Statistical experimental design-driven discovery of room-temperature conditions for palladium-catalyzed cyanation of aryl bromides

Article abstract of DOI:10.1016/j.tetlet.2005.01.116

A combination of Pd₂(dba)₃·CHCl₃ (0.5 mol %) and commercially available, air-stable phosphonium salt [(t-Bu) ₃PH]BF₄ (1.4 mol %) in a presence of Zn powder and Zn(CN)₂ as the cyanide source comprises an extremely efficient catalyst system for the cyanation of a diverse array of aryl bromides, at room temperature. This result emerged from an experimental strategy that combines the advantages of parallel, automated experimentation with the design of experiments (DOE) for the effective definition of an optimal set of reaction conditions.

Full text of DOI:10.1016/j.tetlet.2005.01.116

Tetrahedron
Letters
Tetrahedron Letters 46 (2005) 1815–1818
Statistical experimental design-driven discovery
of room-temperature conditions for palladium-catalyzed
cyanation of aryl bromides
Federica Stazi,a Giovanni Palmisano,a Marco Turconib and Marco Santagostinob,*
a
`
Dipartimento di Scienze Chimiche Fisiche e Matematiche, Universita dell’Insubria, Via Valleggio 11, 22100 Como, Italy
bChemistry Research Centre, Boehringer Ingelheim Italia, Via Lorenzini 8, 20139 Milano, Italy
Received 25 November 2004;revised 21 January 2005;accepted 24 January 2005
Abstract—A combination of Pd2(dba)3ÆCHCl3 (0.5 mol%) and commercially available, air-stable phosphonium salt [(t-Bu)3PH]BF4
(1.4 mol%) in a presence of Zn powder and Zn(CN)2 as the cyanide source comprises an extremely efficient catalyst system for the
cyanation of a diverse array of aryl bromides, at room temperature. This result emerged from an experimental strategy that com-
bines the advantages of parallel, automated experimentation with the design of experiments (DOE) for the effective definition of an
optimal set of reaction conditions.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Mainly propelled by the quest for milder reaction condi-
tions as compared to the classic Rosenmund-von Braun
reaction,1 transition metal-catalyzed cyanation of aryl
halides is experiencing a remarkable renaissance.2 Nev-
ertheless, despite the discovery of new and more active
catalytic systems, efficient halogen-cyanide exchange
still necessitates temperatures substantially above the
ambient temperature and/or high catalyst loadings. Pre-
sumably, this is required to compensate for catalyst
deactivation by an excess of cyanide ions in the reaction
mixture.2a Obviously, the development of room-temper-
ature reactions poses significant advantages with regard
to selectivity/stability issues and becomes an especially
attractive objective from an industrial standpoint.
couples cyanide ions, from Zn(CN)2, with 1 at 50 ꢁC
to produce 1-cyanonaphthalene 2.5 Although at this
stage catalyst efficiency was still moderate, this crystal-
line solid and commercially available phosphonium salt,
recently emerged as a user-friendly alternative to air-sen-
sitive P(t-Bu)3,6 furnished the cleanest reaction profile
and was selected for further study. Ancillary experiment-
ation showed that conversion could be improved by
working at higher dilutions (0.3 g/mL) in wet N-methyl-
pyrrolidone (NMP, 1 vol% water content7) and includ-
ing Zn powder (25 mol%) in the reaction mixture.8
Under these conditions, however, considerable homo-
coupling of 1 was observed in addition to the usual
hydridodebromination side reaction producing naph-
thalene. The large number of potential parameters influ-
encing, often contradictorily, conversion and yield
convinced us of the appropriateness of a multivariate
approach in order to develop a reliable protocol within
a reasonable number of independent experimental trials.
In this light, the technique of design of experiments
(DOE) stands as a precious tool for the effective investi-
gation of an appropriate ꢀexperimental spaceꢁ as delim-
ited by the low and high values of the experimental
variables under study.9 By virtue of statistical experi-
mental design approach, the influencing variables and,
appealingly, their interactions are easily revealed and
their effect quantified. Thus, generation of inferential
models from aptly designed sets of experiments
furnishes the experimentalist an optimum set of reaction
In this paper we report a general procedure for room-
temperature palladium-catalyzed cyanation of a wide
range of aromatic bromides.3 Initial attempts focussed
around selecting pertinent ligands for moderate-temper-
ature cyanation of 1-bromonaphthalene 1. An evalua-
tion of several ligands4 showed that a combination of
Pd2(dba)3ÆCHCl3 (1 mol%) and [(t-Bu)3PH]BF4
(2 mol%) in DMF (1 g/mL) generated a catalyst that
Keywords: Cyanation;Aryl bromides;Room temperature;Electron-
rich phosphines;Design of experiments.
*
Corresponding author. Tel.: +39 025355383;fax: +39 025355205;
e-mail: msantagostino@mil.boehringer-ingelheim.com
0040-4039/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2005.01.116

1816
F. Stazi et al. / Tetrahedron Letters 46 (2005) 1815–1818
conditions. A preliminary fractional factorial design10
unveiled robust reaction conditions and set the basis
for the subsequent optimization design. Accordingly, a
Box–Behnken design was implemented (Fig. 1) to allow
a more precise quantification of the effect of significant
factors on the cyanation yield with a minimum number
of additional experimental runs.11
Three factors, namely, amount of Zn(CN)2 (variable A:
from 1.0 to 1.2 equiv), [(t-Bu)3PH]BF4/Pd(0) ratio (li-
gand to palladium ratio, L:Pd;variable B: from 0.8 to
2.0) and water content (variable C: from 0.1 to
2 vol%) were varied over three levels, while the amount
of palladium pre-catalyst Pd2 (dba)3ÆCHCl3 (0.5 mol%),
of Zn (5 mol%)12 and the temperature (25 ꢁC) were kept
at their best level as emerged from the previous factorial
screening. Experiments were performed in parallel on an
automated workstation (Anachem SK 233) and ana-
lyzed after 24 h.13 The plot, illustrated in Figure 2, was
generated according to the quadratic model obtained
from statistical analysis of the data and describes the
dependence of in situ yield of 1-cyanonaphthalene 2
on the amounts of added water and the L:Pd ratio.
On the contrary, the amount of Zn(CN)2 was found
not to have a significant influence on the reaction yield
over the examined range.
Figure 2. Response surface plot, as produced by MODDE, describing
the variation of the predicted in situ yield of 1-cyanonaphthalene 2 as a
function of the two experimental variables which were found signifi-
cant in the statistical analysis, namely B (L:Pd ratio) and C (amount of
added water relative to the amount of solvent used).
independently observed the strong connection between
cross-coupling reaction rates and L:Pd ratio with steri-
cally congested ligands.14 While the resting state for pal-
ladium in the Pd2(dba)3/t-Bu3P system has been
identified as Pd[P(t-Bu)3]2 (L:Pd = 2) across a wide
range of L:Pd ratios, oxidative addition has been shown
to occur to a highly unsaturated monophosphine inter-
mediate (L:Pd = 1) through a preequilibrium involving
ligand dissociation.15 It was also noted that higher, up
to 1.5, L:Pd ratio appears to stabilize the catalyst and
decrease the precipitation of palladium metal.14b
A region of predicted maximum yield is observed at low
levels of added water (0.1–0.3 vol%), where best perform-
ances are realized by adopting a L:Pd ratio equal to 1.2–
1.5. Higher L:Pd ratios (1.8–2.0) consistently hamper the
catalyst system performances. This effect is particularly
marked at high levels of added water, which lead to an
almost complete inactivation of the catalyst, probably
as a consequence of the increased concentration of
dissolved cyanide ions. This well fits with the recent
findings from Fuꢁs and from Hartwigꢁs group who
As a final confirmation, the cyanation of 1 was con-
ducted in quadruplicate, on a 5 g scale, under the sug-
gested optimal variables setting. An average in situ
yield of 2 of 98.0% (isolated yield of 88%, Table 1, entry
1) was obtained when 1 was treated with Zn(CN)2
(1.1 equiv) in the presence of Pd2(dba)3ÆCHCl3
(0.5 mol%), [(t-Bu)3PH]BF4 (1.4 mol%) and Zn powder
(5 mol%) in wet NMP (water content 0.1 vol%) at room
temperature for 24 h.
99.0
97.8
5.3
98.7
96.0
4.6
93.4
84.4
99.2
91.8
Having established a proper set of conditions for the
cyanation of 1-bromonaphthalene, the scope of the pres-
ent findings was tested on a range of differently substi-
tuted aryl bromides. As illustrated in Table 1,
Pd2(dba)3ÆCHCl3/[(t-Bu)3PH]BF4 effectively promotes
room-temperature cyanation of a sterically and electroni-
cally diverse array of aryl bromides. Particularly note-
worthy are the reactions with sterically congested,
electron-rich 2-bromoanisole 9 and with extremely elec-
tron-demanding 4-bromoaniline 12. This underlines the
fundamental role of bulky, electron-rich phosphine li-
gands in facilitating oxidative addition to otherwise
problematic substrates.16 However, these conditions
were not completely general. Neither relatively acidic
3-bromophenol 13 nor potentially ligating17 3- and 2-
bromopyridine 14 and 15 were effectively cross-coupled
under these room-temperature conditions. Pleasingly
+1 (2)
C: H2O (%)
-1 (0.1)
+1 (2)
A:
4.8
76.7
90.4
-1 (0.8)
B: L:Pd (ratio)
Figure 1. Box–Behnken optimization design. Spheres indicate the
position of single experimental runs, that is, the setting of the
experimental variables, within the experimental space. In turn, this is
defined as the cubic region delimited by the coded value (actual values
are reported in brackets) for the reaction variables under study. The
grey dot positioned in the centre of the cubic region indicates four
replicate experiments (factors setting: Zn(CN)2 = 1.1 equiv,
L:Pd = 1.4, water = 1 vol%), which allow experimental error to be
estimated. Measured response values for in situ yield of 1-cyanonaph-
thalene 2 for each experimental run are reported within the sphere
area;the average value is reported for the replicate centre points.

F. Stazi et al. / Tetrahedron Letters 46 (2005) 1815–1818
1817
Table 1. Room temperature Pd2(dba)3ÆCHCl3/[(t-Bu3)PH]BF4-cata-
unveiled a versatile catalyst system, which allows cyana-
tion of a wide range of aryl bromides under mild condi-
tions, typically at room temperature.19 This procedure
fares well when compared with the recently reported
room-temperature palladium-catalyzed cyanation of
aryl bromides3 in terms of higher turnover number
(100 vs 20), stability of the ligand (air-stable [(t-Bu)3PH]BF4
vs air-sensitive P(t-Bu)3) and use of quasi-stoichiometric
amounts of Zn(CN)2 (1.1 vs 1.8 equiv) and provides an
excellent complement to existing methods which oper-
ates under harsher conditions. Moreover, the present
approach reveals the ability of statistical experimental
design (DOE) in efficiently disclosing and understanding
relationships between experimental variables in complex
reaction environments and represents the first example
of empirical multivariate modelling of the non-linear
relationship between L:Pd ratio and reaction yields, in
palladium catalyzed reactions. On-going efforts are di-
rected at expanding the utility of this catalyst system
with special focus on aryl chlorides and triflates.
lyzed cyanation of aryl bromidesa
Entry Aryl bromide
Conv. (%)b Yieldc (%)
Br
1
1
98
88
Br
2
3
4
5
6
7
>99
>99
>99
>99
>99
89
79
65
75
98
O
Br
3
F
OHC
Br
4
5
6
MeOOC
Br
OHC
MeO
MeO
Br
Br
Acknowledgments
7
8
8
9
>99
>99
74
85
Appreciation is expressed to Mr. Angelo Girola of the
Dipartimento di Scienze Chimiche, Fisiche e Matema-
Br
tiche, Universita dellꢁInsubria for the 31P NMR
experiments.
`
OMe
Br
9d
10
>99
86
Supplementary data
Br
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.tet-
let.2005.01.116. Results for the ligands screening experi-
ment;design matrix, measured responses and ANOVA
table for the screening design and for the Box–Behnken
optimization design. Experimental procedures using the
automated workstation Anachem SK 233. This material
is available free of charge via the internet at http://
pubs.acs.org.
10e
11
11
12
>99
>99
49
88
N
H
Br
H2N
HO
Br
12f
13f
13
70
55
Br
3-Br 14
2-Br 15
47
0
40
NR
N
References and notes
a Conditions: Pd2(dba)3ÆCHCl3 0.5 mol%, [(t-Bu)3PH]BF4 1.4 mol%,
5% Zn powder, 1.1 equiv Zn(CN)2, wet NMP, 24 h, rt. Reaction
scale: 23 mmol.
1. Ellis, G. P.;Romney-Alexander, T. M. Chem. Rev. 1987,
87, 779–794.
b Measured by HPLC after 24 h reaction time.
2. Pd-catalyzed: (a) Schareina, T.;Zapf, A.;Beller, M. Chem.
Commun. 2004, 12, 1388–1389;(b) Schareina, T.;Zapf, A.;
Beller, M. J. Organomet. Chem. 2004, 689, 4576–4583;(c)
Yang, C.;Williams, J. M. Org. Lett. 2004, 6(17), 2837–
2840;(d) Marcantonio, M. K.;Frey, F. L.;Liu, Y.;Chen,
c Isolated yield by column chromatography or crystallization.
d Pd2(dba)3ÆCHCl3 1.5 mol%, [(t-Bu)3PH]BF4 4.2 mol%.
e Indole was also formed as the major side product (42%, HPLC).
f Rxn run at 80 ꢁC for 3 h.
Y.;Strine, J.;Phenix, B.;Wallace, J. D.;Chen, C.
Org.
Lett. 2004, 6(21), 3723–3725;(e) Sundermeier, M.;Zapf,
A.;Beller, M. Angew. Chem., Int. Ed. 2003, 42, 1661–1664;
(f) Sundermeier, M.;Zapf, A.;Mutyala, S.;Baumann, W.;
Sans, J.;Weiss, S.;Beller, M. Chem. Eur. J. 2003, 9(8),
1828–1836;(g) Sundermeier, M.;Mutyala, S.;Zapf, A.;
Spannenberg, A.;Beller, M. J. Organomet. Chem. 2003,
684, 50–55;For a review, see (h) Sundermeier, M.;Zapf,
A.;Beller, M. Eur. J. Inorg. Chem. 2003, 3513–3526, and
references cited therein;Cu-catalyzed: (i) Zanon, J.;
Klapars, A.;Buchwald, S. J. Am. Chem. Soc. 2003, 125,
2890–2891.
enough, cyanation of these substrates, with the excep-
tion of 2-bromopyridine,18 does proceed rapidly at
80 ꢁC (Table 1, entries 12 and 13), thus substantially
expanding the scope of the present method to less reac-
tive aryl halides.
In summary, a strategy based on parallel ligands screen-
ing succeeded by investigation of the experimental vari-
ables by means of statistical experimental design has

1818
F. Stazi et al. / Tetrahedron Letters 46 (2005) 1815–1818
3. While this work was well in progress, room-temperature
and MODDE, version 7.0.0, by Umetrics (http://
www.umetrics.com) were used for generation and analysis
of experimental designs.
cyanation of aryl bromides/iodides using P(t-Bu)3 as
ligand was reported. Under the described conditions, the
catalyst system performances were highly dependent on
catalyst loading and no reaction was observed on 4-
bromoacetophenone using 1.25 mol% of Pd catalyst.
Ramnauth, J.;Bhardwaj, N.;Renton, P.;Rakhit, S.;
Maddaford, S. P. Synlett 2003, 2237–2239.
11. With three factors, Box–Behnken designs represent an
attractive alternative to the more commonly utilized
central composite designs since they still allow to model
quadratic response surfaces with a reduced number of
experimental runs (12 vs 14 non-center points).
4. Eighteen commercially available ligands were screened in
parallel. Besides [(t-Bu)3PH]BF4, the following bulky
phosphines-catalyzed cyanation of 1-bromonaphthalene
at 50 ꢁC in comparable yield albeit at the expense of more
consistent dehalogenation to generate naphthalene, which
is the major side product under the conditions tested: P(o-
Tol)3, 2-(di-t-butylphosphino)biphenyl;2-(di- t-butylphos-
phino)-1,10-binaphthyl;2-(diphenylphosphino)-2 0-(N,N-
dimethylamino)biphenyl and P(t-Bu)3. See the supplemen-
tary data.
5. Such a cyanation reaction proceeds without any added
base, which is regarded as necessary in order to deproto-
nate [(t-Bu)3PH]BF4 and promote formation of the active
catalyst, see Ref. 6. While cyanide ion can be considered a
competent Brønsted base, we observed in the 31P NMR
spectrum partial formation of Pd[P(t-Bu)3]2 (d = 86 ppm)
upon mixing Pd2(dba)3 and [(t-Bu)3PH]BF4 (d = 52 ppm)
in DMF in the absence of any base.
12. In the absence of Zn powder under otherwise identical
conditions, we observe cyanation of 1-bromonaphthalene
to occur in comparable yields and purity. However,
inconsistent results were often obtained. We believe this
may indicate an easier inactivation of the catalyst system
by adventitious oxygen in the reaction mixture.
13. Reactions were run in a random order under nitrogen
on 1.0 g scale, in 3 mL of NMP and quantified by HPLC
gauging the in situ yield of 2 and the amount of
residual starting material 1. See the supplementary data
for details. The final equation (scaled experimental
variables) for the in situ yield of
2
is given
by yield = 84.2 À 25.5 · B À 12.2 · C À 17.1 · B2 + 8.0 ·
C2 À 13.6 · B · C. The model parameters were estimated
by means of the partial least squares regression method
(PLS) including the result of one run from the 25À1
factorial design which falls in the actual design space (i.e.,
run 3, factors setting: Zn(CN)2 = 1.2 equiv, L:Pd = 1.2,
water = 0.1% with Pd amount = 1%, Zn amount = 5% and
temperature = 25 ꢁC. The measured in situ yield of 1-
cyanonaphthalene 2 for this run was 83.5%). The product
statistics indicate a significant model with fairly good
predictability (R2 = 0.840, Ra2dj ¼ 0:767 and Q2 = 0.584).
See the supplementary data.
6. Netherton, M. R.;Fu, G. C. Org. Lett. 2001, 26, 4295–
4298. [(t-Bu)3PH]BF4 is commercially available from
STREM and Aldrich.
7. Hydroxide ions are known to stabilize low ligated palla-
dium complexes and facilitate palladium-catalyzed cyana-
tion of aromatic halides. It is likely that addition of small
quantities of water results in the formation of catalytically
more active palladium(0) hydroxide species. Takagi, K.;
Okamoto, T.;Sakakibara, Y.;Ohno, A.;Oka, S.;
Hayama, N. Bull. Chem. Soc. Jpn. 1976, 49, 3177–3180.
8. The co-catalytic activity of Zn powder is based on its
ability to reduce palladium(II) species formed to some
extent in side reactions. (a) Okano, T.;Iwahara, M.;Kiji,
J. Synlett 1998, 243–244;(b) Jin, F.;Confalone, P. N.
Tetrahedron Lett. 2000, 41, 3271–3273.
9. (a) Anderson, M.;Whitcomb, P. J. DOE simplified:
Practical Tools for effective Experimentation;Productivity
Inc.: Portland, OR, 2000;(b) Montgomery, D. C. Design
and Analysis of Experiments;Wiley: New York, 2001;(c)
Carlson, R. Design and Optimization in Organic Synthesis;
Elsevier: New York, 2000.
14. (a) Littke, A. F.;Fu, G. J. Am. Chem. Soc. 2001, 123,
6989–7000;(b) Littke, A. F.;Dai, C.;Fu, G. J. Am. Chem.
Soc. 2000, 122, 4020–4028;(c) Lee, S.;Hartwig, J. J. Org.
Chem. 2001, 66, 3402–3415;(d) Hartwig, J.;Kawatsura,
M.;Hauck, S. I.;Shaughnessy, K. H.;Alcazar-Roman, L.
M. J. Org. Chem. 1999, 64, 5575–5580.
15. Stambuli, J. P.;Bu hl, M.;Hartwig, F. J. Am. Chem. Soc.
¨
2002, 124, 9346–9347.
16. For a review on electron-rich phosphines, see: Valentine,
D. H., Jr.;Hillhouse, J. H. Synthesis 2003, 16, 2437–2460.
17. Wagaw, S.;Buchwald, S. L. J. Org. Chem. 1996, 61, 7240–
7241.
18. Unreactive pyridyl-bridged palladium dimers, formed
after oxidative addition, account for the persistently
observed reduced reactivity of 2-halopyridines in a num-
ber of cross-coupling reactions. See for example Bozell, J.
J.;Vogt, C. E.;Gozum, J. J. Org. Chem. 1991, 56, 2584–
2587.
19. General procedure. A two-necked reaction flask was
charged with Pd2(dba)3ÆCHCl3 (120 mg, 0.116 mmol,
0.5 mol%), [(t-Bu)3PH]BF4 (94 mg, 0.324 mmol, 1.4
mol%), Zn powder (76 mg, 1.16 mmol, 5 mol%) and
Zn(CN)2 (1.50 g, 12.77 mmol, 1.1 equiv) and purged with
nitrogen. A solution of the aryl bromide (23.30 mmol) in
wet NMP (0.1% water content, 15 mL) was added and the
mixture was stirred at rt. After 24 h, the reaction was
diluted with EtOAc and then washed with a 2 N NH4OH
solution and brine. The organic phase was concentrated
and the residue purified on silica gel or by crystallization.
10. This two-level factorial design (25À1, 16 randomized
experiments plus three center points) was performed on
an automated workstation (Anachem SK 233). Five
factors were screened at two levels, namely amount of
Pd (range: 1–5 mol%), L:Pd ratio (range: 0.8–2), amount
of Zn powder (range: 5–50 mol%), amount of water
(range: 0.1–1 vol%) and temperature (range: 25–75 ꢁC).
See the supplementary data. Based on results from this
screening, two-variable interactions could be revealed and
uniformly high yields were observed by setting the reaction
variables Pd, Zn and temperature at their low level. As for
the amount of water and L:Pd ratio, the situation was less
clear cut and warranted further investigation. Both Design
Expert, version 6.0.4, by Stat-Ease (www.statease.com)