Regioselectivity for the Rh(I)‐catalyzed Annulation of 1,2,3‐Thiadiazoles with Alkynes: Experimental and Computational Analysis Reveal the Surprising Role of the Alkyne Substituent

The Rh(I)‐catalyzed denitrogenative annulation reactions of 1,2,3‐thiadiazoles are a new direct approach to synthesizing densely functionalized heterocycles. The synthesis of multisubstituted thiophenes is possible using this methodology, however the difficulty to predict which regioisomer will form is a current limitation. In this current work, systematic computational and experimental studies were performed addressing how the terminal alkyne substituent effects regioselectivity. The data revealed that the electronic and steric properties of the alkyne substituent effects regioselectivity, including whether the group is aryl or alkyl. The insight gained from this study has allowed the development of a framework for predicting regioselectivity for the Rh(I)‐catalyzed denitrogenative transannulation of 1,2,3‐thiadiazoles with terminal alkynes.


Introduction
Heterocycles are essential to life and function, ranging from key scaffolds in nucleic acids, enzymes and vitamins, to their role in fundamental biochemical processes. [1]1a] Due to their vast applications, the development of efficient synthetic approaches to functionalized heterocycles has been at the forefront of chemical research for many decades.Of the various methods developed, annulation [6] and denitrogenative annulation have emerged as important strategies as they permit streamlined synthesis of heterocyclic compounds from a common, stable building block.
Among the different types of building blocks examined for denitrogenative annulation, diazocarbonyl compounds and 1,2,3-triazoles are relatively well explored, allowing a range of oxygen-and nitrogen-containing heterocycles to be accessed. [7]8a,9a,c,11a] In 2021, Bao examined the reaction mechanism using density functional theory (DFT) calculations, allowing a general reaction mechanism to be established (Figure 1a), as well as rationalizing the effect that the coupling partner (alkyne, alkene or nitrile) had on regioselectivity. [17]Recent experimental and computational studies by our group uncovered key insight into the role that the C5-substituent on the 1,2,3-thiadiazole plays in determining reactivity and regioselectivity (Figure 1b). [18]In combination, these recent reports highlight that the regiochemical outcome of the rhodium-catalyzed denitrogenative transannulation of 1,2,3-thiadiazoles is influenced by factors such as the ligand, solvent, and the electronic and steric properties of the 1,2,3thiadiazole.However, one factor that has not been considered is the role that the alkyne substituent plays in determining regioselectivity.
To address this limitation in our understanding of 1,2,3thiadiazole reactivity, we used a combination of computational and experimental studies to examine how the steric and electronic properties of the substituent on the terminal alkyne influences regioselectivity for the denitrogenative transannulation reaction (Figure 1c).This new insight, combined with insight from previous studies, [17][18] allowed development of a framework for a priori prediction of regioselectivity, which, in turn, allows more rational synthesis of the desired thiophene regioisomer.

Results and Discussion
Computational studies DFT calculations were employed to probe general trends on how the terminal alkyne substituent influences regioselectivity.
The general reaction mechanism for the Rh-catalyzed 1,2,3thiadiazole transannulation with alkynes involves initial 1,2,3thiadiazole ring-opening and subsequent denitrogenation, leading to formation of a cyclometalated Rh(III) intermediate (Figure 1a).Migratory insertion follows, proceeding through competing pathways leading to regioisomers 3 and 4 after reductive elimination.This current study is focused on the effect of the alkyne on regioselectivity, thus the relevant parts of the mechanism to consider in the computational studies are the migratory insertion and reductive elimination steps.
Thiadiazole 1 a was chosen as the model compound to examine (Figure 1c): previous studies reported that a mixture of regioisomeric products form using thiadiazole 1 a and phenylacetylene 2 a, [18] making it an ideal compound to probe the effect that the alkyne substituent has on regioselectivity.The aromatic alkynes chosen for computational assessment were: 1ethynyl-4-methoxybenzene 2 b, 1-ethynyl-4-(trifluoromethyl)benzene 2 c and 2-ethynyl-1,3,5-trimethylbenzene 2 d, to allow both electronic and steric effects to be probed.The aliphatic alkynes but-1-yne 2 e and 3,3-dimethylbut-1-yne 2 f were also selected to assess their influences on regioselectivity relative to the aryl alkynes, as well as the effect of steric hindrance.The reaction mechanism (based on that proposed by Bao [17] and recent studies in our group using 1 a and 2 a [18] ) was analyzed using CPCM(THF)PBE0 + D3/def2TZVP//ωB97XD/6-31G(d) + SDD for the reaction of thiadiazole 1 a with alkynes 2 a-f (Scheme 1, with key data summarized in Table 1

below).
Initially the data for phenylacetylene 2 a will be discussed in detail as a representative example (Scheme 1a).
The Gibbs free energy of activation for the reductive elimination step was found to be lower for thiophene 4 a formation via TS3a (ΔG � of 15.3 kcal mol À 1 ) relative to thiophene 3 a formation via TS4a (ΔG � of 19.2 kcal mol À 1 ).While this difference in ΔG � for reductive elimination could indicate a kinetic preference for thiophene 4 a formation via TS3a, this is unlikely to be the case as migratory insertion is not reversible: migratory insertion is exergonic, and from INT2a or INT3a reductive elimination is more kinetically feasible than the reverse migratory insertion process.As such, regioselectivity is likely determined at the migratory insertion step.Overall, the DFT studies suggest that for phenylacetylene 2 a, a mixture of products 4 a and 3 a are expected to form experimentally, due to the similar Gibbs free energy of activation for the competing migratory insertion pathways (ΔΔG of 0.6 kcal mol À 1 ).
To understand how changing the electronic properties of the alkyne substituent influences reactivity and regioselectivity, 1-ethynyl-4-methoxybenzene 2 b and 1-ethynyl-4-(trifluoromethyl)benzene 2 c were examined (Scheme 1b-c).Inclusion of the electron donating methoxy substituent (σ p = À 0.27) resulted in a slight decrease in the Gibbs free energy of activation for migratory insertion (TS1b and TS2b, Scheme 1b), relative to phenylacetylene 2 a (TS1a and TS2a, Scheme 1a), whereas the electron withdrawing trifluoromethyl group (σ p = 0.54) led to an increase in ΔG � (TS1c and TS2c, Scheme 1c).This data suggests that the rate of migratory insertion is increased when using an electron-rich alkyne substituent, and decreased when an electron-poor alkyne substituent is employed.In terms of regioselectivity, for both 2 b and 2 c, ΔG � for migratory insertion via TS1 or TS2 were similar (ΔΔG of 0.1 kcal mol À 1 and 0.6 kcal mol À 1 , respectively), and the process was exergonic.The similar ΔG � for migratory insertion suggests that a mixture of regioisomers would be expected to form for both 2 b and 2 c.However, it is important to highlight that for phenylacetylene 2 a, migratory insertion via TS1a (leading to 4 a) is more favourable, while for 1-ethynyl-4-(trifluoromethyl)benzene 2 c insertion via TS2c (leading to 3 c) is lower in energy.While these energy differences are minor, and difficult to distinguish due to the error associated with the DFT calculations, this trend may suggest that inclusion of an electron-poor substituent on the alkyne favours formation of regioisomer 3.
For 1-ethynyl-4-(trifluoromethyl)benzene 2 c, reductive elimination from either INT2c or INT3c is more kinetically feasible than the reverse migratory insertion process (Scheme 1c).Thus, migratory insertion is unlikely to be reversible for 2 c, and regioselectivity is likely induced at the migratory insertion step.In contrast, for 1-ethynyl-4-methoxybenzene 2 b, reductive elimination and reverse migratory insertion from INT3b have an identical ΔG � (17.8 kcal mol À 1 , Scheme 1b).This indicates that the migratory insertion step involved in regioisomer 3 b formation (black path) may be reversible, whereas formation of regioisomer 4 b does not involve a reversible migratory insertion step (blue path).Overall, the potential for reversible migratory insertion may lead to decreased formation of regioisomer 3 b, and thus an increase in selectivity for regioisomer 4 b formation may be expected.
For 2-ethynyl-1,3,5-trimethylbenzene 2 d, a substantial difference in the Gibbs free energy of activation for the competing migratory insertion steps was observed (ΔΔG of 3.4 kcal mol À 1 ), indicating that steric hindrance on the alkyne favours migratory insertion via TS2d (leading to regioisomer 3 d).While migratory insertion via both TS1d and TS2d is exergonic, reversible migratory insertion is only likely for TS2d (black path).In particular, from INT3d, reverse migratory insertion via TS2d is much more kinetically feasible than reductive elimination via TS4d (ΔΔG of 4.8 kcal mol À 1 ).This presents an interesting scenario -formation of regioisomer 3 d (black path) is predicted to be favoured over 4 d based on the lower ΔG � for migratory insertion, however the route to 3 d involves a reversible migratory insertion step, and thus 3 d formation is likely to be slow.
The computational assessment using the aliphatic alkynes but-1-yne 2 e and 3,3-dimethylbut-1-yne 2 f revealed a marked difference in the Gibbs free energy of activation for the competing migratory insertion steps (Scheme 1e-f), with insertion via TS2 (black path) favoured in both cases.The energy difference between TS1 and TS2 was more substantial for 2 f (ΔΔG of 4.5 kcal mol À 1 ) than for 2 e (ΔΔG of 2.3 kcal mol À 1 ), suggesting that steric hindrance favours migratory insertion via TS2, similar to the result for the aromatic alkyne 2 d.For alkynes 2 e and 2 f, migratory insertion was exergonic, and from INT2 and INT3 reductive elimination was more kinetically feasible than the reverse migratory insertion process.As such, migratory insertion is unlikely to be reversible, and regioselectivity for thiophene 3 e/3 f formation is expected to be induced at the migratory insertion step.Importantly, the data suggests that selective formation of regioisomer 3 is expected when using aliphatic alkynes.The key data and conclusions from the DFT studies are summarised in Table 1.

Experimental studies
To validate the results of the computational analysis, experimental studies were performed to probe how the alkyne substituent influenced the regiochemical outcome of the Rh(I)catalyzed denitrogenative transannulation of 1,2,3-thiadiazole 1 a with a series of alkynes 2. A direct comparison with reported studies was facilitated by employing reaction conditions developed by our group previously. [18]In particular, [Rh(COD)Cl] 2 was used as the Rh source and DPEPhos as the ligand, with the reaction performed in Me-THF for 6 h at 80 °C (Table 2).A series of terminal alkynes were examined, which were compared with reported data for phenylacetylene 2 a (Entry 1, Table 2).The aryl alkynes ethynyl-4-methylbenzene 2 g (σ p = À 0.17) and ethynyl-4-methoxybenzene 2 b (σ p = À 0.27) featuring an electrondonating group were examined first (Entries 2-3, Table 2).While a mixture of regioisomers 3 and 4 formed in both cases, there was a minor increase in selectivity for regioisomer 4, relative to that observed for phenylacetylene 2 a.This increased regioselectivity for 4 is consistent with our DFT studies, and likely arises from the formation of isomer 3 being less favourable due to the reversible migratory insertion step.Interestingly, for alkyne 2 b a lower combined NMR yield of products 3 b and 4 b was observed (74 %), relative to 2 a and 2 g.While no unreacted thiadiazole 1 a remained, the NMR spectrum for the reaction with 2 b showed a number of minor by-products (see Supporting Information), accounting for the slightly lower yield.The identity and origin of these by-products was not determined.
Inclusion of an electron-withdrawing trifluoromethyl substituent (σ p = 0.54) on the aryl alkyne (2 c, Entry 4, Table 2) led to an increased regioselectivity for product 3 c.Importantly, regioisomer 3 c is significantly favored over the formation of regioisomer 4 c (ratio 3 c : 4 c of 8.9 : 1).While this effect was difficult to conclude based on the computational studies, highlighting a limitation in the computational model used, the observed regioselectivity is likely due to the Gibbs free energy of activation for the migratory insertion step being significantly lower for the pathway that leads to 3 c, relative to the pathway that results in 4 c formation.High selectivity for isomer 3 was also observed when using the sterically hindered 2-ethynyl- 1,3,5-trimethylbenzene 2 d (Entry 5, Table 2), with a ratio 9.7 : 1 of 3 d : 4 d.This result confirms that steric hindrance on the alkyne results in migratory insertion into the RhÀ S bond via TS2 being substantially lower in energy than migratory insertion into the RhÀ C3 bond via TS1, leading to regioselective formation of isomer 3.In addition, the reaction between thiadiazole 1 a and alkyne 2 d was found to be very slow experimentally (Entry 5 required heating for 7 days), consistent with that predicted computationally -the migratory insertion step is reversible making product 3 d formation challenging.
Table 2. Experimental studies to determine how the substituent on the alkyne 2 influences regioselectivity.
[b] Yield determined by 1 H NMR spectroscopic analysis of the crude reaction mixture against the internal standard 1,3,5-trimethoxybenzene (0.1 mmol), see Supporting Information for all crude spectra, and characterization data of the isolated products 3 and/or 4.
[c] Reported data. [18][d] Only a trace amount of product formed after 6 h, thus the reaction was heated at 90 °C for 7 days -see Supporting Information for additional details.Unreacted thiadiazole 1 a (25 %) was present in the NMR spectrum.
[e] Unreacted thiadiazole 1 a (11 %) was present in the NMR spectrum as well as unidentified by-products, however no characteristic doublet for 4 i (J = 3.5 Hz) was detected.
The aliphatic alkynes 2 h-i were all expected to lead to regioselective formation of isomer 3 based on the DFT data (migratory insertion into the RhÀ S bond is favored over RhÀ C3 insertion).In all cases, high selectivity for product 3 was observed, with the ratio of 3 : 4 greater than 11 : 1 (Entries 6-8, Table 2).The reaction with methyl propiolate 2 i was examined as this alkyne features an electron-withdrawing ester group.The exclusive formation of 3 i observed (Entry 7, Table 2) confirms that electron-withdrawing groups favor isomer 3 formation for both aryl (2 c) and alkyl (2 i) substituted terminal alkynes.Structural confirmation of regioisomer 3 i was obtained via single crystal X-ray diffraction (Figure 2).A relatively low NMR yield of 3 i was observed (47 %), with the NMR spectrum showing unreacted thiadiazole 1 a (11 %) was present as well as unidentified by-products (see Supporting Information).
Changing from the linear heptyne 2 h to the bulkier ethynylcyclohexane 2 j did not lead to increased selectivity for regioisomer 3: this differs from the DFT studies, where increased bulk of the substituent was predicted to increase selectivity for 3 (Scheme 1e vs 1 f).However, the bulkier 3,3-dimethylbut-1yne 2 f was considered in the DFT studies, whereas the less bulky ethynylcyclohexane 2 j was examined experimentally (due to 2 f being unavailable), and thus greater regioselectivity for 3 might be expected when using a more bulky alkyl substituent.It is reasonable to conclude that use of either an alkyl substituent (e. g. 2 h-j) or a very bulky substituent on the alkyne (e. g. 2 d) results in high regioselectivity for regioisomer 3.In addition, a lower combined NMR yield was observed for ethynylcyclohexane 2 j (74 %) relative to heptyne 2 h (97 %).We tentatively suggest that the origin of this lower NMR yield is the increased steric bulk of the cyclic alkyne 2 j, relative to the linear alkyne 2 h, as the electronic properties of these alkynes are comparable.
Overall, there was generally good agreement between the regioselectivity trends predicted computationally and that found in the experimental studies.However, the high selectivity for isomer 3 when using the electron-withdrawing alkyne 2 c was not accurately predicted, nor was the differing reactivity observed in some cases (for example, the lower NMR yield when using alkyne 2 j).As such, while the computational model employed was sufficient for the current study, especially as experimental data was used in combination with the computational analysis, more robust computational predictions could likely be achieved through use of a higher level of theory and inclusion of explicit solvent effects.It is also important to highlight that in all cases examined, with the exception of alkyne 2 i, small amounts of regioisomer 3 formed, indicating that achieving exclusive formation of either regioisomer 3 or 4 is challenging.
Using the insight gained from this current work, in combination with previous studies, [8a,18] a guiding framework for predicting transannulation regioselectivity of commonly used C4 ester-substituted 1,2,3-thiadiazole 1 with terminal alkynes 2 has been developed (Figure 3).In particular, this figure summarizes: 1) the data presented in this current study (y-axis); and 2) data from a previous study [18] by our group, which demonstrated how the R 1 substituent on thiadiazole 1 impacts regioselectivity when using the alkyne 2 a (x-axis).Based on the insight gained from these two studies, reasonable extrapolations have been made to complete the table, noting that this model is mainly applicable to use of the Rh-DPEPhos catalytic system, as the ligand is known to have an impact, albeit relatively minor, on regioselectivity. [18]Figure 3 reinforces that in the majority of cases the isomer 3 is expected to be the major product, highlighting that regioselective synthesis of isomer 4 via alkyne insertion is considerably more challenging.9a]

Conclusions
In summary, we have reported a series of computational and experimental studies on how the alkyne substituent effects regioselectivity for the denitrogenative transannulation of 1,2,3thiadiazole 1 a with terminal alkynes 2. Use of an arylsubstituent featuring an electron-donating group was shown to  result in a mixture of regioisomers 3 and 4, however formation of isomer 4 is less favored as this pathway involves a reversible migratory insertion step.With this being the case, slightly increased selectivity for isomer 3 can be achieved through introduction of an electron-donating group on the aryl alkyne.
Inclusion of an electron-withdrawing group on the aryl alkyne had a more marked impact on regioselectivity, with high selectivity for isomer 3 observed experimentally.In addition, use of an aryl-substituent with greater steric bulk also led to high selectivity for regioisomer 3. The likely origin of the selectivity observed in each case was the Gibbs free energy of activation for the competing migratory insertion steps: RhÀ S insertion via TS2 (leading to 3) is more favored than RhÀ C3 insertion via TS1 (leading to 4) for both electron poor and bulky aryl substituents.A similar trend was observed when using alkyl-substituted alkynes, with high regioselectivity for isomer 3 observed for all aliphatic alkynes examined.
The mechanistic studies presented in this manuscript uncovered the central role the electronic and steric nature of the terminal alkyne plays in controlling regioselectivity.In combination with reported studies, it can be concluded that the type of coupling partner (alkyne, alkene or nitrile), [17] and the nature of the substituent on both the 1,2,3-thiadiazole [18] and the alkyne (this work), have a substantial impact on regioselectivity.This variety of factors shown to influence regioselectivity clearly highlights how susceptible the regiochemical outcome of the Rh(I)-catalyzed denitrogenation transannulation reactions of 1,2,3-thiadiazoles 1 is to reactant structure.The increased understanding of how each factor influences regiochemical outcome can now allow rational targeted regioisomer synthesis.We suggest that developing robust and widely applicable methodology to selectively access single regioisomers will require catalytic systems beyond Rh(I), which is an on-going focus in our group.
Figure 1.a) General reaction mechanism based on calculations by Bao; b) previous studies on how the 1,2,3-thiadiazole substituent influences regioselectivity; c) this work.
Scheme 1. Calculated reaction coordinate diagrams for the migratory insertion and reductive elimination steps of the Rh(I)-catalyzed denitrogenative transannulation of 1,2,3-thiadiazole 1 a with alkynes 2 a-f.Gibbs free energy is given in kcal mol À 1 .See Supporting Information for additional details.

Table 1 .
Summary of key data from the DFT studies presented in Scheme 1.