Using intelligent/random library screening to design focused libraries for the optimization of homogeneous catalysts: Ullmann ether formation

Article abstract of DOI:10.1021/ja000094c

A 96-member 'pyridine' library consisting of both rationally chosen and 'random' members was used to screen Ullmann ether forming reactions. The reaction of 2-bromo-4,6-dimethylaniline and other substrates with a variety of alkoxides was investigated under different conditions with the aid of an automated liquid handler. From the results of the 96-member library screening, a structure activity profile was determined which led to the design of smaller 'focused' ligand libraries. The focused libraries produced a higher frequency of hits compared to the original 96-member library. Some of the more effective ligands discovered in this work were found to be generally useful for alkoxylation of a variety of substrates, and also functioned in intramolecular ether forming reactions. This work demonstrates for homogeneous catalysis the analogy to the pharmacological model of drug discovery. By using a large library to screen for a lead compound followed by screening the diversity space closest to the lead, a larger fraction of increased performance ligands was discovered.

Full text of DOI:10.1021/ja000094c

J. Am. Chem. Soc. 2000, 122, 5043-5051
5043
Using Intelligent/Random Library Screening To Design Focused
Libraries for the Optimization of Homogeneous Catalysts: Ullmann
Ether Formation
Paul J. Fagan, Elisabeth Hauptman,* Rafael Shapiro, and Albert Casalnuovo
Contribution from The Dupont Company, Central Research and DeVelopment Department,
P.O. Box 80328, Experimental Station, Wilmington, Delaware 19880-0328
ReceiVed January 7, 2000
Abstract: A 96-member “pyridine” library consisting of both rationally chosen and “random” members was
used to screen Ullmann ether forming reactions. The reaction of 2-bromo-4,6-dimethylaniline and other substrates
with a variety of alkoxides was investigated under different conditions with the aid of an automated liquid
handler. From the results of the 96-member library screening, a structure activity profile was determined which
led to the design of smaller “focused” ligand libraries. The focused libraries produced a higher frequency of
hits compared to the original 96-member library. Some of the more effective ligands discovered in this work
were found to be generally useful for alkoxylation of a variety of substrates, and also functioned in intramolecular
ether forming reactions. This work demonstrates for homogeneous catalysis the analogy to the pharmacological
model of drug discovery. By using a large library to screen for a lead compound followed by screening the
diversity space closest to the lead, a larger fraction of increased performance ligands was discovered.
Introduction
Borrowing from the early work directed at pharmaceutical
discovery, ligand libraries have been derived from building
blocks of amino acids with a few notable exceptions.5,6j,n In
some catalyst studies, the libraries were biased with cases that
were already known to work well to demonstrate principle.
Combinatorial chemistry is now routinely used for lead
discovery in pharmaceutical and agrochemical companies.1
Combinatorial chemistry techniques have more recently been
applied for the discovery of materials2 and catalysts.3-7 For
catalysis, the strategy usually relies on optimization of reaction
conditions using different temperatures, pressures, additives,
ligands, and/or metal ions. Most of the ligand synthesis effort
to date has focused on the use of polymer-bound ligands.6
(4) For heterogeneous catalysis see: (a) Cong, P.; Doolen, R. D.; Fan,
Q.; Gianquinta, D. M.; Guan, S.; McFarland, E. W.; Poojary, D. M.; Self,
K.; Turner, H. W.; Weinberg, W. H. Angew. Chem., Int. Ed. 1999, 38,
484. (b) Senkan, S. M.; Ozturk, S. Angew. Chem., Int. Ed. 1999, 38, 791.
(c) Holzwarth, A.; Schmidt, H.-W.; Maier, W. F. Angew. Chem., Int. Ed.
1998, 37, 2644. (d) Reddington, E.; Sapienza, A.; Gurau, B.; Viswanathan,
R.; Saranggapani, S.; Smotkin, E. S.; Mallouk, T. E. Science 1998, 280,
1735. (e) Senkan, S. M. Nature 1998, 394, 350.
(5) For homogeneous ligands synthesis and catalysis see: (a) Moreau,
C.; Frost, C. G.; Murrer, B. Tetrahedron Lett. 1999, 40, 5617. (b) Lapointe,
A. M. J. Comb. Chem. 1999, 1, 101. (c) Brandli, C.; Ward, T. R. HelV.
Chim. Acta 1998, 81, 1616. (d) Gennari, C.; Ceccarelli, S.; Piarulli, U.;
Montalbetti, C. A. G. N.; Jackson, R. F. W. J. Org. Chem. 1998, 63, 5312.
(e) Gilbertson, S. R.; Chang, C.-W. T. J. Org. Chem. 1998, 63, 8424. (f)
Moye-Sherman, D.; Welch, M. B.; Reibenspies, J.; Burgess, K. Chem.
Commun. 1998, 2377. (g) Porte, A. M.; Reibenspies, J.; Burgess, K. J. Am.
Chem. Soc. 1998, 120, 9180. (h) Buck, R. T.; Coe, D. M.; Drysdale, M. J.;
Moody, C. J.; Pearson, N. D. Tetrahedron Lett. 1998, 39, 7181. (i) Burgess,
K.; Lim, H.-J.; Porte, A. M.; Sulikowski, G. A. Angew. Chem., Int. Ed.
1996, 35, 220.
(6) For combinatorial polymer-supported synthesis and catalysis see: (a)
Jacobsen, E. N.; Francis, M. B. Angew. Chem., Int. Ed. 1999, 38, 937. (b)
Boussie, T. R.; Coutard, C.; Turner, H.; Murphy, V.; Powers, T. S. Angew.
Chem., Int. Ed. 1998, 37, 3272. (c) Annis, D. A.; Helluin, O.; Jacobsen, E.
N. Angew. Chem., Int. Ed. 1998, 37, 1907. (d) Sigman, M. S.; Jacobsen, E.
N. J. Am. Chem. Soc. 1998, 120, 4901. (e) Menger, F. M.; Ding, J.;
Barragan, V. J. Org. Chem. 1998, 63, 7578. (f) Bergbreiter, D. E.; Case,
B. L.; Liu, Y.-S.; Caraway, J. W. Macromolecules 1998, 31, 6053. (g) Cole,
B. M.; Shmizu, K. D.; Krueger, C. A.; Harrity, J. P.; Snapper, M. L.;
Hoveyda, A. H. Angew. Chem., Int. Ed. 1996, 35, 1668. (h) Gilbertson, S.
R.; Wang, X. Tetrahedron Lett. 1996, 37, 6475. (i) Gilbertson, S. R.; Wang,
X.; Hoge, G. S.; Klug, C. A.; Schaefer, J. Organometallics 1996, 15, 4678.
(j) Gilbertson, S. R.; Wang, X. J. Org. Chem. 1996, 61, 434. (k) Liu, G.;
Ellman, J. A. J. Org. Chem. 1995, 60, 7712. (l) Menger, F. M.; Eliseev, A.
V.; Migulin, V. A. J. Org. Chem. 1995, 60, 6666. (m) Gilbertson, S. R.;
Chen, G.; McLoughlin, M. J. Am. Chem. Soc. 1994, 116, 4481. (n) Boussie,
T. R.; Murphy, V.; Hall, K. A.; Coutard, C.; Dales, C.; Petro, M.; Carlson,
E.; Turner, H. W.; Powers, T. S. Tetrahedron 1999, 55, 11699.
(1) For recent reviews see: (a) Liu, D. R.; Schultz, P. G. Angew. Chem.,
Int. Ed. Engl. 1999, 38, 37. (b) Weber, L.; Illgen, K.; Almstetter, M. Synlett
1999, 3, 366. For special issues and additional reviews see: (c) Chem. ReV.
1997, 97, No. 2. (d) Verdine, G. L.; Broach, J. R.; Thorner, J.; Hogan, J.
C. Nature 1996, 384 (Suppl.), 11. (e) Service, R. F. Science 1996, 272,
1266. (f) Balkenhohl, F.; von dem Bussche-Hunnefeld, C.; Lansky, A.;
Zechel, C. Angew. Chem., Int. Ed. Engl. 1996, 35, 2288.
(2) (a) Jandeleit, B.; Schaefer, D. J.; Powers, T. S.; Turner, H. W.;
Weinberg, W. H. Angew. Chem., Int. Ed. Engl. 1999, 38, 2495 (b) Reynolds,
C. H. J. Comb. Chem. 1999, 1, 297. (c) Klein, J.; Lehmann, C. W.; Schmidt,
H.-W.; Maier, W. F. Angew. Chem., Int. Ed. Engl. 1998, 37, 3369. (d)
Akporiaye, D. E.; Dahl, I. M.; Karlsson, A.; Wendelbo, R. Angew. Chem.,
Int. Ed. Engl. 1998, 37, 609. (e) Wang, J.; Yoo, Y.; Gao, C.; Takeuchi, I.;
Sun, X.; Hauyee, C.; Xiang, X.-D.; Schultz, P. G. Science 1998, 279, 1712.
(f) Danielson, E.; Devenney, M.; Giaquinta, D. M.; Golden, J. H.;
Haushalter, R. C.; McFarland, E. W.; Poojary, D. M.; Reaves, C. M.;
Weinberg, W. H.; Wu, X. D. Science 1998, 279, 837. (g) Gravert, D. J.;
Datta, A.; Wentworth, P., Jr.; Janda, K. D. J. Am. Chem. Soc. 1998, 120,
9481. (h) Danielson, E.; Golden, J. H.; McFarland, E. W.; Reaves, C. M.;
Weinberg, W. H.; Wu, X. D. Nature 1997, 389, 944. (i) Baker, B. E.; Kline,
N. J.; Treado, P. J.; Natan, M. J. J. Am. Chem. Soc. 1996, 118, 8721. (j)
Briceno, G.; Chang, H.; Sun, X.; Schultz, P. G.; Diang, X.-D. Science 1995,
270, 273. (k) Brocchini, S.; James, K.; Tangpasuthadol, V.; Kohn, J. J.
Am. Chem. Soc. 1997, 119, 4553.
(3) For general catalysis reviews: (a) Crabtree, R. H. Chem. Commun.
1999, 1611. (b) Maier, W. F. Angew. Chem., Int. Ed. 1999, 38, 1216. (c)
Bein, T. Angew. Chem., Int. Ed. 1999, 38, 323. (d) Hoveyda, A. H. Curr.
Opin. Chem. Biol. 1999, 3, 313. (e) Jandeleit, B.; Turner, H. W.; Uno, T.;
van Beek, J. A. M.; Weinberg, H. W. Cattech 1998, 2, 101. (f) Dagani, R.
Chem. Eng. News 1998, June 15, 6. (g) Francis, M. B.; Jamison, T. F.;
Jacobsen, E. N. Curr. Opin. Chem. Biol. 1998, 2, 422. (h) Gennari, C.;
Nestler, H. P.; Piarulli, U.; Salom, B. Liebigs Ann./Recl. 1997, 637. (i)
Burgess, K.; Porte, A. M. AdV. Catal. Processes 1997, 2, 69.
10.1021/ja000094c CCC: $19.00 © 2000 American Chemical Society
Published on Web 05/10/2000

5044 J. Am. Chem. Soc., Vol. 122, No. 21, 2000
Fagan et al.
With the goal of discovering the best catalyst for an Ullmann
alkoxylation reaction, we used a strategy that parallels enzyme
inhibitor optimization. To our knowledge, this approach has not
yet been reported to be successful for the discoVery and
optimization of homogeneous catalysts. Our strategy was to
screen a parent set of ligands that were designed to contain both
rationally and randomly chosen members. After screening this
library and discovering successful hits, smaller focused libraries
were created based on ligand structure-activity relationships.
These daughter libraries were screened, and resulted in a higher
“hit frequency” than the parent library. This method of screening
an intelligent/random parent library followed by focused libraries
was demonstrated to be a fast and logical approach for
optimization of the ligand component of the catalytic Ullmann
reactions studied here.
The discovery of the best heterogeneous catalyst for a reaction
by combinatorial methods is fraught with difficulties and renders
the process of rational library design challenging. Library
catalyst synthesis, history, characterization, and reproducible
performance are important and sometimes difficult to control
experimental parameters.8 However, for homogeneous catalysts
that combine a ligand and a metal, the factors that control
catalyst performance at the molecular level are better defined
and easier to incorporate when creating a ligand library. We
felt that the design of a ligand library for homogeneous catalysis
need not be as random or exhaustive in parameter space as the
heterogeneous catalyst case. Incorporating molecular parameters
important to most homogeneous catalyst functions was a natural
way to cover ligand diversity space as completely as possible
with a relatively small parent library.9 For our particular study,
we considered the following as important parameters to vary
when preparing the “intelligent” portion of the parent library:
(1) electron-donating and -withdrawing abilities of the ligands
(σ and π); (2) sterics around the donating atom on the ligands;10
(3) monodentate or multidentate donation, (4) for multidentate
ligands, ligand bite angles;11 and (5) second coordination sphere
effects (molecular recognition potential).12 Taking these pa-
rameters into account, we attempted to maximize the chance
of discovering the best ligand with a minimum set. Although
the bulk of the library members were chosen with these
parameters in mind, we also took the opportunity for serendipity
by incorporating ligands that were not expected to work for
any particular reason. By a combination of intelligent and
random experimentation, we hoped to attain our goal of finding
the best “pyridine-containing” ligands for the Ullmann reactions
in this study. Having found these lead ligands, we then would
prepare and screen a focused library whose members contained
structural elements resembling the discovered lead compounds.
We selected the Ullmann ether forming reaction because it
constitutes a practical approach to form aryl ethers, a structure
common to many agrochemical and pharmaceutical lead com-
pounds. These reactions couple aryl halides and alkoxides and
are usually carried out using Cu(I) or Cu(II) salts in the presence
of ligands (or a solvent that can serve as a ligand). Recent
advances have been made for certain types of Ullman
reactions,13-17 and novel Pd-based catalytic systems have been
shown to be efficient at carrying out C-O and other C-X bond
forming reactions.18 General trends indicate that electron-
deficient aryl halides typically work best in Ullmann chemistry,
but yields and/or rates can rapidly decrease with increasing
electron density on the aromatic ring, increasing steric bulk, or
the presence of deleterious unprotected functionality. An ad-
ditional problem especially encountered in the methoxylation
reaction of haloarenes is the formation of the reduced arene
byproduct; one of our goals was to minimize this side reaction.
In general, the detailed mechanistic steps of copper-catalyzed
Ullmann reactions are poorly understood and dependent on
many variables; this is exactly the type of experimental situation
where combinatorial methods are best suited.
Copper(I)-catalyzed C-O bond-forming reactions have been
reported in the presence of pyridine-type ligands.14,16c Numerous
pyridine-containing compounds are commercially available and
(13) For recent reviews on C-X coupling (X ) N and O) see: (a)
Cristau, H.-J.; Desmurs, J.-R.; Ratton, S.; Rignol, S.; Taillefer, M. In The
Roots of Organic DeVelopment; Desmurs, J-R., Ratton, S., Eds.; Industrial
Chemistry Library, 1996; Vol. 8, p 90. (b) Tonks, L.; Williams, J. M.
Organomet. Chem. 1998, 26, Chapter 8, p 180. (c) Beller, M.; Riermeier,
T. H. In Transition Metals for Organic Synthesis; Beller, M., Bolm, C.,
Eds.; Wiley-VCH: New York, 1998; Chapter 2.11.
(14) Yamakawa, K. U.S. Patent 5,006,660 to Fuji Photo Film Co.
(15) For general Ullmann couplings see: (a) Paine, A. J. J. Am. Chem.
Soc. 1987, 109, 1496. (b) Couture, C.; Paine, A. J. Can. J. Chem. 1985,
63, 111. (c) Lindley, J. Tetrahedron 1984, 40, 1433. (d) Yamamoto, T.;
Kurata, Y. Can. J. Chem. 1983, 61, 86.
(16) For phenoxylation reactions see: (a) Marcoux, J.-F.; Doye, S.;
Buchwald, S. L. J. Am. Chem. Soc. 1997, 119, 10539. (b) Nicolaou, K. C.;
Boddy, C. N. C.; Natarajan, S.; Yue, T.-Y.; Li, H.; Brase, S.; Ramanjulu,
J. M. J. Am. Chem. Soc. 1997, 119, 3421. (c) Pellon, R. F.; Carrasco, R.;
Milian, V.; Rodes, L. Synth. Commun. 1995, 25, 1077. (d) Yeager, G. W.;
Schissel, D. N. Synthesis 1995, 28. (e) Smith, K.; Jones, D. J. Chem. Soc.,
Perkin Trans. 1 1992, 407. (f) Oi, R.; Shimakawa, C.; Takenaka, S. Chem.
Lett. 1988, 899. (g) Afzali, A.; Firouzabadi, H.; Khalafi-nejad, A. Synth.
Commun. 1983, 13, 335.
(17) For methoxylation reactions see: (a) Ragan, J. A.; Malowski, T.
W.; Castaldi, M. J.; Hill, P. D. Synthesis 1998, 1599. (b) Capdevielle, P.;
Maumy, M. Tetrahedron Lett. 1993, 34, 1007. (c) Aalten, H. L.; van Koten,
G.; Grove, D. M.; Kuilman, T.; Piekstra, O. G.; Hulshof, L. A.; Sheldon,
R. A. Tetrahedron 1989, 45, 5565. (d) Keegstra, M. A.; Peters, T. H. A.;
Brandsma, L. Tetrahedron 1992, 48, 3633. (e) Keegstra, M. A.; Brandsma,
L. Recl. TraV. Chim. Pays-Bas 1991, 110, 299. (f) Bacon, R. G. R.;
Rennison, S. C. J. Chem. Soc. (C) 1969, 312. For intramolecular Cu-
catalyzed alkoxylations see refs 17c and 17f and: (g) Castro, C. E.; Havlin,
R.; Honwad, V. K.; Malte, A.; Moje, S. J. Am. Chem. Soc. 1969, 91, 6464.
(18) For palladium-catalyzed phenoxylation and alkoxylation reaction
see: (a) Aranyos, A.; Old, D. W.; Kiyomori, A.; Wolfe, J. P.; Sadighi, J.
P.; Buchwald, S. L. J. Am. Chem. Soc. 1999, 121, 4369. (b) Mann, G.;
Incarvito, C.; Rheingold, A. L.; Hartwig, J. F. J. Am. Chem. Soc. 1999,
121, 3224. (c) Hartwig, J. F. Acc. Chem. Res. 1998, 31, 852. (d) Hartwig,
J. F. Angew. Chem., Int. Ed. 1998, 37, 2047. (e) Mann, G.; Hartwig, J. F.
Tetrahedron Lett. 1997, 38, 8005. (f) Palucki, M.; Wolfe, J. P.; Buchwald,
S. L. J. Am. Chem. Soc. 1997, 119, 3395. (g) Palucki, M.; Wolfe, J. P.;
Buchwald, S. L. J. Am. Chem. Soc. 1996, 118, 10333.
(7) For references emphasizing screening methodologies: (a) Reetz, M.
T.; Becker, M. H.; Klein, H.-W.; Stockigt, D. Angew. Chem., Int. Ed. 1999,
38, 1758. (b) Guo, J.; Wu, J.; Siuzdak, G.; Finn, M. G. Angew. Chem., Int.
Ed. 1999, 38, 1755. (c) Ding, K.; Ishii, A.; Mikami, K. Angew. Chem., Int.
Ed. 1999, 38, 497. (d) Shaughnessy, K. H.; Kim, P.; Hartwig, J. F. J. Am.
Chem. Soc. 1999, 121, 2123. (e) Cooper, A. C.; McAlexander, L. H.; Lee,
D.-H.; Torres, M. T.; Crabtree, R. H. J. Am. Chem. Soc. 1998, 120, 9971.
(f) Reetz, M. T.; Becker, M. H.; Kuhling, K. M.; Holzwarth, A. Angew.
Chem., Int. Ed. 1998, 37, 2647. (g) Taylor, S. J.; Morken, J. P. Science
1998, 280, 267.
(8) Schlogl, R. Angew. Chem., Int. Ed. . 1998, 37, 2333.
(9) Parshall G.; Ittel, S. D. Homogeneous Catalysis. The Applications
and Chemistry of Catalysis by Soluble Transition Metal Complexes, 2nd
ed.; Wiley: New York, 1992; 342 pp (English).
(10) Tolman, C. A. Chem. ReV. 1977, 77, 313.
(11) (a) Van Haaren, R. J.; Oevering, H.; Coussens, B. B.; Van
Stricjdonck, G. P. F.; Reek, J. N. H.; Kamer, P. C. J.; Van Leeuwen, P. W.
N. M. Eur. J. Inorg. Chem. 1999, 8, 1237. (b) Dierkes, P.; Van Leeuwen,
P. W. N. M. J. Chem. Soc., Dalton Trans. 1999, 10, 1519. (c) Van der
Veen, L. A.; Boele, M. D. K.; Bregman, F. R.; Kamer, P. C. J.; Van
Leeuwen, P. W. N. M. J. Am. Chem. Soc. 1998, 120, 11616. (d) Casey, C.
P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B. R.; Petrovich, L. M.;
Matter, B. A.; Powell, D. R. J. Am. Chem. Soc. 1997, 119, 11817. (e) Casey,
C. P.; Whiteker, G.; Melville, M. G.; Petrovich, L. M.; Gavney, J. A., Jr.;
Powell, D. R. J. Am. Chem. Soc. 1992, 114, 5535.
(12) (a) Reetz, M. T. Top. Catal. 1998, 4, 187. (b) Landis, C. R.;
Cleveland, T.; Firman, T. K. J. Am. Chem. Soc. 1998, 120, 2641. (c) Landis,
C. R.; Firman, T. K.; Root, D. M.; Cleveland, T. R. J. Am. Chem. Soc.
1998, 120, 1842. (d) Reetz, M. T.; Waldvogel, S. R. Angew. Chem., Int.
Ed. 1997, 36, 865. (e) Trost, B. M. Acc. Chem. Res. 1996, 29, 355. (f)
Reetz, M. T.; Niemeyer, C. M.; Hermes, M.; Goddard, R. Angew. Chem.,
Int. Ed. 1992, 31, 1017.

Optimization of Homogeneous Catalysts
J. Am. Chem. Soc., Vol. 122, No. 21, 2000 5045
Table 1. Parent Ligand Library of Pyridine and Pyridine-like Ligands
no.
ligand
no.
ligand
no.
ligand
1
2
3
4
5
6
7
8
9
10
11
12
13
14
pyridine
2-picoline
2-aminopyridine
2-hydroxypyridine
3-cyanopyridine
4-cyanopyridine
2,4-lutidine
2,6-lutidine
2-ethylpyridine
2-amino-6-methylpyridine
2-(aminomethyl)pyridine
2-(hydroxymethyl)pyridine
2-hydroxy-6-methylpyridine
2,4,6-collidine
33
34
35
36
37
38
39
40
41
42
43
44
45
46
nicotine
65
66
67
68
69
70
71
72
73
74
75
76
77
78
2,6-lutidine-R-2,3-diol
2,6-pyridinedimethanol
2-quinolinol
1-(2-pyridyl) piperazine
ethyl 2-pyridyl acetate
2-benzylpyridine
2-quinolinethiol
2-anilinopyridine
2,3-pyridine dicarboxylic acid
2,6-pyridine dicarboxylic acid
3,5-pyridine dicarboxylic acid
quinoline-4-carboxylic acid
8-hydroxy-5-nitroquinoline
isoquinoline-3-carboxylic acid hydrate
9-aminoacridine
4-chloroquinaldine
2-(2-diethylaminoethyl)pyridine
4-(2-diethylaminoethyl)pyridine
7,8-benzoquinoline
phenanthridine
acridine
1,10-phenanthroline
4-benzoylpyridine
proflavine
4-(2-pyridylazo) resorcinol
1,2-dihydro-1-(2-(2-pyridyl)-ethyl)-
3,6-pyridazinedione
di-2-pyridyl ketone
15
16
17
18
19
picolinamide
47
48
49
50
51
4,4′-dimethyl-2,2′-dipyridyl
6,6′-bi-2-picoline
8-acetoxyquinoline
2-methyl-8-nitroquinoline
2,6-di-tert-butyl pyridine
79
80
81
82
83
5-nitroquinaldic acid
2-dimethylaminopyridine
4-dimethylaminopyridine
2-(2-hydroxyethyl)pyridine
quinoline
5-nitro-1,10-phenanthroline
9-hydroxy-4-methoxyacridine
4-chloro-7-(trifluoromethyl) quinoline
8-hydroxyquinoline-5-sulfonic acid
monohydrate
20
21
22
23
24
25
isoquinoline
52
53
54
55
56
57
1,3-di(4-pyridyl)propane
2-phenylquinoline
neocuproine
84
85
86
87
88
89
2,3-di-3-pyridyl-2,3-butanediol
2-phenyl-4-quinoline carboxylic acid
acridine orange
cinchonidine
quinine
1,4,5-triazanaphthalene
4-tert-butylpyridine
3-acetoxypyridine
3-pyridinepropanol
quinaldine
2,2′-pyridil
2,2′:6′,2′′-terpyridine
(S,S)-2,6-bis(4-isopropyl-2-
oxazolin-2-yl) pyridine
3-hydroxypyridine
2-mercaptopyridine
bathophenanthroline
26
27
8-hydroxyquinoline
ethylpicolinate
58
59
90
91
spiramycin
2,4-dihydroxy quinoline
monosodium salt
8-ethoxyquinoline-5-sulfonic
acid sodium salt
28
ethylisonicotinate
60
isonicotinamide
92
29
30
31
32
2-phenylpyridine
4-phenylpyridine
2,2′-dipyridyl
61
62
63
64
picolinic acid
93
94
95
96
bicinchonic acid sodium salt (bca)
1-(4-pyridyl) pyridinium chloride
2-pyridylacetic acid hydrochloride
8-mercaptoquinoline hydrochloride
2-(2-methylaminoethyl) pyridine
3-hydroxy picolinamide
3-hydroxypicolinic acid
2-(2-thienyl)pyridine
literature.19 A slight modification of the literature procedure was used
to synthesize 2-bromo-4,6-dimethylaniline:20 it was further purified by
vacuum distillation to remove dark impurities. The compound 3-(2-
bromophenyl)-1-propanol was synthesized according to the literature.21
Solutions of the ligands were prepared with the aid of a Bohdan
automated workstation with liquid handling capability and which was
modified to operate under a dinitrogen atmosphere. The 96 ligands were
prepared as either benzotrifluoride (ligands 1-57) or methanol (ligands
58-96) solutions (0.0200 M) and are listed in Table 1. These solutions
were prepared in glass vials capped with Teflon-lined silicone septa to
maintain inert atmosphere and prevent evaporation. Care was taken to
replace the pierced septa after robotic runs to minimize the amount of
evaporation; the vials were stored in a nitrogen atmosphere to ensure
the integrity of the ligand solutions between runs.
constitute a set of ligands that can be sterically and electronically
varied. We thus set out to screen pyridine and pyridine-like
compounds to find the best ligand for copper to produce the
highest yield with the least amount of reduction byproduct. The
particular reactions chosen for study were alkoxylations of
2-bromo-4,6-dimethylaniline because these were related to a
problem of practical interest to Dupont. This substrate was also
readily synthesized, and presented a somewhat challenging case
being an electron-rich, unprotected aniline that might react
sluggishly under typical Ullmann conditions (eq 1).
Preparation of catalyst solutions was performed on the Bohdan
automated work station. In a typical automated procedure, each sample
was prepared by adding 150((3) µL of a CuCl solution (0.0200 M in
CH3CN) and 150((3) µL of a ligand solution (0.0200 M in benzotri-
fluoride or MeOH) to a septa sealed 2 mL glass vial in a rack. The
solvents were then removed simultaneously from all vials in the rack
by purging with nitrogen gas streams. After briefly drying the rack of
vials in a vacuum chamber (ca. 5 min) the vials were mounted back
on the Bohdan automated liquid handler for addition of the remaining
reagents. First, 150((3) µL of a solution of 2-bromo-4,6-dimethylaniline
and biphenyl (0.400 and 0.00125 M in diglyme, respectively) were
added in sequence into each vial. Finally, 150((3) µL of potassium
phenoxide (0.600 M in diglyme) or sodium methoxide (0.600 M in
methanol) was added in sequence to each vial. Depending on the
Experimental Section
All reactions were prepared and performed under an atmosphere of
nitrogen. The compounds in the ligand libraries were obtained from
commercial sources, and were purified by distillation, sublimation, or
recrystallization; the compounds were stored under nitrogen before use.
Solvents were distilled from drying agents under nitrogen using standard
procedures: methanol from magnesium turnings; benzotrifluoride and
acetonitrile from P2O5; tetrahydrofuran (THF), dimethoxyethane (DME),
diglyme, and toluene from sodium benzophenone ketyl; dimethylfor-
mamide (DMF) from calcium hydride; and dimethylacetamide (DMAc)
from barium oxide. Copper(I) chloride was purified according to the
(19) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory
Chemicals, 3rd ed.; Butterworth-Heinemann: Boston, 1988.
(20) Bamberger, E. Justus Liebig’s Ann. Chem. 1925, 443, 209.
(21) Cooke, M. P., Jr.; Widener, R. K. J. Org. Chem. 1987, 52, 1381.

5046 J. Am. Chem. Soc., Vol. 122, No. 21, 2000
Fagan et al.
experiment, the septa on the 2 mL vials containing the reaction solutions
were either replaced with fresh septa, or the pierced septa were left on
to allow for faster evaporation if desirable as in the case of methanol.
The vials were placed in an orbital shaker/metal block heater. The block
had a sealing cover allowing the vials to be maintained under an inert
dinitrogen atmosphere during the reaction run. The block was heated
to the desired temperature (typically 100 °C), which was accurate to
(1 °C across the entire block. At the end of the run, each reaction
solution was diluted with 0.8 mL of toluene and was analyzed by gas
chromatography (HP6890 with auto sampler and a 5 m × 0.53 mm
HP-1 capillary column). Response factors were calculated and yields
were derived. Variations on this basic procedure were carried out to
test the effects of ligand-to-copper ratio, solvent variation, etc.
Large-Scale Synthesis of 2-Phenoxy-4,6-Dimethylaniline. To a
100-mL sidearmed flask was charged under nitrogen in succession 6.0
g (30 mmol) of 2-bromo-4,6-dimethylaniline (distilled), 15 mL of
diglyme (freshly opened bottle, anhydrous), 165 mg (1.68 mmol) of
CuCl (freshly opened bottle), 210 mg (1.45 mmol) of 8-hydroxyquino-
line, and 4.6 g (35 mmol) of freshly prepared potassium phenoxide
(from KOt-Bu/phenol/MeOH, stripped at reduced pressure and dried
in vacuo) and the mixture was heated at 90-95 °C for 16 h. The
conversion by GC area % was 91.3%, with 82.7% product and 8.7%
2,4-dimethylaniline. Another 1.0 g (7.5 mmol) of KOPh was added
and heating was continued for 8 h, at which time GC-analysis showed
88.9 area % product, 3.1% bromide, and 7.6% 2,4-dimethylaniline. The
mixture was cooled to ambient temperature, 20 mL of 3% aqueous
NH4OH was added, and the mixture was extracted with 100 mL of
hexanes. The organic layer was washed with 3% aqueous NH4OH, then
with water, decanted from tarry impurities, dried (MgSO4), filtered,
and concentrated to dryness to afford 5.6 g of crude product which
was purified by chromatography on silica gel, eluting with hexanes to
obtain 4.4 g (69%) of product which was >99 area % purity by GC-
°C, CDCl3): δ 7.4-6.7 (m, 4H, Ar), 4.57 (t, J ) 9 Hz, 2H, CH2O),
3.22 (t, J ) 9 Hz, 2H, CH2Ar). 13C NMR (125 MHz, 23 °C, CDCl3):
δ 159.5 (s, quaternary C), 127.4 (s, ArCH), 126.3 (s, quaternary C),
124.3 (s, ArCH), 119.7 (s, ArCH), 108.8 (s, ArCH), 70.4 (s, CH2O),
29.2 (CH2Ar).
Synthesis of Chroman. A 50-mL round-bottomed flask was charged
sequentially with 6 mg of Cuprous chloride (0.06 mmol), 8 mg of
2-aminopyridine (0.08 mmol), 268 mg of 3-(2-bromophenyl)-1-propanol
(1.25 mmol), and diglyme (10 mL). Then sodium methoxide (100 mg,
1.85 mmol) was added as a solid. The solution was heated to 100 °C
with stirring and kept there for 20 h. A sample was taken and GC
indicated complete conversion. Once cooled to ambient temperature,
the solution was quenched with 2 N HCl (40 mL) and the product was
extracted with 3 × 20 mL of hexane. The organic layer was washed
with 3 × 20 mL of H2O and the solvent removed. The product was
further purified by preparative TLC (ethyl acetate/hexane, 5:95). The
1
isolated yield is 119 mg (71%). H NMR (300 MHz, 23 °C, CDCl3):
δ 7.2-6.7 (m, 4H, Ar), 4.19 (t, J ) 5 Hz, CH2O), 2.80 (t, J ) 6 Hz,
2H, CH2Ar), 2.01 (m, J ) 9 Hz, 2H, CH2). 13C NMR (125 MHz, 23
°C, CDCl3): δ 154.7 (s, quaternary C), 129.6 (s, ArCH), 127.0 (s,
ArCH), 122.0 (s, quaternary C), 119.9 (s, ArCH), 116.5 (s, ArCH),
66.2 (s, CH2O), 24.7 (s), 22.2 (s).
Results and Discussion
Phenoxylation of 2-Bromo-4,6-dimethylaniline: Parent
Library. Under our standard conditions, we used 1.5 equiv of
potassium phenoxide in diglyme and 5 mol % of copper(I)
chloride and ligand relative to 2-bromo-4,6-dimethylaniline. We
did not explore the use of different copper(I) salts since bromide
ion builds up during the reaction. In one experiment, copper(I)
bromide performed equivalently within experimental error to
copper(I) chloride. Biphenyl was used as an internal gas
chromatography (GC) standard. Diglyme was the preferred
solvent for our initial experiments as this diminishes solvent
loss during the long run times (90 °C, 20 h). Twenty hours was
chosen as the run time to allow for comparison of the lowest
and highest yielding ligands. (For the phenoxylation reaction,
we were primarily interested in identifying the ligands giving
the best yield.)
The results of screening our parent library of 96 ligands are
presented in Figure 1. Because we were running a robotic
protocol and reproducibility was a concern, the data in Figure
1 show two identical sets of experiments for each ligand. The
reproducibility was good with the standard deviation being on
average (2% of the average value of the two experiments for
the two highest values. The average standard deviation for the
remaining ligands in the library was (18% of the average value
of the two measurements. The observation that the errors tend
to be smaller for those ligands that give higher overall yield
occurs in methoxylation data sets as well (see discussion on
errors below). Two ligands which were the most active clearly
stand out in both runs, i.e., 8-hydroxyquinoline and 8-acetoxy-
quinoline. It is likely that under the basic reaction conditions,
8-acetoxyquinoline is hydrolyzed to 8-hydroxyquinoline and this
explains the observed structure-activity relationship.
1
analysis, mp 55-56 °C. H NMR (CDCl3) δ 2.18 (s, 6H), 3.60 (br s,
2H), 6.57 (d, 1H, J ) 1 Hz), 6.70 (d, 1H, J ) 1 Hz), 7.0 (m, 3 H), 7.3
(m, 2H). Elemental Anal. (Calcd) for C14H15NO: C, 78.59 (78.84); H,
7.48 (7.09); N, 6.56 (6.57).
Large-Scale Synthesis of 2-Methoxy-4,6-dimethylaniline. A 250-
mL sidearmed flask containing 10.0 g (50.0 mmol) of distilled 2-bromo-
4,6-dimethylaniline in 15 mL of diglyme was flushed with nitrogen
and charged with 0.20 g (0.21 mmol) of 2-aminopyidine, 0.25 g (0.25
mmol) of cuprous chloride, and after 5 min, 16 mL (74 mmol) of 25%
methanolic sodium methoxide. The mixture was heated under nitrogen
for 24 h at reflux (85 °C), at which time GC-analysis indicated 94%
conversion with about 2% hydrodebromination. The mixture was cooled
to ambient temperature, diluted with water, and extracted with
cyclohexane. The organic layer was concentrated on a rotary evaporator,
diluted with hexanes, and washed with water to remove residual
diglyme. The hexane was removed on the rotovap, and the crude air-
sensitive product (6.9 g, 92%) was purified via the hydrochloride salt
as follows. A solution of HCl in ether (50 mL of a 1 M solution) was
added dropwise to a solution of the crude product in 20 mL of ether,
and the precipitated salt (8.4 g, 94% pure by GC, corrected yield is
84%) was filtered. Tarry impurities were removed by recrystallization
from 2-propanol to afford 4.9 g (52% yield) of 98%-pure product,
1
sublimation at 243-246 °C. H NMR (300 MHz, DMSO-d6): δ 2.28
(s, 3H), 2.34 (s, 3H), 3.85 (s, 3H), 6.64 (br s, 1H), 6.86 (br s, 1H), 9.9
(br s, 2H). Elemental Anal. (Calcd) for C9H14ClNO: C, 57.34 (57.60);
H, 7.49 (7.52); N, 7.43 (7.46).
Synthesis of 2,3-dihydrobenzofuran. A 50-mL round-bottomed
flask was charged sequentially with 18 mg of Cuprous chloride (0.18
mmol), 18 mg of 2-aminopyridine (0.19 mmol), 730 mg of 2-bromo-
phenethyl alcohol (3.63 mmol), and diglyme (20 mL). Then sodium
methoxide (300 mg, 5.55 mmol) was added as a solid. The solution
was heated to 100 °C with stirring and kept there for 20 h. A sample
was taken and GC indicated complete conversion. Once cooled to
ambient temperature, the solution was quenched with 2 N HCl (40
mL) and the product was extracted with 3 × 20 mL hexane. The organic
layer was washed with 3 × 20 mL of H2O and the solvent removed.
The product was further purified by preparative TLC (ethyl acetate/
hexane, 5:95). The isolated yield is 184 mg (42%). The spectroscopic
data fit with a sample obtained from Aldrich. 1H NMR (300 MHz, 23
We propose two methods to evaluate the performance of a
ligand library. One measure is the Average Library Production
(ALP), where j is the number of ligands in the library. For the
j
ALP ) [ (Yield Product) ]/j
∑
i
i)1
experiment in Figure 1, the ALP is 13% yield per ligand (the
average values of the two runs were used to calculate ALP).
Alternatively, one can measure a Library “Hit Frequency” (LHF)
above an arbitrarily chosen yield, that is the number of ligands

Optimization of Homogeneous Catalysts
J. Am. Chem. Soc., Vol. 122, No. 21, 2000 5047
Figure 1. Yields of 2-phenoxy-4,6-dimethylaniline versus pyridine numbers. For the identity of the pyridine ligands refer to Table 1.
Figure 2. Yields of 2-phenoxy-4,6-dimethylaniline versus the best
pyridine numbers.
above a certain threshold divided by the total number of ligands
in the library. The LHF for the library in Figure 1 is (2/96) )
0.02 at 50% yield.
Figure 3. Control experiments. Yields of 2-phenoxy-4,6-dimethyl-
aniline versus conditions. Experiments 1-4 are ligands plus copper,
5-8 are ligands alone, and 9 is with copper only.
Several other ligands give yields greater than the arbitrarily
chosen limit of 23% yield (Figure 2); the LHF is (10/96) )
0.10 at 23% yield. Most of these have structural motifs with a
pyridine donor ligand that is connected to a second donating
group. However, there are several exceptions to this such as
3-hydroxypyridine and 3-acetoxypyridine which are among the
next highest yielding cases relative to 8-hydroxyquinoline.
Control experiments were performed to demonstrate the
requirement for both ligand and copper in the catalysis reactions.
Figure 3 shows the results of runs with ligand and copper added,
ligands alone added, and copper alone added for four ligands
(8-hydroxyquinoline, 8-acetoxyquinoline, 3-hydroxypyridine,
3-acetoxypyridine). A reaction of ca. 5% yield is seen by using
Cu(I) alone as a catalyst, but clearly the combination of ligand
and copper is required to obtain high yields.
The effect of a lower ligand-to-metal ratio was investigated
(0.5:1 vs 1:1 ligand-to-copper:ligand mole ratio). For the best
set of ligands, there was no obvious effect of varying this
parameter, and within experimental error the experiments gave
identical results.22 The effect of having protic impurities was
examined by performing experiments on solutions that contained
10 mol % phenol relative to the phenoxide present. Except for
the best ligands that exhibited a ca. 10-20% decline in overall
yield, the added phenol did not have a large effect.22 (This was
of interest because it was found that the presence of methanol
affects the rate of methoxylation as discussed below).
Phenoxylation of 2-Bromo-4,6-dimethylaniline: Focused
Libraries. The results of the above experiments led us to choose
both 8-hydroxyquinoline and 3-hydroxyquinoline as the basis
for the creation of two focused or daughter libraries. The goal
was to see if an improved ALP or LHF could be obtained for
each of these classes of ligand. Hydroxyquinoline and hydroxy-
pyridine libraries were created incorporating a diverse range of
electronic and steric features. The members of each library are
listed in Tables 2 and 3, respectively.
The results of screening the hydroxyquinoline focused library
are shown in Figure 4. In comparison to Figure 1, an increase
in the number of ligands giving high yields is immediately
apparent. The six most active ligands are shown in Figure 4.
From the results, it is apparent that electron-withdrawing groups
in the 4-quinoline position were detrimental, as well as steric
(22) See the Supporting Information for details.

5048 J. Am. Chem. Soc., Vol. 122, No. 21, 2000
Table 2. Focused 8-Quinolinol Library
Fagan et al.
no.
1
ligand
no.
11
ligand
no.
21
ligand
N-methyl-2,2′-imino
bis(8-hydroxyquinoline)
quinoline
tributyl(8-quinolyloxy)tin
8-quinolinyl trifluoromethanesulfonate
2
12
4-hydroxyacridine
22
8-hydroxy-7-(4-sulfo-1-naphthylazo)-
5-quinolinesulfonic acid, disodium salt
5-nitroso-8-hydroxyquinoline
8-hydroxyquinoline-7-sulfonic acid
indooxime sodium salt
3
4
5
6
7
8-hydroxyquinaldine
5-chloro-8-hydroxyquinoline
8-hydroxy-5-nitroquinoline
8-hydroxyquinoline
13
14
15
16
17
quinoline-8-sulfonic acid sodium salt
2-amino-8-quinolinol
8-hydroxyquinoline-2-carboxaldehyde
8-hydroxyquinoline-2-carboxylic acid
5,7-dimethyl-8-hydroxyquinoline
23
24
25
26
27
quinoline yellow
8-aminoquinoline
8-hydroxyquinoline-5-sulfonic acid
monohydrate
8-hydroxyquinoline-5-sulfonic acid
dihydrate
8
5-amino-8-hydroxyquinoline
dihydrochloride
18
5-octyloxymethyl-8-quinolinol
28
9
10
5,7-dichloro-8-hydroxyquinaldine
8-aminoquinaldine
19
20
8-hydroxyquinoline-2-carbonitrile
8-hydroxyquinoline-2-sulfonic
acid monohydrate
29
30
2,8-quinolinediol
8-hydroxy-7-quinolinecarboxylic acid
Table 3. Focused 3-Hydroxypyridine Library
no.
ligand
no.
ligand
no.
ligand
1
2
3
4
3-ethoxy-2-nitropyridine
9
10
11
12
2-amino-3-hydroxypyridine
pyridoxine
3-hydroxy-6-methylpyridine
2-(dimethylaminomethyl)-
3-hydroxypyridine
17
18
19
20
2H-pyrido[3,2-b]-1,4-oxazin-3(4H)-one
pyridoxamine dihydrochloride
5-chloro-2,3-dihydroxypyridine
3-hydroxy-2-(hydroxymethyl)-
pyridine hydrochloride
3-hydroxy-2-nitropyridine
3-hydroxy-6-methyl-2-nitropyridine
2,3-dihydroxypyridine
5
6
7
8
3-hydroxy-2-mercaptopyridine
3-hydroxypicolinic acid
3-hydroxypicolinamide
13
14
15
16
2,6-lutidine-R-2,3-diol
21
22
23
3-hydroxy-2-methylpyridine
5-hydroxynicotinic acid methyl ester
oxazolo[4,5-b]pyridin-2(3H)thione
3-acetoxypyridine
3-hydroxypyridine
5-chloro-3-hydroxypyridine
2-amino-3-benzyloxypyridine
Figure 5. Yields of 2-phenoxy-4,6-dimethylaniline versus the 3-hy-
droxypyridine focused library. For the identity of the ligands refer to
Table 3.
Figure 4. Yields of 2-phenoxy-4,6-dimethylaniline versus the 8-quino-
linol focused library. For the identity of the ligands refer to Table 2.
hindrance at the 2-position. One ligand, namely the tributyltin-
derivatized 8-hydroxyquinoline, was apparently better than
8-hydroxyquinoline ligand in this experiment, although the basis
for this difference is not understood and is within experimental
error. For this library, the ALP of 23% yield per ligand and the
LHF of 0.20 at 50% yield is higher than that observed for the
parent library (ALP ) 13%; LHF ) 0.02 at 50% yield). From
these metrics we conclude that the daughter library performed
significantly better than the parent library.
The use of 8-hydroxyquinoline has been noted previously in
the patent literature for other Ullmann coupling reactions.14,23
Although by this study we have “rediscovered” this ligand, we
could not have predicted ahead of time this would be the best
ligand for these particular reaction conditions and substrate.
The results of screening the 3-hydroxypyridine library are
presented in Figure 5. For this library, we observed an increase
in ALP (20% yield per ligand) versus the parent library (13%
yield per ligand), and the LHF (7/23 ) 0.30 at 23% yield) was
also larger compared to the parent library (0.10 at 23% yield).
Of significance is that an improved ligand was identified over
the initial 3-hydroxypyridine lead ligand found in the parent
library, i.e., 2-dimethylaminomethyl-3-hydroxypyridine (53%
yield). This particular ligand has not been mentioned before in
the literature on Ullmann reactions.
Methoxylation of 2-Bromo-4,6-dimethylaniline: Parent
Library. For standard conditions, we used 1.5 equiv of sodium
methoxide in methanol, 5 mol % of both copper chloride, and
ligand relative to 2-bromo-4,6-dimethylaniline. Biphenyl was
used as an internal gas chromatography (GC) standard. Reac-
tions were run at 100 °C for 20 h. In this case, we were
interested in discovering catalytic systems giving both high
yields and low amounts of reduction.
(23) For recent applications see: (a) Nagai, R.; Myaki, K. Preparation
of fluoroalkoxybenzenes from bromofluorobenzenes. JP 1991-216630
19910802. (b) Gillet, J.-P. Process for the Preparation of Esters of
Alkoxyphenoxybenzoic Acids and their Dealkylation to Hydroxyphenoxy-
benzoic Acids. EP 1992-401918. (c) Okizaki, A.; Yoshimitsu, M.; Kubo,
M. Preparation of 4,4′-dialkoxybiphenyls. JP 1987-317365 19871217.

Optimization of Homogeneous Catalysts
J. Am. Chem. Soc., Vol. 122, No. 21, 2000 5049
Figure 6. Yields of 2-methoxy-4,6-dimethylaniline versus pyridine numbers. For the identity of the pyridine ligands refer to Table 1.
It was observed that the yield was affected by whether the
vials were capped with pierced septa or sealed septa during the
runs. This suggested the rate of evaporation of methanol affected
the rate of the reactions since runs with pierced septa tended to
exhibit higher yields than those run with sealed septa. Careful
study on large-scale reactions confirmed that the presence of
methanol retarded the rate of the methoxylation reaction. Despite
this effect, the same ligand trends were identified whether the
vials were capped or not. We found that running the reactions
at 90-100 °C was preferrable as this tended to evaporate the
methanol fairly rapidly.
We found the reproducibility of the 96 experiments to be
acceptable by rerunning the robotic protocol under the same
conditions. The results of two identical runs using our 96-
pyridine parent library are presented in Figure 6.
The reproducibility was good with the standard deviation
being on average (13% of the average value of the two
Figure 7. Yields of 2-methoxy-4,6-dimethylaniline versus the best
pyridine numbers. For the identity of the pyridine ligands refer to Table
1.
experiments for the eight highest values. The average standard
deviation for the remaining ligands in the library was (23% of
the average value of the two measurements. Two identical runs
were also carried out at 90 °C and similar results were obtained,
although total yields tended to be lower than the 100 °C run.22
The observation that the highest values tend to have less error
associated with the measurement of activity is an important
observation, and something that should be known when evaluat-
ing combinatorial catalysis results. The implication is that it
would be relatively difficult to pick out lead compounds from
a set of ligands that gave relatively low conversion. Knowing
the errors involved at different levels of activity is critical to
make judgments about the relative merit for choosing which
focused library to prepare. We therefore would recommend
duplicate runs at the minimum when establishing the validity
of a robotic/screening protocol.
The observed ALP for the parent library in the case of
methoxylation was 34% yield per ligand. The LHF was 0.19 at
50% yield. (The average values of the two runs for each ligand
were used to determine ALP and LHF.) Figure 7 illustrates that
among the best ligands for methoxylation in terms of yield and
low reduction are pyridines bearing an amino group in the
2-position. As can be seen in Figure 7, other pyridines that do
not possess this structural motif have a good overall yield of
product with a low amount of reduction byproduct. Some ligands
gave relatively high yields but suffered from an increased yield
of byproduct. The value of the combinatorial approach is shown
by these experiments, since it was impossible to predict which
ligands would give the best combination of yield and low
reduction. By this approach, we have been able to identify
several “lead” ligands for the development of focused libraries.
Intelligent Ligand Performance versus Random Ligand
Performance in Methoxylation. Out of the 96 ligands used,
79 were intelligently chosen by the criteria previously discussed.
Seventeen ligands were random selections since we either
anticipated unwanted reactivity under the very basic reaction
conditions or they had structural features that were considered
to be outside the parameters taken into account in the intelligent
set. Specifically, the “random” ligands were 5, 6, 21, 40, 45,
46, 52, 55, 76, 77, 78, 84, 86, 87, 88, 90, and 94 in Table 1.
We can use the ALP and LHF measures to examine how the
randomly chosen library members performed (ALP ) 30%;
LHF ) 0.18 at 50% yield; LHF ) 0.00 at 80% yield) compared
to the intelligently chosen set (ALP ) 44% and LHF ) 0.16 at
50% yield; LHF ) 0.04 at 80% yield). It would appear from

5050 J. Am. Chem. Soc., Vol. 122, No. 21, 2000
Table 4. Focused 2-Aminopyridine Library
Fagan et al.
no.
ligand
no.
ligand
no.
ligand
1
2
3
4
5
6
7
8
7-azaindole
2,2′-dipyridylamine
2-anilinopyridine
2-hydrazinopyridine
2-(methylamino)pyridine
2-benzylaminopyridine
2-benzylamino-4-methylpyridine
2-aminopyridine
15
16
17
18
19
20
21
22
2-amino-4-methylpyridine
2-amino-4,6-dimethylpyridine
2-amino-5-chloropyridine
2-amino-5-nitropyridine
2-amino-5-methylpyridine
2,6-diaminopyridine
2-amino-6-methylpyridine
2H-pyrido[3,2-b]-1,4-oxazin-
3(4H)-one
29
30
31
32
33
34
35
36
2-amino-3-chloro-5-trifluoromethylpyridine
2-(formylamino)pyridine
2-aminoquinolin-4-ol
2-benzylamino-6-methylpyridine
10H-pyrido(3,2-b)(1,4)benzothiazine
2-amino-8-quinolinol
2-methoxy-6-methylaminopyridine
1′,3′-dihydrospiro(cyclohexane-
1,2′-(2H)imidazo(4,5-b)pyridine)
2-pyridyldithiocarbamic acid ethyl ester
2-(2-aminoethylamino)-5-nitropyridine
2,3-diaminopyridine
9
10
11
12
13
14
2-amino-3,5-dichloropyridine
2-amino-3-nitropyridine
2-amino-4-methyl-3-nitropyridine
2-amino-3-benzyloxypyridine
2-amino-3-hydroxypyridine
2-amino-3-methylpyridine
23
24
25
26
27
28
niflumic acid
37
38
39
40
41
1-phenyl-3-(2-pyridyl)-2-thiourea
5-nitro-2-benzylaminopyridine
1-aminoisoquinoline
2-aminoquinoline
2,5-diaminopyridine dihydrochloride
piroxicam
sulfapyridine
8-quinolinol as a ligand and methoxylation with 2-aminopyridine
as a ligand were scaled up. The product from phenoxylation
was obtained in 69% isolated yield upon scale-up of the reaction.
The isolated yield of product in the case of methoxylation was
52%, and very little reduction sideproduct was observed. The
material was isolated in pure form as the hydrochloride salt (see
the Experimental Section). The moderate isolated yields are due
in part to the formation of tars which were difficult to work up
and losses during crystallization. However, the conversions by
GC area % (which we found closely related to the yields) for
the phenoxylation and methoxylation reactions were 89% and
92%, respectively. These results were similar to what we found
from small-scale screening experiments.
Alkoxide Generality. The parent library was also screened
for ethoxylation, butoxylation, isopropoxylation, and tert-
butoxylation activity. Promising results were found using NaOEt
and NaOBu as alkoxides. The same class of ligands found for
NaOMe operated efficiently.22 However, in the case of the
secondary and tertiary alkoxide (sodium isopropoxide and
potassium tert-butoxide) reactions were sluggish for all ligands
giving either very low conversions or no product at all under
our standard reaction conditions.
Solvent Effects. We examined solvents effects and found
that ether solvents (diglyme or DME) were best for the
methoxylation reaction.22 The use of DMAc or DMF reduces
the yields of product considerably (less than half of the DME
yield).22 It is noteworthy that the most efficient ligands found
for these solvents (8-hydroxyquinoline and phenanthrolines) are
much different from the class found in DME.22 These experi-
ments point out how difficult it is to predict which ligands will
function under a particular set of conditions, and again
demonstrates the usefulness of the combinatorial catalysis
approach in this case.
Substrate Generality. Having found that 2-aminopyridine
was a good ligand for Ullmann reactions in ether solvents, we
explored the generality of this ligand for use with other
substrates. Table 5 summarizes these results for other substrates
run under our standard conditions. Overall, the data suggest this
ligand performs quite well for most of the substrates examined.
We are depending solely on GC/mass spectral data to determine
that the major observed product had the expected mass, but it
should be noted we have not performed these reactions on a
large scale. The low yield observed for â-bromostyrene is
associated with the formation of the byproducts tentatively
identified by GC/mass spectral analysis as 1,4-diphenylbutadiyne
(ca. 59%) and 2-phenylnaphthalene (ca. 7%).
Figure 8. Yields of 2-methoxy-4,6-dimethylaniline versus the 2-ami-
nopyridine focused library. For the identity of the ligands refer to Table
4.
these data that the intelligently chosen library performed better
than the random set. The overall ALP was slightly higher, and
the quality of hits (i.e., those that gave higher yields) was better
in the intelligent set (see LHF’s at 80% yield). One should be
cautious not to overinterpret these results, since the exact line
between randomly chosen versus intelligently chosen ligands
is partially subjective. However, we did try to cover all of the
important parameters (sterics, electronics, bite-angle, etc.) with
the intelligently chosen library, and the higher quality of hits
from the intelligent library members suggests some advantage
to this approach.
Methoxylation of 2-Bromo-4,6-dimethylaniline: Focused
Library. On the basis of the results described above, we
prepared a daughter library of 2-aminopyridine type ligands
which are listed in Table 4. This class of ligands was chosen
because of the observed high yields with low levels of reduced
side product. They were also readily available commercially,
and offered some practical benefits if this chemistry was to be
used in a commercial process. Results of screening this library
are shown in Figure 8. We observed an increase in the ALP
(55% yield per ligand) and LHF (0.68 at 50% yield) relative to
the original library of ligands which had ALP at 34% per ligand
and an LHF of 0.19 at 50% conversion. As in the phenoxylation
reaction, the parent library again led us to a ligand class that
functions in an improved manner by these metrics.
Scale-Up of Reactions. It was important to validate the
findings of small-scale reactions to see if similar results could
be obtained on a large scale. Both the phenoxylation with
Intramolecular Ullmann Ether Formation. Reports of
intramolecular Ullmann ether reactions are rare. There are

Optimization of Homogeneous Catalysts
J. Am. Chem. Soc., Vol. 122, No. 21, 2000 5051
Table 5. Substrates and Observed Gas Chomatographic Yields of Product Using 2-Aminopyridine as Ligand
entry
name
GC yield (%)
entry
name
GC yield (%)
1
2
3
4
5
6
7
8
9
2-bromoanisole
98
99
93
97
93
90
90
34
99
11
12
13
14
15
16
17
18
19
20
4-bromobenzonitrile
1-bromonaphthalene
2-bromothiophenol
1-bromo-2-nitrobenzene
1-bromo-4-nitrobenzene
2-bromoaniline
3-bromoaniline
4-bromoaniline
4-bromoveratrole
2-bromo-4,6-dimethylaniline
100
100
-
100
100
89
85
91
83
81
2-bromothioanisole
4-bromo-m-xylene
2-bromo-m-xylene
3-bromoanisole
4-bromoanisole
4′-bromoacetophenone
â-bromostyrene
9-bromoanthracene
2-bromobenzonitrile
10
100
sporadic reports of copper-catalyzed reactions17c,f,g and a pal-
ladium route using secondary or tertiary alcohols was recently
reported by Buchwald.18g The ligand 2-aminopyridine was
shown to be effective for intramolecular Ullmann reactions as
well. Reaction of 2-bromophenethyl alcohol and 3-(2-bromo-
phenyl)-1-propyl alcohol proceed cleanly and in good yields
forming 2,3-dihydrobenzofuran and chroman, respectively (eq
2).
We believe part of the key to success in the current study
was that our initial design of the parent library included an
intelligent aspect, i.e., we took into account the typical
parameters which affect homogeneous catalyst performance.
Whether the Ullmann reaction is monometallic, bimetallic, or
metal cluster mediated, the same ligand parameters we took into
consideration are likely to be important in one sense or another.
Overall, there is the trend that bidentate chelators with a
relatively small bite angle are favored to work well (2-
aminopyridines, 8-hydroxyquinolinols, etc.). However, there are
many exceptions to this general trend, and the appearance of
ligands such as for example proflavine in some of the screens
defies ready explanation. Because of the complexity of this
reaction, it is likely our intelligently chosen parameters are not
the sole important ones. This would argue for the need for
“random” member selection in any library design aimed at a
poorly understood reaction. The bottom line for this particular
study is that this approach of intelligent/random library screening
led us to the correct ligands for our particular needs.
Conclusion
In summary, using combinatorial chemistry methods, we have
been able to scout a large array of catalyst solutions in a
relatively short period of time. By starting with a ligand library
that contained both intelligently chosen and random members,
we were able to determine some initial lead ligands for both
phenoxylation and methoxylation reactions for the substrate of
interest. From the lead compounds, we created focused libraries
that performed better than the parent library, and in some cases
resulted in the discovery of new ligands.
Acknowledgment. We would like to thank Ronald J. Davis
and Gary Halliday for technical assistance.
The impetus to use combinatorial techniques in catalysis was
the assumed analogy to drug discovery. In the pharmaceutical
paradigm, screening for lead compounds for enzyme inhibition
would lead to structure-activity relationships that could be
expanded into more focused libraries for screening and eventual
optimization of inhibition. No studies to our knowledge have
clearly demonstrated that this approach can be successful for
homogeneous catalysis. The data presented in this paper lead
us to conclude that in certain instances, this paradigm can work
in homogeneous catalysis discovery and optimization.
Supporting Information Available: Graphs of ligand ratio
effects and additive effect for the phenoxylation reaction; graph
of the methoxylation reaction at 90 °C; graphs and tables of
data describing solvent effects (DME, DMAC, and DMF) for
the methoxylation reaction; graphs and tables of data for the
ethoxylation and n-butoxylation reaction (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA000094C