BIOTECHNOLOGY AND BIOENGINEERING, VOL. XIX, PAGES 1125-1143 (1977) zy zyx Rapid Ethanol Fermentations Using Vacuum and Cell Recycle GERALD R. CYSEWSKI* and CHARLES R. WILKE, Department of Chemical Engineering and Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720 Summary zyxw zy Cell recycle and vacuum fermentation systems were developed for continuous ethanol production. Cell recycle was employed in both atmospheric pressure and vacuum fermentations to achieve high cell densities and rapid ethanol fermentation rates. Studies were conducted with Saccharomyces cerevisiae (ATCC No. 4126) at a fermentation temperature of 35°C. Employing a 10% glucose feed, a cell density of 50 g dry wt/liter was obtained in atmospheric-cell recycle fermentations which produced a fermentor ethanol productivity of 29.0 g/liter-hr. The vacuum fermentor eliminated ethanol inhibition by boiling away ethanol from the fermenting beer as it was formed. This permitted the rapid and complete fermentation of concentrated sugar solutions. At a total pressure of 50 mmHg and using a 33.4% glucose feed, ethanol productivities of 82 and 40 g/liter-hr were achieved with the vacuum system with and without cell recycle, respectively. Fermentor ethanol productivities were thus increased as much as twelvefold over conventional continuous fermentations. I n order to maintain a viable yeast culture in the vacuum fermentor, a bleed of fermented broth had to be continuously withdrawn to remove nonvolatile compounds. It was also necessary to sparge the vacuum fermentor with pure oxygen to satisfy the trace oxygen requirement of the fermenting yeast. INTRODUCTION A major constraint of conventional alcohol fermentation processes is ethanol or end-product inhibition. When a concentrated sugar solution is fermented and the ethanol concentration of the fermentation broth increases above 7 to lo%, the specific ethanol production rate and the specific growth rate of the yeast is severely suppressed.'** Ethanol inhibition produces many economic implications when considering industrial ethanol fermentations. In order to maintain * Present address: Dept. of Chemical and Nuclear Engineering, University of California, Santa Barbara, Calif. 93106. 1125 @ 1977 by John Wiley & Sons, Inc. zy 1126 zyxwvutsrq zyxwv CYSEWSKI AND WILKE the ethanol concentration a t the optimal level for ethanol production, concentrated sugar solutions, such as molasses, must be diluted to 10 to 20% sugar. The additional water used to dilute the substrate must then be carried through the fermentation process, increasing the size and cost of pumps, mixing and storage tanks, heat exchangers, and distillation columns. Also, because the cell mass concentration is a direct function of substrate concentration, lower cell densities will be experienced with diluted substrates. This results in lower fermentation rates per unit volume and hence, dictates that larger fermentors must be employed in the fermentation process. However, by use of a cell recycle arrangement the cell mass and the fermentor productivity may be increased. A portion of the cells in the effluent broth from the fermentor is separated from the beer by either sedimentation or centrifugation and recycled back to the fermentor. By this means the cell mass concentration in the fermentor is increased which produces higher fermentation rates per unit volume. To circumvent the problem of ethanol inhibition, ethanol must be removed from the fermenting beer as i t is formed. One means of accomplishing this task is to take advantage of ethanol’s high volatility and t o boil off the ethanol as i t is formed. Vacuum operation is, of course, necessary to achieve boiling of the fermentation broth a t temperatures compatible with the yeast. Besides eliminating ethanol inhibition, the vacuum operation produces an increase in fermentor productivity because the yeast cell mass concentration may be maintained at a high level. Only ethanol and water are boiled away from the fermentor while most of the yeast remains in the fermentor. The one disadvantage of vacuum fermentation is that carbon dioxide produced during fermentation must be compressed up to atmospheric pressure. This work was then undertaken to develop a cell recycle system and a vacuum fermentor for the continuous production of ethanol. The main emphasis of this study was t o assess the advantages of cell recycle and vacuum ethanol fermentations and demonstrate the high ethanol productivities obtainable with each mode of operation. zyx zyxw EXPERIMENTAL PROCEDURES The organism used in the fermentation studies was Saccharomyces cerevisiae, ATCC No. 4126. The standard media employed in all vacuum fermentations are listed in Table I. When the glucose concentration of the media was decreased t o 100 g/liter for the growth of inocula or atmospheric pressure operation, all other components RAPID ETHANOL FERMENTATIONS 1127 were decreased by the same ratio. The media were sterilized by dissolving the glucose in a n amount of water equivalent to 67% of the desired medium volume and by dissolving the salts and yeast extract in the remaining 33% of the water. After steam sterilization at 121°C for 30 min in separate containers, the solutions were allowed to cool to ambient temperature and mixed. Separate sterilization of the glucose and minerals was necessary to avoid carmelieation of the glucose which, while not affecting ethanol or cell mass production, did interfere with the optical determination of cell mass concentrations. TABLE I Base Medium for Vacuum Fermentation Per liter Component” Glucose (anhydrous) Yeast extract (Difco) NH,Cl MgSO4.7HzO CaClt Antifoam (General Electric AF60) Tap water to a 334 28.4 4.4 0.4 0.2 0.6 1 g g g g g zy zy zy ml liter All salts and glucose reagent grade. A 5 liter “Micro Ferm” fermentor (Fermentation Design Model MA501) was used in the atmospheric pressure cell recycle experiments. Details of the operation and arrangement of the continuous fermentor have been previously given.3 The effluent from the fermentor was passed to a jacketed settler, as described below, and a cell concentrate stream was returned to the fermentor. Tubing pumps (Sigmamotor Model TM-2.0-2) were used to control the flow of fermented beer from the fermentor to the settler and the cell recycle stream. The heart of the vacuum fermentor was also the 5 liter “Micro Ferm” fermentor. A schematic diagram of the complete vacuum system is shown in Figure 1. I n order to achieve the required boil-up rate of ethanol and water a 1500 W heater was added to the temperature control loop of the fermentor. The heater was constructed of four 10 in. diam coils of 3 in. copper tubing wrapped with electrical heating tape. The heat input was controlled by adjusting either of two variable autotransformers (Superior Electric Company type 3PN1168). A 1 in. stainless steel pipe connected to the fermentor zyxwv 1128 CYSEWSKI AND WILKE zyxw Vacuum Rrmenter Chilled woter Fig. 1. Schematic diagram of the complete vacuum system. zyx inoculation port led to two shell and tube condensers (American Standard No. 47M200-8A2) arranged in series. The vapor generated in the fermentor was condensed on the shell side of the exchangers by a 10% methanol-water solution chilled t o -4.0% by a Haws Model HR4-24W water cooler. The condensate was then collected for analysis in a 40 liter stainless steel tank which was set in a dry ice bath. The vacuum system was connected to a Kinney Model I(2-8 vacuum pump. The vacuum pump ran continuously and the pressure was controlled by a “Manowatch” Model MW-1 controller (Instruments for Research and Industry, Inc.), which activated a solenoid valve allowing filtered air to be bled into the system when the pressure became too low. Although the fermentor pressure fluctuated 1-2 mmHg with this method of pressure control, it was found superior to placing the solenoid valve in line with the vacuum pump, as recommended by the manufacturer, because the small pressure fluctuations helped to control foaming in the fermentor and allowed better liquid level control. The absolute pressure in the fermentor was measured with a Zimmerli gauge. As the liquid level in the vacuum fermentor dropped owing to the boil-off of vapor and the bleed-off of the fermented broth, a liquid level controller (Cole Palmer Model 7186) opened a solenoid RAPID ETHANOL FERMENTATIONS 1129 zy zy valve connected t o the medium reservoir and sterile broth was sucked into the fermentor to maintain a 2 liter fermentor working volume. The feed rate of fresh medium was thus determined by the boil-up rate and the bleed rate of fermented broth. A liquid level probe for the Cole Palmer controller was constructed of a 3 in. stainless steel rod which was forced down 4 in. Teflon tubing so that both ends were exposed for electrical contacts. The Teflon coating was necessary because its high hydrophobic surface properties did not allow a condensate film t o form on the probe. A liquid film (water) shortcircuits the probe with the fermentor head plate and causes the controller t o sense a high liquid level. However, during long term experiments the antifoam and protein constituents of the medium adsorbed onto the Teflon thereby changing the surface properties and producing a short circuit. This was corrected by wrapping the length of probe above the head plate with heating tape to boil off any surface water on the probe below the head plate. A bleed of fermented broth and cells was withdrawn from the vacuum fermentor by a tubing pump [Sigmamotors Model (TM20-2)] into a 4 liter jar which was maintained a t the same pressure as the fermentor. The cell bleed rate was adjusted by changing the speed of the pump and measured by emptying the 4 liter jar a t timed intervals and measuring the volume. Cell recycle experiments were run with both atmospheric pressure and vacuum fermentations using a jacketed settler vessel. A diagram of the settler arrangement is shown in Figure 2. The pressure in the settler and receiver flask was equalized enabling the clarified liquid to overflow by gravity to the receiver flask. The clarified liquid overflow rate was controlled by adjusting the difference between the pumping rate of the feed to the settler from the fermentor and pumping rate of the cell concentrate recycle stream. A solution of methanol and water chilled to 4.0"C was circulated through the jacket to slow fermentation in the settler. The settler system was operated a t a total pressure of 250 mmHg in the vacuum system and a t atmospheric pressure in the atmospheric fermentation system. Both cooling the settler and operating a t a pressure higher than the vacuum fermentation pressure of 50 mmHg was necessary to minimize mixing effects of COz evolved during fermentation in the settler. The vacuum fermentor was sterilized in place by filling the fermentor with 300 ml of a 70 volyo ethanol-water solution and boiling the solution under 250 mmHg total pressure (house vacuum) for 8 hr. The system was then flushed with air (3 liter/min) for 4 hr to remove zyxwv 1130 zyxwvutsr CYSEWSKI AND WILKE Fomented Broth enter Settled Yeast i To House Vocuum zyxwvuts uRe;mEng Fig. 2. Diagram of the settler arrangement used for cell recycle experiments. the last traces of the sterilizing solution. The fermentor was filled with 3 liter of 10% glucose medium, brought t o 35"C, and inoculated. An air rate of 0.5 liter/min was maintained during batch growth. At the end of batch growth (12 to 16 hr) the air flow was stopped and 0.12 v/v/m (240 ml/min a t STP) of oxygen was sparged through the fermentor. The pressure in the fermentor was slowly decreased (25 mmHg/min) until the fermentation broth began boiling a t 35°C. As the ethanol in the fermentation broth boiled off, the pressure was further lowered t o 50 mmHg to maintain boiling. I n all vacuum experiments the total pressure was 50 mmHg. At this pressure the boiling point of the fermentation broth containing 1% ' ethanol was 35°C-the optimum fermentation temperature of the yeast.3 The p H of the fermentation broth during vacuum operation was maintained between 4.0 and 3.5 by the buffering capacity of the medium. Unless otherwise stated, pure oxygen was sparged into the vacuum fermentor a t a rate of 0.12 v/v/m at STP and a n agitation rate of 500 rpm was used to supply adequate oxygen t o the yeast. zy zyxwvuts zyxw zyxwv RAPID ETHANOL FERMENTATIONS Assay Procedures zyx zy 1131 Ethanol concentrations. Ethanol was measured by gas chromatography using an Aerograph 1520 G-L chromatograph. A six foot t in. column packed with Chromosorb-W acid wash type 60-80 mesh was used with a flame ionization detector. The injector and detector temperatures were 175°C and the column oven operated isothermally a t 105°C. Cell mass. The cell mass concentrations were measured optically using a Fisher electrophotometer with a 650 mp filter. Glucose concentration. Glucose was determined by the dinitrosalicylic acid (DNS) m e t h ~ d . ~ Yeast viability. The percentage of viable yeast cells was determined using a methylene blue stain as described by Townsend and Lindgren.5 RESULTS AND DISCUSSION Semicontinuous Vacuum Operation Figure 3 illustrates the performance of the vacuum system during semicontinuous operation. Fresh medium was continually fed to the fermentor to maintain a constant volume as ethanol and water were boiled away. A bleed stream of fermented broth was not removed from the fermentor. This allowed the rapid accumulation of cell mass within the fermentor. However, components in the medium which were not metabolized by the yeast also accumulated in the fermentor under this mode of operation. The steplike appearance of the ethanol productivity curve in Figure 3 reflects that the productivity (boil-up rate times the ethanol concentration in the condensed product) was increased by manually increasing the boil-up rate and hence the feed rate to the fermentor. The boil-up rate was always adjusted so that the yeast was able t o ferment almost all the glucose in the feed. By this means the glucose concentration in the fermentor was held between 2 to 5 g/liter. The results shown in Figure 3 were obtained using a 33.4y0 glucose feed. No ethanol inhibition was detected and the cell concentration and ethanol productivity steadily increased with time for 48 hr. A maximum ethanol productivity and cell mass of 44 and 68 g/literhr, respectively, were obtained. However, after 48 hr of fermentation, the yeast cell mass concentration began to decline and the feed zyx CYSEWSKI AND WILKE 1132 60 z zyxw zy - 80 - - - 60 L c 0 u - zyxwvutsrqpo - zyxwvuts I 0 z I I I I 40 20 I 60 I -0 80 Time, hours Fig. 3. Performance of the vacuum system during semicontinuous operation. Vacuum fermentation (no cell bleed), 33.4% glucose feed; oxygen bleed rate: 0.12 v/v/m. rate, or boil-up rate, had to be sharply reduced to obtain complete fermentation of the glucose, and, as shown, the ethanol productivity correspondingly decreased. The sharp decrease in cell mass after two days of semicontinuous operation indicated than nonvolatile components were accumulating in the fermentor and killing the yeast. Only 60% of the yeast was found viable by the methylene blue stain method after 55 hr of operation. This required that a bleed stream of fermented broth be continually withdrawn from the fermentor to keep the concentration of nonvolatiles a t a level which did not inhibit yeast growth or ethanol production. zyx zyxwvut Continuous V a c u u m Operation Figure 4 illustrates the effect of removing a bleed of fermented broth. The data in Figure 4 were taken a t steady-state operation of the vacuum fermentor using a 33.4% glucose feed. The cell yield factor, YxIs, and cell concentration are plotted against a concentration factor. The concentration factor, c, is defined as c=’ 100 zyxwvut (FIB)(80) RAPID ETHANOL FERMENTATIONS 1133 z 0.08 0 zyxwvut zyxwvuts zyxwv zyxwv zyx 2 .o 4 .O 6.O 8.0 10.0 zyxwvut zyxw Concentrotion factor. c = F/B (S,)/OO Fig. 4. Effect of removing a bleed of fermented broth. Vacuum fermentation, 33.4% glucose feed; oxygen bleed rate: 0.12 v/v/m. where F is the volumetric feed rate, liter/hr; B is the volumetric bleed rate, liter/hr; So is the initial glucose concentration, g/liter. A decrease in bleed rate, holding the feed rate constant, increases the concentration factor and also increases the concentration of nonvolatiles in the fermentor. The concentration factor in Figure 4 was increased by lowering the bleed rate. Thus, as the concentration factor increased, the cell mass concentration rose because fewer cells were removed in the bleed stream. But when the concentration factor reached 8.5, the cell concentration and cell yield factor dropped. At this concentration factor the bleed stream was not sufficient and the concentration of nonvolatiles reached a critical level which began to inhibit yeast growth. Further increases in the concentration factor had a deleterious effect on yeast growth. The results of Figure 4 show that in order to sustain stable operation of the continuous vacuum fermentation, a bleed of fermented broth had to be removed so that the concentration factor did not rise above 8.5. The cell yield factor remained constant a t 0.064 during this vacuum fermentation a t concentration factors lower than 8.0. However, cell yield factors typically ranged from 0.055 to 0.066 for continuous vacuum fermentations operated a t concentration factors below 8.0. The reason for the variation in cell yield factors between consecutive vacuum experiments is not apparent a t this point. But the cell 1134 zyxwvuts zyx CYSEWSKI AND WILKE yield factors obtained during vacuum operation were always about 50% lower than the yield factors of 0.1 to 0.12 experienced during atmospheric pressure fermentations.2.3 The lower yield factors may be a direct result of increased maintenance energy requirements for yeast growth under vacuum. A lower cell yield factor was the only discernible difference between vacuum and atmospheric pressure fermentations. The results of a long term continuous vacuum fermentation are shown in Figure 5 for a 33.4y0 glucose feed. A constant bleed of fermented broth was withdrawn to maintain a concentration factor of 7.7. The cell mass concentration remained stable a t 50 g dry wt/liter for over 13 days of continuous operation, a t which point the experiment was terminated. With this concentration of yeast, the 33.4% glucose feed was fermented to less than 0.4% residual sugar in a mean fermentor residence time of 3.8 hr. This corresponded to a n ethanol productivity of 40 g/liter-hr. With conventional continuous fermentation a t atmospheric pressure using optimal conditions (pH = 4.0, T = 35"C, 10% glucose feed) the maximum ethanol productivity obtained with this yeast and similar fermentation media was 7.0 g/liter-hr.3 Thus, the vacuum system produced a n almost sixfold increase in ethanol productivity compared t o conventional continuous operation. The specific ethanol productivity in the vacuum fermentor was 0.8 hr-l. This is 38y0 higher than obtained for conventional continuous fermentations a t optimal conditions. The increase in zyxwvu I c .c a c .- c a zyxwv zyxwvut zyxwvuts - - - 20 I I I I I I I I I I *.I zyxw Fig. 5. Results of a long term continuous vacuum fermentation, 33.4% glucose feed. Oxygen bleed rate: 0.12 v/v/m; concentration factor: 7.7. (a) Cell mass, g/liter ; ( 0 )productivity, g/liter-hr. RAPID ETHANOL FERMENTATIONS 1135 specific productivity experienced in the vacuum system seems to be a direct result of lowering ethanol inhibition. The ethanol concentration in the fermentor was always below 10 g/liter, however, during atmospheric continuous operation the ethanol concentration was 46 g/liter for the optimal feed sugar concentration of 10%. If the effluent ethanol concentration was reduced from 46 t o 10 g/liter in the atmospheric fermentations, the specific productivity was increased from 0.58 to 0.8 hr-1.3 This is in direct support of the finding in the vacuum system. It should be remembered, however, that the primary evidence of eliminating ethanol inhibition is the ability to completely ferment a 33.4% glucose feed in the vacuum fermentor. This was not possible in atmospheric fermentations because of ethanol inhibition. The increase in ethanol productivity shown in Figure 5, after 100 hr of fermentation, was achieved by simultaneously increasing the fermentor bleed and feed rate, thus keeping the concentration factor a t 7.7. The productivity could not be increased above 40 g/liter-hr and still maintain stable operation. I n order t o further increase the productivity, a n increase in feed rate was necessary. But from the above discussion, a corresponding increase in bleed rate had to be made to keep the concentration of nonvolatile components a t a level compatible with the yeast. Simultaneously increasing the bleed rate and feed rate to maintain a constant concentration factor dictates a constant cell mass concentration within the fermentor. This fact may be predicted from a simple mass balance and is borne out by experimental results. Since the fermentor ethanol productivity is the product of the specific cell ethanol productivity and the cell mass concentration, the fermentor productivity is limited by the cell mass concentration obtainable a t any given concentration factor. Thus, for a concentration factor of 7.7, the maximum fermentor ethanol productivity is 40 g/liter-hr corresponding to a cell mass concentration of 50 g dry wt/liter and a specific ethanol productivity of 0.8 hr-l ( Y + in this experiment was 0.055). zy zy zyxw zyxwvu zyxwv Cell Recycle in Vacuum Fermentation I n order to remove inhibitory substances and increase concentration, a settler was used in conjunction with the system. The bleed stream from the fermentor was passed the settler and the settled cells returned to the fermentor. manner, a high concentration of cells was maintained in mentor a t high bleed rates. the cell vacuum through I n this the fer- CYSEWSKI AND WILKE 1136 zyxw The settler was not loo’% efficient and some cells were lost in the overflow of clarified product. At steady state the amount of cells lost in the overflow was equal to the amount of cells produced during fermentation. The cell Concentration was adjusted by changing the pumping rate of the recycle stream. The results of the settler-vacuum system are shown in Figure 6. A final cell mass of 124 g dry wt/liter was achieved resulting in a n ethanol productivity of 82 g/liter-hr. This is almost a twelvefold increase in productivity o v u that obtained in conventional continuous operation.3 The specific productivity of the yeast decreased from 0.8 to 0.66 hr-l whcn ccll recycle was used in the vacuum system. This, no doubt, reflccts that som(’ of the yeast died during the extensive recycling. The mcan residence time of the yeast in the fermentor was ten times that in conventional continuous operation. The maximum ethanol productivity of 82 g/liter-hr of thc vacuumrecycle system was limited by the capacity of the settler. When a n attempt was made to increase the productivity by increasing the fermentor through-put, the flow velocity in thc settler became higher than the settling velocity of the yeast. As a result, more cells were lost in the overflow stream than produced during fermentation and the cell mass concentration in the fermentor and the ethanol productivity rapidly declined. The extremely high cthanol productivity obtained with the vacuum-cell recycle system is a direct result of the high cell mass zyxwv I ’ z I zyxwvutsrqpo I I I I I I I I I I zyxwvut I80 - 140 -80 40 I* : 20 0 20 - 60 60 80 I00 la0 140 Time. hours 160 180 P - 0 Y 0 zyx -40 40 40 zy a00 220 240 260 Fig. 6. Results of the settler-vacuum system, 33.4% glucose feed. RAPID ETHANOL FERMENTATIONS 1137 concentration achieved with the recycle system. Whereas 11.0 g dry wt/liter of yeast cell mass are typically obtained in conventional atmospheric continuous culture, over 120 g dry wt/liter of yeast cell mass are obtained in the vacuum-cell recycle system. The reason such high cell densities were achieved in the vacuumrecycle system was the ability of the vacuum system to ferment a concentrated sugar solution. This permitted low flow velocities within the settler becausr a relatively low feed rate and hence a low bleed rate from the settler was required to achieve high productivities. A clarified liquid bleed rate corresponding to a fermentor dilution rate of only 0.23 hr-l was rcquirt.d to achieve a productivity of 82 g/liter-hr when the 33.4y0 glucose fwd was fermented to a concentration of 0.4%. This low flow rat(. allowed the settler to operate efficiently and produce a Concentrated cell recycle stream. As mentioned previously, above this flow rate the settler became inefficient during vacuum operation. A similar advantage would br experienced in an industrial vacuumcell recycle system employing a centrifuge rather than a settler. By fermenting a concentrated sugar solution, the through-put of the centrifuge would be reduced in a vacuum system which would lower both operating and capital costs of the centrifuge. The high productivities obtained in the vacuum system agree with the recent work of Ramalingham and Finn6 on vacuum fermentations. Howww, they used an ergosterol supplemented growth medium to eliminate the oxygen requirement of the yeast and did not employ cell recycle. The ethanol productivity reported by them was 12.5 g/liter-hr. This is much lower than the productivities reported here of 82 and 40 g/liter-hr for the vacuum system with and without cell recycle, respectively. The lower productivity reported by them may be a result of not pushing the vacuum system to its limit. The main emphasis of their work was to demonstrate that a 5Oy0 sugar feed could be fermented in a vacuum fermentor. zy zyxwvu zy h’flect of Oxygen on Vacuum Fermentation As noted by numerous workers, trace amounts of oxygen stimulate alcoholic fermentation r a t e ~ . ~ - 7 However, ,* there is an optimum oxygen tension above which fermentation rates are s u p p r e ~ s e d . ~ . ~ The optimum oxygen tension for the strain of Saccharomyces sp. used in this work was found to be 0.07 mmHg for atmospheric pressure operation after the yeast had been “adapted” to high oxygen tension^.^ It was not possible to measure the oxygen tension of the medium in the vacuum system, although this would have been very desirable. CYSEWSKI AND WILKE 1138 zyxw zyx When a n oxygen probe was put in the vacuum fermentor, a stable reading could not be obtained because of the intense boiling taking place. The optimal oxygen sparging rate was, however, determined for the vacuum system. The results are shown in Figure 7 for the fermentation of a 33.4% glucose feed. The data were obtained a t a concentration factor of 7.7. The ethanol productivity is plotted against the oxygen feed rate to the fermentor for various agitation rpm’s. The highest oxygen feed rate used was 0.37 v/v/m a t STP. This corresponded to 5.6 v/v/m in the fermentor because of gas expansion under vacuum. Above this oxygen feed rate foaming was extensive and interfered with the liquid level control system. The optimum oxygen feed rate for ethanol production was between 0.08 t o 0.14 v/v/m a t STP. At high agitation rates the ethanol productivity declined more rapidly as the oxygen sparging rate was increased. Both increasing the agitation and oxygen feed rate increased the mass transfer rate of oxygen into the medium. This undoubtedly increased the oxygen tension in the fermentor. The productivity curves in Figure 7 may then be viewed as analogous t o the ethanol productivities obtained for atmospheric operation presented in ref. 3. Trace amounts of oxygen stimulated ethanol 10- zyxw zyxwvu - Oxygen bleed (VVM ot operating conditions) 1.0 20 3.0 4.0 50 I I I I I zy I I t I l 60 l RAPID ETHANOL FERMENTATIONS 1139 zy zyx production but if the oxygen concentration became too high, the ethanol productivity decreased. The result of using an air feed rather than oxygen in the vacuum system is also shown in Figure 7. When air was sparged into the fermentor a t a rate of 0.26 v/v/m a t STP, or 4.0v/v/m a t operating conditions, the ethanol productivity substantially decreased after only 12 hr of operation. The datum a t 12 hr shown in Figure 7 does not represent a steady-state point. The productivity and cell mass concentration were declining. An oxygen feed was resumed because conditions of fermentor “washout” were feared. This points out the necessity of using pure oxygen instead of air to maintain a high enough oxygen transfer rate under vacuum to support yeast growth. Figure 8 illustrates the effect of oxygen on yeast viability in the vacuum system. If oxygen was not sparged into the fermentor, the viability of the yeast continually dropped. Whereas, a n oxygen feed rate of 0.12 v/v/m a t S T P maintained yeast viability above 95%. zyxwvut zy zyxwvut Cell Recycle in Atmospheric Pressure Fermentations The results of a continuous fermentation employing cell recycle are shown in Figure 9. A 10% glucose feed was used in these experiments to avoid ethanol inhibition. l . 3 Ethanol remained in the fermentation broth under atmospheric operation and was not 100 I 5 I I I I I 1 10 1 I I 1 I I I I 1 zyxwvu 15 I I 20 Cell mass, Q/I I zyxw I 25 I I 30 I I 35 Fig. 8. Effect of oxygen on yeast viability in the vacuum system, 10%glucose feed. (u)O2bleed, 0.12 v/v/m; ( 0 )anaerobic. 1140 zyxwv zyxw zy CYSEWSKI AND WILKE 50 40 zyx zy -5 0) Y) 0 -3 30 0 0 L c !O zy 8 c .> .c 0 z .u 2 0 n zyxwvut ? Fig. 9. Results of a continuous fermentation employing cell recycle. 10% glucose feed. boiled away as in the vacuum system. The oxygen tension in the fermentor was maintained a t 0.12 mmHg. The cell concentration was adjusted by changing the pumping rate of the cell recycle stream and the system was allowed to rtach steady state before samples were withdrawn for analyses. The data presented in Figure 9 definitely show that a n increase in ethanol productivity was realized by increasing the cell mass concentration in the fermentor with a recycle system. The maximum specific productivity of the yeast in the recycle system was identical to the specific productivity obtained with conventional continuous operation, 0.58 hr-l, a t conditions of complete substrate u t i l i z a t i ~ n . ~ However, a cell mass concentration of 50 g dry wtlliter or four times higher than without cell recycle was achieved. The net effect was a fourfold increase in fermentor ethanol productivity in the recycle system over conventional continuous operation. The yeast did not degenerate or lose viability in the recycle system. This is evident by the samp specific productivities obtained with or RAPID ETHANOL FERMENTATIONS 1141 zy without cell recycle. Also, yeast viability, as determined by methylene blue stain, remained over 967, for the duration of the 14 day experiment. The steep decrease in cell mass and ethanol productivity above a dilution rate of 0.75 hr-l was due, once again, to exceeding the capacity of the settler and not because of a loss of yeast viability. When the dilution rate was increased abovc 0.75 hr-l, the flow velocity in the settler became higher than the settling velocity of the yeast. As a result, more cells were lost in the overflow stream than generated during fermentation, and conditions of ((washout” were experienced. The maximum ethanol productivity obtained with atmosphericcell recycle operation was only about one third that obtained with vacuum operation. A dilute glucose feed, lo?,, was used with the atmospheric fermentation to avoid severe ethanol inhibition. l , 3 The low glucose concentration required high dilution rates and hence high flow velocities through the settler had to be used to achieve high ethanol productivities. The increased load on the settler over the vacuum-recycle fermentation, in which a 33.4% glucose feed was used, did not allow the settler to achieve as effective a separation of cells from the fermentation broth. This fact lowered the cell mass concentration in the recycle stream and thus lowered the cell density in the fermentor during atmospheric operation. The end results were lower ethanol productivities for the atmospheric operation as compared with the vacuum operation when the same size settler was employed. The recycle experiments were conducted to demonstrate the feasibility and advantages of cell recyclc operation for continuous ethanol production. Thc use of a settler was for experimental convenience only. I n an industrial operation a continuous centrifuge would most probably be employed. A centrifuge is not as sensitive to changing flow rates as is a settler and would produce a more stable operation. Also a higher cell mass concentration can be obtained in the recycle stream with a centrifuge. Thus, it may be possible to achieve higher ethanol productivities than shown in Figure 9 or Figure 6 with the use of a ccntrifuge. CONCLUSIONS An overwhelming advantage of the vacuum fermentor is the elimination of ethanol inhibition. This permits concentrated sugar solutions to be fermented a t extremely fast rates. By use of cell recycle in conjunction with the vacuum system, ethanol productivi- 1142 zyxwvuts CYSEWSKI AND WILKE ties of almost twelve times that obtained with conventional fermentations were achieved. The direct consequence of this increased productivity would bc a twelvefold reduction in fermentor volume required for an industrial ethanol fermentation. Another advantage of the vacuum system, owing to the systems ability to utilize highly concentrated sugar solutions, is the production of a concentrated ethanol product ( 16-20y0 ethanol). This high concentration of ethanol in the fermentation product will reduce distillation costs for the final recovery of 95y0 ethanol. I n this respect, when comparing the productivitiw of various fermentation schemes, distillation costs should be taken into account. Atmospheric pressure-cell recycle fermentations produced an increase in ethanol productivity of four times over the conventional continuous operation. This was about one third the productivity achieved with vacuum operation. The productivity of the atmospheric-cell recycle fermentation was limited by the low feed glucose concentration which had to be employed to avoid severe ethanol inhibition. The low substrate concentration increased the flow rate required through the settler and thus limited the cell density and volumetric fermentation rate in the atmospheric system. A major constraint of vacuum fermentation is the accumulation of nonvolatile components in the fermentor. As a result, a bleed of fermented broth must be continually withdrawn from the fermentor to maintain the concentration of nonvolatile components at a level which will not inhibit yeast growth and ethanol production. The required bleed rate will be set by the concentration of nonvolatile components in the fermentation substrate. Thus, experiments should be conducted using industrial fermentation media (i.e., molasses or hydrolysate sugars) before a process design can be finalized. As in atmospheric pressure fermentation, trace amounts of oxygen were found t o be a n important supplement for the alcoholic fermentation during vacuum operation. To satisfy the yeast oxygen requirement a low flow rate (0.12 v/v/m) of pure oxygen had to be sparged through the vacuum fermentor in order to maintain a viable and actively fermenting yeast population. It may, however, be possible to eliminate the pure oxygen requirement by employing a n aerobic atmospheric pressure fermentation stage preceding the vacuum fermentor. The aerobically grown yeast would then be fed to an anaerobic vacuum fermentor. The yeast would be able t o actively ferment during anaerobic conditions in the vacuum zyxw RAPID ETHANOL FERMENTATIONS zyx 1143 fermentor using the pool of unsaturated fats and lipids stored during aerobic growth.9 Vacuum and cell recycle alcoholic fermentations represent new approaches to an age-old fermentation process. Because fermentation derived ethanol may someday serve as a supplement to or even a replacement for conventional petroleum liquid fuels, new and improved fermentation processes are needed. The economic implications of the vacuum ethanol fermentation will be examined in a later paper. zy zyxwvut zyxwvu zyxwv zyxw This work is part of a general program on utilization of cellulose as a chemical and energy resource conducted under the auspices of the Energy Research and Development Administration. References 1. C. Baeua and C. R. Wilke, “Effect of Alcohol Concentration on Kinetics of Ethanol Production by S. eerevisiae,” presented at First Chemical Congress of the North American Continent, Mexico City, Mexico, December 1-5, 1976. 2. S. Aiba, M. Shoda, and M. Nagatave, Bioteehnol. Bioeng., 10, 845 (1968). 3. G. R. Cysewski and C. R. Wilke, Biotechnol. Bioeng., 18, 1297 (1976). 4. J. B. Sumner and G. F. Somers, Laboratory Experiments in Biological Chemistry, Academic, New York, 1944. 5. G. F. Townsend and C. C. Lindgren, Cytologiu, 18, 183 (1953). 6. R. K. Finn and A. Ramalingam, Biotechnol. Bioeng., 19, 583 (1977). 7. T. W. Cowland and D. R. Maule, J. Znst. Brew., 72, 480 (1966). 8. M. D. Akbar, P. D. Richard, and F. J. Moss, Biotechnol. Bioeng., 16, 455 (1974). 9. A. D. Haukeli and S. Lie, J . Znst. Brew., 77, 253 (1971). Accepted for Publication February 25, 1977