Industrial-scale palladium-catalyzed coupling of aryl halides and amines - A personal account

Article abstract of DOI:10.1002/adsc.200505158

The palladium-catalyzed coupling of amines and aryl halides or aryl alcohol derivatives has matured from an exotic small-scale transformation into a very general, efficient and robust reaction during the last ten years. This article reports several applications of this method from an industrial vantage point, including ligand synthesis, synthesis of arylpiperazines, arylhydrazines and diarylamines. Much emphasis in placed on issues of scale-up and safety to underline the potential of C-N couplings as solutions for industrial-scale synthetic problems.

Full text of DOI:10.1002/adsc.200505158

REVIEW
DOI: 10.1002/adsc.200505158
Industrial-Scale Palladium-Catalyzed Couplingof Aryl Halides
and Amines – A Personal Account
Stephen L. Buchwald,a Christelle Mauger,b, e Gerard Mignani,b,* Ulrich Scholzc, d,*
a
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
Fax: (þ1)-617-253-3297, e-mail: sbuchwal@mit.edu
b
`
Rhodia, Centre de Recherche de Lyon, 85 rue des Freres Perret, BP62, 69192 Saint-Fons Cedex, France
Fax: (þ33)-4-72-89-68-55, e-mail: gerard.mignani@eu.rhodia.com
Lanxess Deutschland GmbH, Fine Chemicals Research & Development, 51368 Leverkusen, Germany
Current address: Boehringer Ingelheim Pharma GmbH & Co. KG, Dept. Process Development, 55216 Ingelheim am
Rhein, Germany
c
d
E-mail: ulrich.scholz@ing.boehringer-ingelheim.com
Current address: Reckitt Benckiser, Dauson Lane, Hull HU8 7DS, United Kingdom
E-mail: christelle.mauger@reckittbenckiser.com
e
Received: April 18, 2005; Accepted: July12, 2005
Abstract: The palladium-catalyzed coupling of amines
and aryl halides or aryl alcohol derivatives has ma-
tured from an exotic small-scale transformation into
a verygeneral, efficient and robust reaction during
the last ten years. This article reports several applica-
tions of this method from an industrial vantage point,
including ligand synthesis, synthesis of arylpipera-
zines, arylhydrazines and diarylamines. Much empha-
sis in placed on issues of scale-up and safetyto under-
À
4
5
The Preparation of Arylpiperazines as a Model
Process for the Scale-Up of Palladium-Catalyzed
Aromatic Amination
Synthesis of Industrially Important Diarylamines
À
byC N coupling
6
7
8
9
Synthesis of the Dialkylphosphinobiaryl Ligands
Synthesis of Chlorodicyclohexylphosphine
Synthesis of Ligands A and C
Synthesis of Ligand B
line the potential of C N couplings as solutions for in- 10 Removal/Recycling of Palladium after the Reac-
dustrial-scale synthetic problems.
tion
11 Conclusions and Prospects for Future Work
1
2
3
Introduction
Motivation: Technologygives an Edge
The Synthesis of Arylhydrazones and their use as Keywords: amination; anilines; C N cross-coupling;
Synthetic Intermediates
À
homogeneous catalysis; palladium; phosphane ligands
1 Introduction
rials, polymers, liquid crystals and xerographic materi-
als.[2]
The development of a new, general methodologyfor the
Over the first ten years since its development, small-
formation of aromatic carbon-nitrogen bonds is of great scale applications of the palladium-catalyzed aromatic
significance in manyareas of organic synthesis. As the amination have appeared in publications from academ-
Heck, Suzuki–Miyaura, Stille, Kumada and Negishi pro- ic, medicinal and process chemistrylaboratories. This
cedures have revolutionized the waywe construct sp2C– method is able to connect highlyfunctionalized compo-
sp2C bonds,[1] the palladium-catalyzed aromatic amina- nents and is compatible with a wide varietyof functional
tion has the potential to modifyanalogouslyour ap-
groups. This generalitymakes it well-suited for the con-
proach to the synthesis of aniline derivatives. Aromatic struction of a bodyof amine analogues from a common
carbon-nitrogen bonds are structural constituents for intermediate. The versatilityand applicabilityof the
the preparation of compounds of interest to those method has been thoroughlydocumented in two recent
in the pharmaceutical and agrochemical industries. reviews.[3]
In addition, compounds with this motif are impor-
tant for the preparation of new ligands, electronic mate-
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
23

Stephen L. Buchwald et al.
REVIEW
Stephen L. Buchwald was
born (1955), raised and re-
ceived his precollege edu-
cation in Bloomington, In-
diana. He received his
Sc. B. degree, Magna Cum
Laude, in chemistry, from
Brown Universityin 1977.
During his undergraduate
years he worked in the lab-
oratories of Professors
Kathlyn A. Parker and
David E. Cane at Brown Universityand Professor
Christelle C. Mauger was
ˆ
born in Saint-Lo in Nor-
mandy(France) in 1974.
She completed her under-
graduate degree at the
Universityof Caen and ob-
tained her Ph. D. degree in
organic chemistryin 2000
under the guidance of Dr
Serge Masson (Labora-
´
toire de Chimie Molecu-
laire et Thioorganique,
Caen, France). She then moved to Huddersfield
Gilbert Stork at Columbia University. He entered (England) to join Avecia Pharmaceuticals for an in-
Harvard Universityas a National Science Foundation dustrial postdoctoral position. She worked on the
Predoctoral Fellow in 1977 and received his Ph. D. in rapid development of new routes to pharmaceutical
1982. His thesis work, under the supervision of Pro- intermediates and on the “CaTHy” catalytic transfer
fessor JeremyR. Knowles, concerned the mechanism
hydrogenation reaction. In 2001 she accepted a post-
of phosphoryl transfer reactions in chemistry and bio- doctoral position granted byRhodia Organic at the
chemistry. In early 1982 he took up a position as a Universityof Poitiers (Laboratoire de Catalyse en
Myron A. Bantrell postdoctoral fellow at the Califor- Chimie Organique, France) where she worked in
nia Institute of Technologywhere he worked in the
laboratoryof Professor Robert H. Grubbs. His
work at Caltech concerned the studyof titanocene
the fields of fluorine chemistry. In 2002, she joined
Rhodia Recherches (Lyon Research Centre) as a re-
search engineer in the Process Research Group. Her
methylenes as reagents in organic synthesis. During main area of activities was the development of pharma-
this time he was also involved in work on the mech- ceutical intermediates using organometallic catalysis
anism of the Ziegler–Natta polymerization. In 1984 and aromatic bond forming reactions (ABF). In Janu-
he began as an assistant professor of chemistryat
ary2006, she moved to the Reckitt Benckiser scientific
the Massachusetts Institute of Technology. He was services department in Kingston-upon-Hull (UK).
promoted to the rank of associate professor in 1989
´
and to Professor in 1993. He was named the Camille Gerard Mignani studied
Dreyfus Professor of Chemistry in January of 1997. chemistryin Orsayand
During his time at MIT he has received numerous Rennes Universities where
honors including the Harold Edgerton FacultyAch-
ievement Award of MIT, an Arthur C. Cope Scholar (Docteur Ingenieur) in
he received his Ph. D.
´
`
Award, the 2000 Award in Organometallic Chemistry 1980 and his “These dꢁE-
from the American Chemical Society, a MERIT tat” in 1982 in the field of
award from the National Institutes of Health, the organometallic chemistry
Bristol-Myers Squibb Distinguished Achievement and homogeneous cataly-
Award in Organic Synthesis (2005), the CAS Spot- sis especiallyin steroid
ˆ
light Award (2005) and the American Chemical Soci- chemistry(Rho ne-Poulenc
etyAward for Creative Work in Synthetic Organic
Chemistry(2006). In 2000, he was elected as a mem-
ber in the American Academyof Arts and Sciences.
Grant) in Professor Da-
bardꢁs group. He joined the Rhone-Poulenc Research
group in Lyon in 1980 where he developed new proc-
ˆ
He has been a named lecturer at numerous universi- esses in organic and terpene chemistryand in homo-
ties. He is the coauthor of 245 published or accepted geneous and heterogeneous catalysis. Thereafter he
papers and 30 issued patents. He serves as a consul- performed postdoctoral research with Professor D.
tant to Amgen, Merck, InfinityPharmaeuticals, Rho-
Seyferth at the Massachusetts Institute of Technology
dia Pharmaceutical Solutions, Lanxess, Englehard (Cambridge, U. S. A.) on ceramic precursors and or-
ˆ
and Collegium Pharmaceuticals. He and Eric Jacob- ganosilicon chemistry. He came back to Rhone-Pou-
sen have co-taught a short course on Organometallic lenc research where he developed new ceramic pre-
Chemistryin Organic Synthesis at over 35 companies
in the U. S. A. and in Europe.
cursors, new non-linear optic derivatives, polymers,
and homogeneous catalysis processes. He spent ten
years as a Group Leader in Silicon Chemistry. His re-
search interests were the poly-functionalization of
24
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
polysiloxanes, new organometallic catalysis for orga- chemistryand received his diploma degree in Profes-
nosilicon applications and the functionalization of sor E. Winterfeldtꢁs research group in 1996 on the
mineral charges. Now, his research interests are the synthesis of unsymmetrical pyrazines. He stayed in
new processes and scale-up in organic chemistry, or- Professor Winterfeldtꢁs group and spent the subse-
ganometallic catalysis (homogeneous and heteroge- quent three years on the elaboration of bile acids as
neous) and new methodologyin chemistrysynthesis.
potential building blocks for the synthesis of cepha-
He received the “Prix de la Recherche” in 1995, in lostatin analogues. After graduation he continued as
2001 “Prix RHODIA Group” and in 2004 “Prix Cen- a teaching assistant at Hannover Universityto finally
tre de Recherches-RHODIA”.
join the Central Research Department for homoge-
neous catalysis of Bayer AG at the end of 1999. In
2002, he moved to the process development depart-
ment of the Bayer Chemicals Company, now Lanxess
– Fine Chemicals as a project manager of the Special-
ityChemicals Department. Since 2001, he also head-
ed the Bayer part of an industrial collaboration be-
tween Bayer Chemicals, Professor Steve L. Buchwald
from MIT and Rhodia Pharma Solutions (formerly
Rhodia Chirex) on palladium-catalyzed aromatic
aminations. In 2005, he moved to Boehringer-Ingel-
heimꢁs Process Development Department.
Ulrich Scholz started his
studies in chemistryin
1990 at the Universityof
Hannover, Germany. In
1993, he joined Professor
Paul A. Wenderꢁs research
group at Stanford Univer-
sity, California, USA on a
DAAD scholarship. After
returning to his home
town of Hannover, he fin-
ished his basic studies in
2 Motivation: Technology gives an Edge
confidential. To further this endeavor, one of us (SLB)
was brought in as a consultant for this venture (Figure 1).
We first identified three keysuccess factors: 1) Estab-
lishment of a quick method for the optimization of the
technology. 2) The scale-up of several model targets. 3)
With the increasing competition between custom manu-
facturers of chemicals, one wayfor a companyto differ-
entiate itself is byquick reduction to practice of a new
and powerful technology. To focus on a process as new A safe and reliable supplyroute to the dialkylphosphi-
as the palladium-catalyzed aromatic amination, the lev- nobiaryl ligands that are used for these reactions
el of risk is high for one companyto bear on its own.
Therefore a joint venture between two companies, Bay-
(Scheme 1).
This model, in our opinion, has the potential to speed
er Chemicals, now Lanxess, and Rhodia Pharma Solu- the transfer ofpromising new technologyfrom academic
tions, was undertaken. We decided to join efforts with
complete exchange of knowledge gained byboth sides.
Since confidentialityplays a major role in custom man-
ufacturing, it was decided to set up specific goals inside
the collaboration, while all customer inquiries and proj-
ects remained outside of the collaboration and therefore
Figure 1. Collaboration model: two companies, one universi-
ty.
Scheme 1. Important members of the dialkylphosphinobiaryl
ligand family.
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
25

Stephen L. Buchwald et al.
REVIEW
discoveryinto industrial application. Over the course of
two years, the participants have had a good chance to
scrutinize the chemistryand decide whether it is adapt-
able to industrial requirements. For the palladium-cata-
lyzed aromatic amination reaction, this strategy has
worked well. All three of the goals mentioned above
have been realized. In addition, several customer proj-
ects, not a part of the information exchange or collabo-
ration, have been carried out on a pilot scale.
3 The Synthesis of Arylhydrazones and
their use as Synthetic Intermediates
As an initial example of a class of compounds to be in-
vestigated during this collaboration, we studied the ap-
plication of this methodology to the synthesis of arylhy-
drazones. These are precursors to arylhydrazines and,
thus, to various heterocycles, including indoles via the
Fischer indole synthesis (Scheme 2).
Hydrazines are normally accessed via the intermedia-
cyof diazonium salts. For compounds that contain sensi-
tive functional groups, the palladium-catalyzed route
can be an important alternative.
Scheme 3. Potential applications of arylhydrazines.
Table 1. Selection of pharmaceuticallyimportant indoles.
EntryName
Company Use
1
2
3
4
5
Sumtriptan
GSK
GSK
Novartis
Anti-Migraine
Anti-Emetic
Anti-Cholesterol
Odensetron
Fluvastatin
Zafirlukast
AstraZeneca Asthma
Anti-Inflammatory
Literallythousands of indoles have been prepared and
evaluated as potential pharmaceuticals.[4] A few clinical-
lyimportant examples are shown in Table 1.
Indomethacin Merck
The development of a viable, robust and safe industri-
al process is a challenging problem and a prerequisite for dustrial scale. We first examined a modification of the
large-scale applications of this reaction. We have stud- previouslyreported conditions [5] (Scheme 4).
ied, as a model reaction, the coupling of p-bromotoluene
Screening experiments showed that catalysts based on
or p-chlorotoluene with benzophenone hydrazone in or- dialkylphosphinobiaryl ligands gave the best perform-
der to examine the important parameters for the prepa- ance. With A or B, the coupling reaction was achieved
ration of an N-arylbenzophenone hydrazone on an in- in 94% and 93% yield, respectively, using 1 mol %
Scheme 2. Fischer indole synthesis.
26
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
ing material to product. Given these results, the use of
tert-butanol as the reaction solvent was investigated.
With this solvent, either NaO-t-Bu or NaOH as base
both led to quantitative yield of product.
In view of our results using NaOH in tert-butanol, we
examined the use of a varietyof solvents (Table 3). Us-
ing refluxing tert-amyl alcohol, which has a higher boil-
ing point than tert-butanol, the standard reaction took
place even more rapidlythan before. The use of other
solvents such as ethylene glycol, diethylene glycol,
N,N-dimethylethanolamine, n-butanol and 1,2-dime-
thoxybenzene resulted in low conversions (<30%) of
starting material.
To have an economicallyviable process it was necessa-
ryto reduce the level of catalyst that was necessary. With
C/Pd(OAc)2 in xylene and NaO-t-Bu as base, the reac-
tion reached completion within one hour using 0.1 mol
% catalyst. In order to further improve the efficiency
of the coupling process, we next examined the use of
NaOH as the base. Unfortunately, when using finely
ground NaOH in refluxing toluene with or without a
phase-transfer catalyst, no reaction was observed.
When the reaction solvent was changed to tert-amyl al-
cohol, however, with A as supporting ligand, excellent
results were obtained with solid sodium hydroxide pel-
lets as the base. In xylene, the use of NaO-t-Bu as base
provided the most active system (Table 4). In both cases
the reaction was rapid and no by-products such as ben-
zophenone, benzophenone azine or compounds result-
ing from Wolff–Kishner reduction were detected. In
À
Scheme 4. Indole synthesis by palladium-catalyzed C N cou-
pling.
Table 2. Variation of base and solvent in arylhydrazone for-
mation.
Entry[a] Base
Solvent Time [h] GC Yield [%][b]
1
2
3
4
5
6
7
8
NaOH
Cs2CO3
KO-t-Bu
NaO-t-Am Toluene
NaO-t-Bu Toluene
NaO-t-Bu t-BuOH
KO-t-Bu
NaOH
Toluene 20
Toluene 20
Toluene 20
0
40
<10
100
100
100
20
6
6
6
6
5
t-BuOH
t-BuOH
100
[a]
0.5 mol % Pd(OAc)2, 1 mol % A, 1.4 equivs. base, 1.0
equiv. benzophenone hydrazone, solvent, reflux.
GC Yields determined with dodecane as internal standard.
these procedures only0.05 mol % of Pd(OAc) was
2
[b]
needed for full conversion of chloro- or bromotoluene
to the desired product. Also of note is that the process
Pd(OAc)2 in refluxing toluene with NaO-t-Bu as the was complete after twelve minutes with 0.25 mol % Pd.
base. Thus, both the combinations C/NaO-t-Bu/xylene and
We next examined the effect of changing the base, us- A/NaOH/tert-amyl alcohol represent very good condi-
ing Pd(OAc)2/2 A in toluene or xylene at reflux.[6] With tions for the coupling reaction between aryl halides
potassium tert-butoxide, cesium carbonate or sodium and benzophenone hydrazone.
hydroxide in toluene, low conversion of starting materi-
In fact, the coupling of benzophenone hydrazone with
al (<10%) was observed. In contrast, the use of NaO-t- p-tolyl chloride or bromide in refluxing tert-amyl alco-
Bu or NaO-t-Am provided complete conversion of start- hol using 0.05 mol % of Pd(OAc)2, 0.1 mol % of A
Table 3. Solvent screening for arylhydrazone formation.
Entry[a]
Solvent
Time [h]
T [ 8C]
GC Yield [%][b]
1
2
3
4
5
6
7
8
9
Toluene
Xylene
t-BuOH
t-AmOH
MeOCH2CH2OH
Glycol
Diethylene glycol
N,N-Dimethylethanolamine
Dimethoxybenzene
n-BuOH
6
6
5
4
4
20
20
20
20
20
110
110
90
103
100
100
100
100
100
100
0
0
100
100
100
~30
~30
~30
0
10
15
[a]
0.5 mol % Pd(OAc)2, 1 mol % A, 1.4 equivs. NaOH, 1 equiv. benzophenone hydrazone, solvent, reflux.
GC Yields determined with dodecane as internal standard.
[b]
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
27

Stephen L. Buchwald et al.
REVIEW
Table 4. Time to reach completion of reaction of aryl halide
with benzopheone hydrazone under varying conditions.
[a]
EntryConditions
Pd [mol %]
Time[b] [h]
1
2
3
4
5
6
a
a
a
b
b
b
0.05
0.1
0.25
0.05
0.1
30
1
0.2
4
2
1
0.25
[a]
Conditions a: Pd(OAc)2, ligand C, refluxing xylene, NaO-
t-Bu.
Pd(OAc)2, ligand A, refluxing tert-amyl alcohol, NaOH.
Scheme 5. Arylhydrazone formation by palladium-catalyzed
[b]
À
C N coupling.
Table 5. Yields of arylhydrazones from different starting ma-
terials.
Scheme 5 shows the results using p-CF3PhCl as starting
material.
Entry[a]
Substrate
GC Yield [%][b]
Good results were obtained when the reaction was
performed in the presence of 0.5 mol % of Pd(OAc)2,
1 mol % of C in anisole at 1108C using one equivalent
of benzophenone hydrazone and p-CF3PhCl; use of K3
PO4 (1.4 equivs.) as base gave the best results. After 24
hours of stirring, we observed a quantitative conversion
of aryl chloride to product. Upon crystallization from
ethanol, the N-(4-trifluoromethylphenyl)-benzophe-
none hydrazone was isolated in 88% yield. With a yield
1
2
3
4
5
6
7
8
4-Bromotoluene
4-Bromobenzene
4-Bromochlorobenzene
4-Bromoanisole
2-Bromotoluene
4-Bromofluorobenzene
4-Chlorotoluene
4-Methylphenyl triflate
92
95
97
87
85
91
93
0
[a]
0.5 mol % Pd(OAc)2, 1 mol % A, aryl halide:benzophe- of 93%, similar results were obtained using KOH or so-
dium tert-amylate.[8]
none hydrazone¼1: 1, 1.4 equivs. NaOH, p-MePhBr: 1.5
M in tert-amyl alcohol reflux.
Before applying this protocol on scale, calorimetric
data were collected using a Mettler RC1 apparatus in or-
der to assess safetyissues. The coupling of 4-chloroto-
[b]
GC Yields determined with dodecane as internal standard.
with 1.4 equivs. of sodium hydroxide as base was com- luene with benzophenone hydrazine was used as the
plete in less than two hours.
model reaction. The reaction is veryexothermic
Upon work-up, the crude solution was directlytreated
(DH¼409 kJ/mol), an amount that could potentially
with water to remove inorganic salts. The layers were evaporate 78% of the solvent if the reaction occurred
separated and the aryl hydrazone crystallized from the spontaneously. However, we found that the reaction is
organic layer which had been cooled to 38C. This afford- not instantaneous, with only43% of the reaction heat
ed the p-tolylphenylhydrazone as a pale yellow powder generated bythe end of the addition of the benzophe-
in 92% yield.
none hydrazone. This observation indicates that it is nec-
This protocol was applied to the coupling of other essaryto introduce the benzophenone hydrazone slowly
halogenated aromatics in a 500-mL glass-jacketed ves- to prevent build-up of thermal potential.
sel; results are shown in Table 5. The yields reported cor-
respond to material isolated bycrystallization. In each
case, the reaction proceeded in less than four hours. Al- spectrometry.
The reaction protocol was then studied in an 18-liter
glass reactor and its progress was monitored byRaman
[9]
though 4-chlorotoluene is an excellent substrate, the re-
action is completelyselective for substitution of the bro-
This apparatus can be used in the heterogeneous media
and the probe is stable to water. As shown below, benzo-
mine when using bromochlorobenzene as substrate (Ta- phenone hydrazone, 4-bromotoluene and the final aryl-
ble 5, Entry3). Using this procedure, unfortunately, aryl hydrazone each have characteristic bands appearing at
triflates were not converted to product. In fact, using 4- 180 cmÀ1, 290 cmÀ1 and 1620 cmÀ1, respectively.
methylphenyl triflate, less than 5% product was ob- When the reaction was complete, water was added to
served. In all likelihood, competitive cleavage to the dissolve the salts that had formed. The layers were sep-
phenol occurs at a rapid rate. arated at a temperature of about 408C. Further cooling
Trifluoromethylphenylhydrazines are important in- of the tert-amyl alcohol solution afforded crystals of the
termediates for the preparation of pharmaceuticals.[7] arylhydrazone that could be isolated in quantitative
Thus, we queried whether we could applyour coupling
method to the preparation of these compounds.
yield as a pale yellow solid (Figure 3).
Since N-arylhydrazones cannot be stored indefinitely,
it is more advantageous to isolate and store the corre-
28
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
Figure 2. In situ continuous Raman analysis of final hydrazone formation.
Table 6. Screening of acids for the conversion of the arylhy-
drazone to the arylhydrazine hydrochloride.
Acid
T [ 8C]
Time [h]
Conversion [%]
HCl (10%)
HCl (37%)
HCl (37%)
Acetic acid
PTSA
80
80
r.t.
80
80
2
2
20
2
20
100
100
75
2
100
and ethanol (90:10) at room temperature led to the pre-
cipitation of the hydrazine salt, so that simple filtration
and washing of the solid several times afforded the prod-
uct as a white powder in 92% yield. The reaction was
then examined in a 16-liter Büchi steel reactor with
2.5 kg of N-p-tolylbenzophenone hydrazone in 10 liters
of concentrated hydrochloric acid (Figure 4). The hy-
drazine salt could be isolated from the crude solution
in quantitative yield. The palladium contamination of
the sample was determined to be less than 10 ppm.[10]
Safetydata were also obtained on the hydrazine salt
bydifferential thermal analysis (DTA), which showed
an endothermic behavior at 1008C. This corresponds
to the melting point of the salt. Exothermic behavior
was again observed at 1108C[11] which represents the
thermal decomposition of the salt. From these data we
can discern that the salt is stable at room temperature
and can be easilystored and handled for months. The
precipitation reaction itself is not veryexothermic and
can also be regarded as safe.[12]
Figure 3. Glass reactor (18 liters) for synthesis of arylhydra-
zones.
sponding hydrochloride of the arylhydrazine. The prep- 4 The Preparation of Arylpiperazines as a
aration of p-tolylhydrazine·HCl can be affected by
treatment of the arylhydrazone with acid. Different
sets of conditions were screened and the results are sum-
marized in Table 6. For reactions performed at 808C,
several by-products were generated.
Model Process for the Scale-Up of
Palladium-Catalyzed Aromatic
Amination
A second process selected to examine at scale was the
Treatment of N-p-tolylbenzophenone hydrazone in a synthesis of arylpiperazines. The advantages of using
mixture of concentrated aqueous hydrochloric acid the palladium-catalyzed methodology in lieu of the clas-
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
29

Stephen L. Buchwald et al.
REVIEW
sical reaction of anilines with activated diethanolamines
include: 1) No toxic intermediates are required. 2) A
wide range of aryl halide starting materials, which are
available in bulk, can be used. 3) No side reactions other
than diarylation are observed. Compared to alternative
processes such as the nucleophilic aromatic substitution
of, for example, fluoroaromatics, copper- or nickel-cata-
lyzed processes, the palladium-catalyzed coupling al-
lows the use of milder reaction conditions, has a broader
substrate scope and generallyresults in higher chemical
yields.
As indicated in Table 7, the arylpiperazine structural
motif is common among pharmaceutical substances.
The list of pharmaceuticals alreadylaunched in the mar-
ket containing an arylpiperazine subunit, while not dem-
onstrating the applicabilityof palladium-catalyzed aro-
matic amination, indicates the importance of molecules
with this substructure. Additionally, 39 substances are
currentlyin clinical or preclinical trials. [13] Thus, future
syntheses of these or structurally related molecules
might well be undertaken using a palladium-catalyzed
aromatic amination reaction.
Three arylpiperazine-containing molecules were ini-
tiallyselected as targets. These were chosen such that
the corresponding aniline derivatives either were not
available on scale or could onlybe purchased at a signif-
icantlyhigher cost than the corresponding aryl chloride.
The aromatic amination in these cases not onlyoffers a
new set of starting materials in the making of arylpiper-
azines, it is also an approach with economic advantages
over traditional methodology.
Figure 4. Inox 16-liter Büchi reactor used for hydrolysis of
the arylhydrazone.
Table 7. Arylpiperazines launched in the pharmaceutical market.
EntryName
Company
Use
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Aripiprazole
Ciprofloxacin
Dapiprazole
Etoperidone
Fleroxacin
Iomefloxacin
Itraconazole
Ketoconazole
Levodropropizin
Levofloxacin
Naftopidil
Otsuka
Bayer
Angelini
Angelini
Kyorin
Abbott
J&J
J&J
Dompe
Daiichi
Roche
Bristol-Myers Squibb
Sanofi-Synthelabo
Kyorin
Neuroleptic
Antibacterial
Antiglaucoma
Antidepressant
Antibacterial
Antibacterial
Antifungal
Antifungal
Antitussive
Antibacterial
Antihypertensive
Antidepressant
Insomnia
Antibacterial
Antibacterial
Antibacterial
Antibacterial
Antibacterial
Antifungal
Nefazodone
Niaprazine
Norfloxacin
Ofloxacin
Daiichi
Aventis
Pefloxacin
Prulifloxacin
Rufloxacin
Terconazole
Trazodone
Urapidil
Vesnarinone
Nippon Shinyaku
Mediolanum
J&J
Angelini
Altana
Antidepressant
Antihypertensive
HIV
Otsuka
30
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
yield of 97% could be achieved even with this smaller
excess of piperazine (Entry3). In tert-butanol, a 93%
yield of desired product was obtained (Entry 4). At-
tempts to reduce the catalyst loading in toluene resulted
in incomplete conversion of the starting material using
P(t-Bu)3 as ligand (Entry5).
Since the aryl chloride, in this case, is available in bulk
quantities, and is therefore far better suited for use in an
industrial process, the coupling of chlorobenzotrifluor-
ide with piperazine was also investigated (Table 9).
Using the same conditions that worked well for the re-
Scheme 6. Synthesis of arylpiperazines by palladium-cata-
lyzed aromatic amination.
As a representative example for arylpiperazine syn- action of the aryl bromide, only moderate yields of the
thesis, formation of p-trifluoromethylarylpiperazine by arylpiperazine were observed with P(t-Bu)3 as ligand
the selective coupling of p-bromobenzotrifluoride with (entry1). On changing to D in a mixture of toluene
piperazine was investigated (Scheme 6). Using 1 mol and THF and NaO-t-Bu as base, a 93% yield was ob-
% of Pd(OAc)2 in combination with a varietyof ligands
and bases, good conversion of the starting materials to (Entry2). It is interesting to note that with half of the
product was observed. The results differed mainlyin quantityof catalyst approximatelythe same yield of
tained even when using onlya slight excess of piperazine
the amount of bisarylated piperazine by-product that product was found (Entry3). Use of D in amine solvents
was formed. The most promising ligand/base combina- like tributylamine also provided excellent results (Entry
tions were then investigated on a larger scale and the re- 4). The nature of the base is veryimportant both for the
sults are summarized in Table 8.
substrate scope of a reaction and also for economic rea-
As a general trend, it is necessaryto use strong bases
sons. Using ligand C in a mixture of toluene/methanol
like NaO-t-Bu to achieve high catalytic activity. The gave good results with solid sodium hydroxide (Entry 5).
use of P(t-Bu)3 and several of the dialkylphosphinobi- Summarizing this work, the dialkylphosphinobiaryl li-
phenyl ligands, especially C, gave good results. If weaker gands C and D proved to be especiallyvaluable for the
bases such as cesium carbonate were used, similar re- coupling of 4-chlorobenzotrifluoride with piperazine us-
sults were seen, however, it was necessaryto employ
ing a strong base and this avoids the need to use a large
higher amounts of catalyst. With P(t-Bu)3 and a large ex- excess of piperazine to minimize the formation of dou-
cess of piperazine and NaO-t-Bu as base, no detectable ble arylated product.
product of double arylation was observed and a 97%
Different stages of the coupling of chlorobenzotri-
yield of the desired product was realized (Entry 1). De- fluoride with piperazine in a jacketed glass vessel are
creasing the excess of piperazine to 2.5 equivs. caused a shown in Figure 5.
slight diminution of the yield, largely due to double ary-
It is important to note, from an industrial point of view,
lation (Entry2). With stericallyencumbered ligand C, a that addition of the catalyst in four equal portions yields
Table 8. Synthesis of 4-trifluoromethylarylpiperazine from bromobenzotrifluoride.
EntryPiperazine (equivs.) Solvent
T [ 8C] Catalyst
Pd [mol %] L:Pd Base
GC Yield [%][a]
1
2
3
4
5
6
Xylene/DMF 120 Pd(OAc)2/P(t-Bu)3 0.1
Toluene/DMF 110 Pd(OAc)2/P(t-Bu)3 0.1
4
4
3
3
4
NaO-t-Bu 97
NaO-t-Bu 92
NaO-t-Bu 97
NaO-t-Bu 93
NaO-t-Bu 85[b]
2.5
2.5
2.5
2.5
Toluene
tert-Butanol
Toluene
110
90
110
Pd(OAc)2/C
Pd(OAc)2/C
Pd(OAc)2/P(t-Bu)3 0.05
0.1
0.1
[a]
[b]
GC yields determined with dodecane as internal standard.
Reaction did not go to completion.
Table 9. Synthesis of 4-trifluoromethylarylpiperazine from chlorobenzotrifluoride.
EntryPiperazine (equivs.) Solvent T [ 8C] Catalyst Pd [mol %] L:Pd Base
GC Yield [%][a]
NaO-t-Bu 54
NaO-t-Bu 93
NaO-t-Bu 92
NaO-t-Bu 97
1
2
3
4
5
6
Xylene/DMF
Toluene/THF
Toluene/THF
Tributylamine
120
90
90
Pd(OAc)2/P(t-Bu)3 0.1
4
2
2
2
2
1.5
1.5
1.5
1.5
Pd2dba3/D
Pd2dba3/D
Pd2dba3/D
Pd2dba3/C
0.1
0.05
0.1
110
Toluene/MeOH 78
0.1
NaOH
93
[a]
GC yields determined with dodecane as internal standard.
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
31

Stephen L. Buchwald et al.
REVIEW
Figure 5. Arylpiperazine formation conducted in jacketed reaction vessels, from left to right: (a) at room temperature before
reaction; (b) after heating at reflux for one hour; (c) after completion of the reaction (2 h).
verysimilar results to those obtained when all the cata-
lyst is added at one time. Each portion of added catalyst
converts about one quarter of the starting material to
product. Therefore, if during a manufacturing run, in-
process analytical data indicate that significant catalyst
deactivation has occurred, more catalyst can be added
to complete the reaction. This is a situation that is not al-
ways the case when using homogeneous catalysts.
To obtain a clean, palladium-free product, after the re-
action, a pH-controlled work-up was utilized that also
allowed for the selective removal of anyexcess pipera-
zine. After complete consumption of the starting mate-
rial, the reaction medium was allowed to cool to room
temperature and the inorganic salts were removed byfil-
tration. The organic phase was then mixed with water
and the pH was adjusted to 3. At this pH, product and
the excess piperazine were selectivelyextracted into
the aqueous layer. Phase separation and adjustment of
the pH to 10 caused the selective precipitation of the fi-
nal product in high yield and in a purity of about 99%.[14]
Scheme 7. Alternative ligands for the formation of arylpiper-
azines.
ICP analysis indicated that there was no detectable con- 5 Synthesis of Industrially Important
À
tamination of the product byresidual palladium. An ad-
ditional attribute of this method of work-up is that it
Diarylamines by C N Coupling
yielded very clean product even if the conversion was in- Diarylamines are not only used as pharmaceutical inter-
complete (Table 8, Entry5), since residual aryl halide mediates, but also in applications such as electronic ma-
starting material stayed in the organic phase during terials and as antioxidants. Their usual mode of manu-
the acid extraction and excess piperazine does not pre- facture includes high temperature Ullmann coupling[15]
cipitate from the water layer at a pH of 10.
For the amination of veryactivated aryl chlorides,
even simpler catalyst combinations can be applied. In
or, for example, byvarious nucleophilic aromatic substi-
tution methods.[16]
The palladium-catalyzed aromatic amination reaction
the example shown in Scheme 7, use of a mixture of pal- adds to these known methods and significantlyextends
ladium acetylacetonate and triphenylphosphine with the scope of readilyavailable substrates.
K2CO3 as base led to formation of the desired product
We again began our work in this area byscreening the
in high yield. Since triphenylphosphine is one of the important reaction parameters for the transformation
cheapest of all phosphines, it maybe, in certain cases,
the ligand of choice.
shown in Scheme 8. We chose a simple reaction set-up
and used one mmol of starting material per reaction to
investigate the effect of eight bases and eight ligands
32
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
with toluene as the solvent. A set of these 64 reactions
can be run in a laboratoryin one day, giving rapid infor-
mation on changes brought upon bythe different reac-
tion parameters. Once the best combination of base
In this case purification of the product can be achieved
bydistillation and no palladium was detected in the final
product.[18]
Another application in the synthesis of diarylamines
and catalyst has been identified, other parameters such was a Lanxess in-house product, 4-nitrodiphenyla-
as solvent, additives or the ratio of the different reaction mine[19] (Scheme 9), a precursor of the antioxidant com-
components can be investigated, either in a similar ponent 4-aminodiphenylamine. These diarylamines and
screening manner or using a serial optimization ap- their derivatives are of widespread use in car tyres today.
proach.
The current process at Lanxess uses a copper-cata-
As an example, for N-2,6-dichlorodiphenylamine, the lyzed Ullmann coupling. The palladium-catalyzed
retrosynthetic disconnection can be made on either side method offered the chance to avoid the formation of un-
of the nitrogen (Scheme 8).
wanted triarylamine by-product, which considerably re-
This yields two catalyst/base combinations as shown, duces the overall yield of the current process (Table 10).
one for approach 1, and one for approach 2[17]. For ap- Obviously, the economic constraints for the manufac-
proach 1, starting with 2,6-dichloroaniline, F as ligand ture of such commodityproducts are quite severe.
and NaO-t-Bu as base resulted in almost complete for-
In this case quantitative formation of 4-nitrodiphenyl-
mation of the desired product. For approach 2, with K3 amine could be observed even using 0.005 mol % of the
PO4 as base, B was the most efficient ligand. Onlya
verymoderate yield, however, was realized. Thus, we
chose to pursue approach 1 for further studies.
combination of (þ/À)-BINAP and PdCl2. Moreover,
the subsequent reduction of the nitro group could be ac-
complished bythe addition of palladium on charcoal to
We next carried out the reaction on a 0.5-mol scale the crude reaction solution under hydrogen pressure.
with a concentration of 10% w/w of starting material This makes for a more efficient use of the palladium as
to solvent, 0.25 mol % of Pd2dba3 and a ratio of Pd:L well as facilitating its recycling. Following the procedure
of 1:2. We found it advantageous to first charge the re- described in Entry2, analysis of the aqueous phase after
action flask with Pd2dba3 and the ligand dissolved in tol- work-up of the palladium coupling process showed no
uene at room temperature, this was stirred for 10 mi- detectable amount of palladium.[18] The organic phase,
nutes. Next bromobenzene was added, followed sequen- however, showed a concentration that summed to over
tiallyby2,6-dichloroaniline and the base. The addition 95% of the palladium originallyadded as catalyst. After
of base caused the solution to warm noticeably. The re- hydrogenation with palladium on charcoal, the metal
action mixture was then quicklyheated to reflux. After
one hour, analysis indicated that full conversion of the
starting material to product had occurred. Water was
added to the reaction mixture at 508C, the phases
were separated and the organic layer was washed once
with brine. The crude solution was then distilled under
vacuum to give the final product in 94% yield.
cleanlyprecipitated on the support and a total of 85%
When the reaction was repeated on the same scale but
with 0.1 mol % of catalyst, a level typical for an econom-
icallyviable application of such catalytic processes, full
conversion of the starting material was not observed.
Additional portions of catalyst (0.05 mol % Pd2dba3,
0.2 mol % F) were added to the reaction mixture, to a to-
tal of 0.15 mol % Pd2dba3, 0.6 mol % F and full conver-
sion of the starting material to product was observed.
Scheme 9. Lanxess process for the synthesis of antioxidant
component 4-aminodiphenylamine.
Scheme 8. Synthesis of 2,6-dichlorodiphenylamine.
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
33

Stephen L. Buchwald et al.
Yield [%][a]
REVIEW
Table 10. Formation of 4-nitrodiphenylamine.
EntryConditions
1
2
3
4
5 mol % CuO, K2CO3, CsHCO3, refluxing aniline (1848C)
0.1 mol % Pd(acac)2, 0.4 mol % PPh3, K2CO3, refluxing aniline
0.03 mol % PdCl2, 0.13 mol % PPh3, 5 mol % AcOH, K2CO3, refluxing aniline
0.005 mol % PdCl2, 0.005 mol % (þ/– )-BINAP, K2CO3, refluxing aniline
60–70
99
99
99
[a]
Yields were determined bycalibrated HPLC after work-up.
of the palladium used in the coupling reaction could be
recovered.[20]
With these results in hand we saw the possibilityof
preparing anewclass ofantioxidantadditiveswith a sub-
To summarize this work, an alternative to the copper- stituent on the diarylamine that could be vulcanized into
catalyzed method for the large-scale production of a low a rubber for tyres, therefore rendering longer protection
value-added substance was identified. Even with the from oxidation. 4-Isopropyleneaniline, which is readily
strict economic restrictions for the preparation of a com- available byammonolysis of bisphenol A, [21] a common
modityintermediate like 4-aminodiphenylamine, the
polymer building block, could be cleanly coupled with 4-
palladium-catalyzed technology was competitive with chloronitrobenzene (Scheme 10); no competitive by-
the established route that uses a copper catalyst.
product of Heck arylation could be detected.[22] In this
instance, use of ligand B provided a 96% yield of the de-
sired product.
The versatilityof the palladium-catalyzed aromatic
amination can further be demonstrated bythe synthesis
of N-2-methyl-4-methoxyphenylaniline, another impor-
tant intermediate in the fine chemical industry
(Scheme 11).
Classically, this molecule can be prepared by a transfer
hydrogenation between phenol and 2-methyl-4-me-
thoxyaniline using catalytic amounts of cyclohexa-
none.[23] While thearomatic aminationapproach is prob-
ablynot economicallycompetitive, it offers the chance
to choose from a wider pool of starting materials.
As 2-methyl-4-methoxyaniline is available inexpen-
sivelyon scale, approach I is a realistic alternative to
Scheme 10. Synthesis of alkenyl-diarylamines.
Scheme 11. Synthesis of 2-methyl-4-methoxy-diarylamine by palladium catalysis.
34
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
Table 11. Approaches to 2-methyl-4-methoxy-diarylamine.
[a]
EntryApproach
Conditions
Yield [%]
1
2
3
4
5
I
II
III
IV
II
2 mol % B, 0.5 mol % Pd2dba3, K3PO4, refluxing toluene
95
95
96
93
12
1 mol % BINAP, 0.5 mol % Pd2dba3, Cs2CO3, refluxing toluene
2 mol % B, 0.5 mol % Pd2dba3, K3PO4, refluxing aniline (1848C)
4 mol % PPh3, 0.5 mol % Pd2dba3, Cs2CO3, refluxing aniline (1848C)
5 mol % PPh3, 5 mol % Pd/C, Cs2CO3, refluxing toluene
[a]
Isolated yield after distillation.
the current synthesis. Our usual screening method di- 6 Synthesis of the Dialkylphosphinobiaryl
rectlyyielded a protocol that could be scaled to the mul-
tigram level. Product purification could be achieved by
Ligands
distillation with no detectable amount of palladium re- Dialkylphosphinobiaryls are, in our view, the most gen-
À
maining in the final product (Entry1). However the
amount of catalyst necessary would have to be signifi- processes. Six members of this ligand familywhose use
eral ligands for palladium-catalyzed C N bond-forming
cantlydecreased to render this an economicallycompet-
itive process.
usuallyprovided excellent results are shown in
Scheme 1. Studies at MIT have indicated that substitu-
tion on the bottom (non-phosphorus-containing) ring
Approach II, which is slightlyless attractive due to the
use of bromo- rather than chlorobenzene, provides a is important.[24] It is believed that this lessens the tenden-
nearlyquantitative yield of the product (Entry2).
When looking at other alternatives, approach III, in formation of monoligated palladium. While it is not yet
refluxing toluene as solvent, gave onlya moderate yield possible to identifyone ligand for all amination process-
cyfor the formation of palladacycles, while favoring the
of the product. However, on changing the solvent to ani- es, this small set of ligands available from a single proc-
line and raising the reaction temperature to 1848C (re- ess covers a range of possibilities for a custom manufac-
flux), the desired product is obtained in nearlyquantita-
tive yield (Entry 3).
turer to flexiblyscale-up todaꢁys demanding targets
from pharmaceutical companies.
The last possible combination of starting materials is
represented byapproach IV. On first inspection, an eco-
nomicallyattractive catalyst could be identified to per-
form this reaction, consisting of a palladium precatalyst an equallyattractive, but slightlydifferent route.
with triphenylphosphine as the ligand. Unfortunately, it
Despite the apparent complexityof these ligands, A
through C are available in a one-pot procedure as de-
picted in Scheme 12,[24a] while D can be synthesized by
[25]
À
With two metallation steps, one C C bond formation,
À
is necessaryto employCs CO3 as base, which is quite ex- one C P bond formation and a potentiallycomplex
2
pensive. Last, the use of aryl-4-bromo-3-methylanisole work-up, their scale-up is challenging.
as starting material in approach IV, a compound that
for the time being is available in laboratoryquantities
only, compromises this route (Entry 4).
Unfortunatelyour attempts to use Pd/C with triphe-
nylphosphine as ligand gave a poor yield of 2-methyl-
4-methoxy-diarylamine (Entry 5).
To summarize the results for the synthesis of2-methyl-
4-methoxy-diarylamine, the palladium-catalyzed aro-
matic amination probablywill not supersede the classi-
cal synthesis, via transfer hydrogenation, from an eco-
À
nomic point of view. All four approaches (I IV) showed
almost complete formation of the desired product. This
is an indication of the flexibilityof this technologyand
bodes well for future synthetic endeavors. Of special in-
terest to the custom manufacturer is that the choice of
the synthetic disconnection can be determined by the
availabilityof the starting material, enhancing the versa-
tilityof this method even further.
Scheme 12. Synthesis of dialkylphosphinobiaryl ligands.
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
35

Stephen L. Buchwald et al.
REVIEW
Rhodiaꢁs work resulted in two procedures that, inter-
estingly, only differ from the MIT protocolin theway the
work-up was performed.
In the original procedure, the reaction was carried out
in anhydrous THF using two equivalents of magnesium
powder and an aryl bromide (2-bromotoluene or 1-bro-
Scheme 13. Two possible routes for the synthesis of chlorodi-
cyclohexylphosphine.
7 Synthesis of
mo-2,4,6-triisopropylbenzene).
Grignard reagent is then coupled with benzyne, generat-
ed in situ from 1,2-chlorobromobenzene and magnesi-
The
intermediate
Chlorodicyclohexylphosphine
One of the key issues for the synthesis of the dialkyl- um. The carbon-phosphorus bond is formed bythe
phosphinobiaryl class of ligands is the availability of CuCl-catalyzed reaction between the newly formed
chlorodicyclohexylphosphine. With a very limited num- Grignard reagent and chlorodicyclohexylphosphine.
ber of manufacturers of chlorodicyclohexylphosphine,
usuallyproducing onlyon a kilogram scale, we consid-
For safetyreasons, magnesium powder had to be
avoided and we chose to replace it bymagnesium turn-
ered its in-house synthesis by the controlled addition ings, which are easier to handle and manipulate. We used
of a Grignard reagent to phosphorus trichloride in sol- technical grade THF as the solvent.[26] The chlorodicy-
vents such as THF (Scheme 13).
clohexylphosphine was provided by Rhodia PPD and
Companies specialized in the handling of pyrophoric was at least 92% pure.[27]
and highlytoxic phosphorus intermediates might prefer
In the course of our studies, we found that the addition
a route involving the chlorination of dicyclohexylphos- of magnesium turnings in two batches provides a higher
phine. The Grignard approach, however, is a wayto han-
dle this chemistrywithout specialized expertise.
yield of product. In this way, formation of Wurtz-type
by-products was avoided. Generation of the first
À
À
For the synthesis, the reaction temperature needs to Grignard species, the following C C coupling and C P
be kept at 08C in order to avoid the formation of tricy- coupling are fast reactions and occurred to a large extent
clohexylphosphine. It was even more important that a during the addition of the reactants; one hour after their
strict stoichiometry of 2:1 of cyclohexylmagnesium introduction the reactions are complete. A new proce-
chloride to PCl3 be ensured. Strict exclusion of moisture dure for product isolation was developed. This involved
and oxygen was necessary to prevent formation of im- the addition of aqueous sodium sulfite (37%) to the reac-
À
purities that poisoned the subsequent C P bond-form- tion mixture to hydrolyze any excess Grignard reagents
ing step. During the reaction, several equivalents of and to dissolve the magnesium salts that had formed.
magnesium chloride are formed. Since these magnesium Through the use of this reducing solution, phosphine ox-
salts show considerable solubilityin THF, it was necessa-
ryto change the reaction solvent from THF to toluene.
idation was also avoided. At this point the layers were
separated and, in case of C, THF was replaced by n-bu-
Running the reaction in toluene provides a reaction sol- tanol via solvent exchange. Crystallization afforded C as
ution that, after filtration, can be used directly. Howev- a white powder in 65% yield.[28] The procedure was
er, more reliable results in terms of product qualitywere
slightlydifferent for the preparation of A. In that case,
realized if the crude solution of chlorodicyclohexylphos- at the end of the reaction procedure, after cooling the re-
phine was concentrated and distilled under a high vac- action mixture to room temperature, ethyl acetate was
uum. Interestingly, if the mixture contained more than added. The resulting organic layer was washed several
3% w/w of tricyclohexylphosphine, the distillation was times with hot water to remove the THF. Ligand A
unsuccessful. Dicylohexylphosphinous oxide, a white then crystallizes directly from the ethyl acetate solution
solid, deposited throughout the distillation apparatus upon cooling and is obtained, after filtration, as a white
when the heating reached a temperature of 1708C. powder in a 50% yield.[29]
This is probablydue to a phosphine-catalyzed dispro-
The development of a safe industrial procedure re-
portionation reaction. Even distillation techniques quires the determination of all the thermal characteris-
used for the purification of sensitive materials such as tics of the process that occur or could occur during the
short-path distillation or thin-film distillation did not reaction protocol.[30] We first investigated the synthesis
solve this problem.
of A. We found that the formation of 2-methylphenyl-
magnesium bromide is exothermic (DH¼285 kJ/mol)
which translates into an adiabatic temperature rise of
2458C. This heat is potentiallysufficient to evaporate
93% of the initial THF charge. The reaction is quite
8 Synthesis of Ligands A and C
A reliable protocol for the lab-scale synthesis of A has fast and the conversion, measured at the end of the ad-
[24a]
been published bythe MIT group
and was subse- dition, to the Grignard reagent is 96%. Thus, the slow in-
quentlyused to prepare C, as well, on a 20-gram scale. troduction of the aryl bromide to the magnesium turn-
For industrial purposes, the scale of the synthesis needed ings was required to control anyexotherm and to ensure
to be increased to the multi-kilogram level.
a safe process. The second step, the formation of 2-(2’-
36
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
methyl sulfate in alkaline solution worked well on
both small and large scales. However, even if onlysmall
amounts of overmethylated products were formed,
aqueous work-up proved to be verydifficult, since the
trimethylammonium by-products formed strongly in-
hibited phase separation. To solve this problem, we add-
ed an excess of o-chloroaniline to the reaction mixture
(up to 3% of monomethylated aniline was seen in the
crude material) and the problems observed during
work-up disappeared. However, the monomethylated
product could not be removed bysimple distillation,
since the boiling point differs onlybyone degree from
that of the N,N-dimethyl-2-chloroaniline. The solution,
in this case, was the addition of acetic anhydride to the
distillation vessel, therebytransforming the monome-
thylated aniline to the acetanilide, which allowed N,N-
dimethyl-2-chloroaniline to be separated by distillation.
This procedure was successfullyapplied to produce
~100 kg of N,N-dimethyl-2-chloroaniline as starting ma-
terial for the synthesis of B in an overall yield of 44–
Figure 6. Vessel (250 liters) for the pilot-scale production of
A and C.
methylphenyl)-phenylmagnesium bromide, is also exo- 48%.
thermic (DH¼872 kJ/mol, DTadiabatic ¼978C) and this
The conversion of N,N-dimethyl-2-chloroaniline into
the corresponding Grignard reagent is challenging. To
quantityof heat is able to potentiallyevaporate the en-
tire initial charge of THF. Nevertheless, this exotherm safelyconduct this transformation on a moderate scale
can again be easilycontrolled bythe slow addition of requires its simultaneous addition, along with dibromo-
the 1,2-bromochlorobenzene. Finally, the copper-cata- ethane (10 mol %), to a suspension of activated magne-
À
lyzed C P coupling is also quite exothermic (DH¼ sium turnings in a minimal quantityof refluxing THF.
236 kJ/mol) and rapid (about 98% of the chlorodicyclo- With an adiabatic temperature rise of 1708C this first re-
hexylphosphine has been consumed at the end of its ad- action demanded special care. To avoid build-up of heat
dition). All these data confirm the necessityto slowlyin-
potential, the addition of the aryl halide had to be done
troduce the reagents over the course of at least one hour. in small portions with analytical checks made to deter-
Finallythe work-up protocol has also been studied. The
energyof the neutralization of the reaction solution us-
mine if the reaction had started. Compared with the
preparation of other dialkylbiphenylphosphines start-
ing dilute NaHSO3 is 214 kJ/mol, which is dissipated by ing from aryl bromides rather than aryl chlorides, this
dilution. No corrosion effect of copper or bromide onto first step took considerablylonger, usuallyeight to
the steel surface of the reaction vessel was observed.[31] twelve hours, to reach completion.
In summary, a safe reaction protocol was devised for The addition of 1,2-bromochlorobenzene behaved as
the preparation of A and C in which potential exotherms for the preparation of the other two ligands and was easi-
were controlled bythe rate of addition of the reagents. It
was also necessaryto change the nature of the work-up
lycontrolled byadjusting the rate of its addition. Ex-
tending the reaction time of this step, however, leads
to adapt this protocol from an academic realm to an in- to the formation of several Wurtz-type coupling prod-
dustrial one. The procedure for ligands A and C could be ucts, which could be avoided bykeeping the reaction
carried out in a 250-L apparatus (Figure 6) on a scale of time as short as possible. If an excess of bromochloro-
10 kg of product isolated per batch.
benzene was used in this step, the incorporation of a sec-
ond and third phenyl ring into the system was observed
and work-up became more difficult.
À
9 Synthesis of Ligand B
The catalyst for the C P coupling in this case was a
mixture of CuI and LiBr instead of CuCl.[32] In this
For the synthesis of the last of the three ligands prepared waythe reaction solution was more homogeneous and
À
on a larger scale, o-chloroaniline (produced on a multi- the work-up was easier; the rate of the C P coupling
ton scale at Lanxess) was employed as starting material. was similar in both cases.
Methylation of o-chloroaniline could be achieved by
Analysis of the work-up method described in the orig-
several methods. The reaction of the aniline with form- inal procedure indicated that the addition of water to the
aldehyde under reductive aminationconditions with for- reaction mixture would involve tripling of its volume. To
mic acid provided a good yield of product on a small achieve better volumetric productivity, the addition of a
scale. On a larger scale, however, we observed the for- small amount of methanol was used not onlyto quench
mation of polymeric by-products. Methylation with di- anyexcess Grignard reagent, but also to destroymagne-
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
37

Stephen L. Buchwald et al.
REVIEW
Table 12. Results of adsorption studies, shown is the contam-
ination level of palladium in the organic layer
Adsorbent
Trade name
Pd (ppm)
Charcoal
Charcoal
Charcoal
Charcoal
Charcoal
Charcoal
Charcoal
Charcoal
Resin
NORIT GAC
NORIT SX
NORIT CNI
NORIT C GRAND
CECA Acticarbon CXV
CECA Acticarbon BGX
IRC 50 (CO2H)
IRC 48 (aminodiacetate)
Lewatit SP 112
62–67
75
34
<5
<5
<5
114
116
105
104
32
Resin
Resin
Amberlyst A21
Amberlyst 36 DRY
bond-forming processes. Several methods can be inves-
tigated to remove the palladium from either the organic
or aqueous layer. To our knowledge, for this kind of
chemistrythe oxidation state of the palladium is un-
known. However, we determined that in our process
the majorityof the metal is in the Pd(0) oxidation state
and is therefore soluble in the organic layer. Thus, we
concentrated our efforts on recoveryfrom this phase.
A screening of different charcoals and resins as adsorb-
ents was carried out bythe chemists at Rhodia. The best
results were obtained with three types of charcoal:
CECA Acticarbon BGX, CECA Acticarbon CXV and
NORIT C Grand removed all but 5 ppm of palladium
from the organic layer when it was stirred with the ad-
sorbent at 808C. The abilityto remove palladium de-
creased slightlyat room temperature; 10–14 ppm of pal-
Figure 7. Vessel (150 liters) for the pilot-scale production of
B.
sium complexes still in solution. To achieve complete ladium remained.
precipitation of the magnesium salts, a solvent change
ESCA studies of the charcoal from the reaction mix-
from THF to toluene was again necessary, giving rise ture indicated the presence of carbon, oxygen, palladi-
to an easilyfiltered suspension and with a filter cake
which contained no product.
um, sodium and traces of bromides. The palladium was
found to onlybe in the zero oxidative state.
The toluene solution containing the product was con-
centrated, mixed with methanol, and, upon cooling,
white crystals formed that were isolated by filtration 11 Conclusions and Prospects for Future
and showed a purityof >98%.[33] Analysis of this mate-
rial showed that it contained up to 400 ppm of magnesi-
Work
um-containing impurities, while the copper content was In this paper we have described how the implementation
about 20 ppm. Further purification of the material was of a unique three-waycollaboration between two com-
possiblebyrecrystallization from acetone, where copper
panies and one academic institution has led to the
and magnesium impurities dropped below 10 ppm, and scale-up and commercial utilization (several hundreds
the overall yield of the process ranged between 55 and of kg/product so far) of a new technology. While the
63%. The reaction was performed in the 150-L vessel de- original chemistryemanated from MIT, the technical
picted in Figure 7.
skills available from those in Rhodia and Lanxess were
essential for overcoming issues of safetyand scale.
Moreover, the latter were responsible for bringing to
fruition the production of over 100 kg of ligands, osten-
sibly, via the original academic route. Of importance to
the success of this project was that all participants real-
ized that theyhad much to gain bylistening to the points
10 Removal/Recyclingof Palladium after
the Reaction
According to economic estimates, recoveryof the palla-
dium might become an important factor to reduce the of view of the others. In this waynot onlywas the project
À
overall material cost of palladium-catalyzed C N more successful, but valuable information and lessons
38
asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 23 – 39

Industrial-Scale Palladium-Catalyzed Coupling of Aryl Halides and Amines
REVIEW
probe with TE-cooled CCD detector technologyusing
laser 785 nm.
[10] Quantitative palladium analysis was performed by FX
analysis.
[11] The highest temperature reached in this exotherm was
1928C.
[12] The calorimetric energydisplayed during the reaction
was measured in a RC1 Mettler calorimeter to be
62.4 kJ/mol of starting arylhydrazone. This heat corre-
sponds to an adiabatic temperature rise of þ13.58C,
therefore due to the long reaction time of about 20 h
the reaction can be regarded as safe.
were learned byall involved that will be instrumental in
future endeavors.
The palladium-catalyzed formation of aromatic car-
bon-heteroatom bonds is increasinglybecoming a daily
tool for those in the discoverylaboratories of pharma-
ceutical companies. More recently, a number of projects
have made it to process groups and scale-ups of the
transformations have occurred. Clearly, much work is
still necessaryto overcome a varietyof practical limita-
tions. As the chemistryis more widelypracticed and the
methodologymatures, we are optimistic that it will be
practiced on increasinglylarge scale.
[13] CASIS Search 01/2004.
[14] Determined byquantitative 1H-NMR.
[15] a) F. Ullmann, Ber. dtsch. Chem. Ges. 1901, 34, 2174; b) F.
Ullmann, Ber. dtsch chem. Ges. 1903, 36, 2382–2384; c) F.
Ullmann, P. Sponagel, Ber. dtsch. chem. Ges. 1905, 36,
2211–2212; d) I. Goldberg, Ber. dtsch. chem. Ges. 1906,
39, 1691–1692; e) S. V. Ley, A. W. Thomas, Angew.
Chem. 2003, 115, 5558–5607; Angew. Chem. Int. Ed.
2003, 42, 5400–5449; f) K. Kunz, U. Scholz, D. Ganzer,
Synlett 2003, 2428–2439.
[16] a) M. K. Stern, J. K Bashkin, WO 9300324; b) R. D. Tri-
plett, R. K. Rains, WO 2003010126; c) S. M. S. Chauhan,
R. Singh, Synth. Commun. 2003, 33, 2899–2906.
[17] The ligands investigated were: (þ/–)BINAP, P(t-Bu)3,
PPh3, B, dppf, P(o-Tol)3, E and F. Bases examined includ-
ed K2CO3, NaO-t-Bu, KO-t-Bu, CsF, K3PO4, Na2CO3,
NaOH and Cs2CO3.The palladium precatalyst was
0.5 mol % of Pd2dba3.
References and Notes
[1] For a recent review see, for example: V. Farina, Adv.
Synth. Catal. 2004, 346, 1553–1582.
[2] a) T. Kanbara, A. Honma, K. Hasegawa, Chem. Lett.
1996, 1135; b) F. E. Goodson, J. F. Hartwig, Macromole-
cules 1998, 31, 1700; c) J. Louie, J. F. Hartwig, J. Am.
Chem. Soc. 1997, 119, 11695; d) J. Louie, J. F. Hartwig,
Macromolecules 1998, 31, 6737; e) R. A. Singer, J. P. Sa-
dighi, S. L. Buchwald, J. Am. Chem. Soc. 1998, 120, 213.
[3] a) L. Jiang, S. L. Buchwald, in: Metal-Catalyzed Cross-
Coupling Reactions, (Eds.: A. de Meijere, F. Diederich),
ISBN 3-527-30518-1, Wiley-VCH, Weinheim, 1998;
b) B. Schlummer, U. Scholz, Adv. Synth. Catal. 2004,
346, 1599–1626.
[4] a) B. Robinson, The Fischer Indole Synthesis, John Wiley
and Sons, Chichester, 1982; b) R. Milcent, Chimie Or-
[18] Determined byICP-MS.
[19] German Patent DE 19942394.
´ ´
ganique Heterocyclique, EDP Science, 2002; c) J. Joulie,
[20] 75% of the original charcoal precipitated of the Pd/C and
could be isolated byfiltration. The remaining 10% of
palladium could be recovered byaddition of CECA 4S
activated charcoal and 1 h stirring at 508C.
K. Mills, Heterocyclic Chemistry, Blackwell Publishing,
Oxford, 2000; d) I. Gut, J. Wirz, Angew. Chem. Int. Ed.
Engl. 1994, 33, 1153; e) R. Sundberg, Indoles, Academic
Press, London, 1996.
[21] 2,2-Bis-(4-hydroxybiphenyl)-propane.
[5] C. Mauger, G. Mignani, Org. Proc. Devel. 2004, 8, 1065.
[6] The use of commercial bulk phosphates including Na4P2
O7, Na2HPO4, Na3PO4 ·10.5 H2O, Na3PO4, Na2HPO4 ·
2 H2O, (CaO)10(P2O5)H2O, and K4P2O7, proved to be in-
effective.
[7] Trifluoromethylphenyl hydrazines are important com-
pounds for the preparation of azaheterocycle intermedi-
ates in pharmaceutical and agrochemical domains, for
examples, see a) WO 2003076409 (Syngenta Participa-
tions AG, Switzerland); b) WO 2003068223 (Bayer Cor-
poration, USA); c) FR 2815346 (Les Laboratoires Servi-
er, France); d) WO 2001089457 (SmithKline Beecham
Corporation, USA; Glaxo Group Limited); e) WO
2000069849 (Ortho-McNeil Pharmaceutical, Inc.,
USA); f) Y. Nalavde, V. Joshi, Ind. J. Chem. Sect. B
2000, 39, 634–637; g) EP 1044970 (Adir et Compagnie,
France).
[22] Lanxess, German Patent DE 10235834.
[23] Mitsui, Japanese Patent JP61218560, JP 05003867.
[24] a) S. Kaye, J. M. Fox, F. A. Hicks, S. L. Buchwald, Adv.
Synth. Catal. 2001, 344, 789; b) X. Huang, K. W. Ander-
son, D. Zim, L. Jiang, A. Klapars, S. L. Buchwald, J.
Am. Chem. Soc. 2003, 125, 6653.
[25] Due to the fact that 2,6-dimethoxybromobenzene is not
available as a starting material, a separate route for the
synthesis of ligand was developed: T. E. Barder, S. D.
Walker, J. R. Martinelli, S. L. Buchwald, J. Am. Chem.
Soc. 2005, 127, 4685–4696.
[26] THF was provided bySDS and contained 0.1% of water.
[27] Determined by 31P NMR.
[28] Purityby 31P NMR was>98%.
[29] Purityby 31P NMR was 98%, a second crop yielded an
additional 10% yield of A.
[30] The calorimetric and thermodynamic studies were per-
formed in a 2-liter RC1 Mettler calorimeter.
[31] Similar results were obtained for the synthesis of ligand C.
[32] F. Rampf, H. C.Militzer, (Lanxess), European Patent EP
1354886, 2003.
[8] The use of the following bases resulted in lower conver-
sions: NaO-t-Bu (47%), KO-t-Bu (51%), K2CO3 (78%),
Cs2CO3 (70%). No reaction occurred using A as the ligand.
[9] Description of Raman spectrometer: in situ Raman Rxn1
Analyzer from Kaiser Optical Systems Inc., immersion
[33] Determined bycalibrated HPLC.
Adv. Synth. Catal. 2006, 348, 23 – 39
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
39