Review pubs.acs.org/acscatalysis
Understanding the Hydrodenitrogenation of Heteroaromatics on a Molecular Level Mark Bachrach,†,‡,§ Tobin J. Marks,*,† and Justin M. Notestein*,‡ †
Department of Chemistry and the Center for Catalysis & Surface Science and ‡Department of Chemical & Biological Engineering and the Center for Catalysis & Surface Science, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, United States ABSTRACT: The cleavage of recalcitrant C−N linkages in heteroaromatic molecules is a critical component of hydrodenitrogenation (HDN) reactions, the processes by which N atoms are removed from crude fuels. C−N bond cleavage of heteroaromatics with traditional HDN catalysts such as sulfided Ni/Mo-Al2O3 and Ni/W-Al2O3 requires aggressive conditions to hydrogenate the heterocyclic rings and often unnecessarily saturates the carbocyclic rings as part of the C− N cleavage process. In contrast, small-molecule models have illustrated selective hydrogenation of the heterocyclic ring, providing mechanistic insight into HDN processes and inspiring the design of new, more selective HDN catalysts. KEYWORDS: hydrodenitrogenation, C−N bond cleavage, hydrotreating, petroleum refining, organometallic models
1. INTRODUCTION Hydrotreating is the process by which heteroatom impurities (typically N, S, O, and metals) are removed from crude fuels. Under refinery conditions, removal of N, S, O, and metals occurs concurrently. To date, the emphasis of hydrotreating research has been hydrodesulfurization (HDS), since S is the predominant impurity in crude petroleum at an average of 0.65 wt %, while N is only present at an average of 0.094 wt %.1 Current research efforts typically focus on the production of gasoline and ultralow-sulfur diesel (ULSD) fuel, since most industrialized nations are expected to allow a maximum of 10 ppm of S in transportation fuels by 2020.2 Hydrodenitrogenation (HDN) has been traditionally viewed more as a necessary process in order to perform HDS and less as a goal in its own right. Nevertheless, HDN is increasing in importance with the move toward alternative crudes such as biomass, shale, coal, tar sands, and bitumen, as well as heavier fractions of traditional petroleum crude, which contain greater amounts of nitrogenous impurities. Nitrogen removal from crude fuels is not only essential to suppress NOx emissions and improve fuel combustion performance but also to enhance the performance of zeolitic cracking catalysis and other downstream processes.3 The processing of crudes with higher N content to meet more stringent governmental fuel specifications will require reactor modifications to operate at higher H2 pressures or temperatures or will require longer contact times, unless improved hydroprocessing catalysts are developed.4 Hydrotreating reactors typically operate near their maximum H2 pressures, and end-of-run temperatures often approach the reactor maximum, and thus enhancing their capabilities will © 2016 American Chemical Society
require major infrastructure changes. Since catalyst deactivation is typically offset by increasing the reactor temperature, improved catalysts which enable lower start-of-run temperatures should increase catalyst lifetimes. Note also that increasing catalyst contact times to achieve desired fuel specifications is highly undesirable, since that would significantly reduce reactor throughput. Hydrotreating has traditionally been performed with sulfided Co(Ni)/Mo(W)-Al2O3 catalysts which require aggressive conditions to cleave the C−N bonds: 260−393 °C start-ofrun temperatures and 200−2000 psi of H2, since the N atoms are often located within aromatic rings.2,3,5 In contrast to HDS, where direct desulfurization is often observed, HDN with current catalysts generally requires aromatic ring saturation prior to N atom extrusion and is thus highly dependent on H2 partial pressure.2 While new catalyst development has focused largely on tuning the combination of metals as well as the addition of promoters, there has also been research on using other metal sulfides, nitrides, phosphides, and borides which exhibit promising activity and selectivity.4 However, these bulk or high-metal-loading materials often contain many types of sites, and understanding the operative reaction mechanisms and pathways presents a formidable challenge. While recent TEM studies by Helveg et al. have provided atomic-scale resolution of MoS2 and brought significant insight into the catalyst structures,6−8 the exact role of each site under industrially Received: October 13, 2015 Revised: January 14, 2016 Published: January 19, 2016 1455
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
Figure 1. N-heteroaromatic model compounds and their hydrogenated analogues, representative of N-containing impurities within crude fuels and discussed in this review.
how the heteroaromatic coordination mode affects reactivity and have provided insight into relevant hydrogenation and C− N bond scission processes. Here we review the organometallic HDN models and welldefined heterogeneous catalysts relevant to HDN and aromatic C−N bond cleavage processes more generally, with the aim of discussing and analyzing studies which elucidate the elementary steps associated with metal-catalyzed HDN on a molecular level. Moving beyond older reviews of molecular models for HDN,10−16 which emphasized the coordination and reactivity of N heterocycles with groups III−V and VIII organometallic complexes, we focus on recent, unreviewed advances in homogeneous HDN models, supported molecular complexes for aromatic C−N hydrogenation, and C−N bond cleavage with isolated metal oxides. We also describe a representative sample of older work for context. The elementary steps associated with HDN are presented here in three sections: (1) model compounds that provide insight into the initial coordination modes of substrates to metal centers and the coordination modes enhancing reactivity, (2) reaction steps involved in heterocyclic ring hydrogenation, and (3) C−N bond hydrogenolysis. Ultimately, understanding HDN mechanisms and reaction pathways on a molecular level
relevant temperatures and H2 pressures is still poorly understood. In parallel to the above developments, there have been recent small-molecule studies aimed at understanding the coordination, hydrogenation, and hydrogenolysis steps within HDN catalytic cycles. Nevertheless, there is currently minimal conceptual connectivity between the heterogeneous and homogeneous efforts. Prior to recent work,9 the heterogeneous studies have not attempted to mimic the single-site nature of the homogeneous species under industrially relevant conditions. However, recent work with isolated metal oxide structures aims to bridge these two fields by demonstrating that it is possible to maintain the isolated structures which model the enhanced H2 efficiencies of the homogeneous species yet operate under industrially relevant HDN temperatures and H2 pressures. While the exact mechanism of metal-catalyzed crude fuel HDN is not completely understood, the heterogeneous catalytic studies combined with the organometallic models have helped in developing a general HDN mechanism. Heterogeneous catalytic studies have enabled an understanding of the kinetics and how altering process parameters and catalyst selection affect product yields and selectivities, while homogeneous organometallic studies have afforded an understanding of 1456
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
(CH2Xy-3,5)(THF) (NNfc = 1,1′-ferrocene(NSitBuMe2)2) leads to loss of mesitylene and formation of a metallacyclic intermediate by α-C−H activation without disrupting the pyridine aromaticity (Scheme 1). A second equivalent of 2phenylpyridine then displaces the THF to form the 2phenylpyridine dimer (4) by α-C−H activation.
using organometallic models can inform the development of new and improved catalysts.
2. SUBSTRATE COORDINATION Although there have been significant advances in understanding hydrotreating mechanisms, an exact model has not emerged due to the complexity of hydrotreating catalysts. Prins et al. have identified at least four types of sites on classical hydrotreating catalysts,17 and the substrate coordination mode varies depending on the catalyst and substrate. Proposed coordination modes with heterogeneous catalysts are typically derived from kinetic studies and observed product distributions with model compounds, and such studies have only been conducted with a small number of catalysts. In contrast, the homogeneous studies presented in this review discuss the coordination of N-heterocycles to metal centers across the periodic table. In this review, we discuss the chemistry of the Nheterocycles pyridine, quinoline, acridine, pyrrole, indole, and their derivatives, which are representative of the classes of Ncontaining compounds present in crude fuels. These aromatic N-heterocycles, the distribution of which will vary depending on the fraction and source of the crude, are shown in Figure 1. In general, HDN activity decreases with increasing substitution around the N atom. Aromatic substrates can coordinate via the heterocyclic ring, carbocyclic ring, side-on by the C−C bond, side-on by the C−N bond, or directly through the N atom (Figure 2, illustrating pyridine and pyridyl). The coordination
Scheme 1. Reaction of 2-Phenylpyridine with (NNfc)Sc(CH2Xy-3,5)(THF) (1) To Yield an η1(N) Complex (2) Which Undergoes C−H Activation Yo Yield an η2(N,C) Complex (3) and Subsequent C−H Activation Leading to a Phenylpyridine Dimer (4)
Substituted quinolines (Q) and isoquinolines (IQ) exhibit different coordination modes to group III metal centers. Diaconescu et al. observed that coordination of IQ to (NNfc)M(CH2Ar)(THF) (5; M = Sc, Lu, Y, La) leads to η1(N) coordination to displace a THF ligand (6), as expected for an electron-donating substrate and an electron-deficient metal center (Scheme 2).20 THF then recoordinates, causing the alkyl group to migrate to the isoquinoline α-C atom (7), thereby disrupting the aromaticity of the heterocyclic ring and presenting the first opportunity to selectively disrupt C−N bonds. In contrast, 8-methylquinoline (8-MQ) coordinates via the η2(N,C) mode to the metal center (9) (Figure 3).21 This may reflect differences in quinoline and isoquinoline N atom basicities. 2.2. Groups IV and V. N-containing heterocycles also coordinate in an η2(N,C) mode with group IV metals, which then leads to α-C−H activation and leaves the aromaticity intact. Titanium and zirconium complexes bind substituted pyridines from an η2(N,C) conformation to form metallacycles.22−24 Jordan et al. observed that 2-methylpyridine coordinates to Cp2Zr(CH3)(THF)+ (10) and displaces the THF ligand (11) (Scheme 3A).22 C−H activation of 2methylpyridine leads to the loss of CH4 and recoordination of a THF molecule (12). In the absence of an aromatic α-H, an sp3C−H bond undergoes activation to form a four-membered metallacycle, as observed for 2,6-lutidine (13) (Scheme 3B).23 Klei and Teuben observed that reaction of Cp2TiR (14; R = Me, nBu) with 2-methylpyridine, 2-phenylpyridine, 2-vinylpyridine, quinoline, or 8-methylquinoline in Et2O leads to αmetalation (15) at −40 °C for all of the ligands except 2-
Figure 2. Possible coordination modes of pyridine and pyridyl, as representative of all heteroaromatics, to a metal center.
mode depends on both the substrate sterics and electronics, as well as on the electronic nature of the metal center, and dictates whether the complex will undergo subsequent reactions. Understanding the substrate coordination mode is the first step in understanding the reactivity of the heterocycle for HDN and its mechanism. The coordination mode will ultimately dictate which ring(s) will be saturated prior to N atom(s) extrusion. In particular, if the substrate is coordinated via a carbocyclic ring, it is expected that harsher conditions will be necessary for total denitrogenation, since carbocyclic rings have greater resonance stabilization than heterocyclic rings. 2.1. Group III. Extensive work by Diaconescu et al. and Bercaw et al. has provided insight into the coordination modes and reactivity of N-heterocycles with group III transition-metal complexes. Early transition metals are most often present in their highest oxidation state and are thus highly electrondeficient and expected to prefer a coordination by the electronrich N atom. Scandium complexes (1) coordinate substituted pyridines in either an η1(N) (2) or η2(N,C) fashion (3).18,19 η2(N,C) coordination of 2-phenylpyridine to (NNfc)Sc1457
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
activation of pyridine by (PNP)TiCHtBu(CH2tBu) (PNP = N[2-P(CHMe2)2-4-methylphenyl]2) (16) (Scheme 5).25,26 The Ti alkylidyne eliminates tBuCH3 to form the reactive complex (PNP)TiCtBu (17), which then coordinates pyridine (18). The coordination of pyridine is exergonic with a driving force of 13.3 kcal/mol.25 The Ti alkylidyne disrupts the aromaticity via a cycloaddition reaction with pyridine to form a fourmembered ring (19). Similarly, Wolczanski et al. demonstrated that V(silox)3 and Ti(silox)3 complexes also exhibit η1(N) coordination of pyridine.27 In groundbreaking work for this field, Wolzcanski et al. characterized the binding of pyridine and its derivatives to Ta(silox)3 (21) at 25 °C.27,28 Pyridine binds in an η2(N,C) mode to form a metallaaziridine (22), thereby disrupting the aromaticity of the ring via back-bonding from Ta dπ orbitals to pyridine π* orbitals (Scheme 6). Note that Ta−C and Ta−N σ bond formation compensates for the disruption of the aromaticity. Increasing substrate alkylation from pyridine to 2,6-lutidine leads to C−H bond activation at the 4-position and addition to the Ta complex (25). When the sample was allowed to stand for a few days, the pyridyl hydride was observed to slowly equilibrate with the η2(N,C)-bound isomer (26). The pyridine likely coordinates in an η2(N,C) mode rather than in an η1(N) mode due to electron−electron repulsion between the filled Ta(silox)3 dz2 orbital and the pyridine N-donor orbital. The electron−electron repulsion as well as any entropic penalty from complexation are offset in the η2(N,C) mode by the favorable back-bonding from the Ta(silox)3 into the pyridine π* orbitals.27 Pyridazine also forms the anticipated η2 complex by both N atoms (23), but pyrimidine binds unexpectedly to the 3,4-site (24). Presumably, inductive effects from N(3) disfavor coordination to the 1,2-site. Wolczanski et al. showed that Nb reacts similarly to Ta and also exhibits an η2(N,C) coordination of pyridine.29,30 Wigley et al. studied the coordination of quinoline to Ta(OAr)3Cl2 (27) and similar complexes with pyridine, which are discussed in section 4.2.31,32 Quinoline coordinates to a Ta center in either an η1(N) (28) or an η2(N,C) (29) mode (Scheme 7). The η1(N) coordination is obtained since the Ta(V) ion has a d0 electron configuration, unlike d2 Ta(silox)3, which should exhibit significant electron−electron repulsion upon quinoline coordination in an η1(N) mode. At 100 °C, the η1(N)-bound quinoline is in rapid equilibrium with the free heterocycle in solution. When it is dissolved in pyridine-d5, quinoline is rapidly exchanged with pyridine, which also adopts an η1(N) configuration. In the presence of cold NaHg/Et2O, quinoline is rapidly converted to η2(N,C) coordination to form the metallaaziridine, thereby disrupting the aromaticity of the heterocyclic ring. Complexes 26 and 29 provide evidence that a redox-active Ta (or Nb) site may be a viable design strategy for C−N selective hydrogenolysis catalysts. 2.3. Molybdenum. Understanding possible coordination modes and reactivity at molybdenum centers is particularly relevant, due to the importance of MoS2 and related sulfides as conventional hydrotreating catalysts. Parkin et al. studied the coordination and activation of several N-heterocycles with molybdenum complexes. They demonstrated that pyridine and its derivatives (pyrazine, pyrimidine, and 1,3,5-triazine) coordinate in an η2(N,C) mode to the Mo center (Scheme 8, illustrated for pyridine). Thus, Mo(PMe3)6 (30) reacts with the α-C−H bond to form the complex (η2(N,C)-heterocycle)Mo(PMe3)4H (32).33 α-Activation of the aromatic C−H unit does
Scheme 2. Reaction of Isoquinoline with Group III Metal Complexes To Form η1(N) Complexes Followed by Alkyl Migration from the Metal to IQ To Disrupt the Heterocyclic Ring Aromaticity
Figure 3. Molecular structure of the (NNfc)Sc(η2(N,C)-8-MQ)(THF) complex.
Scheme 3. Coordination and C−H Activation of (A) 2Methylpyridine and (B) 2,6-Lutidine by Cp2Zr(CH3)(THF)+
vinylpyridine, where it occurs at 0 °C (Scheme 4, illustrated for 2-methylpyridine, R = Me).24 α-Metalation with these Scheme 4. Coordination and C−H Activation of 2Methylpyridine by Cp2TiCH3
complexes is limited to substrates with only a single α-H atom. If two α-H atoms are present as in pyridine, 3methylpyridine, or 4-methylpyridine, α-metalation does not occur. They suggest that the α-substituent forces the ligand into an orientation which favors α-metalation. Mindiola et al. proposed on the basis of DFT computations that η1(N)-coordinated pyridine is an intermediate in the 1458
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 5. Coordination and Activation of Pyridine at a (PNP)Ti Alkylidyne Complex
100 °C, these adopt η6-heterocyclic coordination (35, 38), while further heating to 130−150 °C promotes ring slippage to an η6-carbocyclic coordination mode (36, 39). Parkin et al. also reported that quinoline reacts with Mo(PMe3)6 (30) to form the (η6-C5N-Q)Mo(PMe3)3 structure (40) (Scheme 9C).34,35 In contrast to the case for pyridine, the η2(N,C) mode is not an observed intermediate in the pathway to η6-heterocyclic ring coordination for Q. At significantly higher temperatures, coordination to the carbocyclic ring of Q (41) yields a stable complex. Parkin et al. also showed that Mo(PMe3)6 (30) reacts with indole to sequentially yield (η1-indonyl)Mo(PMe3)4H (42), (η5-indonyl)Mo(PMe3)3H (43), and finally (η6-indonyl)Mo(PMe3)3H (44) (Scheme 10).34 Indole typically exhibits η1(N) or η5(C4,N) coordination, not coordination via the arene ring. Addition of the phosphine to the η5 complex shifts the equilibrium back to η1(N) and ultimately liberates indole, demonstrating that the N−H bond oxidative addition is reversible. Isomerization to η6 coordination requires heating to 80 °C and leads to a stable complex,34 since the C−N bond is situated far from the metal center and the aromatic ring occupies multiple coordination sites; this prevents H 2 coordination and oxidative addition to indole.11 However, isomerization from the η6- to the η5-coordinated complex can be achieved by photolysis. Coordination through the arene ring is unexpected, since it requires the formation of a zwitterionic complex (44), with a formal negative charge on the N atom and a formal positive charge of the Mo atom. Interestingly, the hydrogen is not transferred to the N atom to form the neutral (η6-indole)Mo(PMe3)3 complex. Two-ring heterocyclic complexes require high temperatures to achieve η6-carbocyclic coordination, since they prefer η2(N,C) or η6-heterocyclic coordination. In contrast, threering heterocycles most easily adopt η6-carbocyclic coordination. Thus, phenazine (PZ) and acridine (ACR) both coordinate in an η6 fashion to one of the outer carbocyclic rings.34,36,37 PZ forms (η6-C6-PZ)Mo(PMe3)3 (45) and (μ-η6,η6-PZ)[Mo(PMe3)]2 (46), with a PZ ligand bridging the two metal centers, and (η4-C4-PZ)2Mo(PMe3)2 (47), with two PZ ligands (Scheme 11A).36 Single-crystal X-ray diffraction indicates that, in all three configurations, each ligand only coordinates through the carbocyclic ring carbon atoms. ACR also coordinates to Mo(PMe3)6 (30) via one of the outer carbocyclic rings (48) but only in a 1:1 ligand to metal ratio (Scheme 11B). In contrast, η6-coordinated PZ and ACR are obtained under conditions significantly milder than those for the all-carbon analogue anthracene.36,37 This demonstrates that a N atom in the central ring enhances reactivity with respect to Mo(PMe3)6 (30) even though it does not coordinate via the N atom. The authors suggest that the N atom may assist ACR and PZ η6 coordination by initially coordinating the heterocycle in an η1(N) mode. Coordination of pyrrole and N-methylpyrrole has also been studied with a tetranuclear molybdenum sulfide complex (49)
Scheme 6. Coordination of (A) Pyridine (22), Pyridazine (23), and Pyrimidine (24) to Ta(silox)3 and (B) C−H Activation of 2,6-Lutidine (25) and η2(N,C) Coordination of 2,6-Lutidine (26) to Ta(silox)3
Scheme 7. Coordination and Activation of Quinoline by a Ta(OAr)3Cl2 Complex
Scheme 8. Coordination Modes of Pyridine and Pyrrole to Mo(PMe3)6
not disrupt the aromaticity of the ring, and pyridine, pyrazine, and pyrimidine reversibly undergo rearrangement to the more stable η6 coordination mode (33) at elevated temperatures. In contrast, pyrrole reacts with Mo(PMe3)6 to form (η5-pyrrole)Mo(PMe3)3H (31) and not an η2(N,C) complex (Scheme 8).34 The lone pair of pyrrole is conjugated with the ring π-system, and thus exclusive coordination of the N atom would not be expected. Parkin et al. also reported that isoquinoline (IQ) and quinoxaline (QX) react with Mo(PMe3)6 (30) at room temperature to form (η2(N,C)-IQ)Mo(PMe3)4H (34) and (η2(N,C)-QX)Mo(PMe3)4H (37) (Scheme 9A,B),35 similar to the coordination mode observed for pyridine. Upon heating to 1459
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
Scheme 9. Coordination and α-C−H Activation of (A) Isoquinoline and (B) Quinoxaline to Mo(PMe3)6 at Room Temperature, Change in Coordination to the η6-Heterocyclic Ring at 100 °C, and Ring Slip to the η6-Carbocyclic Ring at 130−150 °C and (C) Coordination and α-C−H Activation of Quinoline to Mo(PMe3)6 at 80 °C and Ring Slip to the η6-Carbocyclic ring at 150 °C
Scheme 10. Oxidative Addition and Coordination of Indole to Mo(PMe3)6
Scheme 11. Reported Coordination Modes of (A) PZ and (B) ACR to Mo(PMe3)6
Dinuclear complex 54 was treated under H2 at 25 °C in MeCN to determine whether the N-methylpyrrole ligand could be reduced to N-methylpyrrolidine or other species. However, the reaction led only to free N-methylpyrrole and 52, not Nmethylpyrrolidine (Scheme 13). Mixed-metal pyrrolyl complexes were also studied by Rakowski-DuBois et al. to determine how multiple metal coordination affects substrate activation, as is the case for promoters used in commercial catalysts. Addition of 49 to (PMe2Ph)3(Cl)2Re(NC4H4) (56) in CH2Cl2 at room temperature leads to electrophilic transfer of a sulfido ligand to a pyrrolyl ligand (57) (Scheme 14A), similar to that observed for uncomplexed pyrroles. Upon addition of
by Rakowski-DuBois et al.38 This complex is of particular interest, since it may simulate more of the complexity of classical MoS2 hydrotreating catalysts. Interestingly, electrophilic attack at the pyrrole α-C atom causes C−H activation and formation of two dinuclear Mo complexes, one with the pyrrolyl ligand (53) and one with the abstracted α-H atom (52) (Scheme 12A), implicating at least four Mo atoms as a minimal model in this reactivity pattern. Similarly, electrophilic attack of N-methylpyrrole leads to predominantly α-C−H activation (54) but also a small amount of β-C−H activation (55) (Scheme 12B). Reaction with N-methylindole leads to exclusive activation at the 3-position. 1460
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 12. Activation of (A) Pyrrole and (B) N-Methylpyrrole at a Tetranuclear Molybdenum Sulfide Complexa
a
Counterions are omitted for clarity.
Scheme 13. Reaction of Dinuclear Molybdenum Sulfide NMethylpyrrolyl Complex with H2a
a
plausible reaction pathway in catalysis over industrially relevant metal sulfides, with the role of promoters being to modulate N atom binding. 2.4. Ruthenium and Osmium. Molecular ruthenium complexes and triruthenium clusters have been studied with respect to N-heterocycle reactivity. Thus, Fish et al. studied the coordination of pyridine and 2-methylpyridine to [Cp*Ru(CH3CN)3](OTf) (59) in CH2Cl2.39 They observed that pyridine initially coordinates η1(N) to the Ru center (60) but slowly rearranges to the η6 isomer (61) in CH2Cl2 or more rapidly at 80 °C under vacuum or at room temperature in acetone (Scheme 15). In contrast, 2-methylpyridine exhibits exclusively η6 coordination. In THF it is thought that the methyl group weakens the Ru−N bonding and that the
Counterions are omitted for clarity.
another 0.5 equiv of 49, complex 57 undergoes subsequent reaction to activate another pyrrolyl C−H group, thereby yielding a pyrrolyl ligand σ-bound to two bridging sulfido groups (58) (Scheme 14B). Overall, these results suggest that reaction between N-heterocycles and bridging sulfides is a
Scheme 14. (A) Reaction of (PMe2Ph)3(Cl)2Re(pyrrolyl) (56) with 49 at Room Temperature in Dichloromethane and (B) Reaction of 57 with 0.5 equiv of 49 Leading to Activation of the Second Pyrrolyl C−H Bond To Form a Doubly Bound Pyrrole (58)a
a
Counterions are omitted for clarity. 1461
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 15. Coordination of Pyridine to [Cp*Ru(CH3CN)3](OTf)a
a
Counterions are omitted for clarity.
electron-donating Cp* character softens Ru to be a better π donor. The η6 coordination facilitates back-bonding from the filled metal orbitals to the ligand π* orbitals. Fish et al. also observed that quinoline exhibits η6-carbocyclic ring coordination with [Cp*Ru(CH3CN)3](OTf) (59).39 Chaudret et al. similarly observed that polymeric [Cp*RuCl]n (62) exhibits exclusively η6 coordination to the methyl-substituted pyridines 2,6-lutidine (63) and 3,5-lutidine (64) in CH2Cl2/Et2O in the presence of KPF6 (Scheme 16).40
[Ir(COD)(PPh3)2]PF6 (79) with bubbling H2 in the presence of the N-heterocycle led to the formation of [Ir(H)2(η1(N)C5N)2(PPh3)2]PF6 (80) (Scheme 19, illustrated for pyridine). Here two hydride ligands are cis with respect to the two Nheterocycles, analogous to simultaneous adsorption of an Nheterocycle and H2 activation at a single metal center, a potential model for structures on a hydrogenation/hydrogenolysis reaction coordinate. Dinuclear Ir clusters (81) were studied by Vicic and Jones to explore N-methylpyrrole coordination in the presence of a hydrogen acceptor.47 This system is a potential model for the first steps of hydrogenation on extended metal surfaces. NMethylpyrrole coordinates via α-C−H bond activation to form an Ir−C bond, while the other Ir binds via the olefinic CC bond in an η2 fashion (82) (Scheme 20). 2.6. Rhodium. Fish et al. characterized quinoline, isoquinoline, 1-THQ, and 2-methylquinoline coordination to Cp*Rh(MeCN)3(BF4)2 (83).48 Reaction with quinoline leads to a mono-Q Rh complex with two MeCN ligands (84), while reaction with isoquinoline leads to a tris-IQ complex (85) (Scheme 21). Reaction with 2-methylquinoline yields a mixture of η1(N) (86) and η6-carbocyclic ring bonding (87). This suggests that differences in Rh center steric encumbrance and ligand electronics can affect the Rh−N complex lability. In contrast, reaction with 1-THQ leads to an η6-coordinated mono-1-THQ complex (88) not involving the N atom. This result argues that the quinoline and isoquinoline lone pair electrons, which are orthogonal to the aromatic π system, bond to Rh, while those in 1-THQ overlap with aromatic ring π electrons, increasing the electron density and increased likelihood of π complexation. This carbocyclic ring coordination of 1-THQ also highlights the challenge in suppressing further hydrogenation of 1-THQ to DHQ in quinoline hydrogenolysis pathways (see section 4, Schemes 42 and 43). 2.7. Palladium. Rakowski-Dubois et al. reported that reaction of PdCl2(PPh3)(NCCH3) (89) with indoline in refluxing CH2Cl2 yields the PdCl2(PPh3)(η1-indolinyl) complex (90), a σ-bound complex to the Pd (Scheme 22).49 2.8. Cerium. N-heterocycle coordination is not limited to the d-block transition metals, and ceria surfaces not only are supports but also frequently participate directly in catalytic reductions. Andersen et al. observed that the reaction of [1,2,4(Me3C)3C5H2]2CeH (91) with pyridine in C6D12 forms the complex [1,2,4-(Me3C)3C5H2]2Ce(η2-pyridyl) (92) (Scheme 23).50 2.9. Conclusions. The coordination mode(s) of Nheterocycles depends on the metal−N bond strength, which serves as a significant driving force for complexation. Most group III−V complexes adopt an η1(N) or an η2(N,C) bonding motif, reflecting a strong interaction between the N lone pair or C−N bond electrons with the electron-deficient metal center. Note that in most of these complexes the metal is in its highest
Scheme 16. Coordination of 2,6-Lutidine (63), 3,5-Lutidine (64), and Pyridine (65) to a Cp*Ru Complex in CH2Cl2/ Et2O in the Presence of KPF6 and Coordination of Pyridine to the Ru Complex in Acetone To Cause Displacement of the Cp* Ligand (66)a
a
Counterions are omitted for clarity.
However, in acetone and in the presence of KPF6, pyridine adopts an η1(N) coordination along with concomitant loss of the Cp* ligand (66) since the Cp*Ru−pyridine adduct is not very stable. It is believed that the greater stability of the lutidine π-adduct reflects the electronic effects of the greater electron density in the aromatic ring.41 Osmium was shown by Taube et al. to exhibit η2-(C,C) coordination to N-methylpyridinium (67) and 2,6-lutidinium, a very rare coordination mode.42,43 C−H activation at the 4position can then lead to a σ-bound species (68) (Scheme 17, Scheme 17. Coordination of N-Methylpyridinium and Oxidative Addition of the Para C−H Bond to Os(NH3)5
illustrated for N-methylpyridinium). Triruthenium and triosmium clusters have also been shown by Deeming et al. to coordinate N-heteroaromatics by activating two C−H bonds on the same or adjacent carbon atoms (Scheme 18).44−46 Some of the resulting species disrupt the aromaticity and may signal a pathway to bond cleavage. 2.5. Iridium. Sanchez-Delgado et al. studied pyridine, piperidine, and isoquinoline coordination to an Ir phosphine complex in CH2Cl2 at room temperature. Reaction of 1462
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 18. Coordination and C−H Activation of N-Heteroaromatics by (A) [Ru(CO)3]3, (B) Os3(CO)12, and (C) Os3(CO)10(MeCN)2
Scheme 22. Synthesis of the PdCl2(PPh3)(η1-indolinyl) Complex
Scheme 19. Reaction of H2 and Pyridine with [Ir(COD) (PPh3)2]PF6a
a
Counterions are omitted for clarity.
Scheme 23. Complexation of Pyridine by [1,2,4(Me3C)3C5H2]2CeH
Scheme 20. Coordination and C−H Activation of NMethylpyrrole at an Ir Dimer
later transition metals, many of the complexes coordinate in an η6 fashion via either the heterocyclic or carbocyclic ring. In these complexes, the metal is often electron-rich, is not in its highest oxidation state, and may act primarily as a π donor to
formal oxidation state, which should minimize electron− electron repulsion between the N-heterocycle and the metal center. In progressing across the periodic table to group VI and
Scheme 21. Coordination Modes of Quinoline (84), Isoquinoline (85), 2-Methylquinoline (86 and 87), and 1-THQ (88) to a Cp*Rh(MeCN)3(BF4)2 Complexa
a
Counterions are omitted for clarity. 1463
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis the substrate antibonding orbitals instead of acting as an electron acceptor.
Scheme 25. Reaction of Q-Coordinated Mo Complexes with H2
3. HYDROGENATION PROCESSES Hydrogenation of the heterocyclic ring is considered mechanistically to be a prerequisite for substrate C−N bond cleavage in industrial hydrotreating catalysis.3 Traditional catalytic processes often consume more H2 than is stoichiometrically required for denitrogenation because they also saturate carbocyclic rings.3,12 Note that suppressing these parasitic hydrogenation losses would provide significant energy savings and increases in product selectivity. Thus, the hydrogenation of unsaturated C−N bonds has been studied with both homogeneous and solid-phase immobilized molecular complexes. The use of immobilized complexes has not been reviewed significantly in the prior literature. This section first discusses molecular complexes, separated by element groups, followed by several examples of aromatic C−N bond hydrogenation by solid-phase immobilized molecular complexes. 3.1. Homogeneous Models. 3.1.1. Molybdenum. Parkin et al. studied the hydrogenation of pyridine, quinoline, and their derivatives mediated by a Mo complex. They observed that (η2(N,C)-pyridyl)Mo(PMe3)4H (32) readily reacts with H2 to eliminate pyridine and that hydrogenation does not occur (93) (Scheme 24).33 The η6-coordinated complex 33 is unreactive toward H2 at 60 °C.
to the Mo center is observed with η6-coordinated quinoxaline. In contrast to quinoline derivatives that are unreactive toward hydrogenation when coordinated by the carbocyclic ring, the three-ring N-heterocycles phenazine (PZ) (45) and acridine (ACR) (48) undergo hydrogenation to mixtures of 9,10dihydrophenazine (9,10-DHPZ) and 1,2,3,4-tetrahydrophenazine (1-THPZ), and 9,10-dihydroacridine (9,10-DHACR), respectively, when coordinated by the carbocyclic ring (Scheme 26).36 The greater hydrogenation reactivity of PZ and ACR versus that of quinoline can be attributed to the greater resonance stabilization energy imparted by the N atom in quinoline, thereby reducing its reactivity.51 The greater basicity of quinoline (pKa = 4.85) versus that of ACR (pKa = 4.11) may also contribute.52 A 1:2 PZ:Mo complex is also formed and undergoes H2 oxidative addition to one Mo atom but is inactive for hydrogenation. 3.1.2. Ruthenium and Osmium. Mononuclear Ru complexes selectively hydrogenate the heterocyclic ring of quinoline and its derivatives.53−55 Thus, Rosales et al. observed that [RuH(CO)(NCMe)2(PPh3)2]BF4 (98) is 30 times more active for quinoline hydrogenation than isoquinoline and 15 times more active for quinoline hydrogenation than indole. This may again be related to the greater basicity of the N atom in isoquinoline (pKa = 5.46) versus that in quinoline (pKa = 4.85).52 Similar to the case for Mo, Rosales et al. observed hydrogenation rates of acridine to be 3-fold greater than that of quinoline at the Ru complex (98) and suggested it was due to the greater resonance stabilization energy of the quinoline heteroyclic ring. Benzoquinolines (BQ) have hydrogenation rates lower than those of quinoline, presumably due to steric hindrance. Rosales et al. proposed that ACR hydrogenation proceeds via a mechanism different from that of the other species, exemplified by quinoline in Scheme 27. In both mechanisms, the substrate displaces MeCN, coordinates in an η1(N) mode (99, 101), and is subsequently hydrogenated. For the ACR system, hydrogenation leads to 9,10-DHACR and a Ru complex with a vacant coordination site (100), followed by association of another ACR molecule. For the other substrates, a new substrate molecule displaces the partially hydrogenated substrate molecule. Sánchez-Delgado et al. observed the selective hydrogenation of Q to 1-THQ with RuHCl(CO)(PPh3)3 at 150 °C and 435
Scheme 24. Reaction of Pyridine-Coordinated Mo Complexes with H2
The introduction of a second ring makes the substrate more reactive with respect to hydrogenation. At 80 °C in the presence of H2 from the η6-heterocyclic coordination (35), the quinoline heterocyclic ring is hydrogenated to 1-THQ and forms the complex Mo(PMe3)4H4 (93),34,35 which in turn can coordinate and hydrogenate another quinoline molecule (Scheme 25). The catalytic efficiency of this system is very low, since 35 can rearrange to the carbocyclic ring-coordinated isomer 36, which is unreactive toward hydrogenation. This reactivity difference between bound heterocyclic and carbocyclic rings demonstrates how the coordination mode affects reactivity and is a recurring theme in the complexes of this section. Similar to the case for quinoline, Parkin et al. also reported that η6-heterocyclic coordination of isoquinoline and quinoxaline enables hydrogenation of the heterocyclic ring at elevated temperatures.35 As above, carbocyclic ring coordination of isoquinoline and quinoxaline leads to complexes which cannot be readily hydrogenated, even though oxidative addition of H2 1464
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 26. Reaction of (A) Phenazine- and (B) Acridine-Coordinated Mo Complexes with H2
Scheme 27. Proposed Mechanism for the Hydrogenation of (A) ACR and (B) Q with a [RuH(CO)(NCMe)2(PPh3)2]BF4 Catalysta
a
Counterions are omitted for clarity.
psi of H2.56 Although the carbocyclic ring was not hydrogenated, its presence is necessary for hydrogenation of the heterocyclic ring, since pyridine is not hydrogenated under these conditions. They proposed that the carbocyclic ring activates the heterocyclic ring either by an electronic effect that makes the heterocyclic ring more susceptible to hydride attack or by orientation of the molecule in an optimum coordination mode for hydrogenation. The analogous Os complex was significantly less active, exhibiting a rate 20% of that for the Ru complex for Q hydrogenation. Chan et al. studied the enantioselective reduction of 2-MQ with a Ru/Ts-dpen catalyst (Ts-dpen = N-[(1R,2R)-2-amino1,2-diphenylethyl]-4-methylbenzenesulfonamide) (107) in the ionic liquid [BMIM]PF6 (BMIM = 1-n-butyl-3-methylimidazolium) (Scheme 28).57 They observed that this catalyst is airstable in ionic liquids, in contrast to most phosphine-based Ru catalysts. Reaction leads to quantitative conversion of Q to 1THQ with 99% enatiomeric excess at 735 psi of H2 and 25 °C. This reduction is proposed to proceed via a series of proton and hydride transfers to selectively reduce the heterocyclic ring. In contrast to the previous examples that report nearexclusive saturation of the heterocyclic ring, Glorius et al. demonstrated the ability to control which ring would be hydrogenated on the basis of the ligand choice for the
Scheme 28. Proposed Cycle for 2-Methylquinoline Hydrogenation by a Cationic Ru Complexa
a
Counterions and arene and ethylenediamine substituents are omitted for clarity. Asterisks denote chiral centers.
1465
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
Scheme 31. Proposed η4 Coordination Modes of Quinoline, Isoquinoline, and Acridine to Ru(H)2(η2-H2)2(PCy3)2
asymmetric hydrogenation of substituted quinoxalines with a chiral Ru N-heterocyclic carbene (NHC) complex (Scheme 29).58 Since the electronics of the ligands are relatively similar, the difference in selectivity is likely due to the sterics of the ligand and how it reacts with the metal center. Scheme 29. Hydrogenation of the Heterocyclic or Carbocyclic Ring of a Substituted Quinoxaline by a Ru Complex with the Ligand Determining Which Ring Is Hydrogenated
substrates.60 The authors suggest that the more facile heterocyclic ring hydrogenation is due to aromatic ring coordination similar to the unusual η4 coordination observed in other arene complexes.61−64 Borowski et al. found that 7,8-benzoquinoline (7,8-BQ) as well as its phenanthrene analogue were resistant to hydrogenation under the above conditions. They proposed that reaction with 7,8-BQ leads to the formation of a new stable organometallic complex. Chaudret et al. isolated a similar compound, RuH(η2-H2)(NC13H8)(PiPr3)2 (114), with the benzoquinolyl ligand coordinated to the Ru via the N atom and an ortho carbon on the external aromatic ring (Figure 4).65
Borowski et al. studied the reaction of N-heterocycles with Ru(H)2(η2-H2)2(PCy3)2 (110) at 80 °C and 44 psi of H2 (Scheme 30).59 Similar to the previously discussed studies, pyridine exhibits η1(N) coordination and is not hydrogenated. Pyrrole is similarly unreactive. In contrast, quinoline and isoquinoline can be hydrogenated exclusively to 5-THQ and 5THIQ, respectively. It is proposed that η4-arene coordination (111, 112) leads to selective carbocyclic ring hydrogenation (Scheme 31). Thus, naphthalene hydrogenation is also observed, but only at 30% of the rate of these N-containing
Figure 4. Coordination of 7,8-BQ to RuH(η2-H2)(NC13H8)(PiPr3)2.
This ortho-metalated complex is stable under H2. Hydrogenation of indole to indoline was achieved, but it was 4-fold slower than the hydrogenation of quinoline and isoquinoline. Chaudret et al. also observed ACR to coordinate to the metal center in the same, rare η4 coordination (113) of the
Scheme 30. Hydrogenation of (A) Pyridine, (B) Pyrrole, (C) Quinoline, (D) Isoquinoline, (E) Naphthalene, (F) Acridine, (G) 7,8-Benzoquinoline, (H) Phenanthrene, and (I) Indole by Ru(η2-H2)2(H)2(PCy3)2
1466
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis carbocyclic ring proposed for quinoline and isoquinoline (Scheme 31). Acridine is converted quantitatively to a mixture of the 1-tetrahydro- and octahydroacridines (1-THACR and OHACR). It is proposed that the first carbocyclic ring is hydrogenated and then the second carbocyclic ring follows at a significantly slower rate. Hydrogenation of the heterocyclic ring, the commonly detected product in other systems, was not detected in this study. In addition to hydrogenation by the Ru center, Rakowski Dubois et al. observed that η6 coordination of the arene ring of N-methylindole to the Ru (115) allows for hydrogenation to Nmethylindoline (116) by an external catalyst, in this case 5% Rh/C at 25 °C and 44 psi of H2 (Scheme 32).49 The reaction
Scheme 33. Enantioselective Hydrogenation of a Substituted Quinoxaline by the Indicated Ir Complexa
Scheme 32. Hydrogenation of a Ru Indole Complex by 5% Rh/Ca
a
a
The asterisk denotes a chiral center.
reduction rate by 50%, presumably due to steric factors. Acridine and 7,8-benzoquinone are hydrogenated 30% and 11× more rapidly than quinoline, respectively. However, 5,6-BQ is hydrogenated more slowly than Q. Furthermore, competition experiments with quinoline showed that acridine, isoquinoline, and pyridine are the exclusive substrates. In contrast, 2methylpyridine and 2-MQ are reversible competitors, moderately reducing quinoline hydrogenation rates. Remarkably, benzoquinolines have either no effect or an enhancing effect on quinoline hydrogenation rates. The latter may reflect a hydrogen shuttling mechanism. The reduction of quinoline is also self-inhibiting, as the product 1-THQ decreases the initial hydrogenation rate by a factor of 3. 3.1.6. Cerium. Andersen et al. studied catalytic pyridine hydrogenation mediated by [1,2,4-(Me3C)3C5H2]2CeH (91).50 T h e y o b s e r ve d t h a t w h e n t h e c o m p l ex [ 1 , 2 , 4 (Me3C)3C5H2]2Ce(η2-pyridyl) (92) is dissolved in C6D12 at 60 °C in the presence of 15 psi of H2, the pyridyl ligand is slowly hydrogenated to the 4,5,6-trihydropyridyl complex 118 and then to the piperidyl complex 119 (Scheme 34). Although
Counterions are omitted for clarity.
proceeds quantitatively over 24 h. In contrast, reaction with the free N-methylindole and Rh/C yielded only 10% conversion under the same conditions. This suggests the importance of multimetallic catalysts, where one metal activates the substrate, perhaps by this η6 coordination, and a second activates H2. The η6-coordinated indoline complex 116 has decreased N atom basicity relative to free indoline. NaOH titration of the η6indoline complex in aqueous HCl has an equivalence point at pH 7, wheras the free indolinium cation has a pKa of approximately 5.5 Rakowski Dubois et al. also attempted to use a second metal center to coordinate and form a σ bond with the indoline N atom, but reflux in CH2Cl2 with PdCl2(PPh3)(MeCN) did not lead to Pd complexation. 3.1.3. Cobalt. Alvanipour and Kispert studied the hydrogenation of quinoline derivatives with a Co(stearate)-Et3Al catalyst.66 At 90 °C and 700 psi of H2, they achieved significant conversion with exclusive hydrogenation of the isoquinoline heterocyclic ring but obtained a 3:1 mixture of 1-THQ and 5THQ when hydrogenating quinoline. 3.1.4. Iridium. Vidal-Ferran et al. studied the enantioselective hydrogenation of a prochiral quinoxaline at an Ir complex. They observed selective conversion of the heterocyclic ring and 70% ee (Scheme 33).67 3.1.5. Rhodium. Fish et al. studied the hydrogenation of 2methylpyridine with Cp*Rh(CH3CN)32+.68 They observed conversion of 2-methylpyridine to 2-methylpiperidine at 40 °C and 500 psi of H2. This is a rare example of reduction of a pyridine derivative under relatively mild conditions. However, they were unable to hydrogenate pyridine, isoquinoline, or pyrrole under these conditions, presumably due to the formation of catalytically inactive Cp*Rh(η1(N)-heterocycle)32+ complexes. Fish et al. also observed selective heterocyclic ring hydrogenation for several N-heterocycles at rates significantly greater than that of 2-methylpyridine. For example, quinoline was hydrogenated at a rate 32× faster than 2-methylpyridine due to the high resonance stabilization energy of pyridines. Fish et al. studied other nitrogen heterocycles to probe substrate steric and electronic effects on heterocyclic ring reduction. A methyl group in the 2-position (2-MQ) lowers the
Scheme 34. Hydrogenation of a Pyridyl Ligand to 4,5,6Trihydropyridyl and Piperidyl Ligands at a Ce Organometallic Complex
the exact mechanism of hydrogenation is complex since it involves reversible reactions, they concluded that the hydrogenation involves two units of [1,2,4-(Me3C)3C5H2]2CeH (91). One unit coordinates pyridine, and the second unit adds to a CC double bond. This system is catalytic, since hydrogenolysis of the Ce−C bond then regenerates the Ce−H bond and may be a simple model for hydrogenations on ceria surfaces. 3.1.7. Lanthanum. Marks et al. studied the partial hydrogenation of pyridine with a [Cp*2LaH]2 catalyst (120). They reported that when pyridine is added to [Cp*2LaH]2 (120) in C6D12, the pyridine is partially hydrogenated to a 1,2dihydropyridyl ligand (123) (Scheme 35).69 3.2. Supported Molecular Complexes. Immobilized molecular catalysts have been examined for the hydrogenation 1467
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
resistance to sulfur poisoning is a critical property of all practical catalysts. These supported metallophthalocyanine catalysts are relatively sulfur tolerant, with the rate of quinoline hydrogenation to 1-THQ unaffected by thiophene or benzothiophene addition. This sulfur resistance can be attributed to the greater basicity of N-heterocycles vs Sheterocycles, in accord with the proposed axial substrate coordination to the metal center. The effects of time, quinoline concentration, and H2 partial pressure on the reaction rate are all consistent with the proposed axial quinoline binding mode to the metal center, affording selective reduction of the heterocyclic ring. 3.2.2. Rhodium. Rempel et al. examined a SiO2-supported trirhodium acetate cluster for hydrogenation of quinoline to 1THQ.85 They proposed HRh3O(OCOCH3)5+(OCOCH3)− (125) (Figure 6) to be the active species, and at ambient H2
Scheme 35. Partial Hydrogenation of Pyridine by [Cp*2LaH]2 (120)
of aromatic C−N bonds. Many studies have been conducted to heterogenize homogeneous complexes with the stated goal of enhancing their stability under harsh conditions and allowing them to be more easily recovered after the reaction.70,71 However, new reactivity patterns may also emerge, as described below. Studies of supported molecular complexes have primarily involved Rh and Ru, but studies with other metals have also been reported. Studies with these supported molecular complexes are rarer, due in part to difficulties that arise in characterizing these materials. Similar studies on selective C−N bond hydrogenation have also been conducted with imines and nitriles but are beyond the scope of this review.72−82 For convenience, this section is first organized by the general class of catalyst and then by metal. 3.2.1. Supported Metallophthalocyanine-Based Catalysts. Metallophthalocyanines (M(PC)s) on inorganic oxide supports (124) (Figure 5) have been studied by Boucher et al. as a
Figure 6. Structure of the proposed active species of the Rh acetate cluster. The counterion is omitted for clarity.
pressure and 50 °C, the unsupported cluster is competent to hydrogenate quinoline, naphthalene, and other unsubstituted aromatics, while the supported cluster is proficient at hydrogenating quinoline to 1-THQ but not unsubstituted aromatics. Heteroaromatic hydrogenation is more facile due to the lower aromaticity of the heterocyclic versus the carbocyclic ring, thereby lowering the activation energy for hydrogenation. Hydrogenation with these clusters which contain multiple metal centers in close proximity may be a model for hydrogenation with a classical hydrotreating catalyst, where the H− is not directly bound to the same metal center as the substrate but rather migrates from a sulfhydryl group to a coordinated heteroaromatic substrate.86 Fish et al. studied the phosphinated styrene−divinylbenzene resin-supported complex Rh(PPh3)3Cl (126) as well as the unsupported analogue (127) for quinoline and acridine reduction at 85 °C and 310 psi of H2.87,88 Quinoline hydrogenation with the supported catalyst leads exclusively to 1-THQ, similar to that observed for the homogeneous analogue (Scheme 36A,B). In contrast, acridine hydrogenation exclusively yields 9,10-DHACR for the supported catalyst but a 50/ 50 mixture of 1,2,3,4-tetrahydroacridine (1-THACR) and 9,10DHACR for the homogeneous complex (Scheme 36C,D). In both cases, the supported catalyst hydrogenates quinoline and acridine at 22 and 10 times the rate of the unsupported catalyst, respectively. Fish et al. suggested that the difference in product selectivities for the acridine reduction reactions between the supported and unsupported catalysts reflects steric repulsion of the substrate approaching the heterogeneous catalyst. Reduction of quinoline with D2 also leads to different levels of D incorporation between the homogeneous and heterogeneous catalysts. Although both catalysts lead exclusively to 1-THQ and similar deuterium incorporation in the heterocyclic ring, the homogeneous catalyst also incorporated deuterium at the 8position in the carbocyclic ring (Scheme 36A,B). This poor reactivity of the carbocyclic ring with D2 and the heterogeneous
Figure 5. Possible structure of a supported metallophthalocyanine complex.
means to achieve immobilized isolated molecular catalysts for quinoline hydrogenation. 83,84 The M(PC) molecule is suggested to lie flat on the surface with one axial coordination position interacting with the surface and the other free to bind quinoline. Mg, Mn, Fe, Co, Ni, Cu, Zn, Mo, Sn, and Pb M(PC)s were supported on SiO2, Al2O3, and SiO2-Al2O3 at 4 × 10−5 mol of M(PC)/g of support, corresponding to surface coverages of 19% for SiO2, 40% for Al2O3, and 11% for SiO2Al2O3. At 290 °C and 1000 psi of H2, catalytic activity is independent of the support with >99% selectivity to 1-THQ. The differences in activity between various metals and supports for 24 h reactions are relatively small (conversions to 1-THQ of 6−14%) at 290 °C. However, the differences in activity between metals increases at 385 °C for the SiO2-supported catalysts (conversions to 1-THQ of 40−91%) in the order Mo > Ni > Co > Zn > Mn > Mg > Cu > Fe > Sn > Pb. The authors suggest that the activity differences may reflect the metal’s ability to bind and dissociate H2 as well as the strength of the metal−substrate interaction, since too strong an interaction may lead to an unreactive complex. Although not typically addressed in supported molecular catalyst studies, sulfur-containing compounds are inevitably present during the hydrodenitrogenation of crude fuels, and 1468
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
(128) hydrogenates both the heterocyclic and carbocyclic rings (43% conversion; 32% 1-THQ, 7% 5-THQ, 4% DHQ), but hydrogenation of the heterocyclic ring to 1-THQ is significantly faster than hydrogenation of the carbocyclic ring to 5-THQ or subsequent hydrogenation to DHQ. 3.2.3. Ruthenium. Bianchini et al. also studied the hydrogenation of N- and S-containing heterocycles with unsupported [(MeC(CH2PPh2)3)Ru(NCMe)3](OSO2CF3)2 (129) and SiO2-supported [Ru(NCMe)3(O3S(C6H4)(CH2C(CH2PPh2)3))](OSO2CF3) (130) catalysts (Figure 7).90 They
Scheme 36. (A, B) Quinoline Hydrogenation by PolymerSupported Rh(PPh3)3Cl (126) and Unsupported Rh(PPh3)3Cl (127) and (C, D) Acridine Hydrogenation by Polymer-Supported Rh(PPh3)3Cl (126) and Unsupported Rh(PPh3)3Cl (127)
Figure 7. Structures of the unsupported Ru molecular catalyst (129), Ru molecular catalyst supported on SiO2 (130), and Ru nanoparticles supported on SiO2 (131).
compared these catalysts to a SiO2-supported Ru(0) nanoparticle catalyst (131) to examine reactivity differences between single and contiguous metal sites. Both unsupported and supported Ru(II) molecular catalysts selectively hydrogenate the heterocyclic rings of quinoline and acridine to 1-THQ and 9,10-DHACR, respectively (Scheme 38). In contrast, quinoline hydrogenation by Ru0/SiO2 leads to a mixture of 1-THQ, 5THQ, and DHQ, while ACR hydrogenation by Ru0/SiO2 leads to a mixture of 9,10-DHACR, 1-THACR, and OHACR. The unsupported and supported molecular Ru(II) catalysts also hydrogenate the heterocyclic ring of indole to form indoline exclusively, but the supported analogue exhibits a significantly greater turnover frequency (Scheme 39). The nanoparticle catalyst primarily produces octahydroindole with smaller amounts of hydrogenation of one ring to indoline or 4,5,6,7-tetrahydroindole. Bianchini et al. also carried out competitive hydrogenation experiments with quinoline, indole, and benzothiophene. The supported molecular catalysts do not show competition between quinoline, which is expected to coordinate η1 via the N atom, and indole, which coordinates via the η6 mode from its carbocyclic ring. In contrast, quinoline and indole significantly inhibited each other over the supported nanoparticle catalyst and yielded only the partially hydrogenated products. Benzothiophene, but not its hydrogenated analogue, inhibited turnover frequencies or reduced the total extent of hydrogenation for both the nanoparticle catalyst (Scheme 40) and the supported molecular catalyst, indicating competition for the same sites. Likewise, Bianchini et al. also found the supported Ru molecular catalyst to be significantly less active than the nanoparticle catalyst for toluene and anthracene hydrogenation (Scheme 41). The poor activity of the supported molecular complex for hydrogenating unsubstituted aromatics, in contrast to N-substituted aromatics, is consistent with the greater energetic barrier to overcome the resonance stabilization energy of unsubstituted aromatics and, conversely, the relative ease by which the carbocyclic rings chemisorb on extended
catalyst may be attributed to the steric constraints imposed by the polymer functional groups surrounding the metal center.71 Fish et al. observed the rate of hydrogenation of quinoline to 1-THQ to be enhanced in the presence of p-cresol. They believe that p-cresol helps in the dissociation of PPh3, thereby relieving the congestion around the Rh center and allowing quinoline to more easily coordinate. Alternatively, they suggest that the presence of p-cresol may lead to rate enhancement by stabilizing the electron-deficient Rh center. Bianchini et al. studied quinoline hydrogenation at 80 °C and 435 psi of H2 using Rh(COD)+ supported on a cross-linked styrene−divinylbenzene polymer having a tethered diphosphine (Scheme 37).89 This [Rh(COD)(POLYDIPHOS)]PF6 catalyst Scheme 37. Q Hydrogenation by a [Rh(COD)(POLYDIPHOS)]PF6 Complex Tethered to a Polymer Supporta
a
The counterion is omitted for clarity. 1469
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
Scheme 38. Hydrogenation of (A) Q with Catalyst 129 or 130, (B) Q with Catalyst 131, (C) ACR with Catalyst 129 or 130, and (D) ACR with Catalyst 131
change selectivity in the present case, most likely due to the similar steric bulk around the metal center and the fact that the complex is supported on SiO2, a redox-inactive, sterically unencumbered, and neither strongly basic nor acidic support. If the complex were supported on redox-active TiO2 or CeO2, acidic Al2O3, basic MgO, or sterically encumbered phosphinated styrene−divinylbenzene resin as in Scheme 36, to suggest some examples, the support would likely play a greater role in the catalytic activity/selectivity. 3.3. Conclusions. In contrast to N-heterocycle coordination which is common for metals across the periodic table, catalytic hydrogenation is predominantly limited to group VI and later metal complexes. This is likely due to both the strength of the metal−nitrogen bonding and the inability of the high-valent early metals to dissociate H2. In general, hydrogenation activity increases on progressing to the later transition metals while hydrogenolysis activity decreases. Analogously, Nrich feeds are often hydrotreated with Ni/Mo−Al2O3 catalysts, which excel at hydrogenation/H2 activation, while feeds containing low amounts of N are often hydrotreated with Co/Mo−Al2O3 catalysts, which excel at hydrogenolysis due, among other factors, to Brønsted acidity.91 The strong hydrogenation activity of Ni/Mo−Al2O3 catalysts is unnecessary for feeds containing low amounts of N, since hydrogenation is not a prerequisite for HDS while it is for HDN.
Scheme 39. Hydrogenation of Indole with (A) Catalyst 129 or 130 and (B) Catalyst 131
Scheme 40. Quinoline Hydrogenation in the (A) Presence and (B) Absence of Benzothiophene
metal surfaces. Overall, these results demonstrate the greater selectivity of the supported and unsupported molecular catalysts relative to the nanoparticle catalyst, likely due to a combination of ligand steric factors, ease of access to the metal center, and the use of mononuclear active sites. These studies also show that supporting organometallic complexes do not
Scheme 41. Hydrogenation of (A) Toluene by 130, (B) Toluene by 131, (C) Anthracene by 130, and (D) Anthracene by 131
1470
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
Mo−Al2O3 catalysts,93 but it is a minor contributor overall for this and other typical commercial catalysts. A third possibility is direct cleavage of the sp2 C−N bond, but this has yet to be demonstrated with N-heterocycles over sulfided Co/Mo-Al2O3 or sulfided Ni/Mo-Al2O3 catalysts. The relative paucity of homogeneously catalyzed systems can be attributed to the strength of the C−N bond, which often requires high temperatures and H2 pressures to cleave, conditions where many homogeneous catalysts decompose.71 C−N hydrogenolysis is usually considered to be the turnoverlimiting step in HDN, and thus understanding the mechanism may lead to new and improved catalysts.13 4.1. Group III and Actinides. C−N hydrogenolysis of substituted imidazoles by group III metals Sc and Y, and by U, occurs by coupled σ bond cleavage and formation.18,21,94−97 Diaconescu et al. studied the coupling of two imidazole molecules by (NNfc)Sc(CH2Xy-3,5)(THF) (132) to form a more stable structure with delocalized π bonds (136) (Scheme 44).18,21 They proposed that two imidazole molecules
4. C−N HYDROGENOLYSIS In contrast to aromatic C−N hydrogenation, which has been widely studied for molecular complexes, there are far fewer examples of C−N hydrogenolysis by molecular complexes. Most C−N hydrogenolysis studies are in the context of full HDN reaction networks over bulk sulfides, nitrides, carbides, phosphides, borides, or oxides or at high loadings of these materials on oxide supports.4 Quinoline is a heavily studied model reactant and is generally assumed to follow the reaction network shown in Scheme 42. Without wishing to focus exclusively on these catalysts, they are generally considered to effect quinoline C−N hydrogenolysis via the DHQ-PCHA pathway (Scheme 42, pathway 2).3,5 Scheme 42. Generally Accepted Reaction Network for Quinoline HDN
Scheme 44. Ring Opening of N-Methylimidazole by a (NNfc)Sc(CH2Xy-3,5)(THF) Complex
C−N bond hydrogenolysis with commercial heterogeneous catalysts has been proposed to occur via either a Hofmann elimination or an SN2 mechanism (Scheme 43; illustrated for 1Scheme 43. Proposed Mechanism for Ring Opening of 1THQ via (A) Hofmann Elimination and (B) an SN2 Pathway coordinate to complex 132 and displace the THF ligand to form 133. Subsequent C−H activation of a free 1methylimidazole by 133 leads to the formation of 134. Nucleophilic attack of the imidazolyl carbon of 134 at one of the coordinated 1-methylimidazole molecules leads to dearomatization of the coordinated 1-methylimidazole (135). This subsequently enables coupling of both rings in 135 to yield the metallacycle 136 via a series of C−H activations and C−C bond coupling steps. 4.2. Groups IV and V. Pyridine ring opening has been explored with group IV and V metal complexes. Mindiola et al. demonstrated pyridine ring-opening by a Ti alkyl−alkylidene complex (16) via a nonreductive mechanism (Scheme 45).25,26 They proposed that the Ti reagent eliminates an alkane to form the alkylidyne species 17, opening a coordination site for pyridine. Pyridine coordinates η1(N) to the Ti complex 18, and this complex then undergoes ring-opening metathesis to form an azametallic bicyclic complex by a [2 + 2] cycloaddition (19). This structure then expands to a metallacyclooctane (137), which subsequently contracts to form the bicyclic complex 20. The driving force for this C−N bond cleavage is presumably the strong TiN bond formation. In addition to inserting into the heterocyclic ring, this Ti reagent extrudes the N atom from pyridine to generate the hydrocarbon product and a Ti-imido species (Scheme 46).98 Mindiola et al. proposed that treatment of the titanium complex with trimethylsilyl chloride to silylate the α-N atom in the
THQ). In commercial catalytic systems, there are many species which can function as a nucleophile or a base. Sulfhydryl groups, hydroxide ions, hydride ions, hydrogen sulfide, water, and ammonia present can all fill this role.92 HDN extrusion of the N atom as NH3 most reasonably requires sp3 hybridization of the β-C atom for an SN2 process, while sp3 hybridization of both the α- and β-C atoms is required for a Hofmann elimination mechanism. With commercial catalysts these scenarios usually translate into complete saturation of the carbocyclic ring to yield predominantly n-propylcyclohexane (Scheme 42 pathway 2), consuming 7 mol of H2 in the overall transformation of quinoline to n-propylcyclohexane. In contrast, n-propylbenzene formation (Scheme 42 pathway 1) consumes only 4 mol of H2, and N atom extrusion would be via partial hydrogenation of 2propylaniline to 1,6-dihydro-2-propylaniline, followed by a Hofmann elimination or a SN2 reaction and dehydrogenation. This partial hydrogenation pathway to yield the aromatic product has been demonstrated by Prins et al. for naphthylamine HDN over sulfided Co/Mo−Al2O3 and sulfided Ni/ 1471
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis Scheme 45. Proposed Catalytic Cycle for Pyridine Ring Opening by (PNP)Ti(CHtBu)(CH2tBu)
Scheme 47. Possible Modes of Hydride Attack To RingOpen a Substituted Pyridine at a Ta Complex
Scheme 48. Thermolysis of a Ta Metallacycle To Extrude a Light Hydrocarbon Product
azametallabicyclic ring (138) followed by a 1,3-hydride shift leads to ring enlargement to an eight-membered ring (139). Subsequent electrocyclic rearrangement leads to ring contraction to the bicyclic system (140), which then undergoes a [2 + 2] reaction to exclude the arene product and yield the imide (141). Wigley et al. studied the C−N hydrogenolysis of substituted pyridines with aryloxide Ta complexes (Scheme 47).32,99−101 This substituted complex was not derived from coordination of free pyridine but rather from of a reaction between the metallacyclopentadiene and tert-butyl cyanide.102 Similar to the case for quinoline (Scheme 7),31 η2(N,C) coordination is necessary for hydrogenation to transform the α-C atom from sp2 to sp3 hybridization. The change in hybridization makes the substituted pyridine susceptible to nucleophilic attack. Nucleophilic attack at the α-C atom transforms the Ta-amido complex (143) to a Ta-imido species (145), with the formation of the TaN bond being the driving force to cleave the strong CN bond. Isotopic labeling crossover studies confirmed that an intramolecular endo attack occurs with the nucleophile bound to the Ta prior to cleaving the C−N bond. Direct intermolecular attack at the α-C atom by the nucleophile does not occur. The endo attack provides evidence that the hydride and pyridine can be bound to the same metal center, which is in contrast to that proposed for traditional hydrotreating catalysts, where the hydride migrates from a sulfhydryl group to the heterocycle and is not directly bound to the same metal center.86 This study also shows that, for a C−N bond cleavage via nucleophilic attack, it is not necessary to protonate the N atom, since the electrophilic metal center is able to serve that purpose. Wigley et al. demonstrated that upon thermolysis, complex 146, which is the methylated analogue of 145, undergoes subsequent ring expansion to extrude a light hydrocarbon product (Scheme 48).99,100 Complex 146 is proposed to
undergo a β-elimination to yield 147, which undergoes rapid olefin insertion to yield 148. The authors suggest that an electrocyclic rearrangement of the eight-membered metallacycle leads to formation of the bicyclic complex 149. The metallacyclobutane portion of the bicyclic complex then undergoes a retro [2 + 2] cycloaddition to yield complex 150 and tBuCHCH2. Complex 150 is proposed to undergo rapid dimerization to 151. This demonstrates that ring opening of a substrate which leads to β-hydrogen generation may also facilitate further degradation and hydrocarbon extrusion. This may explain why the principal products of pyridine HDN with traditional sulfided Co/Mo-Al2O3 catalysts are C2 and C3 hydrocarbons with only a small fraction of C5 hydrocarbons.103 Although the above complexes do not undergo N atom extrusion, Wolczanski et al. demonstrated that N atom extrusion from substituted anilines does occur at Ta(silox)3 (21) (Scheme 49).104 Thus, oxidative addition across the sp2 C−N bond is observed to form 152, in contrast to oxidative addition across the N−H bond, which might be expected. Electron-withdrawing substituents on the aromatic ring increase
Scheme 46. Denitrogenation by a Ti Organometallic Complex
1472
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
sp3 α-C atom by another Nb(silox)3 molecule. Pyridine addition inhibits the reaction, consistent with the proposed mechanism. Increasing the substrate steric bulk from pyridine to 3,5-lutidine alters the C−N hydrogenolysis mechanism to a cyclometalation pathway with one of the siloxide ligands. Migratory insertion of the cyclometalated tBu group yields the substituted metallaaziridine complex 158, which then undergoes β-H elimination, reinsertion, and a [2 + 2] retro addition to yield the cyclometalated imido product 161 (Scheme 50). 4.3. Rhenium. In contrast to the previous pyridine ringopening results, Riera et al. demonstrated that pyridine ring opening with Re requires an external base.105,106 Pyridine ring opening results from the deprotonation of 1-methylimidazole to form a carbeniate anion which attacks the pyridine ring, leading to dearomatization (163) (Scheme 51), converting the pyridine ligand into an amide ligand. Two subsequent amide N atom methylations lead to the C−N bond cleavage product (166). 4.4. Supported MOx-Al2O3. Inspired by the above smallmolecule studies, we investigated quinoline, aniline, and cyclohexylamine hydrogenolysis with highly dispersed TiOxAl2O3, TaOx-Al2O3, and MoOx-Al2O3, prepared by grafting sterically encumbered molecular precursors and then calcining the supported complexes.9 Using stable oxide materials allows reactions to be conducted at 275 °C and 290 psi of H2. These catalytic materials are competent to denitrogenate quinoline with yields close to stoichiometric and ∼80% selectivity to alkylbenzenes. Likewise, TaOx-Al2O3 is competent to stoichiometrically denitrogenate 2-propylaniline, again primarily leading to the aromatic products. The materials do not appear to be capable of significant catalytic turnover with these substrates, suggesting that after N atom extrusion to yield alkylbenzenes, a less reactive surface Ta-imide remains, in analogy to the species in Scheme 47. The stability of this Taimide species suggests that this is the thermodynamic end product, regardless of the original Ta precursor. In contrast, TaOx-Al2O3 catalytically deaminates methylcyclohexylamine to methylcyclohexene. When Pd nanoparticles are additionally deposited on the above oxide materials (167), quinoline conversion is comparable to that for unmodified Pd/Al2O3 but with an enhanced selectivity to aromatics versus saturated hydrocarbons, thereby reducing stoichiometric H2 consumption by 10−15% (Scheme 52). This increased aromatic selectivity holds for several ratios of Pd to MOx as well as for physical mixtures, suggesting that the two sites communicate via relatively longlived fluid-phase intermediates such as shown in 170, rather than a new Pd-MOx site. Overall, this suggests that the primary role of the metal nanoparticle is for hydrogenation while the isolated cationic sites on the modified support accelerate deamination of hydrogenated or partially hydrogenated intermediates (Scheme 52).
Scheme 49. Oxidative Addition across the C−N Bond of a Substituted Pyridine to Ta(silox)3
the relative rate of C−N addition while decreasing the relative rate of N−H activation. In contrast, substituents which increase the aniline basicity increase the relative rate of N−H activation. The cleavage of an sp2 C−N bond by a low-valent Ta complex is in accord with observations by Prins et al. that replacement of some surface S2− ions with P3− ions in a sulfided Ni/Mo−Al2O3 catalyst creates a more reduced metal site, more active for sp2 C−N cleavage.17 Overall, these results suggest that it may be possible to design hydrotreating catalysts that directly cleave the sp2 C−N bond and do not require complete saturation to sp3 hybridization prior to N atom extrusion. These results also illustrate the importance of metal identity and oxidation state and demonstrate that the metal can be directly involved in C− N bond cleavage, instead of acting only as a center for coordination of the hydride and substrate. The mechanism of ring opening at Nb has been shown to depend on substrate steric factors. Wolczanski et al. reported that pyridine ring opening requires two Nb centers (Scheme 50).29,30 Heating (silox)3Nb(η2(N,C)-C5N) (153) at 70 °C in Scheme 50. Ring Opening of (A) Pyridine and (B) 3,5Lutidine by Nb(silox)3
benzene yields 0.5 equiv of pyridine and 0.5 equiv of the ringopened product 154. The η2(N,C) coordination/electrophilic activation of pyridine facilitates the nucleophilic attack at the
Scheme 51. Ring Opening of a Pyridine Ring by a Re Complex in the Presence of KN(SiMe3)2a
a
Counterions are omitted for clarity. 1473
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis
atoms for extrusion by withdrawing electron density in a role typically played by a proton. Further, and in contrast to traditional sulfided hydrotreating catalysts where the hydride necessary for hydrogenation is provided by a vicinal sulfhydryl group, the presently reviewed studies also show that it is possible for the hydride and the N-heterocycle to be bound to the same metal center prior to reaction. Several potential approaches to achieve more selective catalysis can also be gleaned from this review. First, it is crucial to select compatible metal and ligand structures that selectively coordinate the heterocyclic ring. For example, complexes 108 and 109 demonstrate that ligand selection for a Ru catalyst can determine whether there is near-exclusive saturation of either the heterocyclic or carbocyclic ring of a substituted phenazine.58 Likewise, there is significant potential for tuning reactivity by immobilizing complexes, such as in 129 and 130.90 Similarly, several examples show that an immobilized complex can be more selective than the corresponding metal nanoparticle. Finally, Scheme 52 suggests how molecule-derived surface oxide centers, which withstand high temperatures and H2 pressures, can display some of the same catalytic properties as individual metal complexes.9 Overall, it remains challenging to enable catalytic turnover with H2 rather than more exotic reductants. This is frequently complicated by the formation of stable metal-imido species. Further, for bridging the organometallic models and commercial catalysts consisting mainly of bulk or high catalyst loadings of sulfides, nitrides, or phosphides, many of the promising ligand sets need to be stabilized or mimicked with sulfide, nitride, phosphide, or oxide structures to enable their utility under the harsh conditions relevant to commercial hydrotreating. Future work should focus on tuning the metal−N bond strengths and the ability of the metal complex to bind and dissociate H2 by selection of metal sulfides, nitrides, phosphides, or oxides. Optimal binding strengths are required, since too weak metal−N bonding will not lead to complexation and N atom removal, while too strong metal−N bonding will lead to metal-imido thermodynamic sinks. Overall, these studies suggest that the development of novel metal complexes has provided valuable insight into, and may ultimately improve, the selectivity of a major industrial process.
Scheme 52. Proposed Mechanism for the C−N Hydrogenolysis of 2-Propylaniline to n-Propylbenzene over a Pd Nanoparticle/Molecular Oxide Catalysta
a In the conversion of quinoline, 2-propylaniline could first be formed over the metal or the oxide site.
4.5. Conclusions. In contrast to N-heterocycle coordination, which is observed with metals throughout the periodic table, and hydrogenation, which is observed primarily with late metals, C−N hydrogenolysis is observed primarily with early metals. While these examples are not catalytic, they demonstrate the potential role of early metals for hydrotreating catalysts. Similar to the case for current industrial catalysts, single-site metal catalysts offer the potential to provide separate sites for hydrogenation and hydrogenolysis for selective N atom removal. In contrast to product distributions over industrial catalysts which are generally consistent with initial saturation of the heterocyclic ring, recent studies reviewed here argue that complete ring hydrogenation is not required for C−N cleavage and that direct denitrogenation may be a viable pathway at H2 pressures lower than those typically employed for denitrogenation via a hydrogenation pathway. Most of these models suggest that only the carbon atom involved in bond cleavage must be sp3 hybridized, regardless of the remainder of the heterocyclic ring. Furthermore, Wolczanski et al. showed that oxidative addition of an sp2 C−N bond to a low-valent complex is possible.104 Finally, molecule-derived oxide materials show that with addition of a second active site for H2 activation, e.g. Pd nanoparticles, these concepts can be incorporated into catalysts that only partially saturate the heterocyclic ring, leading to a significant reduction in H2 consumption and milder reaction conditions and, hence, overall energy efficiency gains for world-scale chemical processes.
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail for T.J.M.:
[email protected]. *E-mail for J.M.N.:
[email protected]. Present Address
5. CONCLUSIONS AND PERSPECTIVE Organometallic models provide insight into alternative mechanisms for HDN processes as well as inspirations that more selective HDN catalysts can be developed, particularly those that would exclusively saturate or partially saturate heterocyclic rings. Selective coordination and saturation of the heterocyclic ring of N-heteroaromatics would be of greatest value for naphtha hydrotreating, where minimizing hydrogenation also minimizes the need for subsequent dehydrogenation via catalytic reforming to achieve high-octane fuel. Conditions which saturate the heterocyclic ring exclusively would likely also minimize saturation of unsubstituted aromatics, thereby preventing a significant temperature rise in hydrotreating reactors from highly exothermic hydrogenation reactions, and thus enhance catalyst lifetimes. These studies also show that an electrophilic metal center can prepare N
§
Department of Chemical Engineering and Materials Science, University of MinnesotaTwin Cities, 421 Washington Avenue Southeast, Minneapolis, MN 55455, USA. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge the ACS Petroleum Research Fund and the DOE Office of Basic Sciences Grants SC-0006718 (M.B., J.M.N.) and 86ER1311 (M.B., T.J.M.) for funding.
■
REFERENCES
(1) Tissot, B. P.; Welte, D. H. Petroleum Formation and Occurrence, 2nd ed.; Springer-Verlag: Berlin, New York, 1984.
1474
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis (2) Kokayeff, P.; Zink, S.; Roxas, P. Handbook of Petroleum Processing 2014, 361−434. (3) Katzer, J. R.; Sivasubramanian, R. Catal. Rev.: Sci. Eng. 1979, 20, 155−208. (4) Furimsky, E.; Massoth, F. E. Catal. Rev.: Sci. Eng. 2005, 47, 297− 489. (5) Ho, T. C. Catal. Rev.: Sci. Eng. 1988, 30, 117−160. (6) Hansen, L. P.; Johnson, E.; Brorson, M.; Helveg, S. J. Phys. Chem. C 2014, 118, 22768−22773. (7) Hansen, L. P.; Ramasse, Q. M.; Kisielowski, C.; Brorson, M.; Johnson, E.; Topsoe, H.; Helveg, S. Angew. Chem., Int. Ed. 2011, 50, 10153−10156. (8) Kisielowski, C.; Ramasse, Q. M.; Hansen, L. P.; Brorson, M.; Carlsson, A.; Molenbroek, A. M.; Topsoe, H.; Helveg, S. Angew. Chem., Int. Ed. 2010, 49, 2708−2710. (9) Bachrach, M.; Morlanes-Sanchez, N.; Canlas, C. P.; Miller, J. T.; Marks, T. J.; Notestein, J. M. Catal. Lett. 2014, 144, 1832−1838. (10) Bianchini, C.; Meli, A.; Vizza, F. Eur. J. Inorg. Chem. 2001, 2001, 43−68. (11) Borowski, A. F. Polish J. Chem. 2006, 80, 205−226. (12) Sánchez-Delgado, R. A. Organometallic Modeling of the Hydrodesulfurization and Hydrodenitrogenation Reactions; Kluwer Academic: Dordrecht, Boston, 2002. (13) Weller, K. J.; Fox, P. A.; Gray, S. D.; Wigley, D. E. Polyhedron 1997, 16, 3139−3163. (14) Diaconescu, P. L. Curr. Org. Chem. 2008, 12, 1388−1405. (15) Laine, R. M. Ann. N. Y. Acad. Sci. 1983, 415, 271−291. (16) Angelici, R. J. Polyhedron 1997, 16, 3073−3088. (17) Jian, M.; Prins, R. Catal. Lett. 1998, 50, 9−13. (18) Carver, C. T.; Diaconescu, P. L. J. Am. Chem. Soc. 2008, 130, 7558−7559. (19) Thompson, M. E.; Baxter, S. M.; Bulls, A. R.; Burger, B. J.; Nolan, M. C.; Santarsiero, B. D.; Schaefer, W. P.; Bercaw, J. E. J. Am. Chem. Soc. 1987, 109, 203−219. (20) Miller, K. L.; Williams, B. N.; Benitez, D.; Carver, C. T.; Ogilby, K. R.; Tkatchouk, E.; Goddard, W. A.; Diaconescu, P. L. J. Am. Chem. Soc. 2010, 132, 342−355. (21) Carver, C. T.; Benitez, D.; Miller, K. L.; Williams, B. N.; Tkatchouk, E.; Goddard, W. A.; Diaconescu, P. L. J. Am. Chem. Soc. 2009, 131, 10269−10278. (22) Jordan, R. F.; Taylor, D. F. J. Am. Chem. Soc. 1989, 111, 778− 779. (23) Jordan, R. F.; Guram, A. S. Organometallics 1990, 9, 2116−2123. (24) Klei, E.; Teuben, J. H. J. Organomet. Chem. 1981, 214, 53−64. (25) Bailey, B. C.; Fan, H.; Huffman, J. C.; Baik, M. H.; Mindiola, D. J. J. Am. Chem. Soc. 2006, 128, 6798−6799. (26) Fout, A. R.; Bailey, B. C.; Buck, D. M.; Tan, H. J.; Huffman, J. C.; Baik, M. H.; Mindiola, D. J. Organometallics 2010, 29, 5409−5422. (27) Covert, K. J.; Neithamer, D. R.; Zonnevylle, M. C.; Lapointe, R. E.; Schaller, C. P.; Wolczanski, P. T. Inorg. Chem. 1991, 30, 2494− 2508. (28) Neithamer, D. R.; Parkanyi, L.; Mitchell, J. F.; Wolczanski, P. T. J. Am. Chem. Soc. 1988, 110, 4421−4423. (29) Bonanno, J. B.; Veige, A. S.; Wolczanski, P. T.; Lobkovsky, E. B. Inorg. Chim. Acta 2003, 345, 173−184. (30) Kleckley, T. S.; Bennett, J. L.; Wolczanski, P. T.; Lobkovsky, E. B. J. Am. Chem. Soc. 1997, 119, 247−248. (31) Allen, K. D.; Bruck, M. A.; Gray, S. D.; Kingsborough, R. P.; Smith, D. P.; Weller, K. J.; Wigley, D. E. Polyhedron 1995, 14, 3315− 3333. (32) Gray, S. D.; Smith, D. P.; Bruck, M. A.; Wigley, D. E. J. Am. Chem. Soc. 1992, 114, 5462−5463. (33) Zhu, G.; Pang, K.; Parkin, G. Inorg. Chim. Acta 2008, 361, 3221−3229. (34) Zhu, G.; Tanski, J. M.; Churchill, D. G.; Janak, K. E.; Parkin, G. J. Am. Chem. Soc. 2002, 124, 13658−13659. (35) Zhu, G.; Pang, K.; Parkin, G. J. Am. Chem. Soc. 2008, 130, 1564−1565.
(36) Sattler, A.; Zhu, G.; Parkin, G. J. Am. Chem. Soc. 2009, 131, 7828−7838. (37) Zhu, G.; Janak, K. E.; Figueroa, J. S.; Parkin, G. J. Am. Chem. Soc. 2006, 128, 5452−5461. (38) Rakowski DuBois, M.; Vasquez, L. D.; Ciancanelli, R. F.; Noll, B. C. Organometallics 2000, 19, 3507−3515. (39) Fish, R. H.; Fong, R. H.; Tran, A.; Baralt, E. Organometallics 1991, 10, 1209−1212. (40) Chaudret, B.; Jalon, F. A. J. Chem. Soc., Chem. Commun. 1988, 711−713. (41) Chaudret, B.; Jalon, F.; Perezmanrique, M.; Lahoz, F.; Plou, F. J.; Sánchez-Delgado, R. Nouv. J. Chim. 1990, 14, 331−338. (42) Cordone, R.; Taube, H. J. Am. Chem. Soc. 1987, 109, 8101− 8102. (43) Cordone, R.; Harman, W. D.; Taube, H. J. Am. Chem. Soc. 1989, 111, 2896−2900. (44) Arce, A. J.; Machado, R.; DeSanctis, Y.; Capparelli, M. V.; Atencio, R.; Manzur, J.; Deeming, A. J. Organometallics 1997, 16, 1735−1742. (45) Deeming, A. J.; Arce, A. J.; Desanctis, Y.; Day, M. W.; Hardcastle, K. I. Organometallics 1989, 8, 1408−1413. (46) Arce, A. J.; Manzur, J.; Marquez, M.; Desanctis, Y.; Deeming, A. J. J. Organomet. Chem. 1991, 412, 177−193. (47) Vicic, D. A.; Jones, W. D. Organometallics 1999, 18, 134−138. (48) Fish, R. H.; Kim, H. S.; Babin, J. E.; Adams, R. D. Organometallics 1988, 7, 2250−2252. (49) Chen, S.; Vasquez, L.; Noll, B. C.; Rakowski DuBois, M. Organometallics 1997, 16, 1757−1764. (50) Perrin, L.; Werkema, E. L.; Eisenstein, O.; Andersen, R. A. Inorg. Chem. 2014, 53, 6361−6373. (51) Rosales, M.; Vallejo, R.; Bastidas, L. J.; Gonzalez, B.; Gonzalez, A. React. Kinet. Catal. Lett. 2007, 92, 99−104. (52) Hosmane, R. S.; Liebman, J. F. Struct. Chem. 2009, 20, 693−697. (53) Rosales, M.; Alvarado, Y.; Boves, M.; Rubio, R.; Soscun, H.; Sánchez-Delgado, R. A. Transition Met. Chem. 1995, 20, 246−251. (54) Rosales, M.; Navarro, J.; Sanchez, L.; Gonzalez, A.; Alvarado, Y.; Rubio, R.; DeLaCruz, C.; Rajmankina, T. Transition Met. Chem. 1996, 21, 11−15. (55) Rosales, M.; Boves, M.; Soscun, H.; Ruette, F. J. Mol. Struct.: THEOCHEM 1998, 433, 319−328. (56) Sánchez-Delgado, R. A.; Gonzalez, E. Polyhedron 1989, 8, 1431−1436. (57) Zhou, H. F.; Li, Z. W.; Wang, Z. J.; Wang, T. L.; Xu, L. J.; He, Y.; Fan, Q. H.; Pan, J.; Gu, L. Q.; Chan, A. S. C. Angew. Chem., Int. Ed. 2008, 47, 8464−8467. (58) Urban, S.; Ortega, N.; Glorius, F. Angew. Chem., Int. Ed. 2011, 50, 3803−3806. (59) Borowski, A. F.; Sabo-Etienne, S.; Donnadieu, B.; Chaudret, B. Organometallics 2003, 22, 1630−1637. (60) Borowski, A. F.; Sabo-Etienne, S.; Chaudret, B. J. Mol. Catal. A: Chem. 2001, 174, 69−79. (61) Albright, J. O.; Datta, S.; Dezube, B.; Kouba, J. K.; Marynick, D. S.; Wreford, S. S.; Foxman, B. M. J. Am. Chem. Soc. 1979, 101, 611− 619. (62) Leong, V. S.; Cooper, N. J. Organometallics 1988, 7, 2058−2060. (63) Thompson, R. L.; Lee, S.; Rheingold, A. L.; Cooper, N. J. Organometallics 1991, 10, 1657−1659. (64) Bennett, M. A.; Lu, Z. B.; Wang, X. Q.; Bown, M.; Hockless, D. C. R. J. Am. Chem. Soc. 1998, 120, 10409−10415. (65) Matthes, J.; Grundemann, S.; Toner, A.; Guari, Y.; Donnadieu, B.; Spandl, J.; Sabo-Etienne, S.; Clot, E.; Limbach, H. H.; Chaudret, B. Organometallics 2004, 23, 1424−1433. (66) Alvanipour, A.; Kispert, L. D. J. Mol. Catal. 1988, 48, 277−283. (67) Nunez-Rico, J. L.; Fernandez-Perez, H.; Benet-Buchholz, J.; Vidal-Ferran, A. Organometallics 2010, 29, 6627−6631. (68) Fish, R. H.; Baralt, E.; Smith, S. J. Organometallics 1991, 10, 54− 56. (69) Dudnik, A. S.; Weidner, V. L.; Motta, A.; Delferro, M.; Marks, T. J. Nat. Chem. 2014, 6, 1100−1107. 1475
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476
Review
ACS Catalysis (70) Coperet, C. Chem. Rev. 2010, 110, 656−680. (71) Hartley, F. R.; Vezey, P. N. Adv. Organomet. Chem. 1977, 15, 189−234. (72) Bianchini, C.; Dal Santo, V.; Meli, A.; Oberhauser, W.; Psaro, R.; Vizza, F. Organometallics 2000, 19, 2433−2444. (73) Sahoo, S.; Kumar, P.; Lefebvre, F.; Halligudi, S. B. J. Mol. Catal. A: Chem. 2007, 273, 102−108. (74) Pertici, P.; Vitulli, G.; Carlini, C.; Ciardelli, F. J. Mol. Catal. 1981, 11, 353−364. (75) Islam, S. M.; Bose, A.; Palit, B. K.; Saha, C. R. J. Catal. 1998, 173, 268−281. (76) Islam, S. M.; Palit, B. K.; Mukherjee, D. K.; Saha, C. R. J. Mol. Catal. A: Chem. 1997, 124, 5−20. (77) Islam, S. M.; Saha, C. R. J. Mol. Catal. A: Chem. 2004, 212, 131− 140. (78) Blaser, H. U.; Pugin, B.; Spindler, F.; Togni, A. C. R. Chim. 2002, 5, 379−385. (79) Pugin, B. J. Mol. Catal. A: Chem. 1996, 107, 273−279. (80) Pugin, B.; Landert, H.; Spindler, F.; Blaser, H. U. Adv. Synth. Catal. 2002, 344, 974−979. (81) Ayala, V.; Corma, A.; Iglesias, M.; Rincon, J. A.; Sanchez, F. J. Catal. 2004, 224, 170−177. (82) Gonzalez-Arellano, C.; Corma, A.; Iglesias, M.; Sanchez, F. Adv. Synth. Catal. 2004, 346, 1316−1328. (83) Boucher, L. J.; Holy, N. L.; Davis, B. H. In New Approaches in Coal Chemistry; American Chemical Society: Washington, DC, 1981; Vol. 169, pp 319−330. (84) Boucher, L. J.; Dadey, E. J. Abstracts of Papers, 184th National Meeting of the American Chemical Society; American Chemical Society: Washington, DC, 1982; 654660 (85) Lamping, A. E.; Guo, X. Y.; Rempel, G. L. J. Mol. Catal. 1994, 87, 75−93. (86) Kwart, H.; Schuit, G. C. A.; Gates, B. C. J. Catal. 1980, 61, 128− 134. (87) Fish, R. H.; Thormodsen, A. D.; Heinemann, H. J. Mol. Catal. 1985, 31, 191−198. (88) Fish, R. H.; Tan, J. L.; Thormodsen, A. D. J. Org. Chem. 1984, 49, 4500−4505. (89) Bianchini, C.; Frediani, M.; Mantovani, G.; Vizza, F. Organometallics 2001, 20, 2660−2662. (90) Bianchini, C.; Dal Santo, V.; Meli, A.; Moneti, S.; Moreno, M.; Oberhauser, W.; Psaro, R.; Sordelli, L.; Vizza, F. J. Catal. 2003, 213, 47−62. (91) Breysse, M.; Berhault, G.; Kasztelan, S.; Lacroix, M.; Mauge, F.; Perot, G. Catal. Today 2001, 66, 15−22. (92) Nelson, N.; Levy, R. B. J. Catal. 1979, 58, 485−488. (93) Zhao, Y.; Czyzniewska, J.; Prins, R. Catal. Lett. 2003, 88, 155− 162. (94) Monreal, M. J.; Khan, S.; Diaconescu, P. L. Angew. Chem., Int. Ed. 2009, 48, 8352−8355. (95) Yi, W. Y.; Zhang, J.; Huang, S. J.; Weng, L. H.; Zhou, X. G. Chem. - Eur. J. 2014, 20, 867−876. (96) Duhovic, S.; Monreal, M. J.; Diaconescu, P. L. J. Organomet. Chem. 2010, 695, 2822−2826. (97) Sharma, M.; Botoshanskii, M.; Bannenberg, T.; Tamm, M.; Eisen, M. S. C. R. Chim. 2010, 13, 767−774. (98) Fout, A. R.; Bailey, B. C.; Tomaszewski, J.; Mindiola, D. J. J. Am. Chem. Soc. 2007, 129, 12640−12461. (99) Gray, S. D.; Weller, K. J.; Bruck, M. A.; Briggs, P. M.; Wigley, D. E. J. Am. Chem. Soc. 1995, 117, 10678−10693. (100) Weller, K. J.; Filippov, I.; Briggs, P. M.; Wigley, D. E. Organometallics 1998, 17, 322−329. (101) Weller, K. J.; Gray, S. D.; Briggs, P. M.; Wigley, D. E. Organometallics 1995, 14, 5588−5597. (102) Smith, D. P.; Strickler, J. R.; Gray, S. D.; Bruck, M. A.; Holmes, R. S.; Wigley, D. E. Organometallics 1992, 11, 1275−1288. (103) Choi, J. G.; Brenner, J. R.; Colling, C. W.; Demczyk, B. G.; Dunning, J. L.; Thompson, L. T. Catal. Today 1992, 15, 201−222.
(104) Bonanno, J. B.; Henry, T. P.; Neithamer, D. R.; Wolczanski, P. T.; Lobkovsky, E. B. J. Am. Chem. Soc. 1996, 118, 5132−5133. (105) Huertos, M. A.; Perez, J.; Riera, L. J. Am. Chem. Soc. 2008, 130, 5662−5663. (106) Huertos, M. A.; Perez, J.; Riera, L.; Menedez-Veldazquez, A. J. Am. Chem. Soc. 2008, 130, 13530−13531.
1476
DOI: 10.1021/acscatal.5b02286 ACS Catal. 2016, 6, 1455−1476