Appl Microbiol Biotechnol (1984) 20:100-104

Ethanol production in a hollow fiber bioreactor using Saccharomyces cerevisiae

Mohamed A. Mehaia and Munir Cheryan
Department of Food Science, 382D Agricultural Engineering Sciences Building, University of Illinois, 1304 W. Pennsylvania Avenue, Urbana, IL 61801, USA

Summary. A new approach for continuous production of ethanol was developed using a Hollow fiber f e r m e n t o r (HFF). Saccharomyces cerevisiae cells were packed into the shell-side of a hollow fiber module. Using 100 g/1 glucose in the feed gave an o p t i m u m ethanol productivity, based on total H F F volume, of 40 g ethanol/1/h at a dilution rate of 3.0 h -1. U n d e r these conditions, glucose utilization was 30%. H o w e v e r , at 85% glucose utilization the productivity was 10 g ethanol/1/h. This compares to batch f e r m e n t o r productivity of 2.1 g ethanol/1/h at 100% glucose utilization.

continuously separate and recycle the biocatalyst (Mehaia et al. 1984; Cheryan and Mehaia 1983; Deeslie and Cheryan 1981) or to entrap it within the m e m b r a n e separation module itself (Inloes et al. 1983; Mehaia and Cheryan 1984; Kohlwey and Cheryan 1981; R o y et al. 1982). This p a p e r discusses the development of a plug-flow type of m e m b r a n e bioreactor, based on hollow fiber m e m b r a n e technology, for the continuous fermentation of glucose to ethanol.

Materials and methods

Microorganism and feed medium. Saccharomyces cerevisiae NRRL-Y-132 was obtained from the U,S. Department of Agriculture, Peoria, IL, USA. It was maintained on malt extract-yeast extract-glucose-peptone (MYGP) slants (3 g malt extract, 3 g yeast extract, 50 g glucose, 5 g peptone and 20 g agar in 11 of water). All ingredients were obtained from Difco Laboratories, Detroit, Michigan, USA, Table i givesthe composition of the medium used in our studies. Batch fermentation. The batch experiments were performed in a New Brunswick Microferm Fermentor using 2 1 volume. The

Introduction
As world p e t r o l e u m reserves are depleted, new sources of carbon and hydrogen must be found to supply our chemical and energy needs. Large quantities of biomass are available in most parts of the world and could be used as an energy mechanism or as raw material for chemical manufacturing. The ancient art of alcohol fermentation has been d o c u m e n t e d for m a n y years, and the well-known fermentation processes for converting sugar to alcohol by using yeast and bacteria can be found readily in the past and current literature. Several new techniques have been developed for economic continuous ethanol production. These include immobilized microbial cells ( M c G h e e et al. 1982; Ghose and B a n d y o p h a d h y a y 1980; W a d a et al. 1980; Margaritis et al. 1981), cell recycle (Cysewski and Wilke 1977; G h o s e and Tyagi 1979; Rogers et al. 1980) and v a c u u m fermentation ( R a m a l i n g a m and Finn 1977; Cysewski and Wilke 1978). A n alternate approach involves the use of synthetic semipermeable m e m branes in the appropriate configuration to either
Offprint requests to: M. A. Mehaia

Table 1. The composition of the mediuma

Compound Glucose (anhydrous) Yeast extract (Difco) NH4C1 Na2HPO4 7 H20 KH2PO4 MgSO4 CaC12 Citric acid b Sodium citrate b Antifoam (Sigma, no. 5758) a All salts and glucose reagent grade b Citric-citrate buffer at pH 4.0 Amount g/1 of tap water 100 8 8 3.9 0.5 0.5 0.28 4.3 1.25 0.5 ml

M. A. Mehaia and M. Cheryan: Hollow fiber fermentor fermentor and its contents were sterilized in an autoclave at 120 C for 15 min. The medium was sterilized through a 0.2 ~tm microfilter (Gelman Acroflux Capsule) and added to the fermentor aseptically before inoculation. Nitrogen was passed through the fermentor vessel to maintain anaerobic conditions during these experiments. All experiments were conducted at 28 C.

101 Bulletin No. 332-UV (Sigma Chemical Co., St. Louis, MO) and by gas chromatography as described earlier (Mehaia et al. 1984). Glucose concentration was determined by the dinitrosalicylic acid (DNS) method (Summer and Somero 1949). Cell concentration is expressed as yeast cell dry weight per unit volume (g/l). The cell concentration was measured optically at 525 nm and converted to g/1 units using a calibration curve.

Hollow fiber fermentor (HFF). The overall scheme of the hollow

fiber fermentor system is shown in Fig. 1. There were basically two stages in the system: a sterilization stage, where the medium is "cold sterilized" by passing it through a cross-flow 0.2 ~tm microfilter (Gelman Acroflux Capsule) and a fermentation stage, where the sterilized medium was converted to ethanol in the hollow fiber fermentor. The hollow fiber fermentor was a hollow fiber module (currently designated the "short-short" hollow fiber cartridge by the manufacturer, Romicon, Inc., Woburn, MA, USA). It contained about 660 fibers each with a 50,000 molecular weight cut-off. Membrane surface area was 0.7 m 2. The shell-side volume was 430 ml, while the fiber volume was 195 ml, giving a total volume of 625 ml. The HFF was loaded with yeast cells by "backflushing" the unit through inoculation port P1 with a yeast suspension that had been grown on MYGP broth as described earlier (Mehaia et al. 1984).

Results and discussion

Batch fermentation
Table 2 shows batch fermentation results using 100 g/1 glucose. With an initial cell concentration of 2.6 g/1, fermentation was completed in about 24 h and the final ethanol concentration was 49.5 g/1 (about 6.1% v/v). Productivity varied from an optimum of about 4.3 g ethanol/1/h at 11 h of fermentation to about 2.1 g/1/h at the end of the fermentation. Maximum specific growth rate was 0.22 h -1 and the yield of conversion of glucose to ethanol was 95% of theoretical yield.

Cleaning procedure. After operation the hollow fiber module was submitted to the following cleaning cycle: (1) rinse with about 201 of distilled water; (2) backflushed with about 10 1of distilled water; (3) cleaned with Tergazyme (Alconox Inc., New York, USA) solution overnight; (4)rinsed with distilled water for 15rain; (5) stored in 1% formaldehyde. The hollow fiber module was rinsed with 20 1 of sterile water before re-use. A hollow fiber cartridge can be used several times. This particular cartridge has been used 50 times. Analyticalprocedures. Ethanol concentration was measured by the
alcohol dehydrogenase method as described in Sigma Technical

Continuous fermentation using HFF

All data reported here for HFF where obtained when the system reached steady-state conditions, after five reactor volumes of medium passed through the system. This is adequate since this bioreactor probably approximates plug-flow behavior (Kohlwey and Cheryan 1981). For comparison purposes, ethanol productivities for the HFF were evaluated using two different reactor volumes, as suggested by Inloes et al. (1983). In one approach, productivity was expressed in terms of the fiber volume (Fig. 2B). The fiber volume is calculated from the fiber outer diameter, the number of fibers and the fiber length between the epoxy end-seals. On this basis, the optimum productivity was 126 g ethanol/1/h. In the other approach, the total reactor volume was used, which includes the fiber

Feed Product FEED TANK BALANCE TANK

Sterilized

Medium

~_~!

Bleed !PI
Hollow Fiber

~c:

Sterilization

)1~:

Fermentation

Fig. 1. Schematic of the hollow fiber fermentor

Table 2. Batch fermentation by Saccharomyces cerevisiae using 100 g/1 glucose

Time (h) Glucose concentration (g/l) 100 91 76 42 0 0 0 Ethanol concentration (g/l) 0.0 4.4 11.2 28.4 47.9 49.0 49.5 Biomass (g/l) Yp/sa Productivity (g/l/h)

0 2 4 7 11 19 24

2.6 2.9 3.7 7.1 8.6 9.5 9.4

0.49 0.47 0.49 0.48 0.49 0.49

0.0 2.2 2.8 4.1 4.3 2.5 2.1

a Theoretical ethanol yield (YP/s) = 0.51

102

M. A. Mehaia and M. Cheryan: Hollow fiber fermentor

volume and the shell-side volume accessible to yeast cells between the epoxy end-seals. On a total volume basis (Fig. 2A), optimum productivity was 40 g/1/h at a dilution rate of 3 h -1. However, glucose utilization at this point was only 30%. Reducing dilution rate to 0.25 h -1 resulted in an increase in glucose utilization to 85% but reduced the ethanol productivity to 10 g/1/h.

80

40 30 20

4O

Figure 3 shows data from an experiment evaluating the long-term stability of the HFF. Feed concentration was 100 g/1 glucose and the dilution rate (D) was 0.5 h -1 (expressed in term of total HFF volume). The productivity during the long-run was about 17 g/t/h and glucose utilization was 66%. The shell-side was initially loaded with 100 g/1 yeast. After 93h of operation, the yeast concentration had increased to about 260 g/l, which probably represents an extremely close-packed arrangement in the shell-side (Inloes et al. 1983). This led to poor cell-substrate contact, inadequate mixing and relatively low productivity. However, productivity was still better than conventional fermentation systems (batch, continuous culture), see Table 3. Table 3 is a summary of ethanol productivity data obtained using different fermentation systems. It is

2c

J=

,
"r.

N
.~ 100
80 _= 60 B i60

80
20 m

20
PD a---O"

.=,

-6
.= 60

80

40
-

- i

Ethanol
-m

lo -' 2
, I , |

20

Ethanol

40

20

Glucose

O0

'

1'2

'1'o' ~o ' ~o' ~o '5~ ' go '/o

Time (h)

80

90

Dilution Rate (D), h - 1

Fig. 2A, B. Fermentation kinetics of hollow fiber fermentor in continuous (single-pass) mode. Feed was synthetic glucose medium (100 g/l). Initial cell concentration = 100 g/1. 0, Glucose; A, ethanol; vq, productivity. A Calculation of productivity based on the total volume of the fermentor. B Calculation of productivity based on fiber volume

Fig. 3. Long-term stability of hollow fiber fermentor. Feed was synthetic glucose medium (100 g/l). Initial cell concentration was 100 g/1 and dilution rate was 0.5 h -1. O, Glucose; A, ethanol; [3, productivity

Table 3. Comparison of ethanol productivity from different fermentation systems (based on total bioreactor volume) System Feed Sugar concentration (g/l) 100 250 150 160 100 197 100 150 100 100 150 50 89 100 Microorganism Ethanol concerttration (g/l) 49 102 75 31 40 71 44 74 44.5 49 71 24 12 40 Sugar utilization (%) 100 80 100 38 80 74 97 98 90 100 97 100 27 85 Productivity (g/l/h) 2,5 3,4 3,5 4.1 8.0 25 29 57 120 100 75 24 26 10 Reference

Batch

Glucose Glucose Lactose Glucose Glucose Glucose Glucose Glucose Glucose Glucose Lactose Lactose Glucose Glucose

S. cerevisiae Z. mobilis K. fragilis S. cerevisiae Z. mobilis S. cerevisiae Z. mobilis Z. rnobilis Z. mobilis S. cerevisiae K. fragilis K. fragilis S. cerevisiae S. cerevisiae

This work Rogers et al. (1980) Mehaia and Cheryan (1984) Ghose and Tyagi (1979) Rogers et al. (1980) Ghose and Bandyopadhyay (1980) Margaritis et al. (1981) Klein and Kressdorf (1983) Rogers et al. (1980) Mehaia and Cheryan (1984) Cheryan and Mehaia (1983) Mehaia and Cheryan (1984) Inloes et al. (1983) This work

Continuous culture Immobilized

Membrane recycle fermentor Hollow fiber fermentor

M. A. Mehaia and M. Cheryan: Hollow fiber fermentor

103

apparent that significant improvements in productivity can be made using continuous systems, especially immobilized cell and membrane bioreactors. Among membrane bioreactors, the membrane recycle fermentor appears to perform better than the hollow fiber fermentor. This could be due to several practical problems that arise during the operation of hollow fiber fermentors. Because the substrate and the biocatalyst are separated by a barrier, the rate-limiting step becomes the diffusion of substrate into the shell-side and diffusion of products back into the lumen (tube) side. We attempted to minimize diffusion limitations by applying a back-pressure of 70-100 kPa (10-15 psig) in the feed stream and leaving Port P2 slightly open to simultaneously remove the gas and a small fraction (about 1 - 5 % ) of the shell-side contents. Visual inspection, however, indicated that the yeast cells were mostly attached to the fibers or were localized to an annular volume just around the fibers, similar to what was observed by Roy et al. (1982). Thus our calculated productivity would have been much higher if we had used the "active" fermentor volume or the "fiber" volume instead of the total reactor volume to calculate dilution rate. The continually growing mass of cells gives rise to several problems. Since the cells far from the fibers were probably starved of nutrients and dying or undergoing lysis, cell debris must be continuously removed, such as through our bleed stream through Port P2. If not, the high cell density will eventually give rise to pumping problems and/or rupture the relatively sensitive fibers (Roy et al. 1982; Inloes et al. 1983). Attempting to slow down cell growth by limiting essential nutrients (Roy et al. 1982; Inoles et al. 1983) is counterproductive in growth-associated Gaden Type-I fermentations since it reduces the activity of the microorganisms, and thus lowers bioreactor productivity. When evaluating the data in Table 3, it should be borne in mind that Zymomonas mobilis has been reported to ferment sugars more rapidly and result in higher ethanol concentrations than yeasts (Rogers et al. 1982). Also, the higher productivity obtained with HFF in our earlier study with Kluyveromyces fragilis was due not only to higher sugar concentration (150 g/1 lactose) but also because we used a hollow fiber cartridge with a much higher fiber density (2,150 fibers vs 660 fibers in this study). This resulted in a higher surface area-to-volume ratio (17.8 cm2/cm3) than the cartridge used in the present trials (11.2 cm2/cm3), which probably improved substrate-cell contact and led to improved productivity. The performance of a hollow fiber fermentor for ethanol production can be improved by decreasing the flow rate of the glucose medium, applying some

back-pressure in the feed stream (to minimize diffusion limitations), continuously removing the gas and some of the shell-side content (to improve mixing) and by increasing the fiber density (a higher surface area-to-volume ratio) in the cartridge. Because of problems discussed above, the hollow fiber fermentor is not as efficient as membrane recycle fermentor. Unless these problems can be worked out, the hollow fiber fermentor will not be practical for ethanol fermentation with real feeds (of "natural" origin) since these problems may be magnified.
Acknowledgement. This research was supported in part by the Illinois Agricultural Experiment Station, University of Illinois, Urbana, USA.

References
Cheryan M, Mehaia MA (1983) A high-performance membrane bioreactor for continuous fermentation of lactose to ethanol. Biotechnol Lett 5 : 519 Cysewski GR, Wilke CR (1977) Rapid ethanol fermentation using vacuum and cell recycle. Biotechnol Bioeng 19:1125 Cysewski GR, Wilke CR (1978) Process design and economic studies of alternative methods for the production of ethanol. Biotechnol Bioeng 20:1421 Deeslie WD, Cheryan M (1981) A CSTR-hollow fiber system for continuous hydrolysis of proteins. Performance and kinetics. Biotechnol Bioeng 23 : 2257 Ghose TK, Bandyopadhyay KK (1980) Rapid ethanol fermentation in immobilized cell reactor. Biotechnol Bioeng 21 : 1387 Ghose T, Tyagi RD (1979) Rapid ethanol fermentation of cellulose hydrolysate. Batch versus continuous system. Biotechnol Bioeng 21:1387 Inloes DS, Taylor DP, Cohen SN, Michaels AS, Robertson CR (1983) Ethanol production by Saccharomyces cerevisiae immobilized in hollow-fiber membrane bioreactors. Appl Environ Microbiol 46 : 264 Klein J, Kressdorf B (1983) Improvement of productivity and efficiency in ethanol production with Ca-alginate immobilized Z. mobitis. Biotechnol Lett 5:497 Kohlwey DK, Cheryan M (1981) Performance of a fl-D-galactosidase hollow fiber reactor. Enz Microbiol Technol 3:64 Margaritis A, Bajpai PK, Wallace JB (1981) High ethanol productivities using small Ca-alginate beads of immobilized cells of Zymomonas mobiIis. Biotechnol Lett 3:616-618 McGhee JE, Julian GST, Detroy RW, Bothast RJ (1982) Ethanol production by immobilized Saccharomyces cerevisiae, Saccharomyces uvarum and Zymomonas mobil&. Biotechnol Bioeng 24:1155-1163 Mehaia MA, Cheryan M (1984) Hollow fiber fermentor for continuous production of ethanol. Application to the conversion of lactose by Kluyveromyces fragilis. Enz Microbiol Technol 6:117-120 Mehaia MA, Cheryan M (1984) Ethanol production in a membrane recycle bioreactor. Conversion of glucose using Saccharomyces cerevisiae. Proc Biochem (in press) Mehaia MA, Cheryan M, Argoudelis CJ (1984) Conversion of whey permeate to ethanol. Improvement of fermentor productivity using synthetic membrane. Cult Dairy Prod J (in press)

104 Ramalingam A, Finn RK (1977) The vacuferm process: a new approach to fermentation alcohol. Biotechnol Bioeng 19 : 585 Rogers PL, Lee KJ, Tribe DE (1980) High productivity ethanol fermentations with Zymomonas mobilis. Proc Biochem 15:7 Rogers PL, Lee KJ, Skotnicki ML, Tribe DE (1982) Ethanol production by Zymomonas mobilis. Adv Biochem Eng 23 : 37 Roy TBV, Blanch HW, Wilke CR (1982) Lactic acid production by Lactobacillus delbreukii in a hollow fiber fermentor. Biotechnol Lett 4 : 483

M . A . Mehaia and M. Cheryan: Hollow fiber fermentor Summer JB, Somero GF (1949) Dinitrosalicylic method for glucose. Lab Exp Biol Chem, Academic Press, New York Wada M, Kato J, Chibata I (1980) Continuous production of ethanol in high concentration using immobilized growing yeast cells. Eur J Appl Microbiol Biotechnol 11:67

Received November 18, 1983