Skip to main content
NCBI home page
iScience logoLink to iScience
. 2019 Sep 27;20:373–391. doi: 10.1016/j.isci.2019.09.035

Relative Strength of Common Directing Groups in Palladium-Catalyzed Aromatic C−H Activation

Anna Tomberg 1, Michael Éric Muratore 2, Magnus Jan Johansson 2, Ina Terstiege 3, Christian Sköld 4, Per-Ola Norrby 5,6,∗
PMCID: PMC6818342  PMID: 31614320

Summary

Efficient functionalization of C−H bonds can be achieved using transition metal catalysts, such as Pd(OAc)2. To better control the regioselectivity in these reactions, some functional groups on the substrate may be used as directing groups, guiding the reactivity to an ortho position. Herein, we describe a methodology to score the relative strength of such directing groups in palladium-catalyzed aromatic C−H activation. The results have been collected into a scale that serves to predict the regioselectivity on molecules with multiple competing directing groups. We demonstrate that this scale yields accurate predictions on over a hundred examples, taken from the literature. In addition to the regioselectivity prediction on complex molecules, the knowledge of the relative strengths of directing groups can also be used to work with new combinations of functionalities, exploring uncharted chemical space.

Subject Areas: Catalysis, Computational Molecular Modelling, Organic Reaction

Graphical Abstract

graphic file with name fx1.jpg

Open in a new tab

Highlights

  • •

    Directing group strength for ortho-palladation can be predicted quantum chemically

  • •

    Correlation with fragments allow regioselectivity predictions in complex molecules

  • •

    Directing strength is enhanced by deprotonation under the reaction conditions

  • •

    Palladation in between two directing groups is disfavored sterically; no synergy

Catalysis; Computational Molecular Modelling; Organic Reaction

Introduction

Synthetic protocols that allow direct activation/functionalization of inert C−H bonds have for a long time remained a Holy Grail in organic synthesis (Gensch et al., 2016). Potential applications would lead to atom economical processes with unmatched step-economy. However, the unreactive nature and high stability of C−H bonds (typical bond energy of C(sp2)−H is 110 kcal/mol) have made them elusive targets for diverse functionalizations under mild conditions (Xue et al., 2017). Nonetheless, the mindset that these bonds are out of reach has changed. Nowadays, C−H bonds are considered functional groups and are utilized to introduce a plethora of functionalities, often with the help of organometallic catalysts (Cernak et al., 2016, Abrams et al., 2018).

The presence of multiple unsubstituted carbons in a given molecule makes controlling regioselectivity in these reactions a challenging task. In catalytic C−H functionalization, two main approaches are used to address this problem: (1) add special ligands on the metal catalyst (Lyons and Sanford, 2010, Wang et al., 2017); (2) use directing groups (DGs) on the substrate able to bind to the metal center and force the reactivity to specific positions (Figure 1) (Sambiagio et al., 2018). In addition to directed C−H activation, there are several elegant ways of overcoming the positional selectivity induced by pre-coordination of the metal to the substrate, including seminal contributions from the Yu and Hartwig labs (Liu et al., 2014, Hartwig and Larsen, 2016, Kiser et al., 2012).

Figure 1.

Figure 1

Open in a new tab

Achieving Regioselectivity in C−H Activation Reactions Is a Challenging Task

The electron-donating/withdrawing character of functional groups (FG = EDG or EWG) leads to the activation of different positions. Increased regioselectivity in metal-catalyzed reactions can be achieved with directing groups (DGs).

One of the most developed C−H activation approaches takes advantage of palladium as catalyst, leading to C(sp2)−C(sp2) bond formation or functionalization with N, O, P, and halogens (Lyons and Sanford, 2010). Using Pd(OAc)2, a variety of couplings can be introduced regioselectively by employing DGs (McMurray et al., 2011, Chen et al., 2015). These need not be specially designed moieties: common motifs of organic molecules such as pyridines and carboxylic acids serve as effective DGs. Most DGs in palladium-catalyzed C−H activations are ortho-directing. In the case of functionalization of more complex molecules, the presence of multiple DGs can lead to activation on several sites. Therefore, the prediction of the regiochemical outcome plays an important role (Davies and Morton, 2017). Although general reactivity trends of common functional groups, steric hindrance, and acidity of the leaving proton can hint to the preferred regioselectivity, accurately predicting the site of reaction in compounds with several DGs of similar reactivity remains difficult.

To put things in perspective, several mechanistically diverse methods are available for activating C(sp2)−H bonds (Scheme 1). At one end of the spectrum, a strong enough base (frequently directed by a coordinating group) is able to abstract a proton directly from an aromatic ring. The immediate reaction product can be an organometallic reagent, e.g., an organolithium (Snieckus, 1990), used as a nucleophile in further reactions. At the other end of the reactivity scale, strong electrophiles can react with the π-system in a reaction in which bond formation to carbon is commonly the rate-limiting step, followed by a facile deprotonation. This is the classical Electrophilic Aromatic Substitution reaction (EAS), whereby selectivity is generally determined by the intrinsic reactivity of the aromatic system (Tomberg et al., 2019). Although the reagent can be an electrophile, a radical also reacts by a similar pathway, but with a different selectivity profile. The principle remains the same: the reagent selects the most reactive carbon and forms an addition product, whereupon the proton at that position is eliminated.

Scheme 1.

Scheme 1

Open in a new tab

Classes of C−H Functionalization

The presented work focuses on palladium-catalyzed aromatic C−H activation through the CMD mechanism.

In between these two extremes, we find reagents that combine a weak electrophile with a weak bidentate base. Reactivity is enabled by the cooperativity between the two moieties of the catalyst, where an initial weak electrophilic attack will activate the hydrogen for deprotonation by the weak base in a concerted metallation deprotonation (CMD). With only a weak base and a weak electrophile, the reagent is compatible with a wide range of functionality. The mechanism of action for the prototypical CMD catalysts, palladium carboxylates (e.g., Pd(OAc)2), was elucidated in pioneering studies by the groups of Fagnou (Gorelsky et al., 2008, Lapointe and Fagnou, 2010), Macgregor (Davies et al., 2005), and others (Davies et al., 2017). As palladium initiates an electrophilic attack on an aromatic carbon, the carboxylate forms a bond with the hydrogen atom on that position. Subsequently, palladium moves into the plane of the aromatic ring, forming a σ-bond to that carbon, while its proton is transferred to the carboxylate (Scheme 1 Step 1). The intrinsic barrier for this reaction is moderately high, but the reaction will be facile if the palladium is stabilized by coordination to a proximal DG. The resulting aryl−palladium complex can then undergo coupling reactions through reductive elimination with another group on palladium, possibly preceded by a transmetallation depending on the exact reaction conditions (Scheme 1 Step 2).

The CMD reaction can be reversible. However, if the forward coupling reaction is favored over the reverse CMD, the reaction will display kinetic selectivity based on the relative stabilities of the plausible C−H activation transition states (TSs). Thus, it has been shown that the reaction selectivity can be predicted by calculating the various possible CMD activation barriers using DFT methods (Davies et al., 2017). However, we are interested in automating the selectivity prediction in a workflow available to bench chemists, as we have previously done for other C−H functionalization reactions (Tomberg et al., 2019, Andersson et al., 2014). To this end, TSs searches are not the method of choice since these calculations are notoriously hard to automate, even though recent approaches show promise (Guan et al., 2018). We therefore wanted to explore if simpler methods show sufficient predictive power for our purposes. Based on the Bell-Evans-Polanyi relationship (Bell and Hinshelwood, 1936, Evans and Polanyi, 1936, Jensen, 1999), and the more specific Hammond postulate (Hammond, 1955), we tested the hypothesis that the selectivity in the CMD TS is reflected in the relative energy of the corresponding palladacycle intermediate in the reaction (Scheme 2.1).

Scheme 2.

Scheme 2

Open in a new tab

Palladacycle Intermediates Were Used to Probe Relative DGs Strengths

(1) Isodesmic pseudo-equilibrium between two transition states reflected in the high energy intermediate.

(2) Illustration of the equation used to calculate the relative energies of palladacycles corresponding to DGs. See also Data S17 for coordinates and DFT energies of the compounds and palladacycles studied in this work.

To further simplify the calculations and put each DG on a convenient scale, we compared each potential group with hydrogen, using the equation illustrated in Scheme 2.2. Note that this comparison changes molecularity: the DG displaces one carboxylic acid from palladium. Even in cases in which the coordination of the DG is enthalpically disfavored, it may still be favored entropically and thus can outcompete the non-directed CMD reaction. This means that, on a scale based on potential energies, even DGs with moderately positive values will outcompete positions without a DG. The primary use for the scale should be to compare different groups, i.e., only relative numbers should be used.

To the best of our knowledge, the directing abilities of DGs toward palladium electrophiles have never been analyzed in depth and/or in a systematic way. Few experimental studies can be found reporting competition experiments with a handful of DGs, providing only qualitative trends in reactivity (Sun et al., 2013, Desai et al., 2008, Dey et al., 2019). The work presented here aims to quantitatively measure the directing strength of common ortho-directing functional groups. Specialized functional groups are able to direct instead to the meta-position (Bera et al., 2014, Wan et al., 2013), but the geometry is expected to differ significantly from the CMD intermediate considered here (Yang et al., 2014) and is thus out of the scope of the current study.

We propose a quantum mechanical approach to compute the relative strengths of DGs in palladium-catalyzed aromatic C−H activation. The results have been assembled into a convenient look-up table, featuring 133 DGs, that can be used to quickly compare which DG would yield the major product. The computed relative strengths of DGs were validated by matching results to 150 examples from the literature, where reactant molecules featured two or more non-equivalent potential sites of activation.

Results and Discussion

We set the goal to develop an approach to quantitatively and systematically score DGs for aromatic C−H activations catalyzed by Pd(OAc)2. Our hypothesis was that there should be a correlation between the stability of the palladacycle formed during CMD and the directing strength of a DG. In other words, if DG1 prevails over DG2, then its relative energy according to the equation shown in Scheme 2(2) should be lower than the one from DG2 (see Transparent Methods section in Supplemental Information). It has to be kept in mind that this approach does not have the capability of predicting a reaction's feasibility but provides a way to score DGs relative to each other.

To probe the validity of our hypothesis, we first tried to reproduce experimental findings from Sanford et al. (Desai et al., 2008). The competition experiments described in their work compared how much of each respective acetoxylation product formed after 12 h in AcOH/Ac2O and in benzene when using different DGs. Experimentally, the orders observed for the two solvents were almost identical (Figure 2). Using our method, the ranking was similar to experimental data with very small differences (within 1 kcal/mol) for heterocycles b, c, and d. Therefore, although the calculated values were not spot on with experimental findings, the overall trend in reactivity was captured.

Figure 2.

Figure 2

Open in a new tab

Competition Experiments Reported by Sanford et al. with the Corresponding Calculated Order in 1,2-dicholoroethane

Although the competition experiments reported by Sanford investigated separate compounds featuring one DG each, our main goal was to study molecules that bear two different DGs. For example, compound 1 features a pyridine and an ester, as shown in Scheme 3, which can both be ortho-directing. Several experiments, taken from different studies (Li et al., 2011, Hull et al., 2006), report that the pyridine group is more strongly directing than the ethyl ester. Indeed, our calculations showed that the coordination through pyridine was over 15 kcal/mol lower in relative energy than the one with directing ethyl ester.

Scheme 3.

Scheme 3

Open in a new tab

Compound 1 and the Two Palladacycles Formed with Its DGs Showing that Pyridine Is a Stronger DG Than Ethyl Ester

Experimental C−H activation site is marked by a black circle (Li et al., 2011); predicted site of activation is marked by a green-filled circle; DGs are highlighted with color. See exact energies in Table S1.

The next type of molecules we investigated were compounds that have different DGs that could “help” each other direct reactivity to the same carbon. This is exemplified in compound 2, in which both the pyridine group and the O-methyl oxime could direct the reaction to position C (Figure 3). Nevertheless, the intermediate directing the reaction onto position A through pyridine was calculated to be more stable (Erel = −14.9 kcal/mol) compared with the two intermediates that direct the reaction to position C (Erel = −11.5 or −7.9 kcal/mol depending on whether pyridine or oxime ether coordinates). Position B, stabilized by only the oxime ether, was also less favored (Erel = −11.6 kcal/mol). Interestingly, the potential synergy between the two DGs was not observed: the relative energy of the palladacycle with both DGs coordinated was much higher than either individual coordination, namely, 4.6 kcal/mol. From this, we can conclude that only the strongest DG coordinates to palladium. For two positions that both can be activated by the strongest group, the least sterically hindered position would be favored. These results are in agreement with experimental data from Kalyani et al. (Kalyani and Sanford, 2005) who also observed that the less sterically hindered position was preferred for palladium-catalyzed C−H activations.

Figure 3.

Figure 3

Open in a new tab

Compound 2 Has Three Positions that Could React

The less hindered position activated by the strongest DG is the preferred reaction site, both computationally and experimentally. Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color. See exact energies in Table S2.

Fragmentation Can Be Used to Compare DGs in a Full Molecule

Encouraged by these results, we sought to simplify the model further: could the regioselectivity of complex molecules be predicted using relevant fragments? In other words, can we compare the relative energies of the metallacycles with fragments featuring only one DG and successfully predict the reaction sites on entire molecules? An example of such fragmentation is illustrated in Scheme 4. Exemplified by compound 1 again, the resulting fragments are methyl benzoate (ethyl was replaced by methyl in the model fragment) and 2-phenylpyridine. When coordinated to palladium, these form metallacycles with relative energies of 1.5 and −15.2 kcal/mol, respectively. In agreement with our previous calculations and with literature precedents (Li et al., 2011, Hull et al., 2006), these energies indicate that pyridine is a stronger DG than the ester.

Scheme 4.

Scheme 4

Open in a new tab

Compound 1 and the Fragments that Can Be Used to Predict the Site of Reaction

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

As discussed earlier, certain DGs can in principle direct the catalyst to more than one aromatic carbon. For example, in compound 3, three positions can potentially be activated (Scheme 5). To investigate these reactive sites, three fragments were created and scored based on the stability of the corresponding organopalladium intermediates. The relative energy obtained for the intermediate leading to the activation at A was −13.3 kcal/mol, whereas palladacycles formed at B and C resulted in Erel = 4.4 and 2.0 kcal/mol, respectively. This indicates that position A is activated by the strongest DG in this case and that positions B and C are much less likely to react, which is in agreement with experimental data (Tredwell et al., 2011).

Scheme 5.

Scheme 5

Open in a new tab

Compound 3 and the Fragments that can Be Used to Predict the Site of Reaction

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

In the fragmentation of the previous molecule, alkyl chains on the reacting aryl group were removed leaving only a mono-substituted benzene. The validity of this approximation was evaluated empirically by observing experimental results for a variety of DGs. The reactivity of DGs overshadows the impact of substituents: irrespective of their electron donating/withdrawing abilities, they cannot shift the reactivity from a strong DG to a weak one. In the case of two competing DGs, substitution can be used to either block a position ortho to a DG (Figure 4.1) or create steric hindrance from a meta position that will direct the reaction to a less sterically hindered carbon available to the DG (Figure 4.2).

Figure 4.

Figure 4

Open in a new tab

Substituents on Aromatic Rings Cannot Be Used to Shift Reactivity Away from a Strong DG to a Weak One

However, they can be used to block an accessible ortho-position (Shan et al., 2012) or to produce steric hindrance at the meta-position(Yang et al., 2007).

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

When the same DG can activate carbons on different rings, strong electron donating or withdrawing groups can be used to impact selectivity: since the reaction has an electrophilic character (Scheme 1), an electron-rich ring is more likely to react than an electron-poor ring. For example, once a nitro group is placed on one ring of a benzophenone (8 versus 9), the activation is observed only on the unsubstituted ring (Figure 5) (Xiao et al., 2011, Shan et al., 2012). Similarly, a cyano group on an azobenzene leads to reactivity on only the unsubstituted ring (10 versus 11) (Dong et al., 2014). Conversely, the presence of an electron donating group directs the reaction to the same ring, as illustrated by the methoxy substituent on the azobenzene (10 versus 12) (Xiong et al., 2013). For these types of compounds, where the directing power is identical for two different positions, selectivity between the two DG-activated positions will be determined by rules similar to EAS.

Figure 5.

Figure 5

Open in a new tab

Examples of Compounds in which the DG Activates Two Different Positions: Selectivity Can Be Narrowed Using Ring Substituents

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

The same fragmentation approach was used to obtain Erel for DGs in 150 other compounds; the results for six compounds are presented in Figure 6, whereas the rest can be found in the Data S1–S4. The DGs for ortho-activation of aromatic carbons were extracted from a review by Chen et al. (2015). To render fragments more transferable, alkyl chains were replaced by methyl groups (e.g., compound 15) and other substituents on the aromatic rings were removed (e.g., compound 18). Applying the reactivity patterns described earlier, a simple analysis can be performed on relatively complex molecules with high accuracy in reaction site prediction: of 150 examples collected, only 4 predictions did not match experimental results.

Figure 6.

Figure 6

Open in a new tab

Examples of Fragmentation of Compounds for DGs Strength Comparison

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color. For experimental results see references Gulevich et al., 2012, Kalyani and Sanford, 2005, Li et al., 2013a, Lou et al., 2014, Tredwell et al., 2011, and Kim et al., 2014. See rest of compounds in Data S5.

So far, we have considered neither the reaction conditions nor the coupling partner (or its absence). In reality, these are important parameters that can affect the reaction outcome. For example, how does the strength of a directing group depend on the protonation state of the compound? How would our approach perform in such cases?

The DG's Protonation State Influences Regioselectivity

N-phenylbenzamide (4) presents a perfect example of a system in which selectivity is highly influenced by reaction conditions (Figure 7). Although experimental studies seem to report contradictory results, some supporting reaction at position A (Boele et al., 2002, Zhu et al., 2018) and others illustrating functionalization at position B (Kametani et al., 2000, Chou et al., 2017), a closer scrutiny at reaction conditions easily rationalizes these divergent reactivity profiles. Under acidic conditions, where the amide is presumably present in its neutral form, transformations take place at position A. In contrast, under mild basic conditions, where the amide may be deprotonated in a kinetically preferred CMD reaction of the N−H group, reactions occur on position B. Using our model, this shifting preference could be captured. In the presence of acid, the neutral DG prefers to coordinate to palladium through the oxygen, leading to a metallacycle intermediate activating position A that is 4.2 kcal/mol more stable than the palladacycle activating B. Under basic conditions, the metallacycle intermediate is generated from deprotonated amide with a formal negative charge on the nitrogen. This coordination is preferred over the cycle with oxygen coordination (Erel(B) = −27.8 kcal/mol versus Erel(A) = −21.3 kcal/mol), leading to activation of position B instead.

Figure 7.

Figure 7

Open in a new tab

Compound 4, N-phenylbenzamide, Shows Acidity-Dependent Reactivity in Palladium-Catalyzed C−H Activation

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

To provide a proof of concept and to validate our predictions, we have synthesized substrate 19 featuring both pyridine and acetanilide DGs. According to our model, pyridine is a very strong DG with Erel = −15.2 kcal/mol, whereas the acetamide is significantly weaker (Erel = −7.4 kcal/mol). Using this compound, we wanted to investigate whether it is possible to shift the reactivity away from the pyridine DG by altering the pH of the reaction. On the one hand, we anticipated that by addition of a strong acid, the pyridine moiety should be protonated and under these conditions the acetanilide should become the strongest DG. On the other hand, we envisioned that deprotonation of the acetanilide functionality would result in the formation of a charged amide DG, which according to our model, should coordinate more strongly to palladium than the pyridine fragment does.

The initial conditions of arylation of 19 were inspired by Sanford's seminal report (Kalyani et al., 2005). Under the typical C−H arylation conditions (in acetic acid), we observed mostly arylation ortho to the pyridine DG (19a-b), whereas products of arylation ortho to the acetanilide (mono- or bis-arylation products, including 19c and d) could not be detected, confirming and supporting that under these « neutral » conditions, the pyridine fragment is a much stronger binder to palladium than the acetanilide moiety (Scheme 6.1, see also competition experiments in Tables S4 and S5). Performing the same transformation in toluene in the presence of a strong Brønsted acid (HBF4 as its diethyl ether complex) resulted in an overall poorer reactivity profile; however, in this reaction small amounts of products of arylation ortho to the acetanilide (19c and d) could be isolated and characterized, whereas no trace of products of arylation ortho to the pyridine could be detected (Scheme 6.2). Although this approach to switch regioselectivity has not been optimized, the latter experiment provides a proof of concept and supports our model's prediction. The final test was the deprotonation of the acetamide DG under strong basic conditions. Stoichiometric deprotonation of 19 in the presence of freshly prepared lithium diisopropyl amide (LDA), addition of this lithium amide to stoichiometric Pd(OAc)2, and subsequent exposure to Ph2IBF4 did not lead to any observable amount of arylation products 19c or 19d, and only traces of 19b were isolated. The identical procedure applied to acetanilide led to much lower reactivity than that typically observed for the same arylation under catalytic and neutral conditions (see Transparent Methods in Supplemental Information). This suggests that most of the palladium presumably forms an unproductive and catalytically inactive complex. Additionally, reproducing the latter experiment in the presence of 2-phenylpyridine led to arylation of 2-phenylpyridine only. Therefore, we can conclude that LDA is not a suitable base to achieve both satisfactory reactivity and a regioselectivity shift under the conditions presented herein.

Scheme 6.

Scheme 6

Open in a new tab

C−H Arylation of Bifunctional Substrate 19 under a Range of Conditions: Proof of Concept of Control of Regioselectivity via Protonation or Deprotonation of DGs

1. C−H Arylation of 19 under « neutral » conditions.

2. C−H Arylation of 19 under strong acidic conditions.

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled and blue-filled circles; DGs are highlighted with color. Also see Tables S4 and S5 and Data S8–S16. [a]1H NMR yield employing 1,1,2,2-tetrachloroethane as internal standard >; [b] Isolated yield (as measured against 1,1,2,2-tetrachloroethane as internal standard).

Coupling Partners Play a Role if the Energy Difference Is Small

When a DG can activate more than one position with similar strength, the nature of the coupling partner starts playing a role. As exemplified by compound 20, a triazole DG on the naphthalene can direct the catalyst to either position A or B (Figure 8). Comparing the relative energies of the corresponding metallacycles suggests that the DG would activate both positions to a similar extent. However, from experiment, a mixture of products is not observed. In the paper by Shi and Kuang (2014), ortho alkoxylations on this aryl triazole were reported to take place on position A. Alternatively, in the study by Tian et al. who investigated the bromination of similar molecules, compound 20 reacted on position B (Tian et al., 2013). Both alkoxide and bromide will have relatively high barriers to reductive elimination. Thus, it is conceivable that in at least one of the cases, the reductive elimination becomes rate limiting, allowing the two palladium intermediates to equilibrate before the irreversible selectivity-determining step. Since our model does not describe the reaction steps after C−H activation, it cannot be used to predict which of the two positions will be reactive if another step becomes selectivity determining.

Figure 8.

Figure 8

Open in a new tab

Compound 20: Relative Energies of Metallacycles for Positions A and B Are within the Margin of Error (1 kcal/mol), so the Model Is Unable to Distinguish between the Two

Other factors such as the nature of the coupling partner will determine regioselectivity. DG is highlighted with color.

Another important aspect of the reaction conditions is the presence or absence of coupling partners. This information is important in biaryls or systems with fused rings. For example, in the compounds shown in Figure 9, the DGs reach two positions, A (on same ring) and B (on neighboring ring). By calculation, the activations on positions A are more favorable and those will react if a coupling partner is available (21a [Daugulis and Chiong, 2009, Chiong et al., 2007] and 22a [Kim et al., 2010]). Alternatively, in the absence of an external coupling partner, there is no energetically accessible pathway from the activation of position A; thus, the system will eventually equilibrate to position B, which allows cyclization (21b [Li et al., 2013b] and 22b [Li et al., 2014]).

Figure 9.

Figure 9

Open in a new tab

Examples of Compounds where the Absence of a Coupling Agent Changes Selectivity

DGs are highlighted with color. See rest of similar compounds in Data S6.

Directing Strength Scale Combining the Results for 133 DGs

Combining the observations from the above-mentioned examples and many more, we have demonstrated that our simplified regioselectivity model for palladium-catalyzed C−H activation is predictive (see Data S5–S7). In molecules with multiple competing DGs, the reaction site can be determined by comparing the relative energy of metallacycles consisting of fragments and palladium formate. To validate our approach on as many compounds as possible, we have (1) assembled a testing set featuring a variety of DGs, (2) selected fragments covering the test molecules, and (3) compiled the results into a directing strength scale (see Data S1–S4 in Supplemental Information).

The scale allows one to easily find fragments that correspond to a studied molecule and compare their relative energies: the one with the lowest energy should lead to the major product of the C−H activation α to the DG (Figure 10). As explained earlier, the protonation state of a DG affects strongly its ability to form a stable metallacycle with palladium. As such, the correct fragments must be compared to obtain an accurate prediction. Evidently, the larger the difference between the energies of two DGs, the more likely it is that the model would detect the appropriate reaction site. Our results indicate that DGs within ca. 1 kcal/mol of each other are indistinguishable. In cases in which a molecule bears DGs of similar strength, other electronic and steric factors prevail, as discussed earlier. DGs in their deprotonated form are generally much stronger than the neutral ones. Among the strongest are amines, alcohols, and bidentate (designer) DGs. An example of a bidentate DG is N-(quinolin-8-yl)benzamide, which binds to palladium through both nitrogens, becoming one of the strongest directing group on the list. Although this moiety performs better under basic conditions, neutral/mildly acidic conditions can still allow for the deprotonation of the amide due to the effect of palladium (Gou et al., 2009). A wide range of different coupling partners can be used with this DG (Kanyiva et al., 2014, Wang et al., 2015, Li et al., 2016, Liao et al., 2018). In general, in both neutral and charged forms, the strongest coordination to palladium takes place through a nitrogen, whereas groups that bind through an oxygen atom seem to be weaker. This tendency is further illustrated by another bidentate DG, 2-(benzylideneamino)acetic acid. This imine is a transient DG (Liu et al., 2017), generally formed in situ from an aldehyde or a ketone and an amino acid (Wang et al., 2018, Xu et al., 2017). In the presence of base, both the carboxylic acid and the imine nitrogen coordinate to palladium, with a relative strength slightly weaker than the N-(quinolin-8-yl)benzamide. Once the coupling step is completed, the aldehyde or ketone can be recovered by addition of acid (Zhang et al., 2019).

Figure 10.

Figure 10

Open in a new tab

DGs Relative Strength Scale: A Few Key Examples

See all studied DGs in Data S1–S4.

Once all results for the fragments were assembled into an ordered list, interesting patterns started emerging. For example, there is a correlation between the strength of a DG and the size of the ring it forms in the corresponding metallacycle. Expectedly, DGs that form four-member rings are the weakest. As exemplified in Figure 11.1 by compound 23, the negatively charged oxygen coordinates to palladium leading to two potential activation sites. Reactive site A forms a four-member ring metallacycle and has a relative energy of 1.9 kcal/mol; reactive site B forms a five-member ring with palladium, which leads to a relative energy of −9.7 kcal/mol. When we examined the experimental results, we were surprised to find that most papers report position A-selective activation. However, all these examples had one thing in common: the presence of phosphines as reagents (Willis and Smith, 2014, Liu and Tzschucke, 2016, Roudesly et al., 2018, Lehecq et al., 2017). As demonstrated by Stephens et al. (2015a), phosphines play an important role in diverting the selectivity from the more energetically favorable reactive site B toward position A. If phosphines are not used, the reactivity is observed on position B, as our model predicted (Stephens et al., 2015a, Stephens et al., 2015b). This example highlights that our approach can be used only with palladium ligands with similar reactivity to acetates.

Figure 11.

Figure 11

Open in a new tab

Trends in Reactivity

1: The presence of phosphines alters the reactivity to the less energetically favorable position A (Willis and Smith, 2014, Liu and Tzschucke, 2016, Roudesly et al., 2018, Lehecq et al., 2017).

2: Positions activated through 6-member palladacycles are more favorable than the ones forming larger rings (Bedford et al., 2009, Zhao et al., 2010, Sun et al., 2015).

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color.

DGs that form large rings have lower energies than four-member rings but still lose to five- or six-member ring forming groups. For example, the carbamate group on compound 24 (Figure 11.2) can direct to both positions A and B, with relative strengths of 3.7 and −1.9 kcal/mol, respectively. This is in line with the observed experimental results showing reactivity on position B (six-member ring palladacycle) (Zhao et al., 2010, Sun et al., 2015).

The majority of DGs form either five- or six-member ring palladacycles. From the analysis of our calculations, we found no strong preference toward either. Several examples collected in Figure 12 demonstrate that the computed relative strengths of these DGs differ by less than 1 kcal/mol. The same trend is observed for the deprotonated form as well.

Figure 12.

Figure 12

Open in a new tab

There Is No Clear Preference between DGs Forming Five- and Six-Member Ring Intermediates with Palladium

Examples of Mismatch between Predictions and Experimental Results

Of the 150 examples collected, four predictions did not match experimental results. In this section we will go through these cases and, when possible, rationalize the discrepancies.

The first example is an illustration of the method's limitation: compound 31 was selectively hydroxylated on position A in presence of Pd(OAc)2, TFA/TFAA, and Selectfluor (Shan et al., 2012). The DGs found in this molecule are trifluoroacetamide and benzophenone (Figure 13.1). According to the relative energies corresponding to these DGs, the trifluoroacetamide, activating position C, is slightly stronger than benzophenone. Our model cannot distinguish groups that have relative energies within 1 kcal/mol; thus, the electron richness of each ring should be used to predict which one is the most likely to react. Since the trifluoroacetamide is an electron-withdrawing group (Hansch et al., 1991), we should expect the reaction to take place on the unsubstituted ring of compound 31 on position A. Another possibility that we considered was the influence of TFA on reactivity. A recent paper by Jiří Váňa et al. highlighted the effect of the carboxylates on different aspects controlling reactivity of the palladium-catalyzed C−H activation (Váňa et al., 2019). The authors concluded that TFA can replace acetic acid on the metal, which would change the reactivity of the catalyst by increasing the electrophilicity of the palladium atom. We computed these energies using our approach while replacing the formate ligand by a trifluoroacetate in the palladacycles with the corresponding fragments. Interestingly, the use of TFA as ligand shifted the relative stability of the organopalladium complexes, resulting in the benzophenone fragment being 6.2 kcal/mol lower in energy than the trifluoroacetamide. With this modification of the model, the experimental results are in agreement with the computed values.

Figure 13.

Figure 13

Open in a new tab

Molecules for which the Predicted Reactive Site Did Not Match Experiment

1. Compound 31 with two competing groups of similar strengths at play: benzophenone and trifluoroacetamide. See exact energies in Table S3.

2. Compound 32 where activation on B is wrongly predicted over experimentally observed activation on A.

3. Compound 33: in similar conditions, but with different coupling partners, the reactivity shifted from A to B.

4. Compound 34: the conditions do not seem to allow for deprotonation of the phenol, yet this DG wins over the acetate, which is against predictions.

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color. See also Data S7.

Another case in which our predictions were incorrect is illustrated in Figure 13.2, compound 32. According to the relative energies computed, the position ortho to the sulfonyl moiety (B) has a higher chance of being activated in both the neutral and the deprotonated forms: the nitrogen coordinates to palladium forming a five-member intermediate, which is much more stable than the coordination through the sulfonyl's oxygens. However, experimental results show that the activation takes place on the carbon A ortho to the nitrogen. A recent computational study on a similar palladium catalyst suggested that the activation does indeed proceed through the nitrogen coordination to palladium (Qiao et al., 2019). The following acetate-mediated N−H deprotonation leads to a four-member transition state that directs the reaction to position A. According to their results, the coordination through an oxygen of the sulfonyl moiety is over 10 kcal/mol higher in energy, which is in line with our predictions. However, the reported reaction mechanism does not proceed to a stable palladacycle intermediate following the C−H activation step. It is possible that, in this case, no such intermediate is formed; thus, our model cannot be used on this DG.

The next example where our model predictions differed from experiment is compound 33 (Figure 13.3). From the literature, we found two studies reporting different regioselectivities (Thirunavukkarasu and Cheng, 2011, Yu et al., 2008). However, the relative energies for positions A and B in this molecule are substantially dissimilar (−10.7 and −15.6 kcal/mol, respectively); thus, we would expect reactivity solely on position B. This is in line with the reported product of arylation of 33, reported by Thirunavukkarasu et al. (Thirunavukkarasu and Cheng, 2011). Conversely, a study on oxidative ethoxycarbonylation described activation on position A (Yu et al., 2008), using diethyl azodicarboxylate (DEAD) as coupling partner. The authors proposed that this reagent delivers a CO2Et radical by thermal decomposition and promotes the reaction through a PdIV intermediate, which falls outside of the reactivity predicted by our model.

The last prediction that did not agree with experimental results was for compound 34. The DGs at play are highlighted in Figure 13.4: phenol and acetate. The reaction was performed in AcOH, with benzoquinone as oxidant, so the phenol is expected to be in its neutral form (Zhang et al., 2014). This suggests that the acetate DG should lead to the major product (position B), since its relative energy is −8.8 kcal/mol lower than the one of the palladacycle with the phenol DG. However, the reported product is the alkenylation at position A exclusively. According to our model, this could be possible only if a portion of the phenol DG was deprotonated, which could explain the low yield observed for this reaction (34%).

Final Test: Regioselectivity on Drug-like Compounds

As a final test for the model, we found examples of drug-like molecules that have a palladium-catalyzed C−H activation step in their synthesis and verified that the correct regioselectivity can be predicted using our directing strength scale.

The first example is a natural product, penchinone A, recently isolated from Penthorum chinense, and it was found to have anti-cancer and anti-inflammatory properties (He et al., 2015). The synthesis of this compound and several derivates has since then been achieved through palladium-catalyzed acylation of compound 35 (Oh et al., 2017). In this molecule, two DGs compete: an acetate and an oxime (Figure 14.1). Based on the relative energies of the corresponding fragments' intermediates, the oxime DG is stronger than the acetate, which is in accordance with the reported product of acylation.

Figure 14.

Figure 14

Open in a new tab

Drug Precursors/Analogues That Were Obtained Using Pd-catalyzed C−H Activation

Experimental C−H activation site is marked by a black circle; predicted site of activation is marked by a green-filled circle; DGs are highlighted with color. See rest of compounds in Data S5.

The second example is celecoxib 36, an anti-inflammatory drug, and its analogues (Figure 14.2). The two competing groups are the sulfonamide and the pyrazole, both strong DGs. In the study by Dai et al. (2011), a variety of couplings were performed under basic conditions, such that the sulfonamide is expected to be deprotonated. Negatively charged groups have a stronger coordination to palladium; thus, the sulfonamide will be the winning DG in this case, which is in line with experimental observations.

As demonstrated by the many examples given earlier, the model described herein yields accurate predictions of reactive sites on complex molecules, which should allow chemists to more readily apply this reaction.

With the introduction of late-stage functionalization into mainstream chemistry, regioselectivity prediction became an even more challenging exercise. In the field of metal-catalyzed C−H activation, one of the most successful approaches to increase regioselectivity is to use DGs. However, when such a group can direct reactivity to several sites or when multiple DGs are present in the reactant, accurately predicting which carbon will be activated can be problematic. With little literature reports that compare different DGs, the experimentalist is left to rely on experience and intuition to make synthetic decisions. In our study of palladium-catalyzed directed C–H activations, we offer a scale of the relative strengths of common functional groups and their relative capacity to ortho-direct palladium-catalyzed aromatic C−H activation. We demonstrated that, although the use of fragments and intermediates instead of full molecules and transition states may seem like a dramatic simplification, comparing only the relative energies of corresponding palladacycles allows one to quickly estimate which position is most likely to react. Additionally, our scale is able to capture the shifting reactivity at different pH. With over a hundred common DGs examined, the full scale enables one to make regioselectivity predictions on complex molecules in a flash, as well as encourages to try new unprecedented combinations of functional groups leading to unusual compounds.

Limitation of the Study

The method presented herein was developed to compare the strength of ortho-directing groups for the activation of unsubstituted aromatic carbons by Pd(OAc)2 model catalyst. This approach is not directly transferable to meta-directing activation or to hydrogens bound to non-aromatic carbons or to heteroatoms. Additionally, although we demonstrate that the strength of a directing group depends on its protonation state, the current model does not compute the pKa of directing groups, and users need to decide by themselves whether the group they are interested in would be deprotonated at the reaction conditions they will use to apply the correct scale. Finally, the sensitivity of the model was found to be around 1 kcal/mol: if the relative energies of two directing groups are within this range, then the model cannot be applied to know which group would lead to the main product of the reaction.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Acknowledgments

We would like to thank Pernilla Korsgren for her support. A.T. and M.E.M. are fellows of the AstraZeneca postdoc program.

Author Contributions

Conceptualization, P.-O.N., M.J.J., A.T., and I.T.; Methodology, P.-O.N., M.J.J., M.E.M., and A.T.; Software, A.T.; Validation, A.T., M.E.M., and P.-O.N.; Formal Analysis, A.T. and M.E.M.; Investigation, A.T. and M.E.M.; Data Curation, A.T. and M.E.M.; Writing – Original Draft, A.T., M.E.M., M.J.J., and P.-O.N.; Writing – Review & Editing, A.T., M.E.M., M.J.J., I.T., C.S., and P.-O.N.; Visualization, A.T., M.J.J., and M.E.M.; Supervision, P.-O.N., M.J.J., I.T., and C.S.; Project Administration and Funding, I.T., P.-O.N., and M.J.J.

Declaration of Interests

The authors declare no competing interests.

Published: October 25, 2019

Footnotes

Supplemental Information can be found online at https://doi.org/10.1016/j.isci.2019.09.035.

Supplemental Information

Document S1. Transparent Methods, Tables S1–S5, and Data S1–S16
mmc1.pdf (9.4MB, pdf)
Data S17. Cartesian Coordinates and DFT Energies

This File, in the SDF Format, Contains All the xyz Coordinates and the Corresponding DFT Energies of the Molecules Necessary to Compute the Directing Strengths Used to Get the Values in Data S1–S4, Related to Data S1–S4 and Scheme 2

mmc2.zip (188.8KB, zip)

References

  1. Abrams D.J., Provencher P.A., Sorensen E.J. Recent applications of C–H functionalization in complex natural product synthesis. Chem. Soc. Rev. 2018;47:8925–8967. doi: 10.1039/c8cs00716k. [DOI] [PubMed] [Google Scholar]
  2. Andersson T., Broo A., Evertsson E. Prediction of drug candidates' sensitivity toward autoxidation: computational estimation of C–H dissociation energies of carbon-centered radicals. J. Organomet. Chem. 2014;103:1949–1955. doi: 10.1002/jps.23986. [DOI] [PubMed] [Google Scholar]
  3. Bedford R.B., Webster R.L., Mitchell C.J. Palladium-catalysed ortho-arylation of carbamate-protected phenols. Org. Biomol. Chem. 2009;7:4853–4857. doi: 10.1039/b916724m. [DOI] [PubMed] [Google Scholar]
  4. Bell R.P., Hinshelwood C.N. The theory of reactions involving proton transfers. Proc. R. Soc. Lond. Ser. A. 1936;154:414–429. [Google Scholar]
  5. Bera M., Modak A., Patra T., Maji A., Maiti D. Meta-selective arene C–H bond olefination of arylacetic acid using a nitrile-based directing group. Org. Lett. 2014;16:5760–5763. doi: 10.1021/ol502823c. [DOI] [PubMed] [Google Scholar]
  6. Boele M.D.K., Van Strijdonck G.P.F., De Vries A.H.M., Kamer P.C.J., De Vries J.G., Van Leeuwen P.W.N.M. Selective Pd-catalyzed oxidative coupling of anilides with olefins through C−H bond activation at room temperature. J. Am. Chem. Soc. 2002;124:1586–1587. doi: 10.1021/ja0176907. [DOI] [PubMed] [Google Scholar]
  7. Cernak T., Dykstra K.D., Tyagarajan S., Vachal P., Krska S.W. The medicinal chemist's toolbox for late stage functionalization of drug-like molecules. Chem. Soc. Rev. 2016;45:546–576. doi: 10.1039/c5cs00628g. [DOI] [PubMed] [Google Scholar]
  8. Chen Z., Wang B., Zhang J., Yu W., Liu Z., Zhang Y. Transition metal-catalyzed C–H bond functionalizations by the use of diverse directing groups. Org. Chem. Front. 2015;2:1107–1295. [Google Scholar]
  9. Chiong H.A., Pham Q.-N., Daugulis O. Two methods for direct ortho-arylation of benzoic acids. J. Am. Chem. Soc. 2007;129:9879–9884. doi: 10.1021/ja071845e. [DOI] [PubMed] [Google Scholar]
  10. Chou H.-M., Jhou J.-N., Hong F.-E. Incorporation of norbornene moiety onto the arene of diaryl substituted amides through C–H functionalization. J. Organomet. Chem. 2017;853:178–183. [Google Scholar]
  11. Dai H.-X., Stepan A.F., Plummer M.S., Zhang Y.-H., Yu J.-Q. Divergent C–H functionalizations directed by sulfonamide pharmacophores: late-stage diversification as a tool for drug discovery. J. Am. Chem. Soc. 2011;133:7222–7228. doi: 10.1021/ja201708f. [DOI] [PubMed] [Google Scholar]
  12. Davies H.M.L., Morton D. Collective approach to advancing C–H functionalization. ACS Cent. Sci. 2017;3:936–943. doi: 10.1021/acscentsci.7b00329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Davies D.L., Donald S.M.A., Macgregor S.A. Computational study of the mechanism of cyclometalation by palladium acetate. J. Am. Chem. Soc. 2005;127:13754–13755. doi: 10.1021/ja052047w. [DOI] [PubMed] [Google Scholar]
  14. Davies D.L., Macgregor S.A., Mcmullin C.L. Computational studies of carboxylate-assisted C–H activation and functionalization at group 8–10 transition metal centers. Chem. Rev. 2017;117:8649–8709. doi: 10.1021/acs.chemrev.6b00839. [DOI] [PubMed] [Google Scholar]
  15. Desai L.V., Stowers K.J., Sanford M.S. Insights into directing group ability in palladium-catalyzed C−H bond functionalization. J. Am. Chem. Soc. 2008;130:13285–13293. doi: 10.1021/ja8045519. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dey A., Sinha S.K., Achar T.K., Maiti D. Accessing remote meta- and para-C(sp2)−H bonds with covalently attached directing groups. Angew. Chem. Int. Ed. 2019;58:10820–10843. doi: 10.1002/anie.201812116. [DOI] [PubMed] [Google Scholar]
  17. Dong J., Jin B., Sun P. Palladium-catalyzed direct ortho-nitration of azoarenes using NO2 as nitro source. Org. Lett. 2014;16:4540–4542. doi: 10.1021/ol502090n. [DOI] [PubMed] [Google Scholar]
  18. Evans M.G., Polanyi M. Further considerations on the thermodynamics of chemical equilibria and reaction rates. J. Chem. Soc., Faraday Trans. 1936;32:1333–1360. [Google Scholar]
  19. Gensch T., Hopkinson M.N., Glorius F., Wencel-Delord J. Mild metal-catalyzed C–H activation: examples and concepts. Chem. Soc. Rev. 2016;45:2900–2936. doi: 10.1039/c6cs00075d. [DOI] [PubMed] [Google Scholar]
  20. Gorelsky S.I., Lapointe D., Fagnou K. Analysis of the concerted metalation-deprotonation mechanism in palladium-catalyzed direct arylation across a broad range of aromatic substrates. J. Am. Chem. Soc. 2008;130:10848–10849. doi: 10.1021/ja802533u. [DOI] [PubMed] [Google Scholar]
  21. Gou F.-R., Wang X.-C., Huo P.-F., Bi H.-P., Guan Z.-H., Liang Y.-M. Palladium-catalyzed aryl C−H bonds activation/acetoxylation utilizing a bidentate system. Org. Lett. 2009;11:5726–5729. doi: 10.1021/ol902497k. [DOI] [PubMed] [Google Scholar]
  22. Guan Y., Ingman V.M., Rooks B.J., Wheeler S.E. AARON: an automated reaction optimizer for new catalysts. J. Chem. Theor. Comput. 2018;14:5249–5261. doi: 10.1021/acs.jctc.8b00578. [DOI] [PubMed] [Google Scholar]
  23. Gulevich A.V., Melkonyan F.S., Sarkar D., Gevorgyan V. Double-fold C–H oxygenation of arenes using PyrDipSi: a general and efficient traceless/modifiable silicon-tethered directing group. J. Am. Chem. Soc. 2012;134:5528–5531. doi: 10.1021/ja3010545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hammond G.S. A correlation of reaction rates. J. Am. Chem. Soc. 1955;77:334–338. [Google Scholar]
  25. Hansch C., Leo A., Taft R.W. A survey of Hammett substituent constants and resonance and field parameters. Chem. Rev. 1991;91:165–195. [Google Scholar]
  26. Hartwig J.F., Larsen M.A. Undirected, homogeneous C–H bond functionalization: challenges and opportunities. ACS Cent. Sci. 2016;2:281–292. doi: 10.1021/acscentsci.6b00032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. He Y.-C., Peng C., Xie X.-F., Chen M.-H., Li X.-N., Li M.-T., Zhou Q.-M., Guo L., Xiong L. Penchinones A–D, two pairs of cis-trans isomers with rearranged neolignane carbon skeletons from Penthorum chinense. RSC Adv. 2015;5:76788–76794. [Google Scholar]
  28. Hull K.L., Anani W.Q., Sanford M.S. Palladium-catalyzed fluorination of carbon−hydrogen bonds. J. Am. Chem. Soc. 2006;128:7134–7135. doi: 10.1021/ja061943k. [DOI] [PubMed] [Google Scholar]
  29. Jensen F. Introduction to Computational Chemistry. John Wiley & Sons Ltd; Chichester: 1999. Qualitative Theories. The Bell–Evans–Polanyi Principle/Hammond Postulate/Marcus Theory; pp. 364–368. [Google Scholar]
  30. Kalyani D., Sanford M.S. Regioselectivity in palladium-catalyzed C−H activation/oxygenation reactions. Org. Lett. 2005;7:4149–4152. doi: 10.1021/ol051486x. [DOI] [PubMed] [Google Scholar]
  31. Kalyani D., Deprez N.R., Desai L.V., Sanford M.S. Oxidative C−H activation/C−C bond forming reactions:  synthetic scope and mechanistic insights. J. Am. Chem. Soc. 2005;127:7330–7331. doi: 10.1021/ja051402f. [DOI] [PubMed] [Google Scholar]
  32. Kametani Y., Satoh T., Miura M., Nomura M. Regioselective arylation of benzanilides with aryl triflates or bromides under palladium catalysis. Tetrahedron Lett. 2000;41:2655–2658. [Google Scholar]
  33. Kanyiva K.S., Kuninobu Y., Kanai M. Palladium-catalyzed direct C–H silylation and germanylation of benzamides and carboxamides. Org. Lett. 2014;16:1968–1971. doi: 10.1021/ol500519y. [DOI] [PubMed] [Google Scholar]
  34. Kim B.S., Jang C., Lee D.J., Youn S.W. Highly effective Pd-catalyzed ortho olefination of acetanilides: broad substrate scope and high tolerability. Chem. Asian J. 2010;5:2336–2340. doi: 10.1002/asia.201000613. [DOI] [PubMed] [Google Scholar]
  35. Kim M., Kumar Mishra N., Park J., Han S., Shin Y., Sharma S., Lee Y., Lee E.-K., Kwak J.H., Kim I.S. Decarboxylative acylation of indolines with α-keto acids under palladium catalysis: a facile strategy for the synthesis of 7-substituted indoles. Chem. Commun. 2014;50:14249–14252. doi: 10.1039/c4cc06929c. [DOI] [PubMed] [Google Scholar]
  36. Kiser E.J., Magano J., Shine R.J., Chen M.H. Kilogram-lab-scale oxindole synthesis via palladium-catalyzed C–H functionalization. Org. Process. Res. Dev. 2012;16:255–259. [Google Scholar]
  37. Lapointe D., Fagnou K. Overview of the mechanistic work on the concerted Metallation–Deprotonation pathway. Chem. Lett. 2010;39:1118–1126. [Google Scholar]
  38. Lehecq A., Rousee K., Schneider C., Levacher V., Hoarau C., Pannecoucke X., Bouillon J.-P., Couve-Bonnaire S. Metal-catalyzed direct C–H fluoroalkenylation of pyridine N–oxides and related derivatives. Eur. J. Org. Chem. 2017;2017:3049–3054. [Google Scholar]
  39. Li W., Yin Z., Jiang X., Sun P. Palladium-catalyzed direct ortho C–H arylation of 2-arylpyridine derivatives with aryltrimethoxysilane. J. Org. Chem. 2011;76:8543–8548. doi: 10.1021/jo2016168. [DOI] [PubMed] [Google Scholar]
  40. Li H., Li P., Zhao Q., Wang L. Unprecedented ortho-acylation of azoxybenzenes with α-oxocarboxylic acids by Pd-catalyzed C–H activation and decarboxylation. Chem. Commun. 2013;49:9170–9172. doi: 10.1039/c3cc45492d. [DOI] [PubMed] [Google Scholar]
  41. Li Y., Ding Y.-J., Wang J.-Y., Su Y.-M., Wang X.-S. Pd-catalyzed C–H lactonization for expedient synthesis of biaryl lactones and total synthesis of cannabinol. Org. Lett. 2013;15:2574–2577. doi: 10.1021/ol400877q. [DOI] [PubMed] [Google Scholar]
  42. Li D., Xu N., Zhang Y., Wang L. A highly efficient Pd-catalyzed decarboxylative ortho-arylation of amides with aryl acylperoxides. Chem. Commun. 2014;50:14862–14865. doi: 10.1039/c4cc06457g. [DOI] [PubMed] [Google Scholar]
  43. Li C., Zhang D., Zhu W., Wan P., Liu H. Pd(ii)-catalyzed direct functionalization of C–H bonds of benzamides for synthesis of 1,1-difluoro-1-alkenes. Org. Chem. Front. 2016;3:1080–1083. [Google Scholar]
  44. Liao G., Chen H.-M., Shi B.-F. Synthesis of phthalic acid derivatives via Pd-catalyzed alkoxycarbonylation of aromatic C–H bonds with alkyl chloroformates. Chem. Commun. 2018;54:10859–10862. doi: 10.1039/c8cc06663a. [DOI] [PubMed] [Google Scholar]
  45. Liu S., Tzschucke C.C. Palladium-catalyzed regioselective dehydrogenative C–H/C–H cross-coupling of pyrroles and pyridine N–oxides. Eur. J. Org. Chem. 2016;2016:3509–3513. [Google Scholar]
  46. Liu Y.-J., Xu H., Kong W.-J., Shang M., Dai H.-X., Yu J.-Q. Overcoming the limitations of directed C–H functionalizations of heterocycles. Nature. 2014;515:389. doi: 10.1038/nature13885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Liu X.-H., Park H., Hu J.-H., Hu Y., Zhang Q.-L., Wang B.-L., Sun B., Yeung K.-S., Zhang F.-L., Yu J.-Q. Diverse ortho-C(sp2)–H functionalization of benzaldehydes using transient directing groups. J. Am. Chem. Soc. 2017;139:888–896. doi: 10.1021/jacs.6b11188. [DOI] [PubMed] [Google Scholar]
  48. Lou S.-J., Xu D.-Q., Xu Z.-Y. Mild and versatile nitrate-promoted C–H bond fluorination. Angew. Chem. Int. Ed. 2014;53:10330–10335. doi: 10.1002/anie.201404423. [DOI] [PubMed] [Google Scholar]
  49. Lyons T.W., Sanford M.S. Palladium-catalyzed ligand-directed C−H functionalization reactions. Chem. Rev. 2010;110:1147–1169. doi: 10.1021/cr900184e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. McMurray L., O'hara F., Gaunt M.J. Recent developments in natural product synthesis using metal-catalysed C–H bond functionalisation. Chem. Soc. Rev. 2011;40:1885–1898. doi: 10.1039/c1cs15013h. [DOI] [PubMed] [Google Scholar]
  51. Oh Y., Jang Y.J., Jeon M., Kim H.S., Kwak J.H., Chung K.H., Pyo S., Jung Y.H., Kim I.S. Total synthesis and anti-inflammatory evaluation of penchinone A and its structural analogues. J. Org. Chem. 2017;82:11566–11572. doi: 10.1021/acs.joc.7b02212. [DOI] [PubMed] [Google Scholar]
  52. Qiao Y., Zhao J., Chang J., Wei D. Insights into the oxidative palladium-catalyzed regioselective synthesis of 3-arylindoles from N−Ts–anilines and styrenes: a computational study. ChemCatChem. 2019;11:780–789. [Google Scholar]
  53. Roudesly F., Veiros L.F., Oble J., Poli G. Pd-catalyzed direct C–H alkenylation and allylation of azine N-oxides. Org. Lett. 2018;20:2346–2350. doi: 10.1021/acs.orglett.8b00689. [DOI] [PubMed] [Google Scholar]
  54. Sambiagio C., Schönbauer D., Blieck R., Dao-Huy T., Pototschnig G., Schaaf P., Wiesinger T., Zia M.F., Wencel-Delord J., Besset T. A comprehensive overview of directing groups applied in metal-catalysed C–H functionalisation chemistry. Chem. Soc. Rev. 2018;47:6603–6743. doi: 10.1039/c8cs00201k. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Shan G., Yang X., Ma L., Rao Y. Pd-catalyzed C–H oxygenation with TFA/TFAA: expedient access to oxygen-containing heterocycles and late-stage drug modification. Angew. Chem. Int. Ed. 2012;51:13070–13074. doi: 10.1002/anie.201207458. [DOI] [PubMed] [Google Scholar]
  56. Shi S., Kuang C. Palladium-catalyzed ortho-alkoxylation of 2-aryl-1,2,3-triazoles. J. Org. Chem. 2014;79:6105–6112. doi: 10.1021/jo5008306. [DOI] [PubMed] [Google Scholar]
  57. Snieckus V. Directed ortho metalation. Tertiary amide and O–carbamate directors in synthetic strategies for polysubstituted aromatics. Chem. Rev. 1990;90:879–933. [Google Scholar]
  58. Stephens D.E., Lakey-beitia J., Atesin A.C., Ateşin T.A., Chavez G., Arman H.D., Larionov O.V. Palladium-catalyzed C8-selective C–H arylation of quinoline N–oxides: insights into the electronic, steric, and solvation effects on the site selectivity by mechanistic and DFT computational studies. ACS Catal. 2015;5:167–175. doi: 10.1021/cs501813v. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Stephens D.E., Lakey-Beitia J., Chavez G., Ilie C., Arman H.D., Larionov O.V. Experimental and mechanistic analysis of the palladium-catalyzed oxidative C8-selective C–H homocoupling of quinoline N–oxides. Chem. Commun. 2015;51:9507–9510. doi: 10.1039/c5cc02227d. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Sun X., Shan G., Sun Y., Rao Y. Regio- and chemoselective C−H chlorination/bromination of electron-deficient arenes by weak coordination and study of relative directing-group abilities. Angew. Chem. Int. Ed. 2013;52:4440–4444. doi: 10.1002/anie.201300176. [DOI] [PubMed] [Google Scholar]
  61. Sun X., Yao X., Zhang C., Rao Y. Pd(ii) catalyzed ortho C–H iodination of phenylcarbamates at room temperature using cyclic hypervalent iodine reagents. Chem. Commun. 2015;51:10014–10017. doi: 10.1039/c5cc02533h. [DOI] [PubMed] [Google Scholar]
  62. Thirunavukkarasu V.S., Cheng C.-H. Pd-catalyzed multiple C−H functionalization to construct biologically active compounds from aryl aldoxime ethers with arenes. Chem. Eur. J. 2011;17:14723–14726. doi: 10.1002/chem.201102996. [DOI] [PubMed] [Google Scholar]
  63. Tian Q., Chen X., Liu W., Wang Z., Shi S., Kuang C. Regioselective halogenation of 2-substituted-1,2,3-triazoles via sp2 C–H activation. Org. Biomol. Chem. 2013;11:7830–7833. doi: 10.1039/c3ob41558a. [DOI] [PubMed] [Google Scholar]
  64. Tomberg A., Johansson M.J., Norrby P.-O. A predictive tool for electrophilic aromatic substitutions using machine learning. J. Org. Chem. 2019;84:4695–4703. doi: 10.1021/acs.joc.8b02270. [DOI] [PubMed] [Google Scholar]
  65. Tredwell M.J., Gulias M., Gaunt Bremeyer N., Johansson C.C.C., Collins B.S.L., Gaunt M.J. Palladium(II)-catalyzed C–H bond arylation of electron-deficient arenes at room temperature. Angew. Chem. Int. Ed. 2011;50:1076–1079. doi: 10.1002/anie.201005990. [DOI] [PubMed] [Google Scholar]
  66. Váňa J., Bartáček J., Hanusek J., Roithová J., Sedlák M. C–H functionalizations by palladium carboxylates: the acid effect. J. Org. Chem. 2019 doi: 10.1021/acs.joc.9b00462. [DOI] [PubMed] [Google Scholar]
  67. Wan L., Dastbaravardeh N., Li G., Yu J.-Q. Cross-coupling of remote meta-C–H bonds directed by a U-shaped template. J. Am. Chem. Soc. 2013;135:18056–18059. doi: 10.1021/ja410760f. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Wang Z., Kuninobu Y., Kanai M. Palladium-catalyzed oxirane-opening reaction with arenes via C–H bond activation. J. Am. Chem. Soc. 2015;137:6140–6143. doi: 10.1021/jacs.5b02435. [DOI] [PubMed] [Google Scholar]
  69. Wang P., Verma P., Xia G., Shi J., Qiao J.X., Tao S., Cheng P.T.W., Poss M.A., Farmer M.E., Yeung K.-S. Ligand-accelerated non-directed C–H functionalization of arenes. Nature. 2017;551:489. doi: 10.1038/nature24632. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Wang D.-Y., Guo S.-H., Pan G.-F., Zhu X.-Q., Gao Y.-R., Wang Y.-Q. Direct dehydrogenative arylation of benzaldehydes with arenes using transient directing groups. Org. Lett. 2018;20:1794–1797. doi: 10.1021/acs.orglett.8b00292. [DOI] [PubMed] [Google Scholar]
  71. Willis N.J., Smith J.M. An operationally simple, palladium catalysed dehydrogenative cross-coupling reaction of pyridine N–oxides and thiazoles “on water”. RSC Adv. 2014;4:11059–11063. [Google Scholar]
  72. Xiao B., Gong T.-J., Xu J., Liu Z.-J., Liu L. Palladium-catalyzed intermolecular directed C−H amidation of aromatic ketones. J. Am. Chem. Soc. 2011;133:1466–1474. doi: 10.1021/ja108450m. [DOI] [PubMed] [Google Scholar]
  73. Xiong F., Qian C., Lin D., Zeng W., Lu X. Palladium-catalyzed cascade oxidation/sp2 C–H acylation of azoarenes with aryl methanes. Org. Lett. 2013;15:5444–5447. doi: 10.1021/ol402537t. [DOI] [PubMed] [Google Scholar]
  74. Xu J., Liu Y., Wang Y., Li Y., Xu X., Jin Z. Pd-catalyzed direct ortho-C–H arylation of aromatic ketones enabled by a transient directing group. Org. Lett. 2017;19:1562–1565. doi: 10.1021/acs.orglett.7b00363. [DOI] [PubMed] [Google Scholar]
  75. Xue X.-S., Ji P., Zhou B., Cheng J.-P. The essential role of bond energetics in C–H activation/functionalization. Chem. Rev. 2017;117:8622–8648. doi: 10.1021/acs.chemrev.6b00664. [DOI] [PubMed] [Google Scholar]
  76. Yang S., Li B., Wan X., Shi Z. Ortho arylation of acetanilides via Pd(II)-catalyzed C−H functionalization. J. Am. Chem. Soc. 2007;129:6066–6067. doi: 10.1021/ja070767s. [DOI] [PubMed] [Google Scholar]
  77. Yang Y.-F., Cheng G.-J., Liu P., Leow D., Sun T.-Y., Chen P., Zhang X., Yu J.-Q., Wu Y.-D., Houk K.N. Palladium-catalyzed meta-selective C–H bond activation with a nitrile-containing template: computational study on mechanism and origins of selectivity. J. Am. Chem. Soc. 2014;136:344–355. doi: 10.1021/ja410485g. [DOI] [PubMed] [Google Scholar]
  78. Yu W.-Y., Sit W.N., Lai K.-M., Zhou Z., Chan A.S.C. Palladium-catalyzed oxidative ethoxycarbonylation of aromatic C−H bond with diethyl azodicarboxylate. J. Am. Chem. Soc. 2008;130:3304–3306. doi: 10.1021/ja710555g. [DOI] [PubMed] [Google Scholar]
  79. Zhang C., Ji J., Sun P. Palladium-catalyzed alkenylation via sp2 C–H bond activation using phenolic hydroxyl as the directing group. J. Org. Chem. 2014;79:3200–3205. doi: 10.1021/jo4028825. [DOI] [PubMed] [Google Scholar]
  80. Zhang J., Xu Q., Wu J., Fan J., Xie M. Construction of N–C axial chirality through atroposelective C–H olefination of N-arylindoles by palladium/amino acid cooperative catalysis. Org. Lett. 2019;21:6361–6365. doi: 10.1021/acs.orglett.9b02243. [DOI] [PubMed] [Google Scholar]
  81. Zhao X., Yeung C.S., Dong V.M. Palladium-catalyzed ortho-arylation of O-phenylcarbamates with simple arenes and sodium persulfate. J. Am. Chem. Soc. 2010;132:5837–5844. doi: 10.1021/ja100783c. [DOI] [PubMed] [Google Scholar]
  82. Zhu R., Lu S., Wang Q., Bai J., Wang Y., Yu Q., Huang J. Selectfluor-mediated mono-C–H activation: the syntheses of mono-ortho-substituted anilides. Tetrahedron. 2018;74:3879–3887. [Google Scholar]
  83. Daugulis, O. and Chiong, H. 2009. Use of aryl chlorides in palladium-catalyzed arylation of heterocycles, benzoates, and phenols. https://patents.google.com/patent/US20090012293A1/en. [DOI] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Transparent Methods, Tables S1–S5, and Data S1–S16
mmc1.pdf (9.4MB, pdf)
Data S17. Cartesian Coordinates and DFT Energies

This File, in the SDF Format, Contains All the xyz Coordinates and the Corresponding DFT Energies of the Molecules Necessary to Compute the Directing Strengths Used to Get the Values in Data S1–S4, Related to Data S1–S4 and Scheme 2

mmc2.zip (188.8KB, zip)

Articles from iScience are provided here courtesy of Elsevier

RESOURCES