The method of retro-synthetic analysis practiced by
Professor E.J. Corey changed synthetic organic
chemistry. It produced a quantum leap in
the productivity of synthetic chemistry.
Nevertheless, creativity in organic synthesis will benefit from a
variety of inspiring orthogonal perspectives. The retro-synthetic route is
product-focused. That is, the analysis starts from the connectivity of the
target molecule. Is this the only
rational approach? How could it be any other way, one might ask, since the
target’s connectivity, that is its structure, is the objective? What I am
saying is that the Corey retrosynthetic analysis, at least in the form we have
learned it, looks for inspiration and direction essentially single-mindedly at
this connectivity and positional relatedness of the functional groups and works
unremittingly backward to unique commercially available substances.
Before the modern era, the standard approach was to search
in the product structure for the most complex substructure of a commercially
available starting material and then, working from both sides, attempt to
interconnect starting material and product by some combination of skeletal bond
forming reactions and functional group manipulations.
Molecules related to steroids are still analyzed in this way
both for historical reasons and because our minds intuitively grasp the logic
of the method. Who would look at an
unusual structural feature or functional group arrangement in a steroid and
then think that an entirely different preparation of the tetracyclic steroidal
skeleton would be the preferred approach?
There are other subliminal assumptions shared by both the
Corey-Wipke method and the starting material-focused approach. One is the assumption that efficiency and
elegance in synthesis relate to minimizing the number of chemical reaction
steps. Another is the assumption that the physical properties of intermediate
substances are irrelevant to the simplicity and elegance of the synthesis.
We easily grasp that molecules, which are effectively basic
or effectively acidic in an aqueous solution are especially easy to isolate and
purify. They have the ability to form salts and to be partitioned between
organic and aqueous media when the aqueous pH has been strategically adjusted. Furthermore, they can be formed into a large
number of solid ionic salts which can be recrystallized for purification and
can then be neutralized to recover the origin substance.
Not quite so convenient or obvious are functionalities that
can reversibly form derivatives that are acids and bases, which can, in turn, be
purified by the above-named means and which can then be reconverted back into
the original functional groups. Such
manipulations of course suffer from the increased number of steps. Let us take
the example of an alcohol, which is converted into a potassium O-sulfate salt.
It can then be recrystallized and hydrolyzed back to the alcohol. Three
additional steps have been introduced and one is no further along in terms of
synthetic transformations.
One thing this might be telling us is that there is
inadequate knowledge about the transformations of potassium alcohol
sulfates. If it were possible to change
the synthetic path so that the potassium O-sulfate salt was not just a
substance that allowed easier purification but which also was an intermediate
important in the process of bond-building and/or functional group
transformation, then the sequence would go- alcohol to potassium alcohol sulfate
(with attendant purification) followed by potassium alcohol sulfate to another
more advanced intermediate in the synthetic route.
Non-stoichiometric Derivatives
If we follow the strand of reasoning that has led us to
reversibly form polar or ionic derivatives, we will find that there are
other reactions of neutral functional groups which lead to purification but
without the step of recrystallization of the derivative. These are the organic-inorganic complexes-
most frequently those with lithium chloride, bromide, and iodide, with magnesium
chloride and bromide, with calcium chloride, and most frequently calcium bromide. These complexes are useful because most
organic process intermediates are hydrophobic and are soluble in hydrocarbon,
ether, and ketone solvents and can complex as preferred bases in these solvents
with the above Lewis acids. These
complexes are insoluble in these solvents.
They may be stoichiometric but usually are not and the Lewis acid
selects the neutral function group with which to preferentially complexes based
on the steric environment in the substructure region around the functional
group which is acting as the Lewis base.
The lack of stoichiometric relationship in these complexes is
practically not important because:
- it is
easy to separate hydrophobic organic compounds from inorganic salts
- the
inorganic Lewis acids are relatively inexpensive and readily available
Among common organic functional groups alcohols and phenols
are the most amenable to forming complexes with Lewis acids. Alcohols form complexes with calcium
chloride, calcium bromide, and lithium salts, and phenols form complexes particularly
with calcium bromide. KiloMentor has
already written blogs about such complexes: Inorganic Non-Stoichiometric
Metal Salt Complexes with Organic Molecules as a particularly Useful Method for
Purifying Neutral Substances and More Information about using Inorganic Salt
complexes to simplify purifying Alcohols at Scale.
Alcohol Reversible Derivatization for Acid-Base Extraction
Here I would like to focus on the use of actual covalent
derivatives of alcohols, which convert them into acid substances. In particular
I would like to talk about alcohol phthalates.
Alcohol Phthalates
Phthalic anhydride is an inexpensive alcohol derivatizing
agent, which produces a product that is a carboxylic acid-containing, hydrolysable ester functionality. The
presence of the carboxylic acid allows salt formation and that provides some
water or water-alcohol solubility, which can be used for extractive separations
from by-product contaminants.
Sodium salts of straight-chain alcohols larger than octyl
tend to form emulsions so the process for their preparation needs to be chosen
with this possibility in mind. The
alcohol is heated with twice the molar amount of phthalic anhydride at 105-110
C for a period between thirty minutes to two hours. The higher the molecular
weight of the alcohol the longer the reaction typically takes. The m.p. of phthalic anhydride is 131 C.
The reaction mass is treated with ether, 50 ml/gm of starting
alcohol. The excess phthalic anhydride which is insoluble in ether is
filtered off. The ether solvent is then
removed and the residue is treated with about 120 ml of water and the bi or
tri-phasic mixture is warmed at 60-65 C for 45 minutes to hydrolyze any
residual phthalic anhydride. Then the water is also stripped off or if the
residue is solid it can be filtered and the residue dried. The dried residue is dissolved in 20 cc of
chloroform in which phthalic acid is insoluble and the solution filtered. Again the solution is evaporated under vacuum
and the esters recrystallized three times from petroleum ether. It is often found advantageous to use dry-ice
for cooling the solutions of the esters and using an inverted siphon
filtration. This procedure is for a laboratory sample as one can easily see.
When the shape of the molecule makes emulsion formation less
likely, a simpler procedure can be used.
One gram of the alcohol is heated as before with twice the
stoichiometric amount of phthalic anhydride for 30 minutes to two hours at
105-110 C as before. If the alcohol is
not liquid some toluene is used to create liquidity and prevent charring in the
flask. For purification the mixture is
shaken with 50 ml of toluene diluted if necessary with hexanes filtered from
excess anhydride and the filtrate neutralized by dilute sodium carbonate
leaving the mixture slightly acidic. The
aqueous layer was extracted three times with 100 ml portions of toluene too
remove the unreacted alcohol and possible diesters (the extraction solvent must
take the starting material solubility in mind).
Some alcohol may be needed in the aqueous phase to hold some high
molecular weight or multifunctional alcohol mono-phthalates in the polar
medium. The esters are then precipitated
from the aqueous solution by dilute hydrochloric acid and recrystallized from
an appropriate solvent mixture. Often a
mixture of predominantly petroleum ether mixed with around 10% of toluene will
work well (hydrocarbons).
James F. Goggans, Jr. and J.E. Copenhaver, J. Am. Chem..
Soc. 61, 2909 (1939).
Procedures for making the monophthalate esters of secondary
alcohols have been described by Pickard and Kenyon, J. Chem. Soc., 91,
2058-2061 (1907); 99, 58 (1911): 103, 1937 (1913).
For preparing monophthalates of tertiary alcohols a good
procedure involves the treatment with ethereal triphenylmethyl sodium at room
temperature. The alcohol is dissolved in
a convenient amount (30 parts) of anhydrous ether and to it was added rapidly
with stirring an ethereal solution of triphenylmethylsodium until a persistent red coloration was present
in the solution. This shows a residual slight excess of the colored base.
Phthalic anhydride was added all at once in an equimolar amount with the
alcohol and the stirring was continued for 1-2 hours. Water (200 ml) was added
for each 5 gm of starting alcohol, the layers separated, and the water layer
was poured over cracked ice and hydrochloric acid. The acidic cold medium neutralizes and
precipitates the free acid. The
precipitate formed was filtered cold, air-dried, and recrystallized or
triturated. In the old literature, the
derivative is repeatedly recrystallized.
Kenneth G. Rutherford, Joseph M. Prokipcak, and David P.C.
Fung, J. Am. Chem. Soc., 28, 582, (1963).
It is not clear at what point the yield was lost. Smaller alcohols gave lower yields. The
average yield for several derivatizations of t-butanol was 65%. It would not be surprising if the
preponderance of the yield loss was in the recrystallization step. If the sequence is used to purify the
material from non-alcohol non-reacting substances, then the crude can be
immediatedly hydrolyzed and the alcohol taken back into an organic solvent and
recovered from there or used as such in the next reaction. The phase shifts will have done their job of
removing impurities and the derivative will have served as a stopping point in
the process. In a process, we must bear in mind, an appropriate intermediate purity needs only to be the practical
purity which is required to give a final product meeting the
specifications. The ruggedness of the
process is not so much determined by the absolute purity of the recovered
intermediates but by the number and the discrimination of the phase shifts that
the process provides.
Another way to prepare monophthalate esters is to use dialkylaminopyridine
catalysis of the acylation process. A
method is provided in Synthesis Communications, 1972, 619.
Thus t-butyl hydrogen phthalate is obtained under very mild
conditions using phthalic anhydride and 1.2 equivalents of
4-pyrollidinopyridine in dichloromethane at room temperature. Using
triethylamine requires long heating. In the actual procedure t-butanol (1.0 gm
13.5 mmoles) was added to a solution of phthalic anhydride (1.48 g., 10 mmol)
and 4-pyrollidinopyridine( 1.8 g, 12.5 mmol) in dichloromethane (10 ml). The
NMR spectrum indicated 95% conversion of the anhydride after 1 day. After 2 days, ether was added and the base
extracted with 2N hydrochloric acid. The
ether layer was dried (MgSO4) and evaporated in vacuo. To remove phthalic
acid, the oily residue was dissolved in carbon tetrachloride (10 ml) and the
solution filtered. The filtrate was evaporated and the remaining oil dried at
50 C/0.05 torr.; yield 2.15 gm (97%) mp. 70-75 C.
This yield is based on phthalic anhydride, which was the
limiting starting material. We would be
using the reaction to perform a phase separation to remove alcohols from
non-alcohol constituents and the phthalic anhydride would be used in an
excess. The reaction mixture would
contain neutral, non-alcohol, base soluble alcohol monophthalate and the acid
soluble dialkylaminopyridine catalyst.
In the reference it says, “In preparations on a larger scale
(>2 g) in addition to catalytic amounts of 4-dialkylaminopyridine one
equivalent of triethylamine was used to bind the acid formed in the reaction.
In this case longer reaction times or heating was necessary. As a reaction medium there may be used a
non-protic solvent, excess triethylamine or excess anhydride.
An example of a sec-alcohol is given in which the phthalic
anhydride is used in excess
(-) Menthyl Hydrogen Phthalate
A solution of (-) menthol (3.2 g, 20 mmol) phthalic
anhydride (6.0 g. 34 mmol), triethylamine (4.1 ml, 30 mmol), and 4-dimethylaminopyridine
(0.5 g, 4.1 mmol) in dioxane (20 ml) was kept at room temperature for 6 hours.
Work-up was done as for making the t-butyl phthalate by adding an excess of
ether and extracting with 2N hydrochloric acid to remove the catalyst and the
triethylamine hydrochloride and the dioxane solvent. The ether solution was
dried with magnesium sulfate and after filtering evaporated to an oil which was
redissolved in carbon tetrachloride on any suitable solvent which dissolves the
product but leaves the phthalic acid insoluble. Carbon tetrachloride is
appropriate. Filtering the insoluble phthalic acid leaves a solution of the
monophthalate in carbon tetrachloride. The yield was 5.6 g (92%) m.p. 90-93 C;
after one recrystallization from methanol/water m.p. 111 C. Phthalic acid and the menthol monophthalate
are likely to be even more simply separable. By extractive crystallization if
the aqueous solution in contact with an organic solvent is partially
neutralized with base. The sodium salt
of phthalic acid will selectively partition into the water and the menthol
monophthalate will be retained in the organic phase. Thereafter the derivative can be immediately
hydrolyzed with base, acidified and the menthol extracted into an organic
phase.
Another application of the same technology is the removal of
a small amount of an alcohol: mono, secondary, or tertiary from a main product,
which is not an alcohol and is unreactive with phthalic anhydride.
An example might be the removal of residual alcohol from the
pinocol rearrangement products from a diol.
The product ketone will be unreactive to the phthalic anhydride,
catalyst and base. When the impurity has
been derivatized at either one or both of the alcohol functions; then one
extraction to remove the catalyst and base, followed by an alkaline extraction
to remove the monophthalate will leave the neutral ketone alone in the organic
solvent. The ketone product may at this
stage be of sufficient practical purity to be used in the subsequent reactions
without by-products, which reduce the final product quality analysis.
In another example the mixture of primary and secondary
alcohols formed in the Fischer Tropsch synthesis was first separated into
primary alcohols and secondary alcohols by preferential separation of the
primary alcohols with phthalic anhydride. [Graves, Ind. Eng. Chem. 23, 1381
(1931).
Besides using phthalic anhydride, other anhydrides can be
used in the same way. In an example from
the recent patent literature (WO2005084643A1) the drug substance escitalopram
(not an alcohol) is purified to remove an alcohol impurity by reaction with
succinic anhydride. The alcohol alone reacts and the hemi-succinic acid
derivative is extracted into an aqueous phase with an ammonia solution leaving
the purified escitalopram in the organic solvent.
Pros and Cons
You may say that a lot of labor and reactor time has been
spent removing a by-product when crystallization of the crude mixture would
have produced the same result. The
payback comes when one looks at the recovery yield from the crystallization of
a product that has been purified by derivative formation and phase shifting.
To make this point let me ask you a general question. If one has a product mixture that is 90%
desired product and 10% by-product or co-product and one recrystallizes; what
yield can you realistically expect as a median result? Remember that when 8/9ths of the product has
crystallized the remaining mixture will be 50:50 product and by-product. Even with the by-product being entirely retained
in solution, it is most likely that the last 10% of what you want either will
not crystallize or will crystallize so slowly that it will not be economical to
wait for it. Thus the median yield will
be 80%. Now-suppose that by phase
shifting of a derivative the purity before crystallization is 99% product 1%
coproduct.
Now when one recrystallizes the 50:50 mother liquor
situation only occurs after 98% of the desired product has crystallized. The
potential yield will be 98% in a short time. Much less solvent will be used and
simple slurrying rather than crystallization make be satisfactory to get
practical purity. Now suppose one lost 10% of the material during the
operations of the phase-shifting derivatization and this is a very generous
assumption of loss. Still one would have 88% product compared to 80% saving
of 8% of the material. An advanced intermediate moreover is costly because a
lot of materials and labor has been invested in it. Spending some phthalic anhydride, DMAP,
triethylamine and solvent, and inorganic acids and bases can be a very good
investment.