Efficient and scalable synthesis of ketones via nucleophilic Grignard addition to nitriles using continuous flow chemistry

Article abstract of DOI:10.1016/j.tetlet.2013.02.069

In the present Letter we report the development of efficient continuous flow chemistry conditions for the scalable preparation of ketones. This transformation is achieved via nucleophilic addition of Grignard reagents to the corresponding nitriles and imine hydrolysis by means of in-series plug flow reactors.

Full text of DOI:10.1016/j.tetlet.2013.02.069

Tetrahedron Letters 54 (2013) 2226–2230
Contents lists available at SciVerse ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Efficient and scalable synthesis of ketones via nucleophilic Grignard addition
to nitriles using continuous flow chemistry
⇑
Carlos Mateos , Juan A. Rincón, José Villanueva
Centro de Investigación Lilly S.A., Avda. de la Industria, 30, Alcobendas, Madrid 28108, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
In the present Letter we report the development of efficient continuous flow chemistry conditions for the
scalable preparation of ketones. This transformation is achieved via nucleophilic addition of Grignard
reagents to the corresponding nitriles and imine hydrolysis by means of in-series plug flow reactors.
Ó 2013 Elsevier Ltd. All rights reserved.
Received 10 January 2013
Revised 18 February 2013
Accepted 19 February 2013
Available online 27 February 2013
Keywords:
Ketones
Grignard reagents
Flow chemistry
Continuous process
Ketones in general, and especially arylketones, have been
widely used in the agrochemical, fragrance, dye, and pharmaceuti-
cal industries.1 More importantly, they are among the most versa-
tile intermediates available in the synthetic organic toolbox. To
address some of the needs of one of our drug discovery programs,
we needed ready access to gram-scale intermediates of the general
structure A to facilitate structure–activity relationship (SAR) stud-
ies (Fig. 1).
One of the most straightforward methods to prepare these scaf-
folds consists of the nucleophilic addition of the corresponding
Grignard reagent to the nitrile and subsequent imine hydrolysis
(Scheme 1).2 The nitrile substrates are in many cases readily com-
mercially available.
X
Mg
NH
O
H3O+
N
Ar2
Ar1
Ar1
Ar2
Ar1
Ar2
Scheme 1. General preparation of ketones via nucleophilic addition of Grignards to
nitriles.
reproducible and safer than a batch process.4 From our previous
experience, the hydrolysis of the intermediate imines is highly exo-
thermic and can be difficult to control during batch scale-up. To
overcome this difficulty we envisioned a continuous flow approach
in which both Grignard addition and subsequent hydrolysis could
be achieved continuously.5
There is a growing interest within the organic synthesis com-
munity to find new procedures for translating known transforma-
tions from batch to continuous flow.3 Due to the improved mass
and heat transfer that is a key feature of flow technology, the
scale-up of the desired reaction is potentially higher yielding, more
As the starting point for our investigation we used 2-aminoben-
zonitrile (1) and phenylmagnesium bromide (2) as model reagents
for the preparation of imine 3, to find optimal conditions in flow
mode. For comparison purposes, we first ran the reaction in stan-
dard batch fashion. The reaction was complete in THF as solvent,
at RT in 15 h using 3.5 equiv of PhMgBr. For the hydrolysis step,
an inverse addition of the reaction mixture over 6 N HCl at rt
was needed to control an exotherm (from rt to 60 °C). Additional
stirring at 50 °C for 1 h was required for complete conversion
(Scheme 2).
O
Ar
NH2
R
With this batch result in hand, we designed a flow reaction set-
up for this transformation: a solution of the nitrile in THF was
pumped and mixed using a T-union with a solution of Grignard,6
with reaction taking place in a PTFE-tubular flow reactor (5.5 m,
1 mm ID, total volume 4.32 mL)7 using standard pressure syringe
pumps.8 After allowing this mixture to react at the selected tem-
perature, the output of the reactor was analyzed by LC–MS (Fig. 2)
A
Figure 1. General structure of SAR key intermediates.
⇑
Corresponding author. Tel.: +34 91 6233394; fax: +34 91 6633411.
E-mail address: c.mateos@lilly.com (C. Mateos).
0040-4039/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tetlet.2013.02.069

C. Mateos et al. / Tetrahedron Letters 54 (2013) 2226–2230
2227
NH2
O
NH2
NH2
N
MgBr
H3O+
1h, 60°C
NH
15h, RT
THF
+
3.5 eq
1
2
3
4
Scheme 2. Model reaction performed in batch mode.
Initially, a small set of flow chemistry conditions was screened,
such as reaction temperature, residence time for the Grignard
addition, and organomagnesium reagent excess (Table 1). The best
conditions obtained were T = 60 °C, residence time = 30 min, and
5 equiv of Grignard reagent (entry 5).
Table 2
General methods for continuous flow process
Method
A
B
C
D
Temp (°C)
Residence time (min)
60
5
60
30
0.4
60
60
0.4
60
240
1.2
The next step was to implement the hydrolysis step in a further
continuous process. Thus, a stream of 6 N HCl was pumped and
mixed with the output of the first reactor, entering a second flow
reactor in which imine hydrolysis took place to afford the final de-
sired ketone. We designed the second reactor with the same
parameters as the first one for convenience (5.5 m, 1 mm ID, total
volume 4.32 mL) (Fig. 3).
Nitrile concn (M)
0.1
To assess the scope of the addition and hydrolysis process, a set
of aromatic and aliphatic nitriles were reacted with selected Grig-
nard reagents. The results obtained are summarized in tables and
discussed under the subheadings below. In some cases, fine tuning
of the previously optimized conditions was needed. For the sake of
clarity, the methodology developed was categorized in four differ-
ent condition sets: A, B, C, and D, depending on the values used
(residence time and concentration of starting nitrile). These meth-
ods were named alphabetically based on harshness (D > C > B > A),
(Table 2).
Addition of phenylmagnesium bromides to aromatic nitriles
With the optimized conditions found for the reaction of our
model aminobenzonitrile (general method A), we first studied
the scope and limitations of this chemical transformation with re-
gard to substitution of the aromatic nitrile. Thus, a new set of
substituted aromatic nitriles with differing electronic properties
Figure 3. Final flow process set-up.
was chosen to react with the same Grignard reagent (PhMgBr).
As it can be seen in Table 3, the reaction is of quite broad scope,
using the general method B (Table 2) in the majority of cases. Puri-
ties obtained ranged from good to excellent, and the reaction toler-
ated aminobenzonitriles, as previously described (entry 1),
electron-withdrawing substituents in o-, m-, and p-positions (en-
tries 3–5), electron-donating substituents in o-, m-, and p-posi-
tions, (entries 6–8) as well as heteroaromatic substitution
(entries 9 and 10). In the case of the reaction of 4-methoxybenzo-
nitrile with PhMgBr (entry 8) some precipitation was observed that
clogged the tubular reactor during the reaction. We circumvented
this problem by placing the reactors in a thermostatic ultrasonic
bath that allowed the reaction mixture to flow without any issue.9
Interestingly, if, during the experiment, the sonication was
stopped, the precipitate was formed again, once more clogging
the reactor line.10
Figure 2. Initial continuous flow Grignard addition.
In order to study more extensively the scope of the flow process
and once we had investigated the pair Ar-CN and PhMgBr, we
decided to test the reaction partners Ar-CN with aliphatic Grignard
reagent, aliphatic-CN with PhMgBr, and aliphatic-CN with aliphatic
organomagnesium reagents.
Table 1
Initial optimization efforts for the Grignard addition to aminobenzonitrile
Entry
T
Residence time (min)
Grignard (equiv)
Conversion (%)
1
2
3
4
5
rt
rt
rt
60 °C
60 °C
5
35
60
10
30
3.5
3.5
3.5
5
20
55
65
60
95
Addition of alkyl Grignards to aromatic nitriles
The conversions obtained using these reagent types ranged
from excellent to modest. Using 3-methoxybenzonitrile as a model,
5

2228
C. Mateos et al. / Tetrahedron Letters 54 (2013) 2226–2230
we employed methylmagnesium bromide, iso-propylmagnesium
bromide, and tert-butylmagnesium chloride to study the influence
of the Grignard substitution. The results showed a drop in conver-
sion to the product when going from the primary (98% purity, entry
11) to the secondary (65% purity, entry 12) reagent, probably due
to steric factors. With these reagents, general method C (Table 2)
had to be used to maximize conversion values. We were not able
to measure the conversion with the tert-butylmagnesium chloride
due to a flow reactor clogging issue (entry 14); despite the use of
sonication we repeated the reaction several times unsuccessfully.
Addition of phenylmagnesium bromide to alkyl nitriles
We selected several alkyl nitriles for the reaction with phenyl-
magnesium bromide and found that conversion to the desired ke-
tone was very modest. To determine the influence of the
Table 3
Continuous-flow synthesis of ketones via Grignard addition to nitriles
Entry Nitrile
Grignarda Product
Purity (%)b Method Entry Nitrile
Grignard Product
Purity (%)b Method
NH2
CN
NH2
O
MeO
MeO
O
1
PhMgBr
72
90
92
B
B
B
13
14
15
iPrMgBr
60 (50)c
C
C
C
CN
Ph
OCH3
OCH3
O
CN
2
PhMgBr
tBuMgCl
0d,e
O
Ph
CN
Cl
Cl
O
Cl
Cl
iPrMgBr
25 (20)c
O
3
PhMgBr
CN
CN
Ph
Cl
Cl
O
CN
CN
O
4
PhMgBr
80
45
B
B
16
17
PhMgBr
45
A
B
CN
Ph
Cl
Cl
O
O
CN
5
PhMgBr
PhMgBr
8f
Ph
OMe
CN
OMe
O
NC Ph
O
Ph
Ph
6
PhMgBr
65
C
B
18
19
PhMgBr
47
0f
C
Ph
OMe
OMe
O
CN
7
PhMgBr
90 (87)c
EtMgBr
D
O
CN
Ph
MeO
MeO
NC Ph
O
Ph
O
8
PhMgBr
95e
85
C
A
20
21
EtMgBr
0d,e
D
D
CN
Ph
N
N
O
0d,e
CN
9
PhMgBr
EtMgBr
EtMgBr
Ph
O
N
CN
O
O
CN
10
PhMgBr
85 (70)c
C
22
0f
C
N
H
N
H
(continued on next page)

C. Mateos et al. / Tetrahedron Letters 54 (2013) 2226–2230
2229
Table 3 (continued)
Entry Nitrile
OCH3
Grignarda Product
Purity (%)b Method Entry Nitrile
Grignard Product
Purity (%)b Method
OCH3
11
EtMgBr
98
65
C
C
O
O
CN
OCH3
OCH3
12
iPrMgBr
CN
a
b
c
d
e
f
1.5 equiv of Grignard reagent were used except with 2-aminobenzonitrile (5 equiv). All as solutions in THF except tBuMgCl (MeTHF).
HPLC peak integration.
Isolated yield.
Flow reactor was clogged during the hydrolysis in flow.
Sonication was applied.
Primarily auto-addition product obtained.
iles and/or arylgrignards providing
a safe and reproducible
procedure.
O
Acknowledgments
NC
We thank Dr. Graham Robert Cumming for his assistance during
the preparation of the manuscript. José Villanueva wants to thank
the Fundación Universidad-Empresa (FUE) for financial support.
B
Figure 4. Auto-addition product.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.tetlet.2013.
02.069.
substitution in the alkyl nitrile we also employed primary and ter-
tiary nitriles. In the case of the 3-phenylpropanonitrile, the use of
milder reaction conditions (method A, entry 16 instead of B, entry
17) increased the conversion substantially. In the case of more
highly substituted alkyl nitrile, the yield obtained was moderate
(entry 18). When using the primary alkyl nitrile, the main by-prod-
uct obtained was compound B generated by auto-condensation of
the nitrile (Fig. 4) whose formation could be prevented with bulk-
ier nitriles due to steric factors (entry 18).
References and notes
1. (a) Frank, H. G.; Stadelhofer, J. W. Industrial Aromatic Chemistry; Springer:
Berlin, 1988; (b) Surburg, H.; Panten, J. Common Fragrance and Flavor Materials,
5th ed.; Wiley-VCH: Weinheim, Germany, 2006.
2. (a) Lochte, H. L.; Horeczy, J.; Pickard, P. L.; Barton, A. D. J. Am. Chem. Soc. 1948,
70, 2012–2015; (b) Pickard, P. L.; Vaughan, D. J. Ibid. 1950, 72, 876–878; (c)
Pickard, P. L.; Engles, E. F. Ibid. 1952, 74, 4607–4608; (d) Mosher, H. S.; Mooney,
W. T. J. Am. Chem. Soc. 1951, 73, 3948–3949.
Addition of ethylmagnesium bromide to alkyl nitriles
3. (a) For a general reference about Flow Chemistry see: Chemical Reactions and
Processes under Flow Conditions, Luis, Santiago V. and Garcia-Verdugo,
Eduardo, RSC Green Chemistry Series, 2009.; (b) For an outstanding reference
work covering continuous processes in organic synthesis using mainly micro-
reactors see Micro Reaction Technology in Organic Synthesis, Charlotte Wile
and Paul Watts, Taylor and Francis Group, LLC., 2011.; (c) For a recent review
on flow chemistry inside the Pharmaceutical Industry see: Malet-Sanz, Laia;
Susanne, Flavien J. Med. Chem. 2012, 55, 4062–4098.
4. In the pharma industry, there are two main types of flow reactors employed,
plug flow reactors (PFR) and continuous stirred tank reactors (CSTR). In the
present Letter, we only used tubular PFRs.
5. Riva, E.; Gagliardi, S.; Martinelli, M.; Passarella, D.; Vigo, D.; Rencurosi, A.
Tetrahedron 2010, 66, 3242–3247.
The reaction between aliphatic nitriles and EtMgBr was unsuc-
cessful in all cases examined. With primary nitriles, auto-conden-
sation was the main operating process (entries 19 and 22)
whereas with more substituted aliphatic nitriles reactor fouling
was obtained, even with the aid of sonication (entries 20 and
21). In these cases of reactor clogging, we believe they are caused
by the lack of reactivity of the reagents, which causes these to pre-
cipitate in the THF/H2O mixture during the hydrolysis step.
In order to more fully assess the validity of the methodology
developed and to quantify the throughput of the processes, a small
number of selected examples were scaled up and purified to obtain
isolated yields. In all cases, the numbers remained close to the cor-
responding purity values obtained by LC–MS integration (Table 3,
entries 7, 10, 13, and 15). In terms of productivity, using the exam-
ple described in entry 7 the throughput obtained was 12 g/day
using the 4.3 mL reactor but could be increased to 144 g/day using
a 50 mL reactor.11
6. Check valves were installed in both lines to prevent fluid return. Back pressure
regulator was installed to prevent gas bubble formation and ensure accurate
residence times.
7. Theoretical total volume is calculated as a function of diameter and length
(V = 3.1416 Â (D/2) Â L). Calculated residence time based on reactor V and flow
rate values are accurate when T is not too high (far from the solvent critical
point). At very high T, phase and density changes make the actual residence
time lower than the calculated value. The calculated length to diameter ratio
value (L/D) is 5500. As a rule of thumb, we considered that L/D must be above
the standard minimum number of 5000 to ensure efficient diffusion rates that
implies good mixing between the two feed lines. See: (a) May, S. A.; Johnson, M.
D.; Braden, T. M.; Calvin, J. R.; Haeberle, B. D.; Jines, A. R.; Miller, R. D.;
Plocharczyk, E. F.; Rener, G. A.; Nichey, R. N.; Schmid, C. R.; Vaid, R. K.; Yu, H.
Org. Process Res. Dev. 2012, 16, 982–1002; (b) Gervais, T.; Jensen, K. F. Chem.
Eng. Sci. 2006, 61, 1102–1121.
In conclusion, we have developed an efficient continuous flow
approach to the synthesis of ketones by nucleophilic addition of
aryl or alkyl Grignard reagents to nitriles and subsequent hydroly-
sis. We have demonstrated that the methodology is amenable for
gram scale preparations, it is fairly general for the case of arylnitr-
8. Normal pressure Nemesys syringe pumps. Cetoni Automatisierung and
Mikrosysteme GmbH, Germany. www.cetoni.de.

2230
C. Mateos et al. / Tetrahedron Letters 54 (2013) 2226–2230
9. Timothy, N.; Naber, J. R.; Hartman, L.; McMullen, J.; Jensen, K. F.; Buchwald, S. L.
Chem. Sci. 2011, 2, 287.
10. For a general review of the management of solid in flow see: Hartman, Ryan L.
Org. Process Res. Dev. 2012, 16, 870–887.
11. For large scale-up purposes syringe pumps must work in pairs continuously to
feed volumes greater than those of a single syringe. Other alternative could be
the use of HPLC-type or peristaltic pumps.