(109a) What Can One Do with a New Separation Process?

It is now clear that we have been caught napping with outmoded technology in many key areas of our economy, and the basic reason for this is our tendency toward conformity. This is a deeply imbedded human trait that may well have originated when distant ancestors left the relative safety of trees to live in savannahs. The fearsome predators there could only be kept at bay by presenting united fronts of several individuals. Loners were quickly weeded out, and most of us are still uncomfortable going our own way. The great utility of original thinkers sufficed to keep a few in the gene pool, but they have always been viewed with suspicion ? and not least by CEO's and granting agencies. As a result we tend not to have effective research strategies until we are forced to develop them by major challenges. Now after a half century of a nearly free ride thanks to favorable post WWII circumstances the challenges have arrived.

As Alan Greenspan [The Age of Turbulence, Penguin 2007] has pointed out a dynamic economy is normally characterized by continuing constructive destruction: the replacement of established processes and products - or whole societies - by new and more economical ones. Some overseas competitors seem to realize this better than we.

Mark Buchanan [The Mathematical Mirror to Animal Nature, 2008, 453, 714-716] has neatly described a closely analogous situation faced by most animals: Food is not, in general, spread equally around the world. It comes in lumps. Foragers thus need a strategy for finding those lumps. So do we, and just following others is not in general an effective strategy. That is proven over and over again, by lemmings for example. And by the stubborn refusal of most industries and granting agencies to fund exploratory research.

Surely this should not be true of downstream processing, a complex mixture of separation processes that dominate production costs in one of our most dynamic industries. We find however, that new separation processes in biotechnology have been remarkably few. This is no doubt true in part because old processes, such as chromatography, are continually being improved and specialized. However, it remains a surprise to this author that true steady counter flow processes, dominant in most other areas of the chemical process industry, do not yet have a foothold. It was decided therefore to produce such a process, ideal counterflow membrane cascades, to see if a place can be found for it, and to suggest a ?foraging strategy?. The strategy is a simple type of emergence: a novel combinations of well established ideas.

This new process has now been reasonably well characterized and shown to give much improved separations relative to the closely related simple tangential flow membrane filters or the inefficient cascades of them reported from time to time in the reviewed literature. It remains to find significant specific applications for which this new process is competitive and to demonstrate its utility.

To do this requires a comprehensive picture of downstream processing requirements and costs for our complex industry, and this is the daunting but important purpose of this talk. Such a task is in essence a mystery, i.e., too complex for complete solution, and we must seek to define a puzzle, a problem simple enough to solve, and powerful enough to provide useful insight. Fortunately we already have a significant body of literature for guidance.

The nature of downstream processing

It is immediately clear for example that a separation trajectory can be drawn on a plot of thermodynamic activity vs. concentration for the desired species (Fig. 1), and it appears that there are three separately identifiable tasks required for success. These are usually defined as

Capture or concentration,

Fractionation, and

Purification or polishing

We shall be most interested in the second of these, but the first provides a very useful framework for our discussion. It should also be noted that it is sometimes possible to combine two or more of these tasks, but that possibility will not be discussed here.

For the conditions of most interest here the first of these three processes dominates manufacturing cost, and it in turn is dominated by simple materials handling [E. N. Lightfoot and MCMC]. It has long been known (Fig. 2, [Sherwood], and [figure author]) that cost is inversely proportional to concentration of the desired product in the feed, and it is remarkably insensitive to the specific nature of the system being investigated. Moreover, the feed is normally a very complex multicomponent mixture and not well suited to processes such as membranes which produce only binary or pseudo-binary separations.

From our present standpoint capture is primarily of interest because it suggests that the major processing cost in the remaining two tasks is yield loss, and that the primary measure of success is yield relative to increase in purity. It is also suggested in this talk that fractionation and polishing can be viewed separately.

The last major insight behind this discussion is the analogy between membrane filtration and the various processes used for isotope fractionation. It suggests that the Dirac value function [see BSL, Ex. 23.1-3, p. 735] may be a useful tool for comparing alternate fractionation processes. This approximation will be examined in the course of discussion.

Finally specific examples will be examined to firm up the discussion and to test the various approximations made in earlier phases of the work. These are still to be selected at the time of writing.