kilomentor | 09 March, 2007 12:09
Training in Chemical Process Research & Development
Besides reading a free blog, there are other ways to obtain training or professional mentoring in chemical process development; some comparable, some complementary.
The best chemical process development book on the market is incontestably, Practical Process Research & Development, by Neal G. Anderson, Academic Press ISBN 0-12-059475-7 (about $115). I unreservedly recommend this book to anyone who has the resources to purchase it.
The only competing book is Principles of Process Research and Development in the Pharmaceutical Industry, by Oljan Repič, Ph.D., A Wiley-Interscience Publication, John Wiley & Sons, Inc. Although this book is nowhere near as useful in the sense of being a training manual there are so few useful books in the subject area that it should be mentioned approvingly. Besides there is a personal connection, Oljan and I were graduate students together at Harvard with Professor Woodward. If you can afford two books, get both of these.
Scientific Update is an organization that sells process development training conferences/seminars/courses that can extend over several days. I have evaluated their material and it is useful, even although they do not teach some of the most useful methods which you will learn from Kilomentor. The cost of their meeting is high at about ₤1050 or $2042US. Purchasing the books recommended above is much better value. For the same money as one course/conference registration you could have 17 copies of Practical Process Research & Development; one for each key staff in an organization or 8 copies of each of these books. I certainly would not recommend that anyone attend the conference, who has not already thoroughly internalized one of the above books’ teachings.
Direct, Inverse and Simultaneous Additions
In this blog I will speak about additions in reactions. Direct additions are defined here as additions in which at least the reagent is added to the substrate.
Inverse additions are defined here as additions in which at least the substrate is added to the reagent.
Simultaneous additions are defined here as additions in which both the substrate and the reagent are added at least partially simultaneously to the reactor.
The mode of addition in a process step becomes important for fast or very fast reactions. A fast reaction will be defined here as one in which there is more than a trace (5%) of product or an intermediate present in the flask at the half-addition point.
A very fast reaction will be here defined as one in which there is only a trace of the substrate or reagent being added if the addition is stopped and the mixture analyzed at some point after at least 5% of the addition has occurred.
These definitions are based on the half-addition test recommended by R. Carlson who teaches that to discover whether the rate, order, and direction of addition is going to affect the purity and/or yield take a sample for analysis at the point when half the addition has been completed. The rate and direction of additions in a process step can be significant for either fast or very fast reactions.
Is the Stoichiometry ever realized in the Reactor?
When we think of a chemical reaction, we normally think of the substrate and the reagent reacting together according to the stoichiometry of the balanced equation that we have written. That is:
x S (Substrate) + y R (Reagent) → z P (Product) + w C (Co product)
where the integers x ,y, z and w are selected to balance the chemical equation.
In fact, except for reactions where the rate of reaction is slow compared to the rate of addition (direct or inverse) of substrate with reagent, this instantaneous stoichiometric ratio is never really approximated at any time in the reactor.
In the laboratory, where it is very easy to cool the reaction flask because it has a high surface to volume ratio, it may be possible to do an addition quickly, without worrying about a runaway exotherm, and to get concentrations close to the reaction equation’s stoichiometry in the reaction flask.
Also in the laboratory one can combine reactants with solvent in the stoichiometric ratios at a low temperature in the reaction vessel and then warm them up to the reaction temperature. When this can be done it is possible because the cooling capability in the lab under favorable circumstances can be greater than the exothermicity: nevertheless, even in the lab for many reactions such as Grignard formation, this protocol more often results in an uncontrollable reaction. In any case, certainly these protocols cannot be considered for reactions on scale. Generally when large quantities of chemicals are involved one reactant is added very gradually to a mixture of all the other reactants at a temperature sufficient to cause all of the added chemical to be essentially immediately consumed so that there is no build up that can fuel an uncontrolled exotherm that might otherwise set in after the addition was complete.
When a solution of the reagent is being slowly added, it encounters the full equivalents of the other reactants, creating ratios of reagent to other reactants very far from the ratios expressed in the balanced equation. The result is that any undesired reactions that have a higher molecularity in the reactants already completely in the flask and an equal molecularity in the reagent being added will have an increased tendency to occur and may produce an undesired by-product. The main point here from the process chemist’s perspective is that this tendency will be exaggerated at the scale increases, because as the scale increases it is mandatory to have longer addition time.
For example, in the situation where reagent R is being slowly added to Substrate S aiming to get product P; if the substrate S can react with an intermediate I, and if the addition rate is slow compared to the rate of reaction; you are very likely to get the reaction:
I + S → I-S (where I-S is an overreaction product) because the intermediate I is being formed quickly at the beginning of the addition when it will be still in the presence of a very large excess of the unreacted substrate S. Thus by the law of mass action intermediate I will have an increased likelihood to react with all the S, which is concentrated in the reactor.
In a similar way, in the situation where reagent R is being slowly added to Substrate S aiming to get product P; if the product P is reactive with the starting material you are very likely to get some reaction:
P + S → P-S (where P-S is an overreaction product) because at the middle of the reaction there is much more substrate S than reagent R to encounter the initially formed product P.
In each of these situations an inverse or simultaneous addition needs to be considered. There are corresponding situations where the direct or simultaneous additions are preferable on this theoretical basis.
Additions in the Laboratory
Additions in the laboratory can be very fast indeed. One can pour 100 ml of solution into a 250 ml r.b. flask in a few seconds. On scale it is not possible to copy this. The absolute volume of solution is much more; it must be pumped in or run in by gravity through a constricted line; besides, the enthalpy change would most likely be unmanageable. For these reasons, the rate of addition becomes a variable of significant concern in scale up because when we go to large reactors the addition rate is severely constrained compared to the lab situation.
Besides the instantaneous addition of a solution, which I have just posed as an example which is simply impossible other possibilities are seriously discouraged on scale. In the laboratory so long as a strong stream of inert gas is maintained over the reactor surface and the reaction vessel is in a fume hood, a glass stopper can be removed briefly and a solid reactant poured in through a powder. The equivalent would not be acceptable in the chemical plant. The operators would be exposed to chemical contamination, as would the atmosphere in the plant and the inertness of the reactor atmosphere would be seriously compromised. Also, the addition would not be adequately reproducible and if there were solvent already in the reactor when the solid addition was made there could be a dangerous splash back.
A number of reactions require the slow and controlled addition of a solution containing one reagent to another. These are ideal for scale up. Slow addition is both necessary and simple to achieve on-scale; rather, the technical difficulty that the process development chemist needs to solve is how to duplicate in the laboratory these slow additions on scale to model the process.
Attempting to control the flow rate over a number of hours using a conventional constant pressure addition funnel is a frustrating exercise. An inexpensive way to increase the sensitivity of an addition funnel is to use a Teflon stopcock modified with a groove cut in it. This will allow one to open the stock-cock more gradually by slowly exposing the groove leading to the cylindrical channel in the plug to the liquid in the funnel rather than having to expose partially a circular hole. I first read about this idea in the Laboratory Manual, Research Techniques in Organic chemistry, Robert B. Bates and John P. Schaefer, Foundations of Modern Organic Chemistry Series Prentice Hall Inc., Englewood cliffs, N.J.1971. p. 14.
The Herschberg dropping funnel is the classical method for slow controlled additions in the laboratory. These have been further improved by making the pressure equalizing side arm to follow the Mariotte principle that assures the drop rate regardless of the liquid level. Internal delivery scale is graduated to allow accurate resetting on successive runs. Such a funnel is shown in a Figure in Fieser & Fieser Vol. 1 pg. 783 (under 1,4-pentadiene). These funnels are now called constant-addition funnels and can be obtained from commercial glassware suppliers.
Syringe pumps are alternative means to provide consistent very slow additions on a laboratory scale that can mimic the rates of addition that may need to be used on-scale.
An (expensive) peristaltic pump or syringe pump overcomes these problems but can introduce other complications. An additional advantage however is that using dual syringe injection two different solutions can be simultaneously slowly metered by the same mechanism. Such a use is described by E.J. Corey and Eric Block, New Synthetic Approaches to Symmetrical Sulfur-Bridged Carbonyls, J. Org. Chem. 31(6) 1665-6 1966.
With the use of a syringe pump, there are obvious difficulties associated with purging the solution and assembling such an apparatus under nitrogen, and Peter Osvath [. J. Chem. Educ. 1995 72 658] reported a simple and inexpensive homemade apparatus that can replace the single syringe pump rate of addition. A Male Luer Lock tip (recovered from a broken syringe) was sweated onto the flattened tip of a pressure-equalizing addition funnel and a syringe needle was attached. Judicious selection of needle length, bore size, and reactant volume can be used to control the addition time simply and reproducibly. With a 250-mL funnel, the flow rate changes by <25% from the beginning to the end of the addition. (In fact, a reduction in the rate of addition may even be advantageous as the reaction proceeds, the reagent in the receiving flask is consumed, its concentration drops, and the rate of reaction will decrease). A piece of fine Teflon tubing of appropriate length attached to the needle can be used to reduce the flow rate even further, but this is only necessary for very slow rates of addition. For example, the time of addition of 200 mL, of an ethanolic solution could be varied from approximately 5 minutes (150mm/17 gauge) to approximately 5 h (200mm/22 gauge), and once the addition time for a particular needle length/bore is determined, the tap on the addition funnel is turned fully on, so no adjustment is necessary. When needles with a particularly fine bore are used, a small plug of glass wool should be inserted in the constriction above the tap, to filter the solution and prevent blockage of the needle. An inert atmosphere is readily maintained throughout the system.
For the slow addition of only about 5 ml another technique has been described [Goran Magnusson, J. Chem. Educ., (56) 410 1979] which is prepared from two disposable pipettes and a short length of latex tubing. A drop rate of as low as one drop every 20 seconds is easily obtained even with low-viscosity solvents such as ether.
High dilution apparatus built with a dilution chamber can also be used. A classical example is the apparatus used in the Acyloin reaction for preparing large ring systems.
[N.J. Leonard and C.W. Schimelpfenig, Jr., J. Org. Chem., 23, 1708 (9158)].
In such high dilution apparatus there are two aspects:
The first factor controls the absolute concentration in the bulk of the reaction solution and the second controls the concentration of reactant in the unmixed addition stream at the point of entry into the reaction mixture. In a completely homogeneous reaction the first is almost exclusively important but for heterogeneous reactions occurring on the surface of catalysts or on solids insoluble in the reaction mixture the second is also of significance. The acyloin condensation of esters to α-ketols is thought to occur at low concentration on the surface of the molten sodium metal.
The addition of liquid reagents with relatively high freezing points should often be diluted with solvent when they are being added into a cold reaction mixture to prevent the reagent from freezing in the addition tube.
Gaseous or low boiling reaction constituents can often be preferentially added below the surface of the reaction mixture because this insures more complete absorbtion into the mixture; otherwise the low boiling material may be lost in the off gases perhaps accelerated by an appreciable local exothermicity at the point of addition.
In very fast reactions where the rte of reaction is competitive with the rate of mixing, it may be advantageous to spray the solution being added unto the surface of the reaction mixture thus providing smaller droplets, faster effective mixing and a higher surface area to promote the reaction.
Reactions that are likely to be sensitive to the rate and mode of addition can often be predicted in advance on the basis of a general understanding of the mechanistic details. A susceptible reaction is characterized by the presence of an intermediate or the final product that once formed can react with another equivalent of starting material. The result is that at a point during the addition the intermediate or final product is present along with a substantial amount of starting material or starting reagent. If this situation persists for a considerable time as it often does when the addition period must be prolonged as in scale up, the by-product production will rise. This situation can be mimicked in the laboratory by stopping the addition at some intermediate point, stirring for a time commensurate with the expected total addition time on scale, and then completing the addition and proceeding with the process step. If the step is susceptible to this problem the amount of impurities will rise.Specific examples of such situations are described in Anderson’s book.