Emulsion Stability and Demulsif ication during the Direct Extraction of Penicillin Broth

Csaba Szabo* BIOGA L Pharmaceutical Works, Fermentation Division, H-4042 Debrecen, Hungary

Received September 10, 1991M ccepted January 2 4 1992

In this article, we present the factors which influence the stability of the emulsion forming under the circumstances of the direct extraction of penicillin fermentation broth and the conditions of demulsification. A laboratory method has been developed for testing demulsifiers and a quantitative evaluation of their properties determining the separation of phases has been worked out as well. It has been found that an emulsion of oil in water (o/w) forms and that this is very stable; it can be broken only by the presence of demulsifier and by using a field of centrifugal force. We have discovered that the most effective demulsifiers are cation-active sur- factants for 1-4 g/L broth concentration. Furthermore, it is advantageous if the demulsifier contains a nonionic surfac- tant and wetting agent as well. From the 19 surface-active compounds studied in the laboratory, we found that the NADAR 107 A and ARMOGARD D-5306 demulsifiers con- tained the best properties for the extraction of the penicil- lin broth. Key words: penicillin solvent extraction demulsification

INTRODUCTION

Several processes are known for the recovery of penicil- lin from fermentation broth. Nevertheless, basic plant scale technology is used in the extraction method. Par- allel with the traditional method for the extraction of the filtered fermentation broth, from the 1980s, the direct extraction of fermentation broth has also been applied in plant scale. The latter process concerning extraction undoubtedly has unfavorable side effects, but from the point of view of the whole technology, enough ad- vantages can be gained.’ With the spread of direct ex- traction of broth, the moves to develop new effective demulsifiers and to choose the right one for the given broth, have become more and more important. For the extraction, the appropriate demulsifier has to be chosen carefully by penicillin producers, i.e. theoretical consid- erations must be taken into account, and follow-up ex- periments performed.

In this study we review the reasons for forming a stable emulsion during the extraction of penicillin fer- mentation broth containing mycelium and address the issue of the choice of appropriate demulsifiers based on laboratory experiments.

* To whom all correspondence should be addressed.

EMULSION STABILITY AND DEMULSIFICATION

The formation of the high emulsion stability comes from the fact that in the fermentation broth, in compo- nents dosed in the culture medium, and in the interme- diates and decomposition products of the biosynthesis there are several agents with surface-active properties and stabilizing macromolecules. Furthermore there is a great deal of solid phase. This is an oil-in-water (o/w) emulsion, and its stability is determined by the factors mentioned below.

Surface-active Compounds

The emulsion is formed mainly by anion-active surfac- tants. The water-soluble salts of fatty acids are the most characteristic representatives of the anion-active agents in the broth. These salts are the enzymatic decomposi- tion products of the lipids from vegetable oils principally put in as antifoaming agents. In this medium phospho- lipids are also important surfactants.

Macromolecules

The macromolecules in the fermentation broth, like the proteins and decomposition products of proteins, are not typical surfactants but are the characteristic stabilizers of o/w emulsions.* The broth with mycelium contains 50-100% more protein than the filtered broth.3 Their stabilizing effect fundamentally originates in “steric hin- d r a n ~ e . ” ~ Other macromolecules influence the stability, somewhat increasing the viscosity of the dispersed phase, but their effect is not considerable in this medium, which has a high viscosity.

Solid Particles

In the case of the direct extraction of the fermentation broth, the biomass concentration is 20-40%. The sub- ject of our investigation is fundamentally a “suspoemul- ion."^ The phase separation is encumbered by the solid phase €or two reasons. First, the colloidal particles (pieces of the fragmented mycelium, residues from flour

Biotechnology and Bioengineering, Vol. 40, Pp. 247-251 (1992) 0 1992 John Wiley & Sons, Inc. CCC 0006-3592/92/020247-05$0400

in the culture medium, and the precipitated protein par- ticles) hardly settle themselves. Second, these solid par- ticles obstruct the contact inside the dispersed phase, hindering coalescence. In addition, the viscosity of the water phase is increased by the biomass and in this way the formation and stability of the o/w emulsion are enhanced.

Initially the cation-active demulsifiers which are suit- able for the breaking of the emulsion (e.g., alkyl amins, quaternary ammonium salts, or N-alkyl pyridinium salts) adsorb on the negatively charged surface and form com- pounds with the anion-active emulsifiers. Thus these compounds lose their surface-active property and dis- solve into one of the phases. It is advantageous if non- ionic demulsifiers are also present in the medium because they adsorb on the interface more easily than the natural emulsifiers, but they have no emulsifying and stabilizing effects in themselves.

MATERIALS AND METHODS

To investigate the effectiveness of demulsifiers, fermen- tation broths inoculated with an effectively producing strain of Penicillium chrysogenum in the fermentation plant were consistently used. i-Butyl acetate containing the ester of minimum 95% (v/v) concentration was used as a solvent in accordance with the circumstances of the extraction in the plant. For acidification 15% (m/m) technical sulfuric acid was employed. Ten percent (m/v) water or solvent stock solutions were made from the de- mulsifiers. The trade names, producers, characters and solubility of these compounds are listed in Table I. We were able to make clear water solutions from the chemi- cally homogeneous quaternary ammonium salts, and the mixed preparations of native substance were dissolved in i-butyl acetate. For laboratory experiments laboratory

Table I. Main characteristics of studied demulsifiers.

propeller mixers and a laboratory centrifuge furnished with a rotor of swinging vessels were used.

The laboratory process has been developed taking into consideration the properties of the industrial extrac- tion decanter system of Westfalia Separator CA 366-290 and the theoretical and practical instructions of two de- mulsifier producers (NADAR CHIMICA, Milano, Italy, AKZO CHEMIE, Littleborough, UK).

Fifty milliters of broth is measured in a narrow- necked plastic glass. Mixing the fermentation broth, a previously determined quantity of sulfuric acid solution is added to adjust the pH in the range 2.1-2.3. Immedi- ately after this, mixing this medium gently, the required quantity of the demulsifier is measured precisely using a micropipette. Then i-butyl acetate is added to the broth in such a quantity that the sum of the volumes of i-butyl acetate and the solution of demulsifier measured pre- viously should be 16.7 mL (the ratio of the volume of the broth measured to the latter one is 3: 1). After this the medium is emulsified with a laboratory mixer with 2000 rpm for 30 s. Immediately after the emulsification, divided centrifuge tubes of 10 mL are filled with samples and are centrifuged at 3000 rpm for 3 min. Then the volumes of the phases are recorded. Experiments were performed on each fermentation broth in four or five different concentrations (0-5 g/L) in such a way that the zero concentration was the starting point of the in- terval (blind test) and its biggest value reached or ex- ceeded the maximum or optimal concentration of the appliance of the demulsifier.

RESULTS

The results of the laboratory experiments are provided by centrifuged emulsion and its determined properties. The usual view can be seen in Fig. 1. Without dosing

Demulsifier Manufacturer Character/structure Solubility

NADAR 107 A NADAR 107 AM NADAR 108/4 NADAR C E N T 0 NADAR 98/A NADAR 98/B NADAR 98/C NADAR 98/D ARMOGARD D-5306 ARMOGARD D-5390 STEROGENOL NITROGENOL CETAZOL CI CETAZOL Br GEMEX Z9 DEMULSO DEHYQUART C ALFAQUAT ALPEG N

Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Nadar Chimica, Milan, Italy Akzo Chemie, Littleborough, UK Akzo Chemie, Littleborough, UK Egyesiilt, Budapest, Hungary CAOLA, Budapest, Hungary Organica, Novokuzneck, Russia Organica, Novokuzneck, Russia British Petrol Ltd., Wallington, UK Petrolite Corp., London, UK HENKEL GmbH, Diisseldorf, Germany CAOLA, Budapest, Hungary CAOLA, Budapest, Hungary

fatty amine polyglycolether fatty amine polyglycolether cationic cationic quaternary ammonium salt quaternary ammonium salt quaternary ammonium salt quaternary ammonium salt polyalcoxilate blend cationic, nonionic blend cetyl pyridinium bromide cetyl pyridinium bromide cetyl pyridinium chloride cetyl pyridinium bromide cationic cationic lauryl pryidinium chloride fatty amine derivate fatty acid polyglycolester

org. solvents, (water) org. solvents org. solvents org. solvents water, (alcohols) water, (alcohols) water, (alcohols) water, (alcohols) org. solvents org. solvents water, (alcohols) water, (alcohols) water, (alcohols) water, (alcohols) org. solvents org. solvents water org. solvents, (water) org. solvents

248 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 40, NO. 2, JUNE 20, 1992

Solvent phase

- - 1 Emulsion phase !!!I ------ Sohd phase ------- - Water phase /

a., without demulsifier

b., with demulsifier

Figure 1. Laboratory trial for demulsification.

demulsifiers (see Fig. la) the clear solvent phase does not appear. The upper phase is a stable emulsion. Between the upper and lower phases the water phase of the fer- mentation broth can be seen as opalescent because of the slow sedimentation of the disperse particles. Mycelium and the solid residues of the culture medium can be found in the lower phase. Dosing the smallest required quantities of demulsifiers, clear i-butyl acetate appears as a light phase (see Fig. lb). Its quantity approaches the volume of i-butyl acetate added previously to the broth if the volumes of the dosed demulsifiers are in- creased. The relation between these two parameters is not always steady, but in some cases the volume of the recovered i-butyl acetate has a local maximum consider- ing a broader interval of the demulsifier concentration. In this case reemulsification appears owing to the over- dose of these demulsifiers. Using a good demulsifier, near the interval of the appliance, this phenomenon does not appear.

An interphase nearly always exists between the sol- vent and water phases, and in fact, the same emulsion appears if demulsifiers are not applied. There is a close connection between the thickness of the interphase and the volume of the light phase; in the case of a fermenta- tion broth the smaller the volume of the interphase, the bigger the volume of the solvent.

Having determined the ratio of these two phases, the separation of the phases can be evaluated. The evalu- ation is limited by the fact that the fermentation broths can be different in quality. Thus, the figures obtained possess some deviation. For a thorough investigation a large number of trials is required. However, the absence of only one of the desired properties of the appropriate demulsifier can easily exclude the given compound from subsequent research.

In qualifying the demulsifier and the phase separa- tion, the nature of the interphase is very charcteristic. The dense, thin boundary layer and well-defined inter- face indicate a good demulsifier. It is advantageous if this is a hydrophilic boundary layer because it promotes the phase separation and the boundary layer is removed together with the water phase. This means that the best demulsifiers are multicomponent materials containing wetting agents. If the interphase is not dense and does not have a sharp interface applying large concentrations of demulsifiers, this also indicates that the fermentation broth has disadvantageous properties. These fermenta-

tion broths are generally quite lysed and the mycelium has a filose and considerably fragmented morphology.

We have applied the following evaluation of the data. Considering the fermentation broth and demulsifier, the volumes of the light phase and interphase are plotted against the demulsifier concentration and the character- istic parameters of the curves are thus determined (see Figs. 2 and 3). The figures for the different fermenta- tion broths in relation to one of the demulsifiers are averages.

The characteristic points of the curves are as follows:

1. The ratio of the maximum solvent recovery (given as percentage of the dosed solvent).

2. The demulsifier concentration (grams per liter of broth) relating to the maximum solvent recovery.

3. The ratio of the possible minimum emulsion inter- phase (given as percentage of the total volume).

4. The emulsifier concentration (grams per liter of broth) relating to the minimum emulsion bound- ary layer.

In order to give a more synoptical representation of the connected data of the above extreme values, and for more convenient evaluation, two index numbers have been introduced.

loo, I

Figure 2. Evaluation of laboratory test for demulsification: sol- vent phase.

Emulsion phase 1: 20-

1G- 4.

0 f 1 f 1 Domulaiflor oonoentration [g/l]

Figure 3. Evaluation of laboratory test for demulsification: emul- sion phase.

SZABO: DEMULSIFICATION DURING PENICILLIN BROTH EXTRACTION 249

5. Solvent index: This is the quotient of the maximum solvent recovery and connected demulsifier concen- tration. A bigger index number indicates a better de- mulsifier because a bigger yield of solvent recovery and/or a smaller minimum demulsifier concentration is equally advantageous.

6. Emulsion index: This is the product of the minimum emulsion boundary layer and the connected demulsi- fier concentration. A smaller index number indicates a better demulsifier because a reduced interphase and/or a smaller quantity of the demulsifier is simi- larly advantageous.

The parameters listed above (1-6) are given in Table 11. The relative order of each demulsifier follows from Figures 1, 3, 5, and 6. The sum of their relative orders by each characteristic gives the relative order of each demulsifier involving all the characteristics (Table 111).

Table 11. Laboratory test results for demulsification.

In addition to the above detailed primary properties, other important parameters also influence the choice of the appropriate demulsifier. However, there is a reason for restricting the discussion to only these demulsifiers which otherwise meet the above requirements. The sec- ondary standpoints of evaluation are as follows: the color of the extract as a function of the demulsifier con- centration; the characteristics of phases, of emulsion boundary layer, and interfaces; and the color, solidifica- tion point, solubility, hazardous and toxic properties of demulsifier. and costs.

DISCUSSION

From the ordered demulsifiers given in Table 111, eight compounds can be considered as appropriate for the direct extraction of penicillin fermentation broth. How- ever, NADAR 107 A and ARMOGARD D-5306 sig-

Ratio of Ratio of

recovery, concentration, interface, concentration, Solvent Emulsion % g/L % g/L index index

maximum solvent Demulsifier minimum emulsion Demulsifier

Demulsifier Solvent (1) (2) (3) (4) (5) (6 )

NADAR 107 A NADAR 107 AM NADAR 108/4 NADAR CENTO NADAR 98/A NADAR 98/B NADAR 98lC NADAR 98/D ARMOGARD D-5306 ARMOGARD D-5390 STEROGENOL NITROGENOL CETAZOL Cl CETAZOL Br GEMEX Z9 DEMULSO DEHYQUART C ALFAQUAT ALFAQUAT ALPEG N

i-BuOAc i-BuOAc

i-BuOAc water water water water i-BuOAc i-BuOAc water water water water i-BuOAc

water water i-BuOAc i-BuOAc

i-BuOAc

i-BuOAc

88.9 88.1 68.0 78.0 78.8 76.7 75.1 0.0

88.3 79.2 79.4 47.3 45.0 68.0 65.0 82.0 74.3 82.1 0.0

88.0

3.1 3.3 3.0 3.0 4.6

10.0 10.0

2.4 3.9 3.3 2.9 3.0 3.7 3.0 4.0 4.0 3.8 3.4 4.0

-

3.6 7.0

11.5 8.7 3.9 5.6 5.6

32.0 2.9 7.4

19.7 4.0

12.7 15.0 21.0 21.0 10.5 11.5 10.1 38.5

3.2 3.0 4.0 3.0

9.0 9.0

12.0 4.0 3.6 3.7 3.3 3.7 3.3 3.0 3.0 4.0 3.8 3.5 2.0

4.8

28.7 11.5 26.7 21.0 22.7 46.0 26.0 26.1

7.7 50.4 7.5 50.4

36.7 11.6 22.6 26.6 24.0 73.0 27.4 13.2 15.8 47.0 12.2 49.5 22.7 63.0 16.3 63.0 20.5 42.0 19.6 43.8 24.1 35.4 - 77.0

17.1 18.7

- 384.0

Table 111. Relative order of good demulsifiers based on laboratory tests.

Ratio of Ratio of

recovery interface index index Summarized maximum solvent minimum emulsion Solvent Emulsion

Demulsifier (1) (3) (5) (6) order

1. NADAR 107 A

3. NITROGENOL 4. NADAR 107 AM

6. NADAR 98lA 7. ALFAQUAT (in i-BuOAc) 8. NADAR CENTO

2. ARMOGARD D-5306

5. ARMOGARD D-5390

1. 4. 7. 3. 2. 9. 5.

10.

2. 1. 4. 7. 8. 3.

10. 9.

2. 1. 3. 4.

10. 13. 6. 5.

1. 2. 3. 5. 7. 4. 8. 6.

6 8

17 19 25 29 29 30

250 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 40, NO. 2, JUNE 20, 1992

nificantly excel over the others. They are excellent considering both their primary properties and most of the secondary standpoints of evaluation. These demulsi- fiers are light-colored and dissolve in i-butyl acetate well. They color the extract only slightly, less than, e.g., GEMEX Z9, DEMULSO, or NADAR CENT0 tradi- tionally employed in penicillin technologies. They can- not be determined in the finished product and have only slight toxicity. Close to the optimal concentration of appliance (1.5-2.5 g/L) relating to the first two com- pounds, the tendancy for reemulsification does not ap- pear and even averaged figures derived from trials show little deviation. Most of the trials give sharp interfaces with dense emulsion interphases. The characteristics of the experiments upon these two compounds are given in Figures 4 and 5. On the basis of laboratory experiments, four demulsifiers (NADAR 107 A, ARMOGARD D-5306, D-5390, and STEROGENOL) were tried and the results were in accord with those of the laboratory.

'3 -I 7n-l /

a "i / - Solvent I - Emulsion phase

El /

P f jmulriiier concentration [g

Figure 4. Laboratory test results of NADAR 107 A.

- Solvent phase - Emulsion phase

a

Demulsitler concentration [g/l]

Figure 5. Laboratory test results of ARMOGARD D-5306.

References

1. Brunner, K-H. 1985. Whole broth solvent extraction. pp. 182- 199 In: M. S. Verral (ed.), Discovery and isolation of microbial product. Ellis Horwood, Chichester, United Kingdom.

2. Lennie, S., Halling, P. J., Bell, G. 1990. Causes of emulsion for- mation during the solvent extraction of fermentation broths and its reduction by surfactants. Biotechnol. Bioeng. 35: 948-950.

3. Likidis, Z., Schlichting, E., Bischoff, L., Schiigerl, K. 1989. Re- active extraction of penicillin G from mycel-containing broth in a countercurrent extraction decanter. Biotechnol. Bioeng. 33: 1385-1392.

4. Szhnt6, F. 1987. A kolloidkimia alapjai. Gondolat, Budapest.

Note added in proof On the basis of laboratory experiments, four demulsifiers (NADAR 107A, ARMOGARD D-5306, D-5390, and STEROGENOL) were also tried out in plant scale and the results were in accord with those of the laboratory.

SZAB6: DEMULSlFlCATlON DURING PENICILLIN BROTH EXTRACTION 251