n BY STEVE HERMAN
Data, Data, Everywhere
A formula has numerous ingredients and can be varied in thousands of ways.
How can the best one be found? The formulating robot and combinatorial chemistry, to start.
“Perhaps some day the precision of the data
will be brought so far that the mathematician
will be able to calculate at his desk the outcome
of any chemical combination, in the same way
as he calculates the motions of celestial bodies.”
A successful product involves many factors, but none is more important han the formulation itself. However
well a product may be marketed, it is the stu;
in the bottle that earns consumer loyalty,
and the chemical toolbox has a constantly
expanding range of new raw materials. A
formula has numerous ingredients and can
be varied in thousands of ways—how can the
best one be found? An even greater challenge
arises when ingredients interact or change
;e old way to formulate involved a
chemist at a bench making multiple trials.
Large companies have an army of R&D
sta; generating new products. Ideally, there
are some general guidelines of how much
of each material to use or the ratios of some
materials—say the ratio of the anionics to the
alkanolamide in a shampoo—but, still, a lot
of trial and error is involved in getting just
the right properties.
Experiments create data, and fortunately
when data is graphed, patterns can emerge.
In a shampoo formulation, the salt curve
is an obvious example. In many systems,
adding salt ;rst makes the viscosity go up,
but too much causes a precipitous decline
(Figure 1). Aiming a bit to the le; of the
peak point gives maximum viscosity, with a
little wiggle room if too much salt is added.
A simple correlation of viscosity to salt
content is a powerful formulation tool.
;e situation gets more challenging when
new materials are involved or the ingredients
can interact. If order
of addition is critical,
suddenly the number
of possibilities jumps
exponentially. ;e poor
chemist might wish for a
robot to generate every
possibility and present
the optimum blend. ;at
robot has arrived.
chemistry” is the fancy
term for ;inging multiple
experiments against a wall
and seeing what sticks. ;e
history of combinatorial
chemistry goes back to the
work of Bruce Merri;eld
of Rockefeller University
in the 1960s, which led
to his Nobel Prize in
Figure 1: Experiments create data, and when data is graphed, patterns can emerge. In a shampoo formulation, the salt curve is an example.
Chemistry in 1984. Early work centered on
peptide synthesis. By the 1990s, it became
a primary method for drug discovery.
;e principles are now being applied to
the complex interactions of some types of
personal care formulations.
;e perfume industry has used robots for
years, both on a lab scale and in production.
;e nature of fragrances involves mixing
hundreds of chemicals for each formula, and
using a di;erent formula for every customer
and application, so a lot of repetitive work is
involved. However, the robotics never became
central to the fragrance creation process, just
an extra pair of hands.
Major fragrance houses do use
combinatorial chemistry to create new aroma
molecules, making thousands of experiments
to locate a handful of useful products.
;is is similar to the use of combinatorial
chemistry in drug discovery. Lack of a
precise structure-odor relationship limits
the possibilities for the rational design of
fragrance molecules, so trial and error can be
the only feasible approach.
Applying the Tools
With robots making samples and the
concepts of combinatorial chemistry and
high-throughput screening, all that remains
to be done is to apply the tools to complex
cosmetic formulation issues. ;e Institute
for Formulation Science is at the heart of just
such an activity.
Robots exist that can make 2,000 formulas
a day—a Beckman-Coulter Liquid Handler,
for example, is capable of 350 an hour. ;ey
aren’t soup to nuts ;nished formulas but
simple mixes that show critical interactions.
;e robot can make thousands of iterations,
but surely a human must evaluate them.
68 Chemical Reaction
GCI April 2010