WO2004070026A1

WO2004070026A1 - Continuous process for immobilization of yeast in k-carrageenan gel beads

Info

Publication number: WO2004070026A1
Application number: PCT/CA2004/000158
Authority: WO
Inventors: Christophe Decamps; Sylvain D. J. M. Norton; Denis J. C. M. Poncelet; Ronald J. Neufeld
Original assignee: Labatt Brewing Company Limited
Priority date: 2003-02-07
Filing date: 2004-02-09
Publication date: 2004-08-19

Abstract

A novel continuous two-phase dispersion process was developed to produce kappa-carrageenan gel microspheres, using static mixers. The yeast-loaded microspheres are particularly suited for use in continuous brewing operations. Bead production processes according to the present invention include control by way of selection of static mixer diameter, the number of mixing elements, the fluid linear velocity and the volumetric fraction of κ-carrageenan, on the mean diameter and size distribution of the resulting gel microspheres. Bead diameter was strongly influenced by the average linear fluid velocity through the mixer, and by the mixer diameter. The number of mixer elements and the mixer diameter governed bead size dispersion. Productivity was proportional to the mixer diameter squared, however, as the coefficient of variability increased with mixer diameter, a plurality of small (i.e. less than 1.27 cm) diameter mixers are preferably operated in parallel, to reduce the size dispersion.

Description

CONTINUOUS PROCESS FOR IMMOBILIZATION OF YEAST IN K-CARRAGEENAN GEL BEADS

Field of the Invention:

The present invention relates to the statically controlled formation of carrageenan- bead immobilized cells, especially yeast cells, and in particular to a process for the use of it in the production of alcohols through fermentation. The process is particularly advantageous for producing immobilized yeast cell beads for use in the production of ethanol, and especially in relation to the production of fermented beverages, such as wine or beer.

Background

As pointed out in US 5,869,117, it has been proposed that commercial fermentation practices might advantageously evolve through the use of immobilized cells. A variety of techniques are available for this purpose, include, by way of example, a simple gelation immobilization technique that was developed for and is used primarily with alginate. Generally, this technique involves the drop-wise addition of the ionic polysaccharide/immobilizant solution, through a syringe needle, into a solution of a divalent cation. The divalent ions cross-link various charged species on the polysaccharide molecule, with the result that insoluble gel beads are formed. Where alginate is selected as the ionic polysaccharide, as is typically the case, divalent calcium ions are employed as cross-linkers. This "syringe extrusion" process has the advantage of producing beads having a narrow, unimodal size distribution. Bead size, however, is limited by the syringe needle bore size and viscosity of the solution. As a result, beads of less than 3 mm, and especially beads of less than 1mm can be difficult to produce. The size of these large beads imposes diffusion limits on the transfer of substrate and product to and from the entrapped yeast cells. For example, in some cases the diffusion problems allow anaerobic fermentations to take place internally of the bead, notwithstanding the fact that to all outward appearances, aerobic fermentation is proceeding normally at or near the bead surface.

Smaller diameter beads are needed to better facilitate both internal and external mass transfer, enhancing fermentation performance and minimizing bead rupture due to gas formation and accumulation. Production of small bead sizes has been attempted previously by modifying the syringe extrusion process, through the use of air jets impinging on the needle, electrostatic pulses, or vibrating needles. On the other hand, while the variations on the syringe extrusion technique mentioned above can be employed for facilitating smaller alginate bead production, they too are-fraught with economic penalties. These include needle like extruders of one of two typical designs.

The first such involves producing small drops of sodium alginate/yeast slurry, by passing the material through the needle, and with a vibratory action, shaking off a smaller drop than would otherwise form in the absence of the vibration.

The second approach uses a coaxial flow needle, in which the solution of sodium alginate/yeast is passed through the centre of the needle, and as droplets form at the end of the needle, a coaxial flow of air pulls a small droplet away from the tip. These approaches can be used to keep alginate bead sizes as low as possible (with standard deviations of about 20%). However, the number of needles needed to maintain the flow rate is inversely proportional to the bead volume. Reducing the bead size to 500 micrometers or 100 micrometers requires the use of several hundreds or even hundreds of thousands of needles operating concurrently: a complex, expensive, and generally awkward solution. On the other hand, however, even these methodologies provide only very limited rates of bead production throughput, and are accordingly very difficult and expensive to scale up sufficiently to supply commercial fermentation processes.

Various other approaches have been taken to increase the productivity of beads suitable for use in fermentation applications. These include the in accordance with a broad aspect of the present invention, there is provided a process for immobilizing viable cells in polymer beads, for use in fermentation processes. The immobilization process described in US 5,869,117, for example, comprises the steps of: preparing an aqueous phase that is a mixture of either a pre-polymeric molecular species or an un-gelled polymeric molecular species; in an aqueous suspension of viable cells; preparing a mixture of said aqueous phase and a food grade oil phase (i.e., one that is not substantially reactive with the other components in the bead forming milieu); subjecting said mixture to shear by passing the same through a static mixer under flow-rate conditions selected to disperse the aqueous phase in the oil phase, such that aqueous phase droplets in a resulting emulsion have a desired droplet size distribution; and, subjecting the selected pre-polymeric or un-gelled polymeric molecular species to polymerizing or gelling conditions (as the case may be), to thereby form polymer beads from the droplets, which have immobilized viable cells entrapped therein. In the preferred form, these beads are small, i.e., less than 3 mm; preferably less than 1.5 mm; and in particular, less than about 0.5 mm. In general, immobilized-yeast beads in the range of about 1.5 to about 0.2 mm, are preferred.

There remains a need in the art, however, to refine the control that can be exercised over the course of bead production, of the mean particle size, variance and productivity of the bead population - to more controllably deliver the desired mass-transfer characteristics that are suited to a given fermentation.

Various prior art references relating to bead production and utilization are included in the following listing, (and to which references are variously made elsewhere herein):

AI Taweel, A, M., Walker L, D. Liquid dispersion in static in-line-mixers. Canadian J. Chem. Eng., 61, 527 (1983).

Audet, P., and Lacroix, C. Two phase dispersion process for the production of biopolymer gel beads: Effects of various parameters on bead size and their distribution. Process Biochem. 24, 217 (1989).

Begin, A., Castaigne, F., Goulet, J. Production of alginate beads by a rotating atomizer. Biotechnol. Techniques. 5, 459 (1991).

Berkman, P, D., Calabrese, R, N. Dispersion of viscous liquids by turbulent flow in a static mixer. AIChE Journal.34, 602 (1998).

Castillo, E., Rodriguez, M., Casas, L., Quintero, R., and Lopez-Munguia, A. Design of two immobilized cell catalysts by entrapment on gelatin: Internal diffusion aspects enzyme. Microbiol. Biotechnol.13, 127 (1991).

Galasso, J, L., Bailey, J, E. In vivo nuclear magnetic resonance analysis of immobilization effects on glucose metabolism of yeast Saccharomyces cerevisiae. Biotechnol. Bioeng.26, 217 (1983).

Linko M, and Kronlof J. Main fermentation with immobilized yeasts. Proceedings of the European Brewery Convention congress, 1991, 353-360

Luong, J, H, T. Cell immobilization in K-carrageenan for ethanol production. Biotechnol. Bioeng.25, 2910 (1983). Masschelein, C, A., Ryder, D, S., and Simon, J, P. Immobilization cell technology in Beer Production. Critical Reviews in Biotechnology, 14 (2): 155-177 (1994).

Matulovic, U, Rasch, D., and Wagner, F. New equipment for the scaled up production of small spherical biocatalysts. Biotechnol. Lett. 8, 485 (1986).

Mensour, N.A., Margaritis, A., Briens, C.L., Pilkington, H., and Russell, I. Application of immobilized yeast cells in the brewing industry. In Immobilized Cells: Basics and Applications, Wijffels, R.H., Buitelaar, R.M., Bucke, D. and Tramper, J., (Eds.), Elsevier Science B.V., 1996, 661-671.

Middelman, S. Drop size distributions produced by turbulent pipe flow of immiscible fluids through a static mixer. Ind. Eng. Chem. Process Des. Develop.13, 78 (1974).

Mutsakis, M., and Robert, R. Advances in static mixing technology. Chem. Eng. Progress, 42-48, June 1986.

Myers, K, J., Bakker, A., and Ryan, D. Avoid Agitation By Selecting Static Mixers. Chemical Engineering Progress, June, 28-38, June 1997.

Nava Saucedo JE, Audras B, Jan S, Bazinet CE and Barbotin J-N. Factors affecting densities, distribution and growth patterns of cells inside immobilization supports. FEMS Microbiology Rev. 1994, 14, 93-98

Ogbonna, J.C., Mastumura, M., Yamagoata, T., Sakuma, H., Kataoka, H. Production of microgel beads by a rotating disc atomizer. J. Ferment. Bioeng. 68, 40, 1989.

Poncelet, D., Lencki, R., Beaulieu, C, Halle, J, P., Neufeld, R.J. and Fournier, A. Production of alginate beads by emulsification/internal gelation. Appl. Microbiol. Biotechnol. 38, (1992).

Poncelet, D., Poncelet de Smet, B., Beaulieu, C, and Neufeld, R. J. Scale up of gel bead and microcapsule production in cell immobilization. Fundamentals of Animal Cell Encapsulation and Immobilization, Goosen, M.F.A., CRC Press Inc., Boca Raton, FL (1993).

Poncelet, D., Desobry, S., Jahnz Ulrich and K. Vorlop. Immobilization at large scale by dispersion. In: Wijffels, R.H. (Ed.) Immobilized cells, Springer Lab Manual, Springer- Verlag, Berlin, pp. 139-149, 2001.

Rehg, T., Dorger, C, and Chau, P,C. Application of an atomizer in producing small alginate beads for cell immobilization. Biotechnol. Lett. 8, 111 (1986).

Scheaffer, R, L., and Chau, P,C. Probability and statistics for engineers, PWS-Kent Publishing Co., Boston, 8, 348 (1990). Scott, C,D. Techniques for producing monodispersed biocatalyst beads for use in columnar bioreactors. Ann. N. Y. Acad. Sci. 501, 487 (1987).

Summary of the Invention:

Immobilized cells that are variously carried on and in beads produced in accordance with the present invention are particularly useful in the conduct of carbohydrate fermentation processes for the production of ethanolic products, (although the present invention can also find application in pharmaceutical, bioremediation, water treatment and other applications where active immobilized agents can be used in processing of a substrate).

With regard to fermented beverages, the present invention has application in particular to the production of malt beverages. Malt beverages herein, includes brewery beverages, and fermented malt brewery beverages in particular. The general process of preparing fermented malt beverages, such as beer, ale, porter, malt liquor, low and non-alcoholic derivatives thereof, and other similar fermented alcoholic brewery beverages, hereinafter referred to simply as "beer" for convenience, is in general well known. As practiced in modern breweries, the process comprises, briefly, preparing a "mash" of malt, usually with cereal adjuncts, and heating the mash to solubilize the proteins and convert the starch into sugar and dextrins. The insoluble grains are filtered off and washed with hot water which is combined with the soluble material and the resulting wort boiled in a brew kettle to inactivate enzymes, sterilize the wort, extract desired hop components from added hops, and coagulate certain protein-like substances. The wort is then strained to remove spent hops and coagulate, cooled and pitched with yeast, and then fermented. The fermented, immature brew known as "green" or "rah" beer is then "finished", aged— which is sometimes referred to as "lagering" and clarified, filtered, and then carbonated to produce the desired beer.

Introduction to the Drawings:

Various notations are employed in relation to the drawings and elsewhere herein is as follows:

D static mixer diameter (mm).

N_e number of elements

V superficial fluid velocity (cm s^"1). ε volumetric fraction of K-carrageenan (%) d_m bead mean diameter (μ ). σ standard deviation (μm).

CV coefficient of variability (%).

Figure 1 : Illustration of static mixer similar in design to the Kenics mixer used in study. Mixer 1, includes tube 2 and a static mixer element 3 with flow of fluid within the tube and over the static mixer surfaces being represented by arrows 4. Individual mixer elements are shown mounted in tube. Diameter of mixer elements equals that of the inner diameter of tube.

Figure 2 : Schematic of process developed to formulate yeast-loaded carrageenan microspheres using continuous flow, static mixer technology.

Figure 3 : Typical bead size distributions determined under the following process conditions: ( Δ ) V = 13.2 cm s^"1, D = 9.5 mm, N_e = 24, ε = 25%; ( 0 ) V = 3.5 cm s^"1, D = 9.5 mm, N_e = 24, ε = 25%.

Figure 4 : Impact of the linear fluid velocity (V) and the number of mixer elements (N_e) on the mean diameter, d_m (a) of the resulting microspheres and on the coefficient of variability, CV (b). D = 12.7 mm and ε = 12.5%. N_e = 12 ( • ); 24 ( ■ ); 48 ( A ); 60 ( ♦ ); 72 ( D ); and 120 ( o ).

Figure 5 : Effect of the linear fluid velocity (V) and the volume fraction of carrageenan, ε, on the resulting microsphere mean diameter, d_m (a), and on the coefficient of variability, CV (b). D = 9.5 mm, N_e = 24 and ε = 8.3% ( • ); 12.5% ( ■ ); 25% ( A ); and 50% ( ♦ )

Figure 6 : Impact of the linear fluid velocity, V, and static mixer diameter, D, on the microbead mean diameter, d_m. N_e = 24. Results are shown for carrageenan volumetric fraction, ε, values of 8.3 (a); 12.5 (b); 25 (c) and 50% (d). Static mixer diameters used were D = 6.4 ( • ); 9.5 ( ■ ); and 12.7 mm ( A ).

Figure 7 : Effect of linear fluid velocity, V, with varying static mixer diameter, D, on the microbead coefficient of variability, CV. N_e = 24 and ε = 8.3% and D = 6.4 ( • ); 9.5 ( ■ ); and 12.7 mm ( A ).

Figure 8 : (Prior art) is a schematic representation of a process according to US 5,869,117 for producing carrageenan encapsulated yeast beads; and,

Figure 9 : (Prior art) is a schematic representation of a gas-lift fluidized bed bioreactor according to US 5,869,117, employing beads produced in accordance with the process as illustrated in Figure 2 hereof.

Detailed Description:

In accordance with the present invention there is provided a process for adding control parameters to the design, construct and operation of a process (and most desirably a continuous one) for the production of yeast inoculated gel beads and to produce beer using immobilized cells in a continuous airlift bioreactor.

Lager beer requires a batch fermentation time of 6-7 days, thus there is international interest in developing a more economical, smaller scale, continuous fermentation process using immobilized cells for the primary and secondary fermentation stages. The advantage of immobilizing yeast is to retain highly concentrated yeast during the continuous brew, resulting in faster process times, and potentially operating the fermentations at throughputs higher than the nominal washout rate (Masschelin, C. A, et al., 1994). The goal then is to reduce processing time without sacrificing product quality.

Common immobilization methods include physical entrapment in a gel matrix or within a membrane bound microcapsule, adsorption or covalent attachment to preformed carriers, and self-aggregation or cross-linking of cells. A variety of gel matrices have been used for the physical entrapment of whole cells including alginate, agarose, and K- carrageenan.

Various methods have been reported for the production of cell-loaded gel beads. For industrial large scale production, the method with the highest capacity is probably the rotating atomizer because of its high throughput. This method was initially designed for alginate gel beads where gel-inducing calcium would diffusion into the polymer droplets, once caught in a gelation bath. Alginate beads may be unstable in some fermentation media, due to the presence of calcium chelators. For brewing purposes, it is then interesting to replace alginate by K-carrageenan, as potassium replaces calcium as the gelling cation. Preliminary trials at the pilot scale (200 L of polymer) indicated that the high viscosity of this polymer was not easily compatible with atomizing nozzles.

K-carrageenan like alginate, is a food grade material, and has been favored for whole cell immobilization because of its superior mechanical strength over other gels. Yeast entrapped in K-carrageenan gel has been used successfully in the brewing industry for continuous primary fermentation (Linko and Kronlof; 1991; Mensour et al., 1996). To be a realistic alternative to traditional free cell fermentation and maturation systems, immobilized cells must be stable for relatively long operational times, characteristically measured in weeks or months. Mass transfer limitations of substrate into, and products out of the immobilized cells and associated matrix is also of critical interest Nava Saucedo (1994).

The process described in this study, involves the formation of an emulsion between a non-aqueous continuous phase (vegetable oil) and an aqueous dispersed phase (κ-carrageenan previously inoculated with yeast), with the use of static mixers. The dispersed emulsified carrageenan-cell droplets undergo rapid gelation by chilling the emulsion. Gel beads are then recovered from the oil by filtration, screening, decanting or phase partitioning, and the oil may then be recycled back into the process. For brewing operations, resultant beads must be fully clean of the oil, following the phase partitioning step.

Static mixers consist of a series of stationary elements placed transversely in a tube, as illustrated in Figure 1. The elements form crossed channels that cause the repeated division, rotation and the longitudinal recombination of the liquid flowing through the static mixer. When two immiscible fluids are pumped through the mixer, the transverse rupture of fluid streamlines into an increasingly homogenously dispersed emulsion is promoted (Mutsakis, M., and Robert, R., 1986). Static mixers enable the formation of a fine emulsion with controllable properties in flows similar to those encountered at an industrial scale (several to hundreds Lh^"1). Furthermore, as it operates in continuous mode, online control of bead properties through adjustment of the operational parameters is possible.

A variety of static mixers are commercially available with varying geometry and special applications. Increasingly, static mixers are replacing dynamic systems because rotating equipment consumes power, requires routine maintenance, and can be a significant investment. The static mixer on the other hand, fits in-line, has no moving parts and no electrical requirements, other than the pumping power required to move the fluids (Myers, K. J et al, 1997).

Four operating parameters were selected for controlling gel bead size distribution: the diameter (D) and the number of elements (N_e) of the static mixer, the linear velocity of the mixture (V) and the volumetric fraction of K-carrageenan (ε). The present invention relates to the use of static mixers operating (esp. in a continuous mode), for generating emulsions yielding yeast-loaded beads. Secondly, the operating conditions yielding an appropriate bead mean diameter and size distribution were evaluated. Smaller bead diameters are preferred, reducing intra-capsular mass transfer limitations. The potential of this immobilization technology in terms of productivity is also assessed to establish that it meets the demands of industrial scale operations.

Referring now to Referring now to Figure 8 of the appended drawings, there is shown a schematic representation of the process useful in the present invention.

The sterilized commercial grade canola oil was transferred from a holding tank 1, by way of peristaltic pump 2, through to heat exchanger 3. On exiting heat exchanger 3, the temperature of the oil was about 40.degree. C.

The yeast inoculum contained in tank 4 at 40.degree. C., and the ungelled carrageenan polymer solution in tank 5 (also at 40.degree. C, were pumped together through a junction manifold 6 (a simple "T" junction in the respective lines leading from each of the above mentioned tanks), by peristaltic pumps 7 and 8. Manifold 6 is connected at the input side of a first static mixer 9, in which the inoculum and the un-gelled carrageenan solution were mixed to provide an aqueous phase.

This aqueous phase and the oil phase are then introduced together, through manifold 10, into a second static mixer, 11.

The emulsion that is produced in the static mixer 11 is then passed through a heat exchanger 12, in which the temperature is reduced to below the gelation temperature for the carrageenan. The newly formed beads, still suspended in the cooled oil, are then passed along to a separation tank 13.

Separation tank 13 is supplied with a KCl solution that is held in tank 14. Oil rises to the top of separation tank 13, while the hardening beads become suspended in the KCl solution that is delivered from tank 14. The oil is then decanted, and pumped by pump 15, for re-use. The beads, suspended in the KCl solution, are then pumped to a holding tank 16, from which they are withdrawn as required for loading a bio-reactor, (such as that which is depicted in FIG. 9).

Referring now to Figure 9 immobilized yeast cell beads were prepared substantially as described in accordance with the present invention and utilized in a gas-lift bio-reactor 17 for carrying out a primary beer fermentation process. The reactor was insulated to manage ambient heat loss, and a thermal glycol jacket 18 was included to provide temperature control as needed. The reactor was sparged through a sparger 19, into the reactor's draft tube 20. Beer wort was introduced at the base 21 of the bioreactor 17. The fermentation was carried to completion (i.e. substantially all of the fermentable wort sugars were consumed).

Reagents

K-carrageenan (type x-909, lot 330360, Copenhagen Pectin, Denmark) is a thermogelling polysaccharide extracted from red algae. Its gelation temperature depends on the concentrations of both K-carrageenan and potassium chloride (KCl). The polymer was dissolved at 80°C to a concentration of 30 gL^"1 in distilled water containing 1.5 gL^"1 of KCl. Under these conditions, the gelation temperature is 28°C. The solution was mixed in a 40 L bioreactor (New Brunswick Scientific Co. Inc. Edison, USA), maintained at 80°C for 20 rnin in order to completely dissolve the polymer, then sterilized at 121°C for 30 min and finally cooled and maintained at 40°C.

A commercial grade of canola oil (Pasquale Bros. Inc. Weston, Ontario, Canada), was sterilized for 1 hour at 121°C, and stored at room temperature (20°C).

Kenics Static mixers

Kenics static mixers (Figure 1) were purchased from Cole Parmer Instrument Company (Niles, Illinois, USA). Three mixer diameters (6.4, 9.5 and 12 mm) were tested. The elements were fixed inside tubes with an internal diameter equivalent to that of the static mixer diameter. The number of elements was varied from 12 to 120.

Bead production process

The 40°C sterilized polymer was pumped (Masterflex peristaltic pump, model 07J23- 20, Cole Palmer Company, Chicago, USA) to a first static mixer as shown in Figure 2, and mixed with 40°C inoculum (10⁷ cells per ml), prior to passing through the mixer. The dispersion of the inoculum into the carrageenan solution was achieved by means of the first mixer.

The inoculated carrageenan solution was then mixed with the canola oil (40°), then passed through the second static mixer to form the carrageenan/oil emulsion. The resulting emulsion was rapidly cooled to 5°C initiating gelation of the polymer droplets. The emulsion was mixed with cold sterile 22 gL^"1 KCl in a separator settler, and the beads recovered by phase partitioning to the KCl solution where gelation was completed. The process oil was recycled back to the oil feed tank and the aqueous bead suspension transferred to a holding tank prior to loading into an airlift bioreactor.

The sterilization of the various lines including the static mixers, and tanks was achieved with 121°C steam for lh. Accumulated condensate was cleared with sterile compressed air injections.

Bead diameter measurement

Beads were sampled into 22 gL^"1 KCl and stored at 4°C prior to analysis. Bead diameter was measured by placing beads in a petri dish containing a thin film of water, observing them with a video camera (Pentax macro 50 mm with possible enlargement to 30 and 80 times) and analyzing the image with Optimas image analysis software (Version 4.02, Bioscan, Inc, Washington, USA). A total of 300 to 400 beads were measured per sample. The Optimas software package capability lies between 100 μm and several mm with a maximum absolute error of 30 μm.

The size distributions were analyzed on a number frequency (%lμm) versus bead diameter curve and characterized by its mean diameter (d_m), standard deviation (σ), and coefficient of variability (CV = σ/d_m).

Examples:

A total of 98 example trials were run, by varying process conditions, and evaluating the resulting bead mean diameter (d_m) , standard deviation (σ) and coefficient of variability (CV). Of the 98 trials, 38 were run in triplicate, giving a total of 174 trials. Examples were grouped into two subsets of trials. In the first subset, the following conditions were tested:

- 6, 12, 24, 48, 60, 72 and 120 elements (N_e) within the static mixer were tested

- linear fluid velocity (V) through mixer was varied from 1.7 to 23.8 cm sec^"1

- static mixer diameter (D) was constant at 12.7 mm

- volumetric fraction of the K-carrageenan solution to oil (ε) was constant at 12.5%

In the second subset:

- volumetric fraction of the K-carrageenan solution (ε) was varied from 8.3 to 50%

- linear fluid velocity (V) was varied from 1.7 to 23.8 cm sec^"1

- 6.4, 9.5 and 12.7 mm static mixer diameters (D) were tested

- number of elements (N_e) was held constant at 24 Liquid linear velocity (V) in the static mixer was calculated as the sum of the oil and K-carrageenan volumetric flow rate divided by the mixer cross-sectional area.

Beads produced by the static mixer were characterized by a unimodal size distribution. Satellite peaks were not apparent, and distribution data was fitted to a normal law curve calculated with the sample mean and standard deviation (Figure 3). Kolmorogof Smirnov's method (Sheaffer and McClave, 1990) was used to confirm normality. Mean diameters of the two typical bead preparations illustrated in Figure 3 were 558 ± 142 and 811 ± 190 μm respectively. Beads were spherical, discrete and ranged in diameter from 20 μm to about 2 mm.

The mean size, d_m, and standard deviation, σ, values obtained from triplicate samples from of the same treatment were calculated for 38 of the 98 experimental trials. For the whole experimental design, the error of d_m values ranged between 10 and 70 μm while the error of standard deviation, σ, values ranged between 10 and 40 μm. These variations were similar to the absolute error generated by the image analysis system (30 μm). Therefore, it can be concluded that the process generated reproducible size distributions.

The number of stationary mixing elements in the static mixer, N_e, effectively defines the mixing time of the emulsion. A small number of elements can lead to incomplete emulsification while too large a number would constitute a loss of energy and investment. Between 6 and 120 elements were tested. Six elements were not sufficient to achieve good emulsification, and 12 elements were insufficient at low linear velocity (less than 7 cm s^"1). For a larger N_e, the bead mean diameter, d_m, decreases asymptotically, both with N_e, and with V, as seen in Figure 4a. It is evident that appropriate choice of N_e and/or V would provide considerable control over the resulting mean diameter of the immobilized cell beads. The range of d_m values possible under the range of configurations described by Figure 4, runs from 480 to 1211 μm.

Increasing the linear liquid velocity, V, through the static mixer, provides more mixing energy, thus the emulsion droplet diameters are inversely proportional to the energy required to break droplets. Thus as the droplets become smaller (d_m decrease), the energy required to further breakup smaller and smaller droplets, increases to some higher power of the droplet diameter. Generally, the effect is more noticeable at low V, than at high V as the bead diameter d_m asymptotically approaches a minimum as seen in Figure 4a. The linear liquid velocity, V, would also affect the mixing time and mixing intensity, in a similar manner to that of increasing N_e.

The coefficient of variability, CV, also decreases asymptotically with N_e, but is independent of V as shown in Figure 4b. The CV ranged from a low of 43% with N_e of 120, to a high of 61%, with the smallest number of mixing elements (N_e = 12).

While increasing the number of elements (effectively, the mixing time), the emulsification equilibrium is approached as a first order kinetics, in good agreement with observations both for d_m, and the CV. In subsequent examples, it was assumed that 24 elements would constitute a good compromise between efficiency and cost.

In the second set of examples, the impact of K-carrageenan volumetric fraction, ε, and static mixer diameter, D, were studied over a range of linear fluid velocity through the mixer. For volumetric fraction (carrageenan-yeast/oil) at or above 50%, the dispersed and continuous phases were inverted, resulting in the inclusion of oil droplets within a carrageenan sol. At the other end of the scale, the productivity was considered unacceptably low below ε = 8%.

Figure 5a illustrates a linear decrease in d_m with increasing V through the 9.5 mm diameter, D, mixer. Bead mean diameters were similar at all carrageenan volumetric fractions, ε, except that beads formed at higher ε values deviated increasingly from linearity at low linear flow velocities, V. The largest deviation from linearity occurred at ε = 50%, the carrageenan volume fraction approaching the inversion point of the emulsion. In all cases, the number of mixer elements, N_e, was fixed at 24. The results were similar (data not shown) for mixer diameters of 6.4 and 12.7 mm, in which the bead mean diameters, d_m, were similar for all values of carrageenan volumetric fraction, ε, with upward deviations from linearity at low volumetric flow rate, V, particularly at high ε.

The coefficient of variability for the data shown in Figure 5a, is illustrated in Figure 5b. CV values covered a fairly narrow range (45 to 57%), without a particular trend evident, either with ε or with V.

The effect of the static mixer diameter, D, on the resulting bead mean diameter, d_m is illustrated in Figure 6, for three mixer diameters and carrageenan volumetric fractions ranging from 8.3 to 50%. It is clear that the effect of increasing D, is to increase d_m, at the same liquid linear flow rate, V. The effect is observed, over most of the range of V tested, although the differences become smaller at higher V. The trend observed in Figure 5, with d_m decreasing linearly with V, particularly at higher V, is also observed in Figure 6. Upwardly increasing deviations from linearity are seen at lower V. The range of possible d_m values possible with variations in D and V, are considerable, extending from less than 400 μm, to beads with mean diameters approaching 1 mm.

The coefficient of variability for the data shown in Figure 6, is illustrated in Figure 7 for a representative set of data (N_e = 24; ε = 8.3%). Results for the other cases (ε values of 12.5, 25 and 50%) were similar to those illustrated in Figure 6. The CV for ε of 8.3% remained relatively stable with increasing V, and it is evident that the smaller mixer diameters resulted in the lowest levels of bead size distribution, as measured through the CV. For example, the CV values for a 6.4 mm mixer diameter ranged from 36 to 44%, while CV values for a 12.7 mm mixer diameter ranged from 54 to 65%.

There are few immobilization technologies available for the encapsulation of living cells, on a scale required for industrial application. This is particularly trae when it is desired to formulate smaller diameter beads, with mean diameters in the tens or hundreds of microns. Single droplet formulation techniques are common in the lab, but are generally not appropriate for industrial use due to the high viscosity of the gelation polymers, making the production of large batches of small diameter beads impractical. Emulsion technologies show promise since they are well suited to large scale, continuous production, and are best suited for the formulation of smaller diameter beads. The drawback with emulsion-gelation or emulsion-polymerization technologies is that a solvent or oil phase is required for dispersion, requiring a more difficult phase separation step than that required when using droplet extrusion methodologies. Secondly, emulsions involve breakup and coalescence of droplets, leading to characteristic droplet size distributions. Sorting beads by diameter through sieving leads to wastage of expensive encapsulant, thus if a size distribution can be tolerated, large batches of controlled mean diameter beads may be formulated when required for processes such as that required for continuous brewing operations.

The characteristic size distribution obtained with a Kenics-type static mixer used in the present study is unimodal, without apparent satellite peaks. Poncelet et al, (1992) observed primary, secondary and satellite peaks corresponding to beads with diameters less than 200 μm, in alginate microspheres produced by emulsion dispersion in a batch stirred tank. With the static mixer, it is possible that very small microspheres, corresponding to the satellite peak region in the case of alginate formed in a stirred tank, are simply washed-out during phase-transfer and washing, and thus may not appear in the size distribution. In this study, beads were formed with mean diameters ranging from a low of 350 to a high of 1200 microns. Because of the diameter range in the size dispersion, beads were produced with diameters ranging from 20 microns up to 2 mm. Thus emulsion technologies provide considerable flexibility in choice of mean diameter, as long as the size distribution can be tolerated.

Several parameters were shown to have an influence on bead mean diameter, including linear fluid velocity, V, static mixer diameter, D, and carrageenan sol volume fraction, ε. Within this combination of emulsification parameters, considerable flexibility in the choice and control of mean bead diameter is possible.

The energy required to form an emulsion is proportional to the interfacial area created between the polymer aqueous dispersed phase and the continuous oil phase. Thus the smaller the bead size, the larger is the energy required for their formation. Berkman and Calbrese (1998) have shown that an increase in the average fluid velocity through a static mixer would promote an increase in the dissipated energy per unit fluid mass, thus favoring a reduction in emulsified droplet size. This was evident in the present study, as an increase in the linear velocity resulted in a decrease in the bead mean diameter. Increases in fluid flow result from a pressure differential across the mixer, proportional to the dissipated energy per unit mass of liquid. An increase in velocity therefore induces an increase in dissipated energy favoring a reduction in bead size. However, the mean bead diameter reduces toward a minimum at equilibrium, at higher fluid velocities. Taweel and Walker (1983) have shown that a dynamic equilibrium is established between the formation of droplets (beads) and the coalescence between droplets for high velocities corresponding to significant turbulence levels. For constant mixer diameter, D, carrageenan volume fraction, ε, and number of mixer elements, N_e, the fluid superficial velocity had no effect on the coefficient of variability. Fluid velocity is therefore a factor that uniquely affects the average bead diameter, but has little or no affect on the size dispersion.

At low fluid velocities, an increase in K-carrageenan volumetric fraction, ε, would favor the coalescence between emulsified droplets, thus entailing an increase in bead mean diameter and size dispersion. Over most of the range of fluid velocity, V, tested, carrageenan volumetric fraction, ε, did not affect the bead mean diameter, d_m or the size dispersion, CV. An increase in the volumetric fraction of polymer solution, at high fluid velocities, would therefore not increase the rate of coalescence between emulsified droplets. For the 12.7 mm diameter static mixer, a maximum ε of 25% was studied due to pump limitations. If these results were confirmed for larger mixers, increases in productivity would be conceivable. Audet and Lacroix (1989) studied this parameter extensively for the production of K- carrageenan beads in vegetable oil using a bi-phase dispersion in a stirred tank and concluded that ε had no affect on the mean bead diameter for a K-carrageenan concentration of 3%. The results therefore with a stirred tank and a static mixer were thus similar in this respect.

An increase in the static mixer diameter, D, would result in a larger range of shear forces at the same linear fluid velocity, increasing the size dispersion of the emulsified droplets, and thus the resultant beads. An increase in D would also decrease the overall intensity of the shear forces, thus increasing the mean bead diameter. Both of these effects were observed in the present study. Pressure differential across the static mixer was not measured in this study, but it would seem to play a role in controlling V, and thus understanding the effect of D in combination with V, on both bead diameter and size distribution.

Increasing the number of mixing elements in the static mixer, results in increased emulsification time, due to the extended residence time within the mixer. An increase in the number of mixing elements also increases local fluid velocities, and thus the intensity of mixing, while minimizing the large shear gradients typical of mechanically mixed vessels such as those using turbine impellers. A more homogenous mixture was expected due to a reduction in the size of the larger beads, ultimately leading to a reduction in d_m and CV. Experimentally, an equilibrium was achieved at between 60 and 72 mixer elements, where no further reduction in d_m or CV was observed. Middlemen (1974) showed that 10 elements were sufficient to attain such an equilibrium in the case of emulsions made up of low viscosity constituents (0.6 to 1 cp). The K-carrageenan solution used in these examples (30 g L^"1) had an average viscosity of 200 cp while the oil viscosity was 25 cp. The large difference in the number of mixing elements needed to reach an equilibrium can be explained by the high viscosity polymer which would inevitably require a longer mixing or residence time in order to reach equilibrium.

The objective for the purpose of manufacturing yeast immobilized gel beads for continuous brewing operations, is to operate the static mixer at high productivity while formulating gel beads with a size distribution between 200 μm and 1.6 mm. In a related study, inoculated bead batches were produced (about 150 L) in order to feed a continuous bioreactor. The results obtained demonstrated that the primary brew fermentation time could be reduced from the classical batch fermentation of six to seven days, to 20 hours (unpublished data). Assuming a static mixer diameter of 12.7 mm, a K-carrageenan volume fraction of 25% and a production of 740 μm diameter beads, one would obtain a productivity of nearly 10 L h^"1 beads. Larger production volumes could be obtained by increasing the mixer diameter. For example, with a 5 cm static mixer, the productivity can exceed 200 kg h^" \ Producing smaller beads will require a higher fluid linear velocity and thus, a higher productivity. Reducing bead size from 600 to 300 μm, would double the productivity. The opposite situation is encountered in more conventional bead production approaches using droplet extrusion technology (Poncelet et al, 2000).

An increase in productivity is necessary in order to operate at an industrial scale. Consequently, an increase in the flow of polymer and oil with the static mixer used in this study would induce the formation of beads too small to be used in the fermentation stage. It is therefore necessary to increase the diameter of the static mixer thus increasing the diameter of the beads. However, if the results obtained in this study can be extrapolated, the use of static mixers with larger diameters will increase the bead size dispersion, producing a larger percentage of beads outside the set guidelines. Another possible alternative would be the implementation of a system using multiple static mixers of medium size (12.7 mm) placed in parallel. Productivity reaching 100 kg h^"1 (with 10 mixers, 600 μm beads) is therefore conceivable. Another solution would be to study the characteristics of different static mixer designs in order to determine their efficiency.

In summary, both the static mixer diameter and the carrageenan volumetric fraction affect the bead mean size but mainly in correlation with the fluid linear velocity through the mixer. This last parameter remains the most important factor, followed by the static mixer diameter. From the point of view of size dispersion, the coefficient of variation is mainly and only affected by the static mixer diameter. This observation raises concerns about scale up. While the coefficient of variation is relatively good for smaller diameter static mixers (35% - similar to that obtained with turbine mixers), it increases with the mixer diameter, and thus is negatively affected on scale-up. Accordingly, in a preferred practice of the present invention a plurality to processes are employed in parallel to produce at a given productivity rate, with the individual mixers being sized to produce a bead population having a predetermined variability.

Claims

Claims:

A process for the statistical control of bead production in static mixer processing of an emulsion wherein bead particle size is controlled by a selected one or more of: average linear velocity through the mixer mixer diameter and bead size distribution controlled by a selected one or more of mixer diameter number of mixer elements

The process according to claim 1, wherein a randomly sampled population of beads produced by said process has a unimodal size distribution.

A process for production of beads comprising a plurality of mixers selected and operated to maximize productivity, given that productivity proportional to the mixer diameter squared, for which mixer diameter is selected to produce a population of beads having a desired mean particle size and particle size variance.