Downstream processing of lactic acid-whey permeate fermentation broths by hollow fiber...

�9 1988 by The Humana Press Inc. All rights of any nature whatsoever reserved. 0273-2289/88/1901--0061502.00

Downstream Processing of Lactic Acid-Whey Permeate Fermentation Broths

by Hollow Fiber Ultrafiltration

SUSY TEJAYADI AND MUNIR CHERYAN*

Biochemical Engineering Laboratory, 382 AES Building, University of lllinois, 1304 W. Pennsylvania Avenue, Urbana, IL 61801

Index Entdes: Cell harvesting; ultrafiltration; Lactobacillus bulgaricus.

I N T R O D U C T I O N

Fermentation processes are able to produce a broad range of chemical and biological products. One major limitation to the commercial ap- plication of biotechnology and fermentation is downstream processing. The primary downstream step is the separation of the microorganisms from the rest of the bioreactor constituents. The complexity and unique physicochemical properties of fermentation broths make separation of their components sometimes difficult. Not only do they contain a mix- ture of soluble and insoluble materials, but they also have a high water content and low specific gravity (1). The difficulty is enhanced when the desired product is present at low concentration and/or degrades easily at extreme temperature and pH (2).

The most common separation techniques are centrifugation and dead-end filtration, which have some limitations (3,4). Crossflow mem-

*Author to whom all correspondence and reprint requests should be addressed.

Applied Biochemistry and Biotechnology 61 Vol. 19, 1988

62 Tejayadi and Cheryan

brane filtration offers a n u m b e r of advantages over convent ional techniques and has been used to harvest microbial cells (3-9) and to increase product ivi ty of fermenta t ion processes (10-12).

The objective of this research was to evaluate the applicability of crossflow filtration for the cont inuous harvest ing or separat ion of bacte- rial suspens ions . The mode l system was a lactic acid fermenta t ion broth with Lactobacillus bulgaricus, a homofermenta t ive lactic acid bacteria, sus- p e n d e d in w h e y permeate , which is a byproduct of the ultrafiltration of cheese whey . Factors affecting the performance of the system, such as t r an smembrane pressure , velocity, and cell concentrat ion, were investigated with a v i ew toward minimizing fouling and concentrat ion polarization effects and de te rmin ing op t imum operat ing conditions.

MATERIALS AND METHODS

Membrane Module

The m e m b r a n e sys tem used for these exper iments was a pilot size hol low fiber ultrafiltration uni t (Romicon Inc., Woburn , MA). The m e m - brane cartr idge 's specifications are given in Table 1. The m e m b r a n e was placed in a test loop consist ing of a 50 gal-steam jacketed hold ing vat (the feed tank), a centrifugal pump , pressure gages, valves at the inlet a n d outlet of the modu le , f lowmeters to measure flow rate (recycling rate) and pe rmea te flux, and a thermocouple to measure tempera ture .

Experimental Design

The pe r fo rmance and fouling characteristics of the hollow-fiber m e m b r a n e ultrafilter were investigated at several levels of cell concentration, velocity, and t r ansmembrane pressure. Fouling and per fo rmance

Table 1 Hollow Fiber Module Specifications

Module designation Membrane material Number of fibers Length of cartridge Outside diameter of cartridge Potting material Effective area Fiber inside diameter Maximum operating pressure Maximum operating temperature Nominal molecular weight cut off pH range

HF06-49-PM50 Polysulfone 600-660 27 cm 7.6 cm Epoxy_ .56 m 2 1.2 mm 172.4 kPa 75~ 50000 1-13

Applied Biochemistry and Biotechnology Vol. 19, 1988

Processing of Lactic Acid-Whey Fermentation Broths 63

studies were conducted at 30~ and at constant feed concentration with both the permeate and retentate recycled back to the feed tank. Under these conditions, flux decline would be a result of membrane-solute interactions. For the concentration experiments, only the retentate was recycled back to the feed tank.

Flux (J), a measure of the filtration capacity of a membrane module, is expressed as volume of permeate per unit membrane surface area per unit time or liters per square meter per hour (LMH).

Membrane Cleaning

After each experiment, the membrane was flushed and backflushed with water at room temperature for 30-45 and 15 min, respectively. Fol- lowing the water flush, an enzyme detergent solution (Tergazyme, Alco- nox Inc., NY) was circulated at 50~ for 5-10 min. The membrane was soaked in this solution overnight. After rinsing out the enzyme solution with warm water, the cleaning efficiency was checked by measuring the water flux under standard conditions. The cleaning procedure was repeated if necessary.

Fermentation Media

The media used in these experiments was reconstituted whey permeate powder (WPP, modified whey solids, Express Foods, Inc., Louis- ville, KY). It is the dried mineral/lactose fraction from the ultrafiltration of liquid whey. Its composition is (per 100 g): 79-81 g lactose, 8-10 g ash, 1-2 g moisture, and 2.5-4.5 g total nitrogen.

Organism

A selected strain of Lactobacillus bulgaricus was obtained from Chr. Hansen's Laboratory, Milwaukee, WI. The culture was first grown in 1 L of presterilized media containing 150 g/L WPP and 10 g/L yeast extract (Difco Lab, Detroit, MI). This seed culture was used to inoculate a 4-L batch, which was then used as seed for a 200-L batch. The 200-L batch media, containing 150 g/L WPP and 2-5 g/L yeast extract, was pasteur- ized at 63~ for 30 min, cooled to 45~ and inoculated with the 4-L batch culture. The fermentations were done at 45~ the optimum temperature for the bacteria, and the pH was controlled at pH 5.6 (+ 0.2) using a pH controller (Model 5997-20, Horizon Ecology Co., Chicago, IL).

To obtain concentrated cell cultures, several cycles of fermentation- ultrafiltration-refermentation were conducted until the required cell density was obtained. Other cell concentrations were obtained by diluting this concentrated cell culture with prepasteurized media containing 150 g/L WPP.



RESULTS AND DISCUSSION

Water Flux

The water flux of the hollow fibers increased linearly with transmembrane pressure and was unaffected by feed flow rate. This relationship can be described by the Hagen-Poiseuille model for flow through channels (3). Water flux also increased with temperature. At 30~ and 1 atm pressure, the water flux was 244 LMH.

Fouling Studies

Flux of a solution or suspension is usually much lower than the pure water flux for several reasons, among them a change in feed solution properties, concentration polarization, and membrane fouling (3). Mem- brane fouling is manifested by a decline in flux with time. It differs from concentration polarization in the sense that fouling is an irreversible and time dependen t process. It is probably a result of specific interactions be- tween the membrane and fermentation broth components.

Our fouling studies were conducted under steady operating conditions, i.e., both the retentate and the permeate were recycled back to the feed tank in order to keep the bulk concentration constant. Therefore, any flux decay will be caused by membrane fouling. The fouling process is often assumed to follow a first-order reaction and can be expressed using power-type function, such as (13)

Jt = lit -b (1)

where Jt is the flux at time t in LMH, J1 is the flux at time t = I h, and b is the fouling index. The larger the b value, the greater the fouling rate.

Figure 1 shows typical fouling of the hollow fiber membrane by the lactic acid fermentation broth that was made up by using unclarified w h e y permeate as the media. The flux dropped quickly initially and then approached a quasisteady state over the next 20-24 h. In general, the higher the cell concentration, the lower the flux. Similar results were obtained with other combinations of velocity and concentration and differ- ent lots of WPP (14). The fouling index (b values) for this set of runs at a velocity of 2.1 m/s was 0.02-0.03, and ]1 was 51 LMH at 10 g/L and 43 LMH at 40 g/L cell concentration (14).

In order to clarify the contribution of the cells per se versus the rest of the broth components to the fouling phenomenon, similar fouling studies were conducted with the media itself with no cells. As seen in Fig. 2, the whey permeate media did show some fouling tendencies. This was surprising since whey permeate should contain only lactose, salts, and nonprotein nitrogen since it is the "permeate" from the ultrafiltration of sweet whey. Thus, little or no fouling was expected by whey permeate since all of its components should have been freely permeable. The effect of velocity on fouling by whey permeate is especially dramatic.


Processing of Lactic Acid. Whey Fermentation Broths

6 0 ,

65

L . b u l g o r i c u s in w h e y p e r m e a t e

50 A

4 0 1 .

30 0

Fig. 1.

~ � 9 - �9 A A

A P T �9 52 k Pa

V �9 2.1 m / s I I !

4

10 o I L " ' " - - - - - - - ~k

2 0 , 30 g / L

40 ,/L ~ " ~ i

I I I 1 I ! ! I

8 12 16 20 24 T I M E (hours)

Fouling of hollow fibers by L. bulgaricus fermentation broth: effect of cell concentration. The transmembrane pressure (&PT) was 52 kPa and the velocity (V) was 2.1 m/s.

150

120

-1-

9 0 .../

X

..~ 60 Ls_

A P T = 121 kPa m

3<X)..CO_ 0~0_0_ 0 0 ~ 0 ~ 0 ~ 0 0 0 q C l a r i f i e d Whey Permeate V = 21 m/s

)

L Whey Permeate V = 2.1 m / s

�9 �9 Q

0 60 120 180 240 T I M E (minutes)

3 0 0

Fig. 2. Effect of velocity on fouling by whey permeate and clarified w h e y permeate. Cell concentration = 0.



The fouling index (b) is typically 0.44-0.48 at a velocity of I m/s, whereas it is 0.02-0.09 at 2.1 m/s. Higher velocities reduce the rate of fouling be- cause fouling layers are removed by higher shear stress.

It appears that commercial WPP contains some suspended matter that is not permeable and fouls the membrane, or at least causes severe polarization effects. To confirm this, the whey permeate feedstream was "clarified" of the suspended matter by ultrafiltration and the fouling study repeated. As shown in Fig. 2, the remarkably stable and high flux obtained was a clear indication that the whey permeate powder obtained from commercial sources contains some fouling components of a macro- molecular or colloidal nature. Comparing the fouling parameters (b and J1 values) with and without cells, it appears that the media components have almost as much of an effect on fouling rates as the cells.

Performance Study Figure 3 shows the flux-transmembrane pressure relationship for

our system. The data were taken when the system had reached an appar- ent steady state (after 24 h of operation). Concentration polarization phenomenon is exhibited by this system, in that the permeate flux initially increases with pressure but becomes independent at higher pressures. Similar curves were obtained at all cell concentrations studied (14).

Figure 4 shows the effect of velocity (expressed as Reynolds number)

"75

6O

x 4 5

3 0 o=

z5

0

L. bul~loricus in w h e y p e r m e o t e _ ~ /

X �9 I0 g /L ~ - ~ j ~ j e ~ - - - �9 ., ~ n ~ . _ ~_.. ,-ira

" # A / '~u

Velocity (m/s)

; 1.0

2.(3 I I ! I I I I

0 4 0 8 0 120 16(

TRANSMEMBRANE PRESSURE (kPa) Fig. 3. Effect of t ransmembrane pressure on flux at a cell concentration of

10 g/L. Variable is linear velocity in m/s.


Processing o f Lactic Acid- Whey Fermentation Broths

I00

67

A

-1"

..J v

x

/ i,+

r,-

0 - r -

oJ

70 m

m

m

5 0 -

3 0 _

20_

I 0 , 3O

Fig. 4.

~ / _ . I e d I J " / i / / /+/ /ram _ / e ,r

~i/+ j e / " /

x (g/L)

L. b u l g a r i c u s in

whey l)errneate

�9 I 0

�9 2 0

�9 3 0

4 0

t m I I ~ l i t i I ! i i , I I

I 0 0 I 0 0 0 3 0 0 0

R E Y N O L D S N U M B E R

Effect of turbulence (expressed as Reynolds number) on flux.

on p e r m e a t e flux. The ho l low fiber sys t em was o p e r a t e d u n d e r laminar f low condi t ions , since the Reynolds n u m b e r s were b e t w e e n 50 and 2500. The da ta can be r ep re sen ted by an equa t ion of the form

l = A(Re)"(Sc) ~ In (Cb/Cg) (2)

w h e r e Re is the Reynolds N u m b e r , Sc is the Schmid t N u m b e r , a n d A, a, are cons tan t s d e p e n d i n g on the state of t u rbu lence a n d d e v e l o p m e n t of the veloci ty profile. CB is the bulk concen t ra t ion of re ta ined solids in the feed s t r eam and Cg is the theoretical m a x i m u m concen t ra t ion of the re- t a ined solids. For a part icular cell concent ra t ion at a cons tan t t empera - ture , we can simplify the m ode l to

] = A'(Re)" (3)

The value of a var ied f rom 0.25 to 0.34 (Table 2). In theory , for a lami-

Table 2 Values of Constants (a and A') in the Flux

Model (Eq. [3])

Cell concentration, g/L a A'

10 .34 .72 20 .25 1.05 30 .31 .85 40 .26 1.16 Average .29 .95



nar flow system such as the hollow fibers, the value of a should be 0.33--0.50, depending on the state of development of the velocity profile (3).

Concentration of L. bulgaricus Fermentation Broth

The effect of cell concentration on permeate flux is shown in Fig. 5 for one transmembrane pressure and four velocities. As cell concentration increased, the permeate flux declined rapidly in a semilogarithmic manner. Extrapolating these lines to zero flux gives an average Cg value of 800 g/L. However, in practice, such high cell concentrations result in sharp increases in viscosity (Fig. 6) that would make pumping difficult and would require pressures greater than the burst pressure of the hollow fibers. Our studies (14) indicate that cell concentrations in this hollow fiber system would be limited to about 100-150 g/L before the flux became too low to be economical.

Membrane Deterioration

In our studies, the original water flux was never fully regained even with rigorous cleaning procedures using 0.1N NaOH or 25% ethanol. Figure 7 shows that after 300 h of use, the water flux was only about 80% of its "new" value. Sims and Cheryan (5) also observed a gradual decay in membrane water flux with the Membrana microporous tubular membrane that had been used for harvesting Aspergillus niger, perhaps a re-

80

A

"1-

._l v

X

d b_

OC

0 I

Od

70

60

50

4 0 _

3O 5

Fig. 5.

L. bulgoricus in whey permeote

A P T -- 7 7 kPa

i Oii! ~ 0.7

I ! I !

I0 20 3 0 4 0 I00

CELL CONCENTRATION ( g / L ) Effect of cell concentration and velocity on flux (APT = 77 kPa).


Processing of Lactic Acid- Whey Fermentation Broths

30

U

0

tO >

2 5

20

15

I 0 _

5

0 0

Run I Run 2

D e n s i t y �9 0

A I I I

69

I . I I

Fig. 6. cosity was measured by the Hoeppler falling ball viscometer.

1.09

1.07

1 .05 >- I--

tO

1.03 z Lg a

I.OI

IO 20 30 40 50

CELL CONCENTRATION ( g / L )

Effect of cell concentration on viscosity and density at 30~ Vis-

sult of long- term exposure to acidic conditions. This apparen t change in the intrinsic permeabil i ty of the m e m b r a n e affected the per fo rmance with our fe rmenta t ion broth. For example, after 300 h of use, the fe rmen- tation broth flux had declined by almost 50% u n d e r o therwise identical operat ing condi t ions (14). This deter iorat ion mus t be cons idered w h e n des igning a crossflow membrane sys tem for harvest ing of L. bulgaricus.

500

"1"

"~ 300 m

X

-J 200 I.L

r ~ b J 1--

I00 I0

Fig. 7.

A P T �9 58 kPa

~ e ~ O _ ~ . O . .

�9 "-"i-,-o o ~

t i x u x I 1 1 I v J 30 100 200 500

TIME (hrs )

Decline in hollow fiber membrane performance with use. Each point represents the water flux of the cleaned membrane after the indicated hours of processing the L. bulgaricus fermentation broth. Water flux was measured at 30~ and 58 kPa.


70 Tejayadi a n d C h e r y a n

CONCLUSIONS

Membrane filtration is a suitable method for cell harvesting and clarification of fermentation broths. Hollow fiber ultrafilters gave essen- tially 100% rejection of L. bulgaricus cells from a whey permeate fermentation broth. A combination of low pressures and high velocity generally gave the best permeate flux. Fermentation media components ' (in this case, from the whey permeate) contributed significantly to fouling. Con- sidering the pressure limitations of the current generation of asymmetric hollow fiber modules and the changes in physical properties of the fermentation broths, a cell concentration of 100-150 gfL could be obtained with the flux still relatively high (above 20 LMH), although the chemical compatibility of the membrane module itself under long-term exposure to high acid conditions should be considered.

ACKNOWLEDGMENTS

This research project was supported by the Illinois Agricultural Ex- periment Station, the Ferguson award from the University of Illinois Re- search Board, and by a contribution from Romicon, Inc. ST also acknowl- edges the Ralph O. and Mabel F. Hunter fellowship and financial support from Snow Brand Milk Products Co., Ltd., Tokyo, Japan.

REFERENCES

1. Michaels, A. S. (1980), Desalination 35, 329. 2. Atkinson, B., and Sainter, P. (1982), J. Chem. Tech. Biotechnol. 32, 100. 3. Cheryan, M. (1986), Ultrafiltration Handbook, Technomic, Lancaster, PA. 4. Tutunjian, R. S. (1984), Dev. Ind. Microbiol. 25, 415. 5. Sims, K. A., and Cheryan, M. (1986), Biotechnol. Bioeng. Symp. 17, 495. 6. Patel, P. N., Mehaia, M. A., and Cheryan, M. (1987), J. Biotechnol. 5, 1. 7. Hanisch, W. (1986), in Membrane Separations in Biotechnology, McGregor W.

C., ed., Marcell-Decker, New York, pp. 61-88. 8. Lasky, M., and Grant, D. (1985), Am. Biotechnol. Lab. 3(6), 16. 9. Gabler, F. R. (1984), Dev. Ind. Microbiol. 25, 381.

10. Cheryan, M., and Mehaia, M. A. (1985), in Membrane Separations in Biotech~ nology, McGregor W. C., ed., Marcell-Decker, New York, pp. 255-310.

11. Khorakiwala, K. H., Cheryan, M., and Mehaia, M. A. (1985), Biotechnol. Bioeng. Syrup. 15, 249.

12. Cheryan, M., and Mehaia, M. A. (1986), CHEMTECH, 16, 676. 13. Kuo, K. P., and Cheryan, M. (1983), ]. Food Sci. 48, 1113. 14. Tejayadi, S. (1987), M. S. thesis, University of Illinois, Urbana, IL.


Downstream processing of lactic acid-whey permeate fermentation broths by hollow fiber...

Documents

Transcript of Downstream processing of lactic acid-whey permeate fermentation broths by hollow fiber...