Developing an operating scheme
The following guidelines are useful for determining a satisfactory scheme: "
Start at conditions of high product solubility. In the application, the solubility of the reaction product is reasonably high at pH 2 (Figure 1). Relatively large crystals of CaHPO4•2H2O can be grown in a batch reaction by mixing the phosphoric acid and slaked lime in the proper proportions for pH 2, allowing sufficient time for everything to go into solution, and then slowly adding slaked lime until solids begin to form. In this way, nucleation of crystals can be made to occur near saturation conditions— that is, at a reasonably low supersaturation level. A low level of supersaturation can be maintained by continuing slow addition of Ca(OH)2, gradually increasing the pH. This approach allows the system to stay close to the saturation conditions illustrated in Figure 1, thereby promoting growth of existing CaHPO4•2H2O crystals and avoiding rapid nucleation of new ones. In this case, adding the base over a period of about one hour yielded a filter cake consisting of rod-shaped crystals as large as 20 microns in diameter by 100 microns long (Figure 2), plus many smaller crystals on the order of 10 microns in length.
The example thus involves careful negotiation of a steep pH-dependent solubility curve and, in this respect, resembles crystallization of organic acids and bases from aqueous solution by adjustment of pH across the pKa value  to form the non-ionized species. For example, we are familiar with an industrial crystallization process involving the addition of an inorganic base to an aqueous solution containing an ionized organic compound with amine functionality (R-NH3+). Addition of base deprotonates the ionized species to crystallize the free base form (R-NH2) as pH is raised above the pKa for that compound. Another example involves crystallization of organic acids from an aqueous solution by slow addition of a mineral acid . In that case, the ionized form of the organic acid (R-COO-) is protonated as the pH is reduced below the pKa value, and the protonated form (R-COOH) crystallizes from solution.
Avoid starting at conditions of low product solubility. Approaching the precipitation from basic conditions by adding acid to a pool of slaked lime forced rapid nucleation. This is because hydroxylapatite, the product obtained at alkaline conditions, is essentially insoluble. Because product solubility was so low, the level of supersaturation was always high, even if acid was added slowly. As a result, nucleation dominated growth. We could not grow particles any larger than about 10 microns in size (Figure 3), and subsequent filtration was slow. Comparing Figures 2 and 3 provides a dramatic illustration of the impact that supersaturation level can have on crystal size.
To succeed at commercial scale, it is crucial to take three factors into account: Cycle time adjustments. In scaling up a batch processing scheme developed in the laboratory, adjustments in cycle time may be necessary. This is because mixing quality generally declines on scale up, and a decrease in mixing quality can cause an increase in localized supersaturation levels in the vessel. As mentioned earlier (and worth emphasizing), slowing down the base addition rate (or diluting the base) may help mitigate this effect. Slow addition of reagent is most important at the beginning of the crystallization; the addition rate may be increased toward the end of the batch for increased productivity.
Mixing and heat transfer effects. It is also important to note that performance at the commercial scale can vary significantly depending upon the mixing and heat transfer capabilities of the equipment selected for the operation. Often it makes sense to use an existing vessel to minimize installation costs — however, it is essential to ensure that reasonably good performance can be achieved. For example, mixing impellers should be selected to provide good axial-flow circulation within the vessel with minimal shear. Retrofitting an existing vessel with a more-efficient impeller (or set of impellers) can provide substantial improvement in crystallizer performance. For a glass-lined vessel with a retreat-blade or “crowfoot” impeller, consider a more-efficient axial-flow impeller as a replacement. Also consider adding baffles to the tank for improved mixing performance. (Consult the vessel manufacturer to explore available options.) Although retrofitting a new impeller or baffles into a glass-lined vessel may be a difficult proposition, doing so can yield significant improvements in crystallizer performance. Replacing an existing pitched-blade impeller with a more-efficient and properly-sized hydrofoil design may also boost performance depending upon the application. In addition, we have found it useful to specify a variable-speed motor or mechanical drive – to allow adjustment of impeller rotational speed on startup in the plant. This provides some added flexibility for optimizing the operation in the field. (For more detailed discussions of mixing considerations, consult Refs. 3-5.)
Heat transfer is particularly important in the present example because of the highly exothermic reaction. A pump-around heat exchanger sometimes is used to improve heat transfer capability instead of relying on a heat transfer jacket alone. However, keep in mind that the circulation pump should be selected to avoid generating too much shear; excessive shear can result in unwanted crystal nucleation as well as crystal attrition or breakage. Vendors offer a number of low-shear pump designs as alternatives to standard centrifugal-pump designs.
Feed pipe location. For the present application as well as similar reactive crystallizations, we adding reagent to the crystallizer vessel through a dip pipe located below the liquid level – to facilitate rapid distribution of reagent throughout the vessel during crystallization. Locating the outlet of the feed pipe near the suction side of an axial-flow impeller works well. This can yield much better results than simply pumping reagent into the vessel through the head space on top of the liquid. In our application, we installed only a base-addition dip pipe. An acid-addition dip pipe is not needed in this case.
The authors thank Joseph Bonadies, Jr., Ray Collins, Sumnesh Gupta, Kishore Kar and Tracy Kiefer for many valuable discussions.
1. Gard, D.R., “Phosphoric Acids and Phosphates,” Volume 18 in “Kirk-Othmer Encyclopedia of Science and Technology,” 4th Ed., J.I. Kroschwitz and M. Howe-Grant, eds., John Wiley & Sons, New York, pp. 669-718 (1996).
2. Elliot, J.C., “Structure and Chemistry of the Apatites and Other Calcium Orthophosphates,” Elsevier, Amsterdam (1994).
3. Genck, W.J., “Crystal Clear – Part 1,” Chemical Processing, Vol. 66 (10), pp. 63-67 (Oct. 2003).
4. Genck, W.J., “Crystal Clear – Part 2,” Chemical Processing, Vol. 66 (12), pp. 37-42 (Dec. 2003).
5. “Handbook of Industrial Crystallization,” 2nd Ed., A.S. Myerson, ed., Butterworth-Heinemann, Boston (2002).
6. Tadayyon, A., S.M. Arifuzzaman and S. Rohani, “Reactive Crystallization of Brushite Under Steady State and Transient Conditions: Modeling and Experiment,” Ind. Eng. Chem. Res., Vol. 42, pp. 6,774-6,785 (2003).
7. Kelkar, V.V. and K.M. Ng, “Design of Reactive Crystallization Systems Incorporating Kinetics and Mass-Transfer Effects,” A.I.Ch.E. J., Vol. 45(1), pp. 69-81 (1999).
8. Bechtloff, B., P. Juesten and J. Ulrich, “The Kinetics of Heterogeneous Solid-Liquid Reaction Crystallizations – An Overview and Examples,” Chemie Ingenieur Technik, Vol. 73(5), pp. 453-460 (2001).
9. Butler, J.N., “Ionic Equilibrium: Solubility and pH Calculations,” John Wiley & Sons, New York (1998).
10. Wang, F. and K.A. Berglund, “Monitoring pH Swing Crystallization of Nicotinic Acid by the Use of Attenuated Total Reflection Fourier Transform Infrared Spectrometry,” Ind. Eng. Chem. Res., Vol. 39, pp. 2,101-2,104 (2000).
Tim Frank is a research scientist and senior technical leader in the Engineering and Process Sciences Laboratory at The Dow Chemical Company, Midland, Mich. E-mail him at firstname.lastname@example.org.
Wayne Fort is a senior research engineer in the Agricultural Chemicals Technology Center at The Dow Chemical Company, Midland, Mich. E-mail him at email@example.com.
Chris Jones is a senior research engineer in the Engineering and Process Sciences Laboratory at The Dow Chemical Company, Midland, Mich. E-mail him at firstname.lastname@example.org.