The dutch resolution variant of the classical resolution of racemates by formation of diastereomeric salts: Family behaviour in nucleation inhibition

Article abstract of DOI:10.1002/chem.200500440

The resolution of racemates through their diastereomeric salts can be positively affected by the addition of small amounts of suitable nucleation inhibitors. This discovery is a logical extension of "Dutch Resolution", in which equimolar amounts of resolving agents that are members of the same family (i.e., structurally related) are used. We conducted a systematic search for nucleation inhibitors of the resolving agent 1-phenylethylamine. A wide range of amines that bear possible family resemblances to 1-phenylethylamine was investigated. It was found that (R)-1-phenylbutylamine is a good inhibitor of (R)-1-phenylethylamine. Results of turbidity measurements showed that, for the model case of mandelic acid resolution, the chief effect of this inhibitor was to widen the metastable zone for the more soluble diastereomer. This observation is in accordance with previous experience. Further scouting for possible family members revealed a wide variation in the effectiveness of inhibitors, dependent on their structure. By far the most effective inhibitors are bifunctional 1-phenylethylamine and/or 1-phenylbutylamine analogues. The effect of racemic inhibitors was found to approach that of enantiomerically pure inhibitors of the same absolute configuration of the 1-phenylethylamine used for resolution. The most effective inhibitors were tested for the resolution of a structural variety of racemates, and were shown to be broadly applicable.

Full text of DOI:10.1002/chem.200500440

FULL PAPER
DOI: 10.1002/chem.200500440
The Dutch Resolution Variant of the Classical Resolution of Racemates by
Formation of Diastereomeric Salts: Family Behaviour in Nucleation
Inhibition
Jan Dalmolen,[a] Theodora D. Tiemersma-Wegman,[a] JosØ W. Nieuwenhuijzen,[b]
Marcel van der Sluis,[b] Erik van Echten,[b] Ton R. Vries,[b] Bernard Kaptein,[c]
Quirinius B. Broxterman,[c] and Richard M. Kellogg*[b]
Abstract: The resolution of racemates
through their diastereomeric salts can
be positively affected by the addition
of small amounts of suitable nucleation
inhibitors. This discovery is a logical
extension of “Dutch Resolution”, in
which equimolar amounts of resolving
agents that are members of the same
family (i.e., structurally related) are
that (R)-1-phenylbutylamine is a good
inhibitor of (R)-1-phenylethylamine.
Results of turbidity measurements
showed that, for the model case of
mandelic acid resolution, the chief
effect of this inhibitor was to widen the
metastable zone for the more soluble
diastereomer. This observation is in ac-
cordance with previous experience.
Further scouting for possible family
members revealed a wide variation in
the effectiveness of inhibitors, depen-
dent on their structure. By far the most
effective inhibitors are bifunctional 1-
phenylethylamine and/or 1-phenylbu-
tylamine analogues. The effect of race-
mic inhibitors was found to approach
that of enantiomerically pure inhibitors
of the same absolute configuration of
the 1-phenylethylamine used for reso-
lution. The most effective inhibitors
were tested for the resolution of a
structural variety of racemates, and
were shown to be broadly applicable.
used. We conducted
a
systematic
search for nucleation inhibitors of the
resolving agent 1-phenylethylamine. A
wide range of amines that bear possible
family resemblances to 1-phenylethyl-
amine was investigated. It was found
Keywords:
chiral resolution
·
chirality · crystal growth · diastereo-
selectivity · enantioselectivity
Introduction
equimolar mixture of, in general, three resolving agents of
the same family (i.e., with close structural analogy, common
absolute stereochemistry) was used. Non-stoichiometric in-
corporation of resolving agents and often improved diaster-
eomeric excesses of the first salts were observed.
“Dutch Resolution” is the use of families of resolving agents
in the otherwise classical Pasteur separation of racemates
through their diastereomeric salts.[1,2] As first described an
Further investigations revealed that certain family mem-
bers of resolving agents failed to be, or were only slightly,
incorporated into the salts, although they had a positive
effect on the resolution.[3] An example is the ortho/para-
nitro mixture (R)-2b and the classical resolving agent 1-phe-
nylethylamine (R)-2a, as illustrated for the case of the reso-
lution of mandelic acid (1) in Scheme 1. On the basis of tur-
bidity measurements, it was shown that 2b is an effective
nucleation inhibitor. It is not detectably incorporated into
the less soluble salt, which is obtained in significantly higher
diastereomeric excess in the presence of 2b (55% de com-
pared to 14% de without inhibitor under the experimental
conditions used). Inhibitor 2b acts by widening the width of
the metastable zone of supersaturation (the temperature
zone between dissolution and the lower temperature at
[a] J. Dalmolen, T. D. Tiemersma-Wegman
Department of Organic and Molecular Inorganic Chemistry
University of Groningen, Nijenborgh 4
9747 AG Groningen (The Netherlands)
[b] J. W. Nieuwenhuijzen, M. van der Sluis, E. van Echten, T. R. Vries,
Prof. R. M. Kellogg
Syncom B.V., Kadijk 3, 9747 AT Groningen (The Netherlands)
Fax : (+31)50-5757399
E-mail: r.m.kellogg@syncom.nl.
[c] B. Kaptein, Q. B. Broxterman
DSM Pharma Chemicals-Advanced Synthesis and Catalysis
P.O. Box 18, 6160 MD Geleen (The Netherlands)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2005, 11, 5619 – 5624
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5619

the “wrong” enantiomer, (S)-2c, also has some positive
effect (entry 4). Notably, (R)-2c alone is not a resolving
agent for 1 (entry 5). The yield, de and S-factor of the reac-
tion corresponding to entry 2, performed on
a 10 g
(65.8 mmol) scale, were within 1% of the results obtained
on the 2 mmol scale reported in Table 1.
Turbidity measurements (see Supporting Information) re-
vealed that, for (R)-2c, analogously to 2b, the metastable
zone width for the more soluble (S)-1/(R)-2a diastereomeric
salt is increased from 7.38C without inhibitor to 28.58C
(zone width 25.38C for racemic 2c and 13.08C for (S)-2c),
whereas for the less soluble (R)-1/(R)-2a diastereomeric
salt, the zone width increases from 3.98C without inhibitor
to only 7.08C (5.98C for racemic 2c and 7.08C for (S)-2c).
The conclusion is clear: 2c, of the same relative configura-
tion as resolving agent 2a, is an effective nucleation inhibi-
tor that affects chiefly the more soluble diastereomeric salt
by increasing the metastable temperature zone width; the
more soluble diastereomer remains in solution for a longer
period of time.[4c] An analogous effect is observed in slightly
diminished form for racemic 2c, and relatively weakly for
the “wrong” (S) enantiomer of 2c.
Scheme 1. Resolution of mandelic acid (1) with (R)-1-phenylethylamine
2a: additives are (R)-2b and (R)-2c.
which precipitation begins) to a greater extent for the more
soluble diastereomer than for the less soluble diastereomer.
This results in, with correct temperature control, more op-
portunities for selective precipitation of the less soluble dia-
stereomeric salt.[4a–c]
The use of a single resolving agent with a small amount of
additive (nucleation inhibitor) instead of a mixture is tech-
nologically simpler, but how can nucleation inhibitors in the
context of Dutch Resolution be identified? Here, we de-
scribe some guidelines derived from the study of amines,
and indicate how these can be applied.
Although 2c is clearly an inhibitor, it seems unlikely that
every small change in the structure of a resolving agent will
result in a nucleation inhibitor. Are there predictable struc-
tural considerations and, if so, what are they? To examine
further the effects of substitution in the aryl ring, the com-
pounds 3–25 (all enantiomerically pure (R)) shown in
Figure 1 were investigated as inhibitors for the resolution of
mandelic acid (1) by 1-phenylethylamine (2a), as depicted
in Scheme 1.[8] All experiments were performed in triplicate
and the values of the S-factors are reproducible to within (Æ)
5%. If a “hit” is arbitrarily defined as Sꢀ0.35, relative to
S=0.24 without additive (see farthest right bar in Figure 1,
in which the concentration of resolving agent 2a is identical
to that in the inhibition experiment), then 2c itself, the
3-nitro (26), 2-hydroxy (23), 1-naphthyl (24), 2-fluoro (22),
2-methoxy (21) and 2-naphthyl (20) derivatives are active
inhibitors (7 out of a library of 24). Note that the halo-sub-
stituted derivatives 3–7 as additives lead to significantly
lower S-values. The other compounds have only modest ef-
fects.
By contrast, the effects of structural modification of the
side-chain are more profound (Figure 2). In this case, not
only compounds with the (R) absolute configuration analo-
gous to (R)-2a were examined, but also racemates and
“wrong” (S)-enantiomers. By using the same arbitrary defi-
nition of a “hit”, 17 of the 21 compounds examined show
significant effects. The achiral additives 26, 28, 32 and 36
have virtually no effect. Cyclization of the side-chain (33–
35) produces no dramatic improvements. Branching of the
side-chain can, however, be quite effective, as exemplified
by 41.[9]
Results and Discussion
The model system that was used for scanning is that shown
in Scheme 1, and the primary goal was to search as quickly
as possible for nucleation inhibitors among other potential
“family members” related to the parent resolving agent 2a.
We first looked at analogues of resolving agent 2a, in which
the side-chain rather than the aryl ring is modified structur-
ally. In orientational experiments, it was observed that readi-
ly available[5] (in both enantiomeric forms) 1-phenylbutyl-
amine (R)-2c induced significant improvements if used as
an additive with (R)-2a. Notably, no (R)-2c can be detected
in the precipitated salt. Yields and de values are shown in
Table 1.[6] Diastereomerically pure salts cannot be obtained
in more than 50% yield. We prefer the use of Fogassy S-fac-
tors (2”yield”de) to express the overall efficiencies.[7] Scan-
ning of 2a/2c ratios revealed that the optimal concentration
for nucleation inhibition was 6 mol% of 2c (see Supporting
Information).
Rather unexpectedly, it was observed that racemic 2c is
also nearly as effective as (R)-2c (Table 1, entry 3) and that
Table 1. Resolution of (Æ)-1[a] with (R)-2a in the absence and presence
of 6 mol% 2c.
Entry
Resolving
agent
Additive
Yield
[%]
de
[%]
S-factor
1
2
3
4
5
(R)-2a
(R)-2a
(R)-2a
(R)-2a
(R)-2c
–
68
60
62
61
14
42
35
30
0.19
0.50
0.43
0.37
(R)-2c
(Æ)-2c
(S)-2c
–
However, by far the most potent inhibitors are the di-
amine derivatives (42–45).[10] We included these compounds
in the screen because molecules with repeating functionality
in their structure often have a greater tendency to aggregate,
no salts precipitate
[a] Concentration=0.40 mmolmLÀ1 in CH3CH(OH)CH3.
5620
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5619 – 5624

Dutch Resolution of Racemates
FULL PAPER
Figure 1. S-Factors of (R)-aryl substituted analogues of (R)-2c as possible nucleation inhibitors.
Figure 2. S-Factors of (R)-side-chain-substituted analogues of (R)-2a as possible nucleation inhibitors.
and might influence nucleation in a different manner
(gemini effect).[11]
The proof of the pudding is in the tasting. Do the better
nucleation inhibitors identified by screening also work in
the resolution of other compounds? Those shown in
Figure 3 are clearly the most potent. Selected results for the
resolution of a-methylphenylacetic acid (46),[12] a-methoxy-
phenylacetic acid (47)[13] and N-acetylleucine (48)[14] with
(R)-2a are summarised in Table 2.[15]
Chem. Eur. J. 2005, 11, 5619 – 5624
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5621

R. M. Kellogg et al.
2) The effect of nucleation in-
hibition is greatest on the
more soluble diastereomeric
salt.[4c,16]
3) A search for inhibitors is
best carried out among
“family members” of the re-
solving agent.
4) There are, at least at this
stage, no hard and fast rules
for the identification with
regard to structure, and all
of the suspects must be sub-
jected to screening.
5) Absolute stereochemistry is
not an absolute prerequisite,
and racemates may be used
for screening purposes.
6) Although we are now able
to identify inhibitors by
using screening methods,
our understanding of the in-
hibition phenomenon itself and of the stereochemical as-
pects in particular remains insufficient.
Figure 3. S-Factors of (R)- and racemic diamine inhibitors as possible nucleation inhibitors.
Table 2. Selected results for the resolution of racemates 46–48 with (R)-
2a, obtained by using the most effective nucleation inhibitors shown in
Figure 3.
In future work, we intend to develop automated protocols
for the screening of inhibitors, to develop specific inhibitors
for the most commonly used resolving agents for which fam-
ilies exist, to find methodologies based on the discoveries
described in this paper that permit application to large scale
resolutions and to understand better the mechanism of
action of nucleation inhibitors.
Entry Racemate Additive (mol%) Yield [%] de [%] S-factor
1
2
3
4
5
6
7
8
9
46
46
46
46
47
47
48
48
48
48
–
60
36
54
18
65
16[a]
58
35
44
56
7
59
52
97
1
96
13
75
48
32
0.08
0.43
0.56
0.35
0.02
0.30
0.15
0.53
0.43
0.36
(Æ)-42 (3%)
(Æ)-43 (3%)
(R,R)-44 (6%)
–
Experimental Section
(Æ)-42 (6%)
–
General procedure for the small-scale nucleation inhibition experiments
described in Table 1: In a Kimble reactor tube (dimensions 1 25”
150 mm), provided with a cylindrical, PTFE-coated magnetic stirring bar
(10”6 mm), 0.12 mmol (0.06 mol equiv) of additive 2c and 1.88 mmol
(0.94 mol equiv) of (R)-1-phenylethylamine (2a) were mixed in 3.33 mL
of CH3CH(OH)CH3. Subsequently, 2.0 mmol racemic mandelic acid (1)
(1.0 mol equiv) in 1.67 mL CH3CH(OH)CH3 was added. The mixture was
heated until a clear solution was obtained. After the reactor tube was
sealed with a rubber stopper, it was placed in the Varian thermostatted
bath and mechanically stirred at 788C for 30 min. The tubes were gradu-
ally cooled to 208C at a ramp rate of À108C hÀ1 and stirred at that tem-
perature for 12 h. The precipitated salts were collected by filtration by
(Æ)-43 (3%)
(R,R)-44 (3%)
(R,R)-45 (6%)
10
[a] Salt isolated after 6 days.
In all cases, the de values of the first-isolated salts in-
crease to very acceptable values, as the yields decrease. In
our experience, this often, but not always, happens. We also
emphasise that the experiments described here may not be
optimal with regard to the concentration of inhibitor and
temperature. Nevertheless, we consider these non-optimised
results to be extremely promising.
using
a
VacMasterꢁ-20, then each was washed with 1.5 mL of
CH3CH(OH)CH3 and dried. HPLC analysis was used to determine the
diastereomeric excesses of the salts. To ensure accurate de determination,
the racemic substrate was measured first in each case. The composition
of the salt was determined by conducting mass spectrometry, and 1H and
13C NMR spectroscopy. The resolution experiments described in Tables
S4, S5 and S6 of the Supporting Information were performed analogously
to this general procedure, with either 0.03 or 0.06 mol equivalent of addi-
tive (and, respectively, 0.97 or 0.94 mol equiv of parent resolving agent
2a) with respect to the 1.0 mol equivalent of racemic substrate. The sol-
vent(s) and concentrations used are listed at the bottom of each table.
The conditions used in the HPLC analysis for each substrate are given in
Summary
We conclude that:
1) Nucleation inhibition is often involved in Dutch Resolu-
tion.
5622
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5619 – 5624

Dutch Resolution of Racemates
FULL PAPER
Table S7. All experiments were performed in triplicate and the error
sis. After bulb-to-bulb distillation, the only products isolated were race-
margin was Æ5%.
mic 42 and 43 (rac:meso>99:1).
(Æ)-1-[3-(1-Aminoethyl)phenyl]-1-ethanamine (42): After workup, the
The additives used in the screening for possible nucleation inhibitors, as
described in Tables S5 and S6 of the Supporting Information, were either
commercially available (additives 28 and 34–36) or were synthesised by
using a simple one-pot Leuckart synthesis (additives 2c, 26, 27, 29, 31–33,
37, 38 and 40–43). The syntheses of enantiopure 2c, 30, 39 and 40 were
as previously reported.[5]
reaction mixture contained
a rac:meso ratio of 85:15, according to
1H NMR analysis. After Kugelrohr distillation, the only product isolated
was racemic 42 (rac:meso>99:1) as a colourless oil, 53% yield after
bulb-to-bulb distillation; 1H NMR (300 MHz, CDCl3): d=1.28 (d, J=
6.59 Hz, 6H), 2.63 (brs, 4H), 3.98 (q, J=6.59 Hz, 2H), 7.07–7.17 ppm (m,
4H); 13C NMR (50 MHz, CDCl3): d=25.07 (q), 49.64 (d), 122.85 (d),
124.10 (d), 128.60 (d), 147.29 ppm (s); MS (CI): m/z: 165 [M+H]+.
General procedure for the Leuckart synthesis of additives 2c, 26, 27, 29,
31–33, 37, 38 and 40–43: A mixture of the corresponding aldehyde or
ketone (10 mmol), formamide (20 mL) and formic acid (10 mL) was
heated to reflux and then refluxed for 1 h. After cooling to ambient tem-
perature, 30 mL of water was added and the mixture was extracted with
diethyl ether (3”10 mL). The combined organic layers were washed with
brine, dried over Na2SO4 and concentrated to furnish the intermediate
formamide. Aqueous HCl (20 mL, 30%) was added and the reaction
mixture was refluxed for 1 h. After cooling to ambient temperature,
20 mL of water was added. The reaction mixture was carefully adjusted
to pH 10 with aqueous NaOH (33%) and extracted with diethyl ether
(3”20 mL). The combined organic layers were washed with brine, dried
over Na2SO4 and concentrated to furnish the corresponding primary
amines.
(Æ)-1-[4-(1-Aminoethyl)phenyl]-1-ethanamine (43): After workup, the
reaction mixture contained a {(R,R)+(S,S)}:meso ratio of 80:20, accord-
ing to 1H NMR analysis. After Kugelrohr distillation, the only product
isolated was racemic 43 ({(R,R)+(S,S)}:meso>99:1) as a colourless oil,
47% yield after Kugelrohr distillation; 1H NMR (300 MHz, CDCl3): d=
1.33 (d, J=6.59 Hz, 6H), 1.51 (brs, 4H), 4.05 (q, J=6.59 Hz, 2H),
7.26 ppm (s, 4H); 13C NMR (50 MHz, CDCl3): d=25.50 (q), 50.82 (d),
125.60 (d), 146.16 ppm (s); MS (CI): m/z: 165 [M+H]+.
Bifunctional additives (R,R)-44 and (R,R)-45 were synthesised as enan-
tiopure compounds by using a three-step procedure reported previous-
ly.[5]
Procedure for the synthesis of (R)-phenylglycine amide (PGA)-diimines:
The disubstituted benzaldehyde (100 mmol) was added to a suspension of
(R)-phenylglycine amide (200 mmol, 30.0 g) in CH2Cl2 (200 mL) at ambi-
ent temperature. The reaction mixture was stirred overnight at room
temperature. After removal of the CH2Cl2, the residual solid was recrys-
tallised once from acetone/hexane (1:20).
1-Propylbutylamine (26): Colourless liquid, 55% yield after Kugelrohr
1
distillation; H NMR (400 MHz, CDCl3): d=0.75–0.80 (m, 6H), 1.17–1.36
(m, 8H), 2.78 (t, J=5.32 Hz, 1H), 5.02 ppm (brs, 2H); 13C NMR
(50 MHz, CDCl3): d=13.76 (q), 18.50 (t), 37.50 (t), 51.71 ppm (d); MS
(CI): m/z: 116 [M+H]+.
(Æ)-2,2-Dimethyl-1-phenyl-1-propanamine (27): Pale yellow oil, 95%
yield; 1H NMR (300 MHz, CDCl3): d=0.86 (s, 9H), 1.40 (brs, 2H), 3.65
(s, 1H), 7.17–7.25 ppm (m, 5H); 13C NMR (50 MHz, CDCl3): d=26.43
(q), 34.91 (s), 65.22 (d), 126.62 (d), 127.38 (d), 128.14 (d), 143.68 ppm (s);
MS (EI): m/z: 163 [M]+.
(2R)-2-({(E)-[3-({[(1R)-2-Amino-2-oxo-1-phenylethyl]imino}methyl)phe-
nyl]methylidene}amino)-2-phenylethanamide (49): White solid, 96%
yield. M.p. 143.8–144.58C; 1H NMR (300 MHz, CDCl3/[D6]DMSO): d=
4.55 (s, 2H), 6.60 (brs, 2H), 6.63 (brs, 2H), 6.82–6.91 (m, 7H), 7.02–7.09
(m, 4H), 7.49 (d, J=7.99 Hz, 2H), 7.79 (s, 1H), 7.94 ppm (s, 2H);
13C NMR (50 MHz, CDCl3/[D6]DMSO): d=76.24 (d), 126.42 (d), 126.85
(d), 127.64 (d), 128.15 (d), 130.16 (d), 135.03 (s), 138.59 (s), 161.37 (d),
172.60 ppm (s); elemental analysis calcd (%) for C24H22N4O2: C 72.34, H
5.57, N 14.06; found: C 72.21, H 5.66, N 14.03; MS (CI): m/z: 399
[M+H]+.
(Æ)-Cyclobutyl(phenyl)-methanamine (29): Orange oil, 61% yield;
1H NMR (300 MHz, CDCl3): d=1.39 (brs, 2H), 1.60–1.85 (m, 5H), 2.07–
2.14 (m, 1H), 2.41–2.49 (m, 1H), 3.73 (d, J=9.15 Hz, 1H), 7.15–7.28 ppm
(m, 5H); 13C NMR (50 MHz, CDCl3): d=17.41 (t), 25.33 (t), 26.04 (t),
43.21 (d), 61.64 (d), 126.51 (d), 126.77 (d), 128.17 (d), 144.70 ppm (s); MS
(CI): m/z: 162 [M+H]+.
(Æ)-1-Phenyl-1-pentanamine (31): Yellow oil, 87% yield; 1H NMR
(300 MHz, CDCl3): d=0.82 (t, J=6.78 Hz, 3H), 1.06–1.27 (m, 4H), 1.41
(brs, 2H), 1.57–1.63 (m, 2H), 3.80 (t, J=6.78 Hz, 1H), 7.13–7.31 ppm (m,
5H); 13C NMR (50 MHz, CDCl3): d=13.79 (q), 22.46 (t), 28.55 (t), 39.21
(t), 56.08 (d), 126.07 (d), 126.54 (d), 128.14 (d), 146.69 ppm (s); MS (EI):
m/z: 163 [M]+.
(2R)-2-({(E)-[4-({[(1R)-2-Amino-2-oxo-1-phenylethyl]imino}methyl)phe-
nyl]ethylidene}amino)-2-phenylethanamide (50): White solid, 98% yield.
M.p. 168.28C; 1H NMR (300 MHz, CDCl3/[D6]DMSO): d=4.53 (s, 2H),
6.58 (brs, 4H), 6.82–6.93 (m, 6H), 7.04 (dd, J=8.42, J=1.46 Hz, 4H),
7.45 (s, 4H), 7.93 ppm (s, 2H); 13C NMR (50 MHz, CDCl3/[D6]DMSO):
d=76.39 (d), 126.42 (d), 126.88 (d), 127.67 (d), 127.83 (d), 137.01 (s),
138.61 (s), 161.32 (d), 172.483 ppm (s); elemental analysis calcd (%) for
C24H22N4O2: C 72.34, H 5.57, N 14.06; found: C 71.95, H 5.58, N 13.87;
MS (CI): m/z: 399 [M+H]+.
(Æ)-6,7,8,9-Tetrahydro-5H-benzo[a]cyclohepten-5-amine (33): Colourless
liquid, 53% yield after Kugelrohr distillation; 1H NMR (300 MHz,
CDCl3): d=1.52–2.02 (m, brs, 8H), 2.82–2.87 (m, 2H), 4.21–4.25 (m,
1H), 7.10–7.26 (m, 4H), 7.42–7.45 ppm (m, 1H); 13C NMR (50 MHz,
CDCl3): d=27.42 (t), 28.69 (t), 35.67 (t), 37.11 (t), 54.60 (d), 124.00 (d),
126.00 (d), 126.19 (d), 129.28 (d), 141.25 (s), 145.57 ppm (s); MS (EI):
m/z: 161 [M]+.
Procedure for the allylation of (R)-PGA diimines 49 and 50: A solution
of allylzinc bromide (3.0 equiv) was prepared by adding allyl bromide
(25.7 mL, 292 mmol) to finely cut zinc wool (19.1 g, 292 mmol) in THF
(150 mL). The solution of allylzinc bromide was cooled to room tempera-
ture and 97.3 mmol of the imine in THF (50 mL) was added at 08C. The
reaction mixture was warmed to room temperature and then poured into
water (100 mL). Ethyl acetate (70 mL) was added and the mixture was
stirred vigorously. After filtration through Celite, the organic phase was
separated and the water layer was extracted with ethyl acetate (2”
100 mL). The combined organic phase was dried over sodium sulfate and
the ethyl acetate was evaporated to furnish the PGA allylamines 51 or
52, which in both cases crystallised on standing.
(Æ)-1-Methylbutylamine (37): Colourless oil, 65% yield after Kugelrohr
distillation; 1H NMR (200 MHz, CDCl3): d=0.81–0.88 (m, 3H), 0.98 (d,
J=6.34 Hz, 3H), 1.10–1.39 (m, 4H), 1.63 (brs, 2H), 2.77–2.86 ppm (m,
1H); 13C NMR (50 MHz, CDCl3): d=13.86 (q), 19.29 (t), 23.63 (q), 42.18
(t), 46.37 ppm (d); MS (EI): m/z: 88 [M]+.
(Æ)-2-Methyl-1-phenyl-1-propanamine (41): Yellow oil, 83% yield;
1H NMR (200 MHz, CDCl3): d=0.72 (d, J=6.59 Hz, 3H), 0.92 (d, J=
6.59 Hz, 3H), 1.41 (brs, 2H), 1.74–1.85 (m, 1H), 3.54 (d, J=7.32 Hz,
1H), 7.17–7.28 ppm (m, 5H); 13C NMR (50 MHz, CDCl3): d=18.81 (q),
19.70 (q), 35.36 (d), 62.37 (d), 128.03 (d), 126.91 (d), 128.03 (d),
150.91 ppm (s); MS (CI): m/z: 150 [M+H]+.
(2R)-2-({(1R)-1-[3-((1R)-1-{[(1R)-2-Amino-2-oxo-1-phenylethyl]ami-
no}-3-butenyl)phenyl]-3-butenyl}amino)-2-phenylethanamide
(51):
Yellow needles, 99% yield, >99:1 dr. M.p. 126.6–127.98C; 1H NMR
(300 MHz, CDCl3): d=2.22 (brs, 2H), 2.36 (t, J=6.59 Hz, 4H), 3.67 (t,
J=6.59 Hz, 2H), 3.94 (s, 2H), 4.97–5.03 (m, 4H), 5.59–5.72 (m, 2H), 6.58
(brs, 2H), 7.01–7.25 ppm (m, 16H); 13C NMR (50 MHz, CDCl3): d=
42.11 (t), 61.32 (d), 63.88 (d), 117.69 (t), 124.97 (d), 126.41 (d), 127.04
(d), 127.88 (d), 128.60 (d), 134.45 (d), 138.79 (s), 142.64 (s), 150.70 (d),
175.84 ppm (s); elemental analysis calcd (%) for C30H34N4O2: C 74.66, H
Note that in the cases of 42 and 43, two stereocenters are present and a
meso compound is possible. After the Leuckart reductive amination of
the corresponding bis-aldehydes, the reaction mixture contained a rac:-
meso ratio of 85:15 and 80:20, respectively, according to 1H NMR analy-
Chem. Eur. J. 2005, 11, 5619 – 5624
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5623

R. M. Kellogg et al.
7.10, N 11.61; found: C 74.66, H 7.43, N 11.53; MS (CI): m/z: 483
[M+H]+.
[3] a) J. W. Nieuwenhuijzen, R. F. P. Grimbergen, C. Koopman, R. M.
Kellogg, T. R. Vries, K. Pouwer, E. van Echten, B. Kaptein, L. A.
Hulshof, Q. B. Broxterman, Angew. Chem. 2002, 114, 4457–4462;
Angew. Chem. Int. Ed. 2002, 41, 4281–4286; b) R. M. Kellogg, J. W.
Nieuwenhuijzen, K. Pouwer, T. R. Vries, Q. B. Broxterman, R. F. P.
Grimbergen, B. Kaptein, R. M. La Crois, E. de Wever, K. Zwaag-
stra, A. C. van der Laan, Synthesis 2003, 1626–1638.
[4] Calculations support the formation of solid solutions of (roughly iso-
morphous) diastereomeric salts: a) C. Gervais, R. F. P. Grimbergen,
I. Markovits, G. J. A. Ariaans, B. Kaptein, A. Bruggink, Q. B. Brox-
terman, J. Am. Chem. Soc. 2004, 126, 655–662; see also: b) B. Kap-
tein, H. Elsenberg, R. F. P. Grimbergen, Q. B. Broxterman, L. Hul-
shof, K. L. Pouwer, T. R. Vries, Tetrahedron: Asymmetry 2000, 11,
1343–1351; c) This conclusion is an extrapolation from turbidity
measurements taken by necessity at different concentrations for the
pure diastereomers; a more exact explanation can be derived from
the ternary phase diagram discussed in ref. [3a].
[5] a) M. van der Sluis, J. Dalmolen, B. de Lange, B. Kaptein, R. M. Kel-
logg, Q. B. Broxterman, Org. Lett. 2001, 3, 3943–3946; b) J. Dalmo-
len, M. van der Sluis, J. W. Nieuwenhuijzen, A. Meetsma, B. de -
Lange, B. Kaptein, R. M. Kellogg, Q. B. Broxterman, Eur. J. Org.
Chem. 2004, 1544–1557.
[6] Details of an efficient HPLC method to allow rapid and accurate
analysis are given in the Supporting Information.
[7] E. Fogassy, A. Lopata, F. Faigl, F. Darvas, M. ꢂcs, L. Tꢃke, Tetrahe-
dron Lett. 1980, 21, 647–650.
(2R)-2-({(1R)-1-[4-((1R)-1-{[(1R)-2-Amino-2-oxo-1-phenylethyl]ami-
no}-3-butenyl)phenyl]-3-butenyl}amino)-2-phenylethanamide (52): Pale
yellow brittle solid, 99% yield, >99% dr. M.p. 96.5–98.98C; 1H NMR
(300 MHz, CDCl3): d=2.16–2.44 (m, brs, 5H), 3.70 (dt, J=6.2 Hz, 2H),
4.05 (s, 2H), 4.95–5.10 (m, 4H), 5.61–5.69 (m, 2H), 6.26 (brs, 2H), 7.31–
7.09 ppm (m, 16H); 13C NMR (50 MHz, CDCl3): d=42.01 (d), 61.46 (d),
64.08 (d), 118.11 (t), 127.36 (d), 127.49 (d), 128.38 (d), 128.96 (d), 134.46
(d), 138.43 (s), 141.23 (s), 175.90 ppm (s); elemental analysis cald (%) for
C30H34N4O2: C 74.66, H 7.10, N 11.61; found: C 74.70, H 7.06, N 11.62;
MS (CI): m/z: 483 [M+H]+.
Procedure for the synthesis of diamino additives (R,R)-44 and (R,R)-45
by catalytic hydrogenation of PGA allylamines 51 and 52: The PGA al-
lylamine (5.0 mmol) was dissolved in CH3CH(OH)CH3 (50 mL). Water
(50 mL), acetic acid (50 mL) and Pd-C (10%) (0.4 g, cat) were added
successively. After application of two vacuum/H2 cycles to remove air
from the reaction flask, the stirred mixture of the substrate was hydro-
genated at an ambient pressure of H2 and room temperature for 5 d.
After filtration, the CH3CH(OH)CH3 was evaporated under reduced
pressure. The residue was diluted with water (50 mL) and the acidic reac-
tion mixture was washed once with diethyl ether to remove any byprod-
ucts. The aqueous phase was adjusted to pH 10 with 10% NaOH and
then extracted with CH2Cl2 (3”40 mL). The combined organic phase was
washed with brine, dried over sodium sulfate and filtered. After evapora-
tion of the CH2Cl2, pentane was added to the residue. After filtration
through a glass funnel, the pentane was removed under reduced pressure.
Bulb-to-bulb distillation yielded primary amines (R,R)-44 and (R,R)-45.
[8] All these compounds were available by means of procedures de-
scribed in ref. [5].
[9] A 1:1:1 mixture of enantiopure 2a, 37 and 42, the latter two of
which appear to be nucleation inhibitors (Table 2) has been used in
the past (ref. [1]) as a family of resolving agents (“PE-III mix”).
[10] The preparation of these materials by means of a Leuckart reaction
and subsequent easy isolation is described in the Supporting Infor-
mation.
1-{3-[(1R)-1-Aminobutyl]phenyl}-1-butylamine (44): Colourless oil, 80%
yield; 1H NMR (300 MHz, CDCl3): d=0.84 (t, J=7.33 Hz, 6H), 1.11–
1.34 (m, 4H), 1.52–1.61 (m, brs, 8H), 3.82 (t, J=6.96 Hz, 2H), 7.11 (d,
J=7.69 Hz, 2H), 7.18 (s, 1H), 7.21 ppm (t, J=6.96 Hz, 1H); 13C NMR
(50 MHz, CDCl3): d=13.96 (q), 19.70 (t), 41.84 (t), 55.97 (d), 124.22 (d),
124.69 (d), 128.38 (d), 155.35 ppm (s); MS (CI): m/z: 441 [2M+H]+.
[11] Catalysts: a) R. G. Konsler, J. Karl, E. N. Jacobsen, J. Am. Chem.
Soc. 1998, 120, 10780–10781; bis-urea gelators: b) J. Brinksma, B. L.
Feringa, R. M. Kellogg, R. Vreeker, J. van Esch, Langmuir 2000, 16,
9249–9255; amide-containing Gemini surfactants: c) M. Johnsson,
J. B. F. N. Engberts, J. Phys. Org. Chem. 2004, 17, 934–944.
[12] Conc.=0.30 mmolmLÀ1 in CH3CH(OH)CH3:H2O (20:1). On addi-
tion of additive, the crystallised material contained the (R)-2a/(S)-
46 in excess.
1-{4-[(1R)-1-Aminobutyl]phenyl}-1-butylamine (45): Pale yellow oil,
74% yield; 1H NMR (300 MHz, CDCl3): d=0.75 (t, J=7.33 Hz, 6H),
1.04–1.25 (m, 4H), 1.46–1.51 (m, brs, 8H), 3.70 (t, J=6.78 Hz, 2H),
7.09 ppm (s, 4H); 13C NMR (50 MHz, CDCl3): d=13.64 (q), 19.34 (t),
41.47 (t), 55.83 (d), 125.91 (d), 144.89 ppm (s); MS (CI): m/z: 441
[2M+H]+.
[13] Conc.=0.25 mmolmLÀ1 in CH3CH(OH)CH3:H2O (20:1). On addi-
tion of additive, the crystallised material contained the (R)-2a/(R)-
47 in excess.
Acknowledgements
[14] Conc.=0.75 mmolmLÀ1 in CH3CH(OH)CH3:H2O (5:2). On addi-
tion of additive, the crystallised material contained the (R)-2a/(R)-
d-48 in excess.
We are most grateful to Prof. Elias Vlieg of the Radboud University in
Nijmegen for fruitful discussions about Dutch Resolution and its link to
nucleation inhibition, and to Dr. Huub Grooten of the DSM Centre for
Particle Technology for aid with turbidity measurements. Partial support
(to J.D.) was provided by the Foundation for Applied Research (STW), a
division of the Dutch National Science Foundation (NWO).
[15] The “habit modifier” (R,R)-bis(a-methylbenzyl)amine [K. Sakai, Y.
Maekawa, K. Saigo, M. Sukegawa, H. Murakami, H. Nohira, Bull.
Chem. Soc. Jpn. 1992, 65, 1747–1750], the secondary amine analog
of 1-phenylethylamine 2a, was only modestly effective and raised
the S-factor for the resolution of 1 to 0.26 and that of 46 to 0.41;
these results were not sufficient to encourage further investigation.
[16] From turbidity measurements performed to date, we have observed
that “Dutch Resolution inhibitors” have the greatest effect on the
more soluble diastereomeric salt. However, not every inhibitor has
been subjected to turbidity measurements.
[1] T. R. Vries, H. Wynberg, E. van Echten, J. Koek, W. ten Hoeve,
R. M. Kellogg, Q. B. Broxterman, A. Minnaard, B. Kaptein, S. van -
der Sluis, L. A. Hulshof, J. Kooistra, Angew. Chem. 1998, 110, 2491–
2496; Angew. Chem. Int. Ed. 1998, 37, 2349–2354.
[2] There is a possible relationship to “tailor-made additives”: a) L.
Addadi, Z. Berkovitch-Yellin, N. Domb, E. Gati, M. Lahav, L. Lei-
serowitz, Nature 1982, 296, 21–26; b) Z. Berkovitch-Yellin, L.
Addadi, M. Idelson, L. Leiserowitz, M. Lahav, Nature 1982, 296, 27–
34.
Received: April 19, 2005
Published online: July 20, 2005
5624
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5619 – 5624