Efficient Pd‐Catalyzed Direct Coupling of Aryl Chlorides with Alkyllithium Reagents

Abstract

Organolithium compounds are amongst the most important organometallic reagents and frequently used in difficult metallation reactions. However, their direct use in the formation of C−C bonds is less established. Although remarkable advances in the coupling of aryllithium compounds have been achieved, Csp²−Csp³ coupling reactions are very limited. Herein, we report the first general protocol for the coupling or aryl chlorides with alkyllithium reagents. Palladium catalysts based on ylide‐substituted phosphines (YPhos) were found to be excellently suited for this transformation giving high selectivities at room temperature with a variety of aryl chlorides without the need for an additional transmetallation reagent. This is demonstrated in gram‐scale synthesis including building blocks for materials chemistry and pharmaceutical industry. Furthermore, the direct coupling of aryllithiums as well as Grignard reagents with aryl chlorides was also easily accomplished at room temperature.

A general protocol for the Pd‐catalyzed coupling of aryl chlorides with alkyllithium reagents is reported. The use of the ylide‐substituted phosphines joYPhos allowed for high selectivities at room temperature, thus minimizing side reactions typical for these highly reactive reagents. This was demonstrated for a large substrate scope including the gram‐scale synthesis of building blocks for pharmaceutical industry.

Introduction

Palladium‐catalyzed cross‐coupling reactions have become a powerful tool for the formation of C−C and C−X bonds in organic synthesis and are nowadays frequently used to produce fine chemicals such as agrochemicals and pharmaceuticals. ^[1] The success of these methodologies lies in the breadth of electrophiles and nucleophiles which can be used in this protocol and the myriad of reagents which are accessible. Albeit many limitations in coupling chemistry have been overcome over the years by the development of new, more sophisticated and specialized ligands, ^[2] still many challenges remain, which seem to be related to the intrinsic limitations in the ligand design. ^[3] In case of C${}_{\mathrm{sp}{}^2}$ −C${}_{\mathrm{sp}{}^2}$ and C${}_{\mathrm{sp}{}^2}$ −C${}_{\mathrm{sp}{}^3}$ coupling reactions the Suzuki–Miyaura, Stille, Kumada and Negishi couplings are the most important protocols. ^[4] Particularly, the Suzuki reaction which makes use of boron nucleophiles has become the method of choice for the formation of biaryl compounds, especially in pharmaceutical industry. This is mainly based on the high group tolerance and the facile access of boronic acids, esters or borates, which are typically generated from readily available organolithium or Grignard reagents. The same is also true for organotin compounds used in Stille couplings, which however are often toxic and thus less frequently applied. To facilitate synthetic protocols by preventing a further transmetallation step the development of reliable protocols with magnesium and lithium nucleophiles would be highly desirable. This would also allow for the reduction of inorganic waste (tin or boron salts) and production costs. While Kumada couplings ^[5] are routinely used since several years and reliable protocols for sp²‐sp² and sp²‐sp³ couplings have been developed with phosphines ^[6] as well as N‐heterocyclic carbenes[ ⁷ , ⁸ ] using nickel or palladium as well as base metals,[ ⁹ , ¹⁰ ] the corresponding reactions with organolithium compounds are far less developed. Feringa and co‐workers greatly improved the first protocols reported by Murahashi, ^[11] and in the case of aryl/alkenyl bromides, protocols exist to couple almost any conceivable lithium reagent even at low temperatures (Figure 1). ^[12] For the more desirable and readily available aryl chlorides, the coupling is much less general. Reagents like aryl/alkenyl lithium or (trimethylsilyl)methyllithium, that cannot undergo β‐hydride elimination (BHE) work efficiently, whereas primary and secondary alkyl lithium reagents are problematic. ^[13] Attempts with nickel‐based system were somewhat more successful, but the reaction seems to be restricted to polyaromatic aryl halides ^[14] and electron‐poor aryl chlorides. ^[15] Until today, there is no general protocol for the coupling of organolithium reagents with aryl chlorides available.

Figure 1.

Cross coupling of aryl and alkyl lithium reagents with aryl halides.

This limitation is mainly due to two obstacles in the mechanism (Figure 2) of the coupling reaction: i) Aryl chlorides are difficult substrates since oxidative addition of the C−Cl bond is usually rate‐limiting and slow at low temperatures. Thus, many side reactions with the highly reactive organolithium reagents can occur (e.g. Cl/Li exchange) prior to the addition step, leading to protodehalogenation and homocoupling. ii) The second limitation concerns the alkyllithium reagents. These reagents are more reactive than aryllithium species and hence more readily undergo the undesired Cl/Li exchange (homocoupling product). This was for example used in the synthesis of symmetric biaryls, where tert‐butyllithium instead of acting as nucleophile solely enabled lithiation of the aryl halide and hence biaryl formation. ^[16] Furthermore, the alkyl palladium species formed after the transmetallation step are prone to BHE, which can lead to isomerization and/or protodehalogenation. This competition between “normal” and migratory cross‐coupling of alkylmetal reagents has been addressed in recent studies, in particular in the context of Negishi coupling reactions, ^[17] but remained unexplored with lithium nucleophiles. Thus, the coupling of aryl chlorides with alkyllithium reagents represents the most challenging combination and an unresolved problem.

Figure 2.

Catalytic cycle of the Murahashi cross‐coupling reaction with possible side‐reactions.

To promote the coupling of aryl chlorides, very electron‐rich ligands are required which facilitate fast oxidative addition at mild temperatures. Therefore, Feringa and co‐workers used highly electron‐rich phosphines or N‐heterocyclic carbenes (NHC) to couple aryl and alkenyllithiums.[ ¹² , ¹³ ] Yet, these ligands seem to reach a limit in case of alkyllithium compounds. Recently, we reported on a new class of electron‐rich phosphines for homogenous catalysis, the ylide‐substituted phosphines, YPhos. ^[18] Due to the carbanionic charge next to the phosphorus center, these ligands are particularly strong donors, also surpassing the donor strength of NHCs. Accordingly, high activities in Buchwald Hartwig aminations ^[19] and α‐arylations of aryl chlorides at room temperatures were observed.[ ²⁰ , ²¹ ] This encouraged us to test these ligands for their activities in the coupling of organolithium compounds including alkyllithiums with aryl chlorides, which led us to the development of the first generally applicable reaction protocol.

Results and Discussion

As a first step, the coupling of sec‐butyllithium with 4‐chloroanisole (1) was studied using the three YPhos ligands L1–L3 together with Pd₂dba₃⋅dba as Pd source. The combination of an electron‐rich aryl chloride with a secondary alkyllithium reagent represents one of the most challenging reactions and—if being successful—promised a broad applicability of the optimized reaction protocol. We started with reaction conditions similar to those reported by Feringa et al., that is, room temperature and slow addition of the diluted organolithium reagent to a toluene solution of the aryl chloride and the catalyst. Optimization of the reaction conditions showed that 3 mol % catalyst loading are necessary for high conversion (Figure 3, see Supporting Information for details on the optimization). To our delight, YPhos ligand joYPhos (L3) gave high conversions of 78 % to the desired coupling product 2 a after only 1 h reaction time. While some protodechlorination product 3 (12 %) was formed, very small amounts of the homocoupling product 4 (4 %) and the isomerized product 2 b (3 %) were obtained. This is especially important, since separation of these isomers are often laborious using standard techniques. Other palladium sources and other YPhos ligands (L1 and L2) also showed high activities in this reaction, albeit with a slightly lower selectivity. In addition, air‐stable palladium complexes of L3, which are easy to apply showed similar results. ^[22] Here, [L3‐Pd(indenyl)Cl] (L3–P3) showed the highest activity, similar to L3 and Pd₂(dba)₃, and thus represents a convenient alternative to using the free ligand and an additional Pd source.

Figure 3.

Comparison of catalytic activity of different ligands and palladium complexes in the cross‐coupling of s‐BuLi with 4‐chloroanisole. Reaction conditions: 4‐Chloroanisole (1 mmol), Pd (3 mol %), ligand (3 mol %) and toluene (1 mL). s‐Buli (1.3 m in cyclohexane/hexane, 1.2 mmol; diluted with toluene), 1 h addition period at 22 °C (or 35 °C). Yields were determined by GC using n‐tetradecane as internal standard; Pd(dba)=[Pd₂(dba)₃⋅dba].

We compared the activity of the YPhos ligands with that of other catalysts, including those reported in the literature for the coupling of aryl and alkenyllithium compounds. As shown in Figure 3, none of these catalysts delivered comparable good results. While other phosphines (PtBu₃, QPhos, XPhos, DavePhos) gave no product at all, the NHC‐based PEPPSI catalysts delivered small amounts of product. Presumably, only the NHC catalysts can perform the oxidative addition at room temperature under the coupling conditions. However, conversions and selectivities were low with these catalysts, and for sBuLi, several unidentified higher molecular weight products were found, indicating further undesired side reactions. Increasing the reaction temperature to 35 °C to facilitate the oxidative addition of the aryl chloride didn't improve the selectivity or yield of the reaction. The YPhos ligand joYPhos (L3) is the only ligand that allows for a sufficiently fast oxidative addition of the aryl chloride at room temperature thus preventing or minimizing the competing homocoupling or protodechlorination, and at the same time minimizes BHE. To our delight, this optimized protocol with L3 was also capable of effecting the coupling with nBuLi (Table 1, Entry 1). In this case, only trace amounts (<0.1 %) of the isomerized product were found, and the amount of protodehalogenation (6 %) was likewise small. Again, none of the other catalyst systems (Figure S2) could produce similar good results. Only the NHC‐derived catalysts yielded the desired coupling product, but with low selectivities.

Table 1.

Results of the reaction of different organometallic reagents with aryl halides.

Entry	X	R−M	Conversion [%]	Product ratio			br:ln ratio
				2	3	4
1	Cl	nBuLi	93	90	6	4	–
2	Cl	sBuLi	95	85	12	3	96:4
3	Br	nBuLi	100	90	6	4	–
4	Br	sBuLi	100	84	9	7	88:12
5	I	nBuLi	100	57	6	37	–
6	I	sBuLi	92	3	4	93	78:22
7	Cl	tBuLi	99	65	20	16	0:100
8	Cl	iPrLi	86	83	12	5	95:5
9	Cl	PhLi	81	95	0	5	–
10	Cl	MeLi	83	98	0	2	–
11	Cl	CyMgCl	96	95	2	3	–
12	Cl	iPrMgCl	100	100	0	0	95:5
13	Cl	tBuMgCl	100	97	3	0	0:100
14	Cl	tBuMgCl[a]	100	94	6	0	0:100
15	Cl	MeMgCl	100	100	0	0	–
16	Cl	PhMgCl	100	100	0	0	–

Reaction conditions: 4‐Haloanisole (1 mmol), Pd₂dba₃⋅xdba (15.62 wt % Pd, 3 mol % [Pd]), L3 (3 mol %) and toluene (1 mL), lithium/Grignard reagent (see SI, 1.2 mmol) diluted with toluene (0.36 m) and added over 1 h at 22 °C. Yields and br:ln ratios were determined by GC using n‐tetradecane as internal standard. [a] Addition over 30 s, 1 h reaction time.

Motivated by the initial catalysis results, we turned our attention towards the examination of the generality of the established reaction protocol and a screening of different commercially available alkyllithium and magnesium reagents and aryl halides (Table 1). Aryl bromides were similar efficiently coupled to the desired products (entry 3 and 4), but interestingly the coupling with sBuLi showed slightly higher amounts of the isomerized product. Aryl iodides on the other hand showed only low conversion and poor selectivities (entry 5, 6), presumably due to fast iodine/lithium exchange. This suggests that for the successful coupling of alkyllithium reagents, activity towards aryl chlorides is particularly important to obtain high selectivities. While methyllithium and phenyllithium were also efficiently coupled (entry 8, 9), lower selectivities were found with tert‐butyllithium, which is more reactive towards lithium‐halogen exchange. Nonetheless, 65 % of product were formed (entry 7). Interestingly, instead of the expected product, complete isomerization to the iso‐butyl product was observed. No traces of the expected tert‐butyl product could be detected. Such a complete isomerization has been reported earlier for similar reactions. ^[22]

Besides C−C couplings with organolithium reagents, L3⋅Pd₂dba₃ is also an extremely efficient catalyst for Kumada coupling reactions (Entries 11–16). Due to the lower reactivity of Grignard reagents higher selectivities for the coupling products were obtained (c.f. entry 8 and 12 or 9 and 116) and secondary alkyl Grignard reagents, like iPrMgCl, also gave low amounts of the isomerized products. It is worth mentioning that tBuMgCl also selectively provided the isomerization product, but in this case, almost no proto‐dehalogenation or biphenyl formation was observed (entry 12). This led us to conclude that the lower selectivities observed wit tBuLi were mostly the result of fast side reactions, mainly lithium‐halogen exchange, while the complete isomerization of the alkyl group results from an inherent property of the catalytic system. We find it particularly interesting that the same catalyst provides high selectivities for the coupling of secondary alkyllithium and Grignard reagents with little isomerization, while at the same time giving complete isomerization in the coupling of tertiary lithium/Grignard reagents. It is also worth mentioning, that in case of tBuMgCl it was not necessary to dilute and slowly add the magnesium reagent. Instant addition of a 1.7 m solution of tBuMgCl in THF and stirring for 1 hour only slightly lowered the yield compared to a slow addition over a period of 1 hour. In contrast, instant addition of all other lithium and magnesium nucleophiles resulted in the formation of only small amounts of the desired product.

To gain insight into the reaction process and to understand the observed selectivities, we performed mechanistic studies. At first, we tested the oxidative addition of an aryl chloride to the palladium complex with L3 by means of a stoichiometric experiment using Pd₂dba₃ and L3 together with 10 equiv of 4‐chlorotoluene. Typically, elevated temperatures are required to reach significant conversion to the Pd^II chloride complex. With L3 however, the oxidative addition complex L3⋅Pd(Tol)Cl was formed at room temperature albeit one day was required for the reaction to complete as evidenced by monitoring of the reaction with ³¹P NMR spectroscopy. The complex could be isolated in 63 % yield as an air‐stable solid, which is poorly soluble in common organic solvents. In the solid state, L3⋅Pd(Tol)Cl forms a dimer (Figure 4) with a planar (Pd−Cl)₂ four membered ring as was observed for other phosphine and carbene ligands. ^[23] Attempts to obtain any information about the transmetallation process revealed to be impossible at room temperature. Addition of n‐butyllithium to [L3⋅Pd(Tol)Cl]₂ immediately resulted in the final coupling product. This was also the case when using methyllithium, even though the methyl palladium complex should be less prone to reductive elimination due to the decreased steric bulk. No intermediate compounds could be observed by NMR studies even at low temperature, thus confirming the fast reaction process. It is also important to note that, isolated [L3⋅Pd(Tol)Cl]₂ can be used as a potent precatalysts. In the coupling of sec‐butyllithium with 4‐chloroanisole under the optimized reaction conditions as discussed above, 87 % conversion with slightly lower selectivities than those found with L3‐Pd₂(dba)₃⋅dba or precatalyst L3–P3 were observed.

Figure 4.

Molecular structure of L3⋅Pd(Tol)Cl.

Since experimental studies could not provide any further information about the mechanism, DFT calculations (PW6B95D3/def2TZVP) were performed to understand the observed selectivities. To this end, we examined the possible reaction steps after transmetallation (Figure 5). From here onwards, complex I can either undergo reductive elimination (RE) to the desired product Pro or reversible BHE to form the isomerized products IsoPro. These reactions steps were calculated with keYPhos (L1) and joYPhos (L3) with tert‐butyl (a), sec‐butyl (b), iso‐propyl (c) and ethyl (d) as alkyl groups. Figure 5 depicts the potential energy surface for the energetically most favored pathways for alkyl complexes Ia‐d with L3 (see SI for more details, for example, energies related to rotamers and isomers as well as all results with L1). The calculations show that there are three distinct scenarios characterized by the energy differences between the transition states of the two reductive elimination processes (TS_RE and TS_Iso‐RE) and the isomerization (olefin rotation TS_III‐IV) 1). For the tert‐butyl palladium complex Ia (red line) the BHE and isomerization process is the kinetically most favored pathway, showing a considerably lower activation barrier (22.8 kJ mol⁻¹ for TS_Iso‐RE) than the RE to the non‐isomerized product TS_RE (33.9 kJ mol⁻¹). This energetic difference is in line with the complete isomerization observed in experiment. In contrast, for the ethyl complex Id, BHE to the hydrido palladium complex III is energetically uphill and rotation of the olefin more energy‐consuming than TS_RE (ΔΔG ^ǂ=19 kJ mol⁻¹), resulting in the selective RE without isomerization as observed in experiment for n‐BuLi.

Figure 5.

Reaction profile of the reductive elimination and β‐hydride elimination from the alkyl Pd complexes I with L3. Energies (PW6B95D3/def2TZVP) are given relative to the respective alkyl complex Ia–Id.

An intermediate situation to both extremes is found for the iPr and sBu complexes Ic and Ib (green and purple line, only most favored conformers shown). Here, the barrier for the rotation of the olefin is energetically similar to the RE. Thus, all three transition states determine the product formation. No classical Curtin–Hammett conditions are reached. The activation barriers TS_RE, TS_III‐IV and TS_Iso‐RE lie within only 1–5 kJ mol⁻¹, being in line with the observed formation of product mixtures. Although the calculations reflect the trends observed in experiment, the observed selectivities for the secondary alkyl groups should be lower. The reason for this discrepancy is not yet clear. Although inaccuracies of the DFT calculations cannot be ruled out, it might also be possible, that due to the high speed of the reaction (note that all activation barriers are very low) the equilibrium state is not always reached. Therefore, the reaction may simply continue from complex I to product Pro before equilibrating to the other agostic intermediate V.

Overall, steric effects seem to be the main reason for the observed selectivities since the least sterically demanding alkyl groups provide the highest selectivities for the desired product. However, also the difference in the strengths of the agostic interactions between the Pd center and the PCy₃ moiety of the ligand (complex I) and the alkyl group (complex II) seem to play a crucial role in this context. The more stable the agostic interaction to the phosphonium group, the less favored the interaction with the alkyl substituent and hence BHE. Since the agostic interaction to the tert‐butyl group is clearly favored (by 25.7 kJ mol⁻¹), the tert‐butyl group selectively undergoes isomerization to form IsoPro. For the complexes with the primary and secondary alkyl groups, the alkyl palladium species I is either more stable or only slightly disfavored over intermediate II, suggesting that the agostic interactions by the ligand and the alkyl groups are equally strong. Here, the different activation barriers are decisive. The whole reaction profile with the YPhos ligands contrasts the behavior of NHC‐based catalysts, where only the barriers of the reductive elimination and BHE were found to be decisive for the final isomer ratio. ^[24]

To test the broadness of the reaction protocol, we next addressed a detailed study of the substrate scope (Figure 6). To this end, either a combination of L3 and Pd₂(dba)₃⋅dba (method A) or precatalyst [L3‐Pd(indenyl)Cl] (L3–P3) (method B) was used. It is important to note that this does not represent an optimization of the reaction conditions. Rather one of the two systems was randomly chosen before the reaction, since both seem to be equally active. Fortunately, a wide variety of aryl chlorides could be coupled via the optimized reaction protocol and the products isolated in moderate to excellent yields. Electron‐rich and electron‐poor aryl chlorides as well as heteroaryl halides could be alkylated. Likewise, dialkylations such as to 2 k are possible. Most remarkably, also 2‐chlorofluorobenzene was coupled in high isolated yields with n‐hexyllithium (2 m) and iso‐ and cyclopropyllithium (2 h and 2 i), respectively, showing no indication of Li/Cl exchange and competing aryne formation. The magnesium compounds also allowed for a surprising group tolerance. Nitriles and esters could be coupled (2 y, 2 z). For the secondary alkyl reagents, the br:ln ratios were generally good. In general, more electron poor aryl chlorides with substituents in meta‐ and ortho‐position formed larger amounts of the linear product (2 d, 2 h), as was previously observed by Organ and co‐workers in Negishi coupling reactions. ^[24a] Most importantly, sp²‐sp²‐bond formations are in general easily achieved using the corresponding aryllithium and magnesium compounds (2 r–2 u). Here, isolated yields greater than 80 % are obtained also with aryllithium reagent which were produced by direct lithiation, thus allowing the synthesis of more complex biaryls from simple precursors (2 t, 2 u). Furthermore, aryl bromides were also successfully coupled (2 q, 2 w), showing that the protocol is not limited to the chlorides.

Figure 6.

Substrate scope. Reaction conditions: aryl halide (3 mmol), [Pd] (0.09 mmol), nucleophile (3.6 mmol), 1 h, RT. See SI for details. Yields are of the isolated mixture of branched and linear products, if applicable. The br:ln ratios were determined using ¹H‐NMR‐Spectroscopy and GC analysis. [a] GC yield. [b] 7.2 mmol of nucleophile, 2 h addition time. [c] from tBuMgCl.

After having established the successful catalysis protocol, we turned our attention towards the synthesis of substrates useful for the formation of compounds in materials science and pharmaceuticals (Figure 7). The synthesis of these compounds has been described by other routes, sometimes in multiple step procedures. We were pleased to see that for example 3‐chlorothiophene can be readily alkylated with hexyllithium also in gram‐scale to compound 5, which is a typical precursor for the synthesis of polythiophenes. Likewise, compound 6, a building block for the synthesis of Siponimod (brand name Mayzent; a sphingosine 1‐phosphate receptor modulator for treating multiple sclerosis) could be synthesized in one step from ortho‐trifluoromethylchlorobenzene in 60 % isolated yield. This compound was first prepared via a three‐step procedure from cyclohexanone including a Pd‐catalyzed hydrogenation reaction ^[25] and a Negishi‐type coupling of the aryl bromide at elevated temperatures with 5 mol % of Pd(PtBu₃)₂. ^[26] Additionally, the introduction of the cyclopropyl group could be performed in gram‐scale to yield compound 7 which is a building block in the synthesis of Lesinurad, a uric acid salt transport protein 1 (URAT1) inhibitor for the treatment of gout. ^[27] These examples demonstrate that the Pd‐catalyzed alkylation of aryl chlorides with alkyllithium reagents could become a viable method for the synthesis of building blocks in chemical and pharmaceutical industries.

Figure 7.

(left) Synthesis of alkylated aromatics which are important building blocks in material science and pharmaceutical industry, yields of isolated products, reaction conditions according to Figure 5. (right) Selective alkylation of 1‐bromo‐4‐chlorobenzene with n‐butyllithium, GC yield with n‐tetradecane as internal standard.

Due to the facile coupling of aryl chlorides at room temperature, we envisioned that the coupling of aryl bromides should readily proceed at low temperatures. This would provide the possibility of two subsequent, different C−C coupling reactions by temperature control. Such a methodology has recently been reported by Feringa and Organ. ^[12d] However, no dialkylation reactions were possible, due to the reasons highlighted in the introductory section. Fortunately, 1‐bromo‐4‐chlorobenzene was found to be easily alkylated at the bromo position at −40 °C. Subsequent warming to room temperature gave compound 8 in high yields. This shows that the selectivity of the C−C coupling of bromo‐substituted aryl chlorides can easily be controlled by temperature, thus allowing step‐wise coupling reactions for example, different C−C couplings ^[28] (such as to 2 f) or C−C coupling followed by C−N coupling reactions, for which joYPhos (L3) and its palladium complexes also showed high activities. ^[19]

Conclusion

In conclusion, we reported on the first successful and broadly applicable protocol for the direct sp²‐sp³ coupling of aryl chlorides with alkyllithium reagents. While all the previously reported catalysts for Murahashi couplings with aryllithium reagents failed in this transformation, catalysts based on ylide‐substituted phosphines (YPhos) gave good to excellent yields with high selectivities. The reason for this superior performance is the ability of the catalyst to undergo the oxidative addition of the aryl chloride fast enough to prevent side reactions such as lithium halogen exchange, while also minimizing isomerization due to the ligand architecture. The breadth of application was shown by means of a variety of substrates including ortho‐fluorinated arenes as well as heteroaromatics. The catalyst was also shown to be highly effective in Kumada couplings as well as in gram‐scale isolations and the synthesis of building blocks in materials science and pharmaceutical industry. Overall, the described protocol offers the direct use of readily available aryl and alkyllithium reagents in coupling reactions without the need of a further transmetallation step and thus represents an easily applicable protocol, which reduces the production of additional salt wastes by sp²‐sp³ couplings in a single reaction step.

Conflict of interest

The authors have filed patent WO2019030304 covering the YPhos ligands and precatalysts discussed, which is held by UMICORE AG & Co. KG and products will be made commercially available from.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Supplementary

T. Scherpf, H. Steinert, A. Großjohann, K. Dilchert, J. Tappen, I. Rodstein, V. H. Gessner, Angew. Chem. Int. Ed. 2020, 59, 20596.