Targeting of High-Valent Iron-TAML Activators at Hydrocarbons and

May 10, 2017 - Dr. Collins is the Teresa Heinz Professor of Green Chemistry and the Director of the Institute for Green Science at Carnegie Mellon Uni...
2 downloads 8 Views 2MB Size
Download PDF
Review pubs.acs.org/CR

Targeting of High-Valent Iron-TAML Activators at Hydrocarbons and Beyond Terrence J. Collins* and Alexander D. Ryabov* Department of Chemistry, Carnegie Mellon University, 4400 Fifth Avenue, Pittsburgh, Pennsylvania 15213, United States ABSTRACT: TAML activators of peroxides are iron(III) complexes. The ligation by four deprotonated amide nitrogens in macrocyclic motifs is the signature of TAMLs where the macrocyclic structures vary considerably. TAML activators are exceptional f unctional replicas of the peroxidases and cytochrome P450 oxidizing enzymes. In water, they catalyze peroxide oxidation of a broad spectrum of compounds, many of which are micropollutants, compounds that produce undesired effects at low concentrationsas with the enzymes, peroxide is typically activated with near-quantitative efficiency. In nonaqueous solvents such as organic nitriles, the prototype TAML activator gave the structurally authenticated reactive iron(V)oxo units (FeVO), wherein the iron atom is two oxidation equivalents above the FeIII resting state. The iron(V) state can be achieved through the intermediacy of iron(IV) species, which are usually μ-oxo-bridged dimers (FeIVFeIV), and this allows for the reactivity of this potent reactive intermediate to be studied in stoichiometric processes. The present review is primarily focused at the mechanistic features of the oxidation by FeVO of hydrocarbons including cyclohexane. The main topic is preceded by a description of mechanisms of oxidation of thioanisoles by FeVO, because the associated studies provide valuable insight into the ability of FeVO to oxidize organic molecules. The review is opened by a summary of the interconversions between FeIII, FeIVFeIV, and FeVO species, since this information is crucial for interpreting the kinetic data. The highest reactivity in both reaction classes described belongs to FeVO. The resting state FeIII is unreactive oxidatively. Intermediate reactivity is typically found for FeIVFeIV; therefore, kinetic features for these species in interchange and oxidation processes are also reviewed. Examples of using TAML activators for C−H bond cleavage applied to fine organic synthesis conclude the review.

CONTENTS 1. Introduction 2. Formation and Properties of Iron(IV) and Iron(V) oxo TAML Derivatives 3. Mutual Transformations between Iron(III), Iron(IV), and Iron(V) Species: the TAML Triangle 3.1. The TAML Triangle 3.2. From FeIIIOH2 to FeIVFeIV 3.3. From FeIVFeIV to FeVO 3.4. Comproportionation of FeIIIOH2 and FeVO to Form FeIVFeIV 3.5. Spontaneous Reduction of FeVO to FeIVFeIV 4. Kinetics and Mechanisms of Oxidation of Organic Sulfides 5. Oxidation of Hydrocarbons: C−H Bond Cleavage versus Electron Transfer 5.1. Products of Hydrocarbon Oxidations 5.2. Kinetic Studies 5.3. Activation Parameters, Kinetic Isotope Effect, and Tunneling 5.4. Pathway for 9,10-Dihydroanthracene Oxidation 5.5. Comparisons with other Non-Heme HighValent Iron Complexes 5.6. Comparative Reactivity of Ironoxo TAML Species: FeIVFeIV versus FeVO 6. Miscellaneous 6.1. Oxidation of NADH into NAD+ © 2017 American Chemical Society

6.2. Functionalization of Secondary C−H Bonds 6.3. Cleavage of C−N Bonds 7. Implications and Conclusions Author Information Corresponding Authors ORCID Notes Biographies Acknowledgments Abbreviations References

9140 9142 9144 9144 9144 9145 9145 9146

9155 9156 9156 9157 9157 9157 9157 9157 9158 9158 9158

1. INTRODUCTION TAML activators were the first major fruit of an iterative design protocol launched in 19801−3 to achieve effective small molecule mimics of oxidizing enzymes. It took 15 years for the protocol to yield the first TAML activatorthe trademark is registered to Carnegie Mellon University covering compositions of matter captured in an internationalized patent.4 The oxidative chemistry of TAML activators (Chart 1) began appearing in scholarly papers and conferences in 1998,5−7 the year that two additional design elements for strategically achieving novel oxidation catalysis were reported including the penultimate design motifs

9147 9149 9150 9151 9151 9153 9154 9154 9154 9155

Special Issue: CH Activation Received: January 17, 2017 Published: May 10, 2017 9140

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Chart 1. Compounds 1−5 Are TAMLs from the First to the Fifth Generations, Respectively, Shown in Figure 1; TAMLs 6 and 7 Belong to Generations 6 and 7, Respectively

before TAML activators were achieved.8,9 The final design step giving TAML catalysts depended on solving a puzzling mystery. When, at what turned out to be the penultimate design stage before transformative TAML activators could be achieved, complex 1c was reacted with tert-butylhydroperoxide in acetonitrile under ambient conditions, a deep blue complex formed that proved to be the one-electron-oxidized, formally FeIV(CN) complex. However, the infrared spectrum exhibited an inexplicable band at 1723 cm −1 which was eventually demonstrated to originate in a white hydantoin degradation product of a small component (10%) of the macrocyclic complex, intermingled and hidden in the deep blue solid FeIV(CN). The ligand decomposition was associated with the intramolecular methylene oxidative C−H bond cleavage in the tail of 1c (Chart 1), giving the first evidence that oxidized TAML species could attack aliphatic C−H bonds. Fortunately for our program, hydantoin rings exhibit extremely strong IR bands in this region.5 This motivating finding inspired further development of the fascinating oxidation chemistry of TAML activators, which behave as full f unctional (i.e., follow the basic steps in the catalytic cycle) peroxidase and peroxide short-circuited cytochrome P450 replicas to efficiently catalyze a myriad of oxidative degradations by hydrogen peroxide in water. The first substrates investigated were dyes7 where the impact of catalytic performance of exchanging an oxidatively vulnerable −CH2Me (R = Et in 1c) with less vulnerable −CH3 (R = Me in 1a) became graphically evidentthe bleaching of ca. 100 dyes has since been studiedfollowed by a broad grouping of environmentally threatening water contaminants.10−13 From the late 1990s endocrine disruptors became our prioritized targets.14 The low

dose adverse effects of endocrine disruptors on aquatic organisms is key to the concept of the micropollutant, a term for any substance that exhibits undesirable effects at low concentrations. By 2015, while in the pursuit of ever better catalysts that could activate hydrogen peroxide to purify water of oxidizable chemical contaminants and microbial pathogens, we had developed five generations of TAML activators presented in Figure 1 as they have been displayed on the front cover picture of ChemistryA European Journal.15,16 These structures are represented in Chart 1, which details the different ligand substituents in the various formulations of each generation. TAML chemistry has been growing so rapidly that by 2016 Figure 1 was already out of date. Compounds in Figure 1 and those 1−5 in Chart 1 all incorporate an aromatic ring within the ligand structure. TAML activator 6 of the sixth generation (Chart 1) without aromatic unit was synthesized and explored in 2016.17−19 The recognition that the free energies of activation for substrate oxidation and substrate degradation are linearly related over 15 TAML activators17 led us to conclude that our basic hypothesis for catalyst degradation, that oxidation alone was the lifetime controlling chemistry, had ceased to be correct for all TAML activators produced after the prototype, 1a. This insight has led to the development of new catalyst compositions, which we are showing outperform to the point of rendering obsolete TAML activators in the 1−6 generationsthese new catalysts (new-TAML activators 7, Chart 1) have been introduced in 2017.20,21 Most of the experimental work carried out so far was performed using the first generation activator 1a (Chart 1) so 9141

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

been trapped in TAML systems, but we regard this as simply a matter of time. This review is centered just around ironoxo TAML complexes with iron in the oxidation states of 4 and 5. These higher oxidation states are produced by reaction of iron(III) TAMLs with various strong oxidizing agents. It is vital to understand deeply the chemistry of oxidative processes which make it possible for selected TAMLs in Chart 1 to activate C−H bonds and catalyze hydrocarbon conversions of synthetic value. The mechanistic picture of C−H bond activation by high-valent iron TAML activators is never complete without such knowledge. Therefore, this review covers the relevant information on the chemistry that precedes C−H bond activation. Brief surveys follow of (i) known TAML activators with iron in oxidation states above 3 with oxygen-containing ligands, (ii) the formation pathways and properties of iron(IV) and iron(V)oxo TAML derivatives, and (iii) interconversions of FeIII, FeIV, and FeV species under the reaction conditions typically used for C−H bond activation. The results of mechanistic studies of the oxidation of organic sulfides by iron(V) TAMLs are also included, since this information is prerequisite to a deep look at the mechanisms of C−H bond cleavage, the key objective of the present account. Throughout, iron TAML derivatives in oxidation states 3 and above will be abbreviated as symbols employed in Scheme 1. Subscripts in bold will indicate a parent

Figure 1. First five generations of TAML activators. Adapted with permission from ref 16. Copyright 2015 John Wiley and Sons.

Scheme 1. Known TAML Ironoxo Species in Oxidation States 3 and Abovea

that 1a is the most frequent subject catalyst in the present review. This catalyst is commercially available. Many researchers have purchased it to study various catalytic processes, including C−H bond activation as detailed herein. Each TAML generation has brought something new to TAML chemistry. In particular, activators 422,23 are so far the most reactive catalysts for the activation of hydrogen peroxide and oxidation of various targets in aqueous solution.11,24 Comparison of the kinetic behavior of the six generations shows that large reactivity changes are consequent to apparently minor structural modifications. For example, 2 catalysts hold many features in common with 4 activators but contain an extra −CMe2− unit. The latter expands the macrocycle and makes 4 substantially more susceptible to demetalation in waterthe 6,5,6,5-ring system produces a cavity that is too large with four planar amido-N ligands for the ferric ion. The system adjusts by forcing the formation of one distinctly nonplanar amido-N ligand (in the solid state) that we attribute as the cause of the observed enhanced hydrolytic instability.25 For some TAML activators such as 1 and 5, it is convenient to identify the “head” and the “tail” parts of the molecules (see Chart 1). Following these identifications, 6 exemplifies a “beheaded” version of TAML activators.17−19 The major goal of the present account is to describe the potential and highlight the mechanistic peculiarities of TAML species in the oxidation of hydrocarbons involving initial C−H bond cleavage. All TAMLs in Chart 1 contain iron(III). The charge of the cation is effectively neutralized by four amidato nitrogens leaving a residual anionic charge that facilitates solubility in water and other polar solvents, where such solubility can be strongly controlled by the nature of the countercation. As a result of the high σ-donor capacity and four anions akin to a straightjacket of negative charge, the oxidative power of TAMLs in the resting ferric state is effectively nonexistent for common reductants. Oxidative reactivity starts to manifest when TAML iron finds itself in oxidation states higher than 3, i.e. slightly oxidizing FeIV and strongly oxidizing FeV. To date, FeVI has not

a

Apart from the FeIIIOH2 precursor, all have the potential to oxidize various substrates. Compounds set in bold were also characterized in aqueous solutions.

TAML compound which can be found in Chart 1. For example, the symbol FeIVFeIV1a corresponds to the iron(IV) μ-oxo dimer (see Chart 2) produced from TAML activator 1a.

2. FORMATION AND PROPERTIES OF IRON(IV) AND IRON(V)OXO TAML DERIVATIVES The four deprotonated amide nitrogens of TAMLs tune the reduction potentials of the iron such that it typically acquires the oxidation state of 3 under the ambient conditions. Correspondingly, oxo TAMLs containing iron in the oxidation states higher than 3 are plausible intermediates in oxidative catalysis or stoichiometric oxidations including conversion of hydrocarbon 9142

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Chart 2. Structures of the Key Species That Participate in C−H Bond Activations, FeIIIOH2, FeIVFeIV, FeVO, and FeIVFeIV That Is Formed from FeIIIOH2 and FeVOa

a “Z” stands for CR2 (1) or NR (5) (R = alkyl or fluorine). The formal iron−oxygen bond order of 2.5 in FeVO reflects the contracted Fe−O bond distance of 1.60 Å estimated by EXAFS.31

oxidizing water, a reactivity that has now been firmly established.33−35 It is important to note that FeVO species are produced using specific oxidizing agents even in organic solvents at −40 °C.36 The iron(V) state is nearly quantitatively generated from iron(III) either by already mentioned mCPBA or by NaClO usually via FeIVFeIV.36,37 In the case of 1a hydrogen peroxide and organic peroxides (benzoyl peroxide, tert-butylperoxide, and tertbutylhydroperoxide) do convert FeIIIOH2 into the dimeric species FeIVFeIV; i.e. they increase the oxidation state of FeIIIOH2 just by one oxidation equivalent.36 This at first glance lessimportant observation has a significant mechanistic implication for interpreting mechanisms of TAML-catalyzed oxidations. The high oxidation state of 5 is often postulated in the literature in oxidative reactions involving hydrogen or organic peroxides. Extra caution should be taken in such cases because as the data for TAMLs show36 such oxidation states may be unachievable in the presence of certain oxidants. Prior to the development of TAML 6, iron(V) states of TAMLs had not been achieved in pure water, though FeVO5a is reported to be rather stable in the presence of 10% acetonitrile at ambient conditions.35 Typically, the oxidative conversion of iron(III) TAMLs in water affords monomeric iron(IV) FeIVO and/or dimeric FeIVFeIV derivatives.30 The monomeric and dimeric species are interrelated by a pH-controlled equilibrium (eq 1). The monomer dominates at pH >12, whereas the dimer dominates at pH mCPBA > tBuOOH > benzoyl peroxide > H2O2.36 The mechanism of FeIIIOH2 → FeIVFeIV conversion depends on the nature of oxidant used. The kinetic data is consistent with a heterolytic O−O bond cleavage mechanism for the oxidation by mCPBA.37,39 The effective rate constant k3/4 for 1a is lower than that for more electron deficient 1b.39 This might indicate that the higher Lewis acidity expected for iron in 1b increases the reactivity of mCPBA toward FeIIIOH2 by holding the oxidant once bound more tightly. However, a vital feature here is likely to be the Brønsted acidity of the coordinated peroxyacid which is expected to be greater in 1b than in 1a. The removal of the acidic proton from the coordinated oxygen, by either liberation of the proton into solution or its transfer to the carbonyl oxygen within the postulated five-membered cyclic intermediate in Scheme 2, is probably key to driving the heterolytic O−O bond cleavage. The final FeIVFeIV product is formed via rapid comproportionation (Scheme 2). In contrast to all other oxidants, the conversion of 1a into FeIVFeIV by tBuOOH, the stoichiometry of which is depicted in eq 2, is characterized by highly unusual kinetics.36

3.1. The TAML Triangle

Chart 2 brings together the exact structures of TAMLs FeIIIOH2, FeIVFeIV, and FeVO. Formally, FeIVFeIV is an addition product of FeIIIOH2 to FeVO. Not surprisingly, FeIIIOH2 and FeVO usually react very fast to form FeIVFeIV so that FeIIIOH2 formed in oxidation processes may be and sometimes is responsible for the consumption of FeVO in the reaction mixture, quenching the high reactivity of the latter. This is an important feature of TAML catalytic chemistry when the concentration is “high” (>1 μM) and especially when noncatalytic processes are being followed. In particular, the comproportionation should be taken into account when the reactivity of FeVO species toward C−H bonds of hydrocarbons is under investigation in stoichiometric processes. Such processes are studied kinetically following a general routine commonly used for kinetic investigations of C−H bond activations by high-valent iron species. The procedure involves the monitoring of the consumption of FeVO by UV−vis spectroscopy in the presence of an excess of a corresponding oxidation target (donor of electrons). If FeIIIOH2 is generated from FeVO as the target is oxidized, the observed rate constants may be attenuated by the comproportionation process. Thus, developing a full mechanistic picture of oxidation requires the quantitative evaluation of all interconversion pathways involving FeIIIOH2, FeIVFeIV, and FeVO in solution, including the generation of FeIVFeIV and FeVO from FeIIIOH2 in the presence of oxygen transfer oxidants. Such detailed studies were first performed with TAMLs 1a and 1c employing mCPBA as the oxidant in acetonitrile at −40 °C.37 It is convenient to 9144

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Scheme 2. Heterolytic Mechanism of Formation of FeIVFeIV in the Presence of mCPBA39

produced through the comproportionation (step 6). The intimate details of the postulated initiation step 3 involving homolytic transformation(s) of tBuOOH are not fully understood. It is likely that 2,2,6,6-tetramethylpiperidine makes the reaction light-insensitive by deprotonating tBuOOH (step 1) and opening the channel for the formation of the alkylperoxide complex of FeIII (step 2), which collapses to the same product as in step 5 through the heterolysis of the O−O bond. However, the possibilities of invisible electron transfers or O−O bond homolyses involving the various iron species that could be assisted by light are many. Although amines other than 2,2,6,6tetramethylpiperidine may also deprotonate tBuOOH (step 1), the binding of tBuOO− to iron(III) is precluded due to generation of the unreactive species with blocked axial sites. Steps 7 and 8, which are typical of free-radical transformations,41 account for the formation of organic products, viz. tBuOH, acetone, and methanol. Minor quantities of MeOH are likely produced via the recombination of methyl radicals with dioxygen (with the CH3OO• then undergoing reactions mirroring the fate of tBuOO•).

IV IV (H C) COH + 1a + (H3C)3 COOH → Fe Fe + 3 3 100% 42%

(H3C)2 CO + H3COH 45%

traces

(2)

The rate of formation of FeIVFeIV at −40 °C in MeCN solvent increases as the reaction proceeds, being the highest at 100% conversion! Reaction 2 is first order in tBuOOH during all phases and close to zero order in 1a. The speed of reaction 2 is affected by light, O2, TEMPO, and bases (B). Most of the bases tested stop the reaction: 1,8-bis(dimethylamino)naphthalene, tBuOK, and triethylamine. In contrast, 2,2,6,6-tetramethylpiperidine accelerates reaction 2. Acids have no effect on the rate. The experimental data obtained were rationalized in terms of a freeradical process shown in Scheme 3. The diverse effect of bases is presumably due to their variable ability to coordinate the iron(III) center of 1. Amines such as pyridine and imidazole bind to axial sites of iron(III) TAML activators even in water to form mono- and bis-adducts of the type Fe(TAML)B and Fe(TAML)B2.40 By these phenomena, the axial sites of the iron polyhedron become blocked, which inhibits the oxidation of FeIIIOH2 to FeIVFeIV. In our current interpretation, when the amine nitrogen is sterically protected from the binding to iron(III) as in the case of 2,2,6,6tetramethylpiperidine, the retardation mechanism suggested turns off and the amine starts to accelerate the reaction through the deprotonation of tBuOOH (Scheme 3). In the absence of base, the reaction occurs as a free-radical, possibly branched, process where light-induced, O2-dependent activation of t BuOOH may lead to radicals tBuOO• and/or tBuO• and •OH via step 3. In one potential pathway, the tBuOO• radical produced reacts rapidly with FeIII (Scheme 3, step 4) to afford the alkylperoxide complex of FeIV, which undergoes homolytic scission to afford tBuO• and FeVO (step 5). The final product is

3.3. From FeIVFeIV to FeVO

As noted above, this transformation of TAML derivatives 1 occurs with both mCPBA31,37,39,42 and NaClO,36,43 with hypochlorite being more reactive.36 There is spectral evidence that FeVO can also be generated photochemically from FeIVFeIV of TAMLs 5 using Na2S2O8 as the oxidant and [Ru(bpy)3]2+ as a photosensitizer.35 The oxidized species FeIVFeIV5 and FeVO5 are significantly more stable in solution under ambient conditions compared to their counterparts derived from 1. Therefore, their reactivity is lower and can be investigated without deepfreezing.42 Also, they are produced in the binary MeCN−H2O mixtures with water content up to 90%, although water significantly decreases the stability of FeVO5.43 The one-electron oxidation of FeIVFeIV to FeVO is slower than that of FeIIIOH2 to FeIVFeIV. Using the k3/4 and k4/5 formalism (Figure 2), the rate difference is 16 and 24 times for 1a and 1b, respectively.37,39 The value of k4/5 for sterically more demanding 1c is ca. 3 times larger than that for 1a.37 The effect is probably not due to an enhanced tendency of the bulkier FeIVFeIV dimer to produce FeVO through steric destabilization because k3/4 is also 3 times higher for 1c than for 1a. 3.4. Comproportionation of FeIIIOH2 and FeVO to Form FeIVFeIV

The recognition of iron(III/V) comproportionation long precedes the chemistry of TAMLs. The process attracted

Scheme 3. Tentative Mechanistic Description of the Free-Radical Oxidation of 1a into FeIVFeIV by tBuOOH in Wet MeCN at −40 °C

9145

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

3.5. Spontaneous Reduction of FeVO to FeIVFeIV

attention of researchers several decades ago in the still evolving wave of chemical popularity of iron(III)/iron(IV) porphyrins and iron(IV)−porphyrin radical cations (iron(IV)−P•+)the latter are isoelectronic with iron(V). The reaction between iron(IV)−P•+ and iron(III) porphyrins, which results in the formation of iron(IV) porphyrins, was first introduced by Jones et al.,44 to account for some inconsistencies in the titration of deuteroferriheme with peroxo acids. The presence of such comproportionation processes was strengthened by additional data later45 and then flourished in connection with oxidative catalysis by iron porphyrins.46−48 TAML derivatives have been employed for direct quantitative studies of the comproportionation in which several interesting features of the reaction have been revealed, with the most important being the pronounced sensitivity to the steric effects.37 When an equimolar amount of FeIIIOH2 (1a) is added to a solution of freshly prepared Fe V O 1a at −40 °C, the comproportionation leads quantitatively to FeIVFeIV, thus unequivocally proving the existence and viability in complex pathways of such processes in the TAML systems.37 The second order rate constants k3+5/4 (Figure 2), which are summarized in Table 1, indicate that the comproportionation is controlled by

Complexes FeVO have finite lifetimes in organic nitrile solvents even at low temperatures. In the absence of external reducing agents, FeVO transforms exponentially to FeIVFeIV.37,39,42 Accessing the corresponding rate constants k5/4 (Figure 2) is critical before any attempt is made to measure the reactivity of FeVO toward any target in a single turnover experiment. Fortunately, rather low values of k5/4, which are collected in Table 2 together with the available activation parameters ΔH‡ and ΔS‡, do not complicate reliable measurements of rates of reduction of FeVO by a wide spectrum of reagents including hydrocarbons by monitoring a decrease in the concentration of FeVO in excess of reducing agents. The reduction of FeVO into FeIVFeIV does not happen due to the cleavage of C−H bonds of acetonitrile. This is consistent with the measured solvent kinetic isotope effect of 1.09 for FeVO1b using CD3CN.39 Similar secondary kinetic isotope effects were registered for FeVO1a and FeVO1c.49 In the case of FeVO1b, the reduction does not seem to occur at the expense of the ligand system because the cycling between FeVO and FeIVFeIV using mCPBA could be performed several times (Figure 3).39 Since TAML catalysts are able to oxidize water to oxygen under appropriate conditions,33−35,50 dioxygen production may be the course of the FeVO into FeIVFeIV transformation.

Table 1. Rate Constants k3+5/4 (in M−1 s−1) for the Comproportionation of FeIIIOH2 and FeVO in Acetonitrile at −40 °C entry

FeIIIOH2

FeVO precursor

k3+5/4

ref

1 2 3 4 5

1a 1a 1c 1c 5a

1a 1c 1a 1c 5a

(4 ± 1) × 104 (1.50 ± 0.03) × 104 (1.0 ± 0.2) × 104 35 ± 1 ∼1 × 105

37 49 49 37 42

steric factors. The value of k3+5/4 decreases 1000-fold when the R group in both FeIIIOH2 and FeVO partners is changed from CH3 to C2H5 (cf. entries 1 and 4 in Table 1). The reactivity gap is even higher when the data for 1c and 5a are compared (entries 4 and 5). TAMLs 5 have the smallest “tail” part of the molecule and are significantly flatter than other generations, and the resulting muted steric effects are apparently expressed in the rate constant with k3+5/4 being the highest.42 The values of k3+5/4 for the “crosscomproportionation” (entries 2 and 3) when FeIIIOH2 and FeVO have been prepared from different TAMLs 1 of variable bulkiness are expectedly between the extreme values in entries 1 and 4. The finding of divergence of k3+5/4 is exceptionally valuable because it establishes a mechanistic probe for the participation of FeVO in either one- or two-electron reductions. One needs just to compare the relative amounts of FeIVFeIV or FeIIIOH2 formed, respectively (see section 4). Such a comparison is not possible for all TAMLs but is accessible for those with relatively low k3+5/4 such as 1c.

Figure 3. Recycling between FeVO and FeIVFeIV in the case of 1b. Addition of 0.5 equiv of mCPBA to FeIVFeIV caused an increase in absorbance at 613 nm consequent to the formation of FeVO (top black curve) and a decrease in the absorbance at 825 nm consequent to the disappearance of FeIVFeIV (bottom red curve). After reaching its highest concentration after all mCPBA had been consumed, FeVO underwent spontaneous reduction. Addition of the next 0.5 equiv of mCPBA repeated the cycle. Conditions: total iron 2.5 × 10−4 M, mCPBA 1.25 × 10−4 M; MeCN, 0.2% H2O (v/v), −40 °C. Adapted with permission from ref 39. Copyright 2015 John Wiley and Sons.

The ligand design of 1b rules out the mechanistic option for the reduction of FeVO via the intramolecular C−H abstraction as it is shown in Scheme 4. Such hydrogen abstraction was assumed

Table 2. Rate Constants k5/4 for Spontaneous Conversion of FeVO into FeIVFeIV in Acetonitrile at −40 °C

a

entry

FeVO precursor

k5/4/s−1

ΔH‡/kJ mol−1

ΔS‡/J K−1 mol−1

ref

1 2 3 4

1a 1c 1b 5a

(1.0 ± 0.1) × 10−5 (1.11 ± 0.02) × 10−4 (3.42 ± 0.02) × 10−4 4.45 × 10−5a

78 ± 8 61 ± 1

−30 ± 30 −58 ± 3

37, 49 37, 49 39 42

At 25 °C. 9146

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Scheme 4. Hydrogen Atom Abstraction of the Ethyl Group of FeVO Derived from the Ethyl-Tailed 1c As Originally Proposed by Bartos et al.5

to be the first step of the subsequent sequence of events leading to the degradation of the TAML ligand system, which has been noted in the Introduction.5 Complex FeVO1b does not have alkyl, in general, and ethyl, in particular, groups in its “tail” part. For FeVO TAML species with the tail alkyl groups, it is difficult to rule out completely that the reduction of FeVO is always free, at least partially, from ligand fragmentation. Some evidence in favor of the rate-limiting pathway similar to that shown in Scheme 4 is provided by the activation parameters for the k5/4 rate constant in Table 2. The enthalpy of activation ΔH‡ is by 17 kJ mol−1 lower for the ethyl-tailed FeVO1c as compared to the methyl-tailed counterpart FeVO1a, which is consistent with the lower bond dissociation energy of the secondary C−H bond compared to the primary C−H bond. At the same time the entropy of activation for 1c is more negative, suggesting a more pronounced entropy loss in the transition state for the hydrogen abstraction as shown in Scheme 4.

hold even in excess p-XC6H4SMe.37 The matching kinetics for the exponential disappearance of FeVO, which permitted the evaluation of the pseudo-first-order rate constants kobs, and the formation of the C6H5SOMe product has also been confirmed (Figure 4). It is essential to establish that the rate of reduction of

4. KINETICS AND MECHANISMS OF OXIDATION OF ORGANIC SULFIDES Organic sulfides such as thioanisole serve as standard substrates for the preliminary evaluation of the reactivity of oxidized ironoxo species.51,52 If an ironoxo derivative displays a weak reactivity toward thioanisole, there is a low probability that this oxidant will cleave C−H bonds of nonactivated hydrocarbons. Therefore, prior to turning to the C−H bond activation, the reactivity of TAML FeVO derivatives was explored in the oxidation of ring substituted thioanisoles p-XC6H4SMe to the corresponding sulfoxides (eq 3).37

Figure 4. Changes of the UV−vis spectrum of FeVO1c (5 × 10−5 M) on reduction by PhSMe (5 × 10−4 M). Time between spectra recordings was 1 s. Inset shows absorbance variation at 375 nm (○) and the amount of PhMeSO formed (●) measured by HPLC versus time under the same conditions. The solid line was calculated using kobs of 0.084 s−1. Conditions: −40 °C, MeCN/H2O (0.2% v/v). Adapted from ref 37. Copyright 2011 American Chemical Society.

p‐XC6H4SMe + Fe V O → p‐XC6H4SOMe + Fe IIIOH 2

ironoxo species by any substrate matches the kinetics of the product formation and that the ironoxo complex is not reduced by an aggressive impurity in a substrate which is used in excess for ensuring pseudo-first-order conditions. The C6H5SOMe sulfoxide was produced quantitatively from C6H5SMe and FeVO. The oxidation of C6H5SMe by FeVO in the presence of H218O gave 25% incorporation of 18O in the product C6H5SOMe.31 Dependencies of kobs on concentrations of p-XC6H4SMe for reaction 3 are hyperbolic for electron-rich and linear for electronpoor thioanisoles, though the saturation is observed at significantly higher concentrations of p-NCC6H4SMe. This behavioral diversity is consistent with a common rate law shown as eq 4.

(3)

Using iron(V) species, which have been generated from the corresponding iron(III) resting states, sets up an important rule when the reactivity is investigated in single turnover experiments and the reaction progress is followed by monitoring the consumption of iron(V) starting material. The reduction of iron(V) may occur via the intermediacy of iron(IV) species which may be reactive as well. Therefore, the reactivity of highvalent metal species, which are by two oxidation equivalents above the resting state, should be studied by following the protocol previously elaborated for the enzymatic transformations by the peroxidase compounds I and II, which are also by two and one oxidation equivalents above the iron(III) resting state, respectively.53 The reactivity of compound I has been monitored by UV−vis spectroscopy at the isosbestic point between compound II and the resting iron(III) state, and the reactivity of compound II should be monitored at the wavelength equating with the isosbestic point between compound I and iron(III).53 Correspondingly, the reactivity of TAML FeVO species toward thioanisoles in acetonitrile was assayed at the isosbestic points of the corresponding FeIIIOH2 and FeIVFeIV species (at 370 and 375 nm, in particular, for derivatives of 1a and 1c, respectively) which

kobs =

kII[p‐XC6H4SMe] 1 + K1[p‐XC6H4SMe]

(4)

Leveling off of kobs at high [p-XC6H4SMe] requires a fast reversible formation of an [FeVO, thioanisole] adduct, which may be or may be not on the reaction coordinate. Equation 4 corresponds to the nonproductive scenario which is portrayed in Scheme 5. The product sulfoxide C6H5SOMe did not retard kobs in the reaction of FeVO with C6H5SMe, showing that the 9147

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Scheme 5. Mechanism for the Interaction of the FeVO with Thioanisoles p-XC6H4SMe Consistent with eq 4a

Table 4. Selected Second Order Rate Constants for the Oxidation of PhSMe by Diverse Iron(V)oxo and Iron(IV)oxo Complexes Obtained in Acetonitrile as Solventa entry 1 2 3 4 5 6 7 8 9 10 11 12

a

The postulated nonproductive intermediate, within which the oxygen transfer is not feasible, is shown on the right.

sulfoxide formed does not bind to the iron(V)oxo oxidant under the reaction conditions, which agrees with the absence of product inhibition. The second-order rate constants kII for the oxidation of electron-rich thioanisoles by FeVO are particularly large and very sensitive to the electronic effects brought up by the X of pXC6H4SMe (Table 3). The Hammett plots show improved linearity for FeVO1a,c when the log(kII/kII°) values are plotted against σ+ constants instead of σ (cf. ρ+ = −2.15 with ρ = −3.0 for FeVO1c), suggesting that resonance effects are significant. The value of ρ+ of −2.1 for FeVO1a indicates similar electrophilicities of FeVO1a,c. The absolute ρ values are larger than those reported for the oxidation of thioanisoles to sulfoxides by reconstituted cytochrome P450 (ρ+ = −0.16),54 by (salen+•)iron(IV)oxo complexes (ρ = −0.65 to −1.54),55 by [FeIII(13-TMC)(OIPh)]3+ (ρ = −1.9),56 or by mononuclear non-heme iron(IV)oxo complexes (ρ = −0.9 to −1.6).57−61 Closer absolute values of ρ = −2.5 have been measured for [(TMC)FeIVO]2+, which is the least reactive toward C6H5SMe compared to other related iron(IV)oxo complexes,58 and for [(Pytacn)FeIVO]2+, which interacts with C6H5SMe in a fast, not rate-determining step.62 The highest values of ρ+ in the FeVO1a,c case suggest that the thioanisole X substituents are in strongest electronic communication with the reactive iron(V) center. A comparison of the rate constants for the oxidation of C6H5SMe by iron(IV)oxo and iron(V)oxo complexes under similar conditions (Table 4) leads to important conclusions. The data reported for FeIVO by different authors show reasonable agreement (cf. entries 4 and 8; see Chart 3 for ligand names). The reactivity of iron(V) species exceeds that of many iron(IV) complexes investigated so far. More reactive than FeVO TAMLs are the complex produced on the interaction between [FeIII(13TMC)(CF3SO3)2] with iodosylbenzene (PhIO) (entry 11 in Table 4)56 and [(Me3NTB)FeIVO]2+ (entry 12 in Table 4).63 The most reactive FeVO TAML complex was produced from 1b,39 which contains strongly electron withdrawing groups at both the head (NO2) and tail (F) parts of the macrocyclic ligand. TAML 1b has higher catalytic activity than 1a and 1c in

iron complex V

Fe O1a FeIVFeIV1a FeVO1b [(TPA)FeIVO]2+ [(TMC)FeIVO]2+ [(Bn-TPEN)FeIVO]2+ [(N4Py)FeIVO]2+ [(BMBPy2N)FeIVO]2+ [(Pytacn)FeIVO]2+b [(N4Py)FeIVO]2+ [FeIII(13-TMC)(OIPh)]3+ [(Me3NTB)FeIVO]2+

T/°C −40 −40 −40 −45 35 −20 0 −30 0 8 −60c −40

k/M−1 s−1

ref

450 0.06 1920 0.44 0.029 0.075 0.065 0.31 1.0 0.2 3.5 × 103 2.1 × 104

37 37 39 58 58 58 58 65 59 61 56 63

a See Chart 3 for ligands other than TAMLs. bFormulated as [LFeIVO(H2O)]2−. cAcetone/CF3CH2OH (3:1) as solvent in the presence of 1.2 equiv of HClO4.

oxidations by hydrogen peroxide in aqueous media.64 Its highest Lewis acidity obviously accounts for the highest rates in the oxidation by FeVO. Slow comproportionation of 1c and FeVO1c offered a unique opportunity for a mechanistic analysis of the reactivity of FeVO with thioanisoles regarding a long-established dispute about oxygen-transfer versus electron-transfer pathways in general66 and in the sulfoxide formation in particular.67,68 This mechanistic challenge is still rather difficult to resolve despite numerous efforts to establish clear and straightforward criteria for discrimination. Each case is almost certainly unique, and multiple researchers will likely address this mechanistic puzzle. For our TAML studies, the slow comproportionation implied that the concentrations of FeIIIOH2 and FeIVFeIV species could be measured during the course of the thioanisole oxidation when the two species are produced from FeVO via the two-electron (oxygen) transfer (OT) and one-electron transfer (ET) pathways, respectively. Correspondingly, the relative contributions from the two- and one-electron pathways could be established. In the FeVO1c/C6H5SMe system, the yields of iron(III) and iron(IV) products could be estimated by UV−vis spectroscopy.37 The rate constant kobs corresponds to the consumption of FeVO, and hence it is a sum of the parallel reductions to afford FeIII (OT) and/or FeIV (ET), i.e. kobs = kobsOT + kobsET. Correspondingly, the kobsOT/kobsET ratio equals [FeIII]/[FeIV].

Table 3. Rate and Equilibrium Constants for the Oxidation of p-XC6H4SMe by FeVO1 and Relative Contributions of Oxygen Transfer and Electron Transfer Pathways Defined as kobsOT/kobsET for FeVO1ca FeVO1c (R = Et) X MeO Me H Cl CN

KI/M−1 800 ± 100 210 ± 30

FeVO1a (R = Me)

kII/M−1 s−1 5000 ± 400 600 ± 80 190 ± 30 80 ± 2 4.4 ± 0.5

kobsOT/kobsETb

KI/M−1

c

1.9 ± 0.1 0.88 ± 0.07 0.57 ± 0.09 0.20 ± 0.01

500 ± 50 23 ± 4

kII/M−1 s−1 9000 ± 700c 3900 ± 250 450 ± 50 165 ± 5 13 ± 5 11.5 ± 0.5d

See text for details. Conditions: −40 °C in MeCN, H2O (0.2% v/v). Data are from ref 37. bCalculated at the time of 95% conversion of iron(V)oxo at [p-XC6H4SMe] = 5 × 10−4 M. cMeasured at [FeVO] = [p-MeOC6H4SMe] assuming second-order kinetics. dObtained from the slope of kobs versus [p-NCC6H4SMe]. a

9148

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Chart 3. Examples of Ligands Used for Complexing Various Forms of Iron in Biorelevant Model Systemsa

Figure 5. Hammett plots for the rate constants kIIOT (squares) and kIIET (circles) calculated from the data in Table 3 for the iron(V)oxo compound 1c showing a higher sensitivity to electronic effects in pXC6H4SMe of the two-electron oxygen transfer pathway (ρ+ = −2.9) compared to the one-electron transfer pathway (ρ+ = −1.9). Adapted with permission from ref 12. Copyright 2013 Elsevier.

oxygen−sulfur double bond. It should be mentioned that Park et al. have reached an opposite conclusion while investigating similar sulfide-to-sulfoxide oxidation by [(N4Py)FeIVO]2+ promoted by ScIII.69 By applying the Marcus electron transfer approach, the authors have suggested that electron-rich thioanisoles are oxidized via the electron transfer mechanism, whereas p-NCC6H4SMe reacts predominantly via the oxygen transfer mechanism.

a

TMC = 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane; 13TMC = 1,4,7,10-tetramethyl-1,4,7,10-tetraazacyclotridecane; N4Py = N,N-bis(2-pyridylmethyl)-N-bis(2-pyridyl)methylamine; TPA = tris(2-pyridylmethyl)amine; TQA = tris(2-quinolinylmethyl)amine; Me3NTB = tris((N-methylbenzimidazol-2-yl)methyl)amine; Pytacn = 1-(2′-pyridylmethyl)-4,7-dimethyl-1,4,7-triazacyclononane; PyNMe3 = 3,6,9-trimethyl-3,6,9-triaza-1(2,6)-pyridinacyclodecaphane; Bn-TPEN = N-benzyl-N,N′,N′-tris(2-pyridylmethyl)ethane-1,2-diamine; BMBPy2N = [N-bis(1-methyl-2-benzimidazolyl)methyl-N-(bis-2pyridylmethyl)amine].

5. OXIDATION OF HYDROCARBONS: C−H BOND CLEAVAGE VERSUS ELECTRON TRANSFER The year 2019 will mark the half-century anniversary of the first discovery of transition metal complexes cleaving the C−H bonds of methane in solution.70 The technological expectations from this finding were set high, though this very issue of Chemical Reviews indicates clearly that these expectations have not been yet fulfilled despite substantial academic and industrial interest in the topic commonly referred to as “C−H bond activation by transition metal complexes” or simply “C−H bond activation”.71 The 50-year time frame72,73 poses a provocative question concerning the probability of an eventual genuine success. Nevertheless, the flow of publications on “C−H bond activation” does not seem to lose any pace. Moreover, the field now encompasses a huge boost from the “biomimetic” world, the study and mimicking of biological C−H bond activations by metal-containing enzymes such as methane monooxygenase and cytochrome P450.74−81 Though C−H bond activations by heme and non-heme iron(IV)oxo complexes are being intensively discussed and reviewed,51,82−88 much less attention has been paid to the corresponding processes involving iron(V)oxo species.89,90 Theoretical studies performed by Geng et al. point to reactivity advantages of iron(V)oxo species toward hydrocarbons. It has been concluded that, in a distorted octahedral coordination environment, the intrinsic reactivities of ironoxo species toward C−H bonds should increase as the iron oxidation state increases from IV to V and then to VI. The reaction barrier monotonically decreases in this series, consistent with the stepwise enhanced electrophilicity of the metal center.91 This rule is not as obvious as it seems because it does not, in particular, hold for the corresponding ironnitrido complexes.91 The oxidation state of iron is not the only factor that affects the reactivity. Density functional theory (DFT) calculations performed for the

The kobsOT/kobsET ratio could not be determined for the 1a system because of the fast comproportionation, and therefore only 1c provided this unique chance. The [FeIII]/[FeIV] ratios were calculated from the absorbance at 722 nm where the molar absorptivities of FeV and FeIV are the same, but where FeIII does not absorb. When the conversion of iron(V)oxo is 95%, the absorbance at 722 nm is given by (0.05[Fe]tεV + [FeIV]εIV) such that εIV [Fe]t − A 722 [Fe III] = IV A 722 − 0.05εIV [Fe]t [Fe ]

The time at which A722 should be read was calculated from kobs for each sulfide at [p-XC6H4SMe] = 5 × 10−4 M after 95% conversion of FeVO. The [FeIII]/[FeIV] ratios, which equal kobsOT/kobsET (Table 3), indicate that (i) the one-electron (ET) pathway is favored for the electron-poor sulfides, (ii) the twoelectron (OT) pathway is favored for the electron-rich thioanisoles, and (iii) the two pathways respond slightly differently to the electronic effects. The latter statement is illustrated in Figure 5, where the Hammett approach has been applied to the rate constants kIIOT and kIIET calculated from kII and kobsOT/kobsET in Table 3. A larger slope for kIIOT may suggest that the oxygen transfer, at least in the TAML system, is realized then a highly electrophilic FeVO unit interacts with an electron-rich nucleophilic counterpart, organic sulfide in the focal study, creating a sufficient driving force for breaking the iron−oxygen bond and forming the 9149

DOI: 10.1021/acs.chemrev.7b00034 Chem. Rev. 2017, 117, 9140−9162

Chemical Reviews

Review

Table 5. Summary of the Oxidation Products of Ethylbenzene Obtained Using FeVO Species Generated from 1a and 5a in Acetonitrile under Variable Conditions FeIII

T/conditions

C6H5CHOHCH3/%

C6H5COCH3/%

C6H5CHCH2/%

ref

1a 1a 1a 5a

−40 °C/air 0 °C/air 0 °C/argon 25 °C/O2 excluded

83 40 38 30

16 80 0 22

not detected 7 12 not detected

97 96 96 42

iron(IV)oxo catalyzed methane C−H bond activation reactions show that equatorial ligands play a fundamental role in tuning the reactivity of iron(IV)oxo complexes.92 Larger effects could be anticipated in the case of small in size and strongly donating equatorial ligands. Note that the equatorial ligation in TAML activators engages four strongly donating, deprotonated anionic amide nitrogens. This feature plus easy achievable iron(V) derivatives 31 combined make TAML activators exciting candidates for studies of C−H bond activations. Inspired by the enzymatic systems,93 it has recently became fashionable to find evidence that the activity of ironoxo species toward hydrocarbon C−H bonds is dictated by the spin state of iron. High-spin S = 2 complexes are considered to be more aggressive compared to corresponding low-spin counterparts.63,94 It should however be mentioned that the recent study of Bae et al. revealed similar reactivity features of S = 1 and S = 2 structurally similar iron(IV)oxo complexes.95 There is no high versus low spin dilemma for iron(V)oxo complexes derived from TAML activators, which are S = 1/2 species, and their S = 3/2 state is practically unattainable due to a large gap in the lowest energy, doubly occupied dxy and HOMO dxz orbitals of 18 500 cm−1.31 Our major inspiration for probing the ability of FeVO TAMLs to cleave C−H bonds of hydrocarbons stemmed from their high reactivity in the oxidation of thioanisoles and persistent micropollutants of water (see previous sections). Coincidentally, the C−H bond activation topic for TAMLs attracted the attention of two other research groups and the results obtained by all three scientific teams were published within a one year period.42,96,97 Therefore, the reader will find below a symbiotic “C−H FeVO TAML story” which is based on the experimental data from all three, rather independent sources.42,96,97

FeVO5a combined yields of products did not exceed 50%.42 The authors accounted for this result in terms of Scheme 6 assuming

5.1. Products of Hydrocarbon Oxidations

that cyclohexane gave solely bromocyclohexane in the presence of CCl3Br.96 When ethylbenzene was oxidized in the presence of H218O (0.2% v/v), no measurable incorporation of 18O was observed by Kundu et al. in either of the products, viz. 1phenylethanol or acetophenone,37 despite the fact that FeVO is known to undergo a rapid oxygen exchange with traces of water.31 When Kwon et al. used H218O, the product 1phenylethanol contained 15% 18O incorporation96 which may be a consequence of different temperatures employed (viz., 0 °C versus −40 °C). The lack of 18O in products generated in the presence of H218O observed by Kundu et al.97 suggests rapid couplings of O2 with alkyl radicals produced upon H atom abstraction by FeVO1a, and indeed both Ghosh et al.42 and Kwon et al.96 found differing product distributions when O2 was eliminated. It is very likely that using excess of mCPBA mentioned above brings an extra feature to the course of C−H bond activation. When FeVO was generated by the stoichiometric amount of mCPBA,42,97 the iron(V) was reduced into the mixture of FeIVFeIV and FeIIIOH2. In excess of mCPBA, the iron(IV) species were proven to be the monomeric complex [FeIV(mCBA)]−, i.e. with axially coordinated m-chlorobenzoate.96

Scheme 6. Stoichiometric Mechanism of Hydrocarbon Oxidation by FeVO5a Which Accounts for