Yiming Zhao1, Huy van Nguyen1, Louise Male1, Philip Craven1, Benjamin R Buckley2, John S Fossey1. 1. School of Chemistry and X-ray Crystallography Facility, School of Chemistry, University of Birmingham, Edgbaston, Birmingham, West Midlands B15 2TT, United Kingdom. 2. Department of Chemistry, Loughborough University, Loughborough, Leicestershire LE11 3TU, United Kingdom.
Abstract
Twelve 1,5-disubtituted and fourteen 5-substituted 1,2,3-triazole derivatives bearing diaryl or dialkyl phosphines at the 5-position were synthesized and used as ligands for palladium-catalyzed Suzuki-Miyaura cross-coupling reactions. Bulky substrates were tested, and lead-like product formation was demonstrated. The online tool SambVca2.0 was used to assess steric parameters of ligands and preliminary buried volume determination using XRD-obtained data in a small number of cases proved to be informative. Two modeling approaches were compared for the determination of the buried volume of ligands where XRD data was not available. An approach with imposed steric restrictions was found to be superior in leading to buried volume determinations that closely correlate with observed reaction conversions. The online tool LLAMA was used to determine lead-likeness of potential Suzuki-Miyaura cross-coupling products, from which 10 of the most lead-like were successfully synthesized. Thus, confirming these readily accessible triazole-containing phosphines as highly suitable ligands for reaction screening and optimization in drug discovery campaigns.
Twelve n class="Chemical">1,5-disubtituted and fourteenpan> class="Chemical">5-substituted 1,2,3-triazole derivatives bearing diaryl or dialkyl phosphines at the 5-position were synthesized and used as ligands for palladium-catalyzed n class="Disease">Suzuki-Miyaura cross-coupling reactions. Bulky substrates were tested, and lead-like product formation was demonstrated. The online tool SambVca2.0 was used to assess steric parameters of ligands and preliminary buried volume determination using XRD-obtained data in a small number of cases proved to be informative. Two modeling approaches were compared for the determination of the buried volume of ligands where XRD data was not available. An approach with imposed steric restrictions was found to be superior in leading to buried volume determinations that closely correlate with observed reaction conversions. The online tool LLAMA was used to determine lead-likeness of potential Suzuki-Miyaura cross-coupling products, from which 10 of the most lead-like were successfully synthesized. Thus, confirming these readily accessible triazole-containing phosphines as highly suitable ligands for reaction screening and optimization in drug discovery campaigns.
Click chemistry, as
defined by Sharpless and co-workers,[1−4] has transformed the face, and accessibility
to the nonspecialist
of molen class="Chemical">cular linking strategies. Among click approaches, the highly
regioselective n class="Chemical">copper-catalyzed, Huisgen cycloaddition reaction to
form 1,4-triazoles (Figure , 1) by Meldal and Sharpless,[4,5] has
grown in popularity and has been employed in increasingly varied applications
over the intervening years.[6] The copper-catalyzed
azide–alkyne cycloaddition (CuAAC) is ubiquitous,[5,7−9] and often referred to as “the click reaction”,
and the product 1,2,3-triazoles, bearing 1,4-substition patterns,
with reliable fidelity and yields being synonymous with click chemistry
(Figure , 1).[8,10−15] However, the resulting triazoles are not always employed as innocent
bystander linkage motifs. Triazoles of this type have been used as
analogues of peptide linkages (Figure , 2),[16] such
peptidomimetics are physiologically stable, and their modular synthesis
allows access to a broad range of biologically relevant applications.[17,18] The utility of further synthetic transformations has been probed,
particularly in the derivatization at the 5-position (Figure , 3).[19,20]
Figure 1
Upper:
copper-catalyzed azide–alkyne cycloaddition (CuAAC).
Lower: examples of 1,2,3-triazole-containing structures.
Upper:
n class="Chemical">copper-catalyzed n class="Chemical">azide–n class="Chemical">alkyne cycloaddition (CuAAC).
Lower: examples of 1,2,3-triazole-containing structures.
n class="Chemical">Copper-catalyzed n class="Chemical">triazole formation has been exploited
in a wide
range of scenarios[21] and has been the subject
of various mechanistic studies,[22−26] leading to the proposal of a binuclear transition state involving
two n class="Chemical">copper atoms. 1,2,3-Triazole derivatives have been employed as
nitrogen-coordinating ligands,[27] e.g. in N,N′-,[28−34] N,S-,[29] N,Se-[29] and cyclometalated[35−37] bidentate coordination complexes. Furthermore, tris-triazoles,
such as TBTA (Figure , 4) and its analogues, have been used as ligands for
reactions including the CuAAC by both Fokin[38] and Zhu[39] and their co-workers, demonstrating
exceptional ligand-mediated reaction acceleration.
Figure 2
Examples of triazole-containing
or -derived ligands.
Examples of n class="Chemical">triazole-containing
or -derived ligands.
Alkylation of a n class="Chemical">1,2,3-triazole derivative, to furnclass="Chemical">pan>ish a n class="Chemical">1,3,4-trisubstituted
n class="Chemical">triazolium salt, is the first step in the synthesis of a newer class
of N-heterocyclic carbene (NHC) reported in 2008
by Albrecht and co-workers[40] and later
extended by Lee and Crowley[41−43] as ligands for gold and by Grubbs
and Bertrand as ligands for ruthenium-mediated catalysis (Figure , 5).[44] The ready access to a range of ligand scaffolds
through the CuAAC has led the triazole–NHC platform to continue
to gain in popularity.[45−49] These reports demonstrate the 1,2,3-triazole unit is a legitimate
candidate for further exploitation as a key component in ligand design
and catalyst development.[50]
The co-authors
of this report have investigated n class="Chemical">triazoles as chemosensors[51−53] and as products of asymmetric synthesis.[54−58] Furthermore, an interest inn class="Chemical">boronic acid derivatives
as chemosensors,[59−64] including the use of cross-coupling reactions as a means of sensing,[65] means the co-authors of this report are familiar
with boronic acids and esters.[66] As such,
a desire to bring together these streams of research under one umbrella
has led to the research reported in this manuscript. Herein, the development
of bulky and highly active triazole-containing phosphine ligands for
palladium-catalyzed Suzuki–Miyaura cross-coupling reactions
is explored (Scheme ).
Scheme 1
Outline of a General Palladium-Catalyzed Suzuki–Miyaura
Formation
of Biaryl 6
n class="Chemical">Palladium-catalyzed cross-coupling reactions are well-studied
and
offer ready access to an extensive range of (bi)aryl motifs (Scheme , 6).[67−76] The suite of n class="Chemical">palladium-catalyzed cross-coupling chemistry available
for the construction of biologically relevant products has transformed
the field of medicinal chemistry, yet cross-coupling between sterically
hindered and lead-like building blocks, particularly with aryl chlorides,
remains challenging.[77]
Developments
inn class="Chemical">palladium-catalyzed chemistry have been heavily
inclass="Chemical">pan>fluenced by ligand design and optimization. Among the superior ligands
for n class="Chemical">palladium-catalyzed cross-coupling reactions are bulky alkyl phosphines,[78−82] and bulky ortho-substitutedaryl-alkyl phosphines,[83,84] such as S-Phos (Figure , 7)[85] and X-Phos
(Figure , 8).[86] Metallocyclic precatalysts have been
developed which delivered greater stability, more facile manipulation
and enhanced reaction outcomes.[87−89]
Figure 3
Representative examples of bulky phosphine
ligands including: upper:
S-Phos (7) and X-Phos (8); middle left:
heteroaromatic phosphine derivative (9); middle right
and lower row: Phosphino-ferrocene derivatives displaying planar chirality.
Representative examples of bulky n class="Chemical">phosphine
liganclass="Chemical">pan>ds including: upper:
n class="Chemical">S-Phos (7) and X-Phos (8); middle left:
heteroaromatic phosphine derivative (9); middle right
and lower row: n class="Chemical">Phosphino-ferrocene derivatives displaying planar chirality.
n class="Chemical">Phosphines appended to one or
more five-membered, all sp2 rings have been shown to offer
advanpan>tages in some cases. The pan> class="Chemical">pyrrole-appended
phosphines of Beller and co-workers (Figure , 9)[90,91] have proven to be useful ligands in cross-coupling catalysis that
furnishes drug-like products. Furthermore, the bulky, electron-rich,
ferrocene-appended phosphines of Richards (Figure , 10),[92] Johannsen (Figure , 11),[93] Fu (Figure , 12),[94] and their respective co-workers provide access
to highly active palladium-ligand conjugates for cross-coupling some
of the least active substrates.
n class="Chemical">1,4-Disubstituted-1,2,3-triazoles
have been used in the assembly
of n class="Chemical">phosphorus-containing species that have the potential to be employed
as ligands. Ready access to libraries of products using CuAAC reactions
has permitted the development of phosphines connected to n class="Chemical">triazoles
with sp3 linkages to the 1-N and 4-C positions. Examples
include those reported by Dubrovnia, Börner and co-workers
(Figure , 13),[95] Gandelman and co-workers who prepared
PCP′ pincer complexes from bis-phosphinotriazoles (Figure , 14),[96] and Kann and co-workers who used
a borane-protected P-chiral azide to deliver protected P-chiral triazole-containing phosphines (Figure , 15).[97] Since proximal stoichiometric phosphine can
arrest the CuAAC through coordinative saturation of substoichiometic
copper catalyst or unwanted Staudinger-type reactions between phosphine
and azide derivative, alternative approaches are required to deliver P-appended triazoles. Zhang and co-workers reported on the
use of alkynyl Grignard reagents for the synthesis of a phosphine
series where phosphorus is attached directly to a 1,2,3-triazole ring
at the 4-position (Figure , 16).[98−101] As part of an impressive, rigorous, and
detailed study, Balakrishna and co-workers prepared not only expected
aryl phosphinetriazole derivative 17a (Figure ) through lithium-halogen exchange
and subsequent reaction with diphenylphosphorus chloride on the corresponding
bromide under kinetic control but also thermodynamically favored 5-phosphinotriazole 18a (Figure ) from the same aryl bromide starting material under
different conditions.[102] Fukuzawa and co-workers
also employed deprotonation of the 5-position of a 1,2,3-triazole
to facilitate installation of 5-phosphino functionality in their ferroncenyl
bisphosphino ligand synthesis (Figure , 19).[103] Glover
et al.[104] and Austeri et al.[105] also utilized deprotonation of the triazole
5-position to create planar chiral cyclophane-containing analogues
of 18a. In order to successfully synthesize a bis-5-phosphino-triazole
bidentate ligand, Manoury, Virieux, and co-workers employed an ethynyl
phosphine oxide in their homocoupled dimer synthesis, which after
treatment with trichlorosilane resulted in 20 (Figure ).[106] A structurally related 5–18 hybrid system was also recently reported by Cao et al., who showed
the formation of bimetallic complexes of NHC–phosphine mixed
systems.[107]
Figure 4
Phosphorus-containing
1,2,3-triazole derivatives.
n class="Chemical">Phosphorus-containing
n class="Chemical">1,2,3-triazole derivatives.
Bulky, or sterically hindered, n class="Chemical">phosphorus-conclass="Chemical">pan>taining ligands
have
also founpan>d application outside of the n class="Chemical">palladium catalysis arena; steric
parameters appear to be important in a variety of gold-mediated transformations.[108−110] Steric and electronic parameters of phosphine ligands have been
a subject of study for more than 40 years.[111] More recent contributions have built upon Tolman’s concept
of cone angle as a descriptor of the steric bulk a ligand imparts
about a metal, resulting in a parameter known as buried volume (%VBur) coming to the fore.[112] Cavallo and co-workers have developed a free web-based
tool for the calculation of %VBur, named
SambVca.[113] This parametrization of ligands,
using both spectroscopic measurements and calculated properties (using
principle component analysis for example), has facilitated exceptional
ligand design and optimization across a range of catalyzed reactions.[114−116]
While a range of ligand-based solutions for cross-coupling
reactions
exist, a platform to rapidly deliver alternatives and explore chemical
space around novel catalyst constructs, such as through n class="Chemical">CuAAC, offers
approaches and complementary tools to the field. Furthermore, cross-coupling
catalysis manifolds that provide access to increasingly three-dimensional
products,[117] and those directly deliverinclass="Chemical">pan>g
products with lead- or drug-like properties[118] without need of protecting group removal or further derivatization
are desired and less-well explored.[77]
Results
and Discussion
In order to overcome the general incompatibility
of n class="Chemical">phosphines
with the n class="Chemical">CuAAC reaction, an approach other than direct phosphine incorporation
is required. While protection of the phosphine (as a phosphine oxide
for example)[106] is possible, the initial
approach chosen in this program was to probe the potential for triazole-mediated
directed ortho-lithiation or halogen-lithium exchange,
and subsequent reaction with diphenyl phosphorus chloride, as possible
routes to phosphino-triazoles 17a–c. Accordingly, 1,4-diphenyl-1,2,3-triazole (21a) was
selected as a model substrate to test if ortho-lithiation
could deliver the required intermediate to give the desired product 17a. As such, 21a was reacted at −78 °C
with n-butyllithium, followed by addition of the
phosphorus chloride reagent, before being allowed to warm to room
temperature, and being allowed to stir for a period of time. Over
numerous attempts only the unanticipated product 18a was
isolated from such reactions, Scheme . This indicates that deprotonation of the 5-position
of the triazole is facile, resulting in formation of the observed
major product. In order to mitigate against the formation of unanticipated
triazole-phosphine 18a, brominated triazole derivatives 22a–c were employed under a similar protocol,
with varying amounts of n-butyllithium. In all cases,
the same 5-phosphino-triazole product, 18a, was obtained,
shown in Scheme .
In the case of 22a, it was possible to isolate small
quantities of the desired product 17a. X-ray crystal
structures of both 17a and 18a were determined
as shown in Figure .
Scheme 2
Reaction of 1,2,3-Triazoles 21a–d with n-Butyllithium Followed by Treatment with
Diphenyl Phosphorous Chloride Led Primarily to Formation of Phosphino-Triazole 18a
Figure 5
Representation of the
crystal structures of isomeric 17a (left) and 18a (right), ellipsoids drawn at the 50%
probability level (Ortep3 for Windows and PovRay). For 17a the structure contains two crystallographically independent molecules
with only one shown for clarity. For 18a the phenyl-appended
1-nitrogen and 4-carbon atoms of the triazole unit are disordered
such that the triazole ring occupies two opposing orientations, related
by a 180° rotation of the triazole ring about the phosphorus-triazole
bond. The refined percentage occupancy ratio of the two positions
are 59.7 (15) and 40.3 (15); one arbitrary molecule depicted and hydrogen
atoms removed for clarity.
Representation of the
crystal structures of isomeric 17a (left) and 18a (right), ellipsoids drawn at the 50%
probability level (Ortep3 for Windows and PovRay). For 17a the structure contains two crystallographically independent molen class="Chemical">cules
with only one shown for clarity. For 18a the phenyl-appended
n class="Chemical">1-nitrogen and 4-carbon atoms of the triazole unit are disordered
such that the triazole ring occupies two opposing orientations, related
by a 180° rotation of the triazole ring about the phosphorus-triazole
bond. The refined percentage occupancy ratio of the two positions
are 59.7 (15) and 40.3 (15); one arbitrary molecule depicted and hydrogen
atoms removed for clarity.
In fact these observations should not have been at all unanticipated.[102,104] The aforementioned report of Balakrishna and co-workers had previously
probed the reaction of 17a in more detail than us under
analogous conditions[102] and determined
a “kinetic” and “thermodynamic” relationship
betweenn class="Chemical">lithium-n class="Chemical">halogen exchange alone, versus lithium-halogen exchange
followed by lithium (triazole-) proton exchange leading to products 17a and 18a respectively, Scheme .
Scheme 3
Previously Reported Kinetic (Upper) and
Thermodynamic (Lower) Lithiation
and Subsequent Phosphorous Addition to Brominated Triazole 22a Affording P-Aryl and P-Triazole Derivatives, Respectively
Ref (102).
Previously Reported Kinetic (Upper) and
Thermodynamic (Lower) Lithiation
and Subsequent Phosphorous Addition to Brominated Triazole 22a Affording P-Aryl and P-Triazole Derivatives, Respectively
Ref (102).Attempts to block
the n class="Chemical">triazole 5-H position, using the n class="Chemical">deuterium
masking approach deployed by Richards and co-workers in the preparation
of ferrocene derivatives,[119] did not dramatically
modify reaction outcomes in our hands. Variously deuterated products
were always obtained from attempted lithiation-mediated access to
products under routine conditions.
Since the scope of 18a-like ligands had not been investigated
beyond the three n class="Chemical">triazole backbones reported by Glover et al.[104] anpan>d Choubey et al.,[102] we chose to fopan> class="Chemical">cus attention on triazole 5-H lithiation to deliver
a range of potential ligands for cross-coupling catalysis. To this
end, alkynes 23a and 23b reacted smoothly
with azides 24a–f to furnish 1,5-disubstituted1,2,3-triazoles 21a–i in acceptable
to good yields, shown in Scheme .
Scheme 4
CuAAC Reaction to Form 1,4-Disubstituted 1,2,3-Triazole
Derivatives 21a–i
Applying the aforementioned n class="Chemical">triazole lithiation
and subsequenpan>t
quench with dicyclohexyl-, di-iso-propyl-, or diphenyl-
n class="Chemical">phosphorus chloride reagent protocol to isolated 21a–i triazole set, with the express expectation of generating
5-phosphino triazole derivatives, delivered 12 targeted n class="Chemical">phosphino-triazoles 18a–l in acceptable to good yields (Scheme ).
Scheme 5
Deprotonation and
Subsequent Reaction with Phosphorous Chloride Reagent
Protocol for Delivery of 18a–l
In order to benchmark the catalytic
capability of ligands in this
report, the n class="Chemical">palladium-catalyzed Suzuki–Miyaura reaction of
n class="Chemical">2-bromo-m-xylene (25) and ortho-tolylboronic acid (26) under standard conditions (1
mol % palladium, 2 mol % ligand, three equivalents of base, 10 h,
toluene, 90 °C, see the Supporting Information) was compared. The catalyzed formation of compound 27 represents a challenging but achievable cross-coupling: While an
aryl bromide is employed in the reaction, the product (a triply ortho-substitutedbiaryl) is sterically congested about
the formed bond. Diphenyl aryl phosphine 17a and 5-phosphinotriazoles 18a–l were employed as
ligands in the benchmark reaction (Table ). The use of 17a as ligand
(Table , entry 1)
resulted in 29% conversion to product 27. The 5-phosphino
isomer of 17a, 18a, also gave less than
50% conversion to desired product 27 in the same reaction
(Table , entry 2).
As may be expected, switching the diphenylphosphine part of 18a to dialkylphosphine groups di-iso-propyl
(18b) and dicyclohexyl (18c) improved the
reaction outcomes, resulting in 62 and 75% conversion, respectively
(Table , entries 3
and 4). Changing the alkyne-derived part of the triazole from phenyl
(18a) to cyclohexyl (18d) gave a marked
improvement delivering product 27 in 86% conversion (Table , entry 2 versus 5).
Conversion determined by inspection
of the corresponding 1H NMR spectra of crude reaction isolates.
See the Supporting Information for details.
Meaning derived from naphthyl
azide.
Reaction conditions: 2-bromo-m-xylene
(0.4 mmol), o-tolylboronic acid
(0.6 mmol), potassium phosphate (1.2 mmol), Pd2(dba)3 (0.5 mol %), ligand (2 mol %), toluene (3 mL), 10 h, 90 °C.Conversion determined by inspection
of the corresponding n class="Chemical">1H NMR spectra of crude reaction isolates.
See the Supporting Information for details.
Meaning derived from naphthyl
n class="Chemical">azide.
While good results
were obtained in the Suzuki–Miyaura cross-coupling
reactions with 18-derived catalysts the more effective
ligands (generally larger) suffered somewhat from poor solubility.
Thus, ligand modifications that retained activity but allowed for
more ready synthesis and manipulation at larger scale were sought.
From briefly surveying the results in Table it was concluded that changes in the R1 (n class="Chemical">alkyne-derived) part (e.g., entry 8 versus entry 9) were
less influential on the reaction outcome than changes in the R2 (n class="Chemical">azide-derived) part (e.g., entry 5 versus entry 13). Reasoning
that smaller ligands may benefit from enhanced solubility and tractability,
a strategy to retain the N-substituents (azide-derived
parts) while minimizing the alkyne-derived parts was chosen for further
elaboration. Gevorgyan and co-workers have already reported that 1,5-disubstituted1,2,3-triazoles may be accessed by selective reaction at the 5-position
of 1-substituted triazoles (Figure shows the electrostatic potentials of the triazole-carbons
they determined), and coupled with the synthetic strategies reported
by Oki et al.[103] for chiral bisphosphine
synthesis, this led to the conclusion that 1-substitutued triazoles
may be readily converted to a library of 1-substituted, 5-phosphino1,2,3-triazoles
Figure 6
Electrostatic potential charges as determined by Gevorgyan
and
co-workers,[19] indicates rationale for selective
C-5 deprotonation.
Electrostatic potential charges as determined by Gevorgyan
and
co-workers,[19] indicates rationale for selective
C-5 deprotonation.Following the optimized
protocol of Oki et al.[103] deployed in the
synthesis of more complex constructs, a
range of n class="Chemical">1-substituted triazoles were synthesized. Specifically, n class="Chemical">trimethylsilylacetylene
(23c) and aryl azides (24a–i) were exposed to CuAAC reaction conditions (Scheme ) that led to effective triazole
formation and desilylation in one pot. Acceptable to good yields of
1-substituted triazoles 28a–i were
isolated after 24 h room temperature reactions.
Scheme 6
Trimethylsilylacetylene
(23c) and Aryl Azides (24a–i) React under Desilylative CuAAC
Reaction Conditions to Deliver 1-Substituted 1,2,3-Triazoles (28a–i)
n class="Chemical">Triazoles 28a–i reacted
smoothly
under the deprotonation and n class="Chemical">phosphorus chloride reagent quench reaction
conditions described earlier. Deprotonation at −78 °C,
by treatment with n-butyllithium, followed by addition
of dicyclohexyl-, di-iso-propyl-, di-tert-butyl- or diphenyl-phosphorus chloride at the same temperature (Scheme ) resulted in formation
of the desired triazole-containing phosphines 29a–n in acceptable to good yields. The X-ray crystal structure
of 29g was determined (Figure ); the orientation of the molen class="Chemical">cule (in the
solid state) is such that the lone pair of the phosphine is oriented
to the same direction as the 1-aryl substituent of the triazole. In
turn, this orientation about a central five-membered ring describes
a relatively wide binding pocket for metals with potential for arene-metal
interactions alongside primary phosphorus–metal ligation.
Scheme 7
Deprotonation of 1-Aryl 1,2,3-Triazole Derivatices and Subsequent
Reaction with Phosphorous Chloride Reagent Protocol for Delivery of 29a–n
Figure 7
Representation of the crystal structure of 29g, ellipsoids
drawn at the 50% probability level (Ortep3 for Windows and PovRay),
hydrogen atoms omitted for clarity.
Representation of the crystal structure of 29g, ellipsoids
drawn at the 50% probability level (Ortep3 for Windows and PovRay),
n class="Chemical">hydrogen atoms omitted for clarity.
Ligands 29a–n were tested
in the
aforementioned benchmark Suzuki–Miyaura cross-coupling reaction
of 25 with 26 catalyzed by a n class="Chemical">palladium–pan> class="Chemical">phosphine
complex, to produce biaryl 27 (Table , entries 1–14). Under the same conditions,
commercially sourced ligands, S-Phos (7) and X-Phos (8) were also used for comparison (Table , entries 15 and 16, respectively). Diphenyl-phosphino
triazole derived ligands failed to deliver product 27 in good yields, under the conditions employed the best conversion
for this ligand class was only 54% (ligand 29c, Table , entry 3), whereas
dialkyl-phosphino triazoles gave universally excellent conversion,
equal to the commercially sourced S- and X-Phos in performance, under
these conditions.
Table 2
Ligand Screening: 1-Substituted 1,2,3-Triazole-Containing
Phosphine Ligand Mediated, Palladium-Catalyzed, Reaction of Arylbromide 25 in the Formation of 27
Conversion determined
by inspection
of the corresponding 1H NMR spectra of crude reaction isolates.
See the Supporting Information for details.
Derived from 2-naphthyl azide.
Reaction conditions: 2-bromo-m-xylene (0.4 mmol), o-tolylboronic acid
(0.6 mmol), potassium phosphate (1.2 mmol), Pd2(dba)3 (0.5 mol %), ligand (2.0 mol %), toluene (3 mL), 10 h, 90
°C.Conversion determined
by inspection
of the corresponding n class="Chemical">1H NMR spectra of crude reaction isolates.
See the Supporting Information for details.
Derived from n class="Chemical">2-naphthyl azide.
While pleased to have created
ligands offering good performance
in a benchmark reaction, the reaction itself did not offer enough
diversity of outcomes to evaluate dialkyl-n class="Chemical">phosphino ligand performances
against each other nor against readily available commercial ligands 7 and 8. Next, a more sterically demanclass="Chemical">pan>ding test-reaction
was chosen to evaluate the ligands further. The formation of n class="Chemical">biaryl
C–C bonds where the formed bond is flanked by four ortho-substituents presents a particularly challenging yet
attractive transformation, not in the least due to the apparent three-dimensional
nature of the cross-coupled products.[117] The reaction of bromide 30 with boronic acid 31 was selected as one such reaction to probe catalyst effectiveness
(Table ). Conversion
to product 32 may be monitored by gas chromatographic
analysis, facilitating ready comparison of reactions performed in
parallel. Having given quantitative conversions to product 27 (Table , entries
7, 8, 13, and 14), and being both relatively easy to synthesize and
available in sufficient quantities, ligands 29g, 29h, 29m, and 29n were selected
for further investigation. These ligands are triazole-analogues of
leading n class="Chemical">phenylene ligands 7 and 8; thus,
in order to probe any specific advantages of triazole-core ligands
they were compared directly against S-Phos and X-Phosphenylene ligands.
(For retained and compared ligands, see Figure .)
Table 3
Ligand
Screening: 1-Substituted 1,2,3-Triazole-Containing
Phosphine Ligand Mediated, Palladium-Catalyzed, Reaction of Arylbromide 30 in Formation of 32
Determined by GC analysis with n-dodecane
as internal standard.
Pd2(dba)3 (0.5
mol %), ligand (2 mol %).
Figure 8
Ligands compared in Table and Table .
Ligands compared in Table and Table .
Table 4
Ligand Screening: 1-Substituted 1,2,3-Triazole-Containing
Phosphine Ligand-Mediated, Palladium-Catalyzed, Reaction of Arylchloride 37 in the Formation of 27
Determined by GC
analysis with n-dodecane as internal standard.
Pd2(dba)3 (0.25
mol %), ligand (1.0 mol %).
Reaction conditions:
2-bromomesitylene
(0.5 mmol), 2,6-xylylboronic acid (1.0 mmol), potassium phosphate
(2 mmol), Pd2(dba)3 (2.0 mol %), ligand (8.0
mol %), toluene (3 mL), 18 h, reflux.Determined by GC analysis with n class="Chemical">n-dodecane
as internal standard.
n class="Chemical">Pd2(dba)3 (0.5
mol %), ligand (2 mol %).
In order to ensure good reaction conversions, the catalyst loading,
temperature, and reaction time were all increased in comparison to
the earlier Suzuki–Miyaura reactions. The standard conditions
employed in the comparisons of Table (entries 1 and 3–7) were 4 mol % n class="Chemical">palladium
and 8 mol % ligand inn class="Chemical">toluene at reflux for 16 h. Under these conditions,
cyclohexyl-substituted triazole-containing ligands 29g and 29m gave higher conversions to 32 than
did their tert-butyl-substituted analogues 29h and 29n (Table , entries 1 and 4 (99 and 60%) versus entries
3 and 5 (47 and 13%), respectively). In this comparison, the use of
S-Phos (7) as ligand gave 87% conversion (Table , entry 6) and the use of X-Phos
(8) as ligand gave 50% conversion (Table , entry 7) to compound 32 (under
these conditions). Since ligand 29g gave the best conversion
(under the conditions employed), catalyst loading was reduced to 1
mol % palladium (in the form of 0.5 mol % Pd2(dba)3) alongside 2 mol % 29g as ligand (Table , entry 2), and under these
conditions, a conversion of 84% to 32 was achieved.
To further probe the utility of n class="Chemical">1-aryl 5-phosphino 1,2,3-triazoles
as ligands in Suzuki–Miyaura catalysis the synthesis of compounpan>d 27 from n class="Chemical">aryl chloride 37 and boronic acid 26 was investigated (shown in Table ) deploying the same
ligand set as in Table . The reaction conditions mirrored those used earlier for cross-coupling
with bromide analogue 25 in Table , but in order to ensure good reaction conversions
the reaction temperature was increased slightly (toluene at reflux).
Reaction conditions: 2-chloro-m-xylene (1.0 mmol), o-tolyboronic acid
(1.5 mmol), potassium phosphate (3.0 mmol), Pd2(dba)3 (0.5 mol %), ligand (2.0 mol %), toluene (3 mL), 10 h, 90
°C.Determined by GC
analysis with n class="Chemical">n-dodecane as internal standard.
n class="Chemical">Pd2(dba)3 (0.25
mol %), ligand (1.0 mol %).
Under the reaction conditions employed (Table ), catalysts derived from n class="Chemical">triazole-containing
liganclass="Chemical">pan>ds 29g, n class="Chemical">29h, and 29m delivered
compound 27 in quantitative yield (Table , entries 1, 3, and 4, respectively). Using 29g as ligand at a lower catalyst loading of 0.5 mol % of
n class="Chemical">palladium proportionally, quantitative conversion to 27 (isolated yield 92%) was achieved. In this comparison, the use of
S-Phos (7) as ligand gave 33% conversion (Table , entry 6), and the use of X-Phos
(8) as ligand gave quantitative conversion (Table , entry 7) to compound 27.
Next, a demanding n class="Chemical">palladium-catalyzed reaction betweenclass="Chemical">pan> class="Chemical">2-chloro-meta-xylene (37) and 2,4,6-trimethylphenylboronic
acid (38) leading to product 32 was attempted.
Two triazole ligand-based catalyst systems were compared against catalyst
systems derived from S-Phos (7) and X-Phos (8); see Table .
Table 5
Ligand Screening: 1-Substituted 1,2,3-Triazole-Containing
Phosphine Ligand Mediate, Palladium-Catalyzed, Reaction of Arylchloride 37 in Formation of 32
Determined by GC analysis with n-dodecane as internal standard.
Reaction conditions:
2-chloro-m-xylene (0.5 mmol), 2,4,6-trimethylphenylboronic
acid (1.0
mmol), potassium phosphate (2 mmol), Pd2(dba)3 (2.0 mol %), ligand (8.0 mol %), toluene (3 mL), 18 h, reflux.Determined by GC analysis with n class="Chemical">n-dodecane as internal standard.
Under the conditions employed, the four catalyst systems
compared
produced only moderate yields. That is not to say that these reactions
could not be optimized further, but the side-by-side comparison revealed
n class="Chemical">S-Phos (7) to be slightly better thanpan> pan> class="Chemical">triazole-containing
phosphine 29g (55 versus 49% conversion; Table , entry 4 versus entry 1, respectively).
Slightly lower conversions were obtained when 29m or
X-Phos were used as ligands (Table entries 2 and 4 respectively).
One potential
problem with Suzuki–Miyaura catalyzed reactions,
partin class="Chemical">cularly evident when usinclass="Chemical">pan>g less reactive n class="Chemical">aryl chlorides in cross-coupling,
is homocoupling of the n class="Chemical">boronic acid containing reaction partner.[122] In order to test our routine reaction protocols
and our four selected triazole ligands (29g, 29h, 29m, and 29n) for their propensity to
lead to undesired homocoupled product, the following reaction was
probed. Aryl chloride 33 was reacted with 1.5 equiv of
phenyl boronic acid 34. Catalyst loading was 2 mol %
palladium and 4 mol % ligand. The reactions were conducted in toluene
at 90 °C with 3 equiv of potassium phosphate as base; see Scheme . Set up like this,
we can judge a reaction to be successful, i.e., not suffering from
an adventitious homocoupling side reaction leading to product composition,
if conversion of aryl chloride 33 is high (near 100%)
and the amount of formed byproduct 36 is low. Choosing
1.5 equiv of boronic acid gives a chance for the formation of 36, thus evidencing the cross- versus homo-coupling potential
of the ligands in the chosen Suzuki–Miyaura reaction under
the conditions described.
Scheme 8
Percentage Homo-Coupled Product 36 under Described Reaction
Conditions Using 50% Excess Boronic Acid 34
All four of the tested ligands
(Scheme , 29g, n class="Chemical">29h, 29m, anclass="Chemical">pan>d 29n) gave
good conversion to product 35 (liganpan>ds pan> class="Chemical">29h and 29n leading
to complete consumption of stating aryl chloride 33).
In this case the apparently most bulky ligand, 29n, performed
best giving just 6% (mol/mol) homocoupled product (36); the other three ligands gave rise to only slightly elevated amounts
of homocoupled side product (10–12% (mol/mol)). Thus, demonstrating
that under the conditions employed, reaction protocols used throughout
this study do not suffer appreciably from loss of halide-containing
starting materials through unwanted homocoupling.
Lead-like Compounds
While the results disn class="Chemical">cussed thus
far have exemplified the effectiveness of ligands 29g and 29m to catalyze sterically demanding cross-coupling
reactions, the ability to catalyze the cross-coupling of funpan>ctionalities
relevant to medicinal chemistry, to give lead- and drug-like products
remains a critical need in the agrochemical and pharmaceutical sectors.[77] To this end, a range of bis-aromatic products,
containing motifs of the type that are commonly encounpan>tered in medicinal
chemistry,[123−125] were identified, and their synthesis embarked
upon. The products of virtual Suzuki–Miyaura cross-couplings
of 48 n class="Chemical">aromatic halides (iodides, bromides, or chlorides) and 44 aromatic
boronic acids (or esters) from (i) a collection held within the lead
research group; (ii) those curated within the University of Birmingham
Scaffold Diversification Resource;[126] and
(iii) drawn from a boronic acid collection via the GSK Free Building
Blocks resource; were enumerated and analyzed by the online resource
LLAMA (Figure ).[118]
Figure 9
Virtual Suzuki–Miyaura catalysis products generated
and
analyzed in the LLAMA web tool. Left: A log P versus molecular mass, Lipinski and lead-like space indicated.
Right: PMI plot, rod, disc, sphere axis. (See the Supporting Information for data tables.)
Virtual Suzuki–Miyaura catalysis products generated
and
analyzed in the n class="Species">LLAMA web tool. Left: A log P versus molecular mass, Lipinski and lead-like space indicated.
Right: PMI plot, rod, disc, sphere axis. (See the Supporting Information for data tables.)
The open-access web tool n class="Species">LLAMA allows the user to conclass="Chemical">pan>duct
virtual
reactions and analyze the virtual product library for molecular properties
such as molecular weight, A log P, and 3D character. Figure left shows the full virtual library of 1661 compounds created
from the virtual Suzuki–Miyaura cross-coupling of the boronic
acids and aryl halides described above. It shows that while the majority
of these virtual products fall within Lipinkski space (Mw < 500, A log P <
5) only ∼20% lie within lead-like space as defined by Churcher
et al. (200 < Mw < 350, A log P < 3). This fact is illustrated
by the lead-likeness penalty scores of these compounds, which have
an average of 3.47. This penalty scoring system was developed by the
creators of LLAMA to visualize how far away from ideal lead-likeness
a compound may be and incorporates all determined molen class="Chemical">cular properties
into one score. Figure (right side) shows the PMI analysis of the same 1661 virtual compounds
in the library of virtual Suzuki–Miyaura cross-coupled products.
A PMI plot describes the 3D shape of the lowest energy conformation
of a compound on a triangular plot. The upper left corner represents
rod-like compounds, the bottom corner represents disc-like compounds,
and the upper right corner represents spherical compounds. This analysis
shows that the majority of the virtual library resides close to the
rod–disc axis, representing flat compounds.
From the
1661 virtual compounds constructed within the n class="Species">LLAMA tool,
14 possible products that accessed preferable lead-like chemical space
and were selected for testing 29g-mediated Suzuki–Miyaura
cross-couplinclass="Chemical">pan>g reactions, as in Scheme . In this case, the screening conditions involved microwave
heating in a sealed-tube at 100 °C inn class="Chemical">acetonitrile for just 1
h. The possible products include challenging heteroatom-containing
and/or ortho substituents, representing both a set
of possible products displaying favorable characteristics for drug
discovery and a robust challenge for road-testing our best new ligand, 29g.
Scheme 9
Microwave-Heated Synthesis (1 h) of 10 Lead-like Compounds
by Suzuki–Miyaura
Cross-Coupling Reactions Using 2 mol % Palladium and 4 mol % Ligand 29g
Pleasingly, most
of the reactions attempted gave greater than 50%
isolated yield of these challenging cross-coupled products (41a–j); however, n class="Chemical">2-bromo-1-methyl-1H-imidazole 39g and (3,5-dimethylisoxazol-4-yl)boronic
acid 40g failed to deliver detectable amounpan>ts of desired
cross-coupled products in four test scenarios unclass="Chemical">pan>der the conditions
employed.
To determine the utility of the products created in
this analysis,
they were analyzed using the n class="Species">LLAMA web tool to determine their suitability
as lead-like compounpan>ds (Figure ). Figure (left), shows how products 41a–j explore the drug-like space, with all products lying within
the Lipinclass="Chemical">pan>ski space (Mw < 500, A log P < 5) and a significant proportion
lying within the lead-like space (40% of the synthesized compounpan>ds).
This illustrates the potential for this catalyst system to access
both drug- and lead-like chemical space. Figure (right) shows a PMI analysis of products 41a to j, this analysis demonstrates a capability
of this catalyst system to access nonflat Suzuki–Miyaura cross-coupled
products.
Figure 10
Left: Mw vs A log P. Right: PMI analysis of products synthesized in Scheme .
Left: Mw vs A log P. Right: PMI analysis of products synthesized in Scheme .These analyzes show that these compounds can be
described as high-quality
starting points for drug discovery programs. These 10 compounds are
now under evaluation for biological activity across a range of targets.[127]
Palladium Complexes
During the course
of this study
numerous attempts to grow crystals of n class="Chemical">palladium-phosphine complexes
suitable for single crystal X-ray diffractionclass="Chemical">pan> were made. Thus far,
two attempts to generate X-ray quality crystals of n class="Chemical">palladium phosphine
complexes have been achieved, using ligands 29e and 29h with n class="Chemical">palladium(II) chloride.
The combination of 29e and n class="Chemical">trans-Pd(CH3CN)2Cl2 inn class="Chemical">dichloromethane at room temperature led to the
formation of material that was precipitated by addition of pentane
and the residue thus obtained was recrystallized from dichloromethane
and hexane. A representation of the single crystal XRD structure of
the palladium complex (42) thus obtained is depicted
in Figure . A 2:1
ligand/metal square planar trans dichloride palladium(II)
complex 42 was identified. Complex 42 may
offer insight into the structural features of 29 series
complexes as catalysts. The five-membered 1,5-disubstituted1,2,3-triazole
core presents the triazole 2,6-bismethoxy aryl fragment oriented toward
the metal with an aryl-centroid···Pd distance of 3.843(2)
Å, in this solid-state structure. The triazolephosphine ligand
offers a distinct geometric difference to phenylene core ligands (c.f.,
S-Phos and X-Phos, Figure ), being more akin to other five-membered ring core ligands
(c.f., 9 and 10, Figure ), generating a slightly wider metal-binding
pocket while still offering a stabilizing shield about the metal center.
Figure 11
Representation
of the crystal structure of 42, ellipsoids
are drawn at the 50% probability level (Ortep3 for Windows and PovRay).
The structure contains a palladium complex, which is located on an
inversion center and two molecules of dichloromethane per complex.
Only half of the complex and one dichloromethane molecule are unique.
Pd···P bond lengths 2.322(9) Å. 2,6-Bismethoxy
aryl-centroid···Pd 3.843(2) Å. Symmetry code used
to generate equivalent atoms: 1 – x, −y, −z.
Representation
of the crystal structure of 42, ellipsoids
are drawn at the 50% probability level (Ortep3 for Windows and PovRay).
The structure contains a n class="Chemical">palladium complex, which is located on an
inversion center and two moleclass="Chemical">pan> class="Chemical">cules of dichloromethane per complex.
Only half of the complex and one dichloromethane molecule are unique.
Pd···P bond lengths 2.322(9) Å. 2,6-Bismethoxy
aryl-centroid···Pd 3.843(2) Å. Symmetry code used
to generate equivalent atoms: 1 – x, −y, −z.
The combination of n class="Chemical">29h anpan>d class="Chemical">pan> class="Chemical">trans-Pd(CH3CN)2Cl2 in dichloromethane
at room temperature
led to the formation of material that was precepted by addition of
pentane and recrystallized from dichloromethane and hexane. A representation
of the single crystal XRD structure of the palladium complex (43) thus obtained is depicted in Figure . To our surprise, the XRD crystal structure
shows 29h functioning as a 3-N-coordinating
ligand, with two such ligations are present about a trans-dichloridepalladium(II)metal center. Since ligand 29h functions as expected in the aforementioned catalyzed reactions,
it is suggested that a more crystalline and readily formed (kinetic)
complex is formed under the crystallization conditions employed. However,
that this stable complex is formed reminds us of another potentially
important feature of the triazole-core ligands, namely, a rear-side
ancillary coordination point. It is conceivable that the ancillary
nitrogen may offer some advantages in some catalyzed reaction, such
as aggregation suppression for example. Suffice it to say, the sterically
encumbered phosphine face of ligand 29h, as evidenced
by the isolation of 43, bodes well for understanding
differing catalysis modes of action on steric rationales as well as
opportunities for divergence from phenylene core ligands.
Figure 12
Representation
of the crystal structure of 43, ellipsoids
are drawn at the 50% probability level (Ortep3 for Windows and PovRay).
The structure contains two molecules of dichloromethane per palladium
complex (omitted for clarity).
Representation
of the crystal structure of 43, ellipsoids
are drawn at the 50% probability level (Ortep3 for Windows and PovRay).
The structure contains two molecules of dichloromethane per palladium
complex (omitted for clarity).
Describing Phosphines
There is a growing body of literature
disn class="Chemical">cussinclass="Chemical">pan>g the importance of various parameters including steric effects[116] of bulky n class="Chemical">phosphines[83,84,128] relating to suitability and efficacy in
catalysis (primarily as ligands for metals in metal mediated catalysis).[108,115,129] While the Tolman cone angle
has been an effective descriptor of ligand bulkiness for many years,[111] it has been complimented more recently by Nolan’s
percentage buried volume parameter (%Vbur).[112] The %Vbur of ligands can be calculated using the SambVca (2.0) free web tool
from Cavallo and co-workers;[113,130] so we set about determining
some steric parameters of our n class="Chemical">phosphines using this tool.
First,
the crystal structure of the free ligand 29g was investigated,
as follows: The n class="Chemical">PDB file correspondinclass="Chemical">pan>g to the XRD crystal structure
of ligand 29g was edited in Spartan’16 Parallel
Suite (Wave funpan>ction Inc.) by changing the valence of n class="Chemical">phosphorus to
four and adding a palladium atom with a standard bond length of 2.280
Å to generate a metal-coordinated model. The PBD file thus generated
was not minimized or edited further, i.e., the atoms of the ligand
remained in their crystallographically determined free-ligand position,
and uploaded for analysis to the SambVca2.0 web tool for analysis.
The added palladium atom was set as the center and then deleted; a
summary of this analysis is shown in Figure . A steric map generated from this analysis
is shown and a 47.0%Vbur was calculated,
using otherwise default SambVca settings.
Figure 13
Steric map of phosphine-palladium
complex: Derived from crystal
structure of free ligand 29g with a P–Pd distance
of 2.280 Å applied, resulting in a 47.0%Vbur.
Steric map of phosphine-palladium
complex: Derived from crystal
structure of free ligand 29g with a P–Pd distance
of 2.280 Å applied, resulting in a 47.0%Vbur.Following this analysis,
the only n class="Chemical">palladium phosphine complex thus
far successfully analyzed by single crystal X-ray diffractionclass="Chemical">pan> studies
(42) was investigated in a similar manner. Compounpan>d 42 is a square planar n class="Chemical">palladium(II) complex of ligand 29e and as such does not necessarily represent the catalytically
relevant species, the determined %Vbur of 36.2% (Figure i) may not be the best comparator across a number of ligand structures,
so three other forms of the complex were generated and compared. In
order to find an appropriate comparison to fairly evaluate relative
steric parameters of ligands 29e and 29g. The ligand portion of the crystal structure of 42 was
edited in Spartan’16 Parallel Suite, the Pd–P bond length
adjusted to 2.280 Å, and the SambVca-determined %Vbur of 36.9% (Figure ii) was essentially the same as that for the slightly
longer, crystallographically determined, Pd–P bond length in
the earlier analysis. Since a model to allow for comparison of ligands
where crystallographically determined data is not available was ultimately
sought two further analyses were conducted.
Figure 14
(i) Chemical structure
of part of the crystallographically determined
complex 42, with a 2.323(2) Å Pd–P distance
(from the crystal structure) used in the calculation of buried volume,
36.2%Vbur. (ii) Chemical structure of
the bond-length-modified, crystal-structure-informed palladium complex
of ligand 29e in complex 42 (2.280 Å
Pd–P distance was used in the buried volume calculation), 36.9%Vbur.
(i) Chemical structure
of part of the crystallographically determined
complex 42, with a 2.323(2) Å n class="Chemical">Pd–P distanclass="Chemical">pan>ce
(from the crystal structure) used in the calculation of buried volume,
36.2%Vbur. (ii) Chemical structure of
the bond-length-modified, crystal-structure-informed palladium complex
of ligand 29e in complex 42 (2.280 Å
n class="Chemical">Pd–P distance was used in the buried volume calculation), 36.9%Vbur.
Sigman and co-workers have previously used n class="Chemical">phosphine oxides
as
computational models for structural minclass="Chemical">pan>imization proxies of ligand–n class="Chemical">metal
complexes,[115] so it was reasoned that computational
minimization of in silico generated phosphine oxides
may be a sensible starting point to allow for comparison among some
of the ligands in this report. Two approaches were compared for optimizing
and computationally determining the %Vbur of ligand 29e (summarized in Figure ) using Spartan’16 Parallel Suite.
In one case a phosphine oxide structure with no geometric restrictions
was used (Figure , right); in the second case, the dihedral geometries of atoms 1–6
were restricted presenting the ligand’s ancillary aryl group
directly and orthogonally aligned to the coordination vector of the
n class="Chemical">phosphine (the P–O bond in this minimization), as shown in
(Figure , left).
Through comparison of the geometry unrestricted and geometry restricted
protocols, with the structurally determined geometry (as shown in Figure ) a protocol for
analysis across the ligands of this report might be reasonably determined.
In both cases the same minimization and optimization cascade was adopted;
the structures used for computationally determined %Vbur calculations were obtained as follows. First, molecular
mechanics (MMFF) conformer distribution (1000 max conformers) and
subsequent molecular mechanics equilibrium geometries were ranked,
and the 20 lowest energy conformers were retained for DFT investigation.
The 20 retained conformers’ equilibrium geometries were used
as starting points for ground-state, gas phase equilibrium geometry
determination (B3LYP 6-31G*). The lowest energy structure thus obtained
was then edited to replace the oxygen (of the phosphine oxide) with
a palladium atom and a standard P–Pd bond length of 2.280 Å
was applied (Figure , center).
Figure 15
Protocol for obtaining structures for comparison of steric
parameters
via an unrestricted (right) and a dihedral angles restricted model
(left). In both cases, a phosphine oxide model was minimized using
Spartan’16 and the P=O later replaced with a P–Pd
bond of 2.280 Å. The structures thus obtained were then analyzed
by the SambVca2.0 free web tool to determine the %Vbur and to create a steric map.
Protocol for obtaining structures for comparison of steric
parameters
via an unrestricted (right) and a dihedral angles restricted model
(left). In both cases, a n class="Chemical">phosphine oxide model was minimized using
Spartanpan>’16 anpan>d the P=O later replaced with a P–pan> class="Chemical">Pd
bond of 2.280 Å. The structures thus obtained were then analyzed
by the SambVca2.0 free web tool to determine the %Vbur and to create a steric map.
The procedure outlined in Figure was first applied to ligand 29e, and
buried volumes of 44.2 and 47.3% were determined for the restricted
and unrestricted geometries, respectively (Figure ) and compared to the buried volume determined
crystallographically. While this protocol gives slightly higher values,
the restricted geometry model is more closely aligned to the XRD-derived
model than the unrestricted one. With this information alone, it is
diffin class="Chemical">cult to ascertain if anpan>y one model provides a superior approach
for assessing liganpan>ds where solid state data is not available. Furthermore,
it should be remembered that the 42-derived %Vbur numbers are from solid-state structures
anpan>d might not offer the best cross-liganpan>d comparison. As such, the
most active liganpan>ds of our study were examined using both protocols
shown below (Figure ).
Figure 16
(i) Chemical structure and computationally determined steric map
and buried volume, structures derived in silico and calculated with
restricted dihedral angles as describe in Figure (left), 44.2%Vbur. (ii) Chemical structure and computationally determined steric map
and buried volume of structures derived in silico and calculated without
any dihedral restrictions as describe in Figure (right), 47.3%Vbur.
Figure 17
SambVca2.0 derived %Vbur and steric
maps for the: (a) left side: (i) 29g; (ii) 29h; (iii) 29n; (iv) 29m; (v) 7 S-Phos; (vi) 8 X-Phos, derived palladium complexes
of structures derived in silico and calculated with restricted dihedral
angles as describe in Figure (left); (b) right side: (i) 29g; (ii) 29h; (iii) 29n; (iv) 29m; (v) 7 S-Phos; (vi) 8 X-Phos, derived palladium complexes
of structures derived in silico and calculated without any dihedral
restrictions as describe in Figure (right).
(i) Chemical structure and computationally determined steric map
and buried volume, structures derived in silico and caln class="Chemical">culated with
restricted dihedral angles as describe in Figure (left), 44.2%Vbur. (ii) Chemical structure and computationally determined steric map
anclass="Chemical">pan>d buried volume of structures derived in silico and calculated without
any dihedral restrictions as describe in Figure (right), 47.3%Vbur.
SambVca2.0 derived %Vbur and steric
maps for the: (a) left side: (i) 29g; (ii) n class="Chemical">29h; (iii) 29n; (iv) 29m; (v) 7 n class="Chemical">S-Phos; (vi) 8 n class="Chemical">X-Phos, derived palladium complexes
of structures derived in silico and calculated with restricted dihedral
angles as describe in Figure (left); (b) right side: (i) 29g; (ii) 29h; (iii) 29n; (iv) 29m; (v) 7 S-Phos; (vi) 8 X-Phos, derived palladium complexes
of structures derived in silico and calculated without any dihedral
restrictions as describe in Figure (right).
Computed structures, percentage buried volumes, and steric
maps
of the six ligands of Figure , following the protocol outlined in Figure , are shown in Figure . The left side shows the information relating
to restricted dihedral angle restricted complexes and the right side
shows the structures obtained without imposing any geometrical restrictions
(other than the 2.280 Å P–n class="Chemical">Pd distances imposed throughout).
It is interesting and confounpan>ded our expectations that the unpan>restricted
geometry-minimized structures give a higher buried volume than the
restricted dihedral angle structures (Figure column a versus column b). Among the two
data treatments, there is broad agreement with and relative similarity
to each other. While the geometry-unclass="Chemical">pan>restricted structures of n class="Chemical">triazole
ligands bear a striking resemblance (by n class="Disease">rudimentary visual inspection)
to any XRD-derived structures, the buried volume determinations of
previously reported S- and X-Phos (by the P=O minimization
proxy discussed earlier) complexes more closely match the geometry-restricted
models we employed (Table , entries 6 and 7).
Table 6
Calculated (Spartan
v16 and SambVca2.0)
Steric Properties of 29e, 29g, 29h, 29m, 29n, S-Phos 7, and
X-Phos 8
entry
ligand
Vbur calcd [%] (restricted)
Vbur calcd [%] (unrestricted)
Ar centroid–Pd distance [Å] (restricted)
Ar centroid–Pd distance [Å] (unrestricted)
1
29e
44.2 (36.2)a
47.3
3.363 (3.843(2))a
4.176
2
29g
45.5 (46.9)b
49.6
3.403 (3.457)b
3.745
3
29h
47.8
50.1
3.558
3.580
4
29m
47.1
48.0
3.639
3.901
5
29n
47.5
53.3
3.647
3.794
6
7 (S-Phos)
50.8 (49.7)c
55.4
3.127
3.253
7
8 (X-Phos)
53.3 (53.1)c
53.2
3.309
3.481
Measured from the
unadjusted XRD
structure of 42.
Using free ligand 29g XRD and applying Pd–P bond
length 2.280 Å.
Refers
to ligand/metal 1:1 complex
P-M distance 2.8 A (M = Au).[112,131]
Measured from the
unadjusted XRD
structure of 42.Using free ligand 29g XRD and applying n class="Chemical">Pd–P bond
length 2.280 Å.
Refers
to ligand/n class="Chemical">metal 1:1 complex
P-M distance 2.8 A (M = Au).[112,131]
From the steric maps of Figure , it was noted
that the space between the dark red
bulky zone (to the left of the steric maps arising from the ancillary
aryl groups of all ligands) protrudes in a manner to give more space
between the central axis and red zone. Since space near the coordination
site may be crucial in permitting reaction with bulky substrates,
we also report the distance between the ancillary aryl group’s
centroid and the computed n class="Chemical">metal center (Table ) for both the geometry-restricted and unpan>restricted
ligands of Figure are also listed in Table (and contrasted against the data determined for ligand 29e). In Entry 2 the data arising from analysis of the previously
detailed ligand crystal-structure-determined structures of 29g–complexes are given in parentheses. Between the restricted
and unpan>restricted models of ligand–complex analysis, the cenpan>troid
distance correlates more closely with the dihedral angle restricted
model of a 29g–n class="Chemical">Pd complex.
It is notable
that the P=O ligand minimization protocol
adopted gives good structural parameter agreement between those determined
for n class="Chemical">palladium complexes of ligand 29g both derived computationally
and from a modified XRD structure of free ligand 29g (Table , entry 1). By analyzinclass="Chemical">pan>g 29a–n side-by-side using the same restricted
dihedral angle minimization protocol (up to 10 conformers analyzed
by DFT methods abbreviating the earlier minimization cascade for newly
analyzed ligands) and plotting the computed %Vbur values against conversion to 27 (data from Table ), as shown in Figure , it can be seen
that a strong bulkiness versus conversion trend exists. Essentially
any %Vbur value above 46% leads to quantitative
conversion to products, unpan>der the prescribed conditions.
Figure 18
Conversion
to 27 (see Table ) versus computed burried volume (%) as determined
by the protocol discussed in Figure .
Conversion
to 27 (see Table ) versus computed burried volume (%) as determined
by the protocol disn class="Chemical">cussed in Figure .
Further analysis of
the data obtained for conversion in reactions
catalyzed by 29 ligands is only against smaller data
sets, and significant correlations of variances in conversions do
not lead to any meaningfully comparable correlations. However, further
study of cross-couplings of very bulky n class="Chemical">aryl chlorides may be warranted
in the future since an intriguing balance between bulk and centroid
distance is suggested (see the Supporting Information), but across only four data points surveyed to date, it may be too
early to draw conclusion yet.[132] A pre-peer-reviewed
preprint of this article was deposited and may be viewed elsewhere.[133]
Conclusions
Two series of n class="Chemical">1,2,3-triazole-containing
n class="Chemical">5-phosphino ligands were
synthesized and tested as ligands in palladium catalyzed cross-coupling
Suzuki–Miyaura reactions of bulky and heteroatom-containing
substrates. The structural parameters of the 4-H triazole series (29) were determined by a restricted dihedral angle, phosphine
oxide surrogate model, and a strong dependence upon bulkiness and
catalytic activity were noted. Furthermore, a link to the space between
the n class="Chemical">metal and the ancillary aryl group in the computed complexes was
noted, suggesting bulkiness of ligand and space around the metal may
both be implicated in delineating trends in cross-coupling of the
most bulky and challenging substrates. Notably, triazole-nitrogens’
may not be completely innocent in the coordination environment created
by these ligands, with a nitrogen coordination complex being characterized
by XRD in one case. The ligands synthesized were benchmarked against
commercially available ligands (X- and S-Phos) and in some cases the
best triazole ligands match or outperformed them under the employed
conditions, in like-for-like tests in triplicate. The phosphine ligands
reported and characterized in this report represent easy to modify
catalytic scaffolds that could be use in future library generation
efforts and we are looking forward to facilitating access to these
compounds and allowing others to include this type of ligand in their
own catalyst screening campaigns.
Authors: Gian Cesare Tron; Tracey Pirali; Richard A Billington; Pier Luigi Canonico; Giovanni Sorba; Armando A Genazzani Journal: Med Res Rev Date: 2008-03 Impact factor: 12.944
Authors: Rhiannon K Iha; Karen L Wooley; Andreas M Nyström; Daniel J Burke; Matthew J Kade; Craig J Hawker Journal: Chem Rev Date: 2009-11 Impact factor: 60.622
Authors: William D G Brittain; Brette M Chapin; Wenlei Zhai; Vincent M Lynch; Benjamin R Buckley; Eric V Anslyn; John S Fossey Journal: Org Biomol Chem Date: 2016-11-22 Impact factor: 3.876
Authors: Stephan Hohloch; David Schweinfurth; Michael G Sommer; Fritz Weisser; Naina Deibel; Fabian Ehret; Biprajit Sarkar Journal: Dalton Trans Date: 2014-03-21 Impact factor: 4.390
Authors: Yiming Zhao; Matthew G Wakeling; Fernanda Meloni; Tze Jing Sum; Huy van Nguyen; Benjamin R Buckley; Paul W Davies; John S Fossey Journal: European J Org Chem Date: 2019-07-30
Authors: Xingjian Li; Daniel T Payne; Badarinath Ampolu; Nicholas Bland; Jane T Brown; Mark J Dutton; Catherine A Fitton; Abigail Gulliver; Lee Hale; Daniel Hamza; Geraint Jones; Rebecca Lane; Andrew G Leach; Louise Male; Elena G Merisor; Michael J Morton; Alex S Quy; Ruth Roberts; Rosanna Scarll; Timothy Schulz-Utermoehl; Tatjana Stankovic; Brett Stevenson; John S Fossey; Angelo Agathanggelou Journal: Medchemcomm Date: 2019-08-01 Impact factor: 3.597