Feasibility of Propionic Acid Production by Extractive Fermentation Zhong Gu, David A

Home , Propionic acid

Feasibility of propionic acid production by extractive fermentation Zhong Gu, David A. Rickert, Bonita A. Glatz, Charles E. Glatz

To cite this version:

Zhong Gu, David A. Rickert, Bonita A. Glatz, Charles E. Glatz. Feasibility of propionic acid production by extractive fermentation. Le Lait, INRA Editions, 1999, 79 (1), pp.137-148. hal-00929629

HAL Id: hal-00929629 https://hal.archives-ouvertes.fr/hal-00929629 Submitted on 1 Jan 1999

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Lait (1999) 79, 137-148 137 © Inra/Elsevier, Paris

Original article

Feasibility of propionic acid production by extractive fermentation

Zhong Gua, David A. Rickert", Bonita A. Glatz", Charles E. Glatza*

a Department of Chemical Engineering, Iowa State University, Ames. lA, USA b Department of Food Science and Human Nutrition, Iowa State University, Ames, lA, USA

Abstract - Production of propionic acid by fermentation is hindered by low productivity and product inhibition. Cell immobilization to increase productivity and extractive fermentation to reduce product inhibition were investigated. Propionic acid concentration in the extractive fermentation was maintained at 13 g·L-1 by concurrent extraction with a Iiquid extractant consisting of 40 % (v/v) Alamine" 304-1 (trilaurylamine) in Witcohol" 85 NF (oleyl alcohol). A final concentration of71 g-L-1 propionic acid was obtained in non extractive mode. Yields of propionic and acetic acids were dou- bled and higher overall productivities were obtained in the extractive fermentation. The extractant also exhibited selectivity for propionic acid over acetic acid, thus partially purifying the former. In both fermentation modes, productivity was enhanced by cell immobilization in calcium alginate beads. An economie analysis of a modified version of the fermentation based on several favorable assumptions showed that the extractive fermentation can, at best, approach economie feasibility at an annual production of 4.7 x 107 kg. For the assumed conditions, the production cost of the propionic acid was US$ 1.16·kg-'; this cost was reduced to US$ 0.94·kg-' when the value of the acetic acid byproduct was included. © Inra/Elsevier, Paris. extractive fermentation / propionic acid / propionibacteria / economie

Résumé - Étude de faisabilité de la production d'acide propionique par fermentation extractive. La production d'acide propionique par fermentation rencontre deux obstacles: la faible pro- ductivité et l'inhibition par le produit. L'immobilisation de cellule pour augmenter la productivité et la fermentation extractive pour réduire l'inhibition par le produit ont été étudiées. La concentration en acide propionique en fermentation extractive était maintenue à 13 g-L'" par extraction en co-cou- rant avec un liquide d'extraction composé à 40 % (v/v) d'Alumine" 304-1 (trilaurylamine) dans du Witcohol" 85NF (oleyl alcool). Une concentration finale de 71 g-L:' en acide propionique était obtenue en mode non-extractif. Les rendements en acides propionique et acétique étaient doublés et une productivité globale supérieure était obtenue en fermentation extractive. Le liquide d'extraction montrait de plus une sélectivité en faveur de l'acide propionique par rapport à l'acide acétique, conduisant à une purification partielle. Dans les deux modes de fermentation, la productivité était aug-

Oral communication at the 2nd Symposium on Propionibacteria, Cork, lreland, June 25-27, 1998. * Correspondence and reprints. [email protected] 138 Z. Gu et al.

mentée par l'immobilisation des cellules dans des billes d'alginate de calcium. Une analyse écono- mique de la version modifiée de la fermentation basée sur ces hypothèses les plus favorables a mon- tré que la fermentation extractive peut, au mieux, approcher la faisabilité économique pour une production annuelle de 4,7 x 107 kg. Dans les conditions envisagées, le coût de production de l'acide propionique était de 1,16 $·kg-1; ce coût était réduit à 0,94 $·kg-1 lorsque l'on incluait la valeur du coproduit acide acétique. © InraJElsevier, Paris.

fermentation extractive / acide propionique / bactérie propionique / économie

1. INTRODUCTION boxylic acids including propionic acid [l, 7,22-24]. Among separation methods, liq- Propionic acid is a commodity chemical uid-liquid extraction is the most widely stud- with uses in animal feed, grain preserva- ied. Several extractive fermentation pro- tion, antifungal agents (calcium and sodium cesses for propionic acid were developed salts), plasticizers (cellulose acetate propi- using several complex solvent systems [13, onate) and herbicides. The synthesis of pro- 16,21]. Solichien et al. [21] used a hollow- pionic acid has so far been dominated by fiber module for membrane extraction petrochemical routes, including the oxida- of propionic acid with 200 g-L-1 TOPO tion of propane, propionaldehyde, and (tri-n-octyl phosphine oxide, extractant) in propanol. This makes the product very vul- kerosene (diluent). This solvent system was nerable to the sudden priee fluctuations of nontoxic to the propionibacteria. However, propane and natural gas. On the other hand, the integrity of the hollow-fiber module was synthe sis by fermentation is able to utilize hard to maintain because of solvent crys- inexpensive and renewable biomass as sub- tallization at room temperature. Propionic strate [3, 9,10,12,14,17,25]. acid was also extracted with 100 g-L:" Commercialization of a propionic acid TOPO in n-decane (as solvent) in a flat fermentation process must overcome three sheet, supported liquid membrane (SLM) barriers. First, the fermentation is lengthy. apparatus by Ozadali et al. [16]. The stabil- A typical batch fermentation takes about ity of the SLM was limited by leaching of 3 d to reach 20 g·L-1 propionic acid, with the n-decane. Lewis and Yang [13] used a yield (the mass of the acid formed per unit 40 % (w/w) Alamine'" 336 in 2-octanol to mass of the substrate consumed) usually extract propionic acid via two modes of less than 60 % [8]. Second, the fermenta- extraction. With in situ extraction, direct tion is end-product-inhibited [26], which contact of the microorganism and the sol- limits the final propionic acid concentra- vent caused acid productivity to decrease; tion. Third, downstream separation and con- the fermentation was not inhibited by ex centration of the acid are expensive because situ extraction. of its low concentration (usually less than The economie feasibility of the propi- 1 60 g.L- propionic acid) and the presence onic acid fermentation process has been of acetic acid as a byproduct. The low addressed [2,13,15]. Lewis and Yang [13] volatility of propionic acid relative to water reported that 16400 kg of calcium propionate makes recovery by direct distillation prob- could be produced daily from 450 000 kg lematic. of whey for about US$ 0.34·kg-1 of prod- Integration of separation and fermenta- uet. However, they gave no details on down- tion as a means of overcoming sorne of these stream processing eosts. Clausen and Gaddy barriers has been considered for several car- [2] presented preliminary designs for a plant Propionie acid fermentation 139

with an annual production of3.2 x 107 kg at fermentations was 40 %. The fermenter and costs of US$ 0.46·kg-1 andUS$ 0.54·kg-1 accessory controls were previously described [20]. A constant pH at 6 or 7 was maintained via for acetic and propionic acids, respectively the automatie addition of6 mol-L:' NaOH. The (1981 values). Nishikawa et al. [15] stud- temperature was 32 "C and the culture was agi- ied co-production of other valuable prod- tated at ISO rpm. Glucose was monitoréd with a YSI enzymatic glucose/lactate analyzer (Model ucts such as vitamin B12 with propionic acid to improve the economie viability of the fer- 2700, Yellow Springs, Inc., Yellow Springs, OH) 1 mentation process. and was maintained between 35 and 75 g-L- by feeding appropriate amounts of a 500 g-L-1 glu- Previously we found that 40 % (v/v) cose solution. An aliquot of 10 x glucose-free Alamine" 304-1 (trilaurylamine) in FB was also fed every 24 h to replenish other Witcohol'" 85 NF (oleyl alcohol) was non- nutrients. toxic, provided good partitioning of propi- For extractive fermentation in 300 mL FB, onic acid into the solvent, and was compat- the solvent reservoir contained 350 mL of 40 % ible with free acid recovery by distillation (v/v) Alamine 304-1 in Witcohol85 NF, and was replaced every other day. The hollow-fiber mem- [6]. In the present study, this solvent sys- brane extractor consisted of 224 hydrophobie, tem is used in a fed-batch extractive fer- microporous polypropylene hollow fibers mentation with immobilized cells to avoid (Celgard'" X20-400, Hoechst Celanese Corp., product inhibition. Immobilized cells offer Charlotte, NC), each with an effective length of the advantages of high cell densities, elim- 25.4 cm, potted into a glass shell (Iowa State ination of the lag phase, and reduced expo- University Glass Blowing Shop, Ames, lA) with 0.0.,1.0., and length of 14,12.7, and 305 mm, sure to solvent. Fed-batch mode also avoids respectively. The total effective membrane sur- catabolite repression. face area was 716 cm-. Before the fermentation, the shell side of the membrane extractor was chemically sanitized with 500 mL of 3 % (v/v) 2. MATERIALS AND METHODS H202 and then rinsed with 800 mL sterile dei on- ized water. 2.1. Materials Extractive fermentation was started in non extractive fed-batch mode for 22 h at pH 6. At Propionibacterium thoenii P20 from the cul- this time, continuous circulation of the medium ture collection of the Oepartment of Food Sci- at 15 ml-rnirr ' on the shell side of the hollow- ence and Human Nutrition at Iowa State Uni- fiber membrane extractor was begun, while the versity was used. Seed cultures were grown in organie solvent circulated countercurrently at sodium lactate broth (NLB) and sodium lactate 5 ml.smirr ' on the tube side. In addition, the pH in agar (NLA) as described by Rickert et al. [20). the fermenter was allowed to fall to pH 5.5 and Fermentations were conducted in fermentation controlled there for better acid extraction. Solvent broth (FB) containing 75 g·L-1 glucose [20). For leakage from the membrane pores was prevented extractive fermentation, 18 mmol-L-1 calcium by applying back pressure (0.08-0.09 MPa) to chloride was added to FB to maintain alginate the shell side of the extractor. The hydrophobie bead integrity. The solvent for extractive fer- nature of the hollow fibers prevented the medium mentation was 40 % (v/v) Alamine 304-1 (tri- from penetrating into the solvent phase as long as laurylamine, Henkel Corp., Tucson, AZ) in Wit- the back pressure did not exceed approximately cohol 85 NF (> 90 % oleyl alcohol, Witco Corp., 0.13 MPa. Dublin, OH). To enumerate viable immobilized celIs, 8 beads were dissolved in 2 mL of sodium citrate (10 g-L-1) solution at room temperature, diluted, 2.2. Fermentation and plated on NLA; colonies were counted after 4 d of anaerobie incubation at 32 "C. Samples Strain P20 was immobilized in calcium algi- were taken from the fermenter for HPLC analy- nate beads (average diameter:l 2.5 mm in nonex- sis before and after glucose feedings. Before tractive and 2.7 mm in extractive processes) as analysis, solvent-phase samples were back- described by Rickert et al. [20). The bead load extracted with NaOH. Oetailed procedures were (% w/v of beads to fermentation medium) in ail described previously [21]. 140 Z. Gu et al.

Amounts of propionic and acetic acids were used. The fermentables were assumed to be glu- expressed as undissociated acids, and ail calcu- cose equivalents with costs comparable to those lations were performed using the molecular of hydrolyzed cellulose or byproducts such as weights of the undissociated acids. Il should be corn steep Iiquor. Fermentations utilizing corn noted that above pH 6, most of these acids will steep Iiquor as the substrate achieved similar pro- exist in their anion form. pionic acid production to that obtained from glucose [19]. The yield coefficients (mass of products produced per mass of fermentable substrate consumed) used were the values obtained from 2.3. Economie evaluation our earlier kinetic studies for a propionic acid concentration of approximately 3.0 g·L-1 [5]. The economie evaluation was carried out, The pH can be controlled in extractive fer- with sorne modifications on our part, using the mentation by removal of acid instead of addi- process design software package BioPro tion of base. Therefore, base consumption was Designer" (Intelligen Inc., Scotch Plaines, NI). A assumed to be negligible in the CUITentdesign. batch process with an assumed annual capacity of The costs of cell immobilization were evaluated 7 4.7 x 10 kg propionic acid was used as the basis based on the highest cell density achievable by for calculations. A 10-stage mixer-sellier extrac- free-cell fermentation and the initial immobi- tor replaced the hollow-fiber membrane extrac- Iized-cell density required by the production fer- tor used experimentally. This substitution was mentation at various bead usages (the number viewed as technically feasible because of the low of consecutive batches in which the beads can toxicity of the solvent for the cells [6]. Scaled-up be used in production fermentation without cost of the hollow-fiber extractor was too great to replacement). These costs were combined with be feasible for a commodity chemical such as the production fermentation costs to determine propionic acid. The mixer-sellier extractor was the overall process profitability. sized to maintain the fermenter propionic acid concentration at 3 g·L-1 and provide a contact time per stage that was suffi cie nt for 90 % stage efficiency in similar extractions [II]. 3. RESULTS The fermentation batch time was assumed to 3.1. Fermentation be 210 h, the longest duration tested experimentally. During the fermentation, the medium would be recycled through the mixer-settler extractor, Although growth of propionibacteria is where the acids are removed by the 40 % (v/v) optimum at pH 7, it has been previously Alamine 304-1 in Witcohol 85 NF solvent sys- observed that acid production by immobi- tem. The solvent extract from the extractor is lized cells is comparable at pH 6 or pH 7 continuously distilled under vacuum in an acid [20). Thus, non extractive fermentations were stripping column to recover the acids from the overhead. Any co-extracted water has been performed at both pH values for compari- neglected. The acids are further purified by dis- son with the extractive fermentation, which tillation to obtain the propionic and acetic acid was started at pH 6 and continued at pH 5.5 product streams. The solvent lost in distillation for better extraction performance. Figure 1 (0.01 %) is replaced with fresh solvent before shows acid production by nonextractive fer- recycle to the extractor. The bead-immobilized mentation at pH 6. After 168 h fermenta- cells for the production fermentation are sup- 1 plied by a separate cell immobilization process, tion, 71.0g·L-1 propionic acid and 16.1 g-L- which includes fermenters for growing seed cul- acetic acid were produced at pH 6; at pH 7, ture and a decanter-centrifuge for biomass recov- 66.2 g-L-1 propionic acid and 17.4 g·L-1 ery. acetic acid were produced by 209 h. The equipment design parameters and chem- Broth concentrations throughout a 202-h ical priees are detailed elsewhere [4]. Ail priees extractive fermentation are shown infigure 2. were values from 1996-1997, either available in The solvent-side profiles show that propi- the software or provided by the suppliers of the chemicals. For fermentation substrates, best case onic acid was extracted much faster than acetic acid (figure 3 and table 1); this choices of NH3 as the nitrogen source and low- cost 'fermentables' as the carbon source were resulted in partial purification of the propi- Propionic acid fermentation 141

70 ,- 60 ....J È? 50 c 0 ~ 40 ë Q) o 30 c 0 o 20

o 20 40 60 80 100 i20 140 160 180 Time (hours)

Figure 1. Substrate and products concentrations in the fermentation broth during the course of fed- batch fermentation at pH 6: Ce), glucose; C+), propionic acid; CÀ), acetic acid. Decreases in products concentration coinciding with glucose addition are the result of dilution. Figure 1. Concentration en substrat et produits dans le milieu de fermentation au cours d'une fer- mentationfed-batch à pH 6: Ce), glucose; C+), acide propionique ; CÀ), acide acétique. Les diminutions en concentration des produits coïncident avec l'addition de glucose et sont le résultat de la dilution.

onic acid. During each of the four solvent during the course of the fermentation. This replacement periods, the propionic acid confirms the tolerance of strain P20 to the extracted was always close to or more than extraction solvent system. Free cells did not 100 % of the amount produced in the fer- cIog the hollow fiberextractor nor interfere menter during the same period. This shows with organic acid extraction. that the membrane area/fermenter volume The overall performance of the extrac- of 2.4 cm--ml, -1 used was adequate for pro- tive fermentation compared to that of the pionic acid recovery. Less frequent solvent nonextractive process is shown in table II. replacement or a need to maintain lower Acid yields and productivities were higher propionic acid concentration in the fermen- in the extractive fermentation. The influ- tation would increase the membrane area ence of acid concentration on culture per- required. No solvent was detected in the formance can be seen by following the vari- final propionic acid solution. ation in productivity and yields (figure 4) As was the case for the nonextractive fer- during the course of the fermentations. mentation, no overall decrease of viable These data have been smoothed to elimi- cells was observed for either immobilized nate sorne of the variability resulting from or free cells; immobilized cell numbers the need to calculate concentration differ- remained above 1010per mL and viable free ences over short time periods. Fluctuations cells increased from about 108 to 109 per mL that remain are not considered significant 142 Z. Gu et al.

:s 60 -9 c 0 50 ~ ë 40 Q) o c 0 30 Ü 20

020 406080100120140160180200220 Time (hours) Figure 2. Substrate and products concentrations in the fermentation broth during the course of the extractive fermentation. Extractive mode began at 22 h. (e), glucose; (+), propionic acid; (.Â.), acetic acid. Figure 2. Concentrations en substrat et produits dans le milieu de fermentation au cours d'une fermentation extractive. Le mode extractif commençait à 22 h. (e), glucose; (+), acide propionique ; (.Â.) acide acétique. but are likely due to unsteady state condi- duced in 48 h of free-cell fermentation tions in the fermenter combined with exper- (unpublished data). The initial cell density in imental error in analyses. For the extractive au immobilized-cell fermentation with 40 % fermentation, differences were summed both (w/v) bead load is about 24 g-L:'. There- from the broth and from the solvent reser- fore, if beads are used for just a single fer- voir. Productivity and yields were quite sta- mentation, it would require four seed fer- ble during the extractive fermentation, but menters each producing five batches of free decreased steadily during non extractive fer- cells to pro duce enough biomass for an mentation. As the acid con centration built immobilized-cell fermenter of the same size. up, performance declined. An added bene- We have assumed reuse of immobilized fit of the extractive process was that the cells for five consecutive batches of 21O-h extraction of acids to the solvent phase fermentation. reduced base consumption for fermentation An acid productivity of 0.142 g-Lv-lr ' pH control by 80 %. with 1.45 % (w/v) bead load was achieved at 2.77 g-L-1 propionic acid concentration in 3.2. Economie evaluation a previous study [5]. For this analysis we assumed that the same productivity per bead Two major favorable extrapolations have volume would be maintained at 40 % (w/v) been made for the model process, We have bead load, giving a productivity of3.9 g-L-1·h-1 observed that 1.23 g-L-1biomass cau be pro- based on fermenter volume. Propionic acid fermentation 143

24 22 20 :s 18 .9 16 c o 14 ~ ë 12

c~ 10 8 8 6 • 4 / 2

o """'-"=-.!.---'---L--'----'----'---'---,---,----,---,----,---"-----,----,-----,---l.J 40 60 80 100 120 140 160 180 200 Time (hours)

Figure 3. Products concentrations in the solvent reservoir during the course of the extractive fermentation. Breaks in the curves indicate points at which solvent was replaced: (+), propionic acid;. (À), acetic acid. Figure 3. Concentrations en produits dans le réservoir de solvant au cours de la fermentation extractive. Les cassures dans les courbes indiquent les points où le solvant était remplacé. (+), acide propionique ; À), acide acétique.

Table I. Efficiency of acid extraction relative to acid production. Tableau I. Efficacité de l'extraction d'acide relative à l'acide propionique.

Time Period- (h) % Propionic Acid extracted" % Acetic Acid extracted"

22-76.5 97.1 37.7 76.5-18 138.6 84.8 118-166 99.2 81.8 166-202 106.1 87.1

-Period of the fermentation in which a given batch of solvent was circulating in the extractor. b In the given period, the percentage of propionic or acetic acid produced that was extracted. "Période de la fermentation au cours de laquelle une quantité donnée de solvant circulait dans l'extracteur. b Dans la période, % d'acides propionique et acétique extraits.

The resulting economie evaluation is 61 % is for the fermenters and 16 % for the shown in table III. Details of major equip- extractor. Based on the annual operating ment specifications and purchase costs as costs of cell immobilization and production, well as annual operating costs are presented including annual depreciation of investment elsewhere [4]. Of the total equipment cost, at 9.5 %, the profitability analysis indicates ---l

144 Z. Gu et al.

Table II. Overall performance of extractive" and non-extractive fermentation. Tableau II. Performance globale de fermentation extractive" et non extractive.

Extractive Non extractive fermentation fermentation pH = 6.0 pH = 7.0

Immobilized cell density" (cfu-rnl..") 1.3 X IO!' 1.5 X 1011 I.4XJ011 Acetic acid yield (g.g-l glucose) 0.12 0.04 0.05 Propionic acid yield (g.g-l glucose) 0.43 0.20 0.19 Acetic acid volumetrie productivity (g-Lrl-h ") 0.12 0.10 0.08 Propionic acid volumetrie productivity (g-L-1·h-l) 0.46 0.42 0.32

-Jncluding the performance in the first 22 h non extractive mode. b Average viable cell counts of immobilized cells per medium volume. a Comprenant la performance au cours de 22 premières heures en mode non extractif. b Moyenne des dénombrements de cellules immobilisées par volume de milieu.

Table III. Profitability analysis. that the overall proeess becomes profitable Tableau III. Analyse de profit. at production costs of US$ 1.16·kg-1 of propionic acid. When credit for aeetic acid (US$ Total investment 0.84·kg-1) is taken into account, the fer- (thousand US$·y-l) 146671 mentation reaches marginal profitability at the CUITentmarket priee for propionic acid of Revenue streams flowrate US$ 0.94·kg-l. The retum on investment of Propionic acid (thousand kg.y') 47203 the proeess before tax is 2.7 %, at that point. Acetic acid (thousand kg-y ") 16561

Production unit cost- 4. DISCUSSION Propionic acid (US$·kg-1) 1.16 4.1. Fermentation Selling priee Propionic acid (US$·kg-l) 0.92 Acetic acid (US$·kg-1) 0.84 The results from the pH 6 fermentation surpass those reported by Paik and Glatz Revenue [17] for a similar study of propionic acid Propionic acid (thousand US$-y-l) 48807 fermentation with calcium alginate-immo- Acetic acid (thousand US$-y-l) 15302 bilized cells. In their work, lower overall Annual operating cost* (thousand productivities of 0.26 and 0.05 g-L -I·h-I US$-y-l) 60084 were achieved for propionic and aeetic acids, Gross profit (thousand US$-y-I) 718 respectively. A different bacterial strain and onlya 10 % (w/v) bead load were used in a Including the annual costs for cell-immobilization their fed-batch fermentation. In addition, process (assuming immobilized cells can be used for 5 the glucose concentration in the medium consecutive batches of a 210 h fermentation) and depre- was exhausted between feedings to 20 g·L-I. ciation of9.5 % annually. The differenees in bead load and prevailing "Comprenant le coût annuel du procédé d'immobilisation des cellules (en considérant que les cellules glucose concentration were probably the immobilisées peuvent être utilisées pour 5 batchs con- major contributing factors to the differenees sécutifs de 210 h de fermentation) et une dépréciation in performance between the previous [17] annuelle de 9,5 %. and the CUITentfermentation. However, acid Propionic acid fermentation 145

inhibition was still seen in the CUITentfer- cess only reaches the break-even point for mentation. Propionic acid productivity and CUITentacid priees. Our favorable assump- propionic and acetic acid yields decreased as tions are justified. We observed that beads propionic acid accumulated in the medium were undamaged and viable cell concentra- (figure 4). Sorne of the glucose was diverted tions remained high at the end of the 202-h to succinate, which accumulated to approx- extractive fermentation process. Therefore, imately 21 g·L-I in the pH 6 fermentation. multiple usage of the beads is possible, The propionic/acetic acid ratio decreased although we have not done so under these with increasing propionic acid concentra- conditions. Beads have been successfully tion, which suggests that propionic acid pro- reused in repeated batch fermentations [19]; duction was more strongly affected by acid up to 20 12-h batches were carried out with- accumulation than was acetic acid produc- out significant deterioration of the beads. tion. The high concentration of acids in fed- However, there is no process that we are batch fermentation was achieved at the cost aware of for producing calcium alginate of inefficient substrate use and lower pro- beads at the required scale. ductivity. The high acid productivity (3.9 g-Lv-lr") Improved performance was observed assumed for 40 % (w/v) load of immobi- with extractive fermentation. Through con- lized cells at 2.77 g-L-1 propionic acid continuous removal of propionic acid from the centration [5] significantly reduced the vol- fermenter with liquid-liquid extraction, the ume of fermenters required for industrial- propionic acid concentration was maintained scale production; hence, reasonable fer- at about 13 g·L-I. The overall acid yields menter cost was achieved. Cost dictates use were more than double those of the nonex- of a mixer-settler extractor rather than the tractive process. Only 1.6 g-L-lof succinate membrane extractor used for our experi- were produced during the extractive fer- ments. mentation. According to the stoichiometry of Because the production cost of propionic glucose conversion to propionic acid through acid is higher than its selling priee (US$ succinate [18], this decrease in succinate 1.16·kg-1 vs US$ 0.94·kg-I), co-production production would account for 15 % of the of acetic acid is very important to the pro- propionic acid yield improvement. cess' economies, In the current study, the The lower acid concentration of the extrac- costs of substrates account for over 20 % tive fermentation also benefited produc- of the ove rail operating costs; this would be tivity (except in the initial 22 h, figure 4A), higher if glucose at CUITentmarket priee was even though the cell density was slightly used. Therefore, the use of low-cost sub- lower. A similar effect has been observed strate without propionic acid productivity during repeated batch fermentations with decline is very important for process feasi- 40 % (w/v) bead-immobilized strain P20, bility. Research on substrates such as corn where propionic aicd concentration never steep liquor, sulfite waste liquor and whey- exceeded 15 g·L-I [20]; up to 4.06 g-L-1·h-1 based substrates has demonstrated the fea- propionic acid productivity was achieved sibility ofsuch alternatives [3, 9,10,12,14, in that case. Response to lower acid con- 17,25]. A common characteristic of these centrations also was observed in fermenta- processes is that useful products such as tion kinetic studies with the same bead- organic acids can be produced from a waste immobilized P20 cells [5]. material with a subsequent reduction in waste treatment cost. Sale of biomass for 4.2. Economie evaluation animal feed would further improve the eco- nomics. The economie analysis demonstrates that In conclusion, extractive fermentation even with favorable assumptions, the pro- was successfully conducted using irnmobi- 146 Z. Gu et al.

2.0 A ,. 1.8 s: 1.6 ~ ÉJ 1.4 ~ 's TI 1.2 :::J "e 1.0 0.. «"1} 0.8 o ·ë 0.6 0 "Ci e 0.4 • 0.. • 0.2 • À À 0.0 0 20 40 60 80 100120140160180200220 T1me (hours)

0.8 B 0.7

0.6 À • ID en • • 0 o 0.5 :::J • a • r 0.4 Cl À ÉJ 0.3 "Qi >= 0.2

0.1 À À • • • • 0.0 0 20 40 60 80 100 120J40 160 180200220 Tirne (hours)

Figure 4. A) Propionic acid productivity during extractive (e) and non extractive (.) fermentation at pH 6.0. B) Variation in product yield during extractive fermentation [(e), propionic acid; (.), acetic acid] and non extractive fermentation [(.), propionic acid; (+), acetic acid] at pH 6. Figure 4. A) Productivité en acide propionique au cours de fermentation extractive (e) et non extractive (.) à pH 6,0. B) Variation du rendement en produit au cours de la fermentation extractive [(e), acide propionique; (.), acide acétique] et non extractive [(.), acide propionique; (+), acide acétique] à pH 6. Propionic acid fermentation 147 lized ceUs and 40 % (v/v) Alamine 304-1 [6] Gu Z., Glatz B.A., Glatz C.E., Propionic acid in Witcohol 85 NF as the solvent. Higher production by extractive fermentation. 1. Sol- vent considerations, Biotechnol. Bioeng. 57 acid yields and productivities were achieved (1998) 454-461. with the maintenance of low concentrations [7] Hatzinikolaou D.G., Wang H.Y., Extractive fer- of propionic acid in the medium by extrac- mentation systems for organic acids production, tion. Additional advantages included reduc- Cano J. Chem. Eng. 70 (1992) 543-552. tion of base consumption by 80 % and selec- [8] Hsu S.-T., Yang S.-T., Propionic acid fermentation of lactose by Propionibacterium acidipro- tive extraction of propionic acid over acetic pion ici: Effects of pH, Biotechnol. Bioeng. 38 acid. Economie analysis indicated that sev- (1991) 571-578. eral favorable outcomes must be realized [9] Humphrey A.E., Fermentation technology, simultaneously for even marginal prof- Chem. Eng. Progr. 73(5) (1977) 85-91. itability. Whole-ceU extraction in a mixer- [10] Jain D.K., Tyagi R.D., KIuepfel D., Agbebavi TJ., Production of propionic acid from whey ultra- settler extractor, multiple uses of the bead- filtrate by immobilized ce Ils of Propionibac- immobilized ceUs, scale-up of productivity, terium shermanii in batch process, Process use of an inexpensive substrate, and co-pro- Biochem. 26 (1991) 217-223. duction of acetic acid were all necessary. [11] Karr AE., Scheibel E.G., Mass transferbetween immiscible liquids in continuous flow in an agi- tated chamber, Chem. Eng. Progr. Symp. Ser. 50 (10) (1954) 73-92. ACKNOWLEDGEMENTS [12] Lee I.H., Frederickson A.G., Tsuchiya H.M., Diauxic growth of Propionibacterium shermanii, Appl. Microbiol. Biotechnol. 28 (1974) This work was supported in whole or part by 831-835. the Iowa Corn Promotion Board, the Consortium [13] Lewis V.P., Yang S.-T., A nove! extractive fer- for Plant Biotechnology Research Inc., by the mentation process for propionic acid produc- Department of Energy cooperative agreement tion from whey lactose, Biotechnol. Progr. 8 no. DE-FCOS-920R220n, the New Energy Com- (1992) 104-110. pany of Indiana, and the United States Depart- [14] Martin M.E., Wayman M., GrafG., Fermenta- ment of Agriculture through the Biotechnology tion of sulfite waste liquor to produce organic Byproducts Consortium. This support does not acids, Cano J. Microbiol. 7 (1961) 341-346. constitute an endorsement by any of these groups [15] Nishikawa M., Branion R.M.R., Pinder KL., of the views expressed in this article. Strasdine G.A., Fermentation of spent sul phi te liquor to produce acetic acid, propionic acid, and vitamin B12, Pulp Pap. Cano 71 (3) (1970) 31-36. REFERENCES [16] Ozadali F., Glatz B.A, Glatz C.E., Fed-batch fermentation with and without on-line extrac- [1] Boyaval P., Seta J., Gavach c., Concentrated tion for propionic and acetic acid production by propionic acid production by electrodialysis, Propionibacterium acidipropionici, Appl. Enzyme Microb. Technol. 15 (1993) 683--686. Microbiol. Biotechnol. 44 (1996) 710-716. [2J Clausen E.C., Gaddy J.L., Fermentation of [17] Paik H.-D., Glatz B.A., Propionic acid produc- biomass into acetic and propionic acids with tion by immobilized cells of a propionate-tolerant strain of Propionibacterium acidipropi- Propionibacterium acidi-propionici, in: Moo- on ici, Appl. Microbiol. Biotechnol. 42 (1992) Young M., Robinson C.W. (Eds.), Advances in 22-27. Biotechnology, vol. II. Pergamon, Toronto, 1981, pp. 63-69. [18] Playne M.J., Propionic and butyric acids, in: Moo- Young M. (Ed.), Comprehensive Biotech- [3] Czaczyk K, Trojanowska K, Wiszumirska E., nology, vol. 3, Pergamon, New York, 1985, Hydrolyzed whey as a medium for propionic pp. 731-759. acid fermentation, Acta Biotechnol. 16 (1996) 175-183. [19] Rickert D.A., Evaluation of efforts to improve organic acid production by calcium alginate- [4] Gu Z., Process development of propionic acid immobilized propionibacteria, M.Sc. thesis, production by fermentation, Ph.D. dissertation, Iowa State University, Ames, Iowa, USA, 1996. Iowa State University, Ames, Iowa, 1997. [20] Rickert D.A., Glatz C.E., Glatz B.A., Improved [5] Gu Z., Glatz B.A., Glatz C.E., Effects of pro- organic acid production by calcium alginate- pionic acid on propionibacteria fermentation, immobilized propionibacteria, Enzyme Microb. Enzyme Microb. Technol. 22 (1998) 13-18. Technol. 22 (1998) 409-414. 148 Z. Gu et al.

[21] Solichien M.S., O'Brien D., Hammond E.G., [24] Weier AJ., Glatz B.A., Glatz c.E., Recovery Glatz C.E., Membrane-based extractive fer- of propionic and acetic acids from fermentation mentation to produce propionic and acetic acids: broth by electrodialysis, Biotechnol. Progr, 8 Toxicity and mass transfer considerations, (1992) 479-485. Enzyme Microb. Technol. 17 (1995) 23-31. [25] Wood H.G., Werkman C.H., The utilization of agricultural byproducts in the production of pro- [22] Tung L.A., Recovery of carboxylic acid at pH pionic acid by fermentation, J. Agr. Res. 49 greater than pK , Ph.D. dissertation, University a (1934) 1017-1024. ofCalifornia, Berkeley, CA, USA, 1993. [26] Woskow S.A., Glatz B.A., Propionic acid pro- [23] Vickroy V.T., Lactic acid, in: Blanch H.W., duction by a propionic acid-tolerant strain of Drew S., Wang D.I.C. (Eds.), Comprehensive Propionibacterium acidipropionici in batch and Biotechnology, vol. 1, Pergamon, New York, semicontinuous fermentation, Appl. Environ. 1985, pp. 761-776. Microbiol. 57 (1991) 2821-2828.