The challenge of palladium-catalyzed aromatic azidocarbonylation: From mechanistic and catalyst deactivation studies to a highly efficient process

Article

pubs.acs.org/Organometallics

The Challenge of Palladium-Catalyzed Aromatic Azidocarbonylation:

From Mechanistic and Catalyst Deactivation Studies to a Highly

Eﬃcient Process

†

‡

†

́

Fedor M. Miloserdov, Claire L. McMullin, Marta Martınez Belmonte, Jordi Benet-Buchholz,

Vladimir I. Bakhmutov,^§Stuart A. Macgregor,*^,‡and Vladimir V. Grushin*^,†

^†Institute of Chemical Research of Catalonia (ICIQ), Tarragona 43007, Spain

^‡Institute of Chemical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, U.K.

^§Department of Chemistry, Texas A&M University, College Station, Texas 77842, United States

S

* Supporting Information

ABSTRACT: Azidocarbonylation of iodoarenes with CO and

NaN₃, a novel Heck-type carbonylation reaction, readily occurs

in an organic solvent−H₂O biphasic system to furnish aroyl

azides at room temperature and 1 atm. The reaction is

catalyzed by Xantphos-Pd and exhibits high functional group

tolerance. The catalyst deactivation product, [(Xantphos)-

PdI₂], can be reduced in situ with PMHS to Pd(0) to regain

catalytic activity. In this way, the catalyst loading has been

lowered to 0.2% without any losses in selectivity at nearly

100% conversion to synthesize a series of aroyl azides in 80−90% isolated yield on a gram scale. Alternatively, the ArCON₃

product can be used without isolation for further transformations in situ, e.g., to isocyanates, ureas, benzamides, and

iminophosphoranes. A detailed experimental and computational study has identiﬁed two main reaction pathways for the reaction.

For both routes, Ar−I oxidative addition to Pd(0) is the rate-determining step. In the presence of CO in excess, the Ar−I bond is

activated by the less electron-rich Pd center of a mixed carbonyl phosphine complex. Under CO-deﬁcient conditions, a slightly

lower energy barrier pathway is followed that involves Ar−I oxidative addition to a more reactive carbonyl-free (Xantphos)Pd⁰

species. Mass transfer in the triphasic liquid−liquid−gas system employed for the reaction plays an important role in the

competition between these two reaction channels, uniformly leading to a common aroyl azido intermediate that undergoes

exceedingly facile ArCO−N₃reductive elimination. Safety aspects of the method have been investigated.

and oxazoles.⁴Conventional routes to aroyl azides include

INTRODUCTION

■

In 1974, Schoenberg and Heck¹reported their groundbreaking

discovery of Pd-catalyzed carbonylation of haloarenes (eq 1;

e.g., X = I, Br and Nu = OR, NR₂). Since then, this reaction has

become one of the most powerful tools for the synthesis of

aromatic carbonyl compounds in both laboratory and industrial

settings.²A broad variety of nucleophiles have been successfully

employed in this transformation to furnish a diversity of

products.

diazotization of hydrazides and azidation of acid chlorides,

mixed anhydrides, and N-acyl benzotriazoles with NaN₃.^3,5

These synthetic methods employ highly reactive and hazardous

chemicals that limit their scope and functional group

compatibility. Since methodologically new, high functional

group tolerance routes to aroyl azides are certainly in demand,

the following question arises: why is it that Pd-catalyzed

aromatic azidocarbonylation (eq 1, where Nu⁻= N₃) has not

−

been developed?

Upon superﬁcial consideration, there should be no

fundamental diﬀerence between azide and other nucleophiles

that have been successfully used in the carbonylation reactions

of haloarenes (eq 1). In-depth analysis of the literature data on

aroyl azides and mechanisms of palladium catalysis suggests,

however, that the development of Pd-catalyzed aromatic

azidocarbonylation should be extremely challenging, if not

impossible. In particular:

Considering the vast research eﬀort made by numerous

groups working in the area,²it might be surprising that the Pd-

catalyzed carbonylation methodology has not been applied to

the synthesis of azidocarbonyl aromatic compounds ArCON₃

−

(Nu⁻= N₃in eq 1). Aroyl azides are valuable intermediates

and building blocks in the preparation of various useful

Received: November 20, 2013

compounds³such as isocyanates, amides, iminophosphoranes,

A

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Scheme 1

(1) A key step in the catalytic loop governing the

carbonylation reaction (eq 1) is migratory insertion of CO

into the Pd−C bond of [(R₃P)₂Pd(Ar)(X)], the intermediate

that is produced in the preceding step of ArX substrate

oxidative addition to R₃P-stabilized Pd(0).⁶For X = I, Br, Cl,

this reaction usually occurs easily and cleanly at 20−70 °C and

atmospheric pressure (Scheme 1). In sharp contrast, complexes

of the type trans-[(R₃P)₂Pd(Ph)N₃] (R = Me, Et) do not

undergo CO insertion into the Pd−Ph bond but are rather

converted to trans-[(R₃P)₂Pd(Ph)NCO] with concomitant

release of N₂under similar conditions (room temperature, 1

atm of CO),⁷as shown in Scheme 1.

deactivation, which have led to the development of a practical

and robust azidocarbonylation process on a gram scale with a

Pd catalyst loading of only 0.2 mol %. We have also performed

a number of tests to prove that the process is safe to run.

Finally, our combined experimental and computational studies

have uncovered novel and unique mechanistic features that are

characteristic not only of the current process but also of other

Heck-type¹carbonylations. It is apt to note that, surprisingly,

there have been no reports of detailed studies of the mechanism

of palladium catalysis of carbonylative reactions of aryl halides

in general and Ar−X oxidative addition to Pd(0) in the

presence of CO in particular.

(2) Being thermally stable, carboxylic acid derivatives (salts,

esters, amides, etc.) formed in the carbonylation reactions (eq

1) easily survive the elevated temperatures that are needed for

the process to occur. In sharp contrast, aroyl azides are

thermally unstable compounds that rapidly undergo the Curtius

rearrangement (Scheme 2) at such temperatures (80 °C and

above).

EXPERIMENTAL STUDIES

■

Early Studies. Originally, we intended to develop the Pd-

catalyzed synthesis of aryl azides from the corresponding

haloarenes and an inorganic azide source such as NaN₃

(Scheme 4). Previous attempts to perform this transformation

have been unsuccessful, although Pd-catalyzed azidation of

vinylic bromides to give 1H-1,2,3-triazoles has been reported.¹⁰

As both oxidative addition of ArX to Pd(0) and halide/azide

metathesis reactions of Pd(II) complexes^7,11are well-known,

we reasoned that it is the Ar−N₃reductive elimination from

Pd(II) that must be the problematic step of the process

(Scheme 4).

Scheme 2

Scheme 4

(3) The anticipated product, ArCON₃, is not only thermally

unstable but also highly reactive toward tertiary phosphines that

are used as stabilizing ligands for Pd catalysts employed in

carbonylation and other coupling reactions. The facile

Staudinger reaction⁸of the ArCON₃produced with the

reversibly dissociated PR₃ligand would lead to immediate

catalyst deactivation from the quick and irreversible formation

of the corresponding phosphinimine ArC(O)NPR₃(Scheme

3).

To study Ar−N₃reductive elimination from Pd(II), the

product-forming step in the proposed catalytic loop, we

prepared [(Ph₃P)₂Pd(Ph)(N₃)] from [(Ph₃P)₂Pd(Ph)(Br)]

and NaN₃. This complex was found to be exclusively trans

both in solution (¹H and ³¹P NMR) and in the solid state (X-

ray diﬀraction). Interestingly, [(Ph₃P)₂Pd(Ph)(N₃)] was found

to crystallize in two polymorphic forms displaying either a

staggered or an eclipsed conformation along the P−Pd−P axis

(Figures 1 and 2).

Scheme 3

The new azido complex trans-[(Ph₃P)₂Pd(Ph)(N₃)] ap-

peared to be unexpectedly thermally stable, exhibiting no sign

of decomposition at 80−100 °C in benzene or toluene for 15−

24 h. At 135 °C in xylenes, however, [(Ph₃P)₂Pd(Ph)(N₃)]

decomposed to give mostly Ph₂and PhNPPh₃(GC-MS).

The iminophosphorane was evidently produced by the

Staudinger reaction⁸of the transient PhN₃with the PPh₃

ligand. Attempts to perform this reaction catalytically, i.e. in

the presence of excess PPh₃, were unsuccessful. Importantly,

The three reactivity patterns described above and shown in

Schemes 1−3 cast serious doubts on the very possibility of Pd-

catalyzed aromatic azidocarbonylation. Nonetheless, herein we

report that this transformation is not only possible, but can be

highly eﬃcient with a carefully designed and thoroughly studied

catalytic system. In a preliminary communication,⁹we

described the formation of aroyl azides from aryl iodides,

CO, and NaN₃in the presence of a Pd catalyst (2 mol %).

Herein we report our most recent investigations into catalyst

B

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

pure. It was found (¹⁹F NMR) that nearly quantitative

conversions of 4-FC₆H₄I could be obtained by running the

reaction in the presence of water, especially under biphasic

conditions. In THF−H₂O, the reaction stalled after quickly

reaching 80−90% conversion, likely because of catalyst

deactivation with the iodide^18,19released in the transformation

(eq 2). Performing the reaction in a water−organic solvent

biphasic system eﬀectively eliminated the inhibiting eﬀect of

iodide that accumulated in the aqueous layer, whereas the

active Pd catalyst remained in the water-immiscible organic

phase. In H₂O−benzene or H₂O−toluene, the reaction

aﬀorded 4-FC₆H₄CON₃in 97% yield at >99% conversion

with 2% Pd. However, benzene is toxic and the boiling point of

toluene (110 °C) prevents isolation of the volatile and

thermally sensitive benzoyl azide product. As a result of a

series of experiments, it was found that the aromatic organic

phase can be successfully substituted with a 2:1 v/v nontoxic,

low-boiling mixture of THF and hexane. Although high

conversions (>90%) and yields of up to 88% could be obtained

under such conditions with only 1% Pd, we settled on 2 mol %

of the catalyst in order to reach full conversion that is critical for

isolation of the product in pure form. A summary of the

optimization studies described above can be found in Table 2 of

the Supporting Information to the preliminary communica-

tion.⁹

Figure 1. ORTEP drawings of the staggered (left) and eclipsed (right)

conformers of trans-[(Ph₃P)₂Pd(Ph)(N₃)] with all H atoms omitted

and thermal ellipsoids drawn at the 50% probability level.

Azidocarbonylation of Substituted Aryl Iodides.

Having optimized the reaction (eq 2) under biphasic

conditions, we performed the azidocarbonylation on a series

of aryl iodides (Table 1).⁹The reactions smoothly proceeded at

23−50 °C and atmospheric pressure of CO. Quantitative

conversions were achieved for all of the substrates 1a−r,

allowing isolation of the pure aroyl azide products 2a−r in high

yield. Various functional groups were tolerated and the reaction

proceeded smoothly and selectively with substrates bearing

electron-withdrawing (2e−l,p) and electron-donating (2b−

d,n) substituents on the ring. Likewise, heteroaryl iodides, 2-

iodothiophene (1q) and 3-iodopyridine (1r), were cleanly

converted to 2q and 2r in 87% and 75% isolated yields,

respectively. The reaction of 2-iodopyridine, however, gave rise

to a mixture of products (2-PyNCO, 2-PyCONH₂, and 2-

PyNH₂) under similar conditions, evidently because of the low

Figure 2. ORTEP views of the staggered (left) and eclipsed (right)

conformers of trans-[(Ph₃P)₂Pd(Ph)(N₃)] along the P−Pd−P axis

with all H atoms omitted and thermal ellipsoids drawn at the 50%

probability level.

during these studies we found that bubbling CO through a

solution of [(Ph₃P)₂Pd(Ph)(N₃)] containing PhI and PPh₃in

benzene at 50 °C triggered a clean transformation to

12

[(Ph₃P)₂Pd(COPh)I]⁶and PhC(O)NPPh₃(³¹P NMR).

The formation of the latter indicated that PhCON₃was formed

as an intermediate that was scavenged with PPh₃in the

Staudinger reaction. These observations prompted our search

for a Pd catalyst for the synthesis of aroyl azides directly from

−

ArI, CO, and N₃.

Catalyst Screening. A large series of scouting experiments

resulted in the identiﬁcation of Xantphos-stabilized Pd

complexes as active catalysts for the reaction of iodobenzene

with CO and NaN₃to give benzoyl azide (eq 2). Of numerous

ligands (Ph₃P, o-Tol₃P, Cy₃P, t-Bu₃P, dppe, dppp, dppb, dppf,

rac-BINAP, Xantphos, i-Pr-Xantphos, t-Bu-Xantphos, and

DPEphos) screened in combination with Pd(OAc)₂or

20

thermal stability of 2-PyCON₃and possibly catalyst

modiﬁcation via the previously documented²¹dimerization of

2-pyridyl Pd complexes via the N atoms of the Py ring.

With ortho-substituted aryl iodide substrates, the reaction did

not produce aroyl azides but rather gave amides, amines, and

diarylureas.⁹A mixture of all three was formed in the reaction

of 2-iodotoluene. 2-Iodoanisole and 1-iodonaphthalene reacted

more selectively to give predominantly ureas. In this way, N,N′-

bis(1-naphthyl)urea (3a; 76% yield) and N,N′-bis(2-

methoxyphenyl)urea (3b; 74% yield) were isolated from the

reactions of 1-iodonaphthalene and 2-iodoanisole, respectively.

Considering the fact that ortho-substituted ArCON₃are 50−

200 times less stable toward the Curtius rearrangement than

their meta and para isomers,^20,22the diﬀerent reaction outcome

for iodoarenes bearing an ortho substituent is hardly surprising.

The steric bulk of a group in the ortho position weakens the

conjugation between the aromatic ring and the CON₃function,

thereby destabilizing the latter and lowering the barrier to the

Curtius rearrangement. As a result, ortho-substituted aroyl

azides formed in the reaction quickly rearrange to the

corresponding isocyanates. Conversion of the latter to an

13

Pd₂(dba)₅precursors, only Xantphos^14,15gave rise to an

active catalyst.⁹With the other ligands, the yield of PhCON₃

from PhI never exceeded 6% under a variety of conditions.¹⁶

The highest activity was observed for a 1:1 Pd(0) to Xantphos

molar ratio, in accord with the literature data.¹⁷

The Need for Full Conversion: Performing the

Reaction under Biphasic Conditions. In most instances,

the aroyl azide product and the iodoarene starting material

exhibited close R_fparameters and therefore could not be

eﬃciently separated by column chromatography. This obser-

vation dictated the need for driving the azidocarbonylation

reaction to full conversion, so that the product could be isolated

C

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

a

Table 1. Pd-Catalyzed Azidocarbonylation of Iodoarenes

a

13

Reaction conditions: ArI (1 mmol), NaN₃(1.2 mmol), Pd₂(dba)₅(0.01 mmol; 2 mol % Pd), Xantphos (2 mol %) in THF (2 mL), hexane (1

b

c

d

mL), and water (3 mL). 50 °C (oil bath). Isolated product contained ca. 2% of PhCON₃as a result of P−Ar/Pd−Ar′ exchange. Isolated product

contained dba. With 1 mmol of 1,4-diiodobenzene (1m) and double amounts of all reagents and solvents.

e

aniline under the reaction conditions, followed by its addition

to the as yet unreacted isocyanate, gives the urea (Scheme 5).

In Situ Modiﬁcation of Catalytically Produced Aroyl

Azides. As mentioned above, aroyl azides are reactive

compounds that can be used in a variety of useful trans-

formations. We were able to demonstrate that some of these

transformations could be conveniently carried out in situ,

without isolation of the aroyl azide product formed in the

catalytic process (Scheme 6).⁹The substrate of choice was 4-

Scheme 6

Scheme 5

carbamates, anilines, and ureas. Examples of the synthesis of

aromatic isocyanates from the corresponding aryl halides are

extremely rare. Tkatchenko et al.²⁵have patented Ni-catalyzed

reaction of haloarenes with NaOCN to produce aryl

isocyanates and their derivatives in up to 50% yield. Most

recently, Buchwald’s group²⁶demonstrated direct Pd-catalyzed

transformation of aryl chlorides and triﬂates to aryl isocyanates

that were converted in situ to unsymmetrical ureas^26aand

carbamates.^26b

Catalyst Deactivation. The identiﬁcation and in-depth

understanding of the side transformations resulting in catalyst

deactivation can be a powerful tool for the design and

development of eﬃcient catalytic processes.^13b,27We therefore

studied the loss of catalytic activity in the azidocarbonylation

reaction.

As mentioned above, attempts to lower the catalyst loading

from 2% to 1% resulted in incomplete conversion. To gain

insight into the loss of catalytic activity, we set up a reaction

with a larger quantity of iodobenzene (10 mmol of PhI in 0.5

mL of THF) and a lower amount of the Xantphos-Pd₂(dba)₅

catalyst (0.2 mol % Pd). After 2 h under such conditions, the

organic phase turned deep purple and the catalytic process

stopped at ca. 65% conversion. The ³¹P NMR spectrum of the

organic phase displayed a broad resonance at 6.4 ppm that

could not be assigned to any of the identiﬁed intermediates

ﬂuoroiodobenzene (1j), so that the transformations could be

conveniently monitored by ¹⁹F NMR. Conducting the

azidocarbonylation of 1j in the presence of polymethylhy-

drosiloxane (PMHS; 2 equiv) at 50 °C gave rise to 4-

ﬂuorobenzamide (4) in nearly quantitative yield. It is worth

noting that catalytic carbonylation of aryl halides to primary

benzamides represents a considerable challenge.²³In another

experiment, the entire reaction mixture containing freshly

produced 2j was treated with PPh₃(1.1 equiv) to prompt the

Staudinger reaction that furnished iminophosphorane 5 in 85%

overall yield. Furthermore, performing the azidocarbonylation

in toluene−water provided a convenient means to carry out the

Curtius rearrangement by simply separating and heating the

toluene phase containing the ArCON₃product. In this way, 4-

ﬂuorophenyl isocyanate (6) was obtained in 86% yield

(Scheme 6), as calculated from the amount of 1j used. Aryl

isocyanates²⁴are widely used in the production of a broad

variety of polymers, materials, and valuable chemicals, including

D

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Figure 3. ORTEP drawing of cis-[(Xantphos)PdI₂]·CH₂Cl₂(cis-7·CH₂Cl₂, left) and trans-[(Xantphos)PdI₂] (trans-7, right) with the CH₂Cl₂

molecule and all H atoms omitted and thermal ellipsoids drawn at the 50% probability level.

involved in the catalytic loop (see below). The dark purple

color suggested that the signal at 6.4 ppm might be from

[(Xantphos)PdI₂] (7). This assumption was probed by an

independent synthesis of 7 from [(Xantphos)PdCl₂] (8) and

NaI. Indeed, both the color and ³¹P NMR parameters displayed

by the thus-prepared 7 (97% yield) appeared identical with

those displayed by the catalyst deactivation product.

Furthermore, spiking the reaction mixture after catalyst

deactivation with an authentic sample of 7 resulted in growth

in intensity of the ³¹P NMR signal at 6.4 ppm. We therefore

concluded that the loss of catalytic activity should be attributed

to gradual conversion of Pd species in the system to 7. As was

established by an independent run, 7 was indeed poorly

catalytically active in the azidocarbonylation reaction.²⁸As a

new compound, 7 was fully characterized and found to exist as

a mixture of cis and trans isomers both in solution and in the

solid state (Figure 3). In solution, the cis to trans ratio was both

solvent and temperature dependent. While only trans-7 could

be observed in benzene at room temperature, both the cis and

trans isomers of 7 were detected in CH₂Cl₂(³¹P NMR). This

trend is similar to the reported^29,30behavior of [(Xantphos)-

Pd(Ph)(CF₃)]. Furthermore, the cis to trans ratios for 7 in

CH₂Cl₂, as determined by ³¹P NMR, were 1:3.5 and 1:1.2 at 25

and −30 °C, respectively. Such solvent and temperature

dependence of the cis to trans ratio for square-planar d⁸

complexes is well-known.³¹

Table 2. Pd-Catalyzed Azidocarbonylation of Iodoarenes (10

mmol) in the Presence of 8 (0.2%) and PMHS

a

All yields are isolated yields of pure products. Reaction conditions:

ArI (10 mmol), NaN₃(12 mmol), [(Xantphos)PdCl₂·CH₂Cl₂] (8;

0.02 mmol; 0.2 mol % Pd), K₂CO₃(0.2 mmol; 2 mol %), in THF (0.5

b

c

mL) and water (2 mL) at 23 °C. 2 mL THF. 3 mL THF.

Reactivation of the Poisoned Catalyst. Eﬃcient

Catalysis with 0.2% Pd. While it is not entirely clear how

7 is formed during the catalytic process, the revealed catalyst

deactivation pattern diﬀers considerably from that previously

established²⁷for the related Pd-catalyzed aromatic cyanation.

Importantly, unlike cyanide, azide is not capable of displacing

the phosphine ligand on Pd(II) complexes involved in the

transformation. Possible reaction pathways from the catalyti-

cally active Pd complexes discussed below to 7 might involve

single electron transfer or redox processes. Although obviously

occurring to a minor extent, these side reactions eventually lead

to the formation of inactive 7 as the major Pd species in the

reaction medium. Luckily, however, this catalyst deactivation

does not involve structural changes to the phosphine ligand.

Evidently, during the catalysis the P centers of Xantphos remain

inaccessible to engage in the Staudinger reaction (Scheme 3)

and are not oxidized in the Pd(II)/P(III) to Pd(0)/P(V)

process.³²We therefore reasoned that the catalytic activity

could be recovered by reducing the poisoned catalyst in the

form of [(Xantphos)PdI₂] (7) to Xantphos-stabilized Pd(0).

It was found that Zn dust and polymethylhydrosiloxane

(PMHS) can both reduce 7 to Pd(0). Of the two, low-cost,

stable, and environmentally friendly PMHS was not only more

attractive but also showed superior performance. After a series

of catalytic runs, it was established that, in the presence of 0.25

mol equiv of PMHS, iodobenzene can be azidocarbonylated at

>99% conversion in 4−6 h with only 0.2% Pd.³³Performing the

reaction in the presence of PMHS also allowed the use of air-

stable and easily accessible [(Xantphos)PdCl₂] (8) as added

catalyst in place of the oxygen-sensitive Xantphos-Pd₂(dba)₅

system. In this way, 8 is reduced in situ with PMHS to give

catalytically active Xantphos-ligated Pd(0). Another advantage

of using 8 is the avoidance of the presence in the reaction of

dba that might contaminate the aroyl azide product.⁹It was also

taken into consideration that the reduction of [(Xantphos)-

PdCl₂] with PMHS in the presence of water produces HCl that

lowers the pH of the medium, thereby increasing the

concentration of hazardous hydrazoic acid. Although HCl is

generated in only minute quantities, 0.4% (2 equiv per 8 that is

used in the amount of 0.2 mol %), we performed the catalytic

reaction in the presence of K₂CO₃(2%) to neutralize the HCl

and thus prevent the formation of HN₃. Under the optimized

new conditions, a series of aryl iodides were successfully

converted to the corresponding aroyl azides with only 0.2 mol

% of 8 (Table 2). These reactions were performed on a 10

E

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

mmol scale and furnished the corresponding pure products in

78−89% isolated yield.

Numerous attempts to eﬃciently azidocarbonylate less

reactive bromoarenes were unsuccessful. While ligands,

solvents, and temperature were varied in a broad range, the

conversion never exceeded 5% even for more electrophilic,

electron-deﬁcient aryl bromides. At the elevated temperatures

required for Ar−Br oxidative addition to Xantphos-stabilized

Pd(0) to occur, the catalyst is quickly deactivated by the

Staudinger reaction of the ArCON₃product with the

phosphine ligand. In the presence of CO, activation of

bromoarenes with tertiary phosphine complexes of zerovalent

palladium is even more sluggish and therefore even higher

temperatures are needed (see below). Indeed, evidence has

been reported^34,35for the diminished reactivity of phosphine-

ligated Pd(0) under CO pressure. This is in full accord with the

fact that higher CO concentrations result in more eﬃcient

displacement of the phosphine ligands on the metal with π-

acidic carbonyls, thereby lowering the reactivity of the Pd(0)

center toward electrophiles.

Safety. The reactions summarized in Table 2 produced

gram quantities of aroyl azides that were isolated and puriﬁed.

As many azido derivatives are explosive, comments are due on

the safety aspects of the developed catalytic azidocarbonylation

method. The so-called “azidophobia” often originates from the

extreme shock sensitivity of certain metal azides.³⁶While care

must be exercised when handling organic azides, some of them

are less explosive than others and have been safely used on a

large scale.³⁷We have not experienced safety problems running

the azidocarbonylation reactions and manipulating organic

azide derivatives reported herein.

Figure 4. ORTEP drawing of cis-[(Xantphos)Pd(N₃)₂]·CHCl₃(cis-9·

CHCl₃) with the CHCl₃molecule and all H atoms omitted and

thermal ellipsoids drawn at the 50% probability level.

Xantphos at room temperature to give [(Xantphos)Pd(Ph)I]

(10) has been reported by one of us previously.²⁹We reasoned,

however, that in the presence of CO, Ar−I activation with

Xantphos-stabilized Pd(0) might involve mixed phosphino−

carbonyl species and result in a diﬀerent outcome. Surprisingly

little is known about oxidative addition to Pd(0) complexes

bearing both tertiary phosphine and CO ligands.^34,39

Mixing Xantphos with Pd₂(dba)₅in a 1:1 ligand to Pd ratio

in benzene-d₆produced two ³¹P NMR-observable species,

[(Xantphos)Pd(dba)] (11; two doublets at 11.3 and 13.5 ppm;

J

_P−P= 10.6 Hz) and [(Xantphos)₂Pd] (two broad multiplets at

3.5 and 6.0 ppm),²⁹in a 4:1 ratio. This pattern is similar to that

previously reported¹⁷for the 4,7-di-tert-butylXantphos-

Pd₂(dba)₃system. The structure of 11 (Figure 5) was

established by an X-ray diﬀraction study of 11·3THF obtained

from a repeat of the reaction of Xantphos with Pd₂(dba)₅in

THF.

In this work, we dealt not only with organic azides but also

with azido complexes of palladium. The simplest palladium

azide, [Pd(N₃)₂], is highly shock sensitive and explosive.³⁸

Tetraazidopalladates such as [Pd(NH₃)₄]²⁺[Pd(N₃)₄]²⁻are

detonated by friction or shock.^38bOn the other hand,

nonhomoleptic Pd(II) azide derivatives bearing other ligands

can be stable and nonexplosive, e.g., [Py₂Pd(N₃)₂].^38bAll of the

phosphine-ligated mono-azido organopalladium complexes

prepared in this work were stable compounds that did not

exhibit any signs of explosiveness. It was conceivable, however,

that small quantities of diazido Pd(II) species could emerge in

the catalytic process from metathesis reactions of the NaN₃

reagent with [(Xantphos)PdI₂] (7, catalyst deactivation

product) or [(Xantphos)PdCl₂] (8, added catalyst). To assess

the safety of such diazido species, we prepared [(Xantphos)-

Pd(N₃)₂] (9) by reacting 8 with sodium azide in CHCl₃−H₂O

(94% yield). Complex 9 was fully characterized, including by

single-crystal X-ray diﬀraction (Figure 4) and found to be

stable. This diazide was not shock-sensitive (striking with a

hammer) and did not show any signs of decomposition in

DMF solution at 100 °C after 4 h. We conclude that Xantphos-

stabilized azido palladium species involved in the catalytic

process are safe compounds. One must bear in mind, however,

that (i) in the presence of NaN₃and in the absence of

stabilizing tertiary phosphine or other ligands a Pd(II) source

may be converted to highly explosive species such as

[Pd(N₃)₂]³⁸and (ii) azido derivatives should always be

handled with care.

Figure 5. ORTEP drawing of [(Xantphos)Pd(dba)]·3THF (11·

3THF) with the THF molecules and all H atoms omitted for clarity

and thermal ellipsoids drawn at the 50% probability level.

Adding CO (1 atm) to the solution of [(Xantphos)Pd(dba)]

and [(Xantphos)₂Pd] generated from Xantphos and Pd₂(dba)₅

in benzene-d₆or THF resulted in instantaneous full conversion

of both complexes and the appearance of only one singlet

resonance at 10.5 ppm (C₆D₆) in the ³¹P NMR spectrum. Two

strong bands at 1974 and 2014 cm⁻¹in the FT-IR spectrum of

the reaction solution in THF suggested⁴⁰that the species

resonating at 10.5 ppm in the ³¹P NMR spectrum is a

dicarbonyl complex, likely [(Xantphos)Pd(CO)₂] (12). This

structure was indeed established by a single-crystal X-ray

diﬀraction study of 12·hexane (Figure 6) obtained from an

independent experiment (see below).

Azidocarbonylation Mechanism. Like all Pd-catalyzed

cross-coupling reactions of aryl halides, the azidocarbonylation

involves oxidative addition of the haloarene substrate to Pd(0)

as the ﬁrst key step. Oxidative addition of PhI to Pd₂(dba)₃/

F

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Figure 6. ORTEP drawing of [(Xantphos)Pd(CO)₂]·n-C₆H₁₄(12·

hexane) with cocrystallized hexane and all H atoms omitted and

thermal ellipsoids drawn at the 50% probability level.

Figure 7. ORTEP drawing of cis-[(Xantphos)Pd(COPh)I]·0.5THF

(cis-13·0.5THF) with the THF molecule and all H atoms omitted and

thermal ellipsoids drawn at the 50% probability level.⁹

We then found that bubbling argon through a solution of 12

(generated from 11 and CO) quantitatively produced 11 within

1 min. These observations clearly indicated that, although CO

binds more tightly to (Xantphos)Pd(0) than dba, the formation

of 12 from 11 is reversible and that the equilibrium between

the two can be easily shifted to 11 under CO-deﬁcient

conditions. Both 11 and 12 were found to react with PhI. The

reaction of 11 with PhI has been previously shown to produce

[(Xantphos)Pd(Ph)I] (10; see above). In the current work, we

found that 10 readily reacted with CO (1 atm) to give

selectively the product of CO insertion into the Pd−Ph bond,

[(Xantphos)Pd(COPh)I] (13), which was also formed upon

addition of PhI to 12 under CO (Scheme 7). Complex 13 was

isolated and fully characterized, including by single-crystal X-ray

diﬀraction (Figure 7).⁹Like its bromo congener [(Xantphos)-

Pd(COPh)Br],^15a13 was found to be cis in the crystal yet

displayed only one ³¹P NMR singlet in THF, benzene, and

CD₂Cl₂. Although this behavior could be attributed to 13 being

exclusively trans in solution, it is likely that the complex

undergoes extremely facile cis−trans isomerization that may

not be frozen out on the NMR time scale. A very low barrier of

only 6.1 kcal/mol has been computed for this isomerization

(see below).

carbonylated at 99% conversion in 2 h. In a repeat of this run

under 7 atm of CO (CO:Pd = ca. 8), only 25% conversion was

reached after the same period of time, indicating that [CO]

does inﬂuence the rate-limiting oxidative addition step, as

shown in Scheme 7 and conﬁrmed by the computational

studies below.

Both products of the oxidative addition, 10 and 13, were

found to readily react with azide, as shown in Scheme 8.

Treatment of 13 with [Bu₄N]⁺N₃in benzene⁴²under CO

−

resulted in the instantaneous clean formation of PhCON₃along

with [(Xantphos)Pd(CO)₂] (12). In this way, the modeled

catalytic cycle was closed because PhCON₃is the ﬁnal product

of the process and 12 can commence another turnover by Ar−I

activation via the oxidative addition (Scheme 7).

Intermediate 10 that would be produced under CO-deﬁcient

conditions (Scheme 7) was also reactive toward azide (Scheme

−

8). The reaction of 10 with 1.1 equiv of [Bu₄N]⁺N₃in

benzene produced [(Xantphos)Pd(Ph)(N₃)] (14) quantita-

tively. This anionic ligand exchange also readily occurred with

NaN₃under biphasic conditions in the absence of a phase-

transfer agent. In benzene−aqueous NaN₃, thermodynamic

equilibrium between 10 and 14 was reached within 5 min at

vigorous stirring. However, only ca. 60% conversion to 14 was

observed at equilibrium even in the presence of 10 equiv of

NaN₃. To isolate 14, the bromo analogue⁴³of 10 was used,

because in the presence of water the equilibrium between the

Pd−X and Pd−N₃(14) complexes is shifted more toward the

latter for X = Br than for X = I⁴⁴since bromide is more strongly

hydrated than iodide. The anion exchange extraction⁴⁵of the

Br⁻from [(Xantphos)Pd(Ph)Br] in CH₂Cl₂with aqueous

NaN₃gave 14, which was isolated pure and fully characterized

Scheme 7

1

in solution by H and ³¹P NMR data and in the solid state by

single-crystal X-ray diﬀraction (Figure 8).⁹

Bubbling CO through a solution of 14 in benzene-d₆resulted

in the instantaneous formation of PhCON₃and [(Xantphos)-

Pd(CO)₂] (12; Scheme 8). Repeating this reaction in toluene

with subsequent addition of hexanes produced X-ray-quality

crystals for the structure determination of 12 (Figure 6).

The results described above pointed clearly to two reaction

pathways to PhCON₃with simultaneous regeneration of Pd(0),

via 13 and via 14 (Schemes 7 and 8). Both routes likely lead to

[(Xantphos)Pd(COPh)(N₃)] (15) as a common intermediate

that undergoes PhCO−N₃reductive elimination. The com-

petition between these two reaction channels is expected to be

As follows from Scheme 7, there are two oxidative addition

pathways leading to the key intermediate 13. If Ar−I oxidative

addition is the rate-limiting step, the catalytic process should be

slower at higher concentrations of CO, favoring the formation

of 12. The latter is less electron-enriched than 11 because of

the stronger π-acidity of two CO ligands in comparison with

dba. Slower reaction rates under higher CO pressures have

been reported³⁵for Pd-catalyzed alkoxycarbonylation reactions.

The literature data⁴¹on the solubility of CO in benzene

allowed the CO to Pd ratio in our standard catalytic

azidocarbonylation runs (1 atm, 25 °C) to be estimated at ca.

1:1. Under such conditions, 4-ﬂuoroiodobenzene was azido-

−

strongly dependent on the concentrations of CO and N₃, the

nature of the medium, and other factors that control anionic

G

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Scheme 8

performed in the presence of CO in excess ([Pd] ≪ [CO]),

the pathway involving 13 is expected to dominate because the

facile migratory insertion of CO into the Pd−C bond of 10 is a

homogeneous process rapidly occurring in the organic phase,

whereas the alternative route, via 14, requires slower anionic

ligand exchange at the liquid−liquid interface of the biphasic

system. The prevailing catalytic cycle would then involve the

sequence 12 → 10 → 13 → 15 → 12 (loop A in Scheme 9). It

remains unknown if the oxidative addition of ArI to 12 occurs

via predissociation of both carbonyl ligands from the Pd center.

If it does, then 10 is likely involved in the cycle, albeit as a

short-lived intermediate that is quickly transformed to 13 via

facile CO insertion. However, if at least one CO ligand remains

on Pd during the Ar−I oxidative addition, the formation of 10

might be skipped, with 12 going directly to 13, i.e., 12 → 13 →

15 → 12 (Scheme 9, loop B). Under CO-deﬁcient conditions

([Pd] > [CO]), two other catalytic pathways are possible: 11

→ 10 → 13 → 15 → 11 (loop C) and 11 → 10 → 14 → 15 →

11 (loop D). The competition between loops C and D would

obviously depend on relative rates of the transformation of 10

Figure 8. ORTEP drawing of trans-[(Xantphos)Pd(Ph)(N₃)] (trans-

14) with all H atoms omitted and thermal ellipsoids drawn at the 50%

probability level.⁹

ligand exchange and migratory insertion processes as well as

mass transfer in the triphasic liquid−liquid−gas system. Two

extreme cases could be considered. If the catalytic reaction is

Scheme 9

H

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Figure 9. Alternative computed reaction proﬁles for PhI activation at [(Xantphos)Pd(CO)₂] (12). Relative free energies in water are given in kcal/

mol.

Figure 10. Computed transition state geometries for Ph−I activation at (a) [(Xantphos)Pd] and (b) [(Xantphos)Pd(CO)]. Relative free energies in

water are given in kcal/mol as well as selected distances in Å. Xantphos Ph groups are truncated at the ipso carbon and H atoms are omitted for

clarity.

to 13 (CO insertion) and to 14 (I⁻/N₃exchange), as well as

of carbonylation of 14 to 15. As described above, the possibility

for all four pathways A−D (Scheme 9) has been demonstrated

by stoichiometric experiments. Regardless of which of the

reaction channels wins the competition, the process invariably

leads to the desired aroyl azide product. Importantly, unlike

[(R₃P)₂Pd(Ph)(N₃)]⁷(R = Me, Et; see above), 14 does not

form a Pd-NCO species via N₂loss upon treatment with CO.

In our experimental mechanistic studies, we have succeeded

in the isolation and full characterization of all intermediates

involved in the azidocarbonylation process (Schemes 7−9; Ar =

Ph), except for [(Xantphos)Pd(COPh)(N₃)] (15). As 15 is too

unstable for isolation and/or detection, quickly undergoing

reductive elimination of PhCON₃, we studied the various

mechanistic routes to this species and the subsequent product-

forming C−N bond coupling event by computational means.

(BS1) protocol, i.e. based on free energies derived from gas-

phase optimizations computed with a smaller basis set (BS1),

and adding to this corrections for solvation (water, PCM

approach computed with BS1), dispersion, and basis eﬀects, the

last via recomputation of the SCF energy with a larger basis set

(BS2) (see Computational Details).

We ﬁrst considered the diﬀerent species that may be involved

in the initial Ph−I oxidative addition step under conditions of

excess CO (i.e., loop A vs loop B; Scheme 9). The most stable

Pd(0) precursor under these circumstances is [(Xantphos)Pd-

(CO)₂] (12; G = 0.0 kcal/mol), and successive CO

dissociations from this species lead to trigonal-planar

[(Xantphos)Pd(CO)] (G = +12.2 kcal/mol) and [(Xantphos)-

Pd] (G = +38.1 kcal/mol; see Figure 9). PhI can add to the

latter to give an η²-adduct bound through the C_ipso−C_orthobond

(I1; G = +26.7 kcal/mol) which is the direct precursor to C−I

cleavage. This proceeds with a small additional barrier of 4.7

−

COMPUTATIONAL STUDIES

kcal/mol through a pseudotetrahedral transition state (TS_OA1

;

■

see Figure 10(a)) which leads initially to I3 (G = +11.2 kcal/

mol), an isomer of 10 featuring a square-pyramidal geometry

with a weakly bound axial iodide and a κ³-P,O,P Xantphos

ligand.⁴⁶Elongation of the Pd−O bond leads to trans-10 (G =

Density functional theory (DFT) calculations have been

employed to assess the diﬀerent mechanistic possibilities

outlined in Scheme 9. Throughout we report free energies

corrected for water solvent using a BP86-D3(BS2)//BP86-

I

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

−

Figure 11. Computed reaction proﬁles for the formation of [(Xantphos)Pd(CO)] and PhCON₃from I4(I) + N₃. Relative free energies in water

are given in kcal/mol.

Figure 12. Computed transition states for (a) CO migratory insertion in [(Xantphos)Pd(CO)(I)(Ph)] (I4(I)) and (b) C−N reductive coupling in

cis-[(Xantphos)Pd(COPh)(N₃)] (cis-15). Relative free energies in water are given in kcal/mol and selected distances in Å. Xantphos Ph groups are

truncated at the ipso carbon, and hydrogen atoms are omitted for clarity.

+4.2 kcal/mol) with a minimal barrier of 1.7 kcal/mol via

TS_isom. Alternatively, Ph−I cleavage may occur at [(Xantphos)-

Pd(CO)]. In this case no η²-adduct could be located, possibly

due to increased steric encumbrance around the Pd center;

instead, the most stable precursor located is a noncovalently

bound adduct in which PhI lies above the metal coordination

plane with the iodine directed toward the Pd center (I2; Pd···I

= 3.06 Å; G = +8.8 kcal/mol). From here Ph−I bond cleavage

occurs via TS_OA2(G = +23.5 kcal/mol) which displays a

distorted trigonal-bipyramidal geometry with CO and I in the

axial sites (see Figure 10(b)). TS_OA2also features signiﬁcant

elongation of the Pd−P2 distance (by ca. 0.14 Å cf. I2), and

IRC calculations show that this bond breaks completely in

forming intermediate I4(I) (Pd···P2 = 3.66 Å; G = +6.5 kcal/

mol). I4(I) displays a square-planar coordination geometry

around Pd with CO again trans to I, this being favored (as in

TS_OA2) due to the trans arrangement of π-acceptor and π-

donor ligands. Figure 9 shows that Ph−I bond activation is

most accessible via [(Xantphos)Pd(CO)] and proceeds with an

overall barrier of 23.5 kcal/mol to form [(Xantphos)Pd(CO)-

(I)(Ph)] directly, i.e., suggesting that loop B will be in

operation (Scheme 9).

consistent with more facile aryl halide activation in the absence

of CO noted above.

The computed reaction proﬁle for the remainder of the

catalytic cycle along loop B is shown in Figure 11. CO

migratory insertion in I4(I) proceeds via TS_MI(I) with a barrier

of 4.4 kcal/mol and forms trans-13 at −7.8 kcal/mol,⁴⁷which is

computed to be 0.4 kcal/mol less stable than cis-13 in solution.

Formation of cis-15 may proceed via trans−cis isomerization in

13 and I⁻/N₃exchange or via I⁻/N₃exchange in trans-13

followed by trans−cis isomerization in 15. Both possibilities

appear to be readily accessible, for example, trans−cis

isomerization in 13 has a barrier of only 6.1 kcal/mol, while

−

both I⁻/N₃exchange processes are computed to lie close to

−

equilibrium.⁴⁸Once cis-15 is formed, C−N bond formation

proceeds via TS_RCwith a barrier of 11.9 kcal/mol, readily

accessible at room temperature and so consistent with 15 being

unobserved experimentally.⁴⁹This gives I5, the most stable

form of which has PhCON₃bound as an η²-arene fashion

through a C_ipso−C_orthobond (G = −6.4 kcal/mol). The

PhCON₃product can then be displaced from I5 by CO to

regenerate [Pd(Xantphos)(CO)]. Each of these steps is readily

accessible and the calculations suggest that after rate-limiting

oxidative addition the formation of products and regeneration

of the catalyst 12 will be rapid and strongly exergonic (ΔG =

−30.4 kcal/mol). Computed geometries for TS_MI(I) and TS_RC

are both shown in Figure 12, and all other geometries are given

in the Supporting Information.

We have also considered Ph−I activation under low CO

concentrations. Under these conditions the likely Pd(0)

precursor is [(Xantphos)Pd(dba)] (11), which is computed

to lie 10.0 kcal/mol above [(Xantphos)Pd(CO)₂]. The

estimated barrier (via TS_OA1) would therefore be 21.4 kcal/

mol, assuming facile displacement of dba by PhI, and this is

J

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Further calculations probed the eﬀect of I⁻/N₃⁻exchange on

those reactions may be performed in situ, without isolation of

the originally produced ArCON₃(Scheme 6).

the migratory insertion process. I⁻/N₃substitution in I4(I)

−

The biggest drawback of the Pd-catalyzed azidocarbonylation

is its inapplicability to aryl bromides. Being less reactive than

aryl iodides, bromoarenes can be activated via oxidative

addition to Xantphos-stabilized Pd(0) only at elevated

temperatures that aroyl azides do not survive. Furthermore,

diminished stability constants of various Pd(Xantphos) units

involved in the catalysis at higher temperatures facilitate the

Staudinger reaction leading to catalyst deactivation, as

described above. It is particularly important that Ar−X

activation with phosphine-stabilized Pd(0) takes place with a

higher activation barrier in the presence than in the absence of

CO. Our studies shed new light on this poorly studied and

understood key mechanistic feature (see below).

Reaction Mechanism. The azidocarbonylation reaction

occurs under triphasic liquid−liquid−gas conditions. Aside

from mass transfer that clearly depends on such parameters as

agitation rate and pressure, the rate-determining step of the

catalytic process is Ar−I activation with Pd(0). Since the

pioneering work of Fitton and co-workers,⁵⁰oxidative addition

of haloarenes to tertiary phosphine complexes of zerovalent

palladium has been studied in considerable detail.⁵¹Migratory

insertion of CO into a variety of complexes of the type

[(R₃P)M(Ar)X] (M = Pt, Pd, Ni) has been thoroughly studied

in the classic paper by Garrou and Heck.⁶Strikingly, however,

very little mechanistic information has been reported on

oxidative addition of aryl halides to tertiary phosphine-

stabilized Pd(0) complexes in the presence of CO, i.e., under

the catalytic conditions of the Heck carbonylation that has been

known and widely used for nearly 40 years.¹

The experimental and computational results obtained in the

current work allow us to analyze two extreme cases: (1) the

catalytic reaction is carried out with CO in excess ([Pd] <

[CO]; loops A and B in Scheme 9) and (2) the process occurs

under CO-deﬁcient conditions ([Pd] > [CO]; loops C and D

in Scheme 9). The experimentally observed slower reaction rate

at higher concentrations of CO (see above) is in accord with

the computational results. The prohibitively high barrier (38.1

kcal/mol) to the loss of both carbonyl ligands from the

(Xantphos)Pd⁰moiety (Figure 9) rules out Ar−I oxidative

addition to the carbonyl-free Pd(0) in the presence of CO in

excess. Dissociation of only one CO from [(Xantphos)Pd-

(CO)₂], however, occurs with a much lower free energy cost of

only 12.2 kcal/mol, suggesting that [(Xantphos)Pd(CO)] is

accessible. (In the presence of larger quantities of CO, the

equilibrium between the two is obviously shifted to the

unreactive dicarbonyl complex.) Furthermore, it has been

demonstrated by the calculations that the monocarbonyl

[(Xantphos)Pd(CO)] can activate the Ph−I bond with an

overall barrier of 23.5 kcal/mol. This is reasonably consistent

with the estimated value ΔG^⧧≈ 21.5 0.5 kcal/mol at 296 K

from the experimental data for oxidative addition of 4-FC₆H₄I

to [(Xantphos)Pd(CO)₂] (12; see the Supporting Informa-

tion). However, [(Xantphos)Pd(Ar)(I)] is not on the reaction

coordinate in this case and the process is mediated by

[(Xantphos)Pd(CO)(Ar)(I)], leading to [(Xantphos)Pd-

(COAr)I], as shown in Figure 11 for Ar = Ph. Therefore, in

the presence of CO in excess, the reaction is governed by loop

B, not A (Scheme 9).

gives I4(N₃) (G = +6.0 kcal/mol), in which azide is now trans

to CO. Thus, I⁻/N₃exchange is again close to equilibrium

−

(ΔG = −0.5 kcal/mol), although the subsequent migratory

insertion transition state is now less accessible than for the

iodide analogue (TS_MI(N₃): G = +15.4 kcal/mol cf. TS_MI(I) at

+10.9 kcal/mol). Both processes are more accessible than the

preceding (overall rate-limiting) Ph−I oxidative addition, and

so a proportion of the catalysis could therefore proceed through

I4(N₃) leading to the direct formation of trans-15.

DISCUSSION

■

Selectivity, Eﬃciency, Scope, and Limitations of the

Reaction. The azidocarbonylation reaction represents a novel,

methodologically distinct approach to aroyl azides. While all

previously developed methods to synthesize ArCON₃employ

starting materials that already include the carbonyl moiety,³our

reaction constructs the desired product of ArI, CO, and N₃⁻at

the metal center of a Pd catalyst. The uniquely remarkable

catalytic activity of the Pd−Xantphos system identiﬁed here

allows numerous problems to be obviated that otherwise would

prevent the azidocarbonylation from occurring. As discussed in

the Introduction, those include the thermal decomposition of

the product (Curtius rearrangement), catalyst deactivation from

the exceedingly facile Staudinger reaction of the aroyl azide

produced with the stabilizing tertiary phosphine on the metal,

and transformation of the N₃ligand on Pd to NCO with

concomitant loss of N₂(Schemes 1−3).

Eﬃciently catalyzed by the Pd-Xantphos system, the

azidocarbonylation reaction has a broad scope and can be

used with a wide variety of meta- and para-substituted aryl

iodides. Functional groups such as alkyl, alkoxy, acyl,

alkoxycarbonyl, nitro, cyano, and even formyl are easily

tolerated. Although the reaction readily occurs at room

temperature and atmospheric pressure, it is inapplicable to

the synthesis of ortho-substituted aroyl azides that are

intrinsically less stable toward the Curtius rearrangement.

The isocyanates thus formed in the reaction are then involved

in subsequent transformations in situ, leading to other

products, including ureas (Scheme 5).

The catalyst deactivation study identiﬁed [(Xantphos)PdI₂]

(7) as the main product of catalyst poisoning. Since fortuitously

no structural changes to the Xantphos ligand are involved in the

deactivation process, the nonorganometallic Pd(II) complex 7

can be reduced in situ to Pd(0), thereby regaining the catalytic

activity. The identiﬁcation of readily available and cheap PMHS

as the particularly eﬃcient reducing agent for this purpose has

led to the development of the markedly eﬃcient process

employing only 0.2 mol % of easily accessible, air-stable

[(Xantphos)PdCl₂] (8) as added catalyst. Cross-coupling

reactions of haloarenes employing such a low Pd catalyst

loading are rare. Considering the above-explained vulnerability

of the reaction, it is truly remarkable that nearly quantitative

conversions can be achieved with such small quantities of the

catalyst. Furthermore, the reaction exhibits >85% selectivity if

run under optimized conditions. The low catalyst loading in

combination with the high conversion and selectivity allow the

straightforward and safe preparation of pure aroyl azides in ca.

80−90% isolated yield on a 1−2 g scale. Nonetheless, if an

aroyl azide is desired for a further transformation, such as to the

corresponding isocyanate, benzamide, iminophosphorane, etc.,

We deﬁne CO-deﬁcient conditions as [Pd] ≈ [CO] or [Pd]

> [CO] and the CO diﬀusion rate is slower than that of the

catalytic reaction. Under such conditions, loops C and D

K

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

(Scheme 9) are expected to be operational, since the Pd(0)

produced in the product-forming step would undergo fast Ar−I

oxidative addition before the CO consumed in the catalytic

cycle is replenished from the gas phase. With [(Xantphos)Pd-

(dba)] (11) as the resting state of the Pd(0) form of the active

catalyst, the computed barrier is 21.4 kcal/mol (see above).

The resultant stable and isolable²⁹intermediate [(Xantphos)-

Pd(Ph)(I)] (10) can then either undergo carbonylation to

[(Xantphos)Pd(COPh)I] (13) when CO becomes available

(loop C) or react with azide to form [(Xantphos)Pd(Ph)(N₃)],

(14; loop D). Mass transfer is involved in both processes since

the Pd catalyst is located in the organic phase, the azide source

(NaN₃) is in the immiscible aqueous layer, and CO is in the

headspace. Although under anhydrous conditions the exper-

imentally observed equilibrium between 10 and 14 is shifted

entirely to the latter,⁵²in benzene−aqueous NaN₃both are

present in comparable quantities (see above). However, 10 and

14 exhibit diﬀerent reactivities toward CO. The computed data

(in water) suggest that migratory insertion of CO into the Pd−

C bond of 10 is >10³times faster than that of 14. This indicates

that if CO diﬀusion into the organic phase is slower than the

iodide/azide ligand exchange, of the two loops C and D it is the

former that will be by far the largest contributor to the overall

catalytic transformation. The inﬂuence of the anionic ligand X

on both kinetics and thermodynamics of the reactions of

[(R₃P)₂Pd(Ar)X] with CO has long been recognized.⁶For

instance, [(Ph₃P)₂Pd(Ar)X] species undergo full conversion to

the corresponding [(Ph₃P)₂Pd(COAr)X] more rapidly for X =

I than for X = Br (Ar = 4-NO₂C₆H₄or 4-NCC₆H₄). However,

CO insertion into the Pd−Ph bond of [(R₃P)₂Pd(Ph)Cl] is

reversible for both R = Ph⁶and Cy.⁵³

precursors for the preparation of radiotracers.⁵⁴The short

lifetime of the ¹¹C isotope (t_1/2= 20.4 min) dictates the need

for highly eﬃcient and selective organic transformations that

produce the desired ¹¹C-containing product within ca. 10 min.

Pd-catalyzed carbonylation reactions have been used to

produce ¹¹C-labeled aldehydes, ketones, carboxylic acids, and

their derivatives.^54,55Very recently, an eﬃcient protocol with a

Pd-Xantphos catalyst was developed.^55d

Under optimized conditions, the azidocarbonylation reaction

occurs on the time scale that might be suitable for ¹¹C PET. For

instance, 85% conversion of 4-ﬂuoroiodobenzene was achieved

in 10 min, using the standard protocol. In another experiment

using 1 equiv of CO, 4-ﬂuorobenzoyl azide was cleanly

produced in 66% yield in 5 min (see the Supporting

Information). It is believed that the azidocarbonylation reaction

may ﬁnd applications in ¹¹C PET, considering (i) the

appropriate time scale, (ii) the fact that aroyl azides are highly

reactive and versatile reagents in synthesis,³and (iii) the

demonstrated possibility to use the ArCON₃product for

further transformations without isolation (Scheme 6).

Why Xantphos? Of numerous tertiary phosphines screened

in the current work, Xantphos is the only eﬃcient ligand for the

Pd-catalyzed azidocarbonylation reaction. The well-established

remarkable catalytic activity of Xantphos-stabilized transition-

metal complexes¹⁴is often attributed to its wide bite angle. It

has been recently demonstrated,³⁰however, that factors other

than that may be involved. To understand the origin of the

uniquely excellent performance of the Xantphos-Pd catalyst in

the azidocarbonylation reaction, a separate mechanistic study is

needed, similar to that performed in the current work but with

analogous Pd complexes bearing ligands other than Xantphos.

As seen from the above, three diﬀerent catalytic cycles (loops

B−D in Scheme 9) can govern the azidocarbonylation reaction.

With CO in excess in the organic phase, loop B is operational.

Under CO-deﬁcient conditions, loops C and (to a much

smaller extent) D operate the process. It is quite possible that

under the conditions conventionally used in the current work,

i.e. [Pd] ≈ [CO] in the organic phase at 1 atm (see above), the

reaction occurs largely via the C channel at the beginning when

the concentration of the iodoarene substrate is high and the

rate-limiting oxidative addition is faster than CO mass transfer.

The formation of less reactive Pd(0) carbonyls is skipped in

this way and, as a result, it is the more reactive CO-free Pd(0)

that eﬀects the Ar−I oxidative addition. As [ArI] drops during

the reaction, the rate of its oxidative addition ﬁrst becomes

comparable with, and eventually slower than, the CO diﬀusion

rate which is constant. In other words, the mechanism that

operates at the beginning of the reaction (loop C) is gradually

replaced with the other, higher barrier channel (loop B). Poorly

eﬃcient azide/iodide ligand exchange in the organic solvent−

water biphasic system (e.g., slow agitation) would result in

accumulation of stable [(Xantphos)Pd(COPh)I] (13) and

consequently higher concentrations of CO, thus favoring the

more energy demanding channel (loop B). The diminished

reactivity of aryl bromides prohibits the intrinsically faster

catalytic cycles C and D from operation because Ar−Br

oxidative addition to Pd(0) is considerably slower than CO

mass transfer under the standard reaction conditions. The only

mechanistic option left for aryl bromides is loop B that involves

the less reactive form of Pd(0), [(Xantphos)Pd(CO)₂] (12).

Potential Applications in ¹¹C Radiolabeling. ¹¹C

positron emission tomography (PET) is a powerful diagnostic

tool that currently employs ¹¹CO₂, ¹¹CH₃I, and ¹¹CO

CONCLUSIONS

■

We have discovered the ﬁrst aromatic azidocarbonylation

reaction, the formation of aroyl azides from aryl iodides, CO,

and NaN₃in the presence of a Pd catalyst. This transformation

is vastly more challenging than other Heck-type Pd-catalyzed

carbonylation reactions because (i) the product, ArCON₃, is

thermally unstable (Curtius rearrangement) under conven-

tional ArX carbonylation conditions, (ii) the Pd catalyst may be

easily and irreversibly deactivated by the exceedingly facile

Staudinger reaction of the aroyl azide product with the

stabilizing tertiary phosphine ligand, and (iii) the azido ligand

on Pd may react with CO to give isocyanate with concomitant

loss of N₂.⁷Remarkably, all these problems are obviated by the

use of the unique Pd-Xantphos catalytic system that eﬀects the

transformation in an organic solvent−H₂O biphasic system at

room temperature and 1 atm of CO. The catalytic process

exhibits high eﬃciency and functional group tolerance.

A catalyst poisoning study has shown that the deactivation

process does not involve structural changes of the stabilizing

ligand but rather leads to the catalytically inactive non-

organometallic complex [(Xantphos)PdI₂]. This has allowed

us to ﬁnd a way to revitalize the poisoned catalyst by

performing the reaction in the presence of cheap and readily

available PMHS that reduces the inactive Pd(II) to catalytically

active Pd(0). As a result, the catalyst loading has been lowered

from 2% to 0.2% without any losses in selectivity at nearly

100% conversion. The synthetic value of the process with this

uncommonly low catalyst loading has been demonstrated by

the successful synthesis of a series of aroyl azides as

spectroscopically and analytically pure compounds in ca. 80−

90% isolated yield on a gram scale. Alternatively, the products

L

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Center at the Complutense University of Madrid. Detailed procedures

for the azidocarbonylation in the absence of reducing agents (2% Pd;

Table 1) can be found in the preliminary communication.⁹

may not be isolated but rather used in situ for further

transformations such as the synthesis of isocyanates, ureas,

benzamides, iminophosphoranes, etc.

General Procedure for Azidocarbonylation of Aryl Iodides in

the Presence of PMHS (0.2% Pd). A 25 mL round-bottom ﬂask

equipped with a gas inlet and a Teﬂon-coated magnetic stir bar was

charged, in air, with an iodoarene (10.0 mmol), NaN₃(0.78 g; 12

mmol), [(Xantphos)PdCl₂]·CH₂Cl₂(8; 17 mg; 0.2 mol %), K₂CO₃

(28 mg; 2 mol %), and water (2 mL). After two freeze−pump−thaw

cycles, the ﬂask was back-ﬁlled with CO and connected to a CO

balloon via a 1.2 mm diameter stainless steel syringe needle. With

vigorous stirring, a solution of PMHS (150 mg; 25 mol %) in oxygen-

free THF was syringed in and the mixture was agitated at room

temperature. After GC-MS analysis of the organic phase indicated 95−

100% conversion of the iodoarene, the product was isolated and

puriﬁed in air. The reaction mixture was diluted with ether (10 mL)

and transferred to a separatory funnel. The reaction ﬂask was rinsed

with ether (2 × 10 mL) and water (10 mL), and the washings were

added to the separatory funnel. After shaking, the organic phase was

separated and the aqueous phase was washed with Et₂O (2 × 20 mL).

The combined extract and the washings were dried over MgSO₄,

ﬁltered, and rotary-evaporated at room temperature. Silica gel column

chromatography of the residue gave the pure product (see below and

the Supporting Information for speciﬁcs).

Like any other process, the azidocarbonylation reaction has

limitations. First, it is inapplicable to aryl bromides that are

activated by Pd(0) at temperatures that prompt the rearrange-

ment of the aroyl azide product and its reaction with the

stabilizing phosphine ligand, which leads to catalyst deactiva-

tion. Second, ortho-substituted iodoarenes, when azidocarbo-

nylated, give rise to the corresponding ArCON₃that are

innately less stable toward the Curtius rearrangement. In

certain cases, the reaction of ortho-substituted substrates can

produce the corresponding symmetric ureas in good yield.

A detailed mechanistic study of the azidocarbonylation

reaction by experimental and computational means has shown

that two main reaction pathways can operate in the process.

Oxidative addition of the Ar−I bond to Pd(0) is the rate-

determining step of both routes. In the presence of excess CO,

the Ar−I bond is activated by the less electron-rich Pd center of

a mixed carbonyl phosphine complex. Under CO-deﬁcient

conditions, a lower energy barrier pathway is followed,

involving Ar−I oxidative addition to a more reactive carbonyl

-free Xantphos-stabilized Pd(0) species. Mass transfer in the

triphasic liquid−liquid−gas system used in the current work

plays an important role in the competition between these two

mechanistic routes, uniformly leading to a common aroyl azido

intermediate that undergoes ArCO−N₃reductive elimination.

It is hoped that the novel azidocarbonylation reaction will

ﬁnd applications in the synthesis of some otherwise poorly

accessible aroyl azides and that the previously unavailable

mechanistic information will be of help in the design and

optimization of other Pd-catalyzed carbonylation reactions of

aromatic electrophiles.

Benzoyl Azide (2a). Azidocarbonylation of iodobenzene (2.04 g) by

the general procedure using a solution of PMHS in THF (0.5 mL)

gave >99% conversion after 20 h. Pentane−ether (9:1 v/v) was used

for isolation by column chromatography. After solvent removal and

drying on a rotary evaporator (100 mbar for 1 h, then 80 mbar for 0.5

h), 2a was obtained as a colorless oil that crystallized on standing at +8

°C. The yield was 1.29 g (88%). ¹H NMR (500 MHz, CDCl₃): δ 7.46

(m, 2H), 7.62 (m, 1H), 8.03 (m, 2H).⁹

4-Methylbenzoyl Azide (2b). Azidocarbonylation of 4-iodotoluene

(2.18 g) by the general procedure using a solution of PMHS in THF

(0.5 mL) gave >99% conversion after 16 h. Hexane−ether (9:1 v/v)

was used for isolation by column chromatography. After solvent

removal and drying on a rotary evaporator, 2b was obtained as a

colorless oil that crystallized on standing at +8 °C. The yield was 1.33

EXPERIMENTAL SECTION

■

1

g (83%). H NMR (400 MHz, CDCl₃): δ 2.42 (s, 3H, Me), 7.25 (m,

Caution! Care should be exercised when handling azide derivatives that

may be potentially explosive. Exposing phosphine-free palladium

compounds to a source of azide must be avoided in order to eliminate

the risk of generating highly explosive, shock-sensitive materials.³⁸Aroyl

azides, pure or in solution, should not be exposed to temperatures exceeding

2H), 7.92 (m, 2H).⁹

4-Methoxybenzoyl Azide (2d). Azidocarbonylation of 4-iodoanisole

(2.34 g) by the general procedure using a solution of PMHS in THF

(2.0 mL) gave 95% conversion after 44 h. Hexane−ether (14:1 v/v)

was used for isolation by column chromatography. After solvent

removal and drying on a rotary evaporator, 2d was obtained as a white

o

60 C to avoid decomposition via the Curtius rearrangement.

All chemicals were purchased from Aldrich, Alfa Aesar, TCI,

Deutero, and Pressure Chemical. Benzene, toluene, THF, dichloro-

methane, hexanes, and other solvents were used as received, unless

otherwise noted. Anhydrous, oxygen-free benzene, benzene-d₆,

toluene, xylenes, THF, and hexanes for mechanistic studies were

obtained by distillation from Na/OCPh₂. CD₂Cl₂and CDCl₃were

vacuum-transferred from CaH₂. All solvents for experiments under an

inert atmosphere were stored over freshly activated 4 Å molecular

sieves in a glovebox. Column chromatography was performed on 60A

silica gel (40−63 μm). The complexes Pd₂(dba)₅,¹³[(Xantphos)Pd-

(Ph)I] (10),²⁹[(Ph₃P)₂Pd(Ph)Br],⁵⁶[(Xantphos)PdCl₂]·CH₂Cl₂

(8),⁵⁷[(Xantphos)Pd(COPh)I] (13),⁹and [(Xantphos)Pd(Ph)N₃]

1

crystalline solid. The yield was 1.38 g (78%). H NMR (400 MHz,

CDCl₃): δ 3.87 (s, 3H, OMe), 6.93 (m, 2H), 7.99 (m, 2H).⁹

4-Nitrobenzoyl Azide (2e). Azidocarbonylation of 4-nitroiodoben-

zene (2.49 g) by the general procedure using a solution of PMHS in

THF (3.0 mL) gave >99% conversion after 19 h. Hexane−ether (2:1

v/v) was used for isolation by column chromatography. After solvent

removal and drying on a rotary evaporator, 2e was obtained as

1

yellowish crystals. The yield was 1.66 g (86%). H NMR (500 MHz,

CDCl₃): δ 8.21 (m, 2H), 8.31 (m, 2H).⁹

4-Acetylbenzoyl Azide (2f). Azidocarbonylation of 4-iodoacetophe-

none (2.46 g) by the general procedure using a solution of PMHS in

THF (3.0 mL) gave >99% conversion after 16 h. Hexane−ether (2:1

v/v) was used for isolation by column chromatography. After solvent

removal and drying on a rotary evaporator, 2f was obtained as white

(14)⁹were prepared by the literature procedures. H, ¹⁹F, and ³¹P

1

NMR spectra were recorded on Bruker Avance 400 Ultrashield and

Bruker Avance 500 Ultrashield NMR spectrometers. Single-crystal X-

ray diﬀraction studies were performed on a Bruker-Nonius

diﬀractometer and a Bruker Apex DUO Kappa 4-axis diﬀractometer

equipped with APEX 2 4K CCD area detectors. An Agilent

Technologies 7890A-5975C instrument was used for GC-MS analysis.

FT-IR measurements in solution were performed with a Thermo

Nicolet FT-IR 5700 Nexus spectrometer equipped with a DLaTGS

detector and a KBr beam splitter at 4 cm⁻¹resolution. A Bruker Optics

FTIR Alpha spectrometer with a DTGS detector and a KBr beam

splitter at 4 cm⁻¹resolution was used for FT-IR measurements of solid

samples. Elemental analyses were performed by the Microanalysis

1

crystals. The yield was 1.65 g (87%). H NMR (400 MHz, CDCl₃): δ

2.65 (s, 3H, Me), 8.02 (m, 2H), 8.12 (m, 2H).⁹

Ethyl 4-(Azidocarbonyl)benzoate (2g). Azidocarbonylation of

ethyl 4-iodobenzoate (2.76 g) by the general procedure using a

solution of PMHS in THF (0.5 mL) gave 99% conversion after 16 h.

Hexane−ether (14:1 v/v) was used for isolation by column

chromatography. After solvent removal and drying on a rotary

evaporator, 2g was obtained as white crystals. The yield was 1.89 g

(86%). ¹H NMR (400 MHz, CDCl₃): δ 1.41 (t, J = 7.1 Hz, 3H, CH₃),

4.41 (q, J = 7.1 Hz, 2H, CH₂), 8.10 (m, 4H).⁹

M

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

4-Fluorobenzoyl Azide (2j). Azidocarbonylation of 1-ﬂuoro-4-

iodobenzene (2.22 g) by the general procedure using a solution of

PMHS in THF (0.5 mL) gave >99% conversion after 20 h. Pentane−

ether (9:1 v/v) was used for isolation by column chromatography.

After solvent removal and drying (100 mbar for 1 h) on a rotary

evaporator, 2j was obtained as a colorless oil. The yield was 1.42 g

aqueous phase was washed with CH₂Cl₂(2 × 5 mL). The combined

dichloromethane phase and the washings were dried over Na₂SO₄and

ﬁltered through Celite. The ﬁltrate was evaporated to ca. 5 mL, treated

with Et₂O (25 mL), and kept at −32 °C overnight. The dark

precipitate was separated, washed with Et₂O (2 × 5 mL), and dried

under vacuum. The yield of 7 was 137 mg (97%). This complex is air-

stable in the solid state and in solution. Anal. Calcd for

C₃₉H₃₂I₂OP₂Pd: C, 49.9; H, 3.4. Found: C, 49.7; H, 3.7. For the

1

(86%). H NMR (500 MHz, CDCl₃): δ 7.12 (m, 2H), 8.05 (m, 2H).

¹⁹F{¹H} NMR (377 MHz, CDCl₃): δ −102.7 (s).⁹

1

VT H and ³¹P NMR data for 7, see the Supporting Information. X-

4-Chlorobenzoyl Azide (2k). Azidocarbonylation of 1-chloro-4-

iodobenzene (2.39 g) by the general procedure using a solution of

PMHS in THF (0.5 mL) gave >99% conversion after 20 h. Hexane−

ether (9:1 v/v) was used for isolation by column chromatography.

After solvent removal and drying on a rotary evaporator, 2k was

ray-quality single crystals of cis-7·CH₂Cl₂were obtained by layering its

solution (ca. 2 mg in 0.6 mL of CH₂Cl₂) with Et₂O (1 mL) and

keeping the mixture at −32 °C for 3 days. X-ray-quality single crystals

of trans-7 were obtained by layering its solution (ca. 10 mg in 0.6 mL

of CHCl₃) with Et₂O (1.6 mL) and keeping the mixture at −32 °C for

3 days (see Figure 3).

1

obtained as a white crystalline solid. The yield was 1.60 g (88%). H

NMR (500 MHz, CDCl₃): δ 7.44 (m, 2H), 7.97 (m, 2H).⁹

3-Methoxybenzoyl Azide (2o). Azidocarbonylation of 3-iodoanisole

(2.34 g) by the general procedure using a solution of PMHS in THF

(0.5 mL) gave >99% conversion after 19 h. Hexane−ether (9:1 v/v)

was used for isolation by column chromatography. After solvent

removal and drying on a rotary evaporator, 2o was obtained as a

colorless oil. The yield was 1.57 g (89%). ¹H NMR (500 MHz,

CDCl₃): δ 3.86 (s, 3H, OMe), 7.16 (ddd, J = 8.3, 2.6, and 1.0 Hz, 1H),

7.36 (t, J = 8.0 Hz, 1H), 7.54 (dd, J = 2.6 and 1.6 Hz, 1H), 7.62 (ddd, J

= 7.7, 1.6, and 1.0 Hz, 1H).⁹

Preparation of [(Xantphos)Pd(N₃)₂] (9). Chloroform employed

in this preparation was ﬁltered through a short K₂CO₃plug prior to

use. A mixture of [(Xantphos)PdCl₂]·CH₂Cl₂(8; 84 mg; 0.10 mmol),

NaN₃(130 mg; 2.0 mmol), deionized water (3 mL), and CHCl₃(3

mL) was vigorously stirred in air for 30 min. The orange organic phase

was separated and the aqueous phase was washed with CHCl₃(2 × 3

mL). The combined chloroform phase and the washings were

evaporated with a ﬂow of argon to ca. 2 mL, treated with Et₂O (15

mL), and kept at +4 °C overnight. The orange needle crystals were

separated, washed with Et₂O (2 × 3 mL), and dried under vacuum.

The yield of 9 was 72 mg (94%). This complex is air-stable in the solid

state and in solution. Anal. Calcd for C₃₉H₃₂N₆OP₂Pd: C, 60.9; H, 4.2;

2-Thiophenecarbonyl Azide (2q). Azidocarbonylation of 2-

iodothiophene (2.10 g) by the general procedure using a solution of

PMHS in THF (0.5 mL) gave >99% conversion after 17 h. Pentane−

ether (9:1 v/v) was used for isolation by column chromatography.

After solvent removal and drying (100 mbar for 1 h) on a rotary

evaporator, 2q was obtained as a white solid. The yield was 1.31 g

(86%). ¹H NMR (400 MHz, CDCl₃): δ 7.14 (dd, J = 4.9 and 3.8 Hz,

1H), 7.66 (dd, J = 4.9 and 1.3 Hz, 1H), 7.85 (dd, J = 3.8 and 1.3 Hz,

1H).⁹

1

N, 10.9. Found: C, 60.5; H, 4.3; N, 10.7. H NMR (CDCl₃, 500

MHz): δ (for major isomer) 1.83 (s, 6H, 2CH₃), 7.12 (t, 8H, J = 7.5

Hz), 7.19 (m, 2H), 7.25, (m, 6H), 7.31, (m, 8H), 7.68 (dd, J = 7.8 and

1.1 Hz, 2H). ³¹P NMR (CDCl₃, 203 MHz): δ 7.7 (s, minor isomer,

5%), 21.3 (s, major isomer, 95%). IR (neat, cm⁻¹): 2018 (N₃). X-ray-

quality single crystals of cis-9·CHCl₃were obtained by slow

evaporation of its CHCl₃solution (Figure 4).

Reaction of Pd₂(dba)₅with Xantphos. Inside a glovebox, a 5

mm NMR tube was charged with Pd₂(dba)₅(7 mg; 0.01 mmol of Pd),

Xantphos (6 mg; 0.01 mmol), and benzene-d₆(0.6 mL) and sealed

with a rubber septum. ³¹P NMR analysis of the sample showed

complete conversion of Xantphos and the formation of two new

species, [(Xantphos)Pd(dba)] (11; two doublets at 11.3 and 13.5

ppm, J_P−P= 10.6 Hz) and [(Xantphos)₂Pd] (two broad multiplets at

3.5 and 6.0 ppm) in a 4:1 ratio. A similar ³¹P NMR pattern was

observed for the reaction mixture obtained similarly using Pd₂(dba)₅

(28 mg; 0.04 mmol of Pd) and Xantphos (23 mg; 0.04 mmol) in THF

(2 mL). After 1 week at room temperature, the THF solution

produced X-ray-quality crystals of [(Xantphos)Pd(dba)]·3THF (11·

3THF; see Figure 5).

[(PPh₃)₂Pd(Ph)N₃]. [(PPh₃)₂Pd(Ph)Br] (153 mg; 0.194 mmol)

was added to a suspension of NaN₃(252 mg; 3.9 mmol; 20 equiv) in

MeOH (4 mL), and the mixture was sonicated for 1 h. Methanol (15

mL) was added and the liquid phase decanted oﬀ. To the solid residue

were added NaN₃(252 mg; 3.9 mmol; 20 equiv) and MeOH (4 mL),

and the mixture was sonicated for 1 h. Again, MeOH (15 mL) was

added, the solid was separated by decantation, and thoroughly

extracted with warm (50 °C) benzene (3 × 5 mL). The benzene

extract was ﬁltered through a short Celite plug, concentrated to ca. 5

mL, and treated with hexanes (15 mL). After 12 h, the white

precipitate was separated and dried under vacuum. The yield of

[(PPh₃)₂Pd(Ph)N₃] was 129 mg (89%). This complex is air-stable in

the solid state and in solution. Anal. Calcd for C₄₂H₃₅N₃P₂Pd: C, 67.3;

1

H, 4.7; N, 5.6. Found: C, 67.2; H, 4.7; N, 5.6. H NMR (CDCl₃, 400

Reaction of Pd₂(dba)₅with Xantphos and CO. Inside a

glovebox, a 5 mm NMR tube was charged with Pd₂(dba)₅(7 mg; 0.01

mmol of Pd), Xantphos (6 mg; 0.01 mmol), and benzene-d₆(0.6 mL)

and sealed with a rubber septum. After the argon headspace was

ﬂushed with CO via a syringe needle, full conversion of the originally

produced mixture of [XantphosPd(dba)] (11) and [(Xantphos)₂Pd]

to [(Xantphos)Pd(CO)₂], (12; s, 10.5 ppm) was observed within 10

min (³¹P NMR). Identical results were obtained when THF (0.6 mL)

was used in place of benzene-d₆. Two strong bands at 1974 and 2014

cm⁻¹were observed in the FT-IR spectrum of the solution

(background subtraction).

Reaction of [(Xantphos)Pd(Ph)I] (10) with CO. Inside a

glovebox, a 5 mm NMR tube was charged with [(Xantphos)Pd(Ph)I]

(10; 7 mg; 0.008 mmol) and benzene-d₆(0.5 mL) and sealed with a

rubber septum. After the headspace was ﬂushed with CO (syringe

needle) for ca. 0.5 min, full conversion of 10 (s, 13.2 ppm) to

[(Xantphos)Pd(COPh)I] (13; s, 3.2 ppm) was observed by ³¹P NMR.

Reaction of [(Xantphos)Pd(CO)₂] (12) with PhI. Inside a

glovebox, a 5 mm NMR tube was charged with Pd₂(dba)₅(7 mg; 0.01

mmol of Pd), Xantphos (6 mg; 0.01 mmol), and benzene-d₆(0.6 mL)

and sealed with a rubber septum. After the headspace was ﬂushed with

CO (syringe needle) for ca. 0.5 min, full conversion to [(Xantphos)-

Pd(CO)₂], (12; s, 10.5 ppm) was observed within 10 min (³¹P NMR).

MHz): δ 6.29 (t, 2H, J = 7.4 Hz, C₆H₅Pd), 6.43 (t, 1H, J = 7.2 Hz,

C₆H₅Pd), 6.62 (m, 2H, C₆H₅Pd), 7.24−7.33 (m, 12H, PPh₃), 7.33−

7.44 (m, 18H, PPh₃). ³¹P NMR (CDCl₃, 162 MHz): δ 25.2 (s). IR

(neat, cm⁻¹): 2043 (N₃). X-ray-quality crystals of trans-[(PPh₃)₂Pd-

(Ph)N₃] were obtained by slow diﬀusion of hexane into a

concentrated solution of [(PPh₃)₂Pd(Ph)N₃] in chloroform in a 5

mm NMR tube. The complex crystallized in two polymorphic forms

(white and yellow) displaying either a staggered or eclipsed

conformation along the P−Pd−P axis, as shown in Figures 1 and 2.

Reaction of [(PPh₃)₂Pd(Ph)N₃] with CO, PhI, and PPh₃. Inside

a glovebox, a mixture of [(PPh₃)₂Pd(Ph)N₃] (7 mg; 0.01 mmol), PPh₃

(5 mg; 0.02 mmol), PhI (50 μL), and benzene-d₆(0.6 mL) was placed

in a 5 mm NMR tube and sealed with a rubber septum. After CO was

bubbled through this solution via a syringe needle and the tube was

heated at 50 °C (oil bath) for 30 min, full conversion of

[(PPh₃)₂Pd(Ph)N₃] to [(PPh₃)₂Pd(COPh)I] (³¹P NMR: s, 18.8

ppm; lit.⁶19.0 ppm) and PPh₃NCOPh (³¹P NMR: s, 20.8 ppm;

lit.¹²20.6 ppm) in a 1:1 molar ratio was observed.

Preparation of [(Xantphos)PdI₂] (7). A mixture of [(Xantphos)-

PdCl₂]·CH₂Cl₂(8; 126 mg; 0.15 mmol), NaI (225 mg; 1.50 mmol),

deionized water (2 mL), and CH₂Cl₂(2 mL) was stirred in air for 30

min. The deep purple mixture was diluted with deionized water (5

mL) and CH₂Cl₂(5 mL). The organic phase was separated and the

N

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

Addition of iodobenzene (20 μL) to the thus generated 12 resulted in

selective formation of [(Xantphos)Pd(COPh)I] (13; ³¹P NMR: s, 3.2

ppm) at full conversion.

(neat, cm⁻¹): 2130 (N₃); 1691 (CO). No characteristic band from

NCO was observed (2247 cm⁻¹for an authentic sample of PhNCO).

[Bu₄N]N₃. A 50% aqueous solution of tetra-n-butylammonium

sulfate (5.8 g; 5.0 mmol of [NBu₄]₂SO₄) was concentrated on a rotary

evaporator. After a total of 2.1 g of water had been evaporated, NaN₃

(1.95 g; 30 mmol) was added and the resultant mixture was agitated

for 20 h. Acetonitrile (30 mL) and anhydrous sodium sulfate (10 g)

were added, and the mixture was stirred for 30 min and ﬁltered. The

solid on the ﬁlter was washed with acetonitrile (2 × 10 mL). The

combined ﬁltrate and the washings were evaporated on a rotary

evaporator. Drying the residue under vacuum (ca. 1 mmHg) overnight

produced a white solid (2.88 g). The thus obtained crude [Bu₄N]N₃

was brought to a glovebox and dissolved in dry benzene (20 mL). The

resultant solution was ﬁltered through a pad of Celite. The reaction

ﬂask was rinsed with dry benzene (2 × 10 mL) and the washings were

ﬁltered through the same pad. The combined ﬁltrate and the washings

were evaporated and dried under vacuum at ca. 1 mmHg for 20 h.

Trituration of the residue with dry ether (40 mL) produced a white

crystalline solid that was separated and dried under vacuum (ca. 1

mmHg) for 20 h. The yield of [Bu₄N]N₃was 2.79 g (98%). This salt is

hygroscopic and was therefore stored and used in a glovebox. ¹H

NMR (500 MHz, CDCl₃): δ 1.00 (t, J = 7.4 Hz, 3H), 1.45 (tq, J = 7.4

and 7.4 Hz, 2H), 1.68 (m, 2H), 3.35 (m, 2H). IR (neat, cm⁻¹): 1991

Reaction of [(Xantphos)Pd(COPh)I] (13) with [Bu₄N]N₃. Inside

a glovebox, a 5 mm NMR tube was charged with [(Xantphos)Pd-

(Ph)I] (10; 7 mg; 0.008 mmol) and benzene-d₆(0.5 mL) and sealed

with a rubber septum. Replacing the argon headspace with CO

(syringe needle) resulted in an instantaneous reaction leading to full

conversion (³¹P NMR) of 10 (s, 13.2 ppm²⁹) to [(Xantphos)Pd-

(COPh)I] (13; s, 3.2 ppm). A solution of [Bu₄N]N₃(10 mg; 0.035

mmol) in benzene-d₆(0.2 mL) was added via syringe. The ³¹P NMR

spectrum recorded within 10 min indicated full conversion of 13 to

[(Xantphos)Pd(CO)₂] (12; s, 10.5 ppm). As the resultant solution of

12 and PhCON₃was highly air-sensitive, the formation of the latter by

IR was conﬁrmed in a similar experiment performed in the presence of

PhI (see below).

Reaction of [(Xantphos)Pd(COPh)I] (13) with [Bu₄N]N₃in the

Presence of PhI. Inside a glovebox, a 5 mm NMR tube was charged

with Xantphos (6 mg; 0.01 mmol), Pd₂(dba)₅(7 mg; 0.01 mmol of

Pd), and benzene-d₆(0.5 mL) and sealed with a rubber septum. The

headspace was ﬂushed with CO (syringe needle) for ca. 0.5 min. The

addition of iodobenzene (20 μL) to the thus generated [(Xantphos)-

Pd(CO)₂] (12; see above) resulted in the selective formation of

[(Xantphos)Pd(COPh)I] (13; ³¹P NMR: s, 3.2 ppm) at full

conversion. To this sample was added via syringe 0.2 mL (1.2

equiv) of a solution of [Bu₄N]N₃(34 mg) in C₆H₆(2 mL). The FT-IR

spectrum registered 10 min after the addition of azide indicated the

formation of PhCON₃(2a; neat, cm⁻¹): 2133 (N₃); 1695 (CO) and

no presence of PhNCO (2247 cm⁻¹). FT-IR for an authentic sample

of 2a (cm⁻¹): 2130 (N₃); 1691 (CO).

−

(N₃). X-ray-quality single crystals of [Bu₄N]N₃·C₆H₆were obtained

from a concentrated benzene solution on standing at room

temperature (Figure 13).

Reaction of [(Xantphos)Pd(Ph)I] (10) with [Bu₄N]N₃. Inside a

glovebox, a vial was charged with Xantphos (29 mg; 0.05 mmol),

Pd₂(dba)₅(35 mg; 0.05 mmol of Pd), PPh₃O (14 mg; 0.05 mmol;

internal standard), and benzene (5 mL). After this reaction mixture

was stirred for 1 min, PhI (100 μL) was added. ³¹P NMR analysis

indicated full conversion of [(Xantphos)Pd(dba)] (11) to

[(Xantphos)Pd(Ph)I] (10; s, 13.2 ppm) after 30 min. [Bu₄N]N₃

(16 mg; 0.06 mmol) was then added, and the solution was again

analyzed by ³¹P NMR after 10 min to observe full conversion to

[(Xantphos)Pd(Ph)N₃] (14; s, 11.0 ppm).

Figure 13. ORTEP drawing of [Bu₄N]N₃·C₆H₆with benzene and all

H atoms omitted and thermal ellipsoids drawn at the 50% probability

level.

Reaction of [(Xantphos)Pd(Ph)I] (10) with NaN₃. Inside a

glovebox, a vial was charged with Xantphos (29 mg; 0.05 mmol),

Pd₂(dba)₅(35 mg; 0.05 mmol of Pd), PPh₃O (14 mg; 0.05 mmol;

internal standard), and benzene (5 mL). After the mixture was stirred

for 1 min, PhI (100 μL) was added. ³¹P NMR analysis after 30 min

indicated full conversion of [(Xantphos)Pd(dba)] (11) to

[(Xantphos)Pd(Ph)I] (10; s, 13.2 ppm). Argon-saturated water (5

mL) and sodium azide (32 mg; 0.5 mmol) were added. After 5 min of

vigorous agitation, a 0.2 mL aliquot of the organic layer was diluted

with C₆H₆(1 mL) and analyzed by ³¹P NMR (IGD) to show ca. 60%

conversion to [(Xantphos)Pd(Ph)N₃] (14; s, 11.0 ppm). No change

in the composition of the organic layer was observed (³¹P NMR) after

stirring the biphasic system for an additional 1 h.

Reaction of [(Xantphos)Pd(Ph)N₃] (14) with CO. Inside a

glovebox, a 5 mm NMR tube was charged with 14 (7 mg; 0.01 mmol)

and toluene (0.6 mL) and sealed with a rubber septum. After the

mixture was placed under CO by purging via a syringe needle for ca.

0.5 min, full conversion of 14 to [(Xantphos)Pd(CO)₂] (12; s, 10.5

ppm) was observed within 10 min (³¹P NMR). Layering the resultant

solution with hexanes under CO produced X-ray-quality crystals of

[(Xantphos)Pd(CO)₂]·n-C₆H₁₄(12·hexane; see Figure 6).

Computational Details. Calculations were run with Gaussian 03

Revision D.01⁵⁸with PCM solvent corrections run with Gaussian 09

Revision A.02.⁵⁹Geometry optimizations were performed using the

BP86 functional⁶⁰and employed a smaller basis set, BS1, in which the

Pd, P, and I centers were described with the Stuttgart RECPs and

associated basis sets⁶¹(with added d-orbital polarization on P (ζ =

0.387) and I (ζ = 0.289)⁶²) and 6-31G** basis sets for all other atoms.

All stationary points were fully characterized via analytical frequency

calculations as either minima (all positive eigenvalues) or transition

states (one negative eigenvalue) and IRC calculations, and subsequent

geometry optimizations were used to conﬁrm the minima linked by

each transition state. Frequency calculations also provided a free

energy in the gas phase, computed at 298.15 K and 1 atm. As has been

found by others,⁶³reasonable barriers, in particular for the initial

oxidative addition steps, required a correction for dispersion eﬀects to

be included and this was obtained via a single-point energy calculation

on the BP86(BS1)-optimized geometries using Grimme’s D3

parameter set⁶⁴(i.e., BP86-D3). Moreover, the energetics of I⁻/N₃

−

exchange were found to be highly sensitive to the choice of basis set,

this process being unrealistically exergonic with BS1 that was used in

the geometry optimization. All energies were therefore recomputed

with a larger basis set, BS2, featuring aug-CC-pVTZ on Pd, P, and I

and 6-311++G** on all other atoms. Further details on this basis set

dependency are given in Table S5 in the Supporting Information for

Reaction of [(Xantphos)Pd(Ph)N₃] (14) with CO in the

Presence of PhI. Inside a glovebox, a 5 mm NMR tube was charged

with 14 (7 mg; 0.01 mmol), PhI (20 μL), and benzene-d₆(0.6 mL)

and sealed with a rubber septum. After the reaction mixture was placed

under a CO atmosphere by purging via a syringe needle for ca. 0.5

min, full conversion of 14 to [(Xantphos)Pd(COPh)I] (13; ³¹P NMR

s, 3.2 ppm) was observed within 10 min by ³¹P NMR. The formation

of PhCON₃(2a) was conﬁrmed by FT-IR (neat, cm⁻¹): 2133 (N₃);

1694 (CO). FT-IR for independently prepared and isolated 2a

−

I⁻/N₃exchange at trans-10 and cis-13. Starting geometries for the

various minima were taken from experimental crystallographic studies

(e.g., for 11, 12, cis-13, trans-14) reported here, as well as for trans-

O

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

10²⁹or by adapting these structures (e.g., via replacement of I⁻by

N₃). Computed geometries reproduced the available experimental

(d) Lee, K.-E.; Jeon, H.-T.; Han, S.-Y.; Ham, J.; Kim, Y.-J.; Lee, S. W.

Dalton Trans. 2009, 6578.

−

data well (see Table S4 in the Supporting Information). A number of

diﬀerent conformations were tested for each stationary point, and in

particular the transition states TS_RC, TS_MI(I), TS_OA1, and TS_OA2were

subject to extensive conformational searching using our published

protocol.⁶⁵

(12) Albright, T. A.; Freeman, W. J.; Schweizer, E. E. J. Org. Chem.

1976, 41, 2716.

(13) (a) The empirical composition Pd₂(dba)₅was derived from

elemental analysis data.^13b(b) Ushkov, A. V.; Grushin, V. V. J. Am.

Chem. Soc. 2011, 133, 10999.

(14) For selected reviews on Xantphos and related ligands, see:

(a) Kamer, P. C. J.; Van Leeuwen, P. W. N. M.; Reek, J. N. H. Acc.

Chem. Res. 2001, 34, 895. (b) Freixa, Z.; Van Leeuwen, P. W. N. M.

Dalton Trans. 2003, 1890. (c) Freixa, Z.; Van Leeuwen, P. W. N. M.

Coord. Chem. Rev. 2008, 252, 1755. (d) Van Leeuwen, P. W. N. M.;

Freixa, Z. In Modern Carbonylation Methods; Kollar, L., Ed.; Wiley-

VCH: Weinheim, Germany, 2008; p 1.

(15) For eﬃcient catalysis of other carbonylation reactions with

Xantphos/Pd, see: (a) Martinelli, J. R.; Watson, D. A.; Freckmann, D.

M. M.; Barder, T. E.; Buchwald, S. L. J. Org. Chem. 2008, 73, 7102.

(b) Ueda, T.; Konishi, H.; Manabe, K. Org. Lett. 2013, 15, 5370.

(16) In the ligand screening experiments, GC-MS was used for

analysis, which could not allow for direct detection of PhCON₃

because of its instantaneous Curtius rearrangement in the GC injector.

The conversions and yields were therefore derived from the quantities

of PhNCO observed.

ASSOCIATED CONTENT

* Supporting Information

Text, tables, ﬁgures, and CIF ﬁles giving full details of

experimental, computational, and crystallographic studies,

including full refs 58 and 59. This material is available free of

charge via the Internet at http://pubs.acs.org.

■

S

AUTHOR INFORMATION

Corresponding Authors

*E-mail for S.A.M.: s.a.macgregor@hw.ac.uk.

*E-mail for V.V.G.: vgrushin@iciq.es.

Notes

■

The authors declare no competing ﬁnancial interest.

(17) Klingensmith, L. M.; Strieter, E. R.; Barder, T. E.; Buchwald, S.

L. Organometallics 2006, 25, 82.

ACKNOWLEDGMENTS

■

(18) (a) Shen, Q.; Ogata, T.; Hartwig, J. F. J. Am. Chem. Soc. 2008,

130, 6586. (b) Fors, B. P.; Davis, N. R.; Buchwald, S. L. J. Am. Chem.

Soc. 2009, 131, 5766.

(19) It was established in a separate experiment that NaN₃in excess

does not deactivate the catalyst.

The ICIQ Foundation, Consolider Ingenio 2010 (Grant

CSD2006-0003), and The Spanish Government (Grant

CTQ2011-25418) are thankfully acknowledged for support of

this work. S.A.M. and C.L.M. thank the EPSRC for support

through award EP/J021911/1.

(20) Saikachi, H.; Kitagawa, T.; Nasu, A.; Sasaki, H. Chem. Pharm.

Bull. 1981, 29, 237.

REFERENCES

(21) (a) Bertani, R.; Berton, A.; Di Bianca, F.; Crociani, B. J.

Organomet. Chem. 1986, 303, 283. (b) Isobe, K.; Nanjo, K.; Nakamura,

Y.; Kawaguchi, S. Bull. Chem. Soc. Jpn. 1986, 59, 2141.

(22) Yukawa, Y.; Tsuno, Y. J. Am. Chem. Soc. 1958, 80, 6346.

(23) Wu, X.-F.; Neumann, H.; Beller, M. Chem. Eur. J. 2010, 16,

9750.

■

(1) (a) Schoenberg, A.; Bartoletti, I.; Heck, R. F. J. Org. Chem. 1974,

39, 3318. (b) Schoenberg, A.; Heck, R. F. J. Org. Chem. 1974, 39, 3327.

(c) Schoenberg, A.; Heck, R. F. J. Am. Chem. Soc. 1974, 96, 7761.

(2) For recent reviews on Pd-catalyzed carbonylation reactions of aryl

halides, see: (a) Wu, X.-F.; Neumann, H.; Beller, M. Chem. Rev. 2013,

113, 1. (b) Roy, S.; Roy, S.; Gribble, G. W. Tetrahedron 2012, 68,

9867. (c) Wu, X.-F.; Neumann, H.; Beller, M. Chem. Soc. Rev. 2011,

40, 4986. (d) Magano, J.; Dunetz, J. R. Chem. Rev. 2011, 111, 2177.

(e) Grigg, R.; Mutton, S. P. Tetrahedron 2010, 66, 5515.

(24) Ulrich, H. Chemistry and Technology of Isocyanates; Wiley:

Chichester, U.K., 1997.

(25) Tkatchenko, I.; Jaouhari, R.; Bonnet, M.; Dawkins, G.; Lecolier,

S. U.S. Patent 4749806, 1987.

(26) (a) Vinogradova, E. V.; Fors, B. P.; Buchwald, S. L. J. Am. Chem.

Soc. 2012, 134, 11132. (b) Vinogradova, E. V.; Park, N. H.; Fors, B. P.;

Buchwald, S. L. Org. Lett. 2013, 15, 1394.

(f) Brennfuhrer, A.; Neumann, H.; Beller, M. Angew. Chem., Int. Ed.

̈

2009, 48, 4114. (g) Barnard, C. F. J. Organometallics 2008, 27, 5402.

(3) For general reviews, see: (a) Scriven, E. F. V.; Turnbull, K. Chem.

(27) Erhardt, S.; Grushin, V. V.; Kilpatrick, A. H.; Macgregor, S. A.;

Marshall, W. J.; Roe, D. C. J. Am. Chem. Soc. 2008, 130, 4828.

(28) Under identical conditions (10 mmol of PhI in 0.5 mL of THF)

with 7 (0.2 mol %) as added catalyst at room temperature, the reaction

stopped after 2 h at ca. 20% conversion.

Rev. 1988, 88, 297. (b) Brase, S.; Gil, C.; Knepper, K.; Zimmermann,

̈

V. Angew. Chem., Int. Ed. 2005, 44, 5188. (c) Brase, S.; Banert, K.

̈

Organic Azides: Syntheses and Applications; Wiley: Chichester, U.K.,

2010.

́

(4) Cano, I.; Alvarez, E.; Nicasio, M. C.; Perez, P. J. J. Am. Chem. Soc.

2011, 133, 191.

(29) Grushin, V. V.; Marshall, W. J. J. Am. Chem. Soc. 2006, 128,

12644.

(5) Katritzky, A. R.; Widyan, K.; Kirichenko, K. J. Org. Chem. 2007,

72, 5802.

(6) Garrou, P. E.; Heck, R. F. J. Am. Chem. Soc. 1976, 98, 4115.

(7) Kim, Y.-J.; Kwak, Y.-S.; Lee, S.-W. J. Organomet. Chem. 2000, 603,

152.

́ ́

(30) Bakhmutov, V. I.; Bozoglian, F.; Gomez, K.; Gonzalez, G.;

Grushin, V. V.; Macgregor, S. A.; Martin, E.; Miloserdov, F. M.;

Novikov, M. A.; Panetier, J. A.; Romashov, L. V. Organometallics 2012,

31, 1315.

(31) See, for example: (a) Redfield, D. A.; Nelson, J. H. Inorg. Chem.

1973, 12, 15. (b) Anderson, G. K.; Cross, R. J. Chem. Soc. Rev. 1980, 9,

185.

(32) Grushin, V. V. Chem. Rev. 2004, 104, 1629.

(33) In the presence of larger quantities of PMHS (2 equiv) at 50 °C,

the aroyl azide product undergoes reduction to the corresponding

benzamide (see above).

(8) For a recent review of the Staudinger reaction, see: Iula, D. M. In

Name Reactions for Functional Group Transformations; Li, J. J., Corey, E.

J., Eds.; Wiley: Hoboken, NJ, 2007; p 129.

(9) Miloserdov, F. M.; Grushin, V. V. Angew. Chem., Int. Ed. 2012, 51,

3668.

́ ́

(10) Barluenga, J.; Valdes, C.; Beltran, G.; Escribano, M.; Aznar, F.

Angew. Chem., Int. Ed. 2006, 45, 6893.

(34) Ozawa, F.; Soyama, H.; Yanagihara, H.; Aoyama, I.; Takino, H.;

Izawa, K.; Yamamoto, T.; Yamamoto, A. J. Am. Chem. Soc. 1985, 107,

3235.

(11) For selected examples of the facile formation of various

organopalladium azido complexes from the corresponding Pd(II)

halides and NaN₃, see: (a) Kim, Y.-J.; Choi, J.-C.; Park, K.-H. Bull.

Korean Chem. Soc. 1994, 15, 690. (b) Kim, Y.-J.; Kim, D.-H.; Song, S.-

W.; Son, T.-i. Bull. Korean Chem. Soc. 1998, 19, 125. (c) Kim, Y.-J.;

Chang, X.; Han, J.-T.; Lim, M. S.; Lee, S. W. Dalton Trans. 2004, 3699.

(35) (a) Magerlein, W.; Beller, M.; Indolese, A. F. J. Mol. Catal. A:

̈

Chem. 2000, 156, 213. (b) Gavino, R.; Pellegrini, S.; Castanet, Y.;

̃

́

Mortreux, A.; Mentre, O. Appl. Catal., A 2001, 217, 91.

P

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Article

(36) See footnote 77 in Kolb, H. C.; Finn, M. G.; Sharpless, K. B.

Angew. Chem., Int. Ed. 2001, 40, 2004.

(37) Gonzalez-Bobes, F.; Kopp, N.; Li, L.; Deerberg, J.; Sharma, P.;

́

Leung, S.; Davies, M.; Bush, J.; Hamm, J.; Hrytsak, M. Org. Process Res.

Dev. 2012, 16, 2051 and references cited therein.

(55) For recent examples of Pd-catalyzed carbonylation reactions

with ¹¹CO, see: (a) Kealey, S.; Plisson, C.; Collier, T. L.; Long, N. J.;

Husbands, S. M.; Martarello, L.; Gee, A. D. Org. Biomol. Chem. 2011,

9, 3313. (b) Eriksson, J.; van den Hoek, J.; Windhorst, A. D. J. Labelled

Compd. Radiopharm. 2012, 55, 223. (c) Takashima-Hirano, M.; Ishii,

H.; Suzuki, M. ACS Med. Chem. Lett. 2012, 3, 804. (d) Dahl, K.;

Schou, M.; Amini, N.; Halldin, C. Eur. J. Org. Chem. 2013, 1228.

(56) Grushin, V. V. Organometallics 2000, 19, 1888.

(57) Johns, A. M.; Utsunomiya, M.; Incarvito, C. D.; Hartwig, J. F. J.

Am. Chem. Soc. 2006, 128, 1828.

(58) Frisch, M. J., et al. Gaussian 03, Revision D.01; Gaussian Inc.,

Wallingford, CT, 2004.

(59) Frisch, M. J., et al. Gaussian 09, Revision A.02; Gaussian Inc.,

Wallingford, CT, 2009.

(60) (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098. (b) Perdew, J. P.

Phys. Rev. B 1986, 33, 8822.

(38) (a) Beck, W.; Fehlhammer, W. P.; Pollmann, P.; Schuierer, E.;

̈

Feldl, K. Chem. Ber. 1967, 100, 2335. (b) Beck, W.; Klapotke, T. M.;

̈

Knizek, J.; Noth, H.; Schutt, T. Eur. J. Inorg. Chem. 1999, 523.

̈

(39) Reported mechanistic information on oxidative addition of

organic halides to mixed phosphino carbonyl Pd(0) species is

extremely limited. For selected examples, see: (a) Kudo, K.; Sato,

M.; Hidai, M.; Uchida, Y. Bull. Chem. Soc. Jpn. 1973, 46, 2820.

(b) Stille, J. K.; Lau, K. S. Y. J. Am. Chem. Soc. 1976, 98, 5841. (c) Lau,

K. S. Y.; Wong, P. K.; Stille, J. K. J. Am. Chem. Soc. 1976, 98, 5832.

(40) Braterman, P. S. Metal Carbonyl Spectra; Academic Press:

London, 1975.

(61) Andrae, D.; Haußermann, U.; Dolg, M.; Stoll, H.; Preuß, H.

̈

(41) Gjaldbæk, J. Chr. Acta Chem. Scand. 1952, 6, 623.

(42) X-ray-quality crystals of tetrabutylammonium azide formed from

its benzene solution on standing. One of them was used for the crystal

structure determination of [Bu₄N]⁺N₃⁻(see the Experimental Section

and the Supporting Information for details).

(43) Fujita, K.; Yamashita, M.; Puschmann, F.; Martinez Alvarez-

Falcon, M.; Incarvito, C. D.; Hartwig, J. F. J. Am. Chem. Soc. 2006, 128,

9044.

Theor. Chim. Acta 1990, 77, 123.

(62) Hollwarth, A.; Bohme, M.; Dapprich, S.; Ehlers, A. W.; Gobbi,

̈

A.; Jonas, V.; Kohler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G.

̈

Chem. Phys. Lett. 1993, 208, 237.

(63) McMullin, C. L.; Jover, J.; Harvey, J. N.; Fey, N. Dalton Trans.

2010, 39, 10833.

(64) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys.

2010, 132, 154104.

(44) (a) Grushin, V. V. Angew. Chem., Int. Ed. 1998, 37, 994.

(b) Flemming, J. P.; Pilon, M. C.; Borbulevitch, O. Ya.; Antipin, M.

Yu.; Grushin, V. V. Inorg. Chim. Acta 1998, 280, 87.

(65) Haller, L. J. L.; Page, M. J.; Erhardt, S.; Macgregor, S. A.;

̈

Mahon, M. F.; Naser, M. A.; Velez, A.; Whittlesey, M. K. J. Am. Chem.

Soc. 2010, 132, 18408.

(45) Coyle, R. J.; Slovokhotov, Yu. L.; Antipin, M. Yu.; Grushin, V. V.

Polyhedron 1998, 17, 3059.

(46) For previous computational studies of Ph−X activation (X = Cl,

Br) at [Pd(Xantphos)], see: Hicks, J. D.; Hyde, A. M.; Cuezva, A. M.;

Buchwald, S. L. J. Am. Chem. Soc. 2009, 131, 16720. Structure 2a

reported in that work is related to I3 located here, although in that

case no Pd−O interaction was apparent and a different orientation of

the Ph ligand was seen.

(47) Alternative isomers of I4(I) are possible, e.g., with CO trans to

phosphine and Ph trans to I. However, migratory insertion has been

shown to be facilitated when the transferring group (here Ph) is trans

to a highest trans inﬂuence coligand, i.e., as in I4(I), in which Ph is

trans to phosphine. For a related example, see: Macgregor, S. A.;

Neave, G. W. Organometallics 2004, 23, 891.

−

(48) We have not attempted to locate transition states for I⁻/N₃

exchange but have assumed that this readily occurs under the reaction

conditions, as was shown to be the case experimentally for the

conversion of 10 to 14.

(49) We also attempted to locate a pathway for C−N bond

formation via the direct nucleophilic attack of free azide at the benzoyl

ligand in 15. Several linear transits were run to probe this possibility,

with optimizations being run with the water solvent being included in

the SCF procedure in order to treat the free anionic nucleophile.

However, in all cases the approach of azide led to facile formation of a

Pd−N₃bond (with displacement of iodide or one arm of the Xantphos

ligand) prior to C−N bond formation.

(50) (a) Fitton, P.; McKeon, J. E. Chem. Commun. 1968, 4.

(b) Fitton, P.; Johnson, M. P.; McKeon, J. E. Chem. Commun. 1968, 6.

(c) Fitton, P.; Rick, E. A. J. Organomet. Chem. 1971, 28, 287.

(51) Hartwig, J. Organotransition Metal Chemistry: From Bonding to

Catalysis; University Science Books: Sausalito, CA, 2010.

(52) As described above, upon addition of 1 equiv of [Bu₄N]N₃to

[(Xantphos)Pd(Ph)(I)] (10) in benzene [(Xantphos)Pd(Ph)(N₃)]

(14) is formed quantitatively (³¹P NMR). This, translates into 14:10 >

100, considering the sensitivity of the method (ca. 1%).

(53) Huser, M.; Youinou, M.-T.; Osborn, J. A. Angew. Chem., Int. Ed.

Engl. 1989, 28, 1386.

(54) For a general review of ¹¹C labeling, see: Antoni, G.; Kihlberg,

T.; Langstrom, B. In Handbook of Nuclear Chemistry, 2nd ed.; Ver

́

tes,

̊

̈

A., Nagy, S., Klencsar

Dordrecht, The Netherlands, 2011; Vol. 4, p 1977.

́

, Z., Lovas, R. G., Rosch, F., Eds.; Springer:

̈

Q

dx.doi.org/10.1021/om401126m | Organometallics XXXX, XXX, XXX−XXX

Article Doi

The challenge of palladium-catalyzed aromatic azidocarbonylation: From mechanistic and catalyst deactivation studies to a highly efficient process

DOI: 10.1021/om401126m

Source and publish data:

Authors:

Article abstract of DOI:10.1021/om401126m

Full text of DOI:10.1021/om401126m

Products guided by the article

R&D Labs maybe for 1534392-21-6

Relevant to this article

Hot Product