Iron-Catalyzed Arylation of Heterocycles and Transition-Metal Free C-N Bond Formation

The formation of carbon-carbon (C-C) and carbon-nitrogen (C-N) bonds is discussed and efforts towards expanding the known reactions of this type are the primary focus of this work. The iron-catalyzed arylation of aromatic heterocycles, such as pyridines, thiophenes and furans has been achieved. The use of an imine directing group allowed for the ortho-functionalization of these heterocycles with complete conversion in 15 minutes at 0 °C. Yields up to 88% were observed in the synthesis of 15 heterocyclic biaryls. C-N bond formation is achieved using aryl Grignard reagents and N-chloroamines at -78 °C.


vi LIST OF TABLES
Sustainable chemistry might seem like a contradictory term; however this is the only way we can continue to provide the ability to generate the compounds we need as a society on a large scale without damaging our environment, diminishing our resources, or compromising the health of future generations. Consequently, green chemistry is a growing field that has quickly generated a lot of interest. The applications of these practices to synthetic chemistry are highly desirable.
Many physical transformations exist for the modern synthetic chemist. In particular, the formation of new carbon-carbon (C-C) bonds is quite literally the backbone of organic chemistry. Developing and understanding these C-C forming reactions has always been at the forefront of chemical synthesis. Some traditional methods to create these bonds include Grignard reagents 1 , organocuprates 2 , organolithium reagents 3 , and Wittig type reactions. 4 More recently over the past few decades several metal-catalyzed crosscoupling methods have been developed 5 ; including the Suzuki-Miyaura, Stille, Heck, Negishi, Kumada, and Sonogashira coupling reactions (Scheme 1.1). These reactions have shown great promise for expanding the organic chemists' tool box and providing robust methods to obtain a variety of products.
One major downfall of these methods is the need to pre-functionalize the starting compounds in order to obtain both the substituted halides and organometallic reagents required for each reaction to occur. The generation of hazardous waste has always been a concern to the environmentally conscious chemist. These extra steps inherently generate more waste, result in a lower atom economy, and consume more resources than would be necessary if the product could be synthesized in a more direct way.

Scheme 1.1. Modern Cross-Coupling Reactions
This is the driving concept behind carbon-hydrogen (C-H) bond activation. A C-H bond can be directly substituted to form C-C and C-heteroatom bonds without prior functionalization at the reacting carbons or heterocyclic centers. Although sometimes directing groups (such as imines, carbonyls, or carboxylic acids) are required 6 , usually these can still be used for further substitution, or in some cases easily removed.
The activation of C-H bonds provides one excellent remedy to the aforementioned environmental problem; but additional efforts can be made to increase the overall "greenness" of a reaction. The first and easiest alternative is to use "green solvents" such as water, 2-methyltetrahydrofuran, or cylcohexane instead of more undesirable solvents like dichloromethane, benzene, and hexanes. The real challenge lies in variations of the reagents themselves and their role in the reaction mechanism (catalytic vs. stoichiometric). The majority of research in the field of C-H activation has focused on catalysts with relatively easy to understand mechanisms like palladium. More recently work has revolved around the use of cheap, non-toxic transition-metal catalysts to promote these same desired reactions through "green" methodology.
Our primary interest was to work exclusively with iron catalysts to develop novel methods for C-C and C-N bond formation through the C-H activation pathway in particular. We chose to work with iron due to its low toxicity and availability. There was also precedent set by Nakamura that demonstrated the ability of iron to activate the C-H bond. 7 Nakamura's pioneering work involved the nucleophilic displacement of a hydrogen atom ortho-to a nitrogen-containing directing group (Scheme 1.2)

Scheme 1.2. First Example of C-H Activation via Iron
The reaction in Scheme 1.2a shows a hindered system that is sterically fixed for the C-H activation of only one carbon in α-benzoquinoline. The reaction of various substituted substrates gives insight to the limitations of this system. When these reaction conditions are applied to a substrate with two equivalent hydrogens, like in the case of 2-phenyl pyridine, then a mixture of products is obtained (Scheme 1.2b).
Increasing steric hindrance at the ortho-position on the phenyl ring decreases the product yield (Scheme 1.2c and 1.2d). Interestingly, when a methyl substituent is added to the 3-position on the phenyl group the mono-arylated product is formed exclusively (Scheme 1.2e). These results can be attributed to steric hindrance. Scheme 1.3 shows that Grignard reagents bearing both electron donating and electron withdrawing substituents are readily coupled with this methylated 2phenylpyridine in good yields over 36 hours. The exception to these is the ortho-tolyl Grignard reagent which afforded no product, reinforcing the important role of steric restraints on this catalytic system.

Scheme 1.3. Scope of Grignard arylation on 2-phenylpyridine
These initial findings showed that iron could be used for C-H activation and provide arylated products under mild conditions, although long reaction times and large equivalents of Grignard reagents are necessary. The scope of the reaction shows the tolerance of electronic effects on the Grignard reagent and the influence of the steric environment around the hydrogen leaving group. Both electron donating, t-butyl and methoxy substituents, and electron withdrawing fluorine substituents on the Grignard reagent are tolerated. These are key concepts that must be considered when attempting to develop more iron-catalyzed reactions.
The ability to use an imine as a directing group instead of the pyridine ring allows for subsequent hydrolysis, and yields the product with the carbonyl suitable for further functionalization 8 (Scheme 1.4).

Scheme 1.4. Directed C-H activation
However, the use of an organozinc reagent to generate the active transmetalating complex is still necessary. This system is not limited to aromatic hydrogens, and has been shown to effectively arylate sp 2 -hybridized olefins 9 and sp 3hybridized carbons. 10 Carbon-nitrogen (C-N) bond formation is possible if primary amines are converted into an organozinc species. 11 The Grignard reagent can also be generated in situ, with excellent yields. 12 Eventually it was discovered that the reaction can proceed without the use of organozinc reagents 13 and it has also been shown that oxygen can be used as an oxidant. 14 The decrease in metals required results in a higher atom economy; and the use of oxygen is considered environmentally friendly. Further manipulation of the directing group allows for an increase in the product scope of these reactions up to a gram scale 15, 16 (Scheme 1.5).

Scheme 1.5. Gram Scale Reaction
Another example of C-N bond formation is possible using this quinoline directing group. 17 The incorporation of deuterium was demonstrated, which suggests the oxidative addition of the iron species into the original C-H bond of interest (Scheme 1.6).

Scheme 1.6. Carbon-Nitrogen Bond Formation via C-H activation
The variations of this initial system have been significantly manipulated and expanded to allow for substantial coupling reactions to occur predictabley and reliably. 18 The result is a process that affords aryl-aryl, alkenyl-alkenyl, alkenyl-aryl, and (hetero)aryl-aryl coupling. A final testament to the future of iron-catalysis and its abilities is the iron-catalyzed Suzuki-Miyaura coupling of alkyl halides with aryl boronic esters. 19 Using this scaffold, we applied similar conditions and continued to employ the imine directing group to successfully perform C-H activations resulting in the substitution of a variety of heterocycles using an iron catalyst. The focus on heterocycles is imperative and highly desirable because almost all biologically active compounds and pharmaceuticals produced on an industrial scale contain some type of heteroatom with aromatic substituents. The scope and limitations of this reaction are discussed further in Manuscript 1.
New C-N bond formation reactions are in high demand due to the prevalence of these bonds in almost all pharmaceutical products. These bonds are traditionally formed through Buchwald-Hartwig type reactions 20, 21 (Scheme 1.7).

Scheme 1.7. Buchwald-Hartwig Reaction
As previously mentioned, the goal of the work presented herein is to eliminate the dependence on expensive metals, especially palladium catalysts. Variants of the Buchwald-Hartwig reaction are still being investigated, and palladium still plays a dominant role in these reactions. 22 Fortunately, we are not alone in trying to move away from this reagent. Recent work by several groups shows a promising future for C-N bond formation using less toxic transition metals such as copper and cobalt.
The most common methods for C-N bond formation through C-H activation pathways are only accessible through an intramolecular pathway performed with palladium or copper catalysts. 23, 24 An alternative method to these types of reactions was proposed by Glorius and coworkers which described a C-H activation mechanism using an N-pivaloyloxy amide directing group with chloroamines 25 (Scheme 1.8).

Scheme 1.8. C-N Bond Formation via C-H Activation
When C-H bond activation is not possible, the use of prefunctionalized compounds is often necessary to obtain the desired products required. N-chloroamines are commonly used in association with aryl organometallic reagents. In some cases stoichiometric zinc is used with catalytic amounts of cobalt or copper. 26 Generation of an aryl-zinc species through cobalt-catalyzed arylation has been shown to produce the desired aryl-amine product in high yields with good functional group tolerance. The weakness of the nitrogen-oxygen bond has been exploited using O-acyl hydroxylamine derivatives as leaving groups in the copper-catalyzed electrophilic amination of diorganozinc reagents to form substituted amines 27 (Scheme 1.10a).
Alkyl-alkyl bond formation is achieved using copper catalysts with these functionalized amines through a hydroamination mechanism 28 (Scheme 1.10b) and direct amination is possible on highly electron deficient arenes and azoles 29 (Scheme 1.10c).
One of the more interesting amination reactions is a transition-metal free electrophilic amination using aryl-Grignard reagents and N-chloroamines. This is achieved using TMEDA as an additive although high equivalents are needed to drive the reaction to completion 30 (Scheme 1.11).

Scheme 1.11. Transition-Metal Free C-N Bond Formation
Our attempts to repeat and expand on the above method began with experiments using iron-catalysts. Preliminary studies showed that the reaction could proceed at room temperature, resulting in low yields, without any catalyst using Nchloroamines and Griganard reagents. We found that the reaction seemed to work much better at 0 °C in the presence of several iron salts. High throughput screening of additives and both dinitrogen and diphosphine ligands at 0 °C resulted in a moderate increase in yield. We maintained this catalytic system and decided to lower the temperature to prevent unwanted side products. We were initially delighted to discover that a decrease in temperature to -78 °C resulted in an excellent 88% yield in just 5 minutes.
Interestingly, when the reaction was held at -78 °C, and the iron was removed for the control reaction, in the presence of dinitrogen ligands we still produced the product in comparably high yields. Our next reaction involved eliminating the ligands.
Surprisingly, this provided the desired product in quantitative yield (99%). We had discovered that the reaction proceeds smoothly and quickly at this specific temperature. While initially dismayed at the lack of a need for the iron-catalyst, we quickly realized the benefits of a transition-metal and ligand free, temperature controlled amination reaction. Thus we explored the scope and limitations of this reaction, and this is discussed further in Manuscript 2. There is an increasing need in both the fine chemical and pharmaceutical industries for the development of new methods that easily provide substituted heterocycles. One of the methods that have been extensively explored for this function is the direct conversion of carbon−hydrogen (C−H) bonds into carbon−carbon (C−C) bonds. 1 This process is considered a "green" synthetic pathway because it eliminates the prefunctionalization steps required in modern coupling reactions and, therefore, directly reduces time, expenses, and hazardous waste. In fact, the ACS Green Chemistry Roundtable described C−H functionalizations of heterocycles as the most desirable new reactions that could benefit the pharmaceutical industry. 2,3 For decades, precious metals, namely palladium, have been the primary catalysts used for both traditional coupling and C−H arylation reactions. 4 Iron catalysts, which are readily available, cheap, and nontoxic, have been relatively unexplored for coupling reactions. However, new methods are emerging that suggest an important role for this transition metal in modern organic synthesis. 5 Notably, Nakamura has recently developed an iron-catalyzed C−H arylation reaction. 6 Comparison of the metallic catalyst used in two similar methods for the direct C−H arylation of 2-phenylpyridine shows that the iron-catalyzed reaction proceeds at lower temperatures and is higher yielding and the catalyst is 22 times cheaper (Scheme 2.1). 4b,6b,7 Though the utility of iron-catalyzed C−H arylation reactions is apparent, the scope of these potentially transformative reactions has yet to be expanded to include the arylation of highly desired heterocycles, and the mechanism is still not fully understood. Herein, we describe the ability to perform directed C−H arylations of heterocyclic substrates using cheap and nontoxic iron catalysts.
Our initial studies commenced with the pyridine substrate shown in Table 2.1.
Nakamaura's conditions that were previously shown in Scheme 2.1 were not optimal, producing only a 67% yield (entry 3). Also in contrast to Nakamura's work, the monoarylated product was exclusively obtained; the diarylated product was never observed for any of the reactions presented herein. Extended reaction times led to deterioration of the reaction's yield, possibly as a consequence of reduction of the imine; on a few occasions, the corresponding amine was isolated as a minor product.
Careful control of reaction conditions allowed for complete conversion in 15 min. Notable difficulty arose with regards to the drop rate of the Grignard reagent and the stir rate of the reaction. 6b It appears that the size of the reaction vessel can also dramatically alter yield. Dropwise Grignard addition into small, narrow vials provided almost no reaction, with exclusive homocoupling of the Grignard reagent resulting in biphenyl formation. This is likely caused by a combination of small surface area for substrate reactivity and inadequate stir rates. Larger flasks (e.g., 35−50 mL roundbottom flasks for a 0.55mmol reaction), providing more surface area, and high stir rates proved to be the best choice (see Supporting Information for details.) The reactions were very clean; the only compounds that could be observed by GCMS were the starting materials, the biaryl product and biphenyl, arising from homocoupling of the Grignard reagent. To minimize the aerobic iron-catalyzed homocoupling, an inert atmosphere and excess Grignard reagent were required. 8 Additionally, we employed additives such as DMPU 9 or KF 10 which have been previously shown to minimize Grignard homocoupling.

Scheme 2.1. Comparison of C−H Arylation Methods
The best conversion was achieved with a catalyst/ligand ratio of 1:2 (Table 2.1, entry 2). As shown by Nakamura, 4,4′-di-tertbutyl bipyridine (dtbpy) appeared to be the optimal ligand (entries 2, 5, and 6). Interestingly, the use of FeF 3 ·3H 2 O showed 18% product formation, with no biphenyl present (entry 9); but the optimal catalyst was Fe(acac) 3 (entries 7 and 8), so this was used for subsequent experiments. We ultimately chose to perform the reactions in the presence of the KF additive (entry 7) due to a slight suppression of the biphenyl byproduct. Interestingly, an iron(II) catalyst was ineffective (entry 11). Future research efforts in our laboratory will be directed toward identifying the catalytic intermediates in this reaction, including the oxidation state of the iron in this process. Further screening of solvents and oxidants showed that our original choices, chlorobenzene and 1,2-dichloro-2-methylpropane, were optimal.
When our optimized conditions were applied to the nonheterocyclic substrate derived from acetophenone, diarylated products were observed, as previously shown by Nakamura (not shown). 6 A screen of directing groups was performed ( Table 2 Our optimized reaction conditions were then applied to a variety of heterocyclic substrates (Table 2.3). In most cases, the imine group could be easily hydrolyzed to the ketone. 11 Several nitrogen-containing heterocyclic biaryls could only be isolated as imines (entries 1 and 3) because the hydrolysis of these compounds proved more difficult than expected, presumably due to protonation of the heterocycle's basic nitrogen. For reactions that did not reach complete conversion, the isolated yields were reduced considerably due to difficult chromatographic separations.   The yields of the arylations were sterically dependent, and opposing trends were observed for pyridines, thiophenes, and furans. Comparison of sulfur-containing compounds shows that benzothiophene was less reactive than thiophene (entries 10 and 9), and 3-methyl thiophene (entry 11) was completely nonreactive, indicating a decrease in reactivity with increasing steric hindrance.
Analysis of the oxygen-containing heterocycles shows that conversions and yields increased with steric constraints (entries 6−8). Azole substrates appear to be more robust (entries 1−4). Notably, chlorinated pyridines can be readily substituted, allowing for subsequent functionalization (entry 3). A quinoline substrate was nonreactive (entry 5); however, this could be attributed to the aldehyde-derived directing group described in Table 2.2, entry 4.  The formation of C-N bonds is essential for the synthesis of highly desirable pharmaceutical and biologically active targets. Current methods rely on transition metals such as palladium 1 and rhodium. 2 More recently the focus has shifted towards reagents that are more environmentally friendly and affordable catalysts like cobalt, 3 copper, 4,5,6 and nickel. 7,8,9 Recent advances have discovered transition metal free methods that also result in C-N bond formation. 10 We report herein a fast and easy method for the formation of C-N bonds resulting in arylated tertiary amines.
Considering our recent work involving iron and Grignard reagents to directly form carbon-carbon (C-C) bonds on various heterocycles, 11 and the success of other transition metals accomplishing reactions of this type, we envisioned that ironcatalyzed reactions could play a role in these mechanisms as well.
Bolm and Correa have already demonstrated that iron can efficiently form C sp2 -N bonds from aryl iodides and nucleophilic nitrogen sources. 12 Our initial optimization focused on the coupling of N-chloroamines with phenylmagnesium bromide in an effort to form similar C sp2 -N bonds in an Umpolung fashion. The variables investigated included the screening of iron catalysts, nitrogen and phosphine ligands, several additives, and a range of temperatures. None 99 a.) reactions performed on a 1.00 mmol scale in 2.00 mL of 2-MeTHF using 1.5 eq. of Grignard reagent b.) yields determined by GC-MS using dodecane as an internal standard. c.) 1.0 eq. Grignard reagent used. d.) 2.0 Grignard reagents used.

Scheme 3.1. Initial Optimization Outline
Our initial efforts to use iron catalysts appeared successful. Several iron salts were shown to produce the expected product in reasonable yields (entries 1-8).
However attempts to isolate this product significantly decreased the yield (entry 9).
Inorganic salts have been shown to promote the presumed transmetallation step. 13 In our case these were detrimental to the yield (entries 10 and 11), and in the case of MgBr 2 completely shut off the reaction. Contrary to other's reports throughout the process we detected chlorobenzene as a dominant byproduct as well as small amounts of biphenyl. In an attempt to minimize these side reactions we lowered the temperature and varied the equivalents of Grignard reagent used (entries [12][13][14]. We found that with 1.5 equivalents of Grignard reagent at -78 °C the reaction afforded a 75% yield. With these optimized conditions we attempted to investigate the substrate scope. Unfortunately the Fe(acac) 2 catalyzed reactions performed on other substrates showed a significant decrease in yield (Figure 1: 3b and 3g), and other metals, CoBr 2 and Cu(OTf) 2 , were detrimental (3b and 3c).
Not satisfied with these results, we screened several ligands hoping to increase our yield further (entries 16-20). Introducing dinitrogen ligands had a small beneficial effect on the reaction (entries 16-20). However we were aware that these reactants could successfully couple in the presence of these dinitrogen bases without any catalyst, 10 although those reactions were performed at -40 °C and required several hours for completion. Removing the iron as a control afforded the product in nearequal yields (entry 21). This was discouraging; but we quickly realized that the ligand itself was also not necessary for the reaction to occur (entry 22). This has been demonstrated previously by Knochel 14 at -45 °C, however he suggested limitations to the reaction scope and that this was only applicable to benzylic N-chloroamines.
Transition metal free reactions were performed at several temperatures (Table   3.2). The decrease in temperature showed a steady increase in product formation, as well as a decrease in overall byproducts.

General Synthesis of Imines
To an oven dried 50 mL RBF with stir bar was added 5 g of 3 Å molecular sieves and 12.00 mL of toluene. The system was sealed with a rubber septum and flushed with N 2 . The amine (2.5 mmol) and heterocyclic ketone (2.3 mmol) were added successively via syringe. The reaction was stirred at 100 °C for 4 hrs, then cooled to room temp. The mixture was filtered through Celite, and the filtrate was concentrated in vacuo. The crude residue was purified by column chromatography (9:1 Hexanes:EtOAc).

Representative Procedure for Arylation of N-Heterocyclic Imines:
To an oven dried 35 mL RBF with stir bar was added the imine ( 0.55 mmol),

Representative Procedure for Arylation and in situ Hydrolysis of S-and O-Heterocyclic Imines:
To an oven dried 35 mL RBF with stir bar was added the imine (0.55 mmol), HCl were added successively. The reaction was stirred at room temperature overnight.
The mixture was extracted 3x with EtOAc. The organic layer was dried over MgSO 4 , filtered, and concentrated in vacuo. The crude residue was purified by column chromatography.

Reaction Vessel Size Comparison:
Analysis of various reaction vessel sizes was performed using gas chromatography. The percent composition of each compound at the end of the 15 minute reaction was obtained, and a ratio of the biphenyl byproduct to the desired arylated product is shown. Narrow reaction vessels such as vials and Schlenk tubes showed a large biphenyl:product ratio. Wide, round bottom flasks showed significant product formation with comparative efficiency relative to vials and Schlenk tubes; 35 mL RBF's were shown to be the best.          General Procedure for the synthesis of N-chloroamines: To an oven dried 100 mL round bottom flask with a stir bar is added 5.00mL (1 eq.) of the amine and cooled to 0 °C followed by the addition of aqueous 4% NaOCl (2 eq.). The reaction is stirred for 5 minutes, then brought to room temperature and extraction with diethyl ether (3x) followed by a DI water wash (3x), and a brine wash (1x). The organic layer was dried over sodium sulfate, filtered, and concentrated in vacuo. The resulting oil is used as obtained after NMR characterization and purity analysis. We propose the above mechanism to explain our iron-catalyzed reactions. The first step involves reduction of the Fe(III) to Fe(I)-Ar using three equivalents of Grignard reagent. This is followed by coordination between the lone pair on the imine nitrogen and the iron species. Oxidative addition and subsequent reductive elimination yields the desired product and a reactive iron hydride. We have experimentally determined trace amounts of a reduced imine byproduct that supports the generation of this iron hydride. This Fe(I) species is then oxidized using 1,2-dicholoroisobutane followed by a ligand exchange to regenerate the active Fe(I)-Ar. This accounts for the observed biaryl formation and need for excess Grignard reagents. We envision another possible pathway going directly from the Fe(I) hydride to the Fe(I)-Ar. If this could be optimized the need for an oxidant and 3 equivalents of Grignard reagent can be eliminated.