Two polymorphs of 4-hydroxypiperidine with different NH configurations

Article abstract of DOI:10.1039/c4ce02477j

4-Hydroxypiperidine 1 exists in two crystal forms, tetragonal 1t, space group P4ˉ2₁c and orthorhombic 1o, space group Fdd2, both with one molecule in the asymmetric unit. The latter form was obtained only rarely and in small quantities. In form 1t, the NH hydrogen is axial, whereas in 1o it is equatorial; the OH group is equatorial in both structures. The packing of both forms involves one hydrogen bond N-H?O and one O-H?N. In solution, NMR spectra indicate the presence of separate axial and equatorial forms (with respect to the OH group) below ca. -53 °C; however, not even at -104 °C, the lowest temperature reached, could any freezing out of the inversion at nitrogen be observed, implying that the energy barrier for this process is (as expected) small. We were unable to convert 1t, which appears to be the more stable form over the whole temperature range up to the melting point, to 1o by heating or via melting and re-cooling (or by any other method), perhaps because the hydrogen-bonding pattern is resistant to change. The crystalline forms 1t and 1o, despite being polymorphs of 1 with different NH configurations, should not be described as "configurational polymorphs" because of the facile interconversion in solution.

Full text of DOI:10.1039/c4ce02477j

View Article Online
View Journal
CrystEngComm
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: P. G. Jones, C.
Döring, I. Jess, C. Nather and K. Ibrom, CrystEngComm, 2015, DOI: 10.1039/C4CE02477J.
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the Ethical guidelines still
apply. In no event shall the Royal Society of Chemistry be held
responsible for any errors or omissions in this Accepted Manuscript
or any consequences arising from the use of any information it
contains.
www.rsc.org/crystengcomm

Page 1 of 11
CrystEngComm
Dynamic Article Links ►
Journal Name
View Article Online
DOI: 10.1039/C4CE02477J
Cite this: DOI: 10.1039/c0xx00000x
www.rsc.org/xxxxxx
ARTICLE TYPE
Two polymorphs of 4-hydroxypiperidine with different NH
configurations
a
b
b
c
Cindy Döring, Christian Näther, Inke Jess, Kerstin Ibrom and Peter G. Jones
*,a
Received (in XXX, XXX) Xth XXXXXXXXX 200X, Accepted Xth XXXXXXXXX 200X
5
DOI: 10.1039/b000000x
4
-Hydroxypiperidine 1 exists in two crystal forms, tetragonal 1t, space group P 4 21c and orthorhombic 1o, space group Fdd2, both with
one molecule in the asymmetric unit. The latter form was obtained only rarely and in small quantities. In form 1t, the NH hydrogen is
axial, whereas in 1o it is equatorial; the OH group is equatorial in both structures. The packing of both forms involves one hydrogen bond
N–HLO and one O–HLN. In solution, NMR spectra indicate the presence of separate axial and equatorial forms (with respect to the
OH group) below ca. –53°C; however, not even at –104°C, the lowest temperature reached, could any freezing out of the inversion at
nitrogen be observed, implying that the energy barrier for this process is (as expected) small.. We were unable to convert 1t, which
appears to be the more stable form over the whole temperature range up to the melting point, to 1o by heating or via melting and re-
cooling (or by any other method), perhaps because the hydrogen-bonding pattern is resistant to change. The crystalline forms 1t and 1o,
despite being polymorphs of 1 with different NH configurations, should not be described as "configurational polymorphs" because of the
facile interconversion in solution.
1
0
1
5
alternative crystal habits as a precondition for more detailed
Introduction
investigations. Our chemical and crystallographic interests,
9
5
5
6
6
7
7
8
0
5
0
5
0
5
0
including studies of amine complexes of gold and
As witnessed by the appearance of this special edition of
CrystEngComm, the phenomenon of polymorphism, the ability of
a chemical compound to exist in more than one crystalline form,
is the focus of much current interest and has been the subject of
many books and review articles. A famous specific case is that
of benzamide, for which the structure of a second polymorph was
established some 170 years after it was first observed, and the
existence of a third polymorph has recently been established.
One particular aspect of polymorphism research is the attempt
to classify various types of polymorphism. A fascinating recent
10
solvates/adducts of urea derivatives, have provided numerous
examples. The most recent of these is 4-hydroxypiperidine (1),
which we present here. The two crystalline forms differ in the
configuration at the nitrogen-bonded hydrogen atom.
Substituted piperidines can in principle change their ring
conformation by ring inversion and can also undergo inversion at
the nitrogen atom. Relevant energy terms for such processes are
2
2
3
3
4
4
0
5
0
5
0
5
1
2
the free energy difference ꢀꢀG between the mutually inverted
‡
forms and the free energy of activation (
ꢀ
G ) for the inversion
process. The conformations of piperidines in solution have been
the subject of several NMR spectroscopic studies (e.g. Refs. 11-
3
review article, for example, presents many examples of
conformational polymorphism, whereby the two or more forms of
a compound differ significantly in one or more torsion angles.
However, as the authors make clear, there are also a great number
of cases of polymorphism involving virtually identical molecular
2
0). Piperidine itself prefers the chair conformation with a barrier
–1 15
to ring inversion of 10.4 kcal mol . Initially there was a
controversial discussion about the axial-equatorial NH
equilibrium in piperidine.
15a,16
17
Anet and Yavari, however, were
3
conformations.
‡
able to determine
ꢀG for nitrogen inversion as 6.1 ± 0.2 kcal
If conformational polymorphism is an established feature, one
might provocatively ask if configurational polymorphism could
also exist. Since different configurations such as cis/trans imply
–1
1
13
mol in a H and C NMR study at very low temperatures down
to –174 °C, with ꢀꢀG only 0.36 kcal/mol. They concluded that
the more stable orientation of the NH proton is probably
equatorial.
4
different isomers and thus different compounds, this appears to
rule out the term polymorph for configurational isomers. What
happens, however, if the energy barrier is not too great, so that
the isomers might be interconvertible in solution, in the melt or
even in the solid state? In this connection we note that the
philosopher C. E. M. Joad was famous for answering every
question with "It depends what you mean by K"; exact
definitions are important, and we return to this principle below.
It is not easy to create polymorphs to order. We too are
interested in polymorphism (see e.g. Refs. 6–8), whereby we
have generally depended on serendipity and the recognition of
18
Abraham and Medforth studied the conformational equilibria
of piperidine and some of its simple derivatives, including 1, with
a bulky cobalt(III)-porphyrin shift reagent that binds specifically
to the amine nitrogen atom. The spin-spin coupling constants in
1
the H NMR spectra of the resulting complexes revealed that the
equatorial position of the OH group in 1 was, as expected,
favoured over the axial position, whereas the amine proton
necessarily takes the axial position at nitrogen in both
conformers, to allow the central atom of the shift reagent to
coordinate at the equatorial position. This conformational
5
This journal is © The Royal Society of Chemistry [year]
[journal], [year], [vol], 00–00 | 1

CrystEngComm
Page 2 of 11
View Article Online
DOI: 10.1039/C4CE02477J
equilibrium of 1 has also been analysed with the J-value
The
CO2
adduct
4
(4-hydroxypiperidinium
4-
19,20
method,
with similar conclusions; the equatorial conformer
hydroxypiperidine-1-carboxylate) was reproducibly obtained as
single crystals in the form of colourless blocks by overlaying a
with respect to the OH group is some 0.8 kcal/mol more stable
than the axial (perhaps surprisingly, this preference is reversed 60 solution of 1 (150 mg, 1.48 mmol) in 2 ml pyridine with
5
for the corresponding piperidinium cation).
diisopropyl ether. It is well known that some amines can absorb
CO2 from the air to form carbamates, especially when
crystallization tubes stand undisturbed for months. Structures of
The important conclusion for this paper is that the various
conformations and configurations of substituted piperidines are
generally closely similar in energy. We note also, trivially, that
commercially available piperidine derivatives are never listed by
suppliers as displaying a particular configuration, again implying
a delicate energy balance between axial and equatorial forms,
with facile interconversion at least in solution; despite the
interconverting configurations, it is clear that only one compound
is involved.
22
2-4 are presented briefly for the sake of completeness.
65
1
0
15
Experimental
Crystal growth
Our studies of amine complexes of gold(I) have shown that the
use of undiluted liquid amines is generally necessary for
9
successful syntheses. However, because we had access to a
20
25
30
35
40
45
50
55
sample of 4-hydroxypiperidine 1, a solid at room temperature, we
decided to attempt the synthesis of the corresponding AuCl
complex. As we had feared, all crystals thus obtained proved to
be the unchanged starting material 1, but optically there seemed
to be two different habits. Thus, a mixture of thtAuCl (40 mg,
0.125 mmol) in 1 ml dichloromethane and 1 (300 mg, 2.96 mmol)
in 1.5 ml dichloromethane was overlayered with various
precipitants in small test-tubes (diameter ca. 5 mm). With pentane
or heptane, form 1t (later shown to be tetragonal) was obtained as
colourless needles. Crystals of form 1o (orthorhombic) were
obtained with petrol ether as colourless blocks. Both crystal
forms are shown in Fig. 1.
Frustratingly, the production of these polymorphs via the
chemically passive gold "precursor" proved not to be generally
reproducible. Form 1o was obtained on just one further occasion
by overlaying a solution of 1 (150 mg, 1.48 mmol) in 2 ml 3-
picoline with pentane. After this initial brief success, it was never
again obtained pure, but only as the occasional crystal in samples
consisting mostly of 1t. The polymorphs are thus concomitant,
but the yields of 1o become almost vanishingly small. This is not
an unusual observation, and more extreme cases correspond to
Figure 1. Crystals of the two forms of compound 1. Above: orthorhombic
form 1o, part of the crystal actually measured (max. linear dimension ca.
70
0
.4 mm). Below: tetragonal form 1t as grown in the test-tube with
21
the known phenomenon of "disappearing polymorphs"; even if
o does not entirely disappear, it does conceal itself very
pentane precipitant (total width ca. 2 mm).
1
effectively.
The corresponding hydrochloride 2 (4-hydroxypiperidinium
chloride) was deliberately prepared for attempts to obtain
crystalline 1o by neutralisation. Single crystals of 2 in the form of
colourless blocks were obtained by drying a solution of 1 (300
mg, 2.96 mmol) in 1.5 ml water and 0.5 ml conc. HCl over solid
sodium hydroxide in a desiccator.
Further derivatives of 1 were obtained by chance from other
crystallization attempts. Bis(4-hydroxypiperidinyl)methane 3,
presumably resulting from nucleophilic displacement of chloride
from dichloromethane, was obtained as single crystals in the form
of colourless blocks by overlaying a mixture of thtAuCl (40 mg,
0.125 mmol) in 1 ml dichloromethane and 1 (300 mg, 2.96 mmol)
in 1.5 ml dichloromethane with petrol ether.
2
| Journal Name, [year], [vol], 00–00
This journal is © The Royal Society of Chemistry [year]

Page 3 of 11
CrystEngComm
View Article Online
DOI: 10.1039/C4CE02477J
4
0
DSC investigations were performed with the DSC 204/1/F from
Netzsch and the DSC 1 Star System with STARe Excellence
Software from Mettler-Toledo AG. The measurements were
performed in Al pans with different heating rates under nitrogen.
The instruments were calibrated using standard reference
materials.
45
Thermomicroscopy
Thermomicroscopic measurements were performed using a hot
stage FP82 from Mettler and a microscope (BX60) from
Olympus using the Analysis software package from Mettler. The
calibration was checked using benzoic acid.
50
NMR-Spectroscopy
1
H
NMR (399.83 MHz) spectra were recorded in
tetrahydrofuran-d8 (THF-d ) solution on a Bruker Avance
8
DRX400 spectrometer. Internal chemical shift standards were
1
X-ray crystallography
55 tetramethylsilane (TMS, for H, δ = 0.00 ppm) or, in the F1
H
dimension of an H,C-HSQC spectrum, the solvent signal (C-2/-3
Details of intensity measurements and refinements are given in
Table 1. Crystals were mounted in inert oil on glass fibres. Data
13
of THF-d for C, δ = 25.31 ppm). For the low temperature
8
C
1
measurements of 1, H NMR spectra were acquired between 24
°C and –104 °C. The temperature display of the spectrometer was
calibrated against the standard methanol sample (4 % methanol in
5
0
5
0
5
were measured with an Oxford Diffraction Xcalibur E
23
diffractometer using monochromated Mo Kα radiation; multi-
scan absorption corrections were performed. The structures were
solved with direct methods using SHELXS-97, and structure
6
0
26
methanol-d4). H,H-COSY, H,H-NOESY and H,C-HSQC were
acquired at –84 °C to verify signal assignment. Chemical
2
refinement was performed with full-matrix least-squares on F
1
2
4
exchange in the H NMR spectrum of 1 between the conformers
1
1
2
2
using SHELXL-97. NH and OH hydrogens were identified
clearly in difference maps and refined freely (for one exception
see below), other hydrogens using a riding model. Molecular
with an equatorial and an axial OH group was observed by means
of a H,H-NOESY spectrum at –84 °C, a mixing time of 500 ms
was used. Standard Bruker pulse programs were used throughout.
65
25
graphics were prepared with XP.
Special features and exceptions: For 1t, 1o and 3, which all
crystallize in non-centrosymmetric but achiral space groups, the
anomalous dispersion was negligible and Friedel opposite
reflections were therefore merged; the Flack parameters are thus
meaningless. Compound 4 was a non-merohedral twin generated
by 180° rotation about c* and was refined using the HKLF5
method; the relative volume of the smaller component refined
to 0.4150(7). Because equivalent reflections are merged and
reflections from both twin components are present, the number of
reflections should be interpreted with caution. For structure 4,
OH hydrogens were included using rigid OH groups allowed to
rotate but not tip ("AFIX 147").
Results and discussion
Crystal structures of 1t and 1o
The molecule of 1t is shown in Fig. 2. The ring shows the
typical chair form and the OH substituent is, as expected,
equatorially oriented (this is also true for all other structures
presented here except 4, see below). More surprising perhaps is
the axial orientation of the N-bonded hydrogen; it seems that
packing effects, presumably associated with the hydrogen bond
pattern, override the slight preference for the equatorial
orientation observed in solution. One might speculate that,
7
7
8
8
9
0
5
0
5
0
24
1
8
similarly to the behaviour of the shift reagent mentioned above,
X-ray powder diffraction
the donor molecule acts as a bulky group and thus preferentially
occupies the equatorial site at the acceptor nitrogen. A search of
X-Ray powder diffraction measurements were performed using
a STOE STADI P transmission powder diffractometer from
STOE & CIE with an Mythen K1 detector using CuKα1 radiation
(λ = 1.540598 Å).
27
28
the Cambridge Database provided just five structures of
mono-4-substituted piperidines with no further substituents; in all
cases the 4-substituent was equatorial (torsion angle C–C–C–X
ca. 180°), but the configuration of the NH group varied, being
equatorial in five fragments and axial (absolute torsion angle C–
C–N–H ca. 60°) in two. This implies that both configurations are
of similar energy, consistent with the various NMR studies
discussed in this paper.
3
0
Differential thermal analysis and thermogravimetry
DTA-TG measurements were performed in Al O3 crucibles
2
using an STA-409CD thermobalance from Netzsch. Several
measurements under nitrogen (purity 5.0) and air atmospheres
with different heating rates were performed. All measurements
employed a flow rate of 75 mL/min and were corrected for
buoyancy and current effects. The instruments were calibrated
using standard reference materials.
The molecular packing of 1t involves two classical hydrogen
bonds (Table 2). The interaction O1–H02LN1 links the
molecules to form chains parallel to the a and b axes (Fig. 3), and
these chains are linked in the c direction by N1–H01LO1.
3
5
Differential scanning calorimetry
This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00–00 | 3

CrystEngComm
Page 4 of 11
View Article Online
DOI: 10.1039/C4CE02477J
heterosubstituents. Mechanisms of solid-state phase changes have
not often been investigated for molecular compounds; it might be
expected that a complete rearrangement of the hydrogen bond
system for 1 (because the OH donor would have to change from
an equatorial acceptor site at nitrogen in 1t to an axial acceptor
site in 1o) would be unfavourable. Such arguments are
necessarily speculative in the absence of detailed solid-state
energy calculations. We attempted in an earlier publication to
give a tentative explanation for phase changes on cooling in two
urea-lutidine adducts, in which the degree of rearrangement was
3
4
4
5
5
6
5
0
5
0
5
0
8
however limited.
Figure 2. The molecule of 1t in the crystal. Ellipsoids correspond to 50%
probability levels.
Further studies were therefore carried out to address three
problems: (1) Is it possible to obtain a reliable supply of the less
common form 1o? (2) Can the solid state forms 1o and 1t be
interconverted? (3) Do the different configurations interconvert
readily in solution, as would be expected?
Corresponding studies are presented below. These were made
more difficult by the highly hygroscopic nature of 1, and also by
its tendency to form amorphous or gelatinous phases. One
possibility of accessing phase 1o seemed to be available via the
aqueous solution that was formed on exposure of 1 to moist air,
but the solutions thus formed deposited only crystals of 1t and
small amounts of a yellow powder that we were unable to
identify. Another possibility was to neutralise the hydrochloride
2, but storing a solution of this above sodium hydroxide in a
desiccator only yielded crystalline 2. Many attempts to grow
single crystals of 1o from various solvent mixtures only yielded
(
apart from 1t) crystals of the methylene-bridged "dimer" 3 and
the carbamate 4. The structures of 2, 3 and 4 are presented briefly
in the next section.
5
Figure 3. Packing diagram of 1t with view direction parallel to the c axis.
Classical hydrogen bonds are drawn as thick dashed lines. Nitrogen atoms
are drawn blue and oxygens red.
The molecular structure of 1o is shown in Fig. 4. In contrast to
1
t, the N-bonded hydrogen is equatorial; we stress that the OH
10
15
20
25
30
group is also equatorial (as in 1t), so that 1o and 1t are not related
by a simple conformational change, but by the "umbrella"
inversion at nitrogen. The molecular packing of 1o involves two
classical hydrogen bonds (Table 3), which qualitatively are the
same as for 1t. The interaction O1–H02LN1 links the molecules
Figure 4. The molecule of 1o in the crystal. Ellipsoids correspond to 50%
probability levels.
to form helical chains parallel to [01
H01LO1 forms helical chains parallel to [101]. The simplest
resulting pattern is a layer parallel to ( 11) (Fig. 5), but three-
dimensional crosslinking occurs via symmetry-equivalent chains
1], and the interaction N1–
1
(
packing diagrams in Fdd2 are seldom simple).
Having established that there are two crystal forms of 1, the
question arises whether these should be described as polymorphs.
Strictly speaking they represent different configurations and thus
different isomers, at least in the "frozen" form in the solid state,
but interconversion should occur via a modest energy barrier, and
thus be feasible at least in solution (or in the melt). It should
however be remembered that other factors may be important in
1
6
the solid state. An early publication noted that the equatorial
form of unsubstituted piperidine is some 0.4 kcal/mol more stable
than the axial form, but that in other derivatives this preference
could be affected by strong hydrogen bonding or further
4
| Journal Name, [year], [vol], 00–00
This journal is © The Royal Society of Chemistry [year]

Page 5 of 11
CrystEngComm
View Article Online
DOI: 10.1039/C4CE02477J
Figure 7. The molecule of methylene-bridged "dimer"
3 in the crystal.
Ellipsoids correspond to 50% probability levels. The molecule displays
non-crystallographic twofold symmetry (r.m.s. deviation 0.01 Å).
30
Figure 8. One of three independent formula units of carbamate 4 in the
crystal. Ellipsoids correspond to 50% probability levels. The dashed line
represents a classical hydrogen bond.
Polymorphism
of
compound
investigations of forms 1t and 1o
1:
Physico-chemical
35
The commercially available material (obtained from Fluka and
Acros) was investigated by X-ray powder diffraction;
Figure 5. Packing diagram of 1o with view direction perpendicular to the
a
plane (111). Classical hydrogen bonds are drawn as thick dashed lines.
The two types of helical chain mentioned in the text run horizontally (two
chains) and vertically (three chains).
comparison of the experimental pattern with that calculated for
both forms from single crystal data proves that it consists of form
5
1
t exclusively (Fig. S1; all Figs. and Tables denoted "S" are to be
Crystal structures of 4-hydroxypiperidine derivatives 2, 3 and
4
40
found in the Supporting Information). As noted above, only a few
crystals of form 1o were available from one of the crystallization
experiments. In this batch these crystals were embedded in large
amount of a colorless, gelatinous residue, which also contained
crystals of form 1t. Therefore, most of the following experiments
The structures of the hydrochloride 2, the methylene-bridged
"dimer" 3 and the carbamate 4 are shown in Figs. 6, 7 and 8.
1
1
2
0
5
0
Structures 2 and 3 display no unusual features, except perhaps
that the OH group in 2 is equatorial, in contrast to NMR evidence 45 were performed in the hope of accessing larger and if possible
that the axial form should be slightly more stable (see discussion
elsewhere in this paper); presumably hydrogen bonding effects
are responsible, but this is difficult to prove. The structure of 4 is
complex, involving three independent formula units; in one of the
purer samples of form 1o. Because compound 1 is extremely
hygroscopic, the material was pre-dried at 60°C for all
quantitative measurements.
If form 1t is investigated by differential scanning calorimetry
cations and two of the anions, the OH group is axial. All the 50 (DSC) at 1°C/min, an endothermic peak is observed at an onset
carboxylate groups are approximately coplanar with the adjacent
ring C–N–C moieties; whereby the N–Ccarboxylate bonds are ca. 0.1
Å shorter than the N–Cring bonds.
temperature T of 88.9°C, corresponding to the melting point of
this form (Fig. 9). There is no indication of any polymorphic or
other solid-to-solid transition before melting. On cooling,
o
The packing diagrams for 2 and 3, involving classical hydrogen
bonds, are presented in the Supplementary Material.
solidification is observed at T = 27.5°C. The enthalpy for the
o
55 crystallization in this DSC run is significantly lower that the heat
of fusion in the first heating cycle, which indicates that some of
the material has been transformed into an amorphous phase. If the
solidified melt is reheated, the melting peak becomes broader and
the heat of fusion is much lower than for the first heating. On
6
6
7
0
5
0
cooling, strong supercooling is again observed.
If a similar measurement is performed at 3°C/min a similar
behavior is observed in the first heating cycle but on cooling, no
crystallization takes place. If the solidified melt is heated and
cooled again, no signals are detected in the DSC curve, which
shows that no crystallization of the amorphous phase on cooling
is observed (Fig. S2). To rule out decomposition during melting,
the IR spectra of the pristine material and the amorphous residue
obtained after cooling were compared but no differences were
found (Fig. S3). Additional DSC runs at higher heating rates are
similar and in all cases amorphous residues are obtained. The
enthalpy of fusion was determined from 4 DSC runs at 10°C/min
to be 18.3 kJ/mol (Table S1).
Figure 6. The formula unit of hydrochloride
correspond to 50% probability levels.
2 in the crystal. Ellipsoids
2
5
This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00–00 | 5

CrystEngComm
Page 6 of 11
View Article Online
DOI: 10.1039/C4CE02477J
T =89.5°C
DTG
p
endo
T =88.9°C
o
ꢀH=18.1 kJ/mol
1. heating
ꢀ
H= -13.2 kJ/mol
1. cooling
2. heating
Tp=88.5°C
T =86.3°C
TP=217 °C
To=27.7°C
o
T =27.8°C
p
ꢀH=15.8 kJ/mol
endo
2. cooling
ꢀm=98 %
ꢀH= -12.7 kJ/mol
T =30.9 °C
o
T =31.5°C
p
TG
TP=224 °C
TP=99 °C
T
3
0
40
50
60
Temperature °C
70
80
90
100
m
DTA
T
Figure 9. DSC heating and cooling curves for form
I at 1°C/min; the
peak (T
in kJ/mol.
p
) and onset (T
o
) temperatures are given in °C and the enthalpy ꢀH
m
6
5
0
5
0
In our studies of the polymorphism of trimethylthiourea, we
found that an initially elusive second polymorph could be
obtained reproducibly by solidification of the melt. In a further
DSC run of 1 at 1°C/min, the crystalline residue as obtained after
cooling the melt was investigated by X-ray powder diffraction,
which proves that form 1t is obtained exclusively (Fig. 10).
Surprisingly, this residue is of very good crystallinity and there
are no indications of any amorphous content. Because no
5
0
100 150 200 250 300
Temperature /°C
2
5
Figure 11. DTA, TG and DTG curve for form 1t at 4°C/min in a nitrogen
atomosphere
1
1
2
To test whether solvates are accessible that might transform
into form 1o on solvent removal, a saturated solution of form 1t
crystallization takes place at faster cooling rates, there is no 30 with an additional amount of solid was stirred in a number of
chance to obtain form 1o or additional forms by kinetic control.
Additional measurements on form 1t using simultaneous
differential thermoanalysis and thermogravimetry show no
significant mass loss before melting and thus, sublimation of
some amount of this form can be excluded under these conditions 35 of 1 can be added to the solution because gelatinous systems are
Fig. 11). There are also no indications for significant amounts of
solvents for one week and afterwards all residues were
investigated by X-ray powder diffraction. In some solvents
crystalline residues are difficult to obtain because always
amorphous residues are formed. In several cases a large amount
(
formed. In practically no solvents does any transformation take
place and therefore, form 1t is always re-isolated (Fig. S4).
Therefore, form 1t can be concluded to represent the
thermodynamically stable form at room temperature. However, in
2-butanol a powder pattern is obtained that indicates the presence
of an additional crystalline phase. If this residue is investigated
water in the original material. After melting, the complete sample
vaporizes on heating to 300°C.
4
4
5
5
0
5
0
5
by
simultaneous
differential
thermoanalysis
and
thermogravimetry a mass loss of about 5.5% is observed before
melting, indicating the presence of a solvate (Fig. S5). If the
residue formed after solvent removal is investigated by XRPD, it
is shown that modification 1t has formed exclusively (Fig. S6).
If form 1t is stirred in dichloromethane, a residue is formed that
exhibits a powder pattern also different from that of 1t. Further
analysis shows that this pattern is identical to that calculated for
the hydrochloride 2 (Fig. 12). Curiously, only form 1t is obtained
if the commercial compound is stirred in concentrated
hydrochloric acid. This observation is difficult to understand but
we have repeated these experiments several times with consistent
results. If the hydrochloride is investigated by DTA-TG
measurements, it starts to decompose at about 170°C and on
further heating a smooth decrease of the sample mass is observed
5
10
15
20
25
30
35
40
2
-Theta / °
Figure 10. Experimental X-ray powder pattern of the solidified melt of
form 1t (top) and calculated powder pattern for form 1t (bottom).
(
Fig. S7). There is no distinct step corresponding to the removal
of HCl and therefore, there is no chance of obtaining a pure form
of 1.
6
| Journal Name, [year], [vol], 00–00
This journal is © The Royal Society of Chemistry [year]

Page 7 of 11
CrystEngComm
View Article Online
DOI: 10.1039/C4CE02477J
45
5
10
15
20
25
30
35
40
2
-Theta / °
Figure 12. Experimental X-ray powder pattern of the residue obtained
from dichloromethane (top) and calculated powder pattern for the
hydrochloride
2 (bottom).
5
0
5
0
5
0
5
0
Finally, we also performed several crystallization experiments
under kinetic control, e.g. by rapid solvent removal, but in none
of these experiments were even traces of form 1o obtained.
To obtain at least some information on the thermodynamic
relation between 1t and 1o, we tried to measure a DSC curve of
some crystals of form 1o, which were selected by hand from the
batch consisting of both forms. Unfortunately, because of the low
sample mass only a very noisy DSC curve was obtained, which
consisted of a broad maximum at a temperature where melting is
expected. From this experiment one cannot decide which of the
two forms melts at the higher temperature. Moreover, because
these crystals are always coated with the gelatinous residue
associated with this batch, the exact sample mass was unknown
and thus, no reliable values for the heat of fusion were obtained.
Therefore, with the last crystals available of form 1o, two
thermomicroscopic measurements were performed. Beforehand,
the identity of these crystals was confirmed by single crystal
structure analysis, and only afterwards were they prepared for the
measurement.
On heating both forms at 1°C/min, the crystal of form 1t
became continuously smaller and at about 86°C it disappeared
completely (Fig. S8). In contrast, the crystal of form 1o melted at
about 89°C, which is in the range observed for the melting of
form 1t in the DSC experiments. This experiment indicates that
form 1t sublimes under these conditions, which might be possible
because a relatively low heating rate was used and, in contrast to
the DSC measurements, the sample was not enclosed. The fact
that the crystal of form 1o is still present after sublimation of
form 1t does not mean that it is more stable, because it is
embedded in the gelatinous residue.
1
1
2
2
3
3
4
5
0
Figure 13. Microscopic images at various temperatures (36.1, 87.5, 90.5,
9
1.5, 92.5, 93.5°C) obtained by thermomicroscopy with form 1t (crystal
on left) and form 1o (crystal on right).
Solution NMR studies of compound 1
5
6
6
7
5
0
5
0
We wished to find out whether the distinctive crystallographic
features of forms 1o and 1t are in any way reflected in the
1
13
solution H and C NMR spectra. Hence we undertook NMR
8
experiments at low temperature in THF-d solution. As 1 is very
hygroscopic, however, we were not able to obtain a sample free
of water. Proton exchange between water and the amine and
hydroxyl groups, as well as signal overlap with the water signal,
1
considerably complicated the H NMR spectra, even at low
temperature where exchange is slow. As already known, the
conformer with an equatorial hydroxyl group dominates at room
1
temperature, as can be straightforwardly deduced from the H
NMR splitting patterns and coupling constants. Lowering the
temperature decelerates ring inversion. Between –53 °C and –
1
04°C a second set of signals can be observed, corresponding to
the conformer with an axial hydroxyl group, as expected. Lower
temperatures that could have furnished more information about
nitrogen inversion were unfortunately not accessible to us,
irrespective of the solvent used. Within the temperature range
studied (room temperature to –104 °C) no NMR-spectroscopic
peculiarities were observed.
To avoid sublimation, a second measurement was performed at
3
°C/min. No sublimation was observed, and it is clear that form
1
o melts at significantly lower temperatures than 1t (Fig. 13).
75
Conclusions
This experiment proves that form 1t is also more stable up to the
melting point. Moreover, because the density of form 1t (1.149
Compound 1 presents an unusual case in which two different
configurational isomers, axial or equatorial with respect to the
NH group, are observed in the two crystalline forms 1t and 1o
respectively, where the configurations are "frozen" in the solid
state; the hydrogen bonding patterns are necessarily different. To
assess the stereochemical relevance of this observation, we now
3
3
g/cm ) is significantly higher than that of 1o (1.138 g/cm ) it can
29
also be assumed that form 1t is more stable at low temperatures.
In this case it is highly likely that 1t is the thermodynamically
stable form over the complete temperature range and that both
forms are thus monotropically related.
80
This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00–00 | 7

CrystEngComm
Page 8 of 11
View Article Online
DOI: 10.1039/C4CE02477J
return to the Joad principle as mentioned in the Introduction; how
exactly do we define the stereochemical terms involved?
The standard definition states that conformational isomers (or
conformers), in contrast to configurational isomers, differ only by
rotations about single bonds and thus are readily interconvertible
Further Supplementary Material to this paper includes packing diagrams
for structures 2, 3 and 4 and additional Figures from the physico-
chemical investigations of 1.
6
6
7
7
8
8
0
5
0
5
0
5
1
(a) D. Braga and F. Grepioni, Making Crystals by Design: Methods,
Techniques and Applications, Wiley-VCH Verlag GmbH & Co.
KGaA, Weinheim, 2007; (b) R. Hilfiker, Polymorphism in the
Pharmaceutical Industry, WILEY-VCH, 2006; (c) W. C. McCrone,
5
0
5
0
5
0
(
and represent the same substance). One standard text on
30
stereochemistry, however, reminds us that "the distinction
between configuration and conformation is subtle and K not
universally agreed upon" and takes the definition further:
"Conformation can be changed by rapid rotation around single
bonds (and in the definition of some, by rapid inversion at
trigonal pyramidal centres)" (our italics). Following this
principle, the two forms of molecule 1, although strictly speaking
showing a configurational difference, are "only" conformers
because they readily interconvert in solution, even at low
temperature, by inversion at nitrogen. Thus the two crystalline
forms certainly represent different polymorphs of 1, but it would
be unsafe and probably misleading to describe them as
Polymorphism in Physics and Chemistry of the Organic Solid State
,
ed. D. Fox, M. M. Labes and A. Weissenberg, Interscience, New
York, vol. II, pp. 725–767, 1965; (d) G. R. Desiraju, Crystal
Engineering, Mat. Sci. Monogr., Elsevier, Amsterdam, 1989; (e) J.
Bernstein, Polymorphism in Molecular Crystals, Clarendon Press,
Oxford, 2002; (f) J. Bernstein; R. Davey and O. Henck, Angew.
Chem. Int. Ed. 1999, 38, 3440–3461; (g) G. Brittain, Polymorphism
in Pharmaceutical Solids, Vol. 23, Marcel Dekker Inc., New York,
1
1
2
2
3
1999; (h) B. Moulton and M. J. Zaworotko, Chem. Rev., 2001, 101
,
1629–1658.
2
3
C. Butterhof, T. Martin, P. Ector, D. Zahn, P. Niemietz, J. Senker, C.
Näther and J. Breu, Cryst. Growth Des., 2013, 12, 5365–5372; J.
Thun, L. Seyfarth; C. Butterhof; J. Senker, R. E. Dinnebier and J.
Breu, Cryst. Growth Des., 2009,
9, 2435–2441, and references
"configurational polymorphs". Indeed, this category of
therein.
polymorphs can then never, by definition, exist; configurational
isomers that do not readily interconvert are different compounds,
whereas those with facile interconversion are redefined as
conformers.
This is however a definition of convenience, and furthermore is
incomplete; what exactly does "readily" or "facile" mean? How
high must the energy barrier become before the conformers
become separable configurational isomers and the polymorphs
thus become different compounds, and at what temperature will
the separation be possible? Is it even sensible to tie the definition
to this energy value? The studies reported here underline the
importance of these questions, but exact answers may not be
possible and borderline cases will probably be discovered.
A. J. Cruz-Cabeza and J. Bernstein, Chem. Rev., 2014, 114, 2170–
2191.
4
5
e.g. Ref. 1(b), p. 22.
R. K. R. Jetti, R. Boese, J. A. R. P. Sarma, L. S. Reddy, P.
Vishweshwar and G. R. Desiraju, Angew. Chem., 2003, 115, 2008–
2012; Angew. Chem. Int. Ed., 2003, 42, 1963–1967
6
7
C. Näther, I. Jess, P. G. Jones, C. Taouss and N. Teschmit, Cryst.
Growth Des., 2013, 13, 1676–1684.
C. Näther, C. Döring, I. Jess, P. G. Jones and C. Taouss, Acta Cryst.
013, B69, 70–76.
C. Taouss and P. G. Jones, CrystEngComm, 2014, 16, 5695–5704.
,
2
8
9
90
e.g. C. Döring and P. G. Jones, Z. Naturforsch., 2013, 68b, 474–492;
C. Döring and P. G. Jones, Z. Naturforsch., 2014, 69b, 1315–1320.
0 e.g. P. G. Jones, C. Taouss, N. Teschmit and L. Thomas, Acta Cryst.,
1
2
013, B69, 405–413; C. Taouss, L. Thomas and P. G. Jones,
CrystEngComm, 2013, 15, 6829–6836.
11 H. Booth and D. V. Griffiths, J. Chem. Soc., Perkin Trans. 2, 1973,
42–844.
H. Booth and J. R. Everett, J. Chem. Soc., Chem. Commun., 1979,
9
5
8
1
1
1
2
3
4
Dedication
3
4–35.
J. M. Bailey, H. Booth, H. A. R. Y. Al-Shirayda and M. L. Trimble,
J. Chem. Soc., Perkin Trans. 2, 1984, 737–743.
E. L. Eliel, D. Kandasamy, C. Yen and K. D. Hargrave, J. Am.
Chem. Soc., 1980, 102, 3698–3707.
35
This paper is dedicated to the memory of Dr. Frank Allen, a much
respected colleague and friend.
1
1
1
1
1
1
00
Acknowledgement
15 (a) J. B. Lambert, R. G. Keske, R. E. Carhart and A. P. Jovanovich,
J. Am. Chem. Soc., 1967, 89, 3761–3767. (b) Calculation in Ref. 17.
We thank Prof. Dr. Wolfgang Bensch for granting access to his
powder diffractometer. Three referees assisted us by providing
challenging and constructive criticism that enabled us to improve
the manuscript significantly.
05 16 I. D. Blackburne, A. R. Katritzky and Y. Takeuchi, Acc. Chem. Res.,
1975, , 300–306.
F. A. L. Anet and I. Yavari, J. Am. Chem. Soc., 1977, 99, 2794–
6
1
1
7
8
40
2
796.
(a) R. J. Abraham and C. J. Medforth, J. Chem. Soc., Chem.
Commun., 1987, 1637–1638; (b) R. J. Abraham, C. J. Medforth,
Magn. Reson. Chem., 1988, 26, 334–344.
10
15
20
25
Notes and references
1
2
9
0
Y. Terui and K. Tori, J. Chem. Soc., Perkin Trans. 2, 1975, 127–
a
Institut für Anorganische und Analytische Chemie, Technische
1
33.
R. J. Abraham, C. J. Medforth and P. E. Smith, J. Comput.-Aided
Mol. Design, 1991, , 205–212.
Universität Braunschweig, Postfach 3329, D-38023 Braunschweig,
Germany. Fax: +49 531 5387; Tel: +49 531 5382; E-mail: p.jones@tu-
bs.de
45
50
55
5
2
2
2
2
1
2
3
4
J. D. Dunitz and J. Bernstein, Acc. Chem. Res., 1995, 28, 193–200.
M. Freytag and P. G. Jones, Acta Cryst., 1999, C55, 1874–1877.
Agilent, CrysAlis PRO. Agilent Ltd., Yarnton, England, 2014.
(a) G. M. Sheldrick, Acta Cryst., 2008, A64, 112–122; (b) SHELXL-
1997, a Program for refining Crystal Structures, G. M. Sheldrick,
University of Göttingen, Germany, 1997.
b
Institut für Anorganische Chemie, Christian-Albrechts-Universität zu
Kiel, Max-Eyth-Str. 2, D-24118 Kiel, Germany
c
NMR-Laboratorium der Chemischen Institute, Technische Universität
Braunschweig, Postfach 3329, D-38023 Braunschweig, Germany
†
1
CCDC-1037209 (1t), -1037210 (1o), -1037211 (2), -1037212 (3), -
037213 (4) contain the supplementary crystallographic data for this
2
2
2
5
6
7
Siemens XP, version 5.03. Siemens Analytical X–Ray Instruments,
Madison, Wisconsin, U.S.A, 1994.
Bruker AM Series User Manual. Bruker Analytische Messtechnik,
Rheinstetten, Germany, 1987.
paper. These data can be obtained free of charge from the Cambridge
Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/.
CCDC Version 5.35; F. H. Allen, Acta Cryst., 2002, B58, 380–388.
8
| Journal Name, [year], [vol], 00–00
This journal is © The Royal Society of Chemistry [year]

Page 9 of 11
CrystEngComm
View Article Online
DOI: 10.1039/C4CE02477J
2
8
9
CCDC refcodes ICANIK, JARCOV, JARCUB, LIPXOX and
POWZAD. We believe that the NH hydrogen position of POWZAD
may be in error, and it was not considered in our discussion.
30 E. L. Eliel and S. H. Wilen, Stereochemistry of Organic Compounds
,
Wiley-Interscience, 1994; see in particular p. 18 and p. 1195.
2
(a) A. I. Kitaigorodski, Organic Chemical Crystallography
Consultants Bureau, New York, U.S.A., 1961; (b) A. Burger and R.
Ramberger, Mikrochim. Acta, 1979, , 259–271.
,
10
5
2
Table 1. Experimental details of the structure determinations
Solvate
1t
1o
2
3
4
+
+
–
(C
(
H12NO)
5
–
Chemical formula
C H NO
C H NO
(C H NO) Cl
C H N O
2
5
11
5
11
5
12
11 22
2
C H NO )
6 10 3
M
r
101.15
101.15
137.61
214.31
227.28
Crystal system,
space group
orthorhombic,
Fdd2
orthorhombic,
Pca21
monoclinic, P2 /n
tetragonal, P
100
4
2 c
1
1
triclinic, P1
Temperature (K)
a (Å)
100
100
100
100
12.0295(4)
12.0295(4)
8.0824(4)
90
21.3459(7)
18.6413(6)
5.93539(19)
90
5.3800(2)
20.4924(11)
6.4026(3)
90
11.5417(12)
10.4947(10)
11.9000(11)
16.836(2)
76.943(10)
81.326(10)
68.873(9)
1905.0
b (Å)
5.0515(5)
c (Å)
21.0649(15)
α
β
γ
(°)
(°)
90
90
90
101.745(5)
90
90
(°)
90
90
90
3
V(Å )
1169.59
1.149
2361.78
1.138
16
691.10
1.323
1228.13
1.159
4
-
3
dcalc (g/cm )
1.288
Z
8
4
6
Radiation,
wavelength
Mo Kα, λ =
0.71073 Å
Mo Kα, λ =
0.71073 Å
Mo Kα, λ =
0.71073 Å
Mo Kα, λ =
0.71073 Å
Mo Kα, λ =
0.71073 Å
F(000)
448
896
296
472
804
–
1
ꢁ
(mm )
0.08
0.08
0.46
0.08
0.10
Crystal size (mm)
Transmissions
0.2 × 0.2 × 0.1
No abs. corr.
53.5
0.35 × 0.35 × 0.35 0.35 × 0.35 × 0.15 0.2 × 0.2 × 0.1
0.25 × 0.15 × 0.08
0.98–1.00
51.2
No abs. corr.
61.6
0.98–1.00.
61.6
0.96–1.00
56.6
2
θmax (°)
No. of measured and
independent
9202, 729
20678, 987
18001, 2057
23143, 1565
10054
reflections
Completeness
100%
0.040
99.3% to 2θ 61°
98.9% to 2θ 61°
99.9%
0.087
99.9%
–
R
0.026
0.040
int
2
wR(F ) all refl., R1
2
0.072, 0.031, 1.07 0.069, 0.027, 1.06 0.069, 0.030, 1.05 0.088, 0.040, 1.04 0.114, 0.054, 0.91
[
F > 4σ(F)], S(F )
No. of parameters /
restraints
7
2 / 0
72 / 1
85 / 0
144 / 2
491 / 21
–3
ꢀ
ρmax,min (e Å )
0.18, –0.15
0.24, –0.16
0.37, –0.22
0.18, –0.17
0.52, –0.33
This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00–00 | 9

CrystEngComm
Page 10 of 11
View Article Online
DOI: 10.1039/C4CE02477J
Table 2. Details of hydrogen bonding [Å and °] for 1t a
D–H
L
A
d(D–H)
0.86(2)
0.90(2)
d(H
L
A)
d(D
L
A)
<(DHA)
i
1
2
O1–H02
L
L
N1
1.89(3)
2.17(2)
2.7464(18)
2.9794(19)
172(2)
ii
N1–H01
O1
150.1(18)
a
Symmetry transformations used to generate equivalent atoms: (i)
x–0.5, –
y+1.5, –
z+0.5; (ii)
y
, –
x+1, –
z.
5
Table 3. Details of hydrogen bonding [Å and °] for 1oa
D–H
L
A
d(D–H)
0.87(2)
d(H
L
A)
d(D
L
A)
<(DHA)
i
1
2
O1–H02
L
L
N1
1.88(2)
2.7461(13)
3.0079(14)
171.6(18)
167.1(18)
ii
N1–H01
O1
0.862(17)
2.161(18)
a
Symmetry transformation used to generate equivalent atoms: (i) –x+0.75, y+0.75, z–0.25; (ii) x–0.25, –y+0.25, z–0.25.
10
1
0
| Journal Name, [year], [vol], 00–00
This journal is © The Royal Society of Chemistry [year]

Page 11 of 11
CrystEngComm
View Article Online
DOI: 10.1039/C4CE02477J
4-Hydroxypiperidine 1 exists in two crystal forms, tetragonal 1t with axial NH and
orthorhombic 1o with equatorial NH. We describe the two crystal forms as configurational
polymorphs.