Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author. T H E ACETONE BUTANOL ETHANOL FERMENTATION: PRELIMINARY STUDIES ON SOME PRACTICAL ASPECTS by JAMES RICHARD GAPES a thesis presented in partial fulfilment of the requirements for the degree of MASTER OF TECHNOLOGY in Biotechnology Supervised by Drs. V.F. Larsen and I.S. Maddox, Department of Biotechnology, Massey University, New Zealand. 1982 (ii) ABSTRACT The dilute nature of solvents at the end of fermentation and slow overall rate of fermentation are major economic burdens on a commercial plant producing acetone, butanol, and ethanol. A preliminary feasibility costing of such a plant showed the cost of fermenters represents almost 50% of the total purchased equipment cost, and emphasised the need for improvements in the fermentation. Experiments were performed in 10-litre pressure vessels, a 1. 5-1 i tr e vessel pressure, and trial runs in 100 ml bottles. and 30-litre at atmospheric Good correlations were found for the different fermentation headspace pressures (100 to 250 kPa abs.) and minimum observed pH's (pH 4.2 to pH 4.65) with final butanol yields (0.92 to 11.6 g/1); increases in both parameters correlat­ ing with increased butanol concentration. Ethanol was found to be correlated with pressure only, and acetone with neither parameter directly. Other chemical species present in the broth were also correlated with each other. It was found that a tree diagram drawn using the strongest correlations resembled closely the known metabolism of the organism in terms of the metabolic pathways, specification of active forms of the metabolites, and effect of external influences. Use of multiple linear regression in this manner was named The Factor Correlation Method, and is potentially useful for research on metabolism and similar investigation on a much broader basis. Application of this technique showed that the pressure effect was possibly due to more than a single metabolic cause, and further experi­ ments also emphasised the complex nature of the pressure effect. The experimental work also highlighted the potential hazard of culture degeneration leading to substandard (iii) fermentation yields and eventual nonviability. Discussion on the experimental results and of the literature suggests the phenomenon is due to infection by lysogenic phage rather than spontaneous mutation, and an approximate model based on simultaneous partial differential equations parallels some observed characteristics of the phenomenon. Other topics include theoretical exercises with laboratory work on the water tolerance of methanol-petrol mixtures, the error associated with cell enumeration using a haemocytometer, and evaluation of growth and solvent production characteristics and some relevant parameters. (iv) ACKNOWLEDGEMENTS I wish to thank Drs. V.F. Larsen and I.S. Maddox for the opportunity to do the work and for the help provided to me. I would also like to acknowledge the help of Messrs J. Alger and P. Shaw of the Biotechnology Department's Workshop, of Mr M. Stevens and the other staff of the Department's Laboratory, and of Miss S. Douglas who typed this work, all of whom made the experimental work and final presentation so much easier. I am grateful also to my family for their continued support and interest throughout the entire degree course. CONTENTS ABSTRACT -;c ACKNOWLEDGEMENTS CONTENTS LIST OF VARIABLES AND ABBREVIATIONS LIST OF FIGURES LIST OF TABLES CHAPTER 1 INTRODUCTION 1.1 Prelude 1.2 History 1. 3 The Organism 1.4 The Process 1.5 The Biochemistry CHAPTER 2 MATERIALS AND METHODS 2.1 Materials 2. 1. 1 Media 2.1.2 Sugars 2 .1. 3 Chromatographic Standards 2.1.4 Gases 2. 1.5 Other Chemicals 2.2 Organisms 2.3 Media and Equipment Sterilisation 2.4 Inoculurn Preparation 2.5 Fermentation Equipment and Methods ~ ii iv V X xvii xix 1 9 10 12 15 17 17 21 21 21 21 22 22 23 2.5.1 2.5.2 2.5.3 2.5.4 100 ml Scale Fermentations 1.5-litre Fermentations 10-litre Fermentations 30-litre Fermentations 2.6 Analytical Methods 2.6.l 2.6.2 2.6.3 2.6.4 pH Measurement Cell Counts Solvents and Fatty Acids Lactose CHAPTER 3 PRODUCTION OF n-BUTANOL FROM WHEY PERMEATE WITH DIFFERENT PRESSURES 3.1 Introduction 3.2 1.5-litre Experiments 3.2.1 3.2.2 Results Discussion 3.3 10-litre Scale Experiments 3.3.l 3.3.2 Results Analysis of Data 3.3.2.1 3.3.2.2 3.3.2.3 Method of Data Analysis Solvent Relationships with Pressure and Minimum pH Solvent Relationships with Acids, Pressure, Final pH, and 23 23 29 30 31 31 31 33 34 36 38 40 55 56 Minimum pH 62 3.3.3 3.3.2.4 Summary of Correlation Result Discussion of Results 3.3.3.1 3.3.3.2 Overview Anomalies in the Correlations 65 79 81 3.3.4 3.3.3.3 3.3.3.4 3.3.3.5 3.3.3.6 3.3.3.7 3.3.3.8 3.3.3.9 The Typical Fermentation 82 Fluctuations in the Stationary Phase Cell Population 85 Cell and Solvent Correlation 87 Solvent and Acid Correlations 88 The Effect of Pressure 89 The Effect of Minimum pH 94 Maximum Specific Rates for Growth and Butanol Production 97 3.3.3.10 Carbohydrate Usage and Fermentation Efficiency 3 . 3 • 3 . 11 Ru n I Discussion of Factor Correlation Method 99 99 102 3.4 30~litre Scale Experiments 3.4.1 3. 4. 2 Results Discussion CHAPTER 4 A.B.E. FERMENTATION OF WOOD SUGAR HYDROLYSATES AND PENTOSE SUGARS 4.1 Introduction 4.2 WSR and Pentose Sugar Fermentations - 4. 2 .1 4.2.2 Results Discussion 4.3 WSH Fractions Experiments 4.3.1 4.3.2 Results Discussion 104 104 108 109 111 112 113 CHAPTER 5 DECLINING PERFORMANCE IN THE A.B.E. FERMENTATION 5.1 Introduction 114 5. 2 Results 5.3 Discussion CHAPTER 6 PRELIMINARY DESIGN AND COSTING OF AN A.B.E. FERMENTATION PLANT 6 • 1 Plant Design Paraneters 6.2 Plant Layout 6.3 Predistillation Processes 6.3.1 6.3.2 6.3.3 6.3.4 6.3.5 6. 3. 6 YE Addition Sterilisation Holding Section Regenerative Heat Exchanger Fermenters Pumps Otl)er Equipment 6.4 Distillation 6.4.1 6.4.2 6.4.3 Overview of Distillation Section Mass Balances Number of Theoretical Stages 6.4.3.1 Mccabe-'Ihiele Constructions and 116 119 121 123 125 126 127 128 133 134 134 137 Internal Flow Rates 146 6.4.3.2 Number of Transfer Units 169 6.4.4 Hydraulic Design and Column Diameter 6.4.4.1 Column Sl Design for 'l\lrbogrid and Sieve Plates 166 6.4.4.2 Column S2 Design for Tower Packing 173 6.4.5 Efficiencies and Tower Height 6.5 Post-Distillation 178 6.6 Equipment Summary and Costing 178 6.7 Overall Cost Breakdown 6.8 Discussion CHAPTER 7 GENERAL DISCUSSION AND CONCLUSIONS APPENDIX 1 WATER TOLERANCE OF METHANOL/PETROL MIXTURES WITH ADDED N-BUTANOL Al.l Introduction and Method Al.2 Results Al.3 Discussion APPENDIX 2 CELL COUNT ERROR ANALYSIS 183 189 195 197 199 199 A2.l The Theoretical Model 204 A2.2 A Practical Model 205 A2.3 Cell Enumeration for Regression Analyses 209 A2.4 Experimental Design 210 A2.5 Results and Discussion 216 APPENDIX 3 DETERMINATION OF a AND B A3.l Two Methods of Parameter Evaluation 217 A3.2 The Logistic Population Model 219 A3.3 Example of the Two Methods 221 A3.4 Inadequacies of the F.quations 227 A3.5 Discussion 234 APPENDIX 4 POPULATION DYNAMICS FOR LYSOGENIC VIRAL CULTURES A4.l Background A4.2 The Model A4.3 Solving the Model and Discussion A4.4 Conclusions BIBLIOGRAPHY 235 241 243 247 248 Variable a a a acet A Aa [AA] f A.B.E. [A] f A s b b.p. but B [BA] f BOD C , C CM CoA COD cw d.f. dg dl LIST OF VARIABLES AND ABBREVIATIONS Name a constant specific area of packing acetone area active area on the plate final total acetate concentration final, calculated, dissociated acetic acid concentration acetone butanol ethanol fermentation Units $ 2 -3 m m (g/1) (g/1) final acetone concentration (g/1) area free for gas flow m2 a constant $m 3 boiling point 0 c butanol bottoms flow final total butyrate concentration final, calculated, dissociated butyric acid concentration final n-butanol c,9ncentration Biological Oxygen Demand specific heat total cost of fermenters Cooked Meat Medium Coenzyme A Chemical Oxygen Demand cooling water degrees of freedom gas density liquid density mol s -l (g/1) (g/1) (g/1) ppm Jg-1 K-1 $ ppm x. D D' DC DE Dist DPC DR e ea equ. eth E [E] f f f2 a frac. F 9 FC FCI Fd Ferm FFAP FP GE GLC Glu h diameter distillate flow direct cost time taken between viral infection and cell burst distillation section direct product cost rate of death of active infected cells error m -1 mol s $ hr $pa human absolute error cells equation ethanol number of cells per square expected cells final ethanol concentration (g/1) fractional error human fractional error fraction fraction of cross sectional area open to gas flow fixed charges fixed capital investment ferridoxin fermentation section Free Fatty Ac id Phase feed plate position (numbered from the top of the column). general expenses gas-liquid chromatography d-glucose liquid hold up as a fraction of the total bed volume heat of combustion heat of vapourisation % Spa $ Spa -1 Jg Jg-1 xi. [HAA] f [HBA]f I IC !PC k K' LN m mol MC MEK M. 1 MLR final hydrogen ion concentration minimum hydrogen ion concentration observed final, calculated, undissociated acetic acid concentration final, calculated, undissociated butyric acid concentration (M) (M) (g/1) (g/1) height of a transfer unit rn water inhibition power constant inhibitor concentration g/1 indirect cost $ individual purchase cost of item updated to September 1980 $ growth rate estimated from the logistic equation a constant a constant a constant length log mean temperature difference log mean cell concentration mass flow rate slope of feed line for a McCabe-Thiele Construction mole slope of rectification section operating line manufacturing costs methyl ethyl ketone general term for a property Multiple Linear Regression -1 hr m K 10 6 /ml -1 kg hr $pa xii. M m MON Mr MU Ml5 M85 n no. N NT N.C.P. N s NTU NV 0 OD p pa pHf general term for a property of a mixture motor octane number molecular weight growth rate mixture of 15% (v/v) methanol with 85% (v/v) petrol mixture of 85% (v/v) methanol with 15% (v/v) petrol number of squares counted number of operating fermenters (excluding spare). cell concentration number of trays National Chemical Products Ltd cell population estimated by the logistic equation cell concentration imme­ diately after inoculation non-proteinaceous nitrogen average stationary cell population stationary phase cell population cell population smoothed using three point moving averages number of transfer units burst size observed cell population oven dried pressure per annum final broth pH g/mol -1 hr squares 10 6 /ml 10 6 /ml xiii. (X 10 6 cells/ml) -1 cell 10 6 /ml kPa PLB Po Prod PS q q Q rpm R RR s s ' t s.g • std. minimum observed pH purchased equipment cost wetted column pressure drop rate of conversion of lysogenic to virial cell dry column pressure drop product section success constant product concentration smoothed using three point $ Pa m -1 -1 ta -1 Pa m moving averages g/1 specific rate of production of product µg/1 hr cell ratio of heat required to vapour in the feed to a dis- tillation column to its heat of vapourisation average WP flow rate (15 m3 hr-1 ) -liquid flow rate gas flow rate ratio of fermenter filling rate to Q. revs per minute ratio of fermenter emptying rate to Q. minimum reflux ratio research octane number external reflux ratio maximum rate of butanol production RR= 0.8 (r-l + R'-l) absolute standard error shell and tube heat exchanger specific gravity standard mol mol -1 hr -1 s -1 s . -1 rev. m1n cells xiv. 2 s tt T TC! TLV TN u V V w we WP WSH WSR - X x. 1 y y* Ya y YALL YE YL variance of cell counts substrate concentration time doubling time for a healthy cell time smoothed using three point moving averages turnaround time temperature total capital investment threshold limit value total nitrogen overall heat transfer coefficient volume of single fermenter velocity total fermentation volume flooding velocity optimum velocity Total tank volume superficial velocity without downcomers heat flux working capital Sulphuric acid casein whey permeate Wood Sugar Hydrolysate Wood Sugar Hydrolysate Residue arithmetic mean composition cells 2 g/1 s hr hr hr K $ ppm kW m-l K-l m3 -1 m s m3 -1 m s -1 m s m3 -1 m s w $ mol. frac. composition (operating line) mol. frac. equilibrium composition mol. frac. distillate composition mol. frac. cone. of healthy cells 10 6 /ml YALL = Y + YY + YL 10 6 /ml yeast extract powder cone. of lysogenic cells xv. Yo yy YYV a B bottoms composition cone. of infected cells cone. of infected cells a period D before the time of interest 68% confidence level envelope standard change in free energy change in free energy a constant a constant growth rate integral sign voidage a constant mol. frac. 10 6 /ml 10 6 /ml cells J/mol J/mol -1 hr xvi. Figure 1.1 1. 2 1. 3 1. 4 1.5 1. 6 2.la 2.lb 2.lc 2.ld LIST OF FIGURES Title Uses for n-Butanol in 1927 Typical Seasonal flow of Whey in a New Zealand Dairy Factory Process Flow Diagram Progress of the Fermentation Distillation Plant Flow Diagram The Metabolic Pathway to the Solvents The 1.5-litre Fermenter in the Microferm The 10-litre Fermenter in the Microferm The Pressure Controller The 30-litre Fermenter 2.2a Schematic Diagram of the 1.5-litre Fermenter in the Microferm Unit 2.2b 2.2c 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 5.1 5.2 Schematic Diagram of the 10-litre Fermenter in the Microferm Unit Schematic Diagram of the 30-litre Fermenter Run A Run B Run C Run D Run E Run F Run G Run H Run I Run J Solvent Yield versus Pressure Solvent Yield versus pHm Possible Concentration Interrelationships Possible Concentration Interrelationships Detailed Biochemistry A Typical F~rmentation The Effect of H 2 Concentration on Free Energy pH Before and After Autoclaving Results of Kutzenoc and Aschner (1952) Declining Performance with Subculturing xvii. 3 8 13 13 13 16 24 24 25 25 26 27 28 41 42 43 44 45 46 47 48 49 50 57 58 64 78 80 83 92 96 115 118 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 Al A2.l A2.2 A2.3 A3.l A3.2 A3.3 A3.4 A3.5 A4 Plant Flowsheet Mole Balance Results for Sl Mole Balance Results for S2 Mole Balance Results for S3 Mole Balance Results for P and S4 Approximation for S2 Mole Balances Approximation for S2 for McCabe-Thiele McCabe-Thiele Construction for Sl V.L.E. Diagram for Acetone and Ethanol Graph for the Determination of K V Cales. Water Tolerance of Laboratory Petrol Blend Standard Deviation vs. Mean (Raw Data) Standard Error vs. Object Concentration Number of Squares Counted vs. Number of Cells µ, and q versus Time q versusµ Cell Population for Run A Scatter Diagram for the Logistic Equation Dimensionless Growth Curve Simulation of Population Fluctuations xviii. 124 139 140 141 142 149 156 157 163 174 201 207 211 216 225 225 228 230 230 245 Table 1.1 1.2 LIST OF TABLES Title Approximate Recent Costs Characteristics of Some Pure Liquids xix. 1.3 Typical Broth Concentrations at National Chemical Products 2 4 14 2.1 Average Seasonal Composition of Whey Permeate 18 2.2 Composition of Wood Sugar Hydrolysate Fractions 19 2.3 Composition of Combined Fractions 19 2.4 Wood Sugar Hydrolysate Residue Composition 20 2.5 Typical GLC Retention Times 32 3.1 Yield Data for 1.5-litre Experiments 37 3.2 Summary of Typical Microscopical Observation on Cl. acetobutylicum NCIB 2951 During Fermentations 51 3.3 Summary of 10-litre Runs 52 3.4 Summary of Calculated Variables for 10-litre Runs 53 3.5 Summary of Estimated Values for 10-litre Runs 54 3.6 Results of Multiple Linear Regression of Final Butanol Concentration with Headspace Pressure and Minimum Broth pH 59 3.7 Results of MLR on Final Acetone Concentration Final Ethanol Concentration and Final Total 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 Solvent Concentration Summary of Solvent Relationships with Headspace Pressure and Minimum Broth pH Correlations for Final Butanol Concentration Correlations for the Undissociated Butyric Acid Concentration Correlations for the Undissociated Acetic Acid Concentration and Total .Acetic .Acid Correlations for the Final Acetone Concentration Correlations for the Final Ethanol Concentration Correlations for the Stationary Phase Cell Population Correlations including the Specific Growth Rate 60 61 66 67 68 69 70 71 72 3.16 3.17 3.18 3.19 3.20 3.21 4 Correlations including the Specific Rate of Butanol Production Correlation Summary Table Levels of Significance for the T-Statistic Cell Population Data for Run A G 0 Values for Production of Various Chemicals from Pyruvate Summary of 30-litre Runs Summary of WSR Fermentations 5.1 Results of 10-litre Fermentations of Serial Transfer with Heat-Shocking 5.2 Cell Count Date in 10-litre Fermentation after the First Serial Transfer 6.1 Cost of Fermenters for Different Numbers of Fermenters 6.2 Summary of Vessels for Predistillation Section 6.3 Summary of Heat Exchangers for Predistillation Section Summary of Pumps for Predistillation Section Mole Balance for Sl xx. 6.4 6.5 6.6 6.7 6.8 6.9 6.10 Mole Balance for Top of S2 (Imaginary Column S2A) 6.11 6.12 6.13 6.14 6.15 6.16 6.17 6.18 Mole Balance over Columns S2 AND S4 Mole Balance for Co·lumn S3 Mole Balance over Column S4 Mole Balance over Column S4 and the Phase Separator (P) Summary of McCabe-Thiele Calculations Summary of Integral Calculations for NTU Calculations for Column S3 Summary of Column Heights Summary of Hydraulic Calculations for Column Summary of Hydraulic Calculations for Column AND S2B Summary of Hydraulic Calculations for Column Summary of Hydraulic Calculations for Column Summary of Fermentation Section Equipment and Costs Sl S2A S3 S4 73 75 77 84 91 105 110 117 120 132 135 135 136 147 148 150 151 152 155 164 165 167 168 170 171 179 6.19 6.20 6.21 6.22 6.23 6.24 6.25 6.26 Al.l Al.2 Al.3 A2.l A2.2 A3.l A3.2 A3.3 A4 Summary of Distillation Section F.quipment and Costs Summary of Product Storage Section Equipment Costs Summary of Combined Total Costs Capital Investment Costs Total Product Cost Breakeven Cost xxi. 180 and 181 182 184 185 186 Reduction in Fermenter Cost with Fermentation Time 191 Reduction in Breakeven Price with Fermenter Cost Base Blend Mixtures Water-Droplet Size Results Water Tolerance (% Dry Test Blend) at 15°C for Different Base Blends and Blending Agents Table of Regression Results Fitting a Modified Poisson Model Values of Fractional Error and ln e for Varying Cell Concentrations for Fixed nc or nc E Productivity Data Units for Table A3.l Results of Integration Method Program for ISIS Solution of Viral Population Model 191 198 200 202 206 213 222 223 229 245 1. 1. INTRODUCTION 1.1 PRELUDE Since 1973 there has been a rapid rise in the price of petroleum and derived products. This has led to increasing interest in the use of fermentations for production of various chemicals in the belief that such processes may become economic, as they were prior to the era of cheap oil. One such fermentation process is the now obs o lete acetone/ butanol/ ethanol (A. B. E.) fermentation. Imports of acetone, butanol, and ethanol into New Zealand are shown in Table 1.1. Historical uses of n-butanol are shown in Fig. 1.1 but the compound has been used in recent times as a cosurfactant/ coagent in tertiary oil recovery from ex isting wells (Compere and Griffith, 1979), and in s o l vents for nitro- cellulose lacquers (Cheremisinoff, 1980). The greatest incentive to recent research, however, has been for alternative fuels applications. Table 1. 2 com­ pares the fuel characteristics of some lower alcohols and acetone with gasoline. As a complete replacement fuel, the chief advantage of butanol o ver methanol or ethanol is its greater energy per unit mass ( 50% and 20% respectively). As a fuel extender, butanol has two very important advan­ tages over both methanol and ethanol. First, butanol blends in mixtures with petrol for spark ignition engines (as in the private vehicle fleet) have water tolerance levels greatly superior to those of methanol and ethanol blends. In fact, iso-butanol has been considered as a blending agent for blends of methanol with gasoline to in­ crease their water tolerance (Judd and Graham, 1979). Work done by the present author as a prelude to this thesis, has shown n-butanol to be equally effective (Appendix 1). A blend of 5% t-butanol in automotive fuel was once used 2. Table 1.1. Imports of Acetone, Butanol, and Ethanol. Solvent ketone Butanol Ethanol Approx. Import Volunea (litres) 1,338,283 361,133 4,382,137 Est. ex Factory Priceb ($/kg) 0.98 0.90 0.60 N. z. Retail Pricec ($/kg) 1.22 1.12 0.74 Approx 'lbtal Cost of Irr()orts 1,036,099 263,266 2,074,504 a: assume 50% total butanol imports are n-butanol b: 1978-79 figures c: based on estinated ex factory Refs: British Petroleun (1980) N. z. D:partnent of Statistics (1981) -- Fig. 1.1. Uses for n-Butan:)l in i927 (Anon, 1927). USES FOR BUTANOL Rubber AcceJerafOll'"r THIS (HART WAS {ONP!LEO FROM DATA. K11VOLYFuRN1SHED BY rHt CoMMERc,."L SoLvENTS (Ol?POIIA.TION, TEtUU /JAU TE, hvo. or~tS INOICATl l'R'ODUCTS; SQUARtJ,IJJ(J USE 5 FOR BUTANOl N.Hufvl (11/orieftt 8uh1rale fJfers Plosl,'cs lo,qu~n and DoMJ DEil/VA TIVES M,scl Oqe /11/.,-rmed,a/eJ Oulqronl' Jo/vent ir<'lher r,·, ,,.,e. /ndu1lr,n. elc. H,sd f:ul'tl"Olr.r falv,nrr w 4. Table 1. 2 CHARACTERISI'ICS OF SOME FURE LIOOIDS a Units Premium Methanol Ethanol n-Butanol Acetone Property Gasoline -1 32.04 46.07 74.12 58.08 Mr g rnol - s.g. - 0.73-0.74 0.793 0.789 0.810 0.792 b.p. oc 40-200 65 78.4 117 56.5 h -1 350 1,170 879.2 619.7 550.9 Jg V -1 h Jg 47,100 23,492 29,690 36,049 30,838 C Ratiob St. g/g 14.7 6.5 9.0 11.1 9.5 Cx:::tane Rating Rcrf - 96 llO 100 - - ~ - 91 87, 94 94 83 93 TLV ppm 250-700 200 1,000 100 1,000 a: See List of Variables and Abbreviations for explanation. b: St. Ratio =.Stoichiometric Ratio (air:fuel) :Molecular Weight of air= 0.79 x 28 + 0.21 x 32 = 28.9 g rno1-l c: Research Cx:::tane Number d: Motor Cx:::tane Number Refs: Judd (1979) Noon (1981) Perry and Chilton (1973) Thompson and Ceckler (1977) 'Weast (1977) 5. under the trade name ARCONOL by the Atlantic Richf iela company (Cheremisinoff, 1980). At this level there was no loss in 'driveability'. At higher concentrations (10% and 15%) however, some loss was apparent. ARCONOL also appear­ ed to have good carburettor anti-icing characteristics, octane performance, and water tolerance properties (Cheremisinoff, 1980). The Goodyear Tire and Rubber Co. have a patent on a t-butanol/gasoline/water blend reported to have cold weather properties superior to both methanol and ethanol blends (Cheremisinoff, 1980). In fact, up to 20% n-butanol can be aaaea to petrol with only very minor engine modifications (Noon, 1981). Miller et al. (1981) state that butanol's low latent heat of vap­ ourisation (h) relative to methanol and ethanol would V probably avoid cold start problems, its low vapour pressure avoid vapour lock problems, and its longer carbon chain reduce water tolerance problems. For example, a 20% n-butanol mixture does not phase separate with 1% aaaea water whereas a 20% ethanol mixture does (Miller et al., 19 81) • In aaa i tion, the present author has shown in ex­ periments similar to those of Appendix 1 that a blend of 20% ethanol with a laboratory petrol mixture has a much greater water tolerance than similar methanol blends (at either 20% or 8 5%) • This implies that the most stable blend is with butanol, followed by ethanol and then methanol. The second advantage of n-butanol as a fuel extender is that it can be mixed directly with diesel fuels for use in the compression combustion engines so extensively used in industry and transport. This is practicable to a concent­ ration of 30-40% (Miller et al., 1981; Noon, 1981), whereas both methanol and ethanol are very much more troublesome, requiring perhaps spraying the fuels separately into the cylinders (Cheremisinoff, 1980), or the creation of fine, stable emulsions by chemical means. Diesel mixtures are limited to just 30-40% to prevent engine knocking. 6. Slight cetane depression is experienced at these concentra­ tions but the problem is minor. The long term effect of the mixture on lubrication was not examined in either publication. Professor A.L. Titchener has outlined nine major problems with alcohol fuels (McEldowney, 1982) , namely, poor cold weather starting characteristics, low calorific values only partially offset by higher engine efficiencies, greater heats of vapour i sat ion, poor engine lubrication, attack of some engine materials, aldehyde emissions, toxicity, poor diesel substitution, and log i sties of introduction. Butanol would compare little worse than the lower alcohols on few of these, and is vastly superior in diesel applications which he suggested were the only major technical hurdles. In addition to traditional uses and these liquid fuel pos­ sibilities, resurgence of interest in solvent delignifica­ tion of wood (originally for pulp production) (Aronovsky and Gertner, 1936) could also give the A.B.E. fermentation a future in forest utilisation (U.S. Dept. Energy, 1979). Aronovsky and Gertner ( 1936) found aqueous n-butanol solu­ tions to be the best of a range of solvents for the delignification process. It is for these reasons as well as the savings on imports for industrial use that the present study was initiated. The two organisms Clostridium acetobutylicum and Cl. butylicum were used, as strains of these two organisms have proved to be amongst the best microorganisms for the fermentation. They are both motile, saccharolytic bacteria and are obligate anaerobes. These two organisms will be described in more depth in Sections 1.3 and 2.2. The choice of raw material remained. Traditionally, either starch or molasses has been used, but neither is particu­ larly cheap or readily available in New Zealand. 7. Wix and Woodbine ( 1958) mentioned the use of whey as a substrate for the A. B. E. fermentation. New Zealand dairy companies produce large volumes of whey and whey deprotein­ ated by ultrafiltration (whey permeate). The whey production over a year for a typical dairy factory is shown in Fig. 1. 2, and the eighty dairy factories in New Zealand would produce some 7, OOO mill ion 1 i tres annually. Maddox (1980) has shown that whey permeate is a suitable substrate for the A.B.E. fermentation. Due to recent growth in the New Zealand dairy industry this supply of whey and whey permeate will outgrow the traditional outlets, principally spray irrigation onto pasture and lactose production. Dumping to waste (into streams, waste treatment plants, or the sea) puts a vast load on the receiving system (the BOD of whey is 40-60,000 p.p.m. (Wix and Woodbine, 1958)), and represents a loss of potentially useful material. The surplus whey or whey permeate therefore represents a possible substrate. An attractive feature of using whey permeate as a substrate for the A.B.E. fermentation is its relatively low sugar concen­ tration (ea. 50g/l lactose) • This is ideal because product toxicity by n-butanol limits the n-butanol concentration in the spent broth to ea. 12g/l (Hastings, 1978) and total solvents to about 20 g/1, and thereby limits the sugar utilised to ea. 50-60g/l. This, in addition to the presence of cheap readily available whey permeate, made whey permeate an attractive substrate. However, another more abundant substrate would be required for very large volume production. The total production capacity from the whey may be estimated as prod. = (sugar conc.)x(conversion efficiency)x(whey volume) = 50 g/1 x 0.32 g/g x 7 x 10 9 1/yr x 10- 6 tonne/g = 110,000 tonnes/yr. where 0.32 g/g is the lactose to solvent conversion effi­ ciency (Section 3.3.3.9). This represents only ea. 7% of the petrol and ea .10% of the diesel supplied to the N. z. ,: .. 'O f ., E :, 0 > >­.. r:: 3: 600 JUN AUG OCT DEC FEB APR Fig. 1.2. Typical Seasonal Flow of Whey for a New Zealand Dairy Factory (Marshall, 1981). 8. 9. consumer in the early 1980' s. Delivery to the pump was 1.6 x 106 tonnes/yr for petrol and 1 x 10 6 tonnes/yr for diesel in 1980 and 1981 (N.Z. Dept. Statistics, 1982). So unless this rough calculation is out by a factor of 10, or more it can be seen that there is insufficient whey. The calculation for ethanol fermentation is virtual­ ly identical. Alternative fermentation substrates would have to include both wood hydrolysates and sugar beet. Leonard et al. (1947) showed that wood sugar hydrolysate could be used for the fermentation, although difficulties would be encountered. This possibility is considered further in Chapter 4. 1.2 HISTORY The history of the A.B.E. fermentation up to the 1950's is well documented by Wynkoop (1943), Prescott and Dunn (1959), and Ross (1961). At one stage it was second in importance only to ethanol production (Spivey, 1978). The fermentation was first noticed by Pasteur in 1861 (Ross, 19 61) . In 1911, Strange and Graham Limited (U.K.) employed Fernbach and Schoen of the Pasteur Institute and Perkin and Weizmann of Manchester University to study the butanol fermentation as a source of raw material for pos­ sible synthetic rubber production. Weizmann left their employ in 1912 and later isolated an organism capable of producing an acetone yield four times that of previous isolates (Compere and Griffith, 1979}. Unfortunately, ini­ tial hopes of synthetic rubber production from butanol were not fruitful, and it was not until World War I and the great demand for acetone for explosives and nitrocellulose wing dopes for aeroplanes that the first commercial plants were established. After the war the demand for acetone diminished and many plants were closed (Prescott and Dunn, 1959). 10. However, soon after World War I, butyl acetate was found to be an excellent solvent for the nitrocellulose lacquers used on cars and so the rapidly growing car industry made the process viable through n-butanol, which had previously been a waste product (Ross, 1961). Fig. 1.1 shows uses for n-butanol in 1927 (Anon, 1927). Production facilities grew rapidly between the wars and new uses for n-butanol were discovered in wood pulping (Aronovsky and Gertner, 1936), urea-formaldehyde resins and coatings, and as a plasticiser or solvent in other plastics (Wynkoop, 1943). After World War II, when the demand for acetone once again plummetted, the A.B.E. fermentation again decreased in importance. At this time a new synthetic chemicals industry based on ethanol from yeast fermentations began competing in the same market. Developments in the A.B.E. industry resulted in increased efficiency and early compe­ tition, but the synthetic chemical industry soon began to dominate. In an effort to survive, previous "waste­ products" were marketed (Hastings, 1978). Hydrogen gas was used for methanol manufacture and foodstuffs. Carbon dioxide was compressed and used as dry ice, and vitamin B 12 and other growth factors from the residual solids made economic the total evaporation of spent medium to protein rich animal food. Finally, however, simultaneous increases in the price of molasses for fermentation and the sudden availability of cheap petroleum as an alterna­ tive raw material for the synthetic chemicals industry meant that the A.B.E. process was no longer economic. Only in sugar- or molasses-producing "non-petroleum" countries such as some Eastern European countries, Egypt, and South Africa has the fermentation continued. The history to 1978 is recorded by Hastings (1978). 1.3 THE ORGANISMS Clostridium acetobutylicum is an anaerobic, spore-forming, 11. motile, rod-shaped bacterium. Bergey's Manual (Smith and Hobbs, 1974) provides a good description of the species, and describes it as Gram-positive and usually 4 µm x 1 µm. Spivey (1978) describes the species as Gram-positive or Gram-variable depending on culture age. Storage of the organism to prevent culture deterioration requires specific attention. Repeated subculturing of many butyric acid-forming bacteria is reported to result in serious culture deterioration, leading eventually to com­ plete loss of viability (Kutzenok and Aschner, 1952). Storage of Cl. acetobutylicum on sand or soil as a means of culture maintenance has proved successful (Spivey, 1978). Conversely, heat shocking of the spores soon after inocula­ tion has been frequently reported to result in faster fermentation rates and higher yields (Hastings, 1978; Prescott and Dunn, 1959; Ross, 1962; Spivey, 1978). Resistance of the organism to the toxic effects of the sol­ vents produced, in particular n-butanol, is of great importance as this is thought to be one of the limiting factors in the final sol vent yield. Inhibit ion occurs at concentrations as low as 10 g/1, and 12 g/1 or 13 g/1 n-butanol halts the fermentation (Hastings, 1978; Ross, 1961). Efforts to increase the resistance of the organism have met with only slight success (U.S. Dept Energy, 1980). Jerusalemskii seems to have been the only person to in­ crease the n-butanol tolerance significantly (Ross, 1961). He used continuous culture methods over a 200 day period to obtain a culture which could tolerate 25g/l n-butanol from a strain initially tolerating only 8g/l. He also found that the resistance was transferred through further generations. Clostridium butylicum is very similar to Clostridium acetobutylicum, and perhaps the only major difference which can be ascribed to strain variation is Cl. butylicum's inability to hydrolyse gelatine (Smith and Hobbs, 1974). 12 • . 1.4 THE PROCESS Beesch (1952), and Prescott and Dunn (1959) present good reviews of some fermentation processes prior to 1959. Spivey (1978) gives details of an A.B.E. fermentation plant currently in operation. Figs. 1.3, 1.4, and 1.5 show the process flow diagram, distillation system, and the pH, acid, solvent, and gas evolution curves for a typical batch fermentation at the National Chemical Products (N.C.P.) plant in South Africa (Spivey, 1978). Briefly, each pres­ sure fermenter of 90,000-litre working volume (constructed of stainless steel-clad mild steel) and associated piping is left under steam pressure while not in use (for steri­ lising), flushed with sterile co 2 (to prevent vacuum formation during cooling and to maintain anaerobic condi­ tions), and filled with medium. The medium (molasses) is diluted to about 60-65g/l fermentable sugar concentration in order that the normal 30% final yield on sugar would result in less than 20g/l total solvent. This is desirable as solvent toxicity would otherwise halt the fermentation prematurely. The medium is sterilised immediately prior to entering the fermenter using steam injection, a plate heat exchanger, and a holding tank to give retention time of 4 minutes at 128°C. The medium is cooled to 34°c and run into the fermenter. Oxygen is stripped from the medium by co2 agitation for 20 minutes before and after inocu­ lation. Finally, if required, the pH is adjusted to pH 5.8-6.0 using ammonia. The fermenter head-space pressure is then adjusted using co2 to 35 kPa and held for 5 hr., after which it is released to 18 kPa to allow for build up due to gas evolution by the organism. When 35 kPa is again attained, excess gas is drawn off. Gas production ceases abruptly after 30-34 hours and the fermentation ends. The beer is then sent for distillation. Typically, solvent yields are as shown in Table 1.3. About 50% (wt. basis) of the sugar goes to carbon dioxide, and 2% to hydrogen. I ' l ~·/~;. i ~~ i -1 __ _j r ---·----- --i l .,....,, .., ::.Sl ~Sl.,)'11 4 .,, j ----T-- __ L-, I .,.,. : ~~- ~ j + I ..,., .. ,..il.. \ YA !~ ,, + ~ I l :~~·;,, ,.:.,.t..T t ! ,....... ! -i--j I I I I _I 1"1 1 ,- ·--·+ r -• ' .. . ,, ...... .... :-. . ., l ! - "'i :::·,~-~·:·" i I ·• I L_ \ " "" i-1-1 1,,;:;;~:,,. ~ ----S lllt •t,JUS I I I I j Fig.1.3. Process Flow Diagram. pH 60 55 0 ,' / '.8 TIME 27 Fig. 1.4. Progress the Fermentation. Fig. 1.5. Distillation Plant Flow Diagram. tH IIYL Bll e \.",'El, S A :11 •u • u 11~1e,~ ] ...... !>Olvl!nt::. " SltWS I . BUIANOl . 1-j;(j l-, ! : L~[!VfH t--- CAl! Qf AC! l :..' NE RlLLIVI k _j r- • 13. pH V"l ,- z w > 0 V"l 36 of Table 1. 3 TYPICAL BROI'H CX>lONI'AATIONS AT NATIONAL CHEMICAL PRODOC'IS (Spivey, 1978) 14. Species Concentration (g/1) Initial Fermentable Sugar 63 Butanol 11 .Acetone 6 Ethanol 2 The total "turnaround time" per batch fermenter is 48 hours, and multiple vessels allow continual production. 15. Continuous fermentation has not yet been attempted on an industrial scale although some workers showed promising results in the laboratory (Ross, 1961). 1.5 THE BIOCHEMISTRY The organism converts sugar to pyruvate via the glycolytic pathway. The pyruvate is then reduced to acetate which can be used for ethanol production and/or condensed to aceto­ acetate and used for production of acetone and/or n-butanol. The basic fermentative pathways are shown in Fig. 1.6. The biochemistry is outlined in more detail in Section 3.3.3.5 (especially Fig. 3.15). Figure 1. 6 THE METABOLIC PATHWAY 'IO THE SOLVENTS (Coenzyme A forms omitted for convenience) Glucose (glycolysis) F 2 (NAtfl + H+) 2 x Pyruvate .Acetone C~CXXXXE ~co2 + "2 2 x .Acetate CH3CXXH ~ H20 .Acetoacetate c~=r:: 8-Hydroxybutyrate ~am~: Crotonate Butyrate C~CH2CH2CXXlf ~ Butyraldehyde NAO CH 3 CH 2 CH 2 CHJ 2 x .Acetaldehyde Cff:3COO K= NAD U , Butanol CH 3 cH 2 cH 2 CHl)f! 16. 17. 2. MATER~ALS AND METHODS 2.1 MATERIALS 2.1.1 Media Sulphuric acid casein whey permeate (WP) was obtained from the New Zealand Dairy Research Institute (Palmerston North, New Zealand). Its preparation is described by Matthews et al. (1978) and the average seasonal composition by Marshall (1981) (Table 2.1). Yeast Extract (YE) was purchased from Sigma Chemical Co. Ltd. (St. Louis, Missouri, u.s.A.). Cooked Meat Medium (CM) was obtained from Difeo Labora­ tories (Detroit, Michigan, U.S.A.). Wood Sugar Hydrolysate (WSH) fractions (pentose-rich prehy­ drolysate and hexose-rich hydrolysate), and WSH residue from the bottom of a beer still (stripper) after a yeast fermentation (WSR), were obtained from the Forest Research Institute (Rotorua, New Zealand). The compositions of the two hydrolysates are given in Tables 2.2 and 2.3, and for the residue in Table 2.4 (Mackie, 1982). For the preparation of the hydrolysate, the Pinus radiata wood feed was presteamed at 148°C for 35 minutes, then cooked at 160°C for 30 minutes in 6.7-litres of 0.5% e 2 so 4 per kg OD (oven dried) feed to produce the pre­ hydrolysate. Further cooking at 185°C for 140 minutes in 20-litres of 0.5% e 2 so 4 per kg OD feed produced the main hydrolysate. The final sugar yield was approximately 48% of OD feed (Mackie, 1982). 2 .1. 2 Sugars L-arabinose, D-glucose, and D-xylose were obtained from Characteristic 'Ibtal Solids lactose Protein ('IN-NPN) NPN Ash 3- F()4 Ca2+ 2- S04 Mg2+ Na+ K+ -Cl pH Table 2.1 AVERAGE SEASONAL cx:MPOSITION OF WHEY PERMEATE (Marshall, 1981) Concentration 56.4 46.0 0.3 0.32 7.9 2.0 1.4 2.8 0.11 0.50 1.4 0.9 4.5 - 4.6 18. (g/1) 19. Table 2.2 a::MIOSITION OF \mD SillAR HYDROLYSATE FRAC'TIONS (Mackie, 1982) Concentration Prehydrolysate Main Hydrolysate (Pentose Rich) (Hexose Rich) Total Solids 35 26 Suspended Solids 1.4 3.0 Arabinose 2.5 0 Xylose 5.9 0.9 Mannose 12.0 1.3 Galactose 3.4 0.3 Glucose 6.8 20.0 'Ibtal carbohydrate ca.32. O ca.25.5 Table 2.3 CXMFOSITION OF CXMBINED FRACTIONS (Mackie, 1982) Property Concentration (rng/1) Acidity 7,675 caco3 Sulphate 5,000 (as sulphuric acid) Metals: Mn 1 Fe 16 Ni 40 er 18 Cu 0.3 Phenols 0.76 Colour 1,900 (chloroplatinate units) CX>D 32,840 BOO 18,000 20. Table 2.4 vl:OD SffiAR HYDROLYSATE .RF.SIDUE CXMroSITION (Mackie, 1982) Species pH Acidity Sulphate Nitrogen Phospherous ('Ibtal) Calcium Phenols Metals 'Ibtal Solids Suspended Solids CX>D BOD Colour Xylose Arabinose Concentration (g/1) 4.5-5.0 1.25 caco3 (pH 8.3) 1.5 0.018 (Kjeldahl) 0.003 1.4 0.0005 (as for combined hydrolysate) 21.9 1.6 22 13 2,400 (chloroplatinate units) 4.3 2.2 21. B.D.H. Chemicals Ltd (Palrnerston North, New Zealand). Lactose, for use in culture media, was purchased from Wyngate (Hawera, N.Z.) and, for use as an analytical standard, from Sigma Chemical Co. Ltd. 2.1.3 Chromatographic Standards Butan-1-ol, butan-2-ol, ethanol, acetone, and acetic acid (UNIVAR grades) were obtained from Ajax Chemicals (Sydney, Australia). N-butyric acid (ANALAR grade) was purchased from B.D.H. Chemicals Limited. 2.1.4 Gases Hydrogen and oxygen-free nitrogen were purchased from New Zealand Industrial Gases Ltd., (Palmerston North, New Zealand) • 2.1.5 Other Chemicals Industrial grade ammonia solution was obtained from Andrew Chemical Co. Ltd. (Takapuna, New Zealand) , and f orrnal in from Laboratory Supply Pierce (N.Z.) Ltd. (Birkenhead South, New Zealand). 2.2 ORGANISMS Clostr id i urn acetobutyl icurn NCIB 2 9 51 was freeze dried specimen from the National Industrial Bacteria, Aberdeen, Scotland. purchased as a Collection of Clostridiurn butylicurn NRRL B-592 was purchased as a freeze dried specimen from the Northern Regional Research Centre, U.S. Dept. of Agriculture, Peoria, U.S.A. Both strains were reconstituted into 20 ml CM supplemented with 50g/l D-glucose contained in 25 ml screw-capped bottles. Soon after gassing ceased, the organism was inoculated D-glucose, incubated stored at 22. into 90 ml CM, supplemented with 50g/1 0 heat shocked at 75 C for one minute, and at 30°c. After sporulation the culture was 7°c. Stock cultures for regular use in inoculum preparation were prepared from this master stock into CM supplemented with 50g/1 of either D-glucose (for WSH media experiments) or lactose (for WP experiments). Cl. acetobutylicum NCIB 2951 was used for all experiments except those described in Section 4.2 and Chapter 5. Cl. butylicum NRRL B-592 was used for the experiments in Section 4.2 and Chapter 5. 2.3 MEDIUM AND EQUIPMENT STERILISATION All media of volumes less than 3-litres were sterilised in an autoclave at 120°C for 15 min. Greater volumes were sterilised in the autoclave at 120°C for 20 min. Gas filters, used to sterilise gases for fermentation, were filled with glass wool and heated in an oven at 160°C for ea. 3 hours. Electrodes for insertion into fermenters were sterilised in 5% (v/v) formalin solution, and rinsed with sterile distil­ led water immediately before use. 2.4 INOCULUM PREPARATION Unless otherwise noted, inocula were prepared by introduc­ ing 1 ml of stock culture into 100 ml of CM supplemented with 50g/l lactose. After incubation for two days at 30°C in an anaerobic jar (Gas Pak BBL 70304 for carbon dioxide and hydrogen production) fermentations were inoculated either by pipette, or by pouring directly into the ferm­ enter allowing a minimum of solids to leave the inoculum bottle. 23. 2.5 FERMENTATION EQUIPMENT AND METHODS 2.5.1 100 ml Scale Fermentations 100 ml nominal volume experiments were performed in 120 ml screw-capped "medicine bottles" containing ea. 90 ml of the required medium and adjusted to pH 6.5 with aqueous ammonia (SN) prior to autoclaving. After sterilisation and cool­ ing, 1 ml of inoculum was added and the bottles were incubated at 30°C. Approximately 8 ml of sample was aseptically withdrawn by pipette as required. Cell numbers and culture pH were determined immediately, and the sample stored at -20°C prior to chemical analysis. 2. 5. 2 1.5-litre Fermentations 1.5-litre nominal volume fermentations were performed in a Microferm Laboratory Fermenter (New Brunswick Scientific Co. Ltd, New Jersey, U.S.A.) using a 2-litre glass vessel containing 1.5-litre of medium. Continuous measurement and one-way control of pH was performed by a Horizon Model 5997-20 pH Controller (Horizon Ecology Co., Chicago, USA) coupled to an EIL 33 1070 030 toughened glass electrode and an EIL 33 1320 210 laboratory sealed reference elec­ trode. Ammonia (SN) flow, when called by the controller, was stopped every 29 sec. by a cam-timer to prevent gross over-correction. The ammonia was gravity fed from a dropping funnel through a Fluorocarbon Delta model DV2 122A2 solenoid valve (Delta Solenoid Valves, Anaheim, California, USA). Fiq. 2.la is a photograph of the Microferm unit, with ferm­ enter in place, and recording equipment. Fig. 2.2a is a diagram of the same. Immediately after sterilisation, the fermenter and contents rig. 2 . 1~. The 1. 5- l itre Fermenter 1n Lh o M1crofc r m. Fig . 2 . lb . Th e 10-litre Fe rmenter in the Microferm. 2 4 . 2S . fig. 2 . lc. The Pressure Controller . Fi g . 2.ld. Th e JO - litre Ferme nter . 1 @ l D 4 I 5 I 2 1. 1.5-litre fermentation vessel. 2. Microferm support unit. 3. Dropping funnel for ammonia. 4. pH controller. 5. Cam timer. 6. Solenoid valve for pH control. 7. Agitator drive. 8. Gas filter. 9. Gas flow measurement and control. 10. Agitation control and rev. counter. , , ..L ..L. 12. 13. 14. 15. 16. 17. Ternper;::a+1,rt:) ~,...,.T'lt-rfil ~ Main switch, heating, cooling, switches and indicator lights. Retort stands. Pressure control (not used). Antifoam pump (not used). Antifoam controls (not used). E measurement (not used). C and agitator Fig 2.2a. Schematic Diagram of 1.5-litre Fermentation Vessel in the Microferm Unit. 2 6. . 1 2 0 7 oo 7 7 oO I ' 1. 17-litre 316 stainless steel pressure fermenter. 2. Microferm unit (see Fig 2.2a.). 3. Drive unit in raised position. 4. Pressure gauge. 5. Pressure control valve. 6. Valve for pressure release. 7. Openings for continuous fermentation (not used). 3. Wire bands(to restrain plugs under pressure). Fig 2.2b. Schematic Diagram of Microferm Unit and Pressure Fermenter. 27. 8 1 1. 30-litre jacketed pressure fermenter. 2. Automatic temperature controller (controls steam,hot water, or cold water as selected). 3. Pressure safety valve. 4. Vacuum relief valve. 5. Pressure gauge. 6. Pressure relief and free-steaming valve. 7. Sampling valve. 8. Motor and variable speed gear box for agitation (used only for sterilisation and cooling). 9. Draining valve. F~g. 2.2c. Schematic Diagram of 30-litre Industrial Style Fermenter. 28. 29. were placed onto the Microferm unit, electrodes were inserted, gas flow was started (either sparging or sweeping across the broth surface as required), and temperature control (30 °C) and agitation were switched on while the fermenter was still hot (ea. 90°C). The system was then left for ea. 1 hour and allowed to equilibrate to pH 6.5. About 150 ml of inoculum were then poured directly into the fermenter (attempting to retain as much of the inoculum solids in the bottle as possible) and the conditions were set for that particular experiment. Broth samples were taken as required by blocking off gas outlets and forcing liquid out using gas pressure. About 10 ml were drawn to clear the sample line, and 8-10 ml of sample were taken. A cell count was performed immediate­ ly, and the sample was stored at -20°C prior to chemical analysis. 2.5.3 10-litre F~rmentations 10-1 i tre nominal volume fermentations were conducted with 10-litres of the required medium in a 17-litre, 316 stain­ less steel vessel built in the workshop of the Biotech­ nology Department of Massey University, and capable of holding about 170 kPa. Figs. 2.1 (b) and 2. 2 (b) respec­ tively are a photograph and diagram of the fermenter. The fermenter did not have an agitator. The assembled, vented, fermenter was sterilised by autoclaving, and cooled in cold water maintaining a positive pressure of ea. 50 kPa oxygen­ free nitrogen in the head space to prevent formation of a vacuum and to maintain anaerobic conditions in the fermenter. The fermenter was then placed in the Microferm unit for operation. Pressure control was achieved using a modified domestic hot water pressure valve vented to atmosphere by bubbling through water. Unfortunately the valve leaked and so gas flow rate measurements were not accurate. The mod if ica- 30. tions required to allow adjustment of pressure consisted of providing a tapping for a threaded shaft which pressed onto the rubber diaphragm. Fig. 2.l(c) shows a photograph of the controller. Samples were taken as required. About 20 ml was drawn to clear the sample lines, followed by ea. 8 ml of sample. If there was insufficient pressure to expel the culture, the sample was drawn into a sealed flask using a slight vacuum. Cell concentration, pH, and temperature were measured and recorded immediately on sampling, and samples were then stored at -20°C prior to chemical analysis. 2.5.4 30-litre Fermentations 30-li tre nominal volume fermentations were conducted in a ea. 50-li tre vessel registered for 140 kPa broth pressure and 170 kPa jacket pressure, and manufactured by Burns and Ferrall Limited (Auckland, New Zealand). Figs. 2. lc and 2. 2c respectively are a photograph and a diagram of the fermenter. Sterilisation was achieved by passing steam through the jacket and controlling the pressure to hold the broth temp­ erature at 120°C for 15 minutes. Cooling was then achieved by passing cold water through the jacket. When the temper­ ature fell below 100°C, oxygen-free nitrogen was introduced to prevent formation of a vacuum and to maintain anaerobic conditions. Subsequent temperature control throughout the fermentation was by means of hot water passing automatic­ ally through the jacket when the temperature fell below 30°C. The fermenter was inoculated with 150 ml culture and the gas pressure was allowed to build up to that required. Sampling and pressure control was as for the 10-litre fermentations. 31. 2.6 ANALYTICAL METHODS 2.6.1 pH Measurement Routine laboratory pH measurements were performed using a Metrohm Herisau pH-Meter E520 (Switzerland). 2.6.2 Cell Counts These were performed immediately after sampling using a standard haemocytometer under 400X magnification. Only vegetative cells were counted. 2.6.3 Solvents and Fatty Acids Solvents and low molecular weight fatty acid concentrations were determined by gas-liquid chromatography (GLC) using a Shimadzu gas chromatograph (GC - SA). A temperature prog­ ramming unit (TP - 3A), electrometer (EM - 55), control panel (FVS - 5), basic unit (BAS - 5), and Shimadzu chart recorder were also used. A programmable integrator (Varian CDS 111 (Chromatography Data System) performed the internal standard calculations. The column (2m x 3mm ID) consisted of 10% FFAP (Free Fatty Acid Phase) on a 100/120 mesh Chromosorb G AWDMCS support. A nitrogen carrier gas flow rate of 80 ml min-l was used, and the temperature programme held the column at 87°C for 7 minutes, followed by an increase to 180°C at a rate of 20°C . -1 m1n Typical retention times are shown in Table 2.5. Sample (1 ml) was added to an equal volume of internal standard solution containing 1% (v/v) butan-2-ol and either 2 0% ( v /v) formic acid or 10% ( v /v) orthophosphor ic acid. The sample was then centrifuged prior to injection (ea. 5 µl). 32. Table 2.5 TYPICAL GI.C RETENI'ION TIMES Compound !Etention Time (minutes) acetone 1.4 ethanol 2.3 butan-2-ol 3.6 butan-1-ol 7.3 acetic kid 13.4 butyric kid 15.6 33. 2.6.4 Lactose A Yellow Springs Sugar Autoanalyser (YSI Model 27) was used for lactose analysis. The samples were centrifuged to remove particulate matter and diluted as appropriate using distilled water prior to analysis. 34. 3. PRODUCTION OF n-BUTANOL FROM WHEY PERMEATE WITH DIFFERENT HEADSPACE PRESSURES 3.1 INTRODUCTION Maddox (1980) showed sulphuric acid casein whey permeate to be a suitable substrate for the A.B.E. fermentation on a 100 ml scale. The purpose of the work in this chapter was to scale up the fermentation to laboratory scale ferm­ enters and to identify important fermentation variables, in particular the effect of headspace pressure. Spivey (1979) stated that some of the hydrogen gas produced by the organism during fermentation was utilised during the subsequent production of butanol, and that the NCP process was controlled to a minimum pH of pH 5. 5. Mcinerney and Bryant (1982) also mention the effect of the partial pressure of hydrogen, but in relation to methane produc­ t ion. Hence, the effect of head space pressure and the controlling of pH to a minimum of pH 5.5 on the A.B.E. fermentation were studied. First, the effect of composition of sparging gas and pH control to pH 5. 5 on 1. 5-litre atmospheric pressure fermentations followed by the effect of head space gas pressure on a 10-li tre scale fermentation was examined. Secondly, the fermentation was attempted in a 30-litre industrial-style pressure fermenter. Due to the work of Maddox (1980), attention was paid almost entirely to the pressure effect and the fermentation itself rather than the effect of substrate, noting only that the permeate with 5 g/1 added YE gave healthy fermentation. At no stage was any attempt made to optimise the fermentation with respect to medium, inoculum preparation, or fermentation temperature. During analysis of the fermentation results it was noticed that the minimum observed pH as well as pressure was well correlated with yield of the butanol. This result was quantified using multiple linear regression. It was then found that use of multiple linear regression . to find 35. correlations between concentrations of species measured at the end of the fermentation structured these metabolites into a system almost identical to the organism's metabolic pathways. This approach, which was named the Factor Correlation Method, appears novel when used to this extent, and could well be useful in future work on metabolic pathway elucidation. / 36. 3.2 1.5-litre EXPERIMENTS 3.2.1 Results The medium used for all fermentations was sulphuric acid casein whey permeate supplemented with Sg/1 yeast extract, and adjusted to pH 6.5 using SN NH 4 0H prior to auto­ claving. The organism used was Cl. acetobutylicum NCIB 2951. The fermentations were conducted with 1. 5-litre of broth as outlined in Section 2.5.2. Three systems of gas flushing were used in order to vary the level of hydrogen in solution, and pH was either con­ trolled to a minimum of pH 5. 5 or not controlled at all. Sparg ing with oxygen-free nitrogen ( 200 ml/min) with vigorous agitation ( 200 rpm) was intended to remove both hydrogen and carbon dioxide from solution and supply the lowest level of dissolved hydrogen (Gas Level A). An intermediate level of hydrogen was provided by swe eping oxygen-free nitrogen ( 50 ml/min) across the medium surface without agitation, allowing the levels of dissolved hydrogen and co 2 to be controlled by the composition of the fermentation gases (ea. 1:1 H 2 :co 2 (volume basis)) (Gas Level B). Hydrogen sparging (200 ml/min) with agitation ( 2 00 rpm) was used to strip all the CO 2 from solution, but maintain the hydrogen as close to saturation as possible (Gas Level C). This last provided the highest hydrogen concentration. The results of the fermentations were similar to those shown in Fig. 3.16. (page 83.) for a typical fermentation. It shows the typical fall in pH as acids and cells are produced, followed by a rise in pH and temporary "dip" in the acid levels as solvent production begins. The order in which the experiments were done along with the final butanol and cell concentrations are given in Table 3.1. 37. Table 3.1 YIEID fil>.TA FOR 1. 5-LITRE EXPERIMENTS Gas Levels pH Control Species Units A B C Butanol g/1 - 2.4 0.4,0.7** minimum r..og 10 (cells) log cells 8.38 9.5,9.1** -pH 5.5 Chronological - - 2 4,5** Order Butanol g/1 3.4 5.3 5.3 no control r..og 10 (cells) log cells 8.9 8.2 8.8 Order - 6 1 3 -1 *Gas Level A :sparge with oxygen-free nitrogen (200 ml rnin ) :agitate (200 rprn) Gas Level B :sweep oxygen-free nitrogen through head space :no agitation Gas Level C :sparge with hydrogen (200 ml rnin-1) : agitate ( 2 00 rprn) ** One set of duplicates was run. -1 (50ml rnin ) 38. 3.2.2 Discussion Consideration of Table 3.1 clearly shows that controlling pH to not less than pH 5. 5 reduces the final yield of butanol compared to fermentations without pH control. It is quite possible that controlling the pH to a lower value would show an improvement (Andersch et al., 1982), but since encouraging yields were obtained without any pH control it was decided to perform all subsequent fermentations at the "natural" pH. In the case of Gas Level B with pH control the broth pH fluctuated greatly. This fluctuation was caused by lack of mixing in the fermentation broth, and it is likely that some regions in the broth had a pH below pH 5. 5. This possibility may well account for the higher yield of butanol in this run (Andersch et al., 1982). The three runs without pH control suggest that hydrogen does have an effect on the final yield of butanol. Gas Level A showed the lowest butanol yield, while the runs at Gas Levels Band C returned the highest yields. However, if the concentration of dissolved hydrogen were to effect butanol yields, it would be expected that Gas Level C show a yield superior to B. It is possible that the lack of co 2 at Gas Level C caused this effect. This possibility is discussed further in Section 3. 3. 3J.l in reference to a 10-litre run with reduced carbon dioxide concentrations, and in Section 7.2. Alternatively, experimental error may be masking a difference. The exact cause could not be conclusively determined. The possibility that an overall time trend produced the variation in butanol yield was discounted due to high yields in both the first and last runs, with the lowest yields ocurring in between (refer Table 3 .1) • A possible mechanism by which dissolved hydrogen could 39. effect the fermentation is via the equilibrium + + NADH + H ~ NAD + H 2 (Mcinerney and Bryant, 1981} where higher levels of H2 would act in accordance with Le Chatelier' s principle to supply the bacteria with a better supply of NADH. This NADH would then be available for oxidation by the reduction reactions, such as alcohol production. It was decided that the most convenient method of increas­ ing the concentration of hydrogen in solution would be to trap the fermentation gases (H 2 and co 2} in the vessel headspace, so that the headspace pressure would build up. This would then cause an increased concentration of hydrogen in solution. There seemed to be no advantage in attempting to create an atmosphere of pure hydrogen and so the natural fermentation gases were used. Thus, a sealed fermentation pressure vessel was constructed of 316 stain­ less steel for use in subsequent experiments. Pressure control was achieved using a domestic hot water supply pressure relief valve modified to allow variable pressure control. It is interesting to compare the results of Table 3.1 with those of Maddox (1980} who achieved butanol yields in excess of 10 g/1. His work was done on a 100 ml scale in screw-capped bottles which allowed the headspace pressure to increase during fermentation. It might therefore be postulated that increased headspace pressure caused an in­ crease in the concentration of dissolved hydrogen, and was responsible for the higher butanol yields. Mention of a possible effect of co 2 is made later in Section 3.3.3.11. It was thought that the effect of the H+ ion concent­ ration (i.e. pH}, which appears also in the NAD/NADH equilibrium, would be more important in its effect on enzyme and other cellular activity than on the NADH level, the former swamping any possible effect of the last. 40. 3.3 10-litre SCALE EXPERIMENTS 3.3.1 Results Whey permeate (10-1 i tres} supplemented with Sg/1 YE was fermented in a 10-li tre stainless steel pressure vessel with temperature controlled to 30°C. The inoculum was 150 ml of Cl. acetobutylicum NCIB 2951 in CM supplemented with 50 g/1 lactose. The maximum headspace pressure allowed to accumulate during the fermentation was the only variable altered, with the exception of Run I which had a pressure of 101 kPa pure H2 from inoculation. Figs. 3.1 - 3.10 describe the course of the fermentations. The pH after autoclaving started near pH 5. 5 and dropped to between pH 4. 2 and pH 4. 7 as acids and cells were produced. Once solvent production began, the pH rose slightly in spite of an overall increase in acid levels. Only in Run I did the pH show a significant rise after inoculation. Data for times greater than 100 hr. are drawn at 100 hr. Table 3. 2 describes typical microscopical observations of the organism during the fermentations. These observations showea no real variation between different fermentations. Table 3.3 summarises for each fermentation the maximum pressure (p), the average stationary phase cell concentra­ tion (Ns}' final butyric acid ([BA]f}, acetic acid ( [AA] f} , butanol ([BJ f} , acetone ( [A] f} , and ethanol ( [E]'f} concentrations, and also the final pH (pHf} and minimum observed pH (pH} for these runs. m Table 3.4 shows variables calculated directly from the experimental discussion values of Table section. Table 3. 3 and used 3.5 shows some in the results predicted by correlations of the data in Tables 3. 3 and 3.4, and is included here as it gives the quantitative cont. p. 55/ 41. Fi9:. 3.1. Fenrentation Results • butarol for Run A . £acetone 0 ethanol 6 pressure Y butyric acid \] acetic acid • log (cell count) - 0 pH ro ~ 20 - ,-j- I 0 ,-j X 15 - ~ Ul tl) [ 10 .. -,-j ~ .s 5 ~ • ,-j • • @ 0 10 20 30 40 50 60 70 80 90 100 10 •••• • . • • • . . :a • • 8 .. • - ,,· . - .+J § 8 ,-j 6 ~ ,-j ~ 0 - 0'\JB 0 ,-j 0 0 8' ,-j 4 ~ .. \] -,-j ~ • ,gj 2 . ~· • ·r-1 {) ro \] 0 10 20 30 40 $0 60 70 80 90 100 tirre since inoculation (hr.). Ferm211tation Results 42. Fig. 3.2. • butaool for Run B . .& acetone 0 ethanol L:lpressure Y butyric acid '\] acetic acid • log(cell count) - 0 pH ~ ~ 20 - r-l- I 0 r-l 2S. 15 ~ Cl) Ul ~ 10 0. .. -r-l ~ 11 5 ~ r-l • @ 0 10 20 30 40 50 60 70 80 90 100 10 • • • • . n. • 8 ".iJ' ~ ,, r-l r-l 6 li 0 r-l 8' 0 0 00 ,-t 4 .. -r-l § ,¥3 2 · ,, .'\] ·r-i ,, u ~ ,, 'v 'v '\] \] 0 10 20 30 40 so 60 70 80 90 100 tirre since inoculation (hr. ) . Fig:. 3. 3. Ferrrentation Results 43. • butaml for Run C . &acetone O ethanol 6pressure y butyric acid \] acetic acid • log (cell rount) - 0 pH ell ~ 20 - ,.....j- I 0 ,.....j :,<: 15 - ; en en ~ 10 0. .. -,-j • • ~ • • jS 5 ! !~ • @ .A 0 10 20 30 40 50 60 70 80 90 100 10 • • • • • . • ~ 8 .. '.µ' ~ ,-j ,.....j 6 ~ ,, -0 0 00 0 • r-i 0 ,§' 4 ,, ,, .. ,, -,-j "' 'v - \] ~ 2 ·.-t CJ ell 0 10 20 30 40 :i 0 60 70 80 90 100 time since inoculation (hr. ) . 44. Fi~. 3.4. Fernentation Results • butarol for Run D . A acetone 0 ethanol ~pressure 9 butyric acid '\] acetic acid • log(cell rount) - 0 pH (lj ~ 20 - ~ I 0 r-i X 15 C::, - ~ ~ • tll tll ~ 10 0.. • .. -r-i • ~ • 21 5 1 @ 0 10 20 30 40 50 60 70 80 90 100 10 • • • • • . t 8 ".µ § 8 r-i .-l 6 ~ • 0 0 .-l 0 ~ 0 8' 0 • r-i 4 • • .. -...-1 ~ ~ 2 • • 'v •.-t '\] CJ (lj '\] '\] 0 10 20 30 40 50 60 70 80 90 100 tirre since inoculation (hr. ) . Fig. 3.5. Ferrrentation Results 45. • butaool for Run E . A acetone 0 ethanol I:::::. pressure Y butyric acid 'v acetic acid • log(cell rount) - 0 pH ~ 20 - M- I C::. 0 M C::. ~ 15 C::, C::. ~ U) Ul ~ 0. 10 • .. • -r-i ~ 21 5 • ~ r-i g 0 10 20 30 40 50 60 70 80 90 100 10 • • ' • • . =a 8 .. '.µ' • ~ • M r-i 6 ~ 0 ,, -0 M 0 ,g 4 'v • .. -r-i ~ ~ 2 •,-f 'v t) ltj 'v 0 • 10 20 30 40 50 60 70 80 90 100 tirre since inoculation (hr. ) . Fig. 3.6. Ferrrentation Results 46. for Run F . • butaool & acetone 0 ethanol 6 pressure 9 butyric acid V acetic acid • log(cell count) - 0 pH I'd ~ 20 --r-1 I 0 & 66 r-1 66 2S. 15 ~ 6 Cll Cll ~ 10 0.. ~ -...-1 ~ • • • B 5 • 1 • @ ~ 0 10 20 30 40 50 60 70 80 90 100 10 • • • . • • . ~ 8 ~ § 8 r-1 r-1 6 ~ 0 0 0 r-1 0 ~ 4 ~ ,, - ~ r-1 ~ "'~ \] ~ 2 f •,-f C) 113 0 10 20 30 40 50 60 70 80 90 100 tine since inoculation (hr. ) • 47. Fi9:. 3.7. Fermentation Results • butarol for Run G. A acetone 0 ethanol 6 pressure Y butyric acid \] acetic acid • log(cell oount) - 0 pH rcl ~ 20 00 66 --.-i I 0 .-i ~ 15 - ; U) U) ~ 10 0.. • .-i • ~ • ll 5 • ~ .-i @ 0 10 20 30 40 so 60 70 80 90 100 10 • • • ,, • . • :a 8 '.µ' g .-i 6 .-i ~ 0 0 0 0 .-i g 0 4 .. -.-i ~ Q'\J ~~ - .gj 2 ii ,, ·r-1 • u w ro 0 10 2·0 30 40 so 60 70 80 90 100 tin'e since inoculation (hr. ) . 4 8. Fig. 3.8. Fenrentation Results • buta!Dl for Run H • & acetone O ethanol C).pressure 9 but-yric acid \) acetic acid • log(cell rount) -- 0 pH C13 ~ ~ 20 ~ ,-;-- I 0 ,-; >: 15 - ~ U) U) ~ 10 0.. .. --,-; ~ • - • • 21 5 ~ ,-; #. @ ~ 0 10 20 30 40 50 60 70 80 90 100 10 • • • . • :a 8 :jj' • ~ .. ,-; 6 ,-; 0 ~ 0 ,-; 0 0 0 8' 0 ,-; 4 .. -,-; ~ 'W ,gJ 2 ·r-l (.) C13 0 10 20 30 40 50 60 70 80 90 100 tirre since inoculation (hr.) • --:1l 1 butaml acetone 0 ethanol 6 pressure Y butyric acid '\] acetic acid Fig. 3. 9. Ferrrentation Results 4 9 · for Run I. note: 'Ihis fenrentation had lOlkPa µire H applied at inoc. 2 log(cell count) 0 pH ~ 20 D. ,-f-- 1 0 ....-t . =a 0 ....-t . 8' ....-t -r-1 ~ ........ 15 10 • 5 • • • • • 0 10 10 20 30 40 50 60 70 80 90 100 8 • • 4 2 'v o#J • • • • ' •• E) 0 0 • • • 10 20 30 40 so 60 70 80 90 100 time since inoculation (hr.). so. Fig. 3.10. Ferrrentation Results • butanol for Run J. A acetone O ethanol ~ pressure W butyric acid \] acetic acid • log(cell rount) °'°' - 0 pH °' co ~ 20 - .....r I 0 r-t :< 15 - ; • ll) • • ll) ~ 10 c.. • .. -r-i ~ • .s 5 ~ .~ r-t ~ @ A 0 10 20 30 40 50 60 70 80 90 100 10 • • ' ' • . n. • 8 ' :i} • § w 8 w r-i 6 r-i 00~ ~ 0 0 r-t 0 ~ 4 • V} I .. -r-i ~ ~ 2 ...... •• t) ~ \] 0 10 20 30 40 so 60 70 80 90 100 tirre since inoculation (hr.). Time (hrs) 0 0.25 1.5 4 8 10 17 20 23 33 55 130 Table 3.2 StMMARY OF TYPICAL MICEOSCOPICAL Cl3SERVATION OF Cl. ACE'IOBUTYLIC!M N.:IB 2951 DURING Fm-1EN.I'ATI0NS l.el:l3th Motility (µrn) 1.5 - 2 some slightly motile 0.25 - 0.5 15-20% slightly motile 2 - 4* brownian motion ea. 0.1* extremely motile (ea. 5-10% population) 1.5 - 2.5 most motile 1.5- 2.5 all well motile 1 - 1.5 all well motile 1 - 1.5 less motility 1 - 1.5 stationary, wigglil:l3 l - 1.5 almost non-motile 1 - 2 non-motile 1 - 3 non-motile * 'Ihere were two distinctly different populations present. Cne large and non-motile, the other very small and extremely motile. 51. 52. Table 3.3 S'C.MMARY OF 10-LITRE Rut£ Run p pHrn ?if N [BJ f [A]f [E] f [BA] f [AA] f s A 101 4.5 4.63 1093 3.40 0.50 0.46 2.41 2.10 B 101 4.2 4.35 1068 0.92 0.27 0.28 - - C 101 4.6 4.85 1373 6.75 1.34 0.55 5.21 3.25 D 140 4.6 4.6 1977 10.5 0.45 0.82 4.82 2.62 E 157 - 4.6 994 9.40 0.67 1.11 4.24 2.89 F 160 4.65 4.65 1875 6.15 0.93 0.46 3.37 2.73 G 201 4.45 4.85 1685 7.40 0.67 0.51 2.89 2.73 H 201 4.5 4.65 1110 8.20 1.95 1.05 1.16 2.31 I 207 4.6 4.65 1183 4.60 0.47 0.27 3.66 3.15 J 250 4.65 4.75 1317 11.6 0.99 1.80 5.98 3.57 s.e. - - - 26% 48% 61% 65% 40% 16% 53. Table 3.4 SUMMARY OF CALCULATED VARIABLES FOR 10-LITRE RUNS Run p [H+] [H+] log N [HBA]f [BA-]f [HAA]f [AA-]f J.l R' m f s xl0- 5 ) (xl0- 5) (xl0- 9 ) A 101 3.16 2.37 9.0386 1.47 0.93 1.20 0.90 0.33 0.091 B 101 6.31 4.47 9.0284 - - - - - 0.037 C 101 2.51 1.41 9.1377 2.53 2.67 1.44 1.81 - 0.233 D 140 2.51 2.51 9.2960 3.02 1.79 1.53 1.09 - 0.101 E 157 - 2.51 8.9972 2.65 1.57 1.69 1.20 0.28 0.121 F 160 2.24 2.24 9.2730 2.02 1.35 1.52 1.21 0.33 0.101 G 202 3.55 1.41 9.2266 1.40 1.48 1.21 1.52 0.16 0.142 H 202 3.16 2.24 9.0453 0.69 0.46 1.28 1.02 0.18 0.135 I 208 2.51 2.24 9.0730 2.19 1.47 1.75 1.40 0.25 - J 250 2.23 1.78 9.1196 3.24 2.72 1.78 1.78 0.42 0.137 s.e. - - - 1.2% 39% 46% 15% 25% 33% 43% RUN A BC C D Ed F G H Ie J s.e.e. f Notes: Table 3.5 SCME ESTIMATED VALUES (From Correlations in Table 3.7} 54. [BJ a f [BJ b f [HBAJ a f [HBAJ b f [HAA] a f 4.8 4.3 1.22 1.19 1.19 - - - - - 6.6 5.7 2.06 2.02 1.43 8.7 7.3 2.39 2.21 1.48 8.6 - 2.93 - - 7.6 8.7 2.35 2.64 1.60 7.8 7.4 1.26 1.19 1.19 6.5 8.3 1.51 1. 70 1.33 9.3 10.0 3.15 2.54 1.57 12.5 12.3 3.26 3.11 1.74 15% 23% 23% 27% 5% a estimated from measured data b estimated from estimated data c concentrations for Run B could not be estimated as no experimental acid concentrations were available d concentrations for Run E could not be estimated as no reliable pH was available m e Run I had different experimental conditions, the effects of which are emphasised by these estimates f the standard error of the estimated variable compared with the measured or calculated variable (Tables 2.3a or c} divided by the mean variable and expressed as a percentage (s.e.e.} (contd. from p.40) 55. result of the text and correlations of Section 3. 3 (see especially Table 3.17). Table 3.5 may be compared directly with Tables 3.3 and 3.4. Some acid data for Run B were not calculated as levels rose so severely late in the fermentation. was evaluated over the apparent stationary phase. 3.3.2 Analysis of Data 3.3.2.1 Method of Data Analysis acid N s In order that any relationships be clearly established, the data of Table 3.3 were correlated using multiple linear regression on the Massey University PRIME 750 computer (PRIME Computer Inc., U.S.A.). The software package used was the Mini tab package developed by Pennsylvania State University (1981). The use of quadratic terms for curvature in the relationships was tried, but the term was never significant. Relationships were established by trying all combinations of variables possible, retaining only those variables where the correlation showed a high t-statistic. For final correlations, only variables significan~ at greater than the 90% confidence level were retained. Automatic selection of significant variables by the computer software proved unreliable because the measure of significance of a variable depended strongly on the variables present in the correlation due to the non-orthogonality of the data. Careful consideration of anomalous correlations containing different combinations of the same variables clarified problems as shown in section 3.3.3.2. The use of multiple linear chemical concentrations in Factor Correlation Method. regression on the final broth this fashion was named the 56. 3.3.2.2 Solvent Relationships with Pressure and Minimum .e!! Fig. 3.11 shows the relationship between solvent yields and p, while Fig. 3.12 shows the relationship between the solvent yields and pH. The results of a multiple m linear regression fitting the model [B] =a+ b (pressure) + c (pH) f m are given in Table 3.6. Results of a similar treatment of data for acetone, ethanol, and total solvents are shown in Table 3. 7. The equations fitted were [A] f = d + e(p) + f (pH) m [E] f g + h(p) + i (pH) m [T] f = j + k (p) + 1 (pH) m over the ranges 0 kPa < p < 2 50 kPa <="less than or equal to" 4 . 2 < pH < 4 . 6 5 m These results, summarised by Table 3.8, show the effect of p and pHm on the difierent solvents. [T]f is of course very strongly influenced by [B]f' and so their regression coefficients and levels of significance will also be closely related. This can be seen in the tables. The regressions were performed on the data for the eight runs A, B, C, D, F, G, H, and J. Run E was not used as no reliable pH value was available, and Run I was not used m as the fermentation conditions were different. Table 3. 8 nor pH, m written: shows acetone is affected by neither pressure hence the correlation for acetone should be contd. p. 62/ 57. Fi~. 3.11. Solvent Concentration vs. Pressure. Solvent Concentration KEY J.sLJj__ Cl total solvents • butanol A acetone ~/ 0 ethanol 16 regression line / / / total / solvent / / / 14 / ~ / / ./ ~ / ,, / ,, / / • / ~ / / / 12 / / ,, butanol • / / ,, / / / / ,, / / ~ / / 10 / / • / / ~/ ,, / • / ,, / / / / / / / ,, • 8 / ,, / / ,, / / 6 4 A ethanol 0 = = =1 --l-o 0 100 125 130 175 200 225 250 Pressure ( kPa) Fig. 3.12. Solvent Concentration vs. Minimum pH. 58. Solvent C'.Jncentration (g/1) KEY Cl total solvent 14 • butanol • acetone 0 ethanol regression line / 12 / / ./ / / • / total / / 10 solvent / / / / / / / / / butanol / ~/ / ./ / 8 / / / / . / / / / / / / / • / / / 6 • / / / / / / / / / ,, / / / / 4 / / / / / / / / • / / / / / / / 2 / • 0 A acetone - - ---- 0 - - - - - ·------~ --- *- - th 1 _ - - - -- - - - - e an.o ~~~---- 0 O ----.....L.---~----...l,_ ___ -1.. ____ L 4.2 4.3 4.4 4.5 4.6 4 .7 Minimum Observed pH Table 3.6 RESUL'IS OF MULTIPLE LINEAR REGRESSION (MLR) OF FINAL BUTANL CX>N:EN'IBATION (Bf) WITH HEADSPACE PRESSURE (p) AND MINIMlM BROI'H pH (pHrn) Variable* Value t-ratio** a -57.28 -2.51 b 0.0314 2.35 C 13.09 2.52 *a= intercept term b = pressure coefficient c = pHrn coefficient **Confidence limits oft for 5 degrees of freedom are: t.90 = 2.015 t.95 = 2.571 . 59. 60. Table 3. 7 RESUL'IS OF MULTIPLE LINEAR REGRESSION ON FINAL ACE'IONE CON:ENI'RATION (Af), FINAL ETHAOOL CON:ENTRATION (Ef), AND FINAL 'IOI'AL SOLVENT CX>N:ENTRATION (Tf). Solvent acetone ([A]f) ethanol ( [E] f) total solvent ([T] f) * t.90 = 2.015 t .95 = 2.571 Variable d e f g h i j k 1 Value - 4.18 0.00281 1.022 - 3.884 0.00591 o. 8164 -65.34 0.04013 14.93 Note: all correlations have 5 degrees of freedom. t-ratio* -0.61 0.69 0.65 -0.97 2.51 0.90 -2.60 2.72 2.61 Table 3. 8 SlMMARY OF SOLVENT REIATIONSHIPS WITH HEADSPACE PRESSURE AND MINIMl.M BROI'H pH Solvent Pressure pH m n-butanol * . * acetone X X ethanol * X total ** ** x probably unrelated * related at 90% confidence level or better ** related at 95% confidence level or better 61. (contd. from p. 56) 62. [A]f = 0.86 g/1 ± 60% for these runs. Table 3. 8 also shows that ethanol is affected only by p and not by pHm' and so a new correlation for [E] f may be made [E]f = 0.00684 p - 0.2908 i.e., g = -0.2908, h = 0.00684, i = 0 (cf Table 3.7). with r = o. 757 (7 d.f) and r 98 % = 0.750 (7 d.f) (Eaton Tables, 1974) (where r Coefficient). is Pearson's Product Moment Correlation Run E has been included in this correlation as pH data is not required. m The butanol concentration is given by [B]f = -57.3 + 0.03142 p + 13.1 pHm and total solvent by [T]f = [B]f + [A]f + [E]f = -65.34 + 0.04013 p + 14.93 pH m (obtained by summing the above relationships). 3.3.2.3 Solvent Relationships with Acids, Final pH and Minimum pH Pressure, In order that a more complete understanding of the concent­ ration relationships could be constructed, each variable was regressed with all other variables in Table 3. 3. It was then found that certain variables other than p and pH rn explained the observed variation in solvent yields better. 63. For example, the variation in [B] f is better explained by variation in [BA]f and p rather than pHm and p. The overall picture developed initially using experimental data was as shown in Fig. 3.13. However, the final pH (pHf) seemed to have an effect, but at worse than the 90% confidence level. In an effort to further understand the effect of pHf' pHf was combined with the experimentally measured values of [BA] f and [AA] f to calculate four new variables; [HBA]f, [BA ] f ' [HAA] f, and [AA '""]f° It was assumed that the H+ ion was the only cation of importance, and that the following relationships held. KBA = [H+] f [BA ] f - (1) [HBA] f [BA] f = [BA-] f + [HBA]f - ( 2) KAA = [H+] f [AA ] f - ( 3) [HAA]f [AA]f = [AA-]f + [HAA]f - ( 4) + pHf = -loglO [H ] f where KBA = dissociation constant for butyric acid in water = 1.484 X 10- 5 M = 1.484 X 10- 5 MrHBA/(MrBA - Mr H+) = 1.484 X 10- 5 X 8 8. 10 / (8 7. 1 X 1.0) = 1.501 X 10- 5 g/1 = dissociation constant for acetic acid in water = 1.750 X 10- 5 M = 1.780 X 10- 5 g/1 Mr = relative molecular weight (g/mol). The temperature was 30 °C. The results of these calcul- I I I pressure I / / / / I I I I - - - - - - - J>Ar /" pHrnin / Fig 3.13 FOSSIBLE t-CENIRATION INTERREIATIONSHIPS Probable relationships - - - R>ssible relationships 64. 65. ations can be seen in Table 3.4. The regression procedure was then repeated. The resulting correlations using the undissociated acid concentration instead of total acid con­ centrations showed improved correlations, and it was inter­ esting to notice that in regressions where undissociated and dissociated forms were present together, the dissociat­ ed form frequently dropped out but the undissociated form remained. Tables 3.9 - 3.16 show some important correla­ tions for [B]f' [HBA]f' [HAA]f, [E]f' [A]f, Ns' and R'. The specific growth rate, µ , was calculated for the expon­ ential growth phase from a graph of ln N versus time. The maximum specific rate of butanol prod~ction, R', was esti­ mated as the maximum observed rate of production divided by the stationary phase cell population (N ) • These tables s give some of the trial and error correlations attempted be- fore arriving at a correlation containing only significant variables. The t-statistic in conjunction with the degrees of freedom gives a measure of the degree of importance of the corresponding variable. As an approximate rule of thumb, any variable with at-statistic in excess of 2.0 is likely to be significant at greater than the 90% confidence level for any degree of freedom (df) except df = 1. 3.3.2.4 Summary of Correlations A summary of the final correlations selected from Tables 3.9 - 3.16 is given in Table 3.17, and accurate confidence levels of the t-statistic at different degrees of freedom in Table 3.18. Fig. 3.14 shows these relationships diag- rammatically. In brief, ary cell population is then, it seems that the station­ related only to the two alcohol concentrations, increasing with butanol and decreasing with ethanol. In particular the stationary cell population seems best related to the ratio of the alcohols. Ethanol is related to pressure and undissociated acetic acid concentration, while butanol is related to pressure and undissociated butyric acid. The undissociated butyric acid concentration is in turn related to the undissociated contd. p. 79/ 66. Table 3.9 CORREIATIONS FOR FINAL BITTAOOL CDNCENTRATION (Bf) Related Coefficient t-Statistic Degrees Variable of Freedom const 6.574 1.93 5 p 0.0311 2.29 [H+] -144573 -2.47 m const -0.388 -0.15 5 p 0.0303 2.48 [BA] f 0.876 2.16 const -3.58 -0.33 2 p 0.0317 2.00 [H+] m 70904 0.25 [HBA] f 3. 385 0.77 [BA] f -0.787 -0.37 const -1.023 -0.42 4 p 0.0317 2.76 [HBA] f 2.399 1.95 [BA-]f -0.873 -0.62 const -2.278 -0.38 4 p 0.0289 1.98 [HBA] f 1.426 0.92 [HAA] f 1.630 0.24 const -0.958 -0.42 5 p 0.0310 2.89 [HBA] f 1. 762 2.75 no const p 0.0275 4.32 6 [HBA] f 1.605 3.32 67. Table 3.10 (l)RRELATI0N3 FOR THE UNDISSOCIATED BlITYRIC ACID (l)N:ENTRATION (HBA) !Elated Variable const p [H+] m [AA] f const p [H+] m [HAA] f const p [H+] m const [H+] m . [HAA] f const p [HAA] f const [HAA] f Coefficient 2.287 -0.00264 -84456 0.9224 -18.03 -0.0150 255846 10.889 5.8409 0.0004714 -139693 -2.0283 -20832 3.272 -2.662 -0.004724 3.834 -2.907 3.4566 t-Statistic Degrees of Freedom 0.66 3 -0.48 -1.19 1.18 -3. 71 3 -4.08 3.10 4.96 3.27 4 0.09 -2.49 -0.32 4 -0.20 1.27 -2.13 5 -1.32 4.32 -2.22 6 3.88 !Elated Variable *const p [H+] m [H+] f **const p [H+] m [H+] f **const p [H+] m 68. Table 3 .11 O)RREI.ATIONS FOR THE UNDISSO:IATED ACETIC ACID (X)N:ENI'RATION ([HAA] f) AND 'TOTAL ACETIC ACID ( [AA] f) Coefficient t-Statistic 5.853 10.79 0.00210 1.84 -69727 -5.63 -76378 -5.32 2.253 9.24 0.001383 2.71 -36622 -6.59 -2302 -0.36 2.192 14.14 0.00142 3.22 -36325 -7.48 * correlation for [AA]f ** correlation for [HAA]f D:grees of Freedom 3 3 4 Related Variable const [B] f [E]f const p [H+] rn const [H+] rn [BAJ f [AA] f const [BAJ f [AA] f const [BAJ f [AA] f const [H+] f [HBA] f [HAA] f const [HBA) f [HAA] f const Table 3.12 CORREIATION.S FOR THE FINAL ACE'IOOE CON:ENIBATION ([A] f) Coefficient t-Statistic 0.5071 1.10 0.02373 0.24 0.2387 0.34 0.888 0.88 0.002631 0.66 -13010 -0.76 0.523 0.23 -45469 -0.86 -0.5274 -2.18 1.326 1.87 -1.098 -0.90 -0.4331 -2.31 1.321 2.09 -0.3228 -2.30 0.7813 3.87 -0.244 -0.16 -44770 -1.18 -0.7684 -2.12 2.566 1. 70 -0.624 -0.39 -0.6652 -1.82 2.044 1.36 0.8633 (std. error = ± 60%) 6 9. Degrees of Freedom 6 2 3 5 6 4 5 8 70. Table 3.13 (X)RREIATIONS FOR THE FINAL ETHANOL OON:ENTRATION ( [E] f) !elated Coefficient t-Statistic Degrees Variable of Freedom const 0.0371 0.06 5 p 0.006 2.49 [H+] rn -7699 -0.74 const -6.718 -1.23 3 p 0.001606 0.39 [H+] rn 88474 0.96 [HM]f 3.378 1.37 const -1.662 -1.54 4 p 0.00467 1.77 [HBA] f -0 .018 -0.06 [HM]f 1.2164 1.01 const -8.432 -2.94 4 [H+] rn 118094 2.51 [HM]f 4.194 3.63 const -0.2744 -0.76 7 p 0.00667 3.07 const 1.2987 2.87 6 [H+] rn -17365 -1.32 const -1.371 -1.50 6 [HM]f 1.521 2.45 const -1.6142 -2.31 5 p 0.00476 2.34 [HM]f 1.147 2.31 71. Table 3.14 OORRELATI0N3 FOR THE STATIONARY PHASE CELL FOPUIATION {N) s !Elated Variable **const [BJ f [AJ f [EJf [H+J m **const p [BJ f [EJf **const [BJ f [EJ f *const [BJf [EJf **const [BJ f/[EJ f *const Coefficient 9.1360 0.04109 -0.06876 -0.22830 -1314 8.977 0.000835 0.03896 -0. 3310 9.0447 0.04225 -0.2782 1,112 X 106 142 X 106 -942 X 106 8.9 0.02388 6.38 X 108 t-Statistic 54.56 2.64 -1.39 -2.66 -0.45 91.50 0.97 2.55 -2.77 132.04 2.85 -2.64 5.06 2.98 -2.78 130.91 3.60 2.80 [BJf/[EJf 7.82 X 107 3.52 * Ns used for correlation. ** log Ns used for correlation. I:egrees of Freedom 3 5 6 6 7 7 72. Table 3 .15 CX)RREIATION3 IN::WDING THE SPECIFIC GROWI'H RATE ( µ} Variable Pelated Coefficient t-Statistic Cegrees of Variable Freedom µ const 0.3912 1.09 2 [HBA] f 0.0622 1.05 [H+] -7562 -0.80 m [BA] f const -0.1118 -0.07 4 µ 12.188 2.14 [HBA] f const -2.892 -1.30 3 R' 18.84 1.33 . µ 8.856 2.97 Bf const 6.304 1.52 4 µ 4.90 0.35 µ const 0.2855 1.62 4 p -1.22 X 10-5 -0.01 [HBA] f const -0.1280 -0.15 4 µ 7.209 2.43 [HBA] f const -2.599 -2.31 3 µ 2.862 1.11 [HAA] f 2.5584 2.55 µ const 0.7181 3.99 3 [H+] m -15124 -2.45 µ const -4.071 -2.23 3 pH 0.9571 2.39 m 73. Table 3 .16 CDRREIATIONS ncwo:m:; THE SPEX::IFIC RATE OF BUTAOOL PRODlCTION (R') Variable ~lated Coefficient t-Statistic Degrees of Variable Freedom fA]f const 0.1228 0.32 7 R' 6.061 2.05 [AA-] f const 0.5374 1.86 6 R' 5.868 2.82 pHf const 4.35411 59.20 7 R' 2.4900 4.46 R' const -0.7001 -0.83 4 [A]f 0.03109 1.37 [AA-] f 0.05035 0.86 pHf 0.1570 0.81 R' const -1.0818 -3.38 6 pHf 0.2532 3.60 £ AJf 0.0284 1.38 R' const -0.8722 0.96 5 [AA-] f 0.0461 0.73 I pHf 0.2010 0.97 R' const -0.01639 -0.37 5 [A] f 0.03380 1.55 [AA-] f 0.0892 2.84 Bf const 3.667 1.28 7 R' 28.48 1.31 contd./ Variable [HBA] f R' [H+] f [BA] f [HBA] f [HBA] f Bf Table 3.16(contd} CDRREIATION.S IN:WDING THE SPECIFIC RATE OF BUTAOOL PROOU:TION (R ') !Elated Coefficient t-Statistic Variable const 1.338 5.29 R' 0.0067 0.00 const 0.08298 5.73 1..1 -0.08579 -2.69 p 2.475 X 10-4 5.75 const 4.06 X 10-5 8.22 R' -1.42 X 10-4 -3.79 const -4.810 -3.68 1..1 5.962 2.10 [AA] f 2.337 4.32 const 0.593 2.95 R' -4.816 -3.46 [BA] f 0.57851 14.84 const 1.009 0.89 R' -7.341 -0.98 1..1 -0.665 -0.29 [BA] f 0.5936 4.73 const -13.217 -2.55 R' 133. 75 4.08 1..1 16.531 2. 36 74. Degrees of Freedom 6 3 7 3 5 2 3 75. Table 3 .17 (l)RREI.ATION SlMMARY TABLE Dependent Related Coefficient t-statistic d.f. Variable Variable log N s const 8.9 130.91 7 [B]f/[E]f 0.02388 3.60 N const 1,112 X 106 5.06 6 s 142 X 106 [B]f 2.98 [E]f -942 X 105 -2.78 [B]f const - - 6 p 0.0275 4.32 [HBA] f 1.605 3.32 [A]f const 0.86± 60%* (9 values) [A]f const - - 6 [BA] f -0.323 -2.30 [AA] f o. 781 3.87 [E]f const -1.6142 -2.31 5 p 0.004759 2.34 [HAA] f 1.147 2.31 [HBAJ f const -2.907 -2.22 6 [HAA] f 3.457 3.88 [HAA] f const 2.192 14.14 4 p 0.00142 3.22 [H+] m -36327 -7.48 contd./ * standard error (approx. one standard deviation) 76. Table 3.17 (X)RREIATION St.MMARY TABLE (contd) D:pendent !Elated Coefficient t-statistic d.f. Variable Variable pHf const 4.35411 59.20 7 R' 2.4900 4.46 R' const 0.08298 5.73 3 J.J -0.08579 -2.69 -4 5.75 p 2.475 X 10 J.J const o. 7181 3.99 3 [H+] rn -15124 -2.45 Table 3 .18 LEVEI.S OF SIGNIFICANCE FOR THE t-STATISTIC (QUINN, 1974) d.f. Level of Probability 90% 95% 1 6.314 12.71 2 2.920 4.303 3 2.353 3.182 4 2.132 2. 776 5 2.015 2.571 6 1.943 2.447 7 1.895 2.365 8 1.860 2.306 9 1.833 2.262 10 1.812 2.228 77. 99% 63.66 9.925 5.841 4.604 4.032 3.707 3.499 3.355 3.250 3.169 -'ve fHBAJ f [BJ f fHAAJ f -'ve *The independent variables were + Pressure and [H Jm. 78. r R' i J1 (contd. from p. 65) 79. acetic acid concentration and probably not to the pressure or minimum pH (refer sect. 3.3.3.2). Undissociated acetic acid is in turn related to pressure and minimum pH. Acetone, however, appears best related to the total acid concentrations (as opposed to undissociated acid concent­ rations), increasing with butyric acid and decreasing with acetic acid. It appears that pHf is dependent on R', which is depen­ dent on µ and p, and µ is dependent on pH • These last m correlations can be seen in Tables 3.15 and 3.16. 3. 3. 3 Discussion of Correlation Results 3.3.3.1 Overview Confidence in the general validity of the presented in Table 3 .17 and Fig. 3 .14 can relation.ships be gained by their comparison with the basic metabolic pathways of the organism. Fig. 3.15 shows the metabolism in much more detail than presented earlier (Fig. 1.6) (Rhodes and Fletcher, 1g66: Ross, 1962: Thauer et al., 1977). Butanol is made from butyl coenzyme A (butyl CoA) which in turn is made from acetyl CoA. Ethanol and acetone are made from acetyl CoA also. Hence, not only do the t-sta t i st ics in Tables 3 .1 7 and 3.18 suggest that the correlations are strong (with almost all correlations at 95% or 99% confidence levels), but the results resemble the known biochemistry to a large extent (Fig. 3.15). The only correlations showing less than 95% confidence are those for acetone and ethanol (at the 90% confidence level) • This may be ascribed in part perhaps to the low concentrations of these solvents in the broth relative to butanol. Glycolysis 2ATP 2NADH CH3CCXDOH pyruvate Fdox) 2e-, Fdred HSCoA HSCoA H3F04 HSCoA 80. { 2e- + NAD+ + H+ -+ NADH then or 2e- + 2H+ -+ 2H2 CH3COOF03H2~ CH3C0SCoA \,,,.,.,NAD+ > CH3CHO;:=\=:~J:=AD=+==~• CH3CH20H ac1~::fha~e acertyl CoA acetaldehyde ethanol CH3(X)OO=====::::,,... HSCoA acetate '---"' CH 3COCH200SCoA • CH 3CDCH 20JOH ;:::=~ CH 3CO'.:H 3 + C02 acetoacetyl CoA acetoacetate acetone NAD+1r Fig 3.15 DETAILED BIOCHEMISTRY Also: NADH + H+ ~ NAD+ + 2H ~ 2H -+ H 2 and Fd is ferredoxin 81. Discussion of each correlation in Table 3.17 and Fig. 3.14 with its suggested metabolic cause follows in the next few sections. 3.3.3.2 Anomalies in the Correlations It is quite noticeable that when [HAA] f is included in a + correlation, then both P and [H) also appear to be m important. In the case of [HBA]f correlations (Table 3.10) neither p nor pH show great significance unless m both are present. In the case of [E]f correlations (Table 3.13) the opposite is true - in a correlation including [HAA]f, either p or [H+]m will give a significant correlation, but not if both are present together. The source of these perplexities most probably lies in the very strong correlation of [HAA] f with both p and [H+] (Table 3.11). The final correlations for [E] f m and [HBA] f (Table 3 .1 7) were made on the basis of correlations involving only p and [H+] without [HAA] f [HBA] f two, m in the correlation. Such correlations suggest [H+] is the more m important of and however it drops out when nor [H+] was finally m [HAA] f is included used (Tables for the so 3.10 neither p and 3.17). When [E] f is considered in this manner, p appears as the more important variable and is still significant when [HAA] f is included in the correlation. It is not surprising from a metabolic view point that [H+] has no importance in the [E] f correlation as m ethanol reaches its maximum concentration before pH m occurs (Tables 3.13 and 3.17). The correlations for R' also show some results which are difficult to understand, but this also appears due to the nature of the data. Further discussion of these results in metabolic terms can be found later. 82. Another possible cause of a misleading correlation is the effect of coincidence due to random causes alone. Of the significant, it is 16 variables possible one which appear is due only to random coincidence. quite If the confidence level of each are all multiplied together, then the result will be to give an overestimate of the probability that none are due to random coincidence, W, alone: w = 0.99 X 0.95 X 0.95 X 0.99 X 0.95 X 0.9 X 0.99 X 0.9 X 0.9 X 0.99 X 0.95 X 0.99 X 0.99 X 0.9 X 0.95 X 0.9 = 0.99 6 X 0.95 5 X o.9 5 = 0.941 X 0.774 X 0.59 = 0.43 (from Tables 3.17 and 3.18). This means that there is only a 43% chance that all the correlations in Table 3.17 are a true representation of real effects. Hence, while confidence in any particular correlation may be strong, there is a better than even chance that one of the "related" variables identified is in fact unrelated. 3.3.3.3 The Typical Fermentation Fig. 3.16 shows a typical tracing against time expected for an experiment run near one atmosphere gauge pressure. Cell count readings during the first five hours after inoculation usually show a rise then a fall (data for Run A is given in Table 3.19). This effect is due to settling of the cells in the fermenter past the sample point and perhaps to a short lag period. No attempt at agitation was made at any stage during fermentation until after final sampling. If Table 3.2 is considered in conjunction with Fig. 3.16, then the first five hours may be considered a period when the very small cells (ea. 0.1 µrn) observed after 4 hours are multiplying to become a 8 3. . Fig. 3.16. A Typical Ferrrentation at 1 atm. (ga.) Pressure. - pressure Cu ~ 20 -r-l I 0 r-l X 15 - ; Cl) butanol Cl) ~ 10 0.. -r-l '-.. .!!; fl 5 ~ acetone r-l ethanol g 0 10 20 30 40 so 60 70 80 90 100 10 cell population . a ** possible fluctuations in 8 acid concentrations (cf. - Prescott and Dunn, 1959). +l § 8 r-l ,-I 6 ~ 0 pH ~ 8' ,-I 4 .. - butyric acid ,-I ~ ~ 2 acetic acid •,-j u ~ 0 10 20 30 40 50 60 70 80 90 100 ti.me since inoculation (hr.) • Table 3 .19 CELL POPULATION DATA FOR RUN A Time Cell Cbunt (hrs) (10 6 cells/ml) 0.42 29 1.25 45 2.5 48 3.8 60 5.2 45 7.7 59 9.2 108 9.9 195 10.5 210 12.0 407 13.6 480 14.6 767 15.3 900 15.8 973 16.5 1,287 19.5 1,113 20.4 873 21.3 1,320 22.5 697 33.5 1,330 57.8 1,033 130.3 560 (ag.) 713* *The 130 hour sample was agitated and on recountiil3 was found to increase about 30%. 84. 85. significant proportion of the population, finally becoming more numerous than the large (ea. 1-2 \J m) almost non­ motile cells. For this reason, and the lack of agitation (which would be required to make the broth homogeneous and the sample representative), data taken in the first five hours should be used only with great care. Once motility started and gassing began (as indicated by a rise in pressure after about 5 hours), the broth was assumed to be well mixed and the samples representative. Throughout the exponential growth phase (ea. 5 - ea. 25 hr s) the cells were rnot ile and about 1-2 \J rn in length. Early in this phase the acid concentration rose rapidly and the pH dropped. Wheti solvent production began (usually all three solvents almost simultaneously), both of the acid concentrations appeared to drop. This dip is not particularly noticeable in Figs. 3.1 - 3.10 due to the low density of sampling in this region. As solvent production began to slow the acid concentrations increased quickly and continue to rise to a new maximum. Both acetone and ethanol appeared to reach their maximum concentrations early in the stationary phase and before pH is reached. However, n-butanol continued to be m produced until late in the stationary phase. The decline phase is reflected only by a decreasing cell count and not by any other parameter. 3.3.3.4 Fluctuations Population in the Stationary Phase Cell An interesting peculiarity of the stationary cell popula­ tion is the large fluctuations observed after the transi­ tion region between the exponential and stationary phases. This can be seen clearly in data from Run A (Table 3.19; Fig. A3.2) although it is often masked in a graph of ln N vs. t (e.g. Fig. 3.1). The exponential phase seemed to start after about 5 hrs and continued for about 11 hours. 86. After 16 hours (between 16 hours and some time less than 130 hours) the stationary phase was reached, the cell population was seen to oscillate, and finally the decline phase began. For Run A, the standard deviation of cell numbers in the stationary phase is far in excess of the expected error (7 6 values; mean of cell number 1,093 x 10 cells/ml; standard deviation of 243 x 10 6 cells/ml or 22% of the mean). Appendix 2 sh