Unexpected polymorphism during a catalyzed mechanochemical Knoevenagel condensation

Article abstract of DOI:10.3762/bjoc.15.110

The transformation of a base-catalyzed, mechano-assisted Knoevenagel condensation of mono-fluorinated benzaldehyde derivatives (p-, m-, o-benzaldehyde) with malonodinitrile was investigated in situ and in real time. Upon milling, the para-substituted prod

Full text of DOI:10.3762/bjoc.15.110

Unexpected polymorphism during a catalyzed
mechanochemical Knoevenagel condensation
Sebastian Haferkamp1,2, Andrea Paul1, Adam A. L. Michalchuk1
and Franziska Emmerling*1,2
Full Research Paper
Open Access
Address:
BAM Federal Institute for Materials Research and Testing,
Richard-Willstaetter-Straße 11, 12489 Berlin, Germany and
Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2, 12489 Berlin,
Beilstein J. Org. Chem. 2019, 15, 1141–1148.
1
doi:10.3762/bjoc.15.110
2
Received: 10 February 2019
Accepted: 10 May 2019
Published: 21 May 2019
Germany
Email:
Franziska Emmerling* - franziska.emmerling@bam.de
This article is part of the thematic issue "Mechanochemistry II".
Guest Editor: J. G. Hernández
*
Corresponding author
Keywords:
© 2019 Haferkamp et al.; licensee Beilstein-Institut.
License and terms: see end of document.
ball milling; C–C coupling; in situ; mechanochemistry; multivariate
data analysis
Abstract
The transformation of a base-catalyzed, mechano-assisted Knoevenagel condensation of mono-fluorinated benzaldehyde deriva-
tives (p-, m-, o-benzaldehyde) with malonodinitrile was investigated in situ and in real time. Upon milling, the para-substituted
product was found to crystallize initially into two different polymorphic forms, depending on the quantity of catalyst used. For low
catalyst concentrations, a mechanically metastable phase (monoclinic) was initially formed, converting to the mechanically stable
phase (triclinic) upon further grinding. Instead, higher catalyst concentrations crystallize directly as the triclinic product. Inclusion
of catalyst in the final product, as evidenced by mass spectrometric analysis, suggests this complex polymorphic pathway may be
due to seeding effects. Multivariate analysis for the in situ Raman spectra supports this complex formation pathway, and offers a
new approach to monitoring multi-phase reactions during ball milling.
Introduction
Mechanochemistry offers a wide array of applications. It is used Despite the increasing use of mechanochemistry, there is still a
widely for synthesis of inorganic, metal-organic, and organic lack in understanding of the underlying processes involved
molecules and materials [1]. Interest in these methods stems during mechanically-facilitated reactions. This is particularly
largely from the fact that they are efficient and more environ- true of the potential role of transient polymorphic phases [8]
mentally friendly as compared to traditional approaches [2,3]. and seeding effects [9] in understanding reaction kinetics of
Mechanochemistry is a well-established method for the synthe- these processes. Early insight into formation pathways was pro-
sis of coordination polymers, the formation of cocrystals, and in vided ex situ, in which the mechanical treatment was stopped,
C–C coupling reactions [4-7].
and powder removed for analysis [10,11]. More recently,
1141

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
further detail has been gained by monitoring mechanochemical
transformations in real time, using in situ techniques [12,13].
The first in situ and real time study was performed by X-ray
powder diffraction (XRPD) to monitor transformations during a
milling process [12]. In situ studies allow novel insight into the
mechanism of a mechanochemical process, without changing
the reaction environment. It was subsequently demonstrated
how a combination of different in situ methods can provide
more thorough investigation of mechanochemical reaction
mechanisms [14-16]. Of particular benefit to synthetic reac-
tions, such as C–C bond formation [17,18], the use of Raman
spectroscopy is of great interest. The characteristic bands are
usually well separated, and the course of the reaction can be fol-
lowed easily. The advantage of Raman spectroscopy was
recently demonstrated [19], where its combination with XRPD
allowed monitoring of the mechanochemically catalyzed
Knoevenagel condensation in detail.
Scheme 1: Catalyzed mechanochemical Knoevenagel condensation
of fluorobenzaldehydes and malonodinitrile. The milling process is
symbolized by the three balls, proposed by Hanusa et al. [20].
Results and Discussion
The catalyzed Knoevenagel condensations of mono-fluorinated
benzaldehydes 1a–c with malonodinitrile (2) are depicted in
Scheme 1. In contrast to previous work, which reported the un-
catalyzed reaction [19], piperidine was used as a basic catalyst.
This was done as the inclusion of the base led to crystallization
of the corresponding products 3a–c during the milling process.
In contrast, crystallization during the base-free reaction re-
quired an ex situ aging step. Figure 1 shows the XRPD pattern
of the substrate 2 and the product 3a. Due to the liquid state of
the fluorinated benzaldehydes 1a–c no XRPD pattern could be
recorded.
Figure 1: Comparison of XRPD pattern of malonodinitrile (2) and
(p-fluorobenzylidene) malonodinitrile (3a). The patterns are baseline
corrected.
firmed that, independent of the amount of catalyst, the same
bulk product was formed. The intensity of the product powder
For the reaction of 1a with 2, the amount of catalyst was varied color decreased from a deep orange to white, with decreasing
and the reaction monitored in each case. XRPD analysis con- quantity of catalyst (Figure 2b). Importantly, none of the
Figure 2: a) XRPD pattern of (p-fluorobenzylidene)malonodinitrile (3a) direct after the synthesis with different amounts of catalyst. b) Color change of
the product 3a after the synthesis with different amounts of catalyst. The amount is reduced in the direction of the arrow.
1142

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
starting materials are colored, and the color of the product phase In contrast, reactions conducted with 2 µL were found to be
was found to vary systematically with increasing catalyst con- more complex (Figure 4). Shortly after the melting of 2, Bragg
centration. This coloring was observed throughout the bulk, reflections were observed and remained visible for a period of
which therefore suggested that the catalyst was incorporated approximately two minutes (Figure 4, gray box). These reflec-
into the structure of the solid product. Such effects have been tions, however, were found to correspond to the monoclinic
previously reported, and have the potential to seed nucleation of phase of the product (m3a) [22,23]. Phase m3a remained stable
polymorphic phases [21]. For that reason, mass spectrometric under mechanical treatment for a few minutes, before trans-
analyses of the powder phases were performed (Figure 3). The forming abruptly (over a period of 90 seconds) to the t3a phase.
relative intensity of the peak corresponding to piperidine Having observed the inclusion of catalyst into the final product
(
Figure 3 red box) decreases systematically with respect to that phase (see Figure 3), we suggest this change in polymorphic be-
of 3a (Figure 3 blue box). Hence, this suggests that the catalyst havior to result from a templating phenomenon, which domi-
is indeed present in the solid product. Further analysis is re- nates at higher concentrations of catalyst.
quired to understand the nature of catalyst incorporation, and
hence the origin of color.
To better observe this transformation pathway, the reaction
2 µL catalyst) was repeated at 30 Hz in order to extend the life-
(
The reactions containing between 40 µL and 5 µL catalyst show time of m3a. The in situ XRPD pattern and Raman spectra are
conversion of the substrates directly into the mechanically shown in Figure 4b and d, respectively. Decreasing the milling
stable triclinic product phase, according to real-time in situ frequency has a number of notable effects. First, the lifetime of
XRPD analysis (Supporting Information File 1, Figure S1). the initial molten/liquid phase was nearly doubled, suggesting
Consistent with previous reports [15], 2 melted at the begin- that mechanical treatment has an important effect on this largely
ning of the reaction and remained liquid or molten for the first solution-phase reaction. This may be a result of heating, me-
few minutes of milling (Supporting Information File 1). This is chanical activation of the fluid-phase molecules or differences
due to increased temperatures within the jar during milling in energy for nucleation. While further work is required in this
(
Tmelt(2) = 32 °C). Shortly after melting of 2, Bragg reflections area, it is clear that mechanical treatment can have notable in-
of a crystalline product phase were found to form (Figure 4). fluence on ‘solution-phase’ chemistry. Second, the lifetime of
Comparison to literature crystallographic data suggests this to the intermediate phase was extended. This allowed us to collect
be the triclinic phase (t3a). This phase remained stable upon higher resolution XRPD data by extracting a sample from the
continued milling.
milling vessel and confirming this intermediate to be the phase,
Figure 3: Mass spectra of the different products of 3a. Red: peak off the molecular ion [M + H]+ of piperidine (m/z ≈ 86). Blue: peak of the molecular
ion [M + H]+ of 3a (m/z ≈ 173).
1143

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
Figure 4: a) In situ XRPD pattern of the mechanochemical Knoevenagel condensation of 1a and 2 with 2 µL catalyst at 50 Hz. Gray box: intermedi-
ate phase. b) In situ XRPD pattern of the reaction with a milling frequency of 30 Hz. c) In situ Raman spectra of the mechanochemical Knoevenagel
condensation of 1a and 2 with 2 µL catalyst at 50 Hz. d) In situ Raman spectra of the reaction with a milling frequency of 30 Hz. Raman bands:
1576 cm−1 – C-C stretching vibration; 2234 cm−1 – C≡N stretching vibration of the product; 2266 cm−1 – C≡N stretching vibration of malonodinitrile.
m3a (Figure 5). Despite this extension, however, the subse- While the XRPD patterns clearly show the occurrence of a tran-
quent transformation m3a → t3a again occurs abruptly sition phase within the first minutes of the reaction, the Raman
(
ca. 120 seconds). This suggests that the transformation may spectra represent a superposition of all spectral components.
result from the accumulation of defects within m3a, or upon These spectral features often remain unchanged for polymor-
comminution of the product phase [24]. phic series. For this reason, it is difficult to employ reliably
1144

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
Figure 5: Comparison of XRPD patterns of both polymorphs of the product 3a. Red: triclinic polymorph t3a. Blue: monoclinic polymorph m3a. The
light XRPD pattern were calculated from single crystal diffraction data.
Raman spectroscopy for in situ mechanochemical data. Hence, 1a to 3a. Concerning PC1 in Figure 6b, the positive part corre-
despite its availability in many laboratories, in situ real-time sponds to spectral features of 1a and PMMA, and the negative
monitoring of mechanochemistry by Raman scattering has part indicates features of 3a. In contrast, the scores of PC2 are
remained underexplored. We therefore sought to identify a initially negative, rise to a peak value, and subsequently fall to
means to resolve this issue, using multivariate methods to zero (cf. Figure 6a). The spectral loadings of PC2 in Figure 6b
analyze the reaction. Principal component analysis (PCA) is a show a dominant positive structure, which resembles the nega-
bilinear modelling method [25,26] that helps to extract the main tive profile of PC1, albeit shifted slightly towards larger
information from multi-dimensional data (here a time series of wavenumbers. We therefore conclude that PC2 represents an in-
Raman spectra). The information contained within the original dependent (orthogonality among principal components) tran-
spectral variables is projected onto a small number of under- sient species. A comparison with the Raman spectra summa-
lying (‘latent’) variables, called the principal components (PC). rized in Figure S2 indicates that PC2 correlates well with the
Typically, the first PC covers the highest variance, the second formation of m3a. Identical results were found for the PCA of
and following PCs cover less information. All PCs are orthogo- in situ Raman spectra at 50 Hz milling frequency (cf. Support-
nal to one-another. Although the PCs themselves do not repre- ing Information File 1, Figure S2). However, the reduction of
sent quantitative data, they represent the underlying chemical or the milling frequency delays the onset of product formation.
physical processes. In order to understand the results of PCA, The kinetic profile obtained by PCA correlates well with XRPD
both scores and loadings must be scrutinized in parallel. Where- data (cf. Figure 6a), in which the transformation is monitored by
as the scores contain information regarding the samples, the the normalized intensities of Bragg reflections at 2θ = 13.87°
loadings provide information on the variables. High loading and 14.78°. While the XRPD patterns indicate formation of two
values indicate spectral variables of high importance. The same clearly separated compounds, the Raman spectra suggest the
applies for large negative values.
parallel presence of 1a, m3a and t3a.
PCA of the in situ Raman spectra in this work reveals two prin- Multivariate curve resolution (MCR) aims to extract informa-
cipal components (PC) for both milling reactions (Supporting tion on the pure compounds [27]. In our case, the time series of
Information File 1, Figure S2a–d). The score and loading plots Raman spectra may be considered as a superposition of all com-
for the first (PC1) and second (PC2) principal component pounds present at a certain time of the reaction. In contrast to
depicted in Figure 6a,b refer to the reaction performed at 30 Hz. PCA, where only the variance of data is evaluated, in MCR
PC1 and PC2 account for 96% and 3% of the spectral variance. chemical knowledge on the number of involved compounds,
The scores of PC1 exhibit high values at the beginning of the some constraints such as non-negativity are included. As a
reaction. With increased milling time, the PC1 scores decline result, relative component concentrations as a function of time
and eventually become negative, indicating the conversion of and the corresponding spectra can be evaluated. In this case,
1145

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
Figure 6: Results of multivariate data analysis of Raman spectra for 30 Hz milling experiments. Principal component analysis (PCA) of Raman data,
showing (a) scores, and (b) loadings for (black) PC1 and (blue) PC2. (a) PC1 scores show a decrease in spectral component of 1a in the first minutes
–
cf. positive spectral signature of loadings of PC1 in (b). The negative scores of PC1 correspond to the negative part of PC1 loadings, i.e., t3c.
Scores of PC2 indicate the presence of a compound with spectral features resembling m3c. For comparison, intensity of PXRD Bragg peaks at
green) 2θ = 13.87° and (red) 2θ = 14.78° are given in (a) for m3c and t3c, respectively. Multivariate curve resolution (MCR) of (c) composition
(
profiles, and (d) component spectra. Coloring of (c) and (d) are used to indicate corresponding data. (c) Relative phase composition of (black) 1a and
the (grey) PMMA-background decrease within the first minutes of the reaction. Simultaneous formation of (green) m3c, shows pronounced increase
followed immediately by a rapid decrease. The latter is accompanied by the formation of (red) t3c. In the final stage of the reaction, only contributions
from t3c and the PMMA background can be observed.
four spectral compounds were used in form of an initial guess The mechanochemical catalyzed reactions of 1b and 1c with 2
for MCR, namely fluorinated benzaldehyde, PMMA (back- performed with 2 µL piperidine did not exhibit the same poly-
ground from the milling jar), monoclinic and triclinic fluo- morphic transformations as with 1a. Instead, reactions using 1b
robenzylidene malonodinitrile.
and 1c led directly to formation of m3b [23] and m3c [28], re-
spectively, at both 30 Hz and 50 Hz milling frequency (Figures
MCR analysis without an initial guess resulted in models with S4 and S5 in Supporting Information File 1).
three and four components, each containing mixed information
from more than one chemical compound (not shown). However, Conclusion
MCR using spectra from the pure materials for an “initial The mechano-assisted catalyzed Knoevenagel condensation of
guess” leads to component profiles for m3a and t3a (Figure 6c) mono-fluorinated benzaldehydes and malonodinitrile was
that are in good agreement with XRPD results (Figure 6a and explored in situ and in real-time by tandem synchrotron powder
c). The component spectra obtained by MCR can be assigned to X-ray diffraction and Raman spectroscopy. For synthesis of
the m3a, t3a, 1a, and the milling jar, PMMA (Figure 6d, Figure p-fluorobenzylidene malonodinitrile (3a) the reaction product
S3, Supporting Information File 1). Negligible contributions of crystallizes according to two different pathways, depending on
spectral features relating to (2) were observed. It should be the concentration of base catalyst. At high concentrations of
noted that all component spectra, especially those of (1a) catalyst, the triclinic product phase is formed, and remains
contain spectral properties belonging to PMMA (shaded areas stable under continued mechanical treatment. In contrast, at
in Figure S3, Supporting Information File 1).
lower concentrations of catalyst, the product crystallizes first as
1146

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
the monoclinic phase. Subsequent milling causes this phase to processed using FIT2D [30]. In situ real-time Raman data were
transform abruptly to the triclinic phase. Due to the inclusion of collected using a non-contact probe (beam diameter 1 mm) and
base catalyst in the final product, we suggest this difference to excitation wavelength of 785 nm. Raman scattering was
be the result of a templating effect, which dominates at higher collected on a RXN1TM analyzer (Kaiser Optical systems,
concentrations. For the reaction of meta- and ortho-substituted France), equipped with a CCD detector (1024 × 1024 pixel).
substrates, crystallization occurs directly into the monoclinic Each Raman spectrum consists of 5 s accumulated scattering in-
phase, regardless of milling conditions or catalyst concentra- tensity, with successive Raman spectra collected every 30 s.
tion.
Chemometrics: The Raman spectra were evaluated using prin-
Multivariate analysis of in situ Raman spectra by both PCA and ciple component analysis (PCA) and multivariate curve resolu-
MCR suggests the formation of a transient product with almost tion (MCR) with the software The Unscrambler® X Vers. 10.5
identical spectral properties as the final product, the triclinic (CAMO). Prior to multivariate analysis, Raman spectra were
polymorph of 3a. These results are consistent with those of baseline corrected followed by unit vector normalization in the
XRPD analysis. Hence, we here identify a new approach to spectral range of 200–2500 cm−1. PCA was conducted with
monitoring mechanically-induced polymorphic transitions in mean centered data using cross validation with 20 randomly
situ and in real-time.
selected segments. MCR iterations were initialized with the
constraints of "non-negative spectra" and "non-negative concen-
trations", and sensitivity to pure compounds was set to 100. The
Experimental
Materials: All chemicals used in this work were taken as maximum number of iterations was set at 250.
supplied (>97% purity), without further purification.
Mass spectrometry: Mass spectra were recorded with electro-
Syntheses: The following procedure is similar as described in spray ionization time of flight mass spectrometry. A Q-TOF
our previous work [29]. Milling experiments were performed Ultima ESI-TOF mass spectrometer (Micromass, Germany)
using a commercially available vibratory ball mill (Pulverisette running at 4 kV capillary voltage and a cone voltage of 35 V
2
3, Fritsch, Germany). For each experiment, stoichiometric was used. The collision energy was set to 5 eV. The source tem-
quantities of reactants p-, m- and o- fluorobenzaldehyde perature was 120 °C whereas the desolvation temperature was
500 mg, ca. 4.03 mmol) and malonodinitrile (266.1 mg, adjusted to 150 °C. The mass spectrometer was operating in
(
ca. 4.03 mmol) were weighed into Perspex milling jars (10 mL). positive ion mode. Around 0.1 mg of the samples were weigh-in
To each jar, a quantity (defined in the main text) of piperidine and solved in methanol (HPLC grade).
was added as catalyst. Two stainless steel milling balls (4 g,
1
0 mm diameter) were also included in each milling jar. The
Supporting Information
reactions were conducted at 30 Hz or 50 Hz, as indicated in the
main text. The final products were characterized by XRPD.
Supporting Information File 1
XRPD data and multivariate data analysis.
X-ray powder diffraction (XRPD): All samples were charac-
terized by XRPD analysis using a Bruker D8 diffractometer
with Cu-Kα1 radiation (λ = 1.54106 Å) in a range of 5.0° ≤ 2 θ
[
https://www.beilstein-journals.org/bjoc/content/
supplementary/1860-5397-15-110-S1.pdf]
≤
40°. The data were obtained in transmission mode with a step
size of 0.009° and an acquisition time of 3 s per step.
ORCID® iDs
Sebastian Haferkamp - https://orcid.org/0000-0001-6049-1102
Andrea Paul - https://orcid.org/0000-0003-2849-4754
In situ investigations: In situ and real-time monitoring of the
Adam A. L. Michalchuk - https://orcid.org/0000-0001-7405-3269
Franziska Emmerling - https://orcid.org/0000-0001-8528-0301
milling reactions was conducted at the mySpot Beamline
(
BESSY II, Helmholtz Centre Berlin for Materials and Energy).
The same mechanochemical reactor was used for these investi-
gations, as was used for laboratory synthesis reactions; i.e., a References
Pulverisette 23, Fritsch, Germany. Perspex milling jars were
used, which have been previously shown to permit collection of
good quality XRPD data during milling reactions [14]. Diffrac-
tion was collected using an incident beam of 12.4 keV. 2D scat-
tering images were recorded on a MarMosaic, CCD detector
1. James, S. L.; Adams, C. J.; Bolm, C.; Braga, D.; Collier, P.; Friščić, T.;
Grepioni, F.; Harris, K. D. M.; Hyett, G.; Jones, W.; Krebs, A.; Mack, J.;
Maini, L.; Orpen, A. G.; Parkin, I. P.; Shearouse, W. C.; Steed, J. W.;
Waddell, D. C. Chem. Soc. Rev. 2012, 41, 413–447.
doi:10.1039/c1cs15171a
2.
Stolle, A.; Szuppa, T.; Leonhardt, S. E. S.; Ondruschka, B.
Chem. Soc. Rev. 2011, 40, 2317–2329. doi:10.1039/c0cs00195c
(
resolution 3072 × 3072 pixel). All scattering data were
1147

Beilstein J. Org. Chem. 2019, 15, 1141–1148.
3.
4.
5.
6.
7.
8.
Andersen, J.; Mack, J. Green Chem. 2018, 20, 1435–1443.
doi:10.1039/c7gc03797j
27.Casassas, E.; Marqués, I.; Tauler, R. Anal. Chim. Acta 1995, 310,
473–484. doi:10.1016/0003-2670(95)00140-u
Hulvey, Z.; Furman, J. D.; Turner, S. A.; Tang, M.; Cheetham, A. K.
Cryst. Growth Des. 2010, 10, 2041–2043. doi:10.1021/cg100121n
Chadwick, K.; Davey, R.; Cross, W. CrystEngComm 2007, 9, 732–734.
doi:10.1039/b709411f
28.Timofeeva, T. V.; Nesterov, V. N.; Antipin, M. Y.; Clark, R. D.;
Sanghadasa, M.; Cardelino, B. H.; Moore, C. E.; Frazier, D. O.
J. Mol. Struct. 2000, 519, 225–241.
doi:10.1016/s0022-2860(99)00305-1
Nielsen, S. F.; Peters, D.; Axelsson, O. Synth. Commun. 2000, 30,
29.Haferkamp, S.; Fischer, F.; Kraus, W.; Emmerling, F.
Beilstein J. Org. Chem. 2017, 13, 2010–2014. doi:10.3762/bjoc.13.197
30.Hammersley, A. P.; Brown, K.; Burmeister, W.; Claustre, L.;
Gonzalez, A.; McSweeney, S.; Mitchell, E.; Moy, J. P.;
Svensson, S. O.; Thompson, A. W. J. Synchrotron Radiat. 1997, 4,
67–77. doi:10.1107/s0909049596015087
3
501–3509. doi:10.1080/00397910008087262
Kaupp, G.; Reza Naimi-Jamal, M.; Schmeyers, J. Tetrahedron 2003,
9, 3753–3760. doi:10.1016/s0040-4020(03)00554-4
5
Kulla, H.; Greiser, S.; Benemann, S.; Rademann, K.; Emmerling, F.
Cryst. Growth Des. 2017, 17, 1190–1196.
doi:10.1021/acs.cgd.6b01572
9
1
1
1
1
1
1
1
.
Fischer, F.; Greiser, S.; Pfeifer, D.; Jäger, C.; Rademann, K.;
Emmerling, F. Angew. Chem., Int. Ed. 2016, 55, 14281–14285.
doi:10.1002/anie.201607358
License and Terms
0.Caira, M. R.; Nassimbeni, L. R.; Wildervanck, A. F.
J. Chem. Soc., Perkin Trans. 2 1995, 2213–2216.
doi:10.1039/p29950002213
This is an Open Access article under the terms of the
Creative Commons Attribution License
1.Štrukil, V.; Fábián, L.; Reid, D. G.; Duer, M. J.; Jackson, G. J.;
Eckert-Maksić, M.; Friščić, T. Chem. Commun. 2010, 46, 9191–9193.
doi:10.1039/c0cc03822a
(http://creativecommons.org/licenses/by/4.0). Please note
that the reuse, redistribution and reproduction in particular
requires that the authors and source are credited.
2.Friščić, T.; Halasz, I.; Beldon, P. J.; Belenguer, A. M.; Adams, F.;
Kimber, S. A. J.; Honkimäki, V.; Dinnebier, R. E. Nat. Chem. 2013, 5,
66–73. doi:10.1038/nchem.1505
The license is subject to the Beilstein Journal of Organic
Chemistry terms and conditions:
3.Gracin, D.; Štrukil, V.; Friščić, T.; Halasz, I.; Užarević, K.
Angew. Chem., Int. Ed. 2014, 53, 6193–6197.
(https://www.beilstein-journals.org/bjoc)
doi:10.1002/anie.201402334
4.Batzdorf, L.; Fischer, F.; Wilke, M.; Wenzel, K.-J.; Emmerling, F.
Angew. Chem., Int. Ed. 2015, 54, 1799–1802.
The definitive version of this article is the electronic one
which can be found at:
doi:10.1002/anie.201409834
5.Kulla, H.; Haferkamp, S.; Akhmetova, I.; Röllig, M.; Maierhofer, C.;
Rademann, K.; Emmerling, F. Angew. Chem., Int. Ed. 2018, 57,
doi:10.3762/bjoc.15.110
5930–5933. doi:10.1002/anie.201800147
6.Lukin, S.; Tireli, M.; Lončarić, I.; Barišić, D.; Šket, P.; Vrsaljko, D.;
di Michiel, M.; Plavec, J.; Užarević, K.; Halasz, I. Chem. Commun.
2
018, 54, 13216–13219. doi:10.1039/c8cc07853j
7.Trotzki, R.; Hoffmann, M. M.; Ondruschka, B. Green Chem. 2008, 10,
73. doi:10.1039/b808265k
8.Trotzki, R.; Hoffmann, M. M.; Ondruschka, B. Green Chem. 2008, 10,
67–772. doi:10.1039/b801661e
9.Haferkamp, S.; Kraus, W.; Emmerling, F. J. Mater. Sci. 2018, 53,
3713–13718. doi:10.1007/s10853-018-2492-0
1
1
1
2
2
2
8
7
1
0.Rightmire, N. R.; Hanusa, T. P. Dalton Trans. 2016, 45, 2352–2362.
doi:10.1039/c5dt03866a
1.Simone, E.; Steele, G.; Nagy, Z. K. CrystEngComm 2015, 17,
9370–9379. doi:10.1039/c5ce01878a
2.El-Agrody, A. M.; Al-Omar, M. A.; Amr, A. E.-G. E.; Ng, S. W.;
Tiekink, E. R. T. Acta Crystallogr., Sect. E: Struct. Rep. Online 2013,
69, o515. doi:10.1107/s1600536813006235
23.Antipin, M. Y.; Nesterov, V. N.; Jiang, S.; Borbulevych, O. Y.;
Sammeth, D. M.; Sevostianova, E. V.; Timofeeva, T. V. J. Mol. Struct.
2003, 650, 1–20. doi:10.1016/s0022-2860(02)00468-4
2
4.Belenguer, A. M.; Lampronti, G. I.; Cruz-Cabeza, A. J.; Hunter, C. A.;
Sanders, J. K. M. Chem. Sci. 2016, 7, 6617–6627.
doi:10.1039/c6sc03457h
25.Pearson, K. London, Edinburgh Dublin Philos. Mag. J. Sci. 1901, 2,
559–572. doi:10.1080/14786440109462720
26.Hotelling, H. J. Educ. Psychol. 1933, 24, 417–441.
doi:10.1037/h0071325
1148