Practical and scalable synthesis of a benzonitrile derivative via palladium-catalyzed cyanation with potassium ferrocyanide

Article abstract of DOI:10.1021/op400204g

A practical and reproducible synthetic method for the preparation of a benzonitrile derivative (1) from an aryl bromide (2) via Pd-catalyzed cyanation was established with dimethylacetamide (DMAc)-toluene as a mixed solvent. Optimization was examined by a design of experiments (DoE) technique to enhance reaction conversion and reduce the dimeric impurity 3, and desired benzonitrile derivative 1 was successfully obtained in high yields with excellent quality control on a 117 kg scale.

Full text of DOI:10.1021/op400204g

Article
pubs.acs.org/OPRD
Practical and Scalable Synthesis of a Benzonitrile Derivative via
Palladium-Catalyzed Cyanation with Potassium Ferrocyanide
Masayuki Utsugi,*,† Hiroki Ozawa,† Eiji Toyofuku,† and Masanori Hatsuda‡
†Process Development Laboratories, CMC Division, Mitsubishi Tanabe Pharma Corporation, 3-16-89 Kashima, Yodogawa-ku, Osaka
532-8505, Japan
‡Corporate Strategic Planning Department, Mitsubishi Tanabe Pharma Corporation, 2-6-18 Kitahama, Chuo-ku, Osaka 541-8505,
Japan
S
* Supporting Information
ABSTRACT: A practical and reproducible synthetic method for the preparation of a benzonitrile derivative (1) from an aryl
bromide (2) via Pd-catalyzed cyanation was established with dimethylacetamide (DMAc)−toluene as a mixed solvent.
Optimization was examined by a design of experiments (DoE) technique to enhance reaction conversion and reduce the dimeric
impurity 3, and desired benzonitrile derivative 1 was successfully obtained in high yields with excellent quality control on a 117
kg scale.
(CN)6]·3H2O (Scheme 2).7 First, the reaction stalled
irreproducibly (88−100% conversion), and reprocessing was
necessary for completion after work-up. Second, byproduct
INTRODUCTION
■
Benzonitrile derivatives are valuable components of agro-
chemicals, fine chemicals, and pharmaceuticals.1 They are
homodimer 3 was formed in the reaction (up to 0.53 area %,
HPLC) and was difficult to remove during purification,
resulting in the presence of a derivative of 3 in the final API.
To improve the reaction conversion and avoid the use of the
patented ligand XPhos, we first screened several alternative
ligands. As depicted in Table 1, aryl phosphine ligands were
effective at promoting the desired transformation, but the
cyanation of our substrate proceeded very slowly without them
(entry 2).7 A comparison of the results for the different ligands,
most notably with respect to the formation of 3 and other
impurities, indicated that P(o-tol)3 is equivalent to XPhos in
this reaction (entry 7 vs entry 1, respectively).
Improvement of Reaction Reproducibility. Reproduci-
bility remained an issue, however. We speculated that this lack
of reproducibility was due to catalyst deactivation. Grushin and
co-workers reported that catalyst deactivation in Pd-catalyzed
cyanation reactions results from the transformation of the
Pd(II) complex to inactive species such as [ArPd(CN)3]−,
[HPd(CN)3]−, and [Pd(CN)4]2−8,9 via oxidation of Pd(0) and
the presence of a high concentration of the cyanide source in
the reaction solution.8e
The effect of oxygen and other additives on the cyanation
reaction to produce 1 was then investigated, and the results are
presented in Table 2.3a,10 As air inhibited the cyanation, it did
not prevent the reaction under inert atmosphere (entry 1 vs
entry 2). The combination of Zn (4 mol %) and Zn(OAc)2 (4
mol %) as a reducing agent induced completion of the
reaction10b but afforded a higher amount of 3 (0.47 area %)
under a nitrogen atmosphere with an oxygen concentration of
<0.5% (entries 5 and 6).11 Other agents also gave disappointing
widely used in active pharmaceutical ingredients (APIs) and
their synthetic intermediates. Benzonitrile derivative 1 is a key
synthetic intermediate of our API.2 A medicinal chemistry route
is shown in Scheme 1. This method has two main drawbacks:
its multistep nature and the use of cryogenic equipment. This
prompted us to devise a short synthetic method via user-
friendly reactions.
Pd-catalyzed cyanation is a practical approach for synthesiz-
ing benzonitrile derivatives from the corresponding aryl halides
and triflates.3 The reaction has received considerable attention
following Takagi’s discovery of Pd-catalyzed cyanation with
KCN in 1973.4 Various cyanide sources have been developed,
such as NaCN, Zn(CN)2, TMSCN, and acetone cyanohydrin,3
but these reagents are toxic and difficult to handle on an
industrial scale. In 2004, Beller and co-workers described a Pd-
catalyzed cyanation using K4[Fe(CN)6], which is inexpensive
and also used as a food additive, as a nontoxic cyanide source.5,6
Despite these advantages, only a few examples of this reaction
have been reported on a manufacturing scale. Cyanide
potentially deactivates the catalyst. Thus, heterogeneous
conditions in which the cyanide source is insoluble are
generally required, which increases the difficulty of the
transformation on a large scale. Herein, we describe the
application of K4[Fe(CN)6] for the preparation of target
compound 1. We further explain how we overcame these
difficulties and controlled the quality of 1 by modifying the
method, effectively optimizing the reaction conditions with
DoE studies and accomplishing the manufacturing of 1 on a
>100 kg scale.
RESULTS AND DISCUSSION
■
Initial Screening. Compound 1 was obtained using
Weissman’s conditions as a preliminary guide, but two major
issues arose during the Pd-catalyzed cyanation with K4[Fe-
Received: August 1, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Medicinal chemistry route for the synthesis of benzonitrile derivative 1
Scheme 2. Initial conditions for Pd-catalyzed cyanation
a
Therefore, we focused next on the concentration of the
cyanide source as another catalyst deactivator. Frech reported
that an excess of K4[Fe(CN)6] leads to catalyst deactivation.8e
Thus, we reduced the concentration of the cyanide source by
adding toluene as an antisolvent for K4[Fe(CN)6] (Table 3).
Table 1. Ligand screening
a
Table 3. Effect of toluene
bc
conversion (%)
entry
ligand
time (h)
,
bc
,
b
temp time conversion
3
1
2
3
4
5
6
7
XPhos
none
2
95
entry
solvent (v/w)
DMAc (6)
(°C)
(h)
(%)
(area %)
3
24
1
2
3
125
110
125
2
100
6
0.33
d
dppp
2.5
2.5
3
99
toluene (6)
24
2
nd
dppf
100
24
DMAc (4), toluene (2)
100
0.11
PCy3
a
Conditions: 2 (2-g scale), Pd(OAc)2 (2 mol %), P(o-tol)3 (2 mol %),
K4[Fe(CN)6]·3H2O (0.4 equiv), Na2CO3 (1 equiv), magnetic stirring
under a nitrogen atmosphere. Detected by HPLC. Conversion was
PPh3
4
89
b
c
P(o-tol)3
2
100
d
a
calculated from the HPLC area. nd: not detected.
Conditions: 2 (100-mg scale), magnetic stirring. All reagents were
charged followed by a nitrogen purge. Conversion was calculated
b
c
from the HPLC area. Detected by HPLC.
While the reaction proceeded sluggishly with only 6%
conversion at 110 °C when toluene was used exclusively as
the solvent (entry 2), the reaction was reproducible in a mixed
solvent of DMAc−toluene. Furthermore, the addition of
toluene reduced the formation of 3 (entry 3). As a result, the
conditions adopted for the reaction included the use of
DMAc−toluene under O2 <1%. The irreproducible results in
DMAc were likely due to oxygen.12−14
a
Table 2. Effect of oxygen and additives
b
cd
,
c
O2
time conversion
3
entry
additives (mol %)
none
(%)
(h)
(%)
(area %)
1
2
3
4
5
6
21
8
3
8
2
2
2
64
0.50
0.32
0.48
6.73
1.19
0.47
none
<0.5
21
100
95
Zn (4), Zn(OAc)2 (4)
e
PMHS (10 wt %)
21
18
Et3SiH (10)
21
16
OPTIMIZATION
■
Zn (4), Zn(OAc)2 (4)
<0.5
100
DoE Studies. Our interest was then directed towards the
optimization of this cyanation process in order to improve the
reaction conversion and reduce the formation of byproduct 3.
DoE is an excellent statistical tool for estimating the main
effects and interactions of several factors in a reaction in order
to truly determine the optimized values and robust ranges.15 In
our previous studies, it was found that the mole fraction of
Pd(OAc)2 and P(o-tol)3 and the composition ratio of toluene
in the DMAc−toluene have an impact on the conversion and
formation of 3.16 Therefore, a Box−Behnken design was
a
Conditions: 2 (1 g scale), Pd(OAc)2 (1 mol %), P(o-tol)3 (1 mol %),
K4[Fe(CN)6]·3H2O (0.8 equiv), DMAc (8 v/w), 135 °C, magnetic
stirring. Oxygen concentration in the gaseous phase. Detected by
HPLC. Conversion was calculated from the HPLC area. PMHS:
polymethylhydrosilane.
b
c
d
e
results. The need to carefully control the oxygen concentration
was not desirable for the manufacturing process.
B
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
employed considering these three parameters (Table 4, entries
1−14).15,17 The ranges were selected based on the results of
our initial studies (Pd(OAc)2: 0.5−2.0 mol %, P(o-tol)3: 0.25−
2.25 mol %, and toluene: 16−50 vol %).
a
Table 4. DoE studies
bc
,
b
Pd(OAc)2
(mol %)
P(o-tol)3
(mol %)
toluene
(vol %)
conversion
(%)
3
entry
(area %)
1
2
0.5
1.25
1.25
1.25
1.25
0.25
0.25
2.25
2.25
0.25
0.25
2.25
2.25
1.25
1.25
16
16
50
50
33
33
33
33
16
50
16
50
33
33
100
0.46
0.32
0.13
0.12
0.29
0.15
0.34
0.16
0.47
0.10
0.31
0.08
0.22
0.18
2.0
100
3
0.5
79.1
99.6
79.1
97.7
99.9
100
4
2.0
5
0.5
6
2.0
7
0.5
8
2.0
9
1.25
1.25
1.25
1.25
1.25
1.25
93.4
81.1
99.9
90.6
99.9
100
10
11
12
13
14
Figure 2. Overlays of graphical analyses showing the vol % of toluene,
mol % of Pd(OAc)2, reaction conversion, and formation of byproduct
3 (P(o-tol)3: 1.9 mol %). The zone within the red area (conversion)
corresponds to 95−100% conversion; the line, 99% conversion. The
zone within the blue area (dimer) represents 0.12−0.20 area %; the
line, 0.15 area %. The circle represents the optimized conditions
(Pd(OAc)2: 1.9 mol %, toluene: 43 vol %; see Supporting
Information).
a
Conditions: 2 (4 g scale), K4[Fe(CN)6]·3H2O (0.4 equiv), Na2CO3
(1 equiv), DMAc−toluene (6 v/w), 125 °C, 8 h, magnetic stirring.
These experiments were performed with vigorous stirring using a
magnetic stirring tip in a test tube. Detected by HPLC. Conversion
was calculated from the HPLC area.
b
c
stirring were not appropriate for larger scale reactions with
impeller stirring (8 h, 65% conversion, 3: 0.22 area %). It was
hypothesized that the use of the different stirring methods
affected the reactive surface area (particle size) of K4[Fe-
(CN)6]. As the cyanation proceeds, the surfaces of K4[Fe-
(CN)6] particles might be converted to a less reactive cyanide
species. With magnetic stirring, however, the K4[Fe(CN)6]
particles are generally ground by the tip, thus exposing fresh
reactive surfaces that can promote the cyanation. An impeller
would not affect the K4[Fe(CN)6] particles in this manner, and
therefore no new reactive surfaces would be exposed as the
reaction progressed.
In fact, the particle size distribution of K4[Fe(CN)6]·3H2O
was a critical factor for the reaction. Weissman reported that
K4[Fe(CN)6]·3H2O with a mean particle size of 312 μm with a
D90 (maximum particle diameter below which 90% of the
sample exists) of <509 μm provided good results, while
K4[Fe(CN)6]·3H2O with a mean particle size of 464 μm and
D90 <822 μm afforded only 5% conversion to their desired
benzonitrile after 10 h.7 In our case, milled K4[Fe(CN)6]·3H2O
was initially used (mean particle size: ∼87.6 μm, D90: ∼188.7
μm), but it was found that K4[Fe(CN)6]·3H2O with a mean
particle size of ∼227 μm and D90 of ∼496 μm exhibited similar
reactivity, and it was commercially available.
The outcomes were analyzed using the JMP 10.0.2 software
package.18 With respect to conversion, beneficial effects of
increasing the mole fraction of Pd(OAc)2 and P(o-tol)3 and
decreasing the toluene ratio were confirmed. On the other
hand, with regard to the formation of 3, it was found that
increasing the toluene ratio and mole fraction of Pd(OAc)2 was
beneficial (Figure 1). These conflicting results were overlapped
in Figure 2 in order to find the optimized conditions; they were
(Pd(OAc)2: 1.9 mol %, P(o-tol)3: 1.9 mol %, and toluene: 43
vol %). In fact, excellent results were obtained using these
conditions (8 h, 100% conversion, 3: 0.07 area %).
Scale-up Studies with a Stirring Impeller. Unfortu-
nately, we found that the optimized conditions using magnetic
Subsequently, the in situ particle size of K4[Fe(CN)6] in the
DMAc−toluene mixed solvent stirred with an impeller was
investigated. A slurry of K4[Fe(CN)6]·3H2O in each of three
different DMAc−toluene solvent mixtures (0, 33, and 43 vol %
toluene) was heated to 85 °C, and the chord length distribution
(CLD) of the particles (square weight) was measured using a
focused beam reflectance measurement (FBRM) probe (see
Supporting Information). Notably, the CLD of the particles in
the 43 vol % toluene solution was larger than that in the 100%
DMAc solution (Figure 3). While the results of the above-
mentioned DoE studies using magnetic stirring were not
Figure 1. Three-dimensional contour plots showing the conversion,
byproduct 3 level, mole fraction of Pd(OAc)2 and P(o-tol)3, and the
toluene composition in the toluene−DMAc mixed solvent.
C
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
the manufacturing of 1 in high yield and high quality on a 117
kg scale.
EXPERIMENTAL SECTION
■
General Information. Solvents and reagents were used
1
without purification or drying. H and 13C NMR spectra were
acquired on a Bruker spectrometer at frequencies of 400 and
100 MHz, respectively. Infrared spectra were acquired on a
PerkinElmer Spectrum One spectrometer. Mass spectra were
recorded on a Waters ZQ-2000 mass spectrometer. HPLC
chromatograms were recorded on a Shimadzu LC-10 system.
Optical rotations were recorded on a PerkinElmer polarimeter
341. Reaction calorimetry was performed using a Mettler
Toledo RC1e vessel and Omnical SuperCRC calorimeter.
Chord length distributions were measured by a Mettler Toledo
FBRM S-400 probe.
Figure 3. Overlays of chord length distribution (CLD) plots for
K4[Fe(CN)6] (2.4 g) in 0, 33, and 43 vol % toluene in DMAc (30 mL)
at 85 °C (square weight).
completely applicable to the reactions using overhead stirring, it
was concluded that the CLD data for the <43 vol % toluene in
DMAc reaction should be relevant because the results obtained
under these conditions with impeller stirring were similar to
those obtained when using K4[Fe(CN)6] of an appropriate
particle size with magnetic stirring.19 Further CLD analyses of
K4[Fe(CN)6] particles in different solvent mixtures with
impeller stirring then revealed that 33 vol % toluene afforded
a CLD for K4[Fe(CN)6] that was similar to that in DMAc. The
optimum conditions for impeller stirring were thus determined
to be Pd(OAc)2, 2 mol %; P(o-tol)3, 2 mol % for 8 h, and using
these conditions, the target compound 1 was obtained with
≥99% conversion, while 3 was formed with 0.18 area %.
4-{(2S)-4-[(1R)-1-Phenylethyl]morpholin-2-yl}benzonitrile
(1). A deoxygenated solution of 2.1 kg of Pd(OAc)2 (9.2 mol)
and 2.8 kg of P(o-tol)3 (9.2 mol) in 302.7 kg of DMAc was
added to a deoxygenated suspension of 160.4 kg of 2 (463.2
mol), 78.1 kg of K4[Fe(CN)6]·3H2O (184.9 mol, PSD: mean
size ∼ 227 μm, D90 ∼ 496 μm), 49.0 kg of Na2CO3 (463.2
mol) in 277.1 kg of toluene, and 305.0 kg of DMAc at 25 °C
(measured O2: 0.2−0.4%). The resulting suspension was heated
to 120−130 °C over 1 h and stirred for 9 h at 125 °C (bubbling
nitrogen from the bottom of the vessel every 1 h).22 After
cooling, the reaction mixture was filtered and washed with
277.2 kg of toluene. The filtrate was washed with 640 kg of
water. The aqueous layer was extracted with 277.0 kg of
toluene, and the combined organic layers were washed with 320
kg of water and concentrated to 480 L. To the concentrate was
added 753.0 kg of IPA, and the resulting mixture was
concentrated to 400 L. The concentrate was combined with
768.4 kg of IPA, and the mixture was heated to 60 °C and
treated with 16.1 kg of activated charcoal. The resulting mixture
was filtered through a Celite pad, and the residue was washed
with 251.3 kg of IPA. The filtrate was concentrated to 400 L
and combined with 260 kg of IPA, and to the resulting solution
was treated with 6.4 kg of N-methyl thiourea. The mixture was
stirred at 55 °C until dissolved. The product crystallized at 40
°C. After cooling to 20 °C, 800 kg of water was dropped into
the slurry. The slurry was aged over 1 h and filtered, and the
filter cake was washed with a solution of 125.8 kg of IPA and
320 kg of water and dried at 60 °C under reduced pressure to
afford 1 as a white solid (116.6 kg, yield 86%, purity 99.4 area
%).
MANUFACTURING
■
To ensure scalability, we confirmed the reaction conditions
with a vessel similar to the reactor on a 50 g scale. Slow heating
during scale-up did not affect catalyst deactivation in spite of
the early introduction of the catalyst and cyanide source at 25
°C.20 After completion, the reaction was filtered to remove the
catalyst and inorganic salts.21 The filtrate was washed with
water and extracted with toluene. The combined extracts were
concentrated, and 1 was obtained by crystallization from
isopropyl alcohol (IPA)−water incorporating N-methyl thiour-
ea as a palladium scavenger in 84% yield.
Finally, a series of scale-up studies led to reproducible
manufacture of 1 in high yields with excellent quality on 48.2 kg
and 116.6 kg scales, regardless of the impeller type (Table 5).
CONCLUSION
■
In conclusion, we developed a practical and reproducible Pd-
catalyzed cyanation in DMAc−toluene while simultaneously
controlling the formation of byproduct 3. Using DoE studies,
the ideal concentration range for Pd(OAc)2 and P(o-tol)3 and
fraction of toluene in the mixed solvent were identified. A high
percentage of toluene was, however, found not to be
appropriate when using impeller stirring because of the in
situ particle size of K4[Fe(CN)6], as revealed using CLD
analysis with FBRM. As a result of this effort, we accomplished
IR (ATR): 2802, 2229, 1610, 1449, 1296, 1119, 1095, 1020,
956, 834, 759, 699, 565 cm−1. 1H NMR (400 MHz, CDCl3): δ
7.66−7.61 (m, 2H), 7.51−7.45 (m, 2H), 7.37−7.24 (m, 5H),
4.65 (dd, J = 10.2, 2.6 Hz, 1H), 3.94 (ddd, J = 11.3, 3.1, 1.8 Hz,
1H), 3.74 (ddd, J = 11.3, 11.3, 2.3 Hz, 1H), 3.38 (q, J = 6.7
Hz,1H), 3.09 (ddd, J = 11.3, 2.6, 2.3 Hz, 1H), 2.62 (dd, J =
11.5, 1.8 Hz, 1H), 2.14 (ddd, J = 11.5, 11.3, 3.1 Hz, 1H), 2.01
a
Table 5. Results on manufacturing scale
reaction
yield
bc
,
c
b
entry
2 scale (kg)
reactor (L)
impeller type
conversion (%)
3 (area %)
kg
%
purity (%)
1
2
68.0
1000
3000
Pfaudler
anchor
100
100
0.10
0.17
48.2
84
86
99.4
99.4
160.0
116.6
a
Conditions: 2, Pd(OAc)2 (2 mol %), P(o-tol)3 (2 mol %), K4[Fe(CN)6]·3H2O (0.4 equiv), Na2CO3 (1 equiv), DMAc−toluene (toluene 33 vol %,
b
c
6 v/w), 125 °C, 8 h. Detected by HPLC. Conversion was calculated from the HPLC area.
D
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(dd, J = 11.3, 10.2 Hz, 1H), 1.40 (d, J = 6.7 Hz, 3H). 13C NMR
(100 MHz, CDCl3): δ 146.0, 143.2, 132.1, 128.3, 127.6, 127.1,
126.8, 118.8, 111.3, 77.5, 67.2, 65.0, 57.3, 50.5, 19.5. LRMS
(ESI) m/z: calcd for C19H20N2O [M + H]+ 293, found 293.
(3) (a) For Pd-catalyzed cyanation reviews, see: Anbarasan, P.;
Schareina, T.; Beller, M. Chem. Soc. Rev. 2011, 40, 5049−5067. (b)
Recent papers using DMF as a cyanide source were described, see:
Kim, J.; Chang, S. J. Am. Chem. Soc. 2010, 132, 10272−10274. (c)
Another cyanation studies in our company, see: Hatsuda, M.; Seki, M.
Tetrahedron 2005, 61, 9908−9917.
(4) (a) Takagi, K.; Okamoto, T.; Sakakibara, Y.; Oka, S. Chem. Lett.
1973, 471−474. (b) Takagi, K.; Okamoto, T.; Sakakibara, Y.; Ohno,
A.; Oka, S.; Hayama, H. Bull. Chem. Soc. Jpn. 1975, 48, 3298−3301.
(5) K4[Fe(CN)6] is used as a table salt additive and for removal of
metal salts from wine; see: (a) WHO Food Additive Series 6,
Calcium, Potassium, and Sodium Ferrocyanide, 1974. (b) Ministry of
Health, Labour and Welfare. http://www.mhlw.go.jp/shingi/2002/
07/s0726-6.html (accessed July 26, 2003). (c) Food Safety
Commission of Japan, http://www.fsc.go.jp/fsciis/
foodSafetyMaterial/show/syu03410320475 (accessed August 9,
2011).
[α]20 +50.7° (c 2.36, CHCl3).
D
(2S)-2-(4-Bromophenyl)-4-[(1R)-1-phenylethyl]morpholine
(2). IR (ATR): 2801, 1601, 1487, 1450, 1294, 1118, 1098,
1
1069, 1008, 957, 917, 809, 758, 699, 550 cm−1. H NMR (400
MHz, CDCl3): δ 7.50−7.44 (m, 2H), 7.36−7.21 (m, 7H), 7.14
(d, J = 8.4 Hz, 2H), 4.56 (dd, J = 10.2, 2.3 Hz, 1H), 3.91 (ddd, J
= 11.3, 3.1, 1.8 Hz, 1H), 3.73 (ddd, J = 11.3, 11.3, 2.3 Hz, 1H),
3.35 (q, J = 6.7 Hz, 1H), 3.07 (ddd, J = 11.3, 2.6, 2.3 Hz, 1H),
2.59 (dd, J = 11.5, 1.8 Hz, 1H), 2.12 (ddd, J = 11.5, 11.3, 3.1
Hz, 1H), 2.03 (dd, J = 11.3, 10.2 Hz, 1H), 1.36 (d, J = 6.7 Hz,
3H). 13C NMR (100 MHz, CDCl3): δ 143.4, 139.8, 131.3,
128.3, 127.9, 127.6, 127.1, 121.5, 77.7, 67.2, 65.1, 57.6, 50.7,
19.6. LRMS (ESI) m/z: calcd for C18H20BrNO [M + H]+ 346,
(6) It was reported that all cyanide groups in K4[Fe(CN)6] can react.
(a) Schareina, T.; Zapf, A.; Beller, M. J. Organomet. Chem. 2004, 689,
4576−4583. (b) Schareina, T.; Zapf, A.; Beller, M. Chem. Commun.
found 346. [α]20 +44.5° (c 2.00, EtOH).
D
(2S,2′S)-2,2′-Biphenyl-4,4′-diylbis{4-[(1R)-1-phenylethyl]-
morpholine} (3). IR (ATR): 2796, 1741, 1492, 1452, 1288,
1121, 1098, 957, 807, 760, 700, 552 cm−1. 1H NMR (400 MHz,
CDCl3): δ 7.59−7.54 (m, 4H), 7.47−7.42 (m, 4H), 7.35−7.23
(m, 10H), 7.14 (d, J = 8.4 Hz, 4H), 4.65 (dd, J = 10.2, 2.3 Hz,
2H), 3.94 (ddd, J = 11.3, 3.1, 1.8 Hz, 2H), 3.76 (ddd, J = 11.3,
11.3, 2.3 Hz, 2H), 3.37 (q, J = 6.7 Hz, 2H), 3.16 (ddd, J = 11.3,
2.6, 2.3 Hz, 2H), 2.61 (dd, J = 11.5, 1.8 Hz, 2H), 2.22−2.09 (m,
4H), 1.38 (d, J = 6.7 Hz, 6H). 13C NMR (100 MHz, CDCl3): δ
143.6, 140.3, 139.8, 128.3, 127.6, 127.0, 127.0, 126.7, 78.2, 67.3,
65.2, 57.7, 50.8, 19.7. LRMS (ESI) m/z: calcd for C36H40N2O2
2004, 1388−1389. (c) Schareina, T.; Zapf, A.; Magerlein, W.; Muller,
̈
̈
N.; Beller, M. Tetrahedron Lett. 2007, 48, 1087−1090.
(7) Weissman, S. A.; Zewge, D.; Chen, C. J. Org. Chem. 2005, 70,
1508−1510.
(8) (a) Dobbs, K. D.; Marshall, W. J.; Grushin, V. V. J. Am. Chem.
Soc. 2007, 129, 30−31. (b) Erhardt, S.; Grushin, V. V.; Kilpatrick, A.
H.; Macgregor, S. A.; Marshall, W. J.; Roe, D. C. J. Am. Chem. Soc.
2008, 130, 4828−4845. Catalyst deactivation was also reported in the
following papers: (c) Takagi, K.; Okamoto, T.; Sakakibara, Y.; Ohno,
A.; Oka, S.; Hayama, H. Bull. Chem. Soc. Jpn. 1976, 49, 3177−3180.
(d) Sakakibara, Y.; Sasaki, K.; Okuda, F.; Hokimoto, A.; Ueda, T.;
Sakai, M.; Takagi, K. Bull. Chem. Soc. Jpn. 2004, 77, 1013−1019.
(e) Marcone, J. E.; Moloy. J. Am. Chem. Soc. 1998, 120, 8527−8528.
(f) Ryberg, P. Org. Process Res. Dev. 2008, 12, 540−543. (g) Gerber, R.;
Oberholzer, M.; Frech, C. M. Chem.Eur. J. 2012, 18, 2978−2986.
(9) Frech reported that [Pd(CN)4]2− formed slowly using
K4[Fe(CN)6] as a cyanide source in ref 8e.
[M + H]+ 533, found 533. [α]20 +68.0° (c 4.83, CHCl3).
D
ASSOCIATED CONTENT
* Supporting Information
■
S
Experimental details and product characterization. This materi-
al is available free of charge via the Internet at http://pubs.acs.
org.
(10) (a) Martin, M. T.; Liu, B.; Cooley, B. E., Jr.; Eaddy, J. F.
Tetrahedron Lett. 2007, 48, 2555−2557. (b) Chidambaram, R.
Tetrahedron Lett. 2004, 45, 1441−1444. (c) Ren, Y.; Liu, Z.; He, S.;
Zhao, S.; Wang, J.; Niu, R.; Yin, W. Org. Process Res. Dev. 2009, 13,
764−768.
(11) For a review on the synthesis of biaryls with zinc, see: Hassan, J.;
Sevignon, M.; Gozzi, C.; Schultz, E.; Lemaire, M. Chem. Rev. 2002,
102, 1359−1469.
(12) The effect of toluene was unclear, but it worked as expected; in
addition, it also worked as a degassing agent during refluxing.
Degassing due to refluxing toluene may prolong the catalyst life.
According to Merritt, dissolved oxygen should be purged under
refluxing; see: Merritt, J. M.; Buser, J. Y.; Campbell, A. N.; Fennell, J.
W.; Kallman, N. J.; Koenig, T. M.; Moursy, H.; Pietz, M. A.; Scully, N.;
Singh, U. K. Org. Process Res. Dev. 2014, 18, 246−256.
(13) In our earlier studies, we observed uneven results using different
batches of 2 in DMAc. It was thought that impurities in 2 which may
have increased the solubility of the cyanide source were causing
catalyst deactivation, but critical impurities for catalyst deactivation
were not identified. In addition, it should be noted that in the case of
K4[Fe(CN)6]·3H2O, unlike that of Zn(CN)2, NaCl, 1-phenetylamine,
and 4 did not hinder the reaction. For the catalyst deactivation using
Zn(CN)2 with NaCl and dimethylamine, see: Marcantonio, K. M.;
Frey, L. F.; Liu, Y.; Chen, Y.; Strine, J.; Phenix, B.; Wallace, D. J.;
Chen, C. Org. Lett. 2004, 6, 3723−3725.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Utsugi.Masayuki@mp.mt-pharma.co.jp.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We are grateful to API Corporation for manufacturing. We also
thank Mitsubishi Chemical Group Science and Technology
Research Center for residual metal and free cyanide analysis,
Dr. Hajime Hiramatsu for NMR, LC-MS, single crystal XRD,
and IR analysis, Mr. Rikizo Furuya for initial studies, Dr.
Hiroshi Iwamura, Mr. Takashi Naka, Mr. Takao Teshima, Mr.
Takafumi Yamagami, and Mr. Toshiaki Shimizu for helpful
discussions.
REFERENCES
■
(1) Kleemann, A.; Engel, J.; Kutscher, B.; Reichert, D. Pharmaceutical
Substances: Syntheses, Patents, Applications, 4th ed.; Georg Thieme,
Verlag: Stuttgart, 2001.
(14) In our kinetic studies using reaction calorimeter SuperCRC, the
trend of catalytic deactivation in the cyanation with 1.0 equiv of
K4[Fe(CN)6]·3H2O was not observed. The studies described that
K4[Fe(CN)6]·3H2O accelerated the cyanation than the catalyst, which
may suggest that mass transfer of K4[Fe(CN)6] is rate-limiting step.
The reaction stalled using 0.2 equiv of K4[Fe(CN)6]·3H2O at ∼70%
conversion. For our kinetics studies, see Supporting Information.
(2) (a) Watanabe, K.; Kohara, T.; Fukunaga, K.; Uehara, F. World
Patent WO/2008/078837 A1, 2008. (b) Fukunaga, K.; Watanabe, K.;
Barneoud, P.; Benavides, J.; Pratt, J. World Patent WO/2009/035162
A1, 2009. (c) Furuya, R.; Ozawa, H.; Toyofuku, E.; Kusaka, S.;
Iwamura, H.; Sakai, D.; Nakayama, D.; Watanabe, K. World Patent
WO/2010/104193 A2, 2010.
E
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(15) For an example of a book about DoE, see: Carlson, R.; Carlson,
R. Design and Optimization in Organic Synthesis, 2nd ed.; Elsevier:
Amsterdam, 2005.
(16) The stirring was carried out under a fast impeller speed because
it was critical for the reaction. Byproduct 3 increased under a slow
impeller speed, and its formation depended on the mixing efficiency in
the reactor. Although the amount of water also had a great impact on
reaction rate, we did not further investigate for avoiding increase of the
cyanide source.
(17) Box, G. E. P.; Behnken, D. W. Ann. Math. Stat. 1960, 31, 838−
864.
(18) We used JMP 10.0.2 (SAS) and STAVEX (Aicos) as statistical
analysis software.
(19) We did not have enough time to re-examine the DoE studies
using overhead stirring.
(20) Some papers suggested that early addition of the cyanide source
causes catalyst deactivation; see: Ryberg, P. Org. Process Res. Dev. 2008,
12, 540−543 and ref 10c.
(21) Free cyanide was not detected in the corresponding filtrate by
ion chromatography (detectable limit <0.5 ppm). An analytical
method using ion chromatography was referred to ref 13. The residue
was disposed of as toxic waste.
(22) It was observed that heavy inorganic salts (e.g., K4[Fe(CN)6])
remained at the bottom of the vessel just below the impeller, even with
vigorous overhead stirring. We thus were concerned about the mixing
efficiency and clogging of the reactor by these salts during the workup.
F
dx.doi.org/10.1021/op400204g | Org. Process Res. Dev. XXXX, XXX, XXX−XXX