Crystallization of L-Phenylalanine anhydrate for product recovery during fermentation M.C.Cuellar, D.Sanchez Garcia, A.J.J. Straathof, J.J. Heijnen and L.A.M. van der Wielen Delft University of Technology, Department of Biotechnology Julianalaan 67 2628 BC, Delft, The Netherlands M.C.Cuellar@tnw.tudelft.nl L-Phenylalanine (Phe) is an essential amino acid mostly used as an intermediate in the production of the artificial sweetener aspartame. We are studying the crystallization of Phe in relation to its fermentative production with recombinant E. coli. Phe exhibits two crystal forms: a flake-like anhydrate and a needle-like monohydrate. Anhydrate is the commercial form, being easier to handle. Anhydrate is thermodynamically stable above 37 o C. Nevertheless, the monohydrate is the initial form obtained by spontaneous nucleation. Consequently, it becomes difficult to control the quality of the crystals in terms of size and size distribution. In this work, we studied the batch crystallisation of Phe anhydrate above 37 o C in aqueous solution. First, we determined the metastability zone by means of the polythermal technique. Second, the size and the amount of seeds required for obtaining only anhydrate crystals were defined. Finally, anhydrate crystals were grown under isothermal conditions, which allowed us to determine the growth kinetics in water at o C. 1. Introduction L-Phenylalanine (Phe) is one of the most important commercially produced amino acids, being a precursor for the production of the artificial sweetener aspartame. Phe has an estimated market of 10 ktons per year and it is industrially produced by chemo-enzymatic methods or by fermentation. Its production by fermentation is however limited by the metabolic pathway of the micro-organisms, which is highly regulated: the key enzymes involved in the microbial synthesis of Phe suffer from feedback inhibition. This means that the specific productivity of the microorganism decreases as a response to the increasing concentration of product in the fermentation medium. One way of overcoming product inhibition during fermentation is to recover the product from the fermentation medium during the process. Several examples exist in literature describing the use of techniques like extraction [1] and adsorption [2] during the fermentation. However, little attention has been devoted to crystallization. This is remarkable, since most chemicals produced by fermentation are commercialized as solids and therefore, involve one or more crystallization steps in their production process. Buque et al. [3] showed that an external cooling crystallization loop during a bioconversion resulted in improved productivity and selectivity of the catalyst, while product crystals in pure form could be directly harvested. In previous work based on model calculations [4], we showed the potential of such a concept (see Figure 1) when applied in the fermentative production of Phe by a recombinant strain of Escherichia coli. The product concentration in the fermenter could be reduced and 43.6% of the total Phe produced could be directly recovered as crystals.
Feed Liquid medium recycle Cells Product crystals Fermentation S/L separation crystallisation S/L separation Figure 1. Product recovery by crystallization during fermentation by means of an external loop. In the first solid-liquid (S/L) separation step, the microbial cells are retained in the fermenter, while the cell-free medium is sent to a crystallizer. After separating the crystals, the mother liquor is sent back to the fermenter. The success of this concept will depend to a large extent on its ability to produce crystals with the right specifications of purity, shape and size among others. In the case of Phe, two crystal forms are known (see Figure 2). Anhydrate is the commercial form, being easier to handle due to its flake-like structure. The monohydrate, however, is the initial form obtained by spontaneous nucleation in water. The monohydrate may transform into the anhydrate at temperatures where the latter is thermodynamically more stable (> 37 o C). This transformation process can take several hours depending on the process conditions. Consequently, it becomes difficult to control the quality of the crystals in terms of size and size distribution. To overcome this problem, the use of seeds with the right crystal form is often recommended [5]. Using anhydrate crystals as seeds might result in a two-fold effect: on the one hand, the nucleation of monohydrate can be hindered; on the other hand, the size and size distribution of the product crystals can be controlled. anhydrate monohydrate Figure 2. Crystal forms of L-Phenylalanine. In this work, we have studied the batch crystallization of Phe anhydrate above 37 o C in aqueous solution. First, we determined the metastability zone. Second, the size and the
amount of seeds required for obtaining only anhydrate crystals were defined. Finally, anhydrate crystals were grown under isothermal conditions in order to derive the crystal growth kinetics from the desupersaturation curve. 2. Materials and methods For all experiments, Phe anhydrate (Sigma Aldrich) of purity > 99.0% and filtered Milli-Q water were used. Experiments were carried out in a 2 L jacketed vessel with three baffles and two Rushton turbine impellers of six blades operating at 250 rpm. The concentration of dissolved Phe was measured by a gravimetric method using about 2 ml of sample and drying at 60 o C during 38-48 hours until constant mass. Each sample was measured in duplo. Metastable zone width (MZW) determination MZW determination was done by means of the polythermal technique [6], using a set-up as shown in Figure 3. The starting volume in all experiments was 1 L. A Mettler-Toledo Focused Beam Reflectance Measurement (FBRM) probe was used for the detection of the crystallization onset. The solution temperature was controlled with a Lauda programmable water bath. The nucleation point was defined as the temperature at which the increase in fines counts (0-20 µm) was larger than 10%, which also coincides with the increase in the solution temperature due to the heat of crystallization. Two cooling rates (5 and 12 o C/hour) were evaluated. T To PC FBRM probe Liquid sampling (for concentration analysis) programmable water bath 0.2 µm filter Figure 3. Experimental set-up for MZW determination. Size and amount of seeds The seeds were prepared by sieving Phe anhydrate and collecting the fraction between 90 and 212 µm. Based on preliminary experiments (data not shown) the amount of seeds to be added was calculated as 10% of the theoretical yield of crystals. The seeds were added within the metastability zone as slurry prepared with 4.5 ml cold water per gram of seeds.
Determination of crystal growth kinetics The experimental set-up used was the same as for MZW determination but with a working volume of 1.5 L. Three different initial supersaturations (within the metastability zone) were tested (see Table 1). The initial solution was prepared at 54 o C and in the vessel the temperature was reduced to o C. At this point the seeds were added to the solution and samples were taken at different time intervals. Table 1. Conditions for isothermal experiments. Initial supersaturation [-] Amount of seeds [g] 1.08 0.49 2.22 1.10 0.65 2.94 1.17 0.88 4.00 Water for activation [ml] 3. Results and discussion Metastable zone width (MZW) determination The overall MZW results are shown in Figure 4. At the two cooling rates evaluated no significant difference in the metastability limit was observed. From these results it can be seen that the MZW for this system is very narrow, which limits the operational range for the measurement of crystal growth kinetics. 50 Concentration [g/l] 40 30 metastability limit solubility anhydrate 25 solubility monohydrate 20 10 15 20 25 30 40 50 55 Temperature [ o C] Figure 4. Metastability limit and solubility. Markers (connected by the full line) are experimental metastability data; dashed lines are solubility data from the literature [7].
Determination of crystal growth kinetics For each experiment a plot describing the change of solution concentration in time was obtained. A mass balance under the assumption of no nucleation resulted in the following equation: dc dt = 3 M L cr 0 S 0 M g ( k ( S( t) 1) ) g That is, the change of solution concentration is attributed only to the growth of crystals. This equation was integrated analytically, and by minimizing the squared residuals of each experiment, the growth kinetic parameters k g and g were derived. Figure 5 shows the concentration profile and the fitted values for each experiment. In this way, the crystal growth kinetic parameters derived were: g = 0.76 k g = 1.46 10 6 [-] [m/s] Phe concentration [g/kg] 43 41 39 37 S 0 =1.08 43 S 0 =1.10 41 39 37 0 4000 8000 12000 16000 time [s] 0 4000 8000 12000 16000 time [s] 43 41 39 37 S 0 =1.17 0 4000 8000 12000 16000 time [s] Figure 5. Change of solution concentration in time at o C.Markers are experimental data; full line are fitted values. In the integrated fermentation-crystallization system modeled previously [4] the Phe production rate of the micro-organism is about 1.7 g/h. With the obtained kinetic parameters and the mass balance described above it is possible to calculate the crystallization rate at a given supersaturation and seed load. Such a calculation shows that the crystallization rate is much faster than the microbial production rate and consequently, operating at the low supersaturation required for avoiding spontaneous monohydrate crystallization seems to be feasible in the integrated system. 4. Conclusions Spontaneous nucleation of Phe monohydrate above 37 o C can be avoided by seeding anhydrate crystals at a supersaturation within the metastability zone. Maintaining such a low supersaturation during fermentation is expected to be feasible, since the crystallization rate is much faster than the production rate of the micro-organism.
5. Nomenclature C Solution concentration [kg/kg] M S 0 Mass of seeds [kg] M Mass of solution [kg] L cr 0 Mean size of seeds [m] k g Growth constant [m/s] g Growth order [-] S Supersaturation [-] 6. References [1] Maass, D., Gerigk, M.R., Kreutzer, A., Weuster-Botz, D., Wubbolts, M. and Takors, R. Integrated L-phenylalanine separation in an E. coli fed-batch process: from laboratory to pilot scale. Bioprocess and Biosystems Engineering 25 (2002) 85-96. [2] Kusunose, Y. and Wang D.I.C. The enhancement of production of phenylalanine by extractive fermentation with polymeric beads. Chemical Engineering Communications 191 (2004) 1199-1207. [3] Buque-Taboada, E.M, Straathof, A.J.J., Heijnen, J.J. and van der Wielen, L.A.M. In situ product recovery (ISPR) by crystallization: basic principles, design, and potential applications in whole-cell biocatalysis. Applied Microbiology and Biotechnology 71 (2006) 1-12. [4] Cuellar, M.C., Straathof, A.J.J., Heijnen, J.J. and van der Wielen, L.A.M. Towards the integration of fermentation and crystallisation. In: VDI-Berichte 1901 16 th International Symposium on Industrial Crystallization (2005) 975-980 [5] Beckmann, W. Seeding the desired polymorph: background, possibilities, limitations, and case studies. Organic Process Research & Development 4 (2000) 372-383. [6] Barret, P. and Glennon, B. Characterizing the metastable zone width and solubility curve using Lasentec FBRM and PVM. Trans IChemE 80 Part A (2002) 799-805. [7] Mohan, R., Koo, K., Strege, C. and Myerson, A.S. Effect of additives on the transformation behavior of L-phenylalanine in aqueous solution. Industrial and Engineering Chemistry Research 40 (2001) 6111-6117. Acknowledgements These investigations are supported in part by the Netherlands Research Council for Chemical Sciences (CW) and the Netherlands Technology Foundation (STW) in the NWO-research program Separation Technology. The FBRM probe used in this study was kindly provided by DSM Anti-infectives.