By-product inhibition effects on ethanolic fermentation by Saccharomyces cerevisiae

zyxw zy zyxw By-Product Inhibition Effects on Ethanolic Fermentation by Saccharomyces cerevisiae BRIAN MAIORELLA, HARVEY W. BLANCH, and CHARLES R. WILKE, Lawrence Berkeley Laboratoy, and Department of Chemical Engineering, University of California, Berkeley, California 94720 Summary Inhibition by secondary fermentation products may limit the ultimate productivity of new glucose to ethanol fermentation processes. New processes are under development whereby ethanol is selectively removed from the fermenting broth to eliminate ethanol inhibition effects. These processes can concentrate minor secondary products to the point where they become toxic to the yeast. Vacuum fermentation selectively concentrates nonvolatile products in the fermentation broth. Membrane fermentation systems may concentrate large molecules which are sterically blocked from membrane transport. Extractive fermentation systems, employing nonpolar solvents, may concentrate small organic acids. By-product production rates and inhibition levels in continuous fermentation with Succharomyces cerevisiae have been determined for acetaldehyde, glycerol, formic, lactic, and acetic acids, 1-propanol, 2-methyl-l-butanol,and 2,3-butanediol to assess the potential effects of these by-products on new fermentation processes. Mechanisms are proposed for the various inhibition effects observed. INTRODUCTION zyxwvu Cysewski found in operation of a laboratory vacuum fermentation (Fig. 1) that the buildup of some nonvolatile inhibitor limited the ultimate productivity of the fermentation.' In the vacuferm process, the fermentation is conducted at reduced pressure (approximately 50 mm Hg). Ethanol is boiled away at 35OC as it is produced, maintaining the beer ethanol concentration below 3.5 wt. TO.Ethanol end-product inhibition is alleviated and the specific ethanol productivity (g ethanollg cells h) is increased. A concentrated glucose feed can be fully converted. The product leaves as a concentrated vapor stream (thus reducing distillation costs). Cells grow during fermentation but cannot escape the fermentor in the vapor product stream and so the yeast density increases, further increasing the fermentor productivity. Figure 2 shows the results of a continuous vacuum fermentation of a 334-g/ L glucose feed solution. Cell density and fermentor productivity increase for the first 40 h of operation with a maximum ethanol productivity of 44 g L-I h-' (ten times the average for a conventional batch fermentation). After 48 h, however, the density of viable cells, and hence the productivity, sharply de- zyx zyxwvuts zyxwvu Biotechnology and Bioengineering, Vol. XXV, 103-121 (1983) 0 1983John Wiiey & Sons, Inc. CCC 0006-3592/83/010103-19~02.90 104 zyxwvutsr MAIORELLA, BLANCH, AND WILKE Vocuum Comoressor +L4ElhonDl Woter Vapor Product ond C02 Vocuum Ferrnentor &+-~teri~e Fig. 1 . Oxygen zyxw zyxw z zyxwvutsr Continuous vacuum fermentation. Productivity limited by buildup of nonvolatiles. Vaccum Fermentation- 8o (no cell b l e e d ) 33 4 % Glucose feed Oxygen bleed r o t e 0 12 V V M 60 - - L r - 60 --" 21 > U 0 "7 v1 RI c 5 E - zyxwvut zyxwvut zyxwvutsrq zyxw -40 - OJ - u Y a a 203 0 zyxwv - 5 5 40.- - 20 - - I 0 Fig. 2. > 1 20 I I 40 I I 60 I 0 80 Viable cell decline in vacuum fermentation without bleed clined. The sharp decline in viable cells after two days is indicative of the buildup of some nonvolatile component to a level toxic to the yeast. The problem of toxin buildup could be readily overcome by continuously withdrawing a bleed stream of fermentor beer (Fig. 3) to maintain the concentration of nonvolatiles below the level of toxicity. The results of such a continuous vacuum fermentation with bleed are shown in Figure 4 where cell mass and cell yield are plotted versus a concentration factor (the ratio of the feed rate to the bleed rate).' The fermentation was begun with a large bleed. A decrease in BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION Vocuum Compressor zy zy 105 Ethanol Woter Vapor Product and C02 Fermentor %gar s o l u t i o n - i - - e / Feed + Bleed: Dilute Ethonal Beer with Concent rot e d Non-volotile By-products +Sterile Oxygen Fig. 3. Continuous vacuum fermentation with liquid bleed to limit buildup of nonvolatile inhibitors. zyxwvu zyxwvu zyxwvu 0.08- I I Vocuum Ferment0 t ion 334 q/1 Glucose Feed I I Concentrotion Foctor, I/y = - 60 Feed Bleed Fig. 4. Limiting bleed rate in continuous vacuum fermentation. bleed rate, while holding the feed rate constant, then increased the concentration factor. As the concentration factor was first increased from 1 to 2.5, cell growth continued and the cell mass concentration increased with fewer cells being washed out in the bleed. Above a concentration factor of 2.5, however, the cell yield decreased and cell concentration dropped. At this concentration factor, the bleed was insufficient to maintain the concentration of the nonvolatile inhibitor below toxic levels. 106 MAIORELLA, BLANCH, AND WILKE Use of a large bleed limits the productivity of the fermentation process as cell density is diminished. Further, the ethanol product removed in the bleed is very dilute and costly to distill. It is therefore desirable to identify and, if possible, control limiting inhibitors to decrease the bleed and increase productivity. The problem of toxin buildup may be common to many of the new fermentation processes which remove ethanol as a concentrated product from the beer (Fig. 5).2Nonvolatiles are concentrated and may be inhibitory in vacuum fermentation. Selective membrane fermentations3 may concentrate larger molecules which are sterically blocked from membrane transport. Extractive fermentation system^,^ employing nonpolar solvents to remove ethanol, may concentrate organic acids. The source of inhibitors may be feed components which are not fully metabolized and which concentrate in the fermentor, or they may be fermentation by-products. In this article, we consider the effect of by-products as these may be hard to eliminate from the fermentation system and may thus set an ultimate limit on fermentation productivity with a given organism. In a further article, the effects of common feed components, when concentrated to high levels, are p r e ~ e n t e d . ~ zyxwvut EXPERIMENTAL APPROACH zy The products of alcoholic fermentation of a synthetic glucose media by Saccharomyces cerevisiae as given by Neish and Blackwood6and as determined in our own laboratory’ are presented in Table I. When sugar from natural sources containing amino acids (such as corn hydrolysate or molasses) is used, fusel oils will also be produced,8 with up to 5 g of these propyl and butyl alcohols produced per liter of ethan01.~ To test the effects of these by-products, continuous fermentations were conducted with increasing amounts of each individual by-product added to the feed until cell growth and ethanol productivity were inhibited. Those by-products marked by an asterisk in Table I were tested. The six major synthetic media by-products and a representative straight-chained and a branched fusel oil component were used. The inhibition studies were carried out in 5-L (2.4-L working volume) New Brunswick fermentors, arranged as shown in Figure 6. Conditions for these studies are summarized in Table 11. Temperature and pH were controlled at established optima for the yeast strain (S. cerevisiae var anamensis ATCC 4226). A feed glucose concentration of only 20 g/L was chosen to limit ethanol production and prevent the masking of by-product inhibition effects by ethanol inhibition. While the basic fermentation reaction to produce ethanol is anaerobic, oxygen is required for the biosynthesis of unsaturated fatty acids and sterols. A dissolved-oxygen concentration of 5% of air saturation was maintained in the fermentor. The dilution rates were chosen to ensure a substantial residual glucose level as according to Moss et a1.I0 the metabolism will zyxw BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION zyx 107 zyxw zyxwvu zyx zy MAIORELLA, BLANCH, AND WILKE 108 TABLE I Products of the Alcoholic Fermentation of Glucose by Saccharomyces cerevisiae From synthetic media: Product mM product/lWmM glucose Ethanol* Carbon dioxide 2,3-Butanediol* Acetoin Glycerol* Acetic acid* Butryic acid Formic acid* Lactic acid* Succinic acid Acetaldehyde* Additional from molasses media Fuse1 oil: 1-propanol* 2-methyl-l-butanol* 177.0 180.8 0.48 - 6.60 0.69 0.32 0.42 0.38 0.26 5.0 0.34 0.11 *The marked by-products represent those that were tested in this study. Scmpling Bell . Dissolved oxygen probe pH probe Secondory Products Addition s Thermocouole vent t Sterile Plote Filter ClOOgA g l w s e Fig. 6. Sompling Bell Bleed Continuous fermentor used in secondary component inhibition studies. BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION TABLE I1 Conditions for Continuous By-product Inhibition Studies Base medium composition Glucose Yeast extract NHJI (NH&SO, MgS04. 7 H 2 0 CaCI, Antifoam zyx 109 zyxwv 20 g/L 1.7 g/L 0.25 g/L 0.65 g/L 0.022 g/L 0.012 g/L zyxwvuts 0.040 mL/L Fermentation conditions PH Temperature Dissolved O2 Dilution rate 4.0 35OC 5% of air saturation Acetaldehyde and glycerol experiments: 0.11 h-I Other experiments: 0.16 h-l be fermentative and independent of oxygen concentration as long as the glucose level is above 3 g/L. Unfortunately, this was not the case. Initial ethanol yields (before the addition of any by-products) were only approximately 0.37 g ethanol/g glucose consumed compared to the anticipated 0.47. A complete mass balance (including carbon dioxide off-gas analysis) confirmed a high fraction of aerobic metabolism. An interpretation of the meaning of changes in ethanol and cell yield factors is thus compli,cated. Loss of volatile components in the off gas was minimized by venting through a refrigerated (2OC) condenser and monitored by analysis of condensates in a liquid nitrogen trap. A stock solution of feed concentrate made to five times the final feed concentration as given in Table I1 (i.e., 100 g/L glucose) was prepared and sterilized by autoclaving. The final feed, with added by-products, was prepared by measuring 4 L of the sterile feed concentrate into the feed reservoir and then adding the by-products and water though a sterilizing filter to make up to a final 20-L volume (20 g/L glucose concentration). The fermentations were begun in by-product free media, inoculating the fermentors with 100 mL of a dense (approximately 10 g cells/L) actively growing yeast culture. The fermentor was kept in batch growth until actively fermenting (about 12 h) and then switched to continuous flow. A benchmark steady state was achieved without by-product addition. By-products were then added and steady states achieved at successively higher by-product concentrations. Samples from the feed and fermentor overflows were taken at regular intervals and a steady state was noted when three successive fermentor samples separated by at least six hours each gave the same cell density, ethanol concen- MAIORELLA, BLANCH, AND WILKE 110 zyxw tration, and residual glucose concentration. The steady state was normally achieved within six fermentor volume flows. Cell densities were determined by optical density measurement at 600 nm and confirmed by actual dry weight measurements of filtered samples. Glucose concentrations were measured using the dinitrosalicyclic acid method. Ethanol concentrations and the concentrations of the volatile by-products were determined by gas chromatography. The concentrations of lactic and formic acid in the fermentor could not be determined by gas chromatography and thus were not measured. It was assumed that like acetic acid (and all of the by-products except acetaldehyde), the fermentor lactic and formic acid concentrations were the concentrations in the prepared feed. zyxwvu RESULTS The results summarized in Table I11 list the by-products studied and byproduct feed concentrations at high inhibition (where the cell density is reduced by 80%). Results for earlier ethanol and glucose inhibition studied2 are included for comparison. The yeast cell is a complex system with many transport, energy, and biosynthetic pathways. Each inhibitor may have many individual points of metabolic effect. An exact knowledge of the modes of effect of each inhibitor would be of tremendous value, since specifically “engineered” organisms resistant to each mode of attack might then possibly be developed. For the present purpose of assessing the limitations placed upon new selective ethanol removal processes by by-product inhibition effects, a generalization of the modes of inhibitor effect is most desirable. Recognizing that these zyxwv TABLE 111 By-product Inhibition Summary By-product Concentration at high inhibitiona (g/L) Ethanol Formic acid Acetic acid Lactic acid 1-Propanol 2-Methyl-1-butanol 2,3-Butanediol Acetaldehyde Glycerol Glucose aAn 80% reduction in celi mass. 70 2.7 Inhibition mechanism Direct inhibition of ethanol production pathway Chemical interference with cell maintenance functions Chemical interference Chemical interference Chemical interference Chemical interference Chemical interference Chemical interference (Largely reconsumed) Osmotic pressure effects Osmotic pressure effects zyxwv 7.5 38 12 3.5 90 5.0 450 380 BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION zy zyx 111 are not necessarily specific descriptions of the mode of action of the inhibitory by-products, all of the inhibition effects observed can nonetheless be classified into three basic mechanistic schemes, each with particular characteristics. The postulated modes of inhibition of each by-product are given in Table 111. Inhibition by Direct Interference with the Ethanol Production or Cell Growth Pathways zy zyxwvu zyxw z Ethanol inhibition has been shown by BazuaI2 to be by direct noncompetitive inhibition of the glucose to the ethanol pathway. Inhibition begins at about 25 g ethanol/L beer and is total at 95 g/L. The ethanol metabolic pathway generates ATP for cell maintenance and growth. Typical of this direct inhibition of the metabolic pathway is a constant proportional decrease in cell growth rate ( p ) and ethanol productivity (v) as ethanol productivity and, hence, available ATP for biosynthesis decreases with increased inhibition (see Fig. 7). There is no apparent change in cell morphology associated with ethanol inhibition. Direct inhibition of the cell growth pathway has not been observed in these experiments but has been induced by nitrogen starving the yeast. The cell / I 0 0 .360 361 o 320 321 0 280.24 0 24- Doto zyxw zyxw Bozuo Of Po(g/l ) ~ 0 940 0 835 A 910 0 04 08 ".rig. 7 1. 58.54 78.40 12 h r - ' ,l -v f,... inhibition 0 26 22 16 . 2 0 . . . . r . ,. , . *oerween , .r . growrn .. rare, ana ernanoi .. ,. . unaer ernanoi .. neiarionsnip speciric proaucriviry I 112 zyxw zyxwvu MAIORELLA, BLANCH, AND WILKE zyxw growth rate (p)is decreased while ethanol productivity (v) decreases or may be partially maintained as ATP is shunted away for production of by-products such as glycerol and acetaldehyde. Direct inhibition of the metabolic (ethanol) or cell growth pathways was not observed for any of the by-products tested. Inhibition by Chemical Interference with Cell Maintenance Functions Inhibition by chemical interference with cell maintenance functions is well illustrated in the case of acetic acid (which inhibits in the range from 0.5-9 g/L, as in Fig. 8). Acetate is soluble in the lipids of the cell membrane.13 Samson14has shown that acetic acid (or sodium acetate) inhibits by chemical interference with the membrane transport of phosphate. Phosphate transport through the cell membrane is an activated transport process requiring the expenditure of ATP. Acetic acid interference results in an increase in the ATP requirement for this maintenance function. Typical of this type of inhibition, cell production decreases while ethanol production increases to make available sufficient ATP for cell maintenance. The ratio of p/v decreases as the inhibitor concentration increases. Chemical interference effects can typically occur at very low inhibitor concentrations and where membrane disruption is involved (as in acetic acid attack), the cell morphology is altered with the cells becoming irregular and elongated (Fig. 9). Formic acid is very similar in lipid solubility to acetic acid.13 Both acids inhibit at similar concentrations (Fig. 10) and for both, while the cell density is zyxwvu zyxwvu ¶/I A c e t i c Acid in F e e d Fig. 8. z Acetic acid inhibition of cell growth and ethanol productivity. BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION NO R!4A L Fig. 9. zy zyx 113 ACETIC ACID Effect of acetate on cell morphology. Fig. 10. Formic acid inhibition of cell growth and ethanol productivity. zyx decreased, the specific ethanol productivity increases to a maximum of 1.5 h-' as the by-product concentation is increased. It is probable that the mechanisms of inhibition are identical. Unlike acetic acid, no cell morphology change was seen with lactic acid. This may simply have been because the acetic acid experiment was carried up to higher inhibitor concentrations. Lactic acid (Fig. l l ) , with its extra hydroxyl group, is much less soluble in lipids than acetic or formic acids,13and lactic acid inhibition occurs at a much higher concentration (10-40 g/L). Further, Samson reports that lactic acid 114 zyxwvutsrq MAIORELLA, BLANCH, AND WILKE Fig. 1 1 . zyxwvu Lactic acid inhibition of cell growth and ethanol productivity. does not inhibit phosphate transp01-t.'~ Thus, the exact mechanism of lactic acid inhibition is probably different than that for acetic acid. As cell density decreases, however, specific ethanol productivity does increase (from 0.55 to 0.8 h-l), again indicative of some form of chemical interference with cell maintenance functions requiring increased ATP expenditure. Both of the fusel oil components tested-1-propanol (Fig, 12) and 2-methyl1-butanol (Fig. 13)-inhibit at similar low by-product concentrations and increase the specific ethanol productivity from 0.45 to 1.6 h-l. For both components, the cell morphology is changed (Fig. 14). The inhibited yeast are pseudomycelial (long and rod shaped). It appears as if cells have repeatedly budded but that the buds have not pinched off into individual cells after reproduction. Detergents and water immiscible solvents such as butanol are known lipid solvents and will cause the disintegration of the membrane. This is the probable mode of attack by the fusel oil components and may explain the observed morphology change. Under solvolytic conditions, the membrane will become "leaky"-freely permeable to monovalent cations such as K + and NH$ ,l4 more energy must be expended for maintenance, pumping these ions (by active transport) to maintain proper internal levels. The ATP demand for maintenance is increased, cell production decreases, and the specific ethanol productivity is increased as was observed. Note that 2,3-butanediol, with its two hydroxyl groups, is less lipid soluble. Inhibition by 2,3-butanediol occurs at a much higher by-product concentration (40-90 g/L, Fig. 15). Inhibition by 2,3-butanediol may be due to cell membrane disruption similar to propanol. Being only slightly lipid-soluble, there may exist an active transport mechanism for the removal of internally produced 2,3-butanediol through the lipid zyxw BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION zyx 115 Fig. 12. Propanol inhibition of cell growth and ethanol productivity. q/l zyxwvutsr zyxwv 2 M e t h y l - Butonol i n Feed Fig. 13. 2-Methyl-butanol inhibition of cell growth and ethanol productivity. membrane. At high external butanediol concentrations, the required pumping energy might be increased, and this could also explain the increase in specific ethanol productivity (from 0.6 to 1.1 h-l) as the butanediol concentration was increased in the fermentor. Acetaldehyde is an immediate precursor to ethanol in the yeast metabolic 116 MAIORELLA, BLANCH, AND WILKE NORMAL zyx zyxwvutsrq zyxwv zyx 1-PROPANOL Fig. 14. Effect of propanol on cell morphology. zyxwvutsr . 0 Concentrotion of 2-3butonedlol in fermentor o I & I 20 g/l I zyxwvutsrq zyxwv I I 40 60 00 2-3 Butonedial In Feed K Fig. 15. 2,3-Butanediol inhibition of cell growth and ethanol productivity. pathway. A unique feature of the acetaldehyde inhibition studies was consumption of the by-product by the yeast and conversion to alcohol. Thus, as the feed acetaldehyde concentration increased from 0 to 4 g/L, the residual acetaldehyde concentration in the fermentor increased only from 0.25 to 8 g/L. Figure 16 plots the results of the acetaldehyde inhibition experiment, with the specific ethanol productivity (v) and yield ( Y E )based only on ethanol derived from glucose, i.e., with the acetaldehyde-derived ethanol subtracted from the zyxw BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION 117 zy zyxwvut zyxwvut zyx g/1 A c e t a l d e h y d e in F e e d Fig. 16. Acetaldehyde inhibition of cell growth and ethanol productivity. zyxwv total. Acetaldehyde inhibits at about the same fermentor concentration and is structurally similar to formic and acetic acids. Acetaldehyde inhibition is accompanied by a cell morphology change with the cells increasing to over twice their normal diameter and appearing “mushy” (Fig. 17). Acetaldehyde inhibition may be by a mechanism of interference with active transport similar to that for formic and acetic acids. Inhibition by Osmotic Stress Inhibition by osmotic stress occurs when the concentration of some byproduct becomes so high that a large osmotic pressure gradient is established NOFlMAL ACETALDEHYDE Fig. 17. Effect of acetaldehyde on cell morphology. 118 zyx zyxwvu z MAIORELLA, BLANCH, AND WILKE between the interior of the cell and the fermentor broth, and the cell must expend large amounts of energy to maintain a homeostatic balance. The uptake of nutrients will require additional energy. There is no direct interference with any cell chemical process-no direct disruption of the cell membraneand the inhibitor would normally be classed as nontoxic to the yeast. Like the mechanisms of inhibition by direct interference with cell maintenance functions, cell production is first reduced with an increase in specific ethanol productivity. Inhibition by osmotic stress occurs only at very high inhibitor concentration and osmotically stressed cells become small, rigid spheres. Inhibition by osmotic stress is well illustrated by glycerol inhibition (Fig. 18). Glycerol has no effect at 100 g/L and significant cell growth continues at a concentration of 400 g/L. It is instructive to compare glycerol and glucose inhibition effects. Total cell productivity was reduced by 25% at a glycerol concentration of 210 g/L with a corresponding osmolality of 2.96. In batch experiments, cell productivity was reduced 25% by a glucose concentration of 270 g/L corresponding to an osmolality of 2.26. CONCLUSIONS A generalized ethanol removal fermentation system is shown in Figure 19. Feed glucose, nutrients, and water enter in stream 1. The concentrated ethanol product, and some water leave in stream 2. A dilute bleed of water, residual product, unutilized nutrients, concentrated by-products, and possibly zyxwvut g/l Fig. 18. G l y c e r o l in Feed Glycerol inhibition of cell growth and ethanol productivity. BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION zy zyx 119 zyx zyxwvutsrq t (3)BLEED Fig. 19. Generalized ethanol removal fermentation scheme for evaluation of inhibition effects. cells leave in stream 3. For the vacuferm process, stream 2 would be the concentrated vapor product and stream 3, the centrifuged bleed stream. To minimize cost, it is desirable to maintain the bleed stream as small as possible (thus removing most of the ethanol product as a purified concentrated stream). If the ratio of the size of the bleed stream (3) to the size of the feed stream (1) is y,and if the ethanol-recovery stream (2) contains no by-product contaminants and the by-products are not reconsumed by the yeast, then the concentration of any given by-product in the fermentor will be given by zyxwvu zyx where Cby-product is the by-product concentration (g/L); v is the specific ethanol productivity (h-l), X is the cell density (g/L); Yby-prduct is the by-product production ratio [(g by-product/h)/(g ethanollh)]; D is the dilution rate (h-’); y is the bleed-to-feed ratio; VXYby.p&uct is the by-product production rate (g by-product L-I h - I ) . Thus, the by-product concentration is increased by a factor of l/y-the smaller the bleed, the greater by-product concentration effect. This formula is not applicable in the case of acetaldehyde for which additional terms must be added to allow for acetaldehyde consumption. The concentration factor will also be less for by-products which “leak” out in the concentrated product stream (such as volatile acetaldehyde and formic acid, which will partially escape in the vacuferm concentrated vapor stream). Formula (1) can now be used to determine if any of the by-products tested might have been responsible for the decline in cell growth in Cysewski’s fermentation experiments. Using Cysewski’s maximum overall fermentor productivity of 38 g ethanol L-’ h-’ and a dilution rate of 0.27 h-’, taking values of by-products from Table I (converted to gram ratios) and recalling that for Cysewski’ssynthetic media, fuse1 oil components were not produced, the limiting value of y can be calculated. Cell productivity had declined 80% in Cy- 120 MAIORELLA, BLANCH, AND WILKE zyxw zy sewski's experiment at a concentration factor of 3.0 corresponding to a y of 0.333. Based on the inhibition experiments, formic acid would cause the earliest effect, and would not cause an 80% cell productivity decline until y had a value of 0.128. Acetic acid would not cause an 80% decline until a y had a value of 0.098. One may hypothesize that formic and acetic acids (with probably similar inhibition mechanisms) are working together to bring about the decline Cysewski observed. It must be remembered, however, that vclatile formic acid would have been largely carried away in the vacuferm vapor product stream and should not have contributed strongly to inhibition effects. It must be concluded that a fermentation by-product probably was not the primary inhibitor affecting the vacuferm experiments. The buildup of nonmetabolized feed components is a more likely explanation. (Inhibition by concentrated feed components has been studied in our laboratory and is the topic of another article.) Buildup of feed can be controlled; inhibition by fermentation byproducts may then still set an ultimate limit on fermentation processes. For all of the by-products studied, it is seen that specific ethanol productivity can be increased (from about v = 0.5) up to a maximum of v = 1.1-1.6 where the cell reproduction rate drops below the fermentor dilution rate and washout occurs. Assuming a cell recycle system such that a high value of cell growth rate ( p ) and yield is not necessary for high productivity, then by-product concentrations just below the total cell growth inhibition level are desirable. Using these values from our experiments, Yby.product values from Table 11, a high cell density of 100 g/L, a concentrated (300g/L) glucose feed typical of vacuum fermentation, the overall fermentor productivity of 80 g/L found by Cysewski for vacuferm with cell recycle, and formula (l), we find that cell productivity will be reduced at a bleed-to-feed ratio 1 :1.3, but that a bleed-to- zyxw zyxwvuts zyxwvu TABLE IV High Cell Density Maintained by Cell Recycle zyxwvu At onset of inhibitiona At high inhibition' By-product Concentration WL) YC Concentration WL) Y 2.3-Butanediol Glycerol Acetaldehyde Formic acid Acetic acid Lactic acid 1 -Propano1 2-Methyl-1-butanol 55 200 2.8 1.2 0.9 17 2.0 2.8 0.014 0.054 NA 0.289 0.794 0.037 0.184 0.061 90 450 5.0 2.7 7.5 38 12 3.5 0.0086 0.025 NA 0.128 0.098 0.0163 0.030 0.049 aTwenty percent reduction in cell mass production. 'Eighty percent reduction in cell mass production. 'For vacuum fermentation with an ethanol productivity of 80 g L-' h - I . zyxwvu zyxwvu BY-PRODUCT INHIBITION OF ETHANOL FERMENTATION 121 feed ratio of only 1:7.8 should be sufficient to prevent excessive toxic by-product buildup, as long as cell recycle is employed to maintain the high cell density (Table IV). The assistance of Wai K. Lam and Jeff Vincent in conducting the continuous fermentation experiments is gratefully acknowledged. This work was supported by the Department of Energy, Director for Energy Research, Office of Basic Energy Science, contract No. W-7405-ENG-48. zyxwvutsr zyxwv zyxw zyxwvu zyxwvu References 1. G . R. Cysewski and C. R. Wilke, Biotechnol. Bioeng., 19, 1125 (1977). 2. B. L., Maiorella, H. W. Blanch, and C. R. Wilke, Adv. Biochem. Eng., 20,43 (1981). 3. H. Gregor and T. Jefferies, Ann. NYAcad. Sci.. 396,273 (1979). 4. E. K. Pye and A. E. Humphrey, “Production of liquid fuels from cellulosic biomass,“ Proceedings 3rd Annual Biomass Energy Systems Conf., USDOE Solar Energy Research Institute, Golden, CO, J u n e s , 1980, p. 69. 5. B. Maiorella, H. Blanch, and C. Wilke, “Feed components as inhibitors of ethanolic fermentation by Saccharomyces cerevisiae, ” unpublished. 6. A. C. Neish and A. C. Blackwood, Can. J. Technol.. 29, 123 (1951). 7. Unpublished results; acetaldehyde not reported by Neish and given as determined in the authors’ laboratory. 8. A. Webb and J. Ingraham, Adv. Appl. Microbiol., 5,317 (1963). 9. H. Suomalainen, Suom. Kemistilehti, 41A (12), 239 (1968). 10. F. 1. Moss, P. A. D. Richard, and F. E. Bash, Biotechnol. Bioeng., 13, 63 (1971). 11. J. B. Summer and G . E. Somers, Laboratory Experiments in Biological Chemistry (Acade,mic, New York, 1944). 12. C. D. Bazua and C. R. Wilke, Biotechnof. Bioeng. Symp.. 7, 105 (1977). 13. E. J. Conway and M. Downey, Biochem. J . . 47, 347 (1950). 14. F. E. Samson, Arch. Biochem. Biophys.. 54, 406 (1955). 15. R. K. Finn, J. Ferment. Technol., 44, 305 (1966). Accepted for Publication July 12, 1982