Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440

Four automatic substrate feeding strategies were developed and investigated in this study to obtain rapid, repeatable, and reliable high cell densities of Pseudomonas putida KT2440 from glucose. Growth yield data of the key nutrients, Y _X/Glucose, Y _X/NH4, Y _X/PO4, Y _X/Mg, and Y _CO2/Glucose, were determined to be 0.41, 5.44, 13.70, 236, and 0.65 g g⁻¹, respectively. Although standard exponential feeding strategy worked well when the predetermined μ was set at 0.25 h⁻¹, an exponential glucose feeding strategy with online μ _max estimation resulted in a higher average biomass productivity (3.4 vs 2.8 g l⁻¹ h⁻¹). A CO₂ production rate based pulse glucose feeding strategy also resulted in good overall productivity (3.0 g l⁻¹ h⁻¹) and can be used as an alternative to pH-stat or DO-stat feeding. A cumulative CO₂ production based continuous feed with real-time cumulative glucose consumption estimation resulted in much higher biomass productivity (4.3 g l⁻¹ h⁻¹) and appears to be an excellent and reliable approach to fully automating high-cell-density fed-batch cultivation of P. putida.     

Lactic acid production from cellulosic waste by immobilized cells of Lactobacillus delbrueckii

Industrial waste corn cob residue (from xylose manufacturing) without pretreatment was hydrolyzed by cellulase and cellobiase. The cellulosic hydrolysate contained 52.4 g l⁻¹ of glucose and was used as carbon source for lactic acid fermentation by cells of Lactobacillus delbrueckii ZU-S2 immobilized in calcium alginate gel beads. The final concentration of lactic acid and the yield of lactic acid from glucose were 48.7 g l⁻¹ and 95.2%, respectively, which were comparative to the results of pure glucose fermentation. The immobilized cells were quite stable and reusable, and the average yield of lactic acid from glucose in the hydrolysate was 95.0% in 12 repeated batches of fermentation. The suitable dilution rate of continuous fermentation process was 0.13 h⁻¹, and the yield of lactic acid from glucose and the productivity were 92.4% and 5.746 g l⁻¹ h⁻¹, respectively. The production of lactic acid by simultaneous saccharification and fermentation (SSF) process was carried out in a coupling bioreactor, the final concentration of lactic acid was 55.6 g l⁻¹, the conversion efficiency of lactic acid from cellulose was 91.3% and the productivity was 0.927 g l⁻¹ h⁻¹. By using fed-batch technique in the SSF process, the final concentration of lactic acid and the productivity increased to 107.6 g l⁻¹ and 1.345 g l⁻¹ h⁻¹, respectively, while the dosage of cellulase per gram substrate decreased greatly. This research work should advance the bioconversion of renewable cellulosic resources and reduce environmental pollution.     

Application of oscillation for efficiency improvement of continuous ethanol fermentation with Saccharomyces cerevisiae under very-high-gravity conditions

Compared with steady state, oscillation in continuous very-high-gravity ethanol fermentation with Saccharomyces cerevisiae improved process productivity, which was thus introduced for the fermentation system composed of a tank fermentor followed by four-stage packed tubular bioreactors. When the very-high-gravity medium containing 280 g l⁻¹ glucose was fed at the dilution rate of 0.04 h⁻¹, the average ethanol of 15.8% (v/v) and residual glucose of 1.5 g l⁻¹ were achieved under the oscillatory state, with an average ethanol productivity of 2.14 g h⁻¹ l⁻¹. By contrast, only 14.8% (v/v) ethanol was achieved under the steady state at the same dilution rate, and the residual glucose was as high as 17.1 g l⁻¹, with an ethanol productivity of 2.00 g h⁻¹ l⁻¹, indicating a 7% improvement under the oscillatory state. When the fermentation system was operated under the steady state at the dilution rate of 0.027 h⁻¹ to extend the average fermentation time to 88 h from 59 h, the ethanol concentration increased slightly to 15.4% (v/v) and residual glucose decreased to 7.3 g l⁻¹, correspondingly, but the ethanol productivity was decreased drastically to 1.43 g h⁻¹ l⁻¹, indicating a 48% improvement under the oscillatory state at the dilution rate of 0.04 h⁻¹.     

Enhanced hydrogen production from glucose using ldh- and frd-inactivated Escherichia coli strains

We improved the hydrogen yield from glucose using a genetically modified Escherichia coli. E. coli strain SR15 (ΔldhA, ΔfrdBC), in which glucose metabolism was directed to pyruvate formate lyase (PFL), was constructed. The hydrogen yield of wild-type strain of 1.08 mol/mol glucose, was enhanced to 1.82 mol/mol glucose in strain SR15. This figure is greater than 90 % of the theoretical hydrogen yield of facultative anaerobes (2.0 mol/mol glucose). Moreover, the specific hydrogen production rate of strain SR15 (13.4 mmol h⁻¹ g⁻¹ dry cell) was 1.4-fold higher than that of wild-type strain. In addition, the volumetric hydrogen production rate increased using the process where cells behaved as an effective catalyst. At 94.3 g dry cell/l, a productivity of 793 mmol h⁻¹ l⁻¹ (20.2 l h⁻¹ l⁻¹ at 37 °C) was achieved using SR15. The reported productivity substantially surpasses that of conventional biological hydrogen production processes and can be a trigger for practical applications.     

Performance assessment of malolactic fermenting bacteria Oenococcus oeni and Lactobacillus brevis in continuous culture

The growth performance of malolactic fermenting bacteria Oenococcus oeni NCIMB 11648 and Lactobacillus brevis X₂ was assessed in continuous culture. O. oeni grew at a dilution rate range of 0.007 to 0.052 h⁻¹ in a mixture of 5:6 (g l⁻¹) of glucose/fructose at an optimal pH of 4.5, and L. brevis X₂ grew at 0.010 to 0.089 h⁻¹ in 10 g l⁻¹ glucose at an optimal pH of 5.5 in a simple and safe medium. The cell dry weight, substrate uptake and product formation were monitored, as well as growth kinetics, yield parameters and fermentation balances were also evaluated under pH control conditions. A comparison of growth characteristics of two strains was made, and this showed significantly different performance. O. oeni has lower maximum specific growth rate (μ_max=0.073 h⁻¹), lower maximum cell productivity (Q _x ^max=17.6 mg cell l⁻¹ h⁻¹), lower maximum biomass yield (Y _x/s ^max=7.93 g cell mol⁻¹ sugar) and higher maintenance coefficient (m _s=0.45 mmol⁻¹ sugar g⁻¹ cell h⁻¹) as compared with L. brevis X₂ (μ_max=0.110 h⁻¹; Q _x ^max=93.2 g⁻¹ cell mol⁻¹ glucose; Y _x/s ^max=22.3 g cell mol⁻¹ glucose; m _s=0.21 mmol⁻¹ glucose g⁻¹ cell h⁻¹). These data suggest a possible more productive strategy for their combined use in maturation of cider and wine.     

Continuous bio-catalytic conversion of sugar mixture to acetone–butanol–ethanol by immobilized <Emphasis Type="Italic">Clostridium acetobutylicum</Emphasis> DSM 792

Continuous production of acetone, n-butanol, and ethanol (ABE) was carried out using immobilized cells of Clostridium acetobutylicum DSM 792 using glucose and sugar mixture as a substrate. Among various lignocellulosic materials screened as a support matrix, coconut fibers and wood pulp fibers were found to be promising in batch experiments. With a motive of promoting wood-based bio-refinery concept, wood pulp was used as a cell holding material. Glucose and sugar mixture (glucose, mannose, galactose, arabinose, and xylose) comparable to lignocellulose hydrolysate was used as a substrate for continuous production of ABE. We report the best solvent productivity among wild-type strains using column reactor. The maximum total solvent concentration of 14.32 g L⁻¹ was obtained at a dilution rate of 0.22 h⁻¹ with glucose as a substrate compared to 12.64 g L⁻¹ at 0.5 h⁻¹ dilution rate with sugar mixture. The maximum solvent productivity (13.66 g L⁻¹ h⁻¹) was obtained at a dilution rate of 1.9 h⁻¹ with glucose as a substrate whereas solvent productivity (12.14 g L⁻¹ h⁻¹) was obtained at a dilution rate of 1.5 h⁻¹ with sugar mixture. The immobilized column reactor with wood pulp can become an efficient technology to be integrated with existing pulp mills to convert them into wood-based bio-refineries.     

Conversion of glycerol to pyruvate by Escherichia coli using acetate- and acetate/glucose-limited fed-batch processes

We report the conversion of glycerol to pyruvate by E. coli ALS929 containing knockouts in the genes encoding for phosphoenolpyruvate synthase, lactate dehydrogenase, pyruvate formate lyase, the pyruvate dehydrogenase complex, and pyruvate oxidase. As a result of these knockouts, ALS929 has a growth requirement of acetate for the generation of acetyl CoA. In steady-state chemostat experiments using excess glycerol and limited by acetate, lower growth rates favored the formation of pyruvate from glycerol (0.60 g/g at 0.10 h⁻¹ versus 0.44 g/g at 0.25 h⁻¹), while higher growth rates resulted in the maximum specific glycerol consumption rate (0.85 g/g h at 0.25 h⁻¹ versus 0.59 g/g h at 0.10 h⁻¹). The presence of glucose significantly improved pyruvate productivity and yield from glycerol (0.72 g/g at 0.10 h⁻¹). In fed-batch studies using exponential acetate/glucose-limited feeding at a constant growth rate of 0.10 h⁻¹, the final pyruvate concentration achieved was about 40 g/L in 36 h. A derivative of ALS929 which additionally knocked out methylglyoxal synthase did not further increase pyruvate productivity or yield, indicating that pyruvate formation was not limited by accumulation of methylglyoxal.     

Butanol production from wheat straw hydrolysate using <Emphasis Type="Italic">Clostridium beijerinckii</Emphasis>

In these studies, butanol (acetone butanol ethanol or ABE) was produced from wheat straw hydrolysate (WSH) in batch cultures using Clostridium beijerinckii P260. In control fermentation 48.9 g L⁻¹ glucose (initial sugar 62.0 g L⁻¹) was used to produce 20.1 g L⁻¹ ABE with a productivity and yield of 0.28 g L⁻¹ h⁻¹ and 0.41, respectively. In a similar experiment where WSH (60.2 g L⁻¹ total sugars obtained from hydrolysis of 86 g L⁻¹ wheat straw) was used, the culture produced 25.0 g L⁻¹ ABE with a productivity and yield of 0.60 g L⁻¹ h⁻¹ and 0.42, respectively. These results are superior to the control experiment and productivity was improved by 214%. When WSH was supplemented with 35 g L⁻¹ glucose, a reactor productivity was improved to 0.63 g L⁻¹ h⁻¹ with a yield of 0.42. In this case, ABE concentration in the broth was 28.2 g L⁻¹. When WSH was supplemented with 60 g L⁻¹ glucose, the resultant medium containing 128.3 g L⁻¹ sugars was successfully fermented (due to product removal) to produce 47.6 g L⁻¹ ABE, and the culture utilized all the sugars (glucose, xylose, arabinose, galactose, and mannose). These results demonstrate that C. beijerinckii P260 has excellent capacity to convert biomass derived sugars to solvents and can produce over 28 g L⁻¹ (in one case 41.7 g L⁻¹ from glucose) ABE from WSH. Medium containing 250 g L⁻¹ glucose resulted in no growth and no ABE production. Mixtures containing WSH + 140 g L⁻¹ glucose (total sugar approximately 200 g L⁻¹) showed poor growth and poor ABE production. Mention of trade names or commercial products in this article is solely for the purpose of providing scientific information and does not imply recommendation or endorsement by the United States Department of Agriculture.     

Continuous acetone–butanol–ethanol production by corn stalk immobilized cells

Corn stalk was used as a support to immobilize Clostridia beijerinckii ATCC 55025 in the fermentation process of acetone, butanol, and ethanol production. The effect of the dilution rate on solvent production was examined in a steady-state 20-day continuous flow operation. The maximum total solvent concentration of 8.99 g l⁻¹ was obtained at a dilution rate of 0.2 h⁻¹. Increasing the dilution rate between 0.2 and 1.0 h⁻¹ resulted in an increased solvent productivity, and the highest solvent productivity was obtained at 5.06 g l⁻¹ h⁻¹ with a dilution rate of 1 h⁻¹. The maximum solvent yield from glucose of 0.32 g g⁻¹ was observed at 0.25 h⁻¹. The cell adsorption and morphology change during the growth on corn stalk support were examined by the SEM.     

Continuous production of isopropanol and butanol using <Emphasis Type="Italic">Clostridium beijerinckii</Emphasis> DSM 6423

Clostridium beijerinckii DSM 6423 was studied using different continuous production methods to give maximum and stable production of isopropanol and n-butanol. In a single-stage continuous culture, when wood pulp was added as a cell holding material, we could increase the solvent productivity from 0.47 to 5.52 g L⁻¹ h⁻¹ with the yield of 54% from glucose. The overall solvent concentration of 7.51 g L⁻¹ (39.4% isopropanol and 60.6% n-butanol) with the maximum solvent productivity of 0.84 g L⁻¹ h⁻¹ was obtained with two-stage continuous culture. We were able to run the process for more than 48 overall retention times without losing the ability to produce solvents.     

Production of d-arabitol by a newly isolated Kodamaea ohmeri

A new yeast, isolated from natural osmophilic sources, produces d-arabitol as the main metabolic product from glucose. According to 18S rRNA analysis, the NH-9 strain belongs to the genus Kodamaea. The optimal culture conditions for inducing production of d-arabitol were 37 °C, neutral pH, 220 rpm shaking, and 5% inoculum. The yeast produced 81.2 ± 0.67 g L⁻¹ d-arabitol from 200 g L⁻¹ d-glucose in 72 h with a yield of 0.406 g g⁻¹ glucose and volumetric productivity Q_\textP Q_{\text{P}} of 1.128 g L⁻¹ h⁻¹. Semi-continuous repeated-batch fermentation was performed in shaker-flasks to enhance the process of d-arabitol production by Kodamaea ohmeri NH-9 from d-glucose. Under repeated-batch culture conditions, the highest volumetric productivity was 1.380 g L⁻¹ h⁻¹.     

Batch and repeated batch production of l(+)-lactic acid by Enterococcus faecalis RKY1 using wood hydrolyzate and corn steep liquor

Lactic acid production was investigated for batch and repeated batch cultures of Enterococcus faecalis RKY1, using wood hydrolyzate and corn steep liquor. When wood hydrolyzate (equivalent to 50 g l⁻¹ glucose) supplemented with 15–60 g l⁻¹ corn steep liquor was used as a raw material for fermentation, up to 48.6 g l⁻¹ of lactic acid was produced with, volumetric productivities ranging between 0.8 and 1.4 g l⁻¹ h⁻¹. When a medium containing wood hydrolyzate and 15 g l⁻¹ corn steep liquor was supplemented with 1.5 g l⁻¹ yeast extract, we observed 1.9-fold and 1.6-fold increases in lactic acid productivity and cell growth, respectively. In this case, the nitrogen source cost for producing 1 kg lactic acid can be reduced to 23% of that for fermentation from wood hydrolyzate using 15 g l⁻¹ yeast extract as a single nitrogen source. In addition, lactic acid productivity could be maximized by conducting a cell-recycle repeated batch culture of E. faecalis RKY1. The maximum productivity for this process was determined to be 4.0 g l⁻¹ h⁻¹.     

Lipid production by <Emphasis Type="Italic">Rhodosporidium toruloides</Emphasis> Y4 using different substrate feeding strategies

Microbial lipid is a potential alternative feedstock for the biodiesel industry. New culture strategies remain to be developed to improve the economics of microbial lipid technology. This work describes lipid production by the oleaginous yeast Rhodosporidium toruloides Y4 using a 15-l bioreactor with different substrate feeding strategies. Among these strategies, the intermittent feeding mode gave a lipid productivity of 0.36 g l⁻¹ h⁻¹, whereas the constant glucose concentration II (CC-II) mode gave the highest lipid productivity of 0.57 g l⁻¹ h⁻¹. The repeated fed-batch mode according to the CC-II mode was performed with a duration time of 358 h, and the overall lipid productivity was 0.55 g l⁻¹ h⁻¹. Our results suggested that substrate feeding modes had a great impact on lipid productivity and that the repeated fed-batch process was the most appealing method by which to enhance microbial lipid production.     

Glycolytic pathway and hydrogen yield studies of the extreme thermophile <Emphasis Type="Italic">Caldicellulosiruptor saccharolyticus</Emphasis>

NMR analysis of ¹³C-labelling patterns showed that the Embden–Meyerhof (EM) pathway is the main route for glycolysis in the extreme thermophile Caldicellulosiruptor saccharolyticus. Glucose fermentation via the EM pathway to acetate results in a theoretical yield of 4 mol of hydrogen and 2 mol of acetate per mole of glucose. Previously, approximately 70% of the theoretical maximum hydrogen yield has been reached in batch fermentations. In this study, hydrogen and acetate yields have been determined at different dilution rates during continuous cultivation. The yields were dependent on the growth rate. The highest hydrogen yields of 82 to 90% of theoretical maximum (3.3 to 3.6 mol H₂ per mol glucose) were obtained at low growth rates when a relatively larger part of the consumed glucose is used for maintenance. The hydrogen productivity showed the opposite effect. Both the specific and the volumetric hydrogen production rates were highest at the higher growth rates, reaching values of respectively 30 mmol g⁻¹ h⁻¹ and 20 mmol l⁻¹ h⁻¹. An industrial process for biohydrogen production will require a bioreactor design, which enables an optimal mix of high productivity and high yield.     

Increased erythritol production in fed-batch cultures of Torula sp. by controlling glucose concentration

The effect of glucose concentration on erythritol production by Torula sp. was investigated. The maximum volumetric productivity of erythritol was obtained at an initial glucose concentration of 300 g l⁻¹ in batch culture. The volumetric productivity was maximal at a controlled glucose concentration of 225 g l⁻¹, reducing the lag time of the erythritol production. A fed-batch culture was established with an initial glucose concentration of 300 g l⁻¹ and with a controlled glucose concentration of 225 g l⁻¹ in medium containing phytic acid as a phosphate source. In this fed-batch culture, a final erythritol production of 192 g l⁻¹ was obtained from 400 g l⁻¹ glucose in 88 h. This corresponded to a volumetric productivity of 2.26 g l⁻¹ h⁻¹ and a 48% yield. Journal of Industrial Microbiology & Biotechnology (2001) 26, 248–252. Received 26 September 2000/ Accepted in revised form 16 January 2001     

Analysis and optimization of a two-substrate fermentation for xylitol production using Candida tropicalis

Xylitol, a functional sweetener, was produced from xylose using Candida tropicalisATCC 13803. A two-substrate fermentation was designed in order to increase xylitol yield and volumetric productivity. Glucose was used initially for cell growth followed by conversion of xylose to xylitol without cell growth and by-product formation after complete depletion of glucose. High glucose concentrations increased volumetric productivity by reducing conversion time due to high cell mass, but also led to production of ethanol, which, in turn, inhibited cell growth and xylitol production. Computer simulation was undertaken to optimize an initial glucose concentration using kinetic equations describing rates of cell growth and xylose bioconversion as a function of ethanol concentration. Kinetic constants involved in the equations were estimated from the experimental results. Glucose at 32 g L⁻¹ was estimated to be an optimum initial glucose concentration with a final xylose concentration of 86 g L⁻¹ and a volumetric productivity of 5.15 g-xylitol L⁻¹ h⁻¹. The two-substrate fermentation was performed under optimum conditions to verify the computer simulation results. The experimental results were in good agreement with the predicted values of simulation with a xylitol yield of 0.81 g-xylitol g-xylose⁻¹ and a volumetric productivity of 5.06 g-xylitol L⁻¹ h⁻¹. Received 16 June 1998/ Accepted in revised form 28 February 1999     

Continuous butanol production with reduced byproducts formation from glycerol by a hyper producing mutant of <Emphasis Type="Italic">Clostridium pasteurianum</Emphasis>

Butanol, a four-carbon primary alcohol (C₄H₁₀O), is an important industrial chemical and has a good potential to be used as a superior biofuel. Bio-based production of butanol from renewable feedstock is a promising and sustainable alternative to substitute petroleum-based fuels. Here, we report the development of a process for butanol production from glycerol, which is abundantly available as a byproduct of biodiesel production. First, a hyper butanol producing strain of Clostridium pasteurianum was isolated by chemical mutagenesis. The best mutant strain, C. pasteurianum MBEL_GLY2, was able to produce 10.8 g l⁻¹ butanol from 80 g l⁻¹ glycerol as compared to 7.6 g l⁻¹ butanol produced by the parent strain. Next, the process parameters were optimized to maximize butanol production from glycerol. Under the optimized batch condition, the butanol concentration, yield, and productivity of 17.8 g l⁻¹, 0.30 g g⁻¹, and 0.43 g l⁻¹ h⁻¹ could be achieved. Finally, continuous fermentation of C. pasteurianum MBEL_GLY2 with cell recycling was carried out using glycerol as a major carbon source at several different dilution rates. The continuous fermentation was run for 710 h without strain degeneration. The acetone–butanol–ethanol productivity and the butanol productivity of 8.3 and 7.8 g l⁻¹ h⁻¹, respectively, could be achieved at the dilution rate of 0.9 h⁻¹. This study reports continuous production of butanol with reduced byproducts formation from glycerol using C. pasteurianum, and thus could help design a bioprocess for the improved production of butanol.     

1,3-Propanediol production in a two-step process fermentation from renewable feedstock

In this work, the production of 1,3-propanediol from glucose and molasses was studied in a two-step process using two recombinant microorganisms. The first step of the process is the conversion of glucose or other sugar into glycerol by the metabolic engineered Saccharomyces cerevisiae strain HC42 adapted to high (>200 g l⁻¹) glucose concentrations. The second step, carried out in the same bioreactor, was performed by the engineered strain Clostridium acetobutylicum DG1 (pSPD5) that converts glycerol to 1,3-propanediol. This two-step strategy led to a flexible process, resulting in a 1,3-propanediol production and yield that depended on the initial sugar concentration. Below 56.2 g l⁻¹ of sugar concentration, cultivation on molasses or glucose showed no significant differences. However, at higher molasses concentrations, glycerol initially produced by yeast could not be totally converted into 1,3-propanediol by C. acetobutylicum and a lower 1,3-propanediol overall yield was observed. In our hand, the best results were obtained with an initial glucose concentration of 103 g l⁻¹, leading to a final 1,3-propanediol concentration of 25.5 g l⁻¹, a productivity of 0.16 g l⁻¹ h⁻¹ and 1,3-propanediol yields of 0.56 g g⁻¹ glycerol and 0.24 g g⁻¹ sugar, which is the highest value reported for a two-step process. For an initial sugar concentration (from molasses) of 56.2 g l⁻¹, 27.4 g l⁻¹ of glycerol were produced, leading to 14.6 g l⁻¹ of 1.3-propanediol and similar values of productivity, 0.15 g l⁻¹ h⁻¹, and overall yield, 0.26 g g⁻¹ sugar.     

Fermentative production of superoxide dismutase with <Emphasis Type="Italic">Kluyveromyces marxianus</Emphasis>

This work sought to develop a fermentative process for the microbial production of superoxide dismutase (SOD), to overcome extraction from animal tissues. Twenty-eight wild-type yeast strains were screened for SOD productivity. Kluyveromyces marxianus L3 showed the highest SOD activity (62 U mg⁻¹) and was used for process development. Oxidative stress conditions and parameters affecting oxygen transfer rate were exploited to improve production. The effects of dilution rate (0.067 vs 0.2 h⁻¹), aeration pressure (0.3 vs 1.2 bar) and H₂O₂ (0 vs 50 mM) were studied during chemostat experiments. Low dilution rate, high pressure and H₂O₂ resulted in an increase in CuZn–SOD up to 475 U mg⁻¹. When a regulation of oxygen saturation was applied during batch cultures, CuZn–SOD was progressively higher at 60, 80 and 90% dissolved oxygen tension (DOT) (250, 330 and 630 U mg⁻¹, respectively). Furthermore, the highest growth rate and biomass yield were achieved at 90% DOT, this being therefore the best DOT condition for high overall productivity. Growth and productivity on different carbon sources were compared. Specific activity was higher on glycerol than on lactose or glucose (496, 454 and 341 U mg⁻¹, respectively). The highest biomass yield was achieved on lactose. It may be therefore the best substrate for SOD production.     

Diacetyl fermentation coupled with pervaporation using oleyl alcohol supported liquid membrane

Pervaporation using oleyl alcohol supported liquid membrane was successfully applied to diacetyl fermentation by immobilized lactic acid bacteria. Diacetyl productivity was about 10 g·m⁻³·h⁻¹, while productivity during batch fermentation was about 6 g·m⁻³·h⁻¹. Diacetyl yield from consumed glucose was about 0.04 g·g⁻¹ which was 4 times as large as that of batch fermentation. The pervaporation functioned favorably on actual fermentation broth. The flux of the permeate and the diacetyl separation factor for the pervaporation were about 9 g·m⁻²·h⁻¹ and 36, respectively, and these values were maintained at almost constant levels during fermentation. Diacetyl concentration in the permeate was about 2 kg·m⁻³, which is sufficiently high for commercial use.