A robustness screen for the rapid assessment of chemical reactions

Article abstract of DOI:10.1038/nchem.1669

In contrast to the rapidity with which scientific information is published, the application of new knowledge often remains slow, and we believe this to be particularly true of newly developed synthetic organic chemistry methodology. Consequently, methods to assess and identify robust chemical reactions are desirable, and would directly facilitate the application of newly reported synthetic methodology to complex synthetic problems. Here, we describe a simple process for assessing the likely scope and limitations of a chemical reaction beyond the idealized reaction conditions initially reported. Using simple methods and common analytical techniques we demonstrate a rapid assessment of an established chemical reaction, and also propose a simplified analysis that may be reported alongside new synthetic methodology.

Full text of DOI:10.1038/nchem.1669

ARTICLES
PUBLISHED ONLINE: 9 JUNE 2013 | DOI: 10.1038/NCHEM.1669
A robustness screen for the rapid assessment
of chemical reactions
Karl D. Collins and Frank Glorius
*
In contrast to the rapidity with which scientific information is published, the application of new knowledge often remains
slow, and we believe this to be particularly true of newly developed synthetic organic chemistry methodology.
Consequently, methods to assess and identify robust chemical reactions are desirable, and would directly facilitate the
application of newly reported synthetic methodology to complex synthetic problems. Here, we describe a simple process
for assessing the likely scope and limitations of a chemical reaction beyond the idealized reaction conditions initially
reported. Using simple methods and common analytical techniques we demonstrate a rapid assessment of an established
chemical reaction, and also propose a simplified analysis that may be reported alongside new synthetic methodology.
ewly reported synthetic chemistry methodologies strive to of the methodology in a broader context. Naturally, the molecules
provide widely applicable transformations that are more effi- prepared tend to be selected to highlight the strengths of the new
N
cient, are more environmentally benign, or provide routes to methodology, and the derivatization of products has a tendency
previously inaccessible chemical motifs1–7. It is desirable that new towards simple transformations.
reactions are robust and, once reported, incorporated into the syn-
Taking into account the limitations of the current methods, we
thetic organic chemists’ repertoire of useful reactions as quickly as propose that a lack of understanding regarding the application of
possible. The importance of developing robust synthetic method- a given reaction to non-idealized synthetic problems can result in
ology for the preparation of drug-like molecules, and the subsequent a reluctance to apply new methodology. Confidence in the utility
implications for drug development, have recently been reported8. of a new reaction develops over time—often over a number of
We believe a major hurdle to the application of a new chemical years—as the reaction is gradually applied within total syntheses,
methodology to real synthetic problems9–12 is a lack of information follow-up methodological papers are published, or personal experi-
regarding its application beyond the idealized conditions of the ence is developed. Unfortunately, even when this information has
seminal report. Two major considerations in this respect are the evolved, it is often widely dispersed, fragmented and difficult to
functional group tolerance of a reaction and the stability of specific locate. To address this problem, both the tolerance of a reaction
chemical motifs under reaction conditions.
to chemical functionality and of the chemical functionality to the
The utility of new chemical methodology, including its tolerance reaction conditions must be established when appropriate, and
to chemical functionality, is typically demonstrated within the sub- reported in an easily accessible manner, preferably alongside the
strate scope of a methodological paper. Consider a typical palla- new methodology. This will facilitate more rapid incorporation of
dium-catalysed cross-coupling reaction of an arylbromide and an the given reaction into the synthetic chemists’ toolbox. It should
arylboronic acid (Fig. 1)13. Once reaction conditions are established, be noted that, to the best of our knowledge, no discrete method
the nature and position of R and R′ will be varied and, typically, for assessing the stability of chemical functionality or motifs, includ-
successful results will be reported. Alternatively, the preparation of ing heterocycles (of particular importance within the pharma-
natural products or biologically active molecules14,15, or the deriva- ceutical and agrochemical industries), has been reported.
tization of prepared products16,17, is often reported alongside new
We present a facile methodology that can be applied to new and
methodology to demonstrate its potential. Substrate scope in par- existing chemical transformations, which in contrast to established
ticular is a critical element of new methodological papers, often pro- methods can simultaneously provide a fast and discrete assessment
viding information beyond functional group tolerance, particularly of the functional group tolerance of the reaction and the stability of
with regard to mechanism. Despite the utility of these methods, they
have inherent limitations in terms of assessing a reaction for appli-
cation to non-idealized synthetic problems. Within substrate scope,
functional group tolerance is typically incorporated into an assess-
ment of the steric and electronic limitations of the methodology
R'
R'
B(OH)2
Pd
+
Br
R
and is not assessed discretely: indeed, the non-reactive functional
R
group (consider R or R′) typically influences the reactive centre.
We suggest that the non-reactive functionality is generally con- Figure 1 | A typical palladium-catalysed Suzuki–Miyaura cross-coupling.
strained within a small substrate, and that its influence on the reac- The scope of this reaction would traditionally be investigated by assessing
tion may not necessarily be extrapolated to larger or more flexible the nature of R and R′ in terms of their electronic influence (that is, electron-
substrates. The preparation of natural products or the derivatization donating functional groups (OMe, NMe2) and electron-withdrawing
of compounds prepared within a report invariably provides little functional groups (CO2Me, NO2)) and their steric influence by varying the
information with which to facilitate an assessment of the application position of a given substituent (ortho, meta or para) on the reactive centre.
Organisch-Chemisches Institut, Westfa¨lische Wilhelms-Universita¨t Mu¨nster, Corrensstrasse 40, 48149 Mu¨nster, Germany.
e-mail: glorius@uni-muenster.de
*
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.
1

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1669
ARTICLES
Table 1 | Examining functional group compatibility.
PdCl2[P(o-tolyl)3]2
2 mol%, NaOtBu
Br
N
O
HN
O
toluene, 100 °C, 3 h
MeO
MeO
1
2
3
Entry
Additive
Yield of
3 (%)*
Additive
remaining
(%)*
SM remaining
(%)*
Entry
Additive
Yield of
3 (%)*
Additive
remaining
(%)*
SM
remaining
(%)*
89†
–
O
A0
None
–
–
A9
49
60
0
OMe
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
A1
86
0
97
0
A10
A11
75
82
100
99
0
0
9
6
x
A2
100
100
2
2
CN
Cl
ꢀ
ꢀ
x
ꢀ
ꢀ
ꢀ
–
x
ꢀ
x
ꢀ
ꢀ
ꢀ
x
A3
A4
A5
A6
79
84
23
37
100
100
90
0
0
A12
A13
A14
A15
20
82
0
95
85
100
0
75
0
7
N
Cl
10
10
NH2
NH2
61
0
100
0
O
4
O
–
x
49
21
H
4
OH
x
x
x
x
ꢀ
ꢀ
OH
A7
A8
3
9
0
0
A16
A17
17
100
83
65
0
7
H
N
ꢀ
ꢀ
96
82
80
N
Ph
Ph
O
O
The standard reaction is undertaken in the presence of one molar equivalent of the given additive. The yield of 3, and the additive and starting material remaining after reaction is given. Colour coding should help
the ready assessment of the data: green ‘✓’ (above 66%), yellow ‘2’ (34266%), red ‘x’ (below 34%). *GC yield. †Isolated yields. SM is starting material.
these chemical motifs to the reaction conditions. Our method inten- analysed using gas chromatography (GC), providing a quantitative
tionally decouples the functional group from the reactive centre, assessment of the yield and therefore the tolerance of the reaction
although, as a consequence, it is not feasible to simultaneously to the given functionality, and of the stability of the additive to
assess the direct steric/electronic impact of a given functionality the reaction conditions. We propose that the intermolecular intro-
on the reaction centre. We therefore propose this method to be duction of a secondary functionality is more likely to parallel the
highly complementary to the traditional substrate scope, and the behaviour of the larger or more flexible substrates typically
data obtained should be reported alongside the traditional substrate encountered in the synthesis of complex molecules, rather than con-
scope. This approach will provide data that directly correlate to both straining the functionality within small bifunctional molecules as is
the likely tolerance and limitations of a chemical transformation. It typical within reaction scope investigations.
should be noted that as with a traditional substrate scope, a static set
of reaction conditions is assessed, and as such, a negative result in Results
the screen does not preclude that a given functionality/chemical To demonstrate our analytical methodology, we selected the widely
motif may be compatible if reaction conditions are optimized.
applied Buchwald–Hartwig reaction. Using the seminal catalyst
Our approach is conceptually very simple. A standard reaction is system independently reported by both Buchwald and Hartwig—
undertaken in the presence of one molar equivalent of an ‘additive’ PdCl2[P(o-tolyl)3]2 (refs 18,19)—we investigated the reaction of
with a given chemical functionality or structural motif. Reactions are 3-bromoanisole 1 and morpholine 2 (Table 1). In the absence of
2
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1669
ARTICLES
Table 2 | Examining heterocycle compatibility.
PdCl2[P(o-tolyl)3]2
2 mol%, NaOtBu
Br
N
O
HN
O
toluene, 100 °C, 3 h
MeO
MeO
1
2
3
Entry
Additive
Yield of
3 (%)*
Additive
remaining
(%)*
SM remaining
(%)*
Entry
Additive
Yield of
3 (%)*
Additive
remaining
(%)*
SM remaining
(%)*
ꢀ
–
B0
None
89†
–
B9
75
51
0
79
0
N
Bn
N
ꢀ
ꢀ
x
ꢀ
ꢀ
B1
82
0
100
0
0
B10
B11
8
85
85
N
N
S
Piv
S
x
x
ꢀ
B2
96
77
nBu
x
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
B3
B4
N
11
100
83
75
0
B12
B13
81
87
0
0
N
O
N
O
ꢀ
70
83
100
nBu
Cl
x
ꢀ
x
x
B5
0
81
99
B14
4
0
87
N
N
NH
S
N
ꢀ
ꢀ
x
ꢀ
ꢀ
B6
B7
76
10
70
0
5
B15
B16
8
70
81
66
0
N
Cl
O
N
x
x
–
36
54
+
N
–O
O
O
–
x
B8
61
28
13
The standard reaction is undertaken in the presence of one molar equivalent of the given additive. The yield of 3, and the additive and starting material remaining after reaction is given. Colour coding should help
the ready assessment of the data: green ‘✓’ (above 66%), yellow ‘2’ (34266%), red ‘x’ (below 34%). *GC yield. †Isolated yield. SM is starting material.
any additive, the reaction was determined to proceed to give 3 in potentially lead to a bias in the screen, with additives likely to work
89% yield (Table 1, entry A0; Table 2, entry B0).
being chosen in preference to those predicted to fail.
The reaction was then repeated in the presence of various addi-
Before undertaking experimentation, GC calibration of starting
tives, and the yield of product 3 as well as the amount of additive material 2, product 3 and the additives was undertaken. We have
and starting material remaining after the reaction were determined. successfully demonstrated a batch calibration technique inspired
Measuring the remaining starting material is critical so as to deter- by Hartwigs’ analysis of complex mixtures applied in the high-
mine whether the additive merely retards the rate of reaction, or is throughput discovery of new chemical reactivity20 (see also
detrimental to the formation of the product. The additives were ref. 21). This allows for the simultaneous calibration of multiple
divided into two groups: ‘functional groups’ A and ‘heterocycles’ compounds, significantly reducing analysis time (Supplementary
B, with the results reported in Tables 1 and 2, respectively. The addi- Section S3). To minimize experimental time, reactions were run in
tives were selected to cover a broad range of common functionalities parallel: the standard reaction was prepared on scale, and subsequently
and chemical motifs, and are all available commercially. It should be distributed to reaction vessels containing a given additive. After the
noted that the additives were not evaluated with regard to possible predetermined reaction time, the reactions were directly analysed
influence on reactivity or stability to the reaction conditions. We feel using GC. It is critical that this method is simple and fast so as to maxi-
that a more critical selection process should be avoided as it could mize its utility and encourage uptake within the synthetic organic
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.
3

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1669
ARTICLES
O
O
O
N
O
N
O
N
N
O
4
5
6
Yield 4, 75%
Recovered SM, 0%
Yield 5, 75%
Recovered SM, 0%
Yield 6, 0%
Recovered SM, 80%
Additive
Yield remaining
Additive
Yield remaining
Additive
Yield remaining
✓
✓
✓
✓
✓
x
Prediction
Result
Prediction
Result
Prediction
Result
✓
✓
✓
✓
✓
x
O
O
OMe
O
CN
O
O
O
N
N
N
O
O
O
O
7
8
9
Yield 8, 42%
Recovered SM, 34%
Debrominated SM 7%
Yield 7, 24%
Recovered SM, 19%
Yield 9, 0%
Recovered SM, 0%
Additive
Yield remaining
Additive
Yield remaining
Additive
remaining
Yield
x
✓
–
–
–
–
Prediction
Result
Prediction
Result
Prediction
Result
x
x
x
✓
–
–
Figure 2 | Results of the preparation of multifunctional substrates to validate the predictions of the robustness screen. Reaction conditions were identical
✓
x
to those used in the screen. Yields are for isolated compounds. SM, starting material. ‘ ’ (above 66%), ‘2’ (34266%), and ‘ ’ (below 34%).
community, and, as such, it should be noted that experimentation and obtained in the screen (Table 1, entry A1; Table 2, entry B1).
analysis of all additives was undertaken in one week.
Failure to synthesize 6 using the standard conditions also correlates
strongly with the robustness screen, as the terminal alkyne was
shown to completely inhibit reactivity (Table 1, entry A2). The
recovery of the starting material in 80% yield also demonstrates
the stability of the functionality to the reaction conditions, as
predicted by the screen. Compound 7 was synthesized in 24%
yield, in comparison to the 49% predicted in the robustness
screen (Table 1, entry A9). The combined sum of the product
yield and the starting material of 43% is representative of the stab-
ility of the functionality to the reaction conditions, and is compar-
able with the 60% shown in the screen. Compound 8 was
synthesized in 42% yield (compared with the 20% observed in the
screen; Table 1, entry A12), with a total yield of 83%, comparable
to the 95% observed in the screen. These examples demonstrate
that the behaviour of substrates that are not predicted to show excel-
lent reactivity and stability, or complete inhibition of the reaction,
are more difficult to predict using the robustness screen, although
the results obtained correlate to a significant degree with the those
obtained in the screen. The failure to isolate compound 9 is attrib-
uted to polymerization of the starting material and product under
the reaction conditions. The mass balance of the reaction was
16%, with significant amounts of insoluble material also observed.
Analysis of the crude reaction product suggests some product
formation and that starting material remained, although no
clean material was isolated. This result further demonstrates the
complexity of predicting the behaviour of substrates that display
intermediate reactivity and/or stability. Overall, we have
shown that the robustness screen correlates very well with results
for the reaction of the corresponding multifunctional substrates
when these substrates are expected to show excellent reactivity/
stability, or complete inhibition of reaction. Using the screen to
predict the behaviour of substrates that show intermediate
Discussion
The data generated from this screen are simple, easy to interpret,
and of significant value. Consideration of group A clearly demon-
strates that aliphatic and aromatic alcohols and amines are not
well tolerated by the reaction conditions. The secondary amide,
terminal alkyne, aromatic aldehyde and aromatic nitrile are also
not tolerated. In contrast, reactions in the presence of one equivalent
of an internal alkyne, terminal alkene, styrene, aliphatic nitrile, ter-
tiary amide, aryl or aliphatic chloride proceed in excellent yield, and
importantly the additive is not significantly degraded. The reaction
conditions are moderately tolerant of both aromatic esters and ali-
phatic ketones, although the stability of the additive is only moder-
ate. In the same manner, analysis of group B demonstrates which
heterocycles (i) inhibit reactivity and (ii) are stable to the reaction
conditions. Although these results do not provide definitive infor-
mation on the success of a specific reaction as specific electronic,
steric and conformational effects cannot be considered, it provides
an overview of which structural motifs are likely to be tolerated by
the reaction conditions and which structural motifs are stable to
the reaction conditions. This information can be applied in the
assessment of whether the reaction is likely to be suitable for a
given substrate, and in the design of synthetic routes.
To give support to the conclusions of our analysis we prepared a
number of multifunctional substrates in which the secondary func-
tionality should not have direct steric or electronic influence on the
arylbromide moiety. Reaction of the corresponding bromides with
morpholine using standard conditions was undertaken, and the
results are given below in Fig. 2.
Compounds 4 and 5 were obtained in 75% and 81% yields,
respectively, which correlate well with the yields of 86% and 82%
4
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1669
ARTICLES
6. Wender, P. A. & Miller, B. L. Synthesis at the molecular frontier. Nature
460, 197–201 (2009).
reactivity/stability is more challenging, although typically the corre-
lation between the robustness screen and the results shown in Fig. 2
is significant.
`
7. Horvath, I. & Anastas, P. T. Innovations and green chemistry. Chem. Rev.
107, 2169–2173 (2007).
We have provided an extensive and quantitative demonstration
of this methodology as applied to the initial reaction conditions
reported for the Buchwald–Hartwig coupling reaction. This
extended analysis is critical to demonstrate the utility of the meth-
odology, and provides a comprehensive assessment of the investi-
gated reaction in a short time period. However, we understand
8. Nadin, A., Hattotuwagama, C. & Churcher, I. Lead-oriented synthesis: a new
opportunity for synthetic chemistry. Angew. Chem. Int. Ed. 51, 1114–1122 (2012).
9. Nicolaou, K. C., Vourloumis, D., Winssinger, N. & Baran, P. S. The art and
science of total synthesis at the dawn of the twenty-first century. Angew. Chem
Int. Ed. 39, 44–122 (2000).
10. Mu¨ller, T. J. J. & Bunz, U. H. F. Functional Organic Materials: Syntheses,
Strategies and Applications (Wiley, 2007).
that the application of a quantitative analysis may, in some 11. Li, J. J. & Johnson, D. S. Modern Drug Synthesis (Wiley, 2010).
12. Lamberth, C. & Dinges, J. Bioactive Heterocyclic Compound Classes:
Pharmaceuticals and Agrochemicals (Wiley-VCH, 2012).
13. Miyaura, N. & Suzuki, A. Palladium-catalyzed cross-coupling reactions of
instances, be considered unnecessary or too time-consuming, and
consequently we propose a shorter qualitative analysis to be
applied and reported alongside the reaction scope of new method-
organoboron compounds. Chem. Rev. 95, 2457–2483 (1995).
ologies, as appropriate (Supplementary Section S6). Two batches
of ten additives, divided into ‘functional groups’ and ‘heterocycles’,
can be investigated. Batch calibration solutions can be prepared,
analysed once, and then utilized for multiple investigations. Using
a standard for comparison, a semi-quantitative analysis can be
undertaken using any appropriate analytical technique (HPLC,
GC, GC-MS, LC-MS). We propose that, by using this method,
experimentation and analysis can be completed in less than two
days, excluding reaction time, providing valuable data.
14. Baidya, M., Griffin, K. A. & Yamamoto, H. Catalytic enantioselective
O-nitrosocarbonyl aldol reaction of b-dicarbonyl compounds. J. Am.
Chem. Soc. 134, 18566–18569 (2012).
15. Wencel-Delord, J., Nimphius, C., Wang, H. & Glorius, F. Rhodium(III) and
hexabromobenzene—a catalyst system for the cross-dehydrogenative coupling
of simple arenes and heterocycles with arenes bearing directing groups.
Angew. Chem. Int. Ed. 51, 13001–13005 (2012).
16. Wang, H., Grohmann, C., Nimphius, C. & Glorius, F. Mild rhodium(III)-
catalyzed C–H activation and annulation with alkyne MIDA boronates: short
and efficient synthesis of heterocyclic boronic acid derivatives. J. Am. Chem. Soc.
134, 19592–19595 (2012).
This short analysis provides data typically obtained over a signifi-
cant time period from multiple sources. The discrete evaluation of
chemical motifs and functionality with respect to a chemical reac-
tion is highly complementary to the information generated in a
typical reaction scope, and should be reported concurrently.
Importantly, we also envisage that the retrospective application of
this methodology will be of particular use in an industrial environ-
ment where the preparation of structurally similar analogues is
commonplace. This method can be tailored readily to assess a
chemical reaction with respect to specific functionality or motifs
within a project, determining the suitability of the reaction before
the preparation of complex molecules that may prove either intrin-
sically unstable to the reaction conditions, or inhibit reactivity. We
strongly believe that this method will facilitate the application of
new and existing chemical reactions within both academic and
industrial environments.
´
17. Burns, A. R., Gonzalez, J. S. & Lam, H. W. Enantioselective copper(I)-catalyzed
borylative aldol cyclizations of enone diones. Angew. Chem. Int. Ed. 51,
10827–10831 (2012).
18. Guram, A. S., Rennels, R. A. & Buchwald, S. L. A simple catalytic method for
the conversion of aryl bromides to arylamines. Angew. Chem. Int. Ed. Engl.
34, 1348–1350 (1995).
19. Louie, J. & Hartwig, J. F. Palladium-catalyzed synthesis of arylamines from aryl
halides mechanistic studies lead to coupling in the absence of tin reagents.
Tetrahedron Lett. 36, 3609–3612 (1995).
20. Robbins, D. W. & Hartwig, J. F. A simple, multidimensional approach to
high-throughput discovery of catalytic reactions. Science 333, 1423–1427 (2011).
21. McNally, A., Prier, C. K. & MacMillan, D. W. C. Discovery of an a-amino
C–H arylation reaction using the strategy of accelerated serendipity. Science
334, 1114–1117 (2011).
Acknowledgements
Dedicated to Prof. R. W. Hoffmann on the occasion of his 80th birthday. The authors
acknowledge financial support from the European Research Council under the European
Community’s Seventh Framework Program (FP7 2007–2013)/ERC Grant agreement (no.
25936) and the Deutsche Forschungsgemeinschaft (DFG, Leibniz award to F.G.).
The authors thank C. Richter and D. Tang for helpful discussions.
Received 2 January 2013; accepted 27 April 2013;
published online 9 June 2013
Author contributions
F.G. and K.D.C. conceived the concept and experiments. K.D.C. performed all experiments.
Both authors discussed the results and co-wrote the paper.
References
1. Seebach, D. Organic synthesis—where now? Angew. Chem. Int. Ed. Engl.
29, 1320–1367 (1990).
Additional information
2. Newhouse, T., Baran, P. S. & Hoffmann, R. W. The economies of synthesis.
Chem. Soc. Rev. 38, 3010–3021 (2009).
3. Trost, B. M. The atom economy—a search for synthetic efficiency. Science
254, 1471–1477 (1991).
4. Trost, B. M. Selectivity: a key to synthetic efficiency. Science 219,
245–248 (1983).
Supplementary information and chemical compound information are available in the
online version of the paper. Reprints and permissions information is available online
at www.nature.com/reprints. Correspondence and requests for materials should be
addressed to F.G.
5. Trost, B. M. Atom economy. A challenge for organic synthesis. Angew. Chem.
Int. Ed. Engl. 34, 259–281 (1995).
Competing financial interests
The authors declare no competing financial interests.
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
5
© 2013 Macmillan Publishers Limited. All rights reserved.