Article pubs.acs.org/joc
Acid-Catalyzed Skeletal Rearrangements in Arenes: Aryl versus Alkyl Ring Pirouettes in Anthracene and Phenanthrene Sarah L. Skraba-Joiner, Jeffrey W. Brulet, Min K. Song, and Richard P. Johnson* Department of Chemistry and Materials Science Program, University of New Hampshire, Durham, New Hampshire 03824, United States S Supporting Information *
ABSTRACT: In 1 M triflic acid/dichloroethane, anthracene is protonated at C9, and the resulting 9-anthracenium ion is easily observed by NMR at ambient temperature. When heated as a dilute solution in triflic acid/dichloroethane, anthracene undergoes conversion to phenanthrene as the major volatile product. Minor dihydro and tetrahydro products are also observed. MALDI analysis supports the simultaneous formation of oligomers, which represent 10−60% of the product. Phenanthrene is nearly inert to the same superacid conditions. DFT and CCSD(T)//DFT computational models were constructed for isomerization and automerization mechanisms. These reactions are believed to occur by cationic ring pirouettes which pass through spirocyclic intermediates. The direct aryl pirouette mechanism for anthracene has a predicted DFT barrier of 33.6 kcal/mol; this is too high to be consistent with experiment. The ensemble of experimental and computational models supports a multistep isomerization process, which proceeds by reduction to 1,2,3,4-tetrahydroanthracene, acid-catalyzed isomerization to 1,2,3,4-tetrahydrophenanthrene with a predicted DFT barrier of 19.7 kcal/mol, and then reoxidation to phenanthrene. By contrast, DFT computations support a direct pirouette mechanism for automerization of outer ring carbons in phenanthrene, a reaction demonstrated previously by Balaban through isotopic labeling.
■
rearrangements have recently been described.11 We previously studied proton-catalyzed aryl group migrations.12 In the present work, we focus on the mechanisms for skeletal rearrangements in naphthalene and anthracene. The simplest nondegenerate arene skeletal rearrangement interconverts anthracene (16) and phenanthrene (10), the latter more stable thermodynamically by ca. 5 kcal/mol.13 This reaction does not occur during flash vacuum pyrolysis14 at 1200 °C but has been reported in medium pressure pyrolysis experiments,15 presumably as a consequence of radical catalysis. No mechanism was suggested in the conversion of 16 to 10 or a similar isomerization of tetracene to chrysene, which were observed as a consequence of high temperature passage of samples over an alumina/CrO3 catalyst.16 In the most detailed earlier study, Cook and Colgrove reported in 1994 the isomerization of anthracene by “fluid cat cracking” in which anthracene dissolved in petroleum was passed over a silica/ alumina catalyst at 482 °C.17 The observation that 17 and 18 are also formed led Cook and Colgrove to postulate that acidcatalyzed skeletal rearrangement occurs in these reduced structures (17 → 18) rather than by direct protonation of
INTRODUCTION The acid-catalyzed isomerization of substituted aromatic compounds by substituent migrations (Scheme 1) has been known for over a century.1 Interconversion of isomers (1−3, R = alkyl or aryl) is believed to occur through intermediate ipso arenium ions which result from protonation at sites bearing substituents.2 Fused saturated rings also migrate, as has been demonstrated through equilibration of octahydroanthracene (4) and octahydrophenanthrene (5)3 and through automerization of an isotopic label in tetrahydrophenanthrene (6).4 Skeletal rearrangements of fully aromatic polycyclic arenes might also be acid-catalyzed. The first reported example was the AlCl3-catalyzed isomerization of benz[a]anthracene (8) to chrysene (9), which was described by Dansi and Salvioni in 1941.5 Small amounts of phenanthrene (10) were also reported as a product in this reaction. Several more complex examples in longer chains were later reported by Zander6 and Buu-Hoi,7 who noted the generality and potential importance of this type of rearrangement. Following an erroneous report on automerization of 14C labeled naphthalene,8 Balaban demonstrated 13C label automerization in phenanthrene (11 → 15).9 Spirocyclic carbocation 13 is the most logical intermediate to explain transposition of the isotopic label. Carbon automerization in biphenyl occurs by 1,2-phenyl shifts and not by rearrangements within the benzene rings.10 Several more complex skeletal © 2017 American Chemical Society
Received: August 14, 2017 Published: November 14, 2017 13076
DOI: 10.1021/acs.joc.7b02058 J. Org. Chem. 2017, 82, 13076−13083
Article
The Journal of Organic Chemistry Scheme 1. Acid Catalyzed Rearrangements in Arenes
Figure 1. The 400 MHz 1H NMR spectrum of the 9-anthracenium ion.
anthracene. This requires multistep reduction and oxidation to complete rearrangement of 16 to 10. Our work began with the expectation that aryl ring pirouettes such as the mechanism proposed by Balaban (12 → 13 → 14) would provide a universal mechanism for skeletal isomerizations. As will be detailed below, our experimental and computational results support this automerization mechanism in phenanthrene; however, with anthracene, a more complex process17 is better supported by theory and experiment.
■
isotopologue, while the adjacent triplet (J = 4.7 Hz) is assigned as the HD isotopologue, with this multiplicity arising from H-D geminal coupling. Solutions of phenanthrene under the same conditions are bright red-orange, but the NMR spectrum at ambient temperature is only broadened, indicating incomplete protonation (Figure SI-2). We note that the facile formation of 19 does not exclude the formation at equilibrium of higher energy isomeric carbocations which are critical to the mechanisms described below. When dilute (1 mg/mL) solutions of anthracene (16) in 1 M TfOH/DCE were heated to reflux (84 °C) or to ca. 150 °C in a microwave reactor, the main volatile product recovered after neutralization was phenanthrene (10). MALDI spectroscopy of the crude product mixture showed that oligomerization also occurs, giving primarily dimeric material (Figure SI-4). Given the site of protonation and usual behavior of anthracene in electrophilic additions, the dimers are expected to be 9,9′bianthryls in various oxidation states. Minor volatile products 17−18, 20, and 21 were identified in the product mixture by NMR and capillary GC, with comparison to authentic samples. With the use of an internal standard added after reaction, the
RESULTS AND DISCUSSION
In strongly acidic media, anthracene (16) is protonated to give a green solution of the 9-anthracenium ion (19). Olah previously reported the NMR spectrum of this cation, which was generated under stable ion conditions.18 We find that 19-d is easily observed at ambient temperature; the well resolved 1H NMR spectrum (Figure 1) indicates complete cation formation in 1 M TfOD/DCE-d4 solution. Chemical shifts are in good agreement with predictions from DFT computation (Figure SI1) and the earlier report, which did not show the resonance for Hf.18 The upfield singlet at δ 5.2 is attributed to the H2 13077
DOI: 10.1021/acs.joc.7b02058 J. Org. Chem. 2017, 82, 13076−13083
Article
The Journal of Organic Chemistry Scheme 2. Acid Catalyzed Reactions of Anthracene and Phenanthrene: Volatile Product Distributions
Scheme 3. Superacid Catalyzed Reactions of Tetrahydroarenes: Volatile Product Distributions
Scheme 4. Computational Results for the Aryl Pirouette Mechanisma
a
B3LYP/6-31+G(d,p) relative free energies (kcal/mol) with solvation in DCE. In brackets, CCSD(T)/cc-pVDZ//DFT + ZPVE.
absolute yield of 10 did not exceed ca. 40%, with the balance of materials apparently oligomeric. In these experiments, we attribute no unusual effects to microwaves; as we have shown
earlier, the microwave reactor provides a safe means of heating superacid solutions.12 Under these same conditions, heating phenanthrene did not lead to anthracene or to significant 13078
DOI: 10.1021/acs.joc.7b02058 J. Org. Chem. 2017, 82, 13076−13083
Article
The Journal of Organic Chemistry Scheme 5. Computational Results for the Alkyl Pirouette Mechanisma
a
B3LYP/6-31+G(d,p) relative free energies (kcal/mol) with solvation in DCE. In brackets, CCSD(T)/cc-pVDZ//DFT + ZPVE.
5) which would interconvert tetrahydro structures 17 and 18. This indirect mechanism was proposed by Cook and Colgrove in 1994.17 We carried out computations to assess the comparative energetics of these two mechanisms. The potential surface for rearrangements was investigated with density functional theory using Spartan19 and Gaussian.20 All structures were optimized at the B3LYP/6-31+G(d,p) level of theory. This density functional has been widely used to model carbocation chemistry.21 The polarizable continuum model (PCM) was employed to assess implicit solvation in dichloroethane.22 In general, we find that DCE solvation provides significant absolute stabilization of the carbocation but does not significantly change relative energies of stationary points. Single point CCSD(T)/cc-pVDZ calculations were carried out at DFT optimized geometries. The CCSD(T) energies were corrected with DFT zero-point energies. In the rearrangements of naphthalene, MP2 and MP4//MP2 computations were also carried out for comparison. Results of DFT and (in brackets) CCSD(T)//DFT computations on the direct rearrangement surface which connects anthracene and phenanthrene are summarized in Scheme 4. As a reference point, cation 28 was arbitrarily set at zero because rearrangement should begin at this point. This is at energy significantly higher than that of 19, which is observed in solution (Figure 1). In the rearrangement mechanism, protonation of 16 at a ring juncture carbon yields 28, which proceeds through spirocyclic cation 29 and then on to protonated phenanthrene 25. An alternative higher energy bond migration from 28 would give 30. Structure 30 is not an energy minimum but opens spontaneously to give 31, which should deprotonate to give benz[a]azulene. We have seen no experimental evidence for this pathway. The initial barrier to rearrangement through TS5 is 33.6 kcal/mol according to DFT calculations and 27.8 kcal/mol from CCSD(T)//DFT. Energetics along the 28 to 25 pathway are consistent with a unidirectional process in which anthracene is converted to phenanthrene. On the phenanthrene side, the lowest energy pathway for automerization of the outer ring carbons, a rearrangement observed by Balaban and co-workers, passes through cations 24 and 27, as previously suggested.9 The predicted barrier between 24 and 27 is 19.5 kcal/mol according to DFT calculation; this estimate is diminished to 14.2 kcal/ mol by CCSD(T)//DFT.
amounts of oligomer, with a 93% recovery after chromatography. In addition to recovered starting material, small amounts of 21 were observed as the sole volatile product. A typical distribution of volatile reaction products in both reactions is shown in Scheme 2. When tetrahydro isomers 17 and 18 were subjected to the same reaction conditions, a similar mixture of volatile products was observed (Scheme 3). Notably, tetrahydroanthracene (17) afforded primarily rearranged products 10, 18, and 21 (total 67% of volatile material), while 18 gave primarily 10 and dihydro isomers 21 and 22, accompanied by ca. 10% of linear ring products. These experiments reveal the dominant direction of rearrangement (17 to 18) and support reoxidation to fully aromatic products under our reaction conditions. To distinguish among potential mechanisms, we measured the effect of initial anthracene concentration on reaction rate and found a non-first-order relationship, as measured by loss of reactant. In a series of experiments, the initial concentration of anthracene was varied, and the volatile product distribution from reaction at 150 °C was measured by gas chromatography after a constant 15 min of reaction. At the lowest reactant concentration (10−3 M), most of the remaining volatile material was reactant, while tetrahydro products 17 and 18 were barely detectable. As reactant concentration was increased (Figure SI3) to 0.07 M, the product mixture was primarily phenanthrene with ca. 10% each of 17 and 18. The total yield of these volatile products, as determined by addition of an internal standard after reaction, decreased as the initial concentration of anthracene increased, resulting from more material lost to oligomerization. These results support a mechanism in which oligomerization seeds formation of 17 and 18 as a route to phenanthrene. Direct isomerization of 16 to 10 may occur, but this clearly is not the major route.
■
COMPUTATIONAL STUDIES ON POTENTIAL MECHANISMS Two mechanisms seem likely in the rearrangement of anthracene to phenanthrene. The first mechanism, which arises from direct protonation of anthracene at a ring juncture carbon, is an aryl ring pirouette (Scheme 4) to give cation 28 and then on to 29. A similar process beginning with cations 24 or 25 can lead to automerization of phenanthrene ring carbons, as reported by Balaban and co-workers in 1989.9 The second mechanism is best described as an alkyl ring pirouette (Scheme 13079
DOI: 10.1021/acs.joc.7b02058 J. Org. Chem. 2017, 82, 13076−13083
Article
The Journal of Organic Chemistry
interconversion with 9,10-dihydroanthracene (20), one of the minor products observed here. This is likely to be rapid but reversible. Alternatively, protonation at C1 will lead through hydride transfer to dihydro derivatives 39 or 40, which can then proceed to 17 by a second hydride transfer to 41. For 39, this also requires a 1,2-hydride shift after protonation. Oxidation reverses the process. The most likely initial hydride donors to seed this chemistry are dihydroaromatic intermediates which are generated during anthracene dimerization, a competitive process which reduces isomeric product yields. Each new aryl− aryl bond releases one proton and one hydride. This assertion is supported by MALDI characterization of oligomeric products. On the Absence of Rearrangement in Naphthalene. Balaban coined the term automerization to describe the degenerate rearrangement of atoms in naphthalene, a reaction which ultimately proved elusive.8a Why? To confirm that this was not a consequence of low reaction temperature or insufficient acidity, we prepared C1-13C labeled naphthalene (44) by a known route,26 as shown in Scheme 7, and subjected this material to our TfOH/DCE microwave reaction conditions. After 60 min at 180 °C, no label migration was detected in the recovered starting material. Even heating with pure triflic acid at 220 °C for 30 min failed to demonstrate rearrangement in recovered naphthalene 44. The rearrangement of protonated naphthalene (Scheme 8) was modeled at several different levels of theory, and the results
Scheme 5 summarizes results for pirouette rearrangements in tetrahydro derivatives 17 and 18. In each case, migration of either an alkyl group or an aryl group is possible. The lower energy pathway clearly is alkyl migration, which connects cations 32 and 34 through TS7, 33, and TS8. Predicted barriers through spirocycle 33 are ca. 20 kcal/mol at both levels of theory. Reaction energetics again favor the phenanthrene side of the reaction coordinate. Aryl vs Alkyl Pirouette Mechanisms. The acid-catalyzed rearrangement of anthracene to phenanthrene can be explained either by an aryl ring pirouette (Scheme 4) or an alkyl ring pirouette (Scheme 5, top pathway). The latter mechanism requires a multistep redox process passing through 17 and 18, which was first proposed by Cook and Colgrove.17 Thus, including only the rearrangement process, our computations support feasibility of the Cook−Colgrove mechanism.17 Two literature results are also supported by these computations. First is Balaban’s observation of isotopic label automerization in phenanthrene (Scheme 1), which likely proceeds through cations 24 and 27.9 The predicted barriers through TS2 are less than 20 kcal/mol, which is consistent with the reported reaction conditions. Scott’s report4 of a similar skeletal rearrangement of isotopically labeled 18 would pass through 34, TS8, and 33. Our computations predict a barrier of ca. 16 kcal/mol for this process. The redox chemistry which is central to the Cook−Colgrove mechanism has ample precedent. In superacid media23 or under Scholl conditions,24 sequential proton and hydride transfer are well-documented.25 Scheme 6 shows a putative scheme for
Scheme 8. Mechanism for Naphthalene Automerization
Scheme 6. Redox Mechanisms for Formation of Tetrahydroanthracene
Table 1. Relative Free Energies (kcal/mol) for Automerization in Naphthalene structure
DFT (PCM)a
DFTb
CCSD(T)// DFTc
MP2d
MP4//MP2e
47 45 TS11 46
−19.4 0.0 33.4 27.1
−19.4 0.0 32.8 26.9
−17.9 0.0 26.8 22.6
−17.7 0.0 27.3 21.2
−17.6 0.0 26.4 23.0
a
ZPE corrected B3LYP/6-31+G(d,p) energy with PCM solvation in DCE. bZPE corrected B3LYP/6-31+G(d,p) energy. cCorrected energy at the CCSD(T)/cc-pVTZ//B3LYP/6-31+G(d,p) + ZPE (B3LYP/631+G(d,p) level). dCorrected energy at the MP2/cc-pVTZ. e Corrected energy at the MP4(SDTQ)/cc-pVTZ//MP2/cc-pVTZ + ZPE (MP2/cc-pVTZ) level).
double reduction of anthracene. In this scheme, RH is a dihydroaromatic hydride donor, and R+ is an arenium ion. Anthracene protonation at C9 provides a mechanism for Scheme 7. Attempted Automerization of Naphthalene
a
Polyphosphoric acid, 90 °C. bNaBH4. cTsOH/CH2Cl2. dDDQ. 13080
DOI: 10.1021/acs.joc.7b02058 J. Org. Chem. 2017, 82, 13076−13083
Article
The Journal of Organic Chemistry
observed automerization in naphthalene,8 reconfirmed here, is consistent with a high barrier to an aryl pirouette. Earlier literature and the results presented here indicate that acid-catalyzed skeletal isomerizations should be observed in many classes of arenes and can proceed by at least two general mechanisms. We will soon report results on higher homologues of anthracene and phenanthrene.
are summarized in Table 1. PCM solvation (DCE solvent) shows only a small effect on the relative energies for stationary points. The predicted DFT barrier from cation 45 is ca. 33 kcal/mol; this is very similar to TS5 (Scheme 4) for rearrangement of anthracene. CCSD(T)//DFT, MP2/MP2, and MP4SDTQ//MP2 computations all predict a barrier which is ca. 6 kcal/mol lower. Apparently, the cumulative barriers to protonation of naphthalene at C-9 followed by rearrangement cannot be surmounted under our reaction conditions. In this case, the absence of polymerization or detectable reduced naphthalene suggests that the polyhydroaromatic route is not available to naphthalene. This would seem to explain why neither Balaban,8b Staab,8c nor our group have seen evidence for naphthalene automerization.
■
■
EXPERIMENTAL SECTION
General Methods. Trifluoromethanesulfonic acid (TfOH, 99% purity) and dichloroethane (DCE, 99+%) were used as received from commercial sources. Glassware was oven-dried, and all reactions were run under a nitrogen atmosphere. 1H NMR spectra were measured in CDCl3 at 400 MHz. Capillary analytical gas chromatography was performed using a DB-3 column (30 m × 0.320 mm), with temperature programming from 75 to 250 °C. Retention times were as follows: 10: 34.8 min, 16: 35.2 min, 17: 33.8 min, 18: 34.2 min, 20:31.2 min, 21: 31.7 min. Microwave reactions were conducted using a single-mode microwave reactor in 10 mL vessels with temperature monitoring by an external sensor. Absolute yields were obtained via capillary GC with fluorene added after reaction as an internal standard or by isolation using column chromatography with hexane elution. Anthracene, phenanthrene, 9,10-dihydroanthracene and 9,10-dihydrophenanthrene were commercial samples. 1,2,3,4-Tetrahydroanthracene (17) and 1,2,3,4-tetrahydrophenanthrene (18) were prepared as reported previously.27 General Procedure for Rearrangement in Microwave (MW) Reactor. Under a nitrogen atmosphere, the substrate (ca. 4−50 mg) and 1,2-dichloroethane (DCE, 4 mL) were added to a 10 mL reaction vessel. Trifluoromethanesulfonic acid (TfOH) (0.40 mL, 4.5 mmol) was added dropwise by syringe; this typically caused formation of a bright color. The reaction mixture was purged with nitrogen, capped and heated in a microwave reactor. Products were isolated by careful neutralization with saturated aqueous NaHCO3 and extraction with dichloromethane, then analyzed using 1H NMR and capillary GC. The presence of oligomeric material was assessed on crude isolated product through time-of-flight matrix assisted laser desorption ionization (MALDI-TOF-MS) mass spectrometry, using sulfur as a matrix. Isomerization of Anthracene (16). Concentration 14 mg/mL: 16 (56 mg, 0.31 mmol) in DCE/TfOH was heated using a MW reactor (150 °C, 30 min) according to the general rearrangement procedure. Analysis of volatile products via capillary GC showed 10 (80%), 17 (14%), 18 (6%), and
