An Autonomous Chemical Robot Discovers the Rules of Inorganic Coordination Chemistry without Prior Knowledge

Article abstract of DOI:10.1002/anie.202000329

We present a chemical discovery robot for the efficient and reliable discovery of supramolecular architectures through the exploration of a huge reaction space exceeding ten billion combinations. The system was designed to search for areas of reactivity f

Full text of DOI:10.1002/anie.202000329

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
International Edition Chemie
www.angewandte.org
Accepted Article
Title: An autonomous chemical robot discovers the rules of
supramolecular chemistry without prior knowledge
Authors: Lee Cronin, Luzian Porwol, Daniel Kowalski, Alon Henson,
De-Liang Long, and Nicola Bell
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202000329
Link to VoR: https://doi.org/10.1002/anie.202000329

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
An autonomous chemical robot discovers the rules of inorganic
coordination chemistry without prior knowledge
Luzian Porwol, Daniel J. Kowalski, Alon Henson, De-Liang Long, Nicola Bell, and Leroy Cronin*
Dedication ((optional))
self-assembly process.9-10 An increase in the complexity of
coordination compounds can be observed when two or more
Abstract: We present a chemical discovery robot for the efficient
and reliable discovery of supramolecular architectures through the
exploration of
a huge reaction space exceeding ten billion
donor groups per ligand are spatially separated far enough from
combinations. The system was designed to search for areas of
reactivity found through autonomous selection of the reagent types,
amounts, and reaction conditions aiming for combinations that are
reactive. The process consists of two parts where reagents are
mixed together, choosing from one type of aldehyde, one amine and
one azide (from a possible family of two amines, two aldehydes and
four azides) with different volumes, ratios, reaction times, and
temperatures whereby the reagents are passed through a copper
coil reactor under flow. Next, either cobalt or iron is added, again
from a large number of possible quantities. The reactivity was
determined by evaluating differences in pH, UV-vis, and mass
spectra before and after the search was started. The algorithm was
focused on the exploration of interesting regions, as defined by the
outputs from the sensors, and this led to the discovery of a range of
each other to avoid coordination with the same metal ion.11
Therefore, the ligand has to be structurally rigid so that the
different binding sites prefer to coordinate in a bridging mode.12
It is also important that the ligand is flexible so that the
coordination sites of the ligand and metal can adapt to each
other.13-14 Using these considerations it is possible to select a
chemical space with many potential complexes to discover.
Herein we present a fully automated supramolecular discovery
platform that can both discover the ligand (by synthesis) and
perform the complexation reaction an order of magnitude faster
(2 h vs. 20 h) than other approaches due to the completely
autonomous, closed-loop nature of the system, see Figure 1.15
1
-benzyl-(1,2,3-triazol-4-yl)-N-alkyl-(2-pyridinemethanimine) ligands
1
2
and new complexes: [Fe(L )
2
](ClO
4
)
2
(1); [Fe(L )
2
](ClO
4
)
2
(2);
3
3
[
Co
2
(L )
2
4
](ClO )
4
(3); [Fe
2
(L )
2
](ClO (4), which were crystalized and
4 4
)
their structure confirmed via single crystal X-ray diffraction
determination, as well as a range of new supramolecular clusters
discovered in solution using high resolution mass spectrometry.
Exploring supramolecular chemical space for new assemblies is
difficult not only because it is vast and sparse, but because we
are looking for novel outcomes not easily predicted from a given
set of input reagents and conditions. In addition, most reactions
are done with specific possible outcomes in mind or rely on
established heuristics. This means that the discovery of new and
unpredicted coordination complexes is challenging and often
1
occurs serendipitously. One approach to overcome this problem
is by expanding the chemical parameter space to include, but to
not be limited to, the theoretical optimal synthesis conditions of a
known set of chemicals and look for unpredicted outcomes.2
However, expanding the scope of possibilities can result in an
exponential increase in the number of possible experiments,
along with a high cost in terms of time and resources. It is
possible to use automation and suitable exploration algorithms
to overcome these shortcomings.3-6 A key question in terms of
chemistry is how to efficiently synthesize a wide range of ligands
7
8
and complexes, and then how to explore and understand the
Figure 1. (A) Connection diagram with the prominent connections of the
system highlighted. (B) Photograph of the system with major features
annotated.
Dr Luzian Porwol, Mr Daniel J. Kowalski, Dr Alon Henson, Dr De-Liang
Long, Dr Nicola Bell, and Prof Leroy Cronin
School of Chemistry, The University of Glasgow, Glasgow, G12 8QQ (UK)
E-mail: Lee.Cronin@glasgow.ac.uk
Homepage: http://www.croninlab.com
Not only are the reaction operations automated,16 but so is the
process of discovery.17 To have flexibility to explore
supramolecular self-assembly, we aimed at the synthesis of a
Supporting information for this article is given via a link at the end of the
document.
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
modified pyridyl-triazole which can be broken into four parts
giving a large library of potentially synthesizable ligands (see
Scheme 1).19 However, the current syntheses of such
frameworks can be very time consuming taking more than a
day.20 The system we designed is capable of exploring a huge
range of possible self-assemblies constructed from three ligand
classes (totaling 56 ligands) and two different metals. This
is controlled using bespoke code written in Python such that the
chemical robot can perform all the liquid-handling operations,
including cleaning procedures between one experiment and the
next, as well as control of analytical instruments and data
analysis. The design and the connectivity of the system is shown
in Figure 1.
exploration is algorithm-driven and closed-loop, running
a
sequence of experiments under an autonomous regime, which
increases the chances (in a reasonable time scale) of the
discovery of new compounds.
The autonomous decision making21 in this system focuses on
the exploration of the most interesting regions of the chemical
space by using live experimental data of reaction mixtures. For a
robust definition of interest, we use a measure of the change
that has occurred. The chemical space is defined by a selection
of the starting materials consisting of the potential ligand building
blocks and transition metal ions, and by three different reaction
parameters: i) reagent volumes; ii) reaction temperature; and iii)
reaction duration. To exemplify this idea, we had to establish
some synthetic constraints, so we aimed to explore a potential
ligand system with a new coordination motif, see Scheme 1. The
synthesis was carried out as a three-component reaction by
combining one pyridinecarboxaldehyde (from two possibilities),
one aminoalkyne (from a set of two) and one azide (from set of
four), each with a selected volume, for a chosen duration and
temperature. The full set of choices results in 394 million
possible reactions, demonstrating that even these limited inputs
defines a vast number of potential experiments.
Scheme 1. Small organic molecules used as chemical inputs for the synthesis
of ligands, via tandem CuAAC and imine formation reactions, prior to
complexation.
The routing of chemicals and solutions through the system is
conducted in a non-deterministic fashion, see Figure 2.25 This
means that the paths through which the solutions are
transported in the system are not predetermined. For each
processing step the system only knows which solution is needed
and where but not how to perform the operation. Instead the
system decides, using graph operations on the network
connection graph that describes the system. It chooses which
route stochastically (see SI) to utilize and optimizes its own route
for each single material transfer operation. The main benefits of
such a system are increased flexibility, reduced complexity and
high scalability. Such systems can even work around faults
arising in real time.
The reactor for the synthesis step consists of a catalytically
active 10 mL capacity copper coil,22-23 which allows for the
synthesis of the bespoke ligand system within a short duration of
under 2 h (ligand dependent). This activated reactor promotes
the coordination of the in situ formed ligand with leached Cu.
The ligand formation happens due to full or partial imine
condensation
and/or
copper-catalysed
alkyne-azide
cycloaddition (CuAAC),24 depending on the reaction temperature
and reaction duration. The experimental space is seven
dimensional as it is defined by the reagents chosen (one
aldehyde, one amine, one azide and one metal), the reagent
volumes used (from 0.5 mL to 5 mL each), the reaction duration
In order to investigate how best to make discoveries in an
experimental system we designed both the chemical space and
the exploration algorithm to not rely on any knowledge beyond
the initial choice of the ligand reaction envelope.26 These areas
may or may not contain discoveries, but the algorithm was
designed to focus exploration to these regions preferentially.
The algorithm does not build a model of the space during its
exploration through it. In this way the system is designed to
search in a stochastic manner, without biases or heuristics, yet
to find areas of interest. Also, rather than optimize27 the
reactivity, the system is coded to find as many interesting points
in a given space. To be able to evaluate each experiment as a
data point in the chemical space, the algorithm uses a live data
stream from three sensors (UV/Vis, mass spectrometry and pH)
to construct a simple and robust measure of the change
occurring over both ligand synthesis step, and the metal ion
(
from 5 to 120 min) and temperature (from 30 to 80 °C). The
different possible combinations of ligand precursors can make
6 potential ligands, see Scheme 1. The metal-exchange
5
reaction occurs once the ligand mixture collected from the
catalytically active reactor (without purification) undergoes a
second reaction step, consisting of complexation with a chosen
II
volume of one of two metal salt solutions – [Fe (ClO
4 2
) ] or
II
[Co (ClO )
4 2
]. Including the second coordination step, the number
of possible experiments in the chemical system increases to 4 x
14
1
0 . The cleaning, reaction activation and decision-making
operations are all fully automated, allowing the continuous
operation of the system without human involvement. The system
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
coordination step (see ESI). Therefore, the autonomous
exploration of the chemical space is driven by the degree of
change from the starting materials to the ligand and between the
ligand and the coordination complex. This is characterized by an
exploration factor, α, which is inversely proportional to the
amount of change calculated which in turn is related to reactivity
or the degree to which a reaction has occurred. For a low α case
the exploration algorithm will result in the next experiment being
close in parameter space to the previous one, as this indicates a
Thus, this value constrains the exploration range for the next
randomly chosen experiment. The calculation of the differences
for each analytical measurement is detailed in the ESI. In
general, the differences are all based on the comparison of the
two data sets – the ligand mixture and starting materials; or
complex combination and ligand mixture. Specifically, MS
changes are calculated using the most intense and highest m/z
peaks found in each of the two spectra compared; UV-vis
changes are measured as the difference of the two spectra
region of high reactivity. Conversely, if the change was small (i.e. compared after normalization by calculating the root mean
high α and low reactivity) the system will search further away, in
order to actively seek areas with more change. It is possible for
the system to get stuck, so if too many experiments are
performed in a small region, the system is programmed to jump
to an unexplored part of the experimental space far away. In this
way the algorithm is designed to perform a minimum number of
experiments in any given area of interest to maximize the space
explored.
square value of the peak area difference; pH changes are
evaluated using the pH of the starting material or of the ligand as
a reference value, see ESI.
Figure 2. A diagram depicting the connections in our system as a network
graph. Nodes depict either locations, in black, or valves, in blue, and the
edges signify the physical connections between the nodes. LEFT: two
equivalent paths, in red and green, to move from node E to node G, the
system will randomly choose one of them. MIDDLE: the shortest path to move
from node B to node E. RIGHT: if the connections between nodes E and C,
and C and B are not clean (shown in brown), e.g. in response to the
movement of material from B to E (see middle panel), the system chooses a
new route. This is shown in orange, for movement from node G to node E,
avoiding the unclean route from B to E.
Figure 3. Schematic of the exploration algorithm operating in 2D space. The
first point is chosen at random. The measured exploration factor, α for Point 1
determines the radius away at which Point 2 is placed. The larger the value of
α, the larger the radius to the next point. This stepwise exploration is
continued in panels
C and D. Note that, over time, the exploration is
concentrated to areas of higher chemical interest (shown as red).
An explanation of the principles of operation of the exploration
algorithm is shown in Figure 3, which shows an illustration of the
algorithm exploring a 2D space with colour representing the low
change areas (blue) to high (red). The first experiment in the
sequence is a random point yielding a moderate value of
Using this system we have been able to autonomously
synthesise a range of 1-benzyl-(1,2,3-triazol-4-yl)-N-alkyl-(2-
pyridinemethanimine) ligands which were applied in the
discovery of new complexes: [Fe(L1) (1); [Fe(L2)
(L3) (L3)
2); [Co ](ClO (3); [Fe ](ClO ) (4). Exact details for
4 4
](ClO )
2 4 2
2 4 2
](ClO )
1
difference change leading to a corresponding value of α , where
(
2
2
)
4 4
2
2
α is the radius of a circle around the point. The next experiments
selected by the algorithm are points in the parameter space on
this circle line, with the choice of point two (B) from these being
stochastic, as the algorithm knows nothing about the space. The
new experiment will become the centre of the new circle having
the conditions used in the syntheses of any isolated compounds
can be found in the SI. Complex 1 is formed through tridentate
coordination of each ligand to Fe(II), with one N-donor atom
from each of the pyridine, imine, and triazole groups. Despite L1
being more flexible than L2 (due to the additional methylene
group), the favored coordination mode is still tridentate with all
nitrogen containing moieties bound. In the UV-vis spectrum of
the complex mixture (Figure S11), three new absorbance bands
are detected at 356, 476 and 544 nm. This evidence lets us
2
a well-defined radius of α , and so on (panels C and D). It is
important to note that the system is programmed to have no
more than three consecutive experiments in the same region of
space, in order to avoid exploring the same local area in a loop.
conclude that the intermediate [Cu(L1)
reactor undergoes a metal exchange process in the presence of
the Fe(II) salt to give [Fe(L1)
](ClO . Also, the pH values and
the ESI-MS spectra of the two analyzed mixtures are in
]2+ formed in the flow
2
The calculation of α is based on the ratio of differences between
the analytical measurements of the ligand with its starting
materials; and between the ligand and its resulting complex.
2
4 2
)
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
2 2 4 4
Fe ](ClO )
(L3)
[
accordance with this observation. In fact, a significant change in
the acidity of the new complex from 6.29 to 4.34 is detected. The
ESI-MS shows the presence of peaks with m/z of 390.23 in the
ligand mixture and m/z of 387.25 in the complex mixture. These
Using this same methodology [Fe(L2)
](ClO )
2 4 2
(2) was
could be assigned to [Cu(L1)
]2+ and 1 respectively.
discovered from an alternate set of input reagents yielding the
general ligand framework 1-benzyl-(1,2,3-triazol-4-yl)-N-alkyl-(2-
pyridinemethanimine). Solid state structures of these complexes
are shown in Figure 4. Further exploration of the space led to
2
the isolation of helicates [Co
Fe ](ClO (4) forming an M cage using a ligand based
(L3)
on a structure 3 type motif.
2
(L3)
2
](ClO
4
)
4
(3) and
[
2
2
)
4 4
2 2
L
The autonomous analytics run on these reaction mixtures show
that this way of conducting experiments can generate a huge
variety of different complexes in solution, not only because of
the range of potential coordination modes of the ligands, but
especially because more than one ligand can be present in the
reaction solution and reactions may be performed with non-
standard reagent stoichiometries. This is demonstrated by the
fact that we could isolate compound 4 through crystallization
from the reaction solution but the ESI-MS measurements of the
reaction mixture reveals more species to be present in solution,
see Figure 5.
Figure 4. Crystal structures of the isolated compounds. Skeletal structures of
1
-3
1
2
3
2 4 2 2 4 2 2 2 4 4
ligands L ; 1, [Fe(L ) ](ClO ) ; 2, [Fe(L ) ](ClO ) ; 3, [Co (L ) ](ClO ) ; 4,
Figure 5. ESI-MS detection of different possible structures based on the helicate complex 4 (noted here as architecture A). Architectures B and E have the
2
general formula M
, as does A, and would adopt the same κ –coordination motif as the helicates shown in Figure 4, see Table 1 for the calculated and measured peak
data.
2 3
L and would be required to adopt κ –coordination through the pyridyl and imine groups. Architectures C and D have the general formula
3
2 2
M L
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
All the assumed adducts are also summarized in Figure 5 and
one set of signals for the compound indicated as A are
consistent with the assumption that the tridentate motif
coordinates two times per metal center to form the isolated
helicate 4. Changing the coordination mode can allow the
formation of a triple stranded helicate (B) or a cage like
arrangement (E). However, other peaks have m/z values that
can be associated with the mass of specific cages, such as
circular triangular (C) or square (D) arrangements.
Arrangements like C and D may be possible because of the
knowledge. Using this approach, the system actively searches
through the available space based on the information obtained
about the system as it is running. The ligand system targeted
has not been previously reported in the literature, although it is
simple and can lead to many outcomes. Furthermore, to the best
of our knowledge, no coordination architectures are known to be
self-assembled through the presented coordination moiety. Our
studies show the complexity of this ligand system in the outcome
of several Co(II), and Fe(II)- coordination architectures. The
screening of a huge chemical space with the described ligand
system resulted in four new coordination structures, which could
be isolated, and their molecular structure could be ascertained
by X-ray diffraction.
3
flexibility of ligand L . However, to obtain those structures the
conformation of the ligand (especially for the tridentate motif)
must change to a bidentate motif. This makes it possible that
three bidentate motifs are coordinated to one metal center to
form the proposed structures illustrated in Figure 5. These Experimental Section
results also suggest that the bipodal ligands can coordinate with
their different coordination moieties in bidentate or tridentate
fashion with the imine triazole unit excluding or including the
pyridine one. The competition of two different ligand
conformations might be a reasonable explanation for these
observations. By MS, we can clearly distinguish masses that can
be associated to three different proposed structures in solutions
and assumed to be formed from the same ligand and Fe(II)
under one reaction condition.
A
volume explaining the platform, software, search data,
analytical data and crystallographic data is available. Code is
also available from the Croninlab git pages:
https://github.com/croningp/InorganicFinder
Acknowledgements
This work was supported by the EPSRC grants (No.
EP/S017046/1; EP/S019472/1; EP/R020914/1; EP/R009902/1;
EP/R01308X/1; EP/R020892/1; EP/P029329/1; EP/L023652/1;
EP/K038885/1 and the ERC for an Advanced Grant (ERC-ADG,
Table 1. calculated and measured mass spec peaks.
m/z
Peak
A
Proposed cation
charge
calculated
1420.22
660.14
407.45
280.60
1924.47
912.77
575.53
407.45
2178.30
1040.18
660.14
469.75
357.11
280.16
2938.39
1420.22
912.77
660.14
508.93
407.45
335.39
280.16
3946.89
1924.47
1248.6
912.77
710.63
574.75
478.4
measured
1420.29
660.16
407.46
280.11
1924.73
912.29
575.19
407.16
2177.85
---
660.48
469.14
---
280.11
2938.09
1420.29
912.29
660.48
---
670467 SMART-POM).
1+
2+
3+
4+
1+
2+
3+
4+
[Fe
2
(L3)
(L3)
2
](ClO
4
)
)
4
Conflicts of Interest
The work in this publication has be filed as a patent by the
University of Glasgow ref XXX.
B
C
[Fe
2
3
](ClO
4
4
+
Keywords: autonomous discovery robot • supramolecular •
2+
3+
4+
5+
6+
1+
2+
3+
4+
5+
6+
7+
8+
1+
2+
3+
4+
5+
6+
7+
8+
algorithms • artificial intelligence
[Fe
3
(L3)
3 4
](ClO )
6
[
[
1]
2]
R. E. P. Winpenny, Dalton Trans. 2002, 1, 1-10
H. Y. Zang, A. R. de la Oliva, H. N. Miras, D.-L. Long, R. T. McBurney,
L. Cronin, Nat. Commun. 2014, 5, 3715-3722
[
3]
R. Chung, J. E. Hein, Reac. Chem. & Eng., 2019, 4, 1674-1681
A. D. Clayton, J. A. Manson, C. J. Taylor, T. W. Chamberlain, B. A.
Taylor, G. Clemens, R. A. Bourne, Reac. Chem. & Eng., 2019, 4, 1545-
1554
[4]
D
[Fe
4
(L3)
4 4
](ClO )
8
407.16
---
280.11
---
1924.47
---
912.29
---
[
5]
R. J. Kearsey, B. M. Alston, M. E. Briggs, R. L. Greenaway, A. I.
Cooper, Chem. Sci., 2019, 10, 9454-9465
[
6]
F. Häse, L. M. Roch, and A. Aspuru-Guzik, Trends in Chem., 2019, 1,
2
82-291
V. Sans, L. Porwol, V. Dragone, L. Cronin, Chem. Sci. 2015, 6, 1258-
264
[
7]
4 6 4 8
[Fe (ClO )
(L3)
E
1
575.86
---
407.29
[8]
[9]
L. Porwol, A. Henson, P. J. Kitson, D.-L. Long, L. Cronin, Inorg. Chem.
Front. 2016, 3, 919-923
406.25
X. Yan, E. Sokol, X. Li, G. Li, S. Xu, R. G. Cooks, Angew. Chem. Int.
Ed. 2014, 53, 5931-5935
[
[
10] D. M. Engelhard, S. Freye, K. Grohe, M. John, G. H. Clever, Angew.
Chem. Int. Ed. 2012, 51, 4747-4750
In summary we have achieved the exploration of a vast range of
potential experiments without any heuristics or other chemical
11] O. V. Dolomanov, A. J. Blake, N. R. Champness, M. Schröder, C.
Wilson, Chem Commun, 2003, 6, 682-683
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
[
[
[
12] L. R. Holloway, M. C. Young, G. J. O. Beran, R. J. Hooley, Chem. Sci.
015, 6, 4801-4806
2
13] R. Papadakis, E. Rivière, M. Giorgi, H. Jamet, P. Rousselot-Pailley, M.
Réglier, A. J. Simann, T. Tron, Inorg. Chem. 2013, 52, 5824-5830
14] K. J. Kilpin, J. D. Crowley, Polyhedron 2010, 29, 3111-3117; H.
Struthers, B. Spingler, T. L. Mindt, R. Schibli, Chem. Eur. J. 2008, 14,
6173-6183
[
15] P. R. Symmers, M. J. Burke, D. P. August, P. I. T. Thomson, G. S.
Nichol, M. R. Warren, C. J. Campbella and P. J. Lusby, Chem. Sci.,
2015, 6, 756-760
[
[
16] V. Sans, L. Cronin, Chem. Soc. Rev. 2016, 45, 2032-2043
17] J. M. Granda, L. Donina, V. Dragone, D.-L. Long, L. Cronin, Nature,
2018, 559, 377-381
[
[
18] H. Hagiwara, T. Tanakaa and S. Hora, Dalton Trans., 2016, 45, 1713
19] N. Chopin, S. Decamps, A. Gouger, M. Médebielle, S. Picot, A.-L.
Bienvenu, G. Pilet, J. Flourine Chem. 2011, 132, 850-857
[
[
20] O. Fleischel, N. Wu, A. Petitjean, Chem. Commun. 2010, 46, 8454-
8456
21] R. A. Bourne, R. A. Skilton, A. J. Parrott, D. J. Irvine, M. Poliakoff, Org.
Process Res. Dev. 2011, 15, 932-938
[
[
[
22] A. R. Bogdan, N. W. Sach, Adv. Synth. Cat. 2009, 351, 849-854
23] K. L. Chavez, D. W. Hess, J. Electrochem. Soc. 2001, 148, G640-G643
24] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. Int. Ed. 2001,
40, 2004-2021
[
[
[
25] J. Neubert, A. P. Cantwell, S. Constantin, M. Kalontarov, D. Erickson, H.
Lipson, IEEE International Conference on Robotics and Automation
ICRA, 2010, 2479-2484
26] P. R. Symmers, M. J. Burke, D. P. August, P. I. T. Thimson, G. S.
Nichol, M. R. Warren, C. J. Campbell, P. J. Lusby, Chem. Sci., 2015, 6,
756-760
27] V. Duros, J. Grizou, W. M. Xuan, Z. Hosni, D. -L. Long, H. N. Miras, L.
Cronin, Angew. Chem. Int. Ed., 2017, 56, 10815-1082
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202000329
RESEARCH ARTICLE
Entry for the Table of Contents (Please choose one layout)
Layout 1:
RESEARCH ARTICLE
We have designed, assembled
utilized chemical robot for the
discovery of supramolecular
&
Author(s), Corresponding Author(s)*
a
Page No. – Page No.
architectures through the exploration
of a chemical space exceeding 109
possible reactions. By searching for
reactivity differences we have found a
range of new 1-benzyl-(1,2,3-triazol-4-
yl)-N-alkyl-(2-pyridinemethanimine)
ligands and discovered four new
Title
complexes
of
Fe
and
Co,
autonomously discovering the rules of
self-assembly for these systems.
Layout 2:
RESEARCH ARTICLE
Author(s), Corresponding Author(s)*
((Insert TOC Graphic here. Note: Please provide a legible image, 11×2.5 cm (w×h),
Page No. – Page No.
min. font size 6-7 pt.))
Title
Text for Table of Contents, 450 characters with spaces.
This article is protected by copyright. All rights reserved.