Optimization and Validation of a GC FID Method for the Determination of Acetone-Butanol-Ethanol Fermentation Products

Transcription

1 Journal of Chromatographic Science 2014;52: doi: /chromsci/bmt022 Advance Access publication March 25, 2013 Article Optimization and Validation of a GC FID Method for the Determination of Acetone-Butanol-Ethanol Fermentation Products Xiaoqing Lin 1,2,3 *, Jiansheng Fan 1,2 *, Qingshi Wen 1,2, Renjie Li 2,3, Xiaohong Jin 1,2, Jinglan Wu 2,3, Wenbin Qian 2,3, Dong Liu 1,3, Jingjing Xie 1,2, Jianxin Bai 1,3 and Hanjie Ying 1,2,3, 1 State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing, China, 2 College of Biotechnology and Pharmaceutical Engineering, Nanjing University of Technology, Nanjing, China, and 3 National Engineering Technique Research Center for Biotechnology, Nanjing, China * These authors equally contributed to this study. Author to whom correspondence should be addressed. yinghanjie@njut.edu.cn Received 4 June 2012; revised 11 February 2013 An improved, simple gas chromatography flame ionization detection (GC FID) method was developed for measuring the products of acetone-butanol-ethanol (ABE) fermentation and the combined fermentation/separation processes. The analysis time per sample was reduced to less than 10 min compared to those of a conventional GC FID (more than 20 min). The behavior of the compounds in temperature-programmed gas chromatographic runs was predicted using thermodynamic parameters derived from isothermal runs. The optimum temperature programming condition was achieved when the resolution for each peak met the analytical requirement and the analysis time was shortest. With the exception of acetic acid, the detection limits of the presented method for various products were below 10 mg/l. The repeatability and intermediate precision of the method were less than 10% (relative standard deviation). Validation and quantification results demonstrated that this method is a sensitive, reliable and fast alternative for conventional investigation of the adsorption-coupled ABE fermentation process. Introduction Butanol is an important commercial chemical that is widely used in the plastic and textile industry. By the 1940s, butanol fermentation, also called acetone-butanol-ethanol (ABE) fermentation, was the second most important industrial fermentation process next to ethanol fermentation. A rise in the chemical synthesis of butanol led to a decline in ABE fermentation in the 1960s (1). However, due to its higher energy density than ethanol and because it can be used directly as a gasoline substitute in internal combustion engines, fermentation-derived butanol production on an industrial scale has attracted the interest of numerous companies (2). A typical feature of the solvent-producing Clostridium species is the transition from the acidogenic to the solventogenic phase. To obtain a holistic view of ABE fermentation and to maximize butanol productivity, analytical methods are required to measure acetic acid, butyric acid (acidogenic phase), acetone, butanol and ethanol (solventogenic phase) (3). Currently, various methods have been developed to measure the substrate and products of ABE fermentation. High-performance liquid chromatography (HPLC) has been used to analyze both fermentation products and substrates (3, 4). However, due to the poor resolution of butyric acid-acetone and acetone-ethanol in HPLC systems, it is of limited use. Buday et al. (3) described an improved analytical method for ABE fermentation, which involved adjusting the temperature of the Bio-Rad Aminex HPX87H column to 148C. The resolution between acetone and ethanol was also increased to 1.2, which allowed the quantitation of the concentrations of acetone and ethanol; however, the resolution between butyric acid and acetone was not significantly improved. Quantification of the solvent fermentation products based on gas chromatography with a flame ionization detector (GC FID) has also been reported in the literature. In GC, different columns, such as a glass column (10% CW-20M, 0.01% H 3 PO 4, support 80/100 Chromosorb WAW) (5), a DB-WAX capillary column (30 m 0.32 mm 0.50 mm) (6) oraninnowax capillary column (15 m 0.53 mm) (7), have commonly been applied to determine the concentrations of acetone, butanol and ethanol. However, there are few articles about the GC method for the determination of ABE fermentation broth. The analysis time was always more than 20 min, and the details on the performance of these methods are not generally given in the literature, particularly the repeatability, limit of detection (LOD), chromatogram and analysis times. According to Mes-Hartree and Saddler (8) and Doremus et al. (9), the major problem with several of the GC methods was the absorptive tailing of acetic acid, which can result in poor quantitation. Hence, as presented by Tsuey et al. (4), GC was used to analyze solvents, whereas HPLC was used to analyze organic acids. Although an improved GC method and a chromatogram were included in a later work, the analysis time reached min to avoid the overlapping peaks that occurred between acetone and ethanol (4). An improved GC method for the analysis and quantification of the culture broth with no sample pretreatment prior to GC analysis is described; the method also includes a lower limit of detection, less time and a wider linear range. This method can also be applied in the adsorption-coupled ABE fermentation process, which can enhance the fermentation rates and reactor productivity by adsorption resin. Various low boiling point eluents such as methanol were required to desorb the products from the different types of adsorbent. It is necessary for each eluent to improve the method to achieve fast and efficient analysis. In this paper, appropriate offline optimization models were conducted to overcome problems such as the short analysis time and peak overlapping effect without loss of resolution in the analysis of in situ product recovery (ISPR). For varied eluents, rapid and # The Author [2013]. Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oup.com

2 effective adjustment can be made to achieve an effective method and shorten the time. In the prediction process, acetic acid and butyric acid have obvious tailing and poor peak shapes in isothermal runs, leading to the trouble in prediction of the peak widths at half-height. According to the research of Lu et al. (10), the relationship between peak width and retention time is suitable for peaks eluted at different temperatures. Hence, the peak widths of acids obtained from different linear temperatures were used in the prediction of the peak widths of acids during temperature programming. The predicted results agree with the experimental results. This improved method was applied to the monitoring and control of butanol fermentation (the glucose concentration was determined with a glucose analyzer). Materials and Methods Chemicals and reagents HPLC-grade acetone, ethanol, methanol, butanol, acetic acid and butyric acid were purchased from Sinopharm Chemical Reagent Co. Stock solutions of acetone, ethanol, butanol, acetic acid and butyric acid were prepared at concentrations of 60, 40, 60, 20 and 20 g/l, respectively, in distilled water. A series of solutions of each analyte was prepared with isobutanol as internal standard (IS; 6 g/l) for the construction of calibration curves. Chromatographic conditions The experiments were performed using a GC system (Agilent 7890, Santa Clara, CA) equipped with an FID. Separation of compounds was conducted on a 60 m HP-INNOWAX capillary column of 0.25 mm i.d., coated with polyethylene glycol (0.25 mm film thickness), using nitrogen as the carrier gas. The injection volume was 1 ml and the flow rate was 2 ml/min. The injector temperature was 1808C with a split ratio of 90:1 and the FID temperature was 2208C. The oven temperature was programmed as follows: the column was held initially at 708C for 0.5 min, then increased to 1908C at 208C/min and held for 4 min. Chromatographic data were recorded and integrated using Agilent Chemstation software. The P2 medium was a synthetic medium and the concentrations of inorganic salts were very low. As the sample was injected into the GC, the sample first underwent gasification in the injection port liner, and then entered the capillary column with a split ratio of 90:1. The actual injection volume was ml, so there was no column damage. However, the injection port liner was contaminated after continuous use for more than three months. The injection port liner was cleaned using acetone by an ultrasonic cleaner and new glass cotton was placed. Microorganism and fermentation conditions The strain Clostridium acetobutylicum B3 was provided by the National Engineering Technique Research Center for Biotechnology (Nanjing, China). The strain was maintained in a glycerol stock culture in a cryogenic freezer at 708C. The seed medium consisted of the following components (g/l): soluble starch, 10; yeast extract, 3; peptone, 5; ammonium acetate, 2; sodium chloride, 2; KH 2 PO 4,1;K 2 HPO 4, 1; MgSO 4, 3; FeSO 4. 7H 2 O, The P2 medium, with glucose as the carbon source, was sterilized at 1218C for 20 min in the 50 L stirred tank fermenter. The fermentation broth samples were centrifuged (Centrifuge 5804R, Eppendorf, Hamburg, Germany) at 13,000 g at 48C for 3 min to separate sediments and the clear liquid was analyzed for ABE fermentation products. Before injection into the GC instrument, clarified samples and standards were filtered through 0.45 mm Whatman nylon filter (Gelman Science, Ann Arbor, MI) to remove insoluble materials that could block the column. All clear filtrate samples were kept frozen in sealed vials to maintain the stability of volatile components until they could be analyzed. Chromatographic samples were prepared with isobutanol as the IS (6 g/l) in 2 ml screw-cap septum vials, which were then loaded into the autosampler. The growth of Clostridium acetobutylicum B3 was determined by the measurement of the optical density at 660 nm (OD660) with the use of a spectrophotometer. Theory Estimation of retention times To achieve adequate resolution in the shortest possible analysis time, some theoretical and computational procedures were used to predict retention times and peak widths of all analytes. In capillary GC, the capacity factor value is influenced by thermodynamic parameters through the following equation: lnk ¼ DH RT þ DS þ lnb ð1þ where k is the capacity factor of the analyte i, R is the universal gas constant, T is the absolute temperature, b is the column phase ratio and DH and DS are the enthalpy and entropy of vaporization of analyte i from the stationary phase to the carrier gas phase, respectively. The variables DH and DS can be assumed to be temperature independent and depend only on the solute solvent interactions (11). A linear relationship between lnk and 1/T can be represented simply as lnk ¼ A þ B T The basic equation of the retention time in the capillary column under isothermal conditions is given by t r ðt Þ¼t m ½1 þ kðt ÞŠ where t m is the column dead time, a weak function of temperature. It can be considered to be a constant at constant flow rate mode and is not influenced by oven temperature (10). Here, the t m value was determined by Agilent Chemstation software. During a time interval Dt, the compound will move inside the column of a Dl length (12): u t ¼ Dl Dt ¼ u gas 1 þ k ¼ L t m ð1 þ kþ when in a linear gradient temperature step Dl ¼ 1 a T i þdt ð T i L dt t m ½1 þ kðt ÞŠ ð2þ ð3þ ð4þ ð5þ Optimization and Validation of a GC FID Method for the Determination of Acetone-Butanol-Ethanol Fermentation Products 265

3 where a is the heating rate and L is the column length: L ¼ Xn i¼1 Hence, the temperature-programmed retention time can subsequently be determined by assuming that the linear velocity of the carrier gas and the capacity factor are constant within each segment. Dl ð6þ retention time (14, 15). Thus, the peak width is related to the invented retention time, t * r, which is the function of the retention temperature of component it ri. In other words, it is possible to approximate the peak width in temperatureprogrammed GC to that in isothermal GC by means of the concept of invented retention time: tr ¼ t mð1 þ kþ ¼t m 1 þ exp A þ B ð12þ T n Estimation of peak widths Apart from retention time, it is necessary to predict peak width at half height for a programmed temperature run. Several methods based on the relationship in Eq. (7) were proposed for the calculation of w 1/2 : w 1=2 ¼ a þ bt r First method for predicting peak widths The width of the zone for an increment of the column can be described by Eq. (8) by making the following assumptions (13): (i) the plate height is considered to be a constant throughout the capillary column; (ii) the effect of temperature on the diffusion characteristics is neglected; (iii) the zone broadening affected by the decompression is neglected; (iv) the thermal expansion of the carrier gas is neglected: ð7þ pffiffi 0 l dw 1=2 ðlþ ¼W 1=2 pffiffiffi dl ð8þ L The value of the peak width, w 1/2, in capillary GC is related to the actual zone width, W 1/2, and the velocity in the column u by the equation: W 1=2 ðt Þ¼w 1=2 ðt ÞuðT Þ On the basis of Eqs. (7) and (8), the zone width at the end of the column can be expressed as W 1=2 ¼ ð tr 0 ð9þ " pffiffiffiffiffiffiffiffiffiffiffiffiffiffi# 0 l½t ðt ÞŠ w 1=2 ½T ðt ÞŠu½T ðt ÞŠ pffiffiffi dt ð10þ L The value of w is obtained by the following equation: w 1=2 ¼ W 1=2 u n ¼ W 1=2 L t m ½1 þ kðt r ÞŠ ð11þ where u n is the velocity of the solute eluted isothermally at the retention temperature T r. Second method for predicting peak widths Due to the poor peak shapes of acetic acid and butyric acid in isothermal processes, another way of approaching the prediction of programmed temperature widths is described. It has been proved that the relationship described by Eq. (7) in isothermal processes is compatible with that between the peak width in temperature-programmed GC and the invented Results and Discussion Prediction and optimization of GC separation It is possible to obtain the coefficients A and B from a series of isothermal data using Eq. (2). The retention time values for acetone, methanol, ethanol, isobutanol and butanol were measured at 80, 110 and 1408C, whereas the rest were determined at 160, 180 and 2208C. Here, methanol was chosen as an eluent to desorb the fermentation products from resin because of its low boiling point and because it was easily recoverable (16). The concentration of total products was estimated by the concentration of products in the fermentation liquid phase and those adsorbed in the resin. Table I shows that good linearity occurred between ln k and 1/T. Due to uncertainty in the determination of the peak widths of acetic acid and butyric acid in the isothermal processes, the coefficients a and b of these acids were obtained by experiments conducted at an initial temperature of 708C, increased at a rate of 5, 10 and 158C/min; the others were determined from the isothermal data measured at 80, 110 and 1408C (Table I). Using the coefficients from Table I, the retention times, t r, calculated as shown previously and peak widths at half height, w 1/2, were calculated by the method described previously. The second method significantly overestimated the peak widths. Hence, in addition to acetic acid and butyric acid, the peak widths shown in Table II were calculated by the method described previously. Considering the boiling point of acetone (the first eluted peak) and the resolution between acetone and methanol, the initial oven temperature was set at 708C when the retention factor was 0.5. The final oven temperature was optimized and set at 1908C to obtain acceptable LODs and loss of the stationary phase of the GC column. Resolution was measured by Eq. (13) to describe how well compounds were separated: 2½t r ði þ 1Þ t r ðiþš R ¼ 1:7½w 1=2 ði þ 1Þþw 1=2 ðiþš Table I Basic Data of Compounds in Isothermal Processes Compound lnk ¼ A þ B/T R 2 w 1/2 ¼ a þ bt r R 2 A B a 10 2 b 10 2 ð13þ Acetone Methanol Ethanol Isobutanol Butanol Acetic acid Butyric acid Lin et al.

4 Table II Comparison of Predicted and Experimental Retention Parameters Compound Retention time (min) Peak width (s) * Experimental Predicted E pred 10 2 Experimental 10 2 Predicted Predicted E pred E pred Acetone Methanol Ethanol Isobutanol Butanol Acetic acid Butyric acid *E pred ¼ value (Experimental) value (Predicted). To minimize the analysis time, 1.5 was chosen as the required resolution, R req, which was sufficient for utmost accuracy (17). The maximum allowed heating rate was 458C/min, which was used as the initial heating rate in the prediction of peak resolution. The final heating rate was identified by decreasing the maximum allowed heating rate to meet the required peak resolution. A temperature programming rate above 208C/min resulted in partial co-elution of methanol with ethanol (R, 1.3). The hold of 0.5 min at the initial temperature allowed complete separation of the analyte peaks and reduced the analysis time. The flow chart of the proposed optimization algorithm for ABE fermentation products is presented in the Supplementary materials. The predicted and experimental retention times and peak widths at half-heights of all compounds are shown in Table II. Excellent agreement was found between the predicted and experimental retention parameters. Except for the peak widths of acids, which were predicted by the previously described second procedure, the first procedure gave significantly better peak width predictions than the second procedure. The reason for this is partly because the linear velocity of a peak s may change rapidly under temperature programming. The applicability of the proposed method was tested by analyzing three samples. Figure 1 shows the chromatograms obtained from the analysis of model fermentation broth (Figure 1A), ABE fermentation broth (Figure 1B) and model eluent and the mixture of methanol and model fermentation broth, diluted with distilled water (Figure 1C). All compounds were clearly identified with no significant interferences from the sample matrix. Method validation Calibration curve and linearity Calibration curves were prepared by plotting the peak area ratios ( peak area of analytes/peak area of IS) versus concentration. As shown in Table III, good linearity for the method was obtained, with correlation coefficients in the range of , by using a series of solutions containing various concentrations of ABE, acetic acid and butyric acid. The linear ranges of acetone, butanol, ethanol, acetic acid and butyric acid were , , , and , respectively. Compared with the results reported in the previous related paper (the linear ranges of acetone, butanol, ethanol, acetic acid and Figure 1. GC chromatograms obtained from the analysis of reference standards in: model fermentation broth sample (A); fermentation broth sample (B); model eluent (C) sample. Peaks: 1, acetone; 2, methanol; 3, ethanol; 4, butanol; 5, acetic acid; 6, butyric acid; IS, isobutanol. Optimization and Validation of a GC FID Method for the Determination of Acetone-Butanol-Ethanol Fermentation Products 267

5 Table III Features of the Proposed GC Methods for the Determination of Acetone, Ethanol, Butanol, Acetic Acid and Butyric Acid Analyte Slope SD 10 4 Intercept SD 10 4 Linear range (g/l) N R 2 LOD (mg/l) Acetone Ethanol Butanol Acetic acid Butyric acid Table IV Recovery of Compounds in Fermentation Broth Determined at Three Concentration Levels Compounds Concentration in fermentation broth (g/l) Added (g/l) Recovery (%) RSD (%) (n ¼ 3) Acetone Ethanol Butanol Acetic acid Butyric acid butyric acid were 0 8, 0 4, 0 4, 0 4 and 0 8 g/l, respectively) (3), this optimized method has wider linear range. Limits of detection LODs were evaluated from a signal to noise ratio (S/N) equal to 3. As listed in Table III, the method allowed the detection of ABE in the range of mg/l, whereas the assigned values were 18 and 8.3 mg/l for acetic acid and butyric acid, respectively. Compared with the results reported in the previous related paper (LOD ranges of acetone, butanol, ethanol, acetic acid and butyric acid were 100, 80, 80, 90 and 100 mg/l, respectively) (3), this optimized method provided lower detection limits of the analytes. Recovery The recoveries of this method were quantified at three concentration levels, which are shown in Table IV. The recoveries of the five analytes ranged from 100 to 113%. Analysis of variance was performed, which showed no statistically significant difference between the recoveries with respect to concentration. In conclusion, these recoveries were consistent, precise and reproducible in the same samples under different concentrations. Precision and repeatability The precision of the method was determined from multiple analyses of the same sample on different days. The inter-day precision was calculated from the fermentation broth at 25.5, 55 and 76 h (Figure 2). The relative standard deviations (RSDs) were less than 10% at various concentrations of the fermentation broth. These results are shown in Table V. The satisfactory Figure 2. Concentration profiles of the substrate and products from ABE fermentation. Glucose concentration was determined with glucose analyzer and products were measured using the described method. Glucose, white circles; butanol, squares; acetone, diamonds; ethanol, upside-down triangles; butyric acid, triangles; acetic acid, black circles (A); growth of Clostridium acetobutylicum B3 (OD660) from ABE fermentation (B). precision indicated good performance and stability of the method for the quantitative analysis of the compounds in the ABE fermentation broth. Measurement of fermentation samples The control batch fermentation experiment was run at 378C and ph 5 with 60 g/l of glucose in P2 medium. The fermentation samples were removed at time intervals and processed 268 Lin et al.

6 Table V Intra-Day and Inter-Day Precision of the Proposed Method Analyte Concentration in fermentation broth Intra-day assay Inter-day assay (4 days) (g/l) RSD (%) (n ¼ 6) RSD (%) (n ¼ 3) Acetone Ethanol Butanol Acetic acid Butyric acid immediately for chromatography. Figure 2 shows the product profiles, substrate uptake and OD660 of Clostridium acetobutylicum B3 over the course of 97.5 h. ABE production by Clostridium acetobutylicum B3 in a batch culture was characterized by two distinct phases, the acidogenic phase (0 30 h) and the solventogenic phase (30 76 h). During the acidogenic phase, acetic and butyric acid were accumulated to concentrations of approximately 2.20 and 2.64 g/l, respectively. The concentration of acids decreased rapidly and ABEs were accumulated in the culture during the solventogenic phase. After 76 h of fermentation, the culture produced 4.24 g/l of acetone, g/l of butanol and 1.15 g/l of ethanol, resulting in ABE productivity of 0.21 (g/l. h) and ABE yield of 0.27 g/g. The concentrations of acetic and butyric acid at the end of fermentation were 0.59 and 0.22 g/l, respectively. Although the fermentation was run for 97.5 h, the culture stopped producing butanol at 76 h, leaving 2.3 g/l of residual glucose. The major reason for cessation of the fermentation process was the hydrophobic nature of butanol (1, 18). It is toxic and its primary effects appear to be disrupting membrane-linked functions, decreasing intracellular ATP levels and inhibiting sugar uptake (1, 18). In the batch fermentation process, the growth of Clostridium acetobutylicum B3 was not inhibited at a butanol concentration of 4.59 g/l. The growth of Clostridium acetobutylicum B3 was 50% inhibited at a butanol concentration of 7.29 g/l, and then was totally inhibited at a butanol concentration of g/l (Figure 2). Conclusions The present paper demonstrates a simple, sensitive, reliable and fast GC FID method for measuring the components of the ABE fermentation broth (acetone, ethanol, butanol, acetic acid and butyric acid). These features make it a reliable and convenient tool for the analysis of the ABE fermentation process and fermentation coupled with separation. Using the developed model, the predicted values of retention times and peak widths agreed with the experimental values. The optimal chromatographic conditions can be obtained by the predicted retention times and peak widths. Methanol was chosen as a suitable eluent for the characteristics of the absorbent used in this article. Other adsorbents were considered; a fast, reliable method for the direct analysis of the eluent of in situ adsorption in ABE fermentation may also be optimized quickly by the theory described previously in the paper. Acknowledgments This work was supported by Program for Changjiang Scholars and Innovative Research Team in University (Grant No.: IRT1066), National Outstanding Youth Foundation of China (Grant No.: ), National High-Tech Research and Development Plan of China (863 Program, 2012AA021200), Jiangsu Provincial Natural Science Foundation of China (Grant No.: SBK ) and Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). Xiaoqing Lin was supported by the average college graduate student research innovative projects of Jiangsu province (CXZZ11_0362). References 1. Jones, D.T., Woods, D.R.; Acetone-butanol fermentation revisited; Microbiology and Molecular Biology Reviews, (1986); 50: Pfromm, P.H., Amanor-Boadu, V., Nelson, R., Vadlani, P., Madl, R.; Bio-butanol vs. bio-ethanol: A technical and economic assessment for corn and switchgrass fermented by yeast or Clostridium acetobutylicum; Biomass and Bioenergy, (2010); 34: Buday, Z., Linden, J., Karim, M.; Improved acetone-butanol fermentation analysis using subambient HPLC column temperature; Enzyme and Microbial Technology, (1990); 12: Tsuey, L.S., Ariff, A.B., Mohamad, R., Rahim, R.A.; Improvements of GC and HPLC analyses in solvent (acetone-butanol-ethanol) fermentation by Clostridium saccharobutylicum using a mixture of starch and glycerol as carbon source; Biotechnology and Bioprocess Engineering, (2006); 11: Qureshi, N., Ezeji, T.C., Ebener, J., Dien, B.S., Cotta, M.A., Blaschek, H.P.; Butanol production by Clostridium beijerinckii. Part I: Use of acid and enzyme hydrolyzed corn fiber; Bioresource Technology, (2008); 99: Atsumi, S., Cann, A.F., Connor, M.R., Shen, C.R., Smith, K.M., Brynildsen, M.P., et al.; Metabolic engineering of Escherichia coli for 1-butanol production; Metabolic Engineering, (2008); 10: Tashiro, Y., Takeda, K., Kobayashi, G., Sonomoto, K., Ishizaki, A., Yoshino, S.; High butanol production by Clostridium saccharoperbutylacetonicum N1-4 in fed-batch culture with ph-stat continuous butyric acid and glucose feeding method; Journal of Bioscience and Bioengineering, (2004); 98: Mes-Hartree, M., Saddler, J.; Butanol production of Clostridium acetobutylicum grown on sugars found in hemicellulose hydrolysates; Biotechnology Letters, (1982); 4: Doremus, M.G., Linden, J.C., Moreira, A.R.; Agitation and pressure effects on acetone-butanol fermentation; Biotechnology and Bioengineering, (1985); 27: Lu, X., Kong, H., Li, H., Ma, C., Tian, J., Xu, G.; Resolution prediction and optimization of temperature programme in comprehensive twodimensional gas chromatography; Journal of Chromatography A, (2005); 1086: Snijders, H., Janssen, H.-G., Cramers, C.; Optimization of temperatureprogrammed gas chromatographic separations I. Prediction of retention times and peak widths from retention indices; Journal of Chromatography A, (1995); 718: Vezzani, S., Moretti, P., Mazzi, M., Castello, G.; Prediction of retention times in linear gradient temperature and pressure programmed analysis on capillary columns; Journal of Chromatography A, (2004); 1055: Optimization and Validation of a GC FID Method for the Determination of Acetone-Butanol-Ethanol Fermentation Products 269

7 13. Ba rtu, V.; Calculation of the retention time and peak width for the purpose of optimized gas chromatographic analysis; Journal of Chromatography A, (1983); 260: Al-Bajjari, T.I., Le Vent, S., Taylor, D.R.; Calculation of programmed temperature gas chromatography characteristics from isothermal data: IV. Prediction of peak widths; Journal of Chromatography A, (1994); 683: Peichang, L., Bingcheng, L., Xinhua, C., Chunrong, L., Guangda, L., Haochun, L.; Optimization and identification in any kind of multistep temperature programmed gas chromatography; Journal of High Resolution Chromatography, (1986); 9: Yang, X., Tsai, G.J., Tsao, G.T.; Enhancement of in situ adsorption on the acetone-butanol fermentation by Clostridium acetobutylicum; Separations Technology, (1994); 4: Matisova, E., Domotorova, M.; Fast gas chromatography and its use in trace analysis; Journal of Chromatography A, (2003); 1000: Alsaker, K.V., Paredes, C., Papoutsakis, E.T.; Metabolite stress and tolerance in the production of biofuels and chemicals: Gene-expression-based systems analysis of butanol, butyrate, and acetate stresses in the anaerobe Clostridium acetobutylicum; Biotechnology and Bioengineering, (2010); 105: Lin et al.