cassava SSF, so as to master the process of cassava SSF in real time, has important practical significance for ensuring the quality of fermented products. For the past few years, optical sensor has been widely used in fermentation process monitoring, which greatly improves the efficiency of rapid monitoring of fermentation process parameters [14–20]. The detection mechanism of the optical sensor technology is mainly based on the change of the internal components of the fermentation matrix, which causes the change of the spectral absorption front in different bands, and the change state of the key parameters is indirectly reflected through this change information [21,22].
Evaluation of weight loss and high heating value from biomasses during fungal degradation by NIR spectroscopy
2022, FuelCitation Excerpt :Given this, near-infrared (NIR) spectroscopy has been used due to being a cheaper, quicker, non-destructive technique, which reduces technical risk and requires minimal preparation. It can also be applied in real-time [17–20]. The NIR spectroscopy technique uses an electromagnetic spectrum range from 12000 to 4000 cm−1 and corresponds to an analytical technique based on vibrational spectroscopy [21,22].