Synthesis of tertiary amines using a polystyrene (REM) resin

Article abstract of DOI:10.1021/ja963829f

A range of tertiary amines was constructed using a 'traceless' linker on a polystyrene resin (REM resin), starting from seconday amines, primary amines, and resin-bound 'ammonia'. The methodology is characterized by three essential steps conducted under ambient conditions: (1) coupling of the starting amine (Michael addition) to the resin, (2) activation (quaternization), and (3) cleavage of the product amine (Hofmann elimination). The linker is compatible with both acid and base sensitive protecting group strategies. The nature of the chemistry ensures that the tertiary amine products are obtained in consistently high purity (95% or greater). After cleavage of the product, REM resin is regenerated and can be reused for repeat syntheses. The yield and purity of repeat batches is maintained over 5 cycles, allowing the automated synthesis of >0.5 g quantities of pure tertiary amine.

Full text of DOI:10.1021/ja963829f

3288
J. Am. Chem. Soc. 1997, 119, 3288-3295
Synthesis of Tertiary Amines Using a Polystyrene (REM) Resin
Angus R. Brown, David C. Rees, Zoran Rankovic, and J. Richard Morphy*
Contribution from the Medicinal Chemistry Department, Organon Laboratories Ltd.,
Newhouse, ML1 5SH, Scotland, U.K.
ReceiVed NoVember 4, 1996X
Abstract: A range of tertiary amines was constructed using a “traceless” linker on a polystyrene resin (REM resin),
starting from secondary amines, primary amines, and resin-bound “ammonia”. The methodology is characterized
by three essential steps conducted under ambient conditions: (1) coupling of the starting amine (Michael addition)
to the resin, (2) activation (quaternization), and (3) cleavage of the product amine (Hofmann elimination). The
linker is compatible with both acid and base sensitive protecting group strategies. The nature of the chemistry
ensures that the tertiary amine products are obtained in consistently high purity (95% or greater). After cleavage of
the product, REM resin is regenerated and can be reused for repeat syntheses. The yield and purity of repeat batches
is maintained over 5 cycles, allowing the automated synthesis of >0.5 g quantities of pure tertiary amine.
In recent years, solid phase chemistry has evolved from being
Tertiary amines are an extremely important class of compound
from the drug discovery perspective. Indeed no less than a
quarter of registered drugs contain tertiary amines.10 They are
particularly common in drugs which are active within the central
nervous system. Our study was prompted by the conspicuous
lack of a methodology whereby tertiary amines could be
constructed on a resin from commercially available primary and
secondary amines using a traceless linker.11 Such a strategy
would preferably require the nitrogen of the amine to be part
of the bond which is broken during cleavage from the resin.
We concluded that the classical Hofmann elimination reaction
would serve our purposes well.12,13 The reaction is usually
considered to be a useful method for the synthesis of alkenes
and indeed had previously been used by Bradley and co-workers
to synthesize dehydroalanine derivatives on a resin.14 It became
apparent that we could achieve our objective of releasing a
tertiary amine into solution while the alkene component
remained resin-bound.
largely the preserve of peptide and oligonucleotide chemists to
become a mainstream activity carried out by an ever-increasing
number of organic chemists. The principal driving force for
this change of emphasis has been the search within the
pharmaceutical industry for libraries of small organic molecules
to generate new leads and accelerate the process of drug
discovery.1 Solid phase synthesis is particularly suitable for
combinatorial chemistry, and its advantages over traditional
solution phase chemistry have been discussed many times
before.2 One limitation of early solid phase linker strategies
was the presence of a resin-tethering substituent in the final
product. Many strategies involved an ester or amide linkage
to a polystyrene resin, which upon cleavage produced a
carboxylic acid, amide, ester, or alcohol group in the product.3
Such polar and metabolically labile substituents often compro-
mise bioavailability and are best avoided unless they contribute
significantly to binding to the target protein. Hence, much effort
is now being directed toward the development of “traceless
linkers” which leave no evidence of resin attachment in the final
product. For the attachment of aromatic groups to resins,
Ellman4 and Veber5 have described silicon-based traceless
linkers and Sucholeiki has described a thioether linker.6 Another
approach which has proved useful for the synthesis of hetero-
cycles, such as diketopiperazines,7 quinazolinediones,8 benzo-
diazepines, and hydantoins,9 involves an intramolecular cy-
clization step to facilitate cleavage from the resin.
The sequence of reactions required to produce the tertiary
amine is shown in Scheme 1.15 In the first step, the acrylate-
functionalized polystyrene resin 216 is prepared from com-
mercially available (hydroxymethyl)polystyrene resin 1 and
acryloyl chloride. A primary or secondary amine then under-
goes a Michael addition to 2 to give a secondary or tertiary
(10) This estimate was obtained by searching the World Drugs Index
(Derwent Information Ltd.) for compounds with a tradename which also
contain a tertiary amine group.
(11) The parallel synthesis of tertiary amines has been reported previously
using (a) an ester or amide tether to polystyrene resin (Green, J. J. Org.
Chem. 1995, 60, 4287) or (b) solution phase methods incorporating solid
phase scavengers (Kaldor, S. W.; Siegel, M. G.; Fritz, J. E.; Dressman, B.
A.; Hahn, P. J. Tetrahedron Lett. 1996, 37, 7193).
(12) â-Eliminations of carboxylic acids from a solid support have been
reported previously: (a) Katti, S. B.; Misra, P. K.; Haq, W.; Mathur, K. B.
J. Chem. Soc., Chem. Commun. 1992, 843. (b) Eritja, R.; Robles, J.;
Fernandez-Forner, D.; Albericio, F.; Giralt, E.; Pedroso, E. Tetrahedron
Lett. 1991, 32, 1511.
X Abstract published in AdVance ACS Abstracts, April 1, 1997.
(1) Terrett, N. K.; Gardner, M.; Gordon, D. W.; Kobylecki, R. J.; Steele,
J. Tetrahedron 1995, 51, 8135. Gordon, E. M.; Barrett, R. W.; Dower, W.
J.; Fodor, S. P. A.; Gallop, M. A. J. Med. Chem. 1994, 37, 1385.
(2) Fruchtel, J. S.; Jung, G. Angew. Chem., Int. Ed. Engl. 1996, 35, 17.
(3) Frechet, J. M. J. In Polymer-supported Reactions in Organic
Synthesis; Hodge, P., Sherrington, D. C., Eds.; John Wiley & Sons: New
York, 1980; Chapter 6, p 294.
(4) Plunkett, M. J.; Ellman, J. A. J. Org. Chem. 1995, 60, 6006.
(5) Chenera, B.; Finkelstein, J. A.; Veber, D. F. J. Am. Chem. Soc. 1995,
117, 11999.
(13) Hofmann, A. W. Annalen 1851, 78, 253.
(14) Blettner, C.; Bradley, M. Tetrahedron Lett. 1994, 35, 467.
(15) Morphy, J. R.; Rankovic, Z.; Rees, D. C. Tetrahedron Lett. 1996,
37, 3209.
(6) Sucholeiki, I. Tetrahedron Lett. 1994, 35, 7307.
(7) Gordon, D. W.; Steele, J. Bioorg. Med. Chem. Lett. 1995, 5, 47.
(8) Smith, A. L.; Thomson, C. G.; Leeson, P. D. Bioorg. Med. Chem.
Lett. 1996, 6, 1483.
(9) Hobbs DeWitt, S.; Kiely, J. S.; Stankovic, C. J.; Schroeder, M. C.;
Reynolds Cody, D. M.; Pavia, M. R. Proc. Natl. Acad. Sci. U.S.A. 1993,
90, 6909.
(16) The resin has also been prepared from Merrifield resin and cesium
acrylate: Cody, D. R.; DeWitt, S. H. H.; Hodges, J. C.; Kiely, J. S.; Moos,
W. H.; Pavia, M. R.; Roth, B. D.; Schroeder, M. C.; Stankovic, C. J. PCT
Int. App. WO 9408711; Chem. Abstr. 1995, 122, 106536.
(17) Ramage, R.; Stewart, A. S. J. J. Chem. Soc., Perkin Trans. 1 1993,
1947.
S0002-7863(96)03829-2 CCC: $14.00 © 1997 American Chemical Society

Synthesis of Tertiary Amines
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3289
Scheme 1
Table 1. A Library of Tertiary Amines Produced on REM Resin
amine 3, respectively. In the former case, the secondary amine
can be converted to a tertiary amine by adding a reductive
alkylation step. Quaternization of the tertiary amine with an
alkyl halide 4 sets the scene for a facile Hofmann elimination
of the resulting quaternary ammonium salt 5, releasing the
tertiary amine 6 into solution. Since 2 is regenerated after
cleavage of the product and is initially functionalised via a
Michael reaction, we refer to the resin, in our laboratories, as
REM resin.
Initially, we demonstrated the methodology with the synthesis
of a small library of 10 tertiary amines (6a-j) derived from
commercially available secondary amines (Table 1). 1,2,3,4-
Tetrahydroisoquinoline, N-methylphenethylamine, ethyl iso-
nipecotate, ethyl nipecotate, and N-phenylpiperazine were found
to undergo addition to REM resin at room temperature in
dimethylformamide (DMF) (18 h). The reactions were carried
out in disposable polypropylene tubes using 10 equiv of the
amine. The resulting tertiary amines were quaternized using
allyl bromide or p-nitrobenzylbromide (5 equiv) in DMF at room
temperature (18 h). Heating to 50 °C produced no benefit in
terms of either the yield or purity of the final product. The
Hofmann elimination was promoted using 2 equiv of diisopro-
pylethylamine (Hunig’s base, DIEA). Either DMF or dichloro-
methane (DCM) served as suitable solvents. Isolation of the
product was easier when DCM was used, but in this case
increased amounts of aliphatic impurities were also present,
presumably dissolved from the polypropylene tubes and the
polystyrene resin. Fortunately, these were easily removed,
together with any DIEA‚HBr salt, by running the DCM solution
through a silica solid phase extraction (SPE) cartridge containing
powdered potassium carbonate. A combination of 1 equiv of
DIEA and 10 equiv of powdered potassium carbonate was also
effective for promoting the Hofmann elimination. Upon
evaporation of the cleavage solution, only the desired amine
6b and a small amount of plasticizer were observed by NMR.
The DIEA, having been deprotonated by the K2CO3, was
volatile. For product 6b, it was found that triethylamine (TEA)
could be used in place of DIEA without compromising either
yield or purity (Table 1).
a Percent yields for 3 steps based on the resin substitution level
determined by the Fmoc quantitation method.17 b Percent purity as
determined by gas chromatography. c Yield obtained using the 3-bro-
mopropionate ester of hydroxymethylpolystyrene resin in place of the
acrylate ester. d Yield obtained using 2 equiv of TEA in place of DIEA
for the elimination step. e Yield obtained using 1 equiv of DIEA and
10 equiv of K2CO3 for the elimination step.
The purity of the products after SPE was determined by gas
chromatography to be greater than 95%. The high purity is
presumably due to the nature of the chemistry involved. Each
step in the sequence has to be successful in order for the final
Hofmann elimination to occur. Pure product is even obtained

3290 J. Am. Chem. Soc., Vol. 119, No. 14, 1997
Brown et al.
Scheme 2
and the resulting acid was converted to the benzyl ester using
benzyl alcohol, 4-(dimethylamino)pyridine (DMAP), and 1-(3-
(dimethylamino)propyl)-3-ethylcarbodiimide (EDC) in DCM (18
h, 20 °C). After quaternization with iodomethane and cleavage
with DIEA, 8a was obtained in 53% yield and 97% purity (92%
enantiomeric purity).
It is possible to use REM resin to perform syntheses requiring
the monoalkylation of diamines, without the use of protecting
groups. For example, piperazine can be added to REM resin
to give the monoalkylated derivative 9, which can then be
acylated or alkylated cleanly at the second nitrogen (Scheme
3). Subsequent reaction with phenyl isocyanate or benzhydryl
bromide generates 10 or 11, respectively. Quaternization and
elimination provides products 12 and 13 in 36 and 44% overall
yields, respectively. The purity of 12 and 13 was >99% by
gas chromatography. The high purity of 13 demonstrates that
although quaternization at the “wrong” nitrogen in piperazines
may occur, it does not reduce the purity of the final product.
Again, only the correct product can cleave from the resin. The
synthesis of 13, the antiemetic drug, marzine, serves to
demonstrate the utility of REM methodology for preparing
pharmacologically interesting compounds.
We have found that the method works well when the
quaternization step can be conducted at ambient temperature
using reactive alkyl halides, e.g., allyl, benzyl, and methyl
halides. However, if less-reactive alkylating agents are used,
the quaternization step requires heat. For example, quaterniza-
tion of a resin-bound tertiary amine with n-butyl iodide does
not occur to any appreciable extent at 20 °C in DMF, but
requires heating at 60 °C for 24 h. Unfortunately, the elevated
temperature causes the Hofmann elimination to occur pre-
maturely, releasing the n-butyl derivative into solution where
it becomes susceptible to quaternization by a second equivalent
of n-butyl iodide. Nonetheless, we were able to isolate 2-n-
butyl-1,2,3,4-tetrahydroisoquinoline in 56% yield, after SPE
purification to remove a small amount of 2,2-bis(n-butyl)-
1,2,3,4-tetrahydroisoquinolinium iodide. We are currently
working on a new linker which should be more thermally stable
once quaternized and thereby prevent premature cleavage from
the resin occurring when less-reactive alkyl halides are used.
An attempt to quaternize resin-bound 1,2,3,4-tetrahydroiso-
quinoline using isopropyl iodide was unsuccessful at 60 °C,
presumably due to steric hindrance.
in some cases where the quaternary ammonium intermediate
contains more than one possible site for the Hofmann elimina-
tion to occur. For example, in the synthesis of the nipecotate
esters 6g and 6h, the desired products are still obtained in
reasonable yield (41 and 48%, respectively) and excellent purity
(97 and 96%, respectively), presumably because any compound
formed by elimination occurring within the piperidine ring
remains resin-bound.
The synthesis of 6a was repeated using the 3-bromopropionate
ester of (hydroxymethyl)polystyrene resin as the solid phase
starting material in place of the acrylate ester. The subsequent
quaternization and elimination steps were performed as before.
No significant difference in either the yield or purity of product
6a was observed compared to the REM resin synthesis (Table
1). This may suggest that the mechanism of addition of the
secondary amine to the resin is the same: the amine initially
promoting elimination of the 3-bromopropionate ester to the
acrylate, followed by a Michael addition.
Since the methodology is conceivably of value for construct-
ing libraries of N-terminally alkylated R-amino acid derivatives,
the effect of the reaction sequence on the chiral integrity of the
R-carbon was examined. Either L- or D-proline benzyl ester
(10 equiv) in DMF was added to REM resin (Scheme 2). After
quaternization using methyl iodide (10 equiv) in DMF, the
N-methylproline benzyl esters (8a and 8b) were eliminated from
the resin using DIEA (5 equiv) in DCM. We were concerned
that treatment of the quaternized R-amino ester 7 with base may
racemize the R-carbon, in addition to promoting the desired
Hofmann elimination. The chiral purity of the two products
was therefore determined using chiral HPLC (CHIRACEL OJ
column). The L-enantiomer 8a had an enantiomeric purity of
97.8% (61% overall yield). The D-enantiomer 8b had an
enantiomeric purity of 96.4% (69% overall yield).
The primary amines, benzylamine and phenethylamine,
underwent the Michael addition using the same conditions as
the secondary amines. The resulting resin-bound secondary
amines were then reductively alkylated with p-nitrobenzaldehyde
in the presence of sodium triacetoxyborohydride. Both were
quaternized using methyl iodide in DMF and eliminated using
DIEA in dichloromethane. Although the synthesis starting from
phenethylamine gave a good yield of product 6d (75%), the
benzylamine-derived product 14 was isolated in only 8% yield.
This probably reflects the difficulty of quaternizing the poorly
nucleophilic bisbenzylic intermediate. When 14 was synthesized
by reversing the reductive alkylation and quaternization steps,
using paraformaldehyde and p-nitrobenzyl bromide, respectively,
a much improved yield (86%) was obtained.
Tertiary amines can also be made from a resin-bound
equivalent of ammonia, as shown in Scheme 4. The ammonia
equivalent 15 is derived from the coupling of Fmoc-â-alanine
to hydroxymethyl resin, followed by Fmoc deprotection using
piperidine. Double reductive alkylation using paraformalde-
hyde, followed by quaternization using p-nitrobenzyl bromo-
acetate and Hofmann elimination, provided 16 in 38% yield
(for four steps) and 97% purity. Sequential reductive alkylation
We wanted to assess the acid stability of the linker by
attempting to synthesize 8a via a tert-butyl ester intermediate
(Scheme 2). After adding proline tert-butyl ester to REM resin,
the ester was cleaved using 50% TFA in DCM (6 h, 20 °C),

Synthesis of Tertiary Amines
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3291
Scheme 3
Scheme 4
using p-nitrobenzaldehyde and paraformaldehyde, followed by
quaternization using benzyl bromide and Hofmann elimination,
provided 14 in 31% yield (for five steps) and 98% purity.
Reaction Monitoring. The current shortage of reliable and
nondestructive analytical techniques for monitoring on-resin
reactions is often rate limiting in the development of new solid
phase chemistry. Two techniques which have been employed
successfully are FT-IR, in which the resin sample is prepared
as a KBr disk, and gel phase 13C NMR,18 in which the resin is
swollen in a suitable solvent, often CDCl3, DMF, or benzene.
FT-IR is destructive since the resin must be prepared as a
powder and the information content of a spectrum is often
limited. Gel phase NMR produces very broad lines and often
the use of 13C-enriched material is required to improve the
sensitivity of the technique. Fortunately, both FT-IR and 13C
gel phase NMR are reasonably effective for monitoring reactions
carried out on REM resin.
The resin contains a conjugated carbonyl group whose IR
absorption shifts to higher frequency as deconjugation occurs
during the Michael addition. Of course, the technique is not
quantitative, but we can reasonably assume the reaction to be
complete when no further shift occurs. In addition, the band
at 1403 cm-1, due to in-plane bending of the alkene CH,
disappears during the reaction. The carbonyl absorption
frequencies observed during the synthesis of 2-allyl-1,2,3,4-
tetrahydroisoquinoline 6a are shown in Table 2 (see also pp
24-27 of the Supporting Information). Not surprisingly, there
is no significant shift in the position of the carbonyl band during
quaternization; therefore, this reaction cannot be followed by
IR. It is apparent that the carbonyl band returns to its original
position after the Hofmann elimination as REM resin is
regenerated. The presence of other functional groups can
sometimes help reaction monitoring. For example, the reductive
alkylations and quaternizations using nitro-containing com-
pounds were followed by observing the intensity of the NO2
absorption at 1523 cm-1. The poor yield of 14 obtained initially
(8%) was ascribed to the slowness of the quaternization of the
bisbenzylic intermediate, on the basis of the intensity of the
NO2 peak being little changed after the attempted cleavage. This
is supported by the fact that 14 was obtained in much better
yield simply by reversing the reductive alkylation and quater-
nization steps. Since the quaternary ammonium compound is
(18) Look, G. C.; Holmes, C. P.; Chinn, J. P.; Gallop, M. A. J. Org.
Chem. 1994, 59, 7588.

3292 J. Am. Chem. Soc., Vol. 119, No. 14, 1997
Table 2. Reaction Monitoring Using FT-IR and 13C NMR
Brown et al.
common to both routes, the final cleavage step cannot be causing
the problem.
To facilitate the use of gel phase 13C NMR to monitor the
reactions in a more quantitative way, 13C-enriched REM resin
was prepared. Acrylic acid, in which each of the three carbons
was 99% 13C, was diluted to 20% enrichment using natural
abundance acrylic acid and then coupled to (hydroxymethyl)-
polystyrene resin using the carbodiimide method (DIC, DMAP,
DCM, 20 °C, 18 h). The positions of the three labeled acrylate
resonances, after the resin was swollen in CDCl3, are given in
Table 2. The carbonyl carbon in REM resin is clearly visible
at 166 ppm, but the vinyl resonances (128 and 131 ppm) overlap
with the aromatic resonances due to the polystyrene resin (Figure
1). After treatment with a 20% solution of 1,2,3,4-tetrahy-
droisoquinoline in DMF for 18 h at 20 °C, the carbonyl
resonance is shifted downfield due to deconjugation. Two new
resonances at 53 and 32 ppm are due to the methylenes next to
the nitrogen and carbonyl, respectively. The resonances in the
labeled resin show multiplicity, theoretically 2 doublets and 1
triplet, due to coupling between neighboring 13C nuclei. The
absence of a carbonyl resonance at 166 ppm was suggestive of
the Michael reaction having proceeded to completion. To
Figure 1. Gel phase 13C spectra of resin-bound intermediates in the
synthesis of 6a.
identical to the starting resin. We were therefore curious to
discover whether the regenerated resin was reusable and in
particular whether yield and purity would be maintained for a
second batch of 6a.
A single batch of unlabeled REM resin (300 mg) in a single
polypropylene tube was used for the repeat synthesis of four
batches of 6a, using the reaction conditions described earlier
(18 h per reaction at 20 °C). The yields for the four batches
after purification by SPE were as follows: (1) 33 mg (86%);
(2) 36.5 mg (95%); (3) 31.5 mg (82%); (4) 33.8 mg (88%).
Thus, the yield showed no sign of deteriorating over four cycles.
The purity of the products was also maintained, since the 400
demonstrate the value of using 20% 13C enriched resin, the 13
C
spectrum of natural abundance resin is also shown in Figure 1.
After treatment with allyl bromide in DMF (5 equiv, 18 h,
20 °C), the carbonyl resonance was shifted slightly upfield (169
ppm). Somewhat unexpectedly, the the position of the CH2
adjacent to nitrogen was unchanged, but the CH2 next to the
carbonyl was shifted upfield (28 ppm). In solution, the position
of a carbon resonance adjacent to nitrogen is shifted downfield,
by 3-8 ppm, upon quaternization. In this case, where the
compound is resin-bound, the shift does not occur, conceivably
due to shielding effects from the many aromatic groups found
in polystyrene resin. The lack of resonance at 32 ppm is
suggestive of the reaction having gone to completion.
1
MHz H NMR spectra of 6a for each of the four cycles were
virtually identical (see pp 19-22 of the Supporting Information).
Although the above sequence of reactions demonstrated the
reusability of REM resin, a total of 216 h was required to
produce 135 mg of product from 300 mg of REM resin (2.1
mg of product per 1 g of resin per hour). We were eager to
discover whether each cycle could be condensed into a single
working day.
The four cycles were thus repeated over 4 days, allowing
2.5 h for each of the three reactions within a cycle. The yields
and purities for the four batches of 6a are given in Table 3.
Again, the yield and purity shows no sign of deteriorating over
4 cycles. The lower yields, compared to the first recycling
experiment, indicate that one or more of the reactions are not
going to completion. Nonetheless, this protocol is more efficient
The Hofmann elimination was promoted using 2 equiv of
DIEA in DMF (18 h, 20 °C). The carbonyl resonance shifted
back to its original position at 166 ppm, and the two allyl
resonances were visible despite the overlying aromatic reso-
nances of the resin. The resonance at 28 ppm had completely
disappeared, suggesting the elimination was complete.
Resin Recycling. Both FT-IR and gel phase 13C NMR
showed that the resin regenerated after the cleavage of 6a to be

Synthesis of Tertiary Amines
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3293
Table 4. Automated REM Resin Recycling
Table 3. Manual REM Resin Recycling
a Percent yields for 3 steps based on the resin substitution level
determined by the Fmoc quantitation method.17 b Percent purity as
determined by gas chromatography.
a Percent yields for 3 steps based on the resin substitution level
determined by the Fmoc quantitation method.17 b Percent purity as
determined by reverse phase HPLC (see also the Supporting Informa-
tion).
in terms of the amount of compound generated in a given time
(246 mg in 30 h ) 8.2 mg of product per 1 g of resin per hour).
These results also suggest that the high purity of the final
products is not dependent upon individual reactions within the
sequence going to completion, since only the correct product
cleaves from the resin.
through five cycles of addition, quaternization, and elimination.
The crude products from the five cycles were pooled, partitioned
between ethyl acetate and 5% aqueous Na
2CO3, separated, and
evaporated to give 6a, 17, and 18. Purity and yields are given
in Table 4. Although a considerable amount of pure 6a was
obtained (340 mg), the percentage yield is somewhat lower than
expected based on the manual synthesis result. It would appear
that heating the reactions to 50 °C may have had a detrimental
effect on the yield of 6a, rather than the desired effect of driving
the reactions further toward completion. On the other hand, it
is conceivable that the lower yield of 6a was due to a technical
fault with the robotic system since the percentage yields of 17
and 18 are similar to that obtained for 6a in the manual
synthesis. Further optimization of the robotic synthesis is clearly
required to reproduce the yields obtained manually. Nonethe-
less, hundreds of milligrams of pure tertiary amine were
produced by a method which required little human intervention.
In order to attempt to discover which of the three steps was
rate and yield limiting, we performed an experiment in which
REM resin was reacted with two amines in sequence: 1,2,3,4-
tetrahydroisoquinoline (10 equiv) in DMF for 2.5 h at 20 °C,
followed after washing the resin by 4-benzylpiperidine (30
equiv) in DMF for 18 h at 20 °C. The Michael adduct was
then treated with allyl bromide (5 equiv) in DMF for 2.5 h at
20 °C, followed after washing the resin by methyl iodide (25
equiv) in DMF for 18 h at 20 °C. Cleavage was then performed
using DIEA (2 equiv) in DCM for 2.5 h at 20 °C, followed
after washing the resin by more DIEA (10 equiv) in DCM, for
a further 42 h at 20 °C. This experiment could potentially
produce up to four products, 2-allyl-1,2,3,4-tetrahydroisoquino-
line, 2-methyl-1,2,3,4-tetrahydroisoquinoline, 1-allyl-4-benzyl-
piperidine and 1-methyl-4-benzylpiperidine, if both the Michael
addition and quaternization steps are incomplete after 2.5 h.
The crude product from the 2.5 h cleavage step was analyzed
by GC-MS (see p 32 of the Supporting Information). Two
products, 2-allyl-1,2,3,4-tetrahydroisoquinoline and 2-methyl-
1,2,3,4-tetrahydroisoquinoline, are apparent in a ratio of 4.8:1,
respectively. The lack of any 4-benzylpiperidine-containing
products suggests that the Michael addition is progressing to
completion. However, the presence of 2-methyl-1,2,3,4-tetra-
hydroisoquinoline indicates that the quaternization using allyl
bromide is only about 80% complete after 2.5 h. The cleavage
step appeared to be complete after 2.5 h, since no further product
was evident after prolonged treatment of the resin with a large
excess of DIEA. Thus, the yield-limiting step in the manual-
recycling synthesis appears to be the quaternization.
Conclusions. We have designed the first traceless linker
which allows the assembly of tertiary amines on a polystyrene
resin (REM resin) from either ammonia or primary or secondary
amines. The Hofmann elimination is an effective cleavage
mechanism for releasing tertiary amines into solution under
mildly basic conditions. The purity of the products obtained
using REM resin is consistently high, presumably due to the
nature of the chemistry involved. The mild conditions involved
in this new methodology are well suited to multiple parallel
synthesis and the production of combinatorial libraries by
manual or automated methods. Starting from a secondary
amine, two sites of diversity can be introduced during a
synthesis. From a primary amine, three sites of diversity are
introduced. We have found that any racemization occurring
during the synthesis of N-alkylated R-amino ester derivatives
on REM resin is likely to be minimal. The REM linker is stable
to both mildly acidic and basic conditions, making possible
Fmoc protection of amines and tert-butyl ester protection of
carboxylic acids. Boc protection of amines is also likely to be
compatible with the resin, since the cleavage conditions are
likely to be similar to those used for deprotecting tert-butyl
esters. Diamines can be immobilized on REM resin and cleanly
monoalkylated or monoacylated.
The full potential of this recycling strategy would be realized
if the above protocol could be automated using a continuous
flow system capable of running 24 h a day until the desired
amount of product has been synthesized (see Scheme 1).
Toward this goal, we used a MultiSyntech SyRo II robot to
synthesize three tertiary amines, 6a, 17, and 18 (Table 4). The
reaction block comprises 60 individual 5 mL glass vessels. To
each vessel was added 100 mg of REM resin. The resin samples
were then treated, in three groups of 20, with one of the starting
amines: 2-allyl-1,2,3,4-tetrahydroisoquinoline, 4-hydroxy-4-
phenylpiperidine, or 4-benzylpiperidine (5 equiv, DMF, 50 °C,
2 h). All samples were then quaternized with allyl bromide
(10 equiv, DMF, 50 °C, 2 h) and eliminated using DIEA (5
equiv, 1,4-dioxane, 50 °C, 2 h). The dried resin was reused
A combination of FT-IR and 13C NMR enabled efficient
tracking of the reactions involved in REM resin methodology.
Using 20% 13C-enriched resin enabled useful spectra to be
obtained after just 10 min of acquisition time. In the case of
the synthesis of 6a, the three steps were shown to proceed to
completion under ambient conditions, explaining the high yield
and purity of the product.

3294 J. Am. Chem. Soc., Vol. 119, No. 14, 1997
Brown et al.
SPE Purification. The hydrobromide salt of DIEA and a trace
amount of plasticizer was removed using an ISOLUTE-XL solid phase
extraction column, containing 500 mg of silica and 50 mg of dried,
REM methodology would appear to offer good scope for an
effective automated resin recycling strategy, capable of produc-
ing multigram amounts of pure tertiary amine for use in
biological assays where a larger amount of material is required,
e.g., in in ViVo experiments. The purity of the final product
was found to be similar whether or not the individual reactions
were driven to completion. Recycling could also be useful for
producing large amounts of feedstock compounds for daughter
library synthesis. We have also been able to reuse the 20%
13C-enriched resin for reaction monitoring. This is very
desirable in view of the high cost of synthesis. The recycling
of resins for solid phase synthesis is not a new concept, having
previously been reported by Wang and co-workers19 for the
synthesis of peptide hydrazides on (Hydroxymethyl)polystyrene
resin and by McManus and co-workers20 for the synthesis of
carboxylic acids and esters using polymer-bound oxazolines.
Nonetheless, surprisingly little practical reuse of resins is evident
from the literature. Since resin handling is perhaps the most
technically difficult and time-consuming operation within a
automated system, the reuse of a single batch of resin within
an individual reaction vessel becomes an attractive proposition,
even when the resin is not a particularly expensive starting
material.
powdered K
2CO3. The crude material was loaded in DCM (0.7 mL),
eluted with heptane (3 mL, plasticizer elutes) and then ethyl acetate (3
mL, elutes amine). Evaporation of the EtOAc provided the product as
a colorless gum.
Procedure for On-Resin Cleavage of tert-Butyl Esters. The resin
obtained from the coupling of L-proline tert-butyl ester to REM resin
(1.113 g, 0.75 mmol) was suspended in 50% trifluoroacetic acid in
DCM (10 mL) in a polypropylene tube. After agitating on the rotator
for 6 h at 20 °C, the resin was drained and washed using the VacMaster
station with DCM (3 × 10 mL), 25% DIEA in DCM (3 × 10 mL),
DCM (3 × 5 mL), and methanol (2 × 10 mL) and dried in vacuo.
Procedure for Urea Formation from Resin-Bound Piperazine.
Resin 9 (0.50 g, 0.38 mmol) was suspended in a solution of phenyl
isocyanate (410µl; 3.8 mmol) in DMF (7 mL) in a polypropylene tube.
After agitating on the rotator for 20 h at 20 °C, the resin was drained
and washed using the VacMaster station with DMF (3 × 5 mL), DCM
(3 × 5 mL) and methanol (2 × 5 mL) and dried in vacuo.
Procedure for Alkylation of Resin-Bound Piperazine. Resin 9
(0.661 g, 0.486 mmol) was suspended in a solution of DIEA (846 µL,
4.86 mmol) and benzhydryl bromide (1.20 g, 4.86 mmol) in N-methyl-
2-pyrrolidinone (7 mL) in a 15 mL glass test tube. After 18 h of stirring
at 80 °C, the resin was filtered using a sintered glass funnel and washed
with DMF (3 × 5 mL), DCM (3 × 5 mL), and methanol (2 × 5 mL)
and dried in vacuo.
Experimental Section
13C-Enriched REM Resin. To a solution of acrylic acid (6.9 µL
99% [1,2,3-13C3]acrylic acid [Cambridge Isotope Laboratories] and 27.4
µL normal acrylic acid, total 0.5 mmol) and 4-(dimethylamino)pyridine
(15 mg, 0.125 mmol) in DCM (2.5 mL) was added diisopropylcarbo-
diimide (78.5 µL, 0.5 mmol). After 30 min of agitation of the rotator,
(Hydroxymethyl)polystyrene resin (250 mg, 0.15 mmol) was added,
and the suspension was agitated on the rotator for 18 h at 20 °C. The
resin was drained, washed (3 × 2 mL DCM, 3 × 2 mL DMF, 3 × 2
mL DCM, 2 × 2 mL CH3OH), and dried in vacuo.
REM Resin Synthesis. (Hydroxymethyl)polystyrene resin (1 g, 0.58
mmol) [Bachem California, 0.58 mmol/g] was added to a 10 mL Biorad
polypropylene tube. Anhydrous dichloromethane (7 mL) and diiso-
propylethylamine (866 µL, 5 mmol) were added, followed by acryloyl
chloride (404 µL, 5 mmol). The vessel was then placed on a Stuart
Scientific SB1 tube rotator and agitated for 4 h at 20 °C. The tube
was placed on a VacMaster sample processing station, and the resin
was washed with DCM (3 × 3 mL) and methanol (2 × 3 mL) and
then dried in vacuo.
General Procedure for Manual Recycling Synthesis of Tertiary
Amine 6a and for the Reaction-Monitoring Experiment. REM resin
(1.0 g, 0.74 mmol) was suspended in a solution of 1,2,3,4-tetrahy-
droisoquinoline (927µl; 7.4 mmol) in DMF (7 mL) in a 15 mL
polypropylene tube [Crawford Scientific] and agitated on a rotator for
2.5 h. The resin was drained and washed using the VacMaster station
with DMF (4 × 2.5 mL) and then briefly dried under reduced pressure.
The resin was resuspended in a solution of allyl bromide (326 µL, 3.7
mmol) in DMF (7 mL) and agitated on the rotator for 2.5 h. The resin
was drained and washed using the VacMaster station with DMF (4 ×
2.5 mL) and DCM (4 × 2.5 mL), then briefly dried under reduced
pressure. The resin was resuspended in a solution of DIEA (266 µL,
1.48 mmol) in DCM (7 mL) and agitated on the rotator for 2.5 h. The
resin was drained and washed using the VacMaster station with DCM
(3 × 3 mL). The filtrate was collected and evaporated. The residue
was partitioned between diethyl ether (3 mL) and aqueous sodium
carbonate (2 mL, 5% w/v). The organic layer was separated, and the
aqueous layer was reextracted with ether (2 × 2 mL). The combined
organic extracts were dried (potassium carbonate) and evaporated to
dryness. 2-Allyl-1,2,3,4-tetrahydroisoquinoline was obtained as a
colorless oil. [Note: an aqueous extraction was used in this case for
purification, since too much material was obtained for purification using
a 500 mg silica SPE cartridge.]
General Procedure for Automated Recycling Synthesis of Ter-
tiary Amines 6a, 17, and 18. REM resin (100 mg, 0.07 mmol) was
added to the 60 glass vessels (5 mL) of a MultiSynTech SyRo II
synthesis robot as an isopycnic slurry in DCM/DMF (1:1, 5 × 1 mL
transfers per 100 mg of resin). The vessels were then drained, and the
resin was washed with DMF (3 × 2 mL). Michael addition was carried
out at 50 °C for 2 h by treating the swollen REM resin (20 vessels per
amine, i.e., 2 g of resin per cycle) with 5 equiv of the secondary amines
in DMF (2 mL). The vessels were then drained, and the functionalized
resins were washed thoroughly with DMF (5 × 2 mL). Quaternization
was accomplished by treating each resin portion with allyl bromide
(10 equiv in DMF, 2 mL) for 2 h at 50 °C, before the vessels were
again drained and washed with DMF (5 × 2 mL). Cleavage of the
Synthesis of 2-Bromopropionate-Functionalized Polystyrene Resin.
(Hydroxymethyl)polystyrene resin (0.5 g, 0.29 mmol) [Bachem Cali-
fornia, 0.58 mmol/g] was suspended in a solution of 2-bromopropionoyl
chloride (252 µL, 2.5 mmol) and DIEA (95 µL, 0.55 mmol) in
anhydrous DCM (5 mL). After agitating on the rotator for 1.25 h at
20 °C, the resin was washed using the VacMaster station with DCM
(3 × 3 mL) and methanol (2 × 3 mL) and dried in vacuo.
General Procedure for the Michael Addition. REM resin (500
mg, 0.29 mmol) was swollen with a solution of the amine (2.9 mmol,
10 equiv) in DMF (5 mL) in a polypropylene tube. After agitating on
the rotator for 18 h at 20 °C, the resin was washed using the VacMaster
station with DMF (3 × 2 mL), DCM (3 × 2 mL) and methanol (2 ×
2 mL) and dried in vacuo.
General Procedure for the Reductive Alkylation. The resin (0.29
mmol) was swollen with a solution of the aldehyde (1.45 mmol) and
acetic acid (50µl) in DMF (5 mL). Sodium triacetoxyborohydride (1.45
mmol) was added as a solid, and the suspension was agitated on the
rotator for 18 h at 20 °C. The resin was washed using the VacMaster
station with DMF (3 × 2 mL), 5% DIEA in DMF (3 × 2 mL), DMF
(3 × 2 mL), DCM (3 × 2 mL), and methanol (2 × 2 mL) then dried
in vacuo.
General Procedure for the Quaternization. The resin (0.29 mmol)
was swollen with a solution of allyl bromide or p-nitrobenzyl bromide
(1.45 mmol, 5 equiv) in DMF (5 mL) and was agitated on the rotator
for 18 h at 20 °C. The resin was washed using the VacMaster station
with DMF (3 × 2 mL), DCM (3 × 2 mL), and methanol (2 × 2 mL)
and then dried in vacuo.
General Procedure for the Hofmann Elimination. A suspension
of the resin in DCM (5 mL) containing DIEA (0.58 mmol) was agitated
on the rotator for 5 h at 20 °C. The resin was drained and washed
using the VacMaster station with DCM (3 × 3 mL). The filtrate was
collected and evaporated.
(19) Chang, J. K.; Shimizu, M.; Wang, S. J. Org. Chem. 1976, 41, 3255.
(20) Colwell, A. R.; Duckwall, L. R.; Brooks, R.; McManus, S. P. J.
Org. Chem. 1981, 46, 3097.

Synthesis of Tertiary Amines
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3295
crude products from the resin was carried out by treating each individual
resin sample with diisopropylethylamine (5 equiv) in 1,4-dioxane (2
mL) at 50 °C for 2 h. The resin was then drained and washed with
methanol (5 mL), and the combined filtrates were evaporated using a
vacuum centrifuge (Savant SpeedVac). The individual products from
each of the three starting amines were then pooled. This process was
then repeated by returning the reaction vessels to the synthesis robot
and repeating this sequence of Michael addition, quaternization, and
Hofmann elimination a further 4 times (each resin portion was reacted
with the same amine in all five cycles). Each total pooled product 6a,
17, or 18 was then partitioned between ethyl acetate (50 mL) and
aqueous potassium carbonate (50 mL, 2 M). The organic phase was
separated, and the aqueous phase was washed a further 2 times with
ethyl acetate (50 mL). The organics were then combined, dried
(K2CO3), and evaporated. [Note: An aqueous extraction was used in
this case for purification, since too much material was obtained for
purification using a 500 mg silica SPE cartridge.]
Determination of Resin Loading. The loading of (Hydroxymeth-
yl)polystyrene resin was determined by the Fmoc quantitation method,17
i.e., coupling of Fmoc-glycine, followed by deprotection using piperi-
dine. The absorbance due to the liberated fulvene adduct was measured
at 300 nm.
Gas Chromatography Procedure. The compounds were dissolved
in methanol and loaded onto a Shimadzu 14A GC, fitted with a RTX-1
column (100% dimethylpolysiloxane; 30 m × 0.25 mm). Film
thickness: 0.25 µm. Carrier gas: helium. Pressure: 2 kg/cm2. Flame
ionization detector. External standard: n-hexadecane.
30 °C. Compound 8a eluted at 21.86 min; 8b eluted at 18.24 min.
See also p 29 of the Supporting Information.
FT-IR Procedure. The resin sample was prepared as a 13 mm
diameter 2% w/w KBr disk, and the spectrum was recorded on a Perkin
Elmer 16PC Fourier transform infrared spectrometer. Resolution: 2
cm-1. Apodization: weak. Number of scans: 5. Range: 4000-400
cm-1. Smooth: 6.
Gel Phase 13C NMR Procedure. The resin (100 mg) was added
to a 5 mm NMR tube and swollen with CDCl3. A PTFE vortex plug
was inserted into the tube, and the resin was compressed at the bottom
of the tube. The 13C spectrum was recorded on a Bruker DRX400
spectrometer using a 5 mm inverse probe at 100.6 MHz 13C frequency
(non-spinning). Data size: 32K. Sweep width: 30 303 Hz. Acquisi-
tion time: 0.52 s. Relaxation delay: 1.0 s; 30° pulse with Waltz
decoupling. Number of scans: 526 (10 min).
Acknowledgment. We thank the analytical departments at
Organon Newhouse, Scotland, and Organon Oss, Holland for
providing analytical data in support of this paper.
Supporting Information Available: Characterization data
for all compounds, including 400 MHz 1H NMR data and
spectra and HR-MS data; elemental analysis, gas chromatograph
1
trace, and a 13C NMR spectrum for 6a; 400 MHz H NMR
spectra for the four batches of 6a obtained during the manual
recycling experiment; FT-IR spectra for the resin-bound inter-
mediates in the synthesis of 6a; Chiral HPLC data for
compounds 8a and 8b; HPLC data for the compounds 6a, 17,
and 18 obtained by automated resin recycling; and GC-MS data
for the reaction monitoring of the manual recycling experiment
(36 pages). See any current masthead page for ordering and
Internet access instructions.
Chiral HPLC Procedure. After purification by SPE, the compound
(8a or 8b) was dissolved in hexane/ethanol (3:1) and injected onto a
Perkin Elmer HPLC system, comprising a CHIRALCEL OJ column
(25 cm × 0.46 cm), Series 200 LC pump, ISS200 advanced LC sample
processor, Diode array 235C detector, Nelson link 600 interface, and
LC1010 oven. Eluent: hexane/isopropyl alcohol/diethylamine (99:1:
0.1). Flow rate: 1 mL/min. UV detection at 220 nm. Temperature:
JA963829F