Mechanistic Insights for Acid‐catalyzed Rearrangement of Quinoxalin‐2‐one with Diamine and Enamine

Benzimidazoles and benzimidazolones can be efficiently synthesized via acid‐catalyzed rearrangements of 3‐aroylquinoxalin‐2‐ones with various nucleophiles. The detailed mechanisms of typical rearrangements of 3‐benzoylquinoxalin‐2‐one (1 Q) with 1,2‐benzendiamine (1 a, Type I) and with methyl 3‐aminocrotonate (1 b, Type II) in acetic acid solution are explored by extensive DFT calculations. Acetic acid acts as proton source and efficient proton shuttle during the catalysis. The former rearrangement is initiated by site‐selective N⋅⋅⋅C condensation at protonated benzoyl site of 1 Q, while the latter is initiated by two competing C⋅⋅⋅C nucleophilic additions at protonated imine and benzoyl sites of 1 Q that eventually leads to two isomeric products. Both rearrangements proceed via intramolecular SN(ANRORC) mechanism (sequential Addition of Nucleophile, Ring‐Opening and Ring‐Closing) involving spiro‐quinoxalinone intermediates, but with fundamentally different RORC steps via ring‐opening anilide cation and isocyanate cation species, respectively. A simple energetic rule is suggested to determine the type of rearrangement by counting the conjugated π‐electrons within the spiro‐forming ring that may stabilize the ring‐opening anilide cation species, which may enable further rational design of novel spiro‐quinoxalinone based rearrangements.

Benzimidazoles and benzimidazolones can be efficiently synthesized via acid-catalyzed rearrangements of 3-aroylquinoxalin-2ones with various nucleophiles. The detailed mechanisms of typical rearrangements of 3-benzoylquinoxalin-2-one (1 Q) with 1,2-benzendiamine (1 a, Type I) and with methyl 3-aminocrotonate (1 b, Type II) in acetic acid solution are explored by extensive DFT calculations. Acetic acid acts as proton source and efficient proton shuttle during the catalysis. The former rearrangement is initiated by site-selective N···C condensation at protonated benzoyl site of 1 Q, while the latter is initiated by two competing C···C nucleophilic additions at protonated imine and benzoyl sites of 1 Q that eventually leads to two isomeric products. Both rearrangements proceed via intramolecular S N (ANRORC) mechanism (sequential Addition of Nucleophile, Ring-Opening and Ring-Closing) involving spiro-quinoxalinone intermediates, but with fundamentally different RORC steps via ring-opening anilide cation and isocyanate cation species, respectively. A simple energetic rule is suggested to determine the type of rearrangement by counting the conjugated πelectrons within the spiro-forming ring that may stabilize the ring-opening anilide cation species, which may enable further rational design of novel spiro-quinoxalinone based rearrangements.
Benzimidazole is a ubiquitous heterocyclic moiety in natural and synthetic compounds. Benzimidazole derivatives with substituents particularly at the N-1 and/or C-2 positions have received paramount interest in recent times because of their broad range of biological functions and pharmacological applications. [1] 2-Substituted benzimidazoles are usually synthesized via either the Phillips-Ladenburg coupling reaction [2] of 1,2-benzenediamines and carboxylic acid derivatives under acidic and high-temperature conditions or the two-step Weidenhagen condensation-aromatization reaction [3] of 1,2benzenediamines or o-nitroaniline with aldehyde or ketone. Benzimidazol-2-ones also exhibit a wide range of biological activities [4] but the benzimidazolone skeleton has received relatively less attention. [5] In view of their multi-dimensional activities, benzimidazol-2-ones have been the targets for both solid [6] and liquid phase [7] synthetic methods.
In 2000, we discovered the acid-catalyzed rearrangement of 3-aroylquinoxalin-2-ones into 2-(benzimidazol-2-yl)quinoxalines when exposed to 1,2-benzenediamines. [8] In the last two decades, we have published a series of works on the rearrangement of quinoxalin-2-ones via the proposed spiroquinoxalinone [9] intermediates (A) into 2hetarylbenzimidazoles [9a-h] (B, Type I) as well as 1-hetarylbenzimidazolones in cases of enamine nucleophiles (C, Type II, Scheme 1), [9i-k] thus opening a simple way to access various 2hetarylbenzimidazoles and 1-hetarylbenzimidazolones according to the S N (ANRORC) mechanism (i. e., via sequential steps of addition of nucleophile, ring-opening and ring-closing). [10] It was shown that the type of rearrangement may depend on the fact if tautomerizable hydrogen is available or not in the spiroforming ring within A. Although such mechanism was supported by the isolation of spiro-quinoxalinone [9a-c] or their ringopening anilide [9c-f] intermediates for Type I rearrangement, such intermediates were not observed for Type II rearrangement. The complete mechanism for such useful rearrangements remains unclear and highly desirable, especially for Type II rearrangement.
In this theoretical work, extensive DFT calculations at the PW6B95-D3 + COSMO-RS//TPSS-D3 + COSMO level in acetic acid solution (see below for computational details) are performed to explore the detailed free-energy paths for two typical rearrangements of 3-benzoylquinoxalin-2-one (1 Q) with 1,2-benzendiamine (1 a) [8] and methyl 3-amonocrotonate (1 b), [9i] which selectively leads to 2-(benzimidazol-2-yl)quinoxaline (Type I) and two N-pyrrolyl-substituted benzimidazol-2-ones (Type II), respectively. It is shown that while the former reaction is initiated by a site-selective N···C condensation at the protonated benzoyl group of 1 Q, the latter can be initiated by two C···C nucleophilic additions at both protonated benzoyl and imine sites of 1 Q that eventually leads to two isomeric products.
Moreover, though both reactions proceed via spiro-quinoxalinone mediated intramolecular S N (ANRORC) rearrangement, [10] the corresponding rearrangements of spiro-quinoxalinone intermediates are fundamentally different via ring-opening anilide cation [9c-f] and isocyanate cation species, [11] respectively. Based on DFT mechanistic insights, a simple energetic rule is suggested to determine the type of acid-catalyzed rearrangement of quinoxalinones, by counting the number of π-electrons within the spiro-forming ring that is directly related to the stability of ring-opening anilide cation intermediates.
In acetic acid solution, our DFT calculations show that diamine 1 a can only be singly protonated by AcOH into separated ion pair of 1 a + and AcOthat is À 1.2 kcal/mol exergonic, while the protonation of quinoxalinone 1 Q is 9.0 and 11.0 kcal/mol endergonic at the benzoyl C=O and imine C=N sites, respectively. As shown in Figure 1, starting from 1 Q and cation 1 a + , the protonation of 1 Q benzoyl C=O followed by nucleophilic addition of 1 a is 7.9 kcal/mol endergonic via transition structure TSB + over a low free energy barrier of only 16.3 kcal/mol to form the high-lying adduct B + . The addition is aided by the concerted proton transfer between two NH 2 groups of diamine 1 a; further intramolecular proton transfer to the resultant OH may lead to H 2 O-release from B + in two possible ways via transition structures TSC + and TSD + , leading to the E-and Z-isomers C + and D + with respect to the new C=N bond, respectively. The former way via TSC + is dominant due to 2.5 kcal/mol lower barrier. Such condensation reaction between 1 Q and 1 a + is thus À 1.5 kcal/mol exergonic with an overall barrier of 20.7 kcal/mol (TSC + ) to form the slightly more stable E-isomer C + . Alternatively, the condensation reaction can also be induced by the nucleophilic addition of 1 a at the protonated imine C=N site of 1 Q (the red path via Scheme 1. Two types of Mamedov rearrangement of quinoxalinone (top) and our DFT mechanistic insights for typical rearrangements of 3benzoylquinoxalin-2-one (1 Q) with 1,2-benzendiamine (1 a, Type I) and with methyl 3-aminocrotonate (1 b, Type II) in acetic acid solution. Both rearrangements proceed via spiro-quinoxalinone intermediates, with the type of rearrangement determined by the relative stability of ring-opening anilide and isocyanate structures that is usually related to the availability of tautomerizable hydrogen within the spiro-forming ring.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57 tsb + ) and followed by the ring-closing condensation at the benzoyl C=O site to form the 1.2 kcal/mol more stable spirostructure E + , which is however kinetically 3.9 kcal/mol less favorable via TSE + .
Although a high barrier can be expected for direct C=N bond rotation within the E-isomer C + , the E-to-Z conversion into D + is actually quite facile as aided by intramolecular N···C nucleophilic addition/cleavage (TSCd + ) over a low barrier of 13.8 kcal/mol. Facile N···C nucleophilic addition at the protonated imine C=N site (via TSF + ) then leads to the transient spiro-structure F + , which can be rapidly deprotonated by AcO À and then protonated by AcOH into the spiro-structure E + . On the other hand, relatively slower AcOH-aided proton-shift within F + (via TSG + a over a barrier of 21.0 kcal/mol from E + ) may lead to the ring-open anilide cation G + that is only 2.7 kcal/mol less stable than E + and can be further deprotonated by AcO À to form the low-lying neutral G observed in the reaction of 1 Q and 1 a in refluxing n-BuOH solution.
[9d] Further N···C ringclosing (via TSI + ) followed by AcOH-aided H 2 O-realease (via TSJ + a) eventually lead to the N-protonated structure J + , which is À 16.0 kcal/mol exergonic over a sizable barrier of 27.6 kcal/ mol with respect to the lowest-lying intermediate G. The intramolecular S N (ANRORC) rearrangement is kinetically 6.9 kcal/mol slower than the preceding condensation between 1 Q and 1 a + , making the isolation of low-lying spiro (E + ) and ring-opening G intermediates possible at low temperatures. The deprotonation of J + with AcO À is 3.9 kcal/mol endergonic to form 2-(benzimidazol-2-yl)quinoxaline 2 a thus is disfavored in acetic acid, which is however favored in water. [8,9d] Our DFTcomputed mechanism is thus fully consistent with the experimental results for Type I rearrangement of quinoxalinones. [9a-h] The reactions of quinoxalinone 1 Q with enamine 1 b is quite different from that with diamine 1 a, mainly due to the fact that 1 b is a C-centered nucleophile with a less mobile CÀ H proton. As shown in Figure 2, the protonation of 1 b with AcOH is nearly neutral in free energy. The C···C nucleophilic addition of 1 b to the imine C=N site of 1 Q protonated by AcOH is 10.0 kcal/mol endergonic over a moderate barrier of 19.7 kcal/ mol (via nTSA + ) to form the adduct nA + , with the enamine CÀ H proton being somewhat activated by the adjacent ester C=O and positively charged C=N double bonds. Intramolecular protonation of benzoyl C=O with the enamine CÀ H (via nTSB + ) makes further ring-closing (via nTSC + ) and AcOH-aided condensation (via nTSD + ) between enamine NH 2 and benzoyl C=O possible, over a moderate barrier of 21.5 kcal/mol (nTSC + ) above nA + to form the spiro-structure nD + . Note that the cations nB + , nC + and nD + can be further deprotonated by AcO À into to neutral nB, nC and nD. However, the formation of the lowest-lying intermediate nD from the condensation of 1 Q and 1 b is 2.2 kcal/mol endergonic over a sizable barrier of 31.5 kcal/mol (nTSC + ), consistent with the experimental difficulty to isolate such spiro-structures. Further protonation of nD at amine NH with AcOH may lead to the ring-opening cation nDo + , which is however 36.3 kcal/mol endergonic with respect to reactant 1 Q, clearly disfavoring the Type I rearrangement. [10] In contrast, direct amine shift of spiro-intermediate nD + via isocyanate cation [11] like transition structure nTSE + is kinetically about 10 kcal/mol more favorable to form the five-memberedring nE + , which after N-protonation with AcO À eventually leads to the observed [9j] N-(pyrrol-3-yl)benzimidazol-2-one (2 bn) product.
As shown in Figure 3, the condensation of quinoxalinone 1 Q and enamine 1 b can also occur via the facile N···C nucleophilic addition of 1 b at the protonated benzoyl C=O site of 1 Q (via oTSA + ) over a low barrier of 16.0 kcal/mol, which is however followed by slower proton shift from enamine CÀ H to imine C=N (via oTSB + ) and rate-limited by intramolecular N···C nucleophilic addition over a barrier of 31.0 kcal/mol (via oTSC + ) to form the high-lying spiro-intermediate oC + . Further H 2 Oelimination from oC + is exergonic and almost barrierless via AcOH-mediated proton transfer (oTSD + a) to form the spirostructure oD + , which is however 1.5 kcal/mol above the reactant 1 Q and thus thermodynamically slightly disfavored. Further ring-opening of oD + is 43.9 kcal/mol endergonic with respect to the reactant 1 Q, again disfavoring the Type I rearrangement. [10] Again, direct amine shift of spiro-intermediate oD + via isocyanate cation [11] like transition structure oTSE + is kinetically much more favorable to form the five-memberedring oE + , which after N-protonation with AcO À eventually leads to the observed [9j] N-(pyrrol-2-yl)benzimidazol-2-one (2 bo) product. The initial electrophilic additions of enamine 1 b at the protonated imine and benzoyl sites of 1 Q thus lead eventually to two competing N-(pyrrol- 3-yl) and N-(pyrrol-2-yl)-substituted benzimidazolone products 2 bn and 2 bo due to nearly the same rate-limiting barriers via nTSC + and oTSC + , respectively.
It is now clear that both acid-catalyzed rearrangements of 1 Q with diamine 1 a (Type I) and enamine 1 b (Type II) should proceed via spiro-quinoxalinone intermediates E + and nD + (or oD + ). The protonation of the spiro amine NH site of deprotonated E and nD (or oD) can be efficiently mediated by AcOH as proton source and proton shuttle to induce further ringopening and ring-closing steps for Type I rearrangement. It is thus all about the stability of ring-opening anilide cation structures that can be stabilized by the spiro-forming (diamine or enamine) ring: the new carbocation (old spiro-carbon) center can be well stabilized within the six-π-electron benzimidazole cation ring in the former case, but highly destabilized within the anti-aromatic four-π-electron pyrrole cation ring in the latter case. As the results, the latter rearrangement proceeds via energetically more favorable amine shift of spiro-quinoxalinones via isocyanate-like species instead. When tautomerizable hydrogen is present in the spiro-forming ring, ring-opening anilide cations can usually be stabilized via tautomerization to favor Type I rearrangement, as seen in the successful synthesis of 2-(pyrrol-3-yl)benzimidazoles. [9l] By counting the number of conjugated π-electrons within the spiro-forming ring that is directly related to the stability of ring-opening anilide cation intermediates, our new energetic rule is simple and general to determine the type of rearrangement of quinoxalinones. In conclusion, extensive DFT calculations are used to explore the detailed mechanisms of typical acid-catalyzed Mamedov rearrangements of 3-benzoylquinoxalin-2-one with 1,2-benzendiamine (Type I) and with methyl 3-amonocrotonate (Type II) in acetic acid solution. Acetic acid acts as efficient proton source and proton shuttle during the catalysis. Though both reactions proceed via intramolecular S N (ANRORC) mechanism involving spiro-quinoxalinone intermediates, the type of RORC rearrangement from spiro-quinoxalinones is determined by the relative stability of ring-opening anilide and isocyanate intermediates stabilized by the spiro-forming (diamine or enamine) ring. Based on DFT mechanistic insights, a simple energetic rule is suggested to determine the type of rearrangement, which may enable further rational design of novel spiro-quinoxalinone based rearrangement.

Computational Methods
All DFT calculations are performed with the TURBOMOLE 7.3 suite of programs. [12] The structures are fully optimized at the TPSS-D3/ def2-TZVP + COSMO (AcOH) level, which combines the TPSS meta-GGA density functional [13] with the BJ-damped DFT-D3 dispersion correction [14] and the def2-TZVP basis set, [15] using the Conductorlike Screening Model (COSMO) [16] for acetic acid solvent (dielectric constant ɛ = 6.19 and diameter R solv = 2.83 Å). The density-fitting RI-J approach [17] is used to accelerate the calculations. The optimized structures are characterized by frequency analysis (no imaginary frequency for true minima and only one imaginary frequency for transition states) to provide thermal free-energy corrections (at 298.15 K and 1 atm) according to the modified ideal gas-rigid rotorharmonic oscillator model. [18] More accurate solvation free energies in acetic acid are computed with the COSMO-RS model [19] (parameter file: BP_TZVP_C30_ 1601.ctd) using the COSMOtherm package [20] based on the TPSS-D3 optimized structures, corrected by + 1.89 kcal/mol to account for the 1 mol/L reference concentration in solution. To check the effects of the chosen DFT functional on the reaction energies and barriers, single-point calculations at both TPSS-D3 [13] and hybridmeta-GGA PW6B95-D3 [21] levels are performed using the larger def2-QZVP [15] basis set. Final reaction free energies (ΔG) are determined from the electronic single-point energies plus TPSS-D3 thermal corrections and COSMO-RS solvation free energies. As expected, the overall results from both DFT functionals are in good mutual agreement (À 0.2 � 3.0 kcal/mol, mean � standard deviation) though as expected 1.9 � 1.7 kcal/mol higher reaction barriers are found at the PW6B95-D3 level. In our discussion, the more reliable PW6B95-D3 + COSMO-RS free energies (in kcal/mol, at 298.15 K and 1 mol/L concentration) are used unless specified otherwise. The applied DFT methods in combination with the large AO basis set provide usually accurate electronic energies leading to errors for chemical energies (including barriers) on the order of typically 1-2 kcal/mol. This has been tested thoroughly for the huge data base GMTKN55 [22] which is the common standard in the field of DFT benchmarking.