March's Advanced Organic Chemistry: Reactions, Mechanisms, and Structure, 7th Edition (2013)

Chapter 5. Carbocations, Carbanions, Free Radicals, Carbenes, and Nitrenes

5.C. Free Radicals

5.C.i. Stability and Structure²¹¹

A free radical (usually just called a radical) may be defined as a species that contains one or more unpaired electrons. Note that this definition includes certain stable inorganic molecules (e.g., NO and NO₂), as well as many individual atoms (e.g., Na and Cl). As with carbocations and carbanions, simple alkyl radicals are very reactive and are usually transient species. For the most part, their lifetimes are extremely short in solution, but they can be kept frozen for relatively long periods of time within the crystal lattices of other molecules.²¹² There are, however, many stable radicals,²¹³ some of which will be noted below. Many spectral²¹⁴ measurements have been made on radicals trapped in this manner. Even under these conditions the methyl radical decomposes with a half-life of 10–15 min in a methanol lattice at 77 K.²¹⁵ Since the lifetime of a radical depends not only on its inherent stability, but also on the conditions under which it is generated, the terms persistent and stable are usually used for the different senses. A stable radical is inherently stable; a persistent radical has a relatively long lifetime under the conditions at which it is generated, although it may not be very stable.

Radicals can be characterized by several techniques (e.g., mass spectrometry²¹⁶ or the characterization of alkoxycarbonyl radicals by Step–Scan Time-Resolved Infrared Spectroscopy).²¹⁷ Another technique makes use of the magnetic moment that is associated with the spin of an electron, which can be expressed by a quantum number of or . According to the Pauli principle, any two electrons occupying the same orbital must have opposite spins, so the total magnetic moment is zero for any species in which all the electrons are paired. In radicals, however, one or more electrons are unpaired, so there is a net magnetic moment and the species is paramagnetic. Radicals can therefore be detected by magnetic-susceptibility measurements, but for this technique a relatively high concentration of radicals is required.

A much more important technique is electron spin resonance (ESR), also called electron paramagnetic resonance (EPR).²¹⁸ The principle of ESR is similar to that of NMR, except that electron spin is involved rather than nuclear spin. The two electron spin states ( and ) are ordinarily of equal energy, but in a magnetic field the energies are different. As in NMR, a strong external field is applied and electrons are caused to flip from the lower state to the higher by the application of an appropriate radio frequency signal. Inasmuch as two electrons paired in one orbital must have opposite spins that cancel, an ESR spectrum arises only from species that have one or more unpaired electrons (i.e., free radicals).

Since only free radicals give an ESR spectrum, the method can be used to detect the presence of radicals and to determine their concentration.²¹⁹ Furthermore, information concerning the electron distribution (and hence the structure) of free radicals can be obtained from the splitting pattern of the ESR spectrum (ESR peaks are split by nearby protons).²²⁰ Fortunately (for the existence of most free radicals is very short), it is not necessary for a radical to be persistent for an ESR spectrum to be obtained. Electron spin resonance spectra have been observed for radicals with lifetimes considerably <1 s. Failure to observe an ESR spectrum does not prove that radicals are not involved, since the concentration may be too low for direct observation. In such cases, the spin-trapping technique can be used.²²¹ In this technique, a compound is added that is able to combine with very reactive radicals to produce more persistent radicals; the new radicals can be observed by ESR. Azulenyl nitrones have been developed as chromotropic spin-trapping agents.²²² An important class of spin-trapping compounds are nitroso compounds, which react with radicals to give stable nitroxide radicals:²²³ RN=O + R′√ → RR′N–O√. An N-oxide spin trap has been developed [37; 2(diethylphosphino)-5,5-dimethyl-1-pyrroline-N-oxide], and upon trapping a reactive free radical, ³¹P NMR can be used to identify it.²²⁴ This technique is effective, and short-lived species (e.g., the oxiranylmethyl radical) have been detected by spin trapping.²²⁵ Other molecules have been used to probe the intermediacy of radicals via SET processes. They are called SET probes.²²⁶

Because there is an equal probability that a given unpaired electron will have a quantum number of or , radicals are observed as a single line in an ESR spectrum unless they interact with other electronic or nuclear spins or possess magnetic anisotropy, in which case two or more lines may appear in the spectrum.²²⁷

Another magnetic technique for the detection of free radicals uses an ordinary NMR instrument. It was discovered²²⁸ that if an NMR spectrum is taken during the course of a reaction, certain signals might be enhanced, either in a positive or negative direction; others may be reduced. When this type of behavior, called chemically induced dynamic nuclear polarization²²⁹ (CIDNP), is found in the NMR spectrum of the product of a reaction, it means that at least a portion of that product was formed via the intermediacy of a free radical.²³⁰ For example, the question was raised whether radicals were intermediates in the exchange reaction between ethyl iodide and ethyllithium (Reaction 12-39):

Curve a in Fig. 5.1²³¹ shows an NMR spectrum taken during the course of the reaction. Curve b is a reference spectrum of ethyl iodide (CH₃ protons at δ = 1.85; CH₂ protons at δ = 3.2). Note that in curve a some of the ethyl iodide signals are enhanced; others go below the base line (negative enhancement; also called emission). Thus the ethyl iodide formed in the exchange shows CIDNP and so was formed via a free radical intermediate. Chemically induced dynamic nuclear polarization results when protons in a reacting molecule become dynamically coupled to an unpaired electron while traversing the path from reactants to products. Although the presence of CIDNP almost always means that a free radical is involved,²³² its absence does not prove that a free radical intermediate is necessarily absent, since reactions involving free radical intermediates can also take place without observable CIDNP. Also, the presence of CIDNP does not prove that all of a product was formed via a free radical intermediate, only that some of it was. Note that dynamic nuclear polarization (DNP) enhances signal intensities in the NMR spectra of solids and liquids. In a contemporary DNP experiment, a diamagnetic sample is doped with a paramagnet and the large polarization of the electron spins is transferred to the nuclei via microwave irradiation of the EPR spectrum.²³³ Dynamic nuclear polarization has been used to examine biradicals.²³⁴

Fig. 5.1²³¹ (a) The NMR spectrum taken during reaction between EtI and EtLi in benzene (the region between 0.5 and 3.5 δ was scanned with an amplitude twice that of the remainder of the spectrum). The signals at 1.0–1.6 δ are due to butane, some of which is also formed in the reaction. (b) Reference spectrum of EtI. [Reprinted with permission from Ward, H.R.; Lawler, R.G.; Cooper, R.A. J. Am. Chem. Soc. 1969, 91, 746. Copyright © 1969 American Chemical Society.]

As with carbocations, the stability order of free radicals is tertiary > secondary > primary, explainable by field effects and hyperconjugation, analogous to that in carbocations (Sec. 5.A.ii).²³⁵

With resonance possibilities, the stability of free radicals increases;²³⁶ some can be kept indefinitely.²³⁷ Benzylic and allylic²³⁸ radicals for which canonical forms can be drawn similar to those shown for the corresponding cations (Sec. 5.A.ii) and anions (Sec. 5.B.i, category 1) are more stable than simple alkyl radicals, but still have only a transient existence under ordinary conditions. Note that 2-phenylethyl radicals have been shown to exhibit bridging of the phenyl group.²³⁹

The triphenylmethyl and similar radicals²⁴⁰ are stable enough to exist in solution at room temperature, although they are in equilibrium with a dimeric form. The concentration of triphenylmethyl radical in benzene solution is ~2% at room temperature. For many years, it was assumed that Ph₃C√, the first stable free radical known,²⁴¹ dimerized to hexaphenylethane (Ph₃C–CPh₃),²⁴² but UV and NMR investigations have shown that the true structure is 38.²⁴³Although triphenylmethyl-type radicals are stabilized by resonance:

steric hindrance to dimerization and not resonance is the major cause of their stability.²⁴⁴ This was demonstrated by the preparation of the radicals 39 and 40.²⁴⁵ These radicals are electronically very similar, but 39, being planar, has much less steric hindrance to dimerization than Ph₃C√, while 40, with six groups in ortho positions, has much more. On the other hand, the planarity of 35 means that it has a maximum amount of

resonance stabilization, while 40 must have much less, since its degree of planarity should be even less than Ph₃C√, which itself is propeller shaped and not planar. Thus if resonance is the chief cause of the stability of Ph₃C√, 40should dimerize and 39 should not, but if steric hindrance is the major cause, the reverse should happen. It was found²³³ that 40 gave no evidence of dimerization, even in the solid state, while 39 existed primarily in the dimeric form, which is dissociated to only a small extent in solution.²⁴⁶ This result indicates that steric hindrance to dimerization is the major cause for the stability of triarylmethyl radicals. A similar conclusion was reached in the case of (NC)₃C√, which dimerizes readily although it is considerably stabilized by resonance.²⁴⁷ Nevertheless, that resonance is still an important contributing factor to the stability of radicals is shown by the facts that (1) the radical t-Bu(Ph)₂C√ dimerizes more than Ph₃C√, while p-PhCOC₆H₄(Ph₂)C√ dimerizes less.²⁴⁸ The latter has more canonical forms than Ph₃C√, but steric hindrance should be about the same (for attack at one of the two rings). (2) A number of radicals (p-XC₆H₄)₃C√, with X = F, Cl, O₂N, CN, and so on, do not dimerize, but are kinetically stable.²⁴⁹ Completely chlorinated triarylmethyl radicals are more stable than the unsubstituted kind, probably for steric reasons, and many are quite inert in solution and in the solid state.²⁵⁰

Allylic radical are relatively stable, and the pentadienyl radical is particularly stable, but (E,E)-, (E,Z)-, and (Z,Z)-stereoisomers can form. It has been calculated that the (Z,Z)-pentadienyl radical is 5.6 kcal mol⁻¹ less stable than the (E,E)-pentadienyl radical.²⁵¹ Note that vinyl radicals have (E)- and (Z)-forms and the inversion barrier from one to the other increases as the electronegativity of substituents increase.²⁵² Conjugated propargylic radicals are calculated to have diminished stability as the conjugation increases, in contrast to the behavior of alkenes.²⁵³ Cyclopropyl alkynes have been used as mechanistic probes to distinguish between vinyl radicals and ionic intermediates.²⁵⁴ Enolate radicals are also known.²⁵⁵

It has been postulated that the stability of free radicals is enhanced by the presence at the radical center of both an electron-donating and an electron-withdrawing group.²⁵⁶ This finding is called the push–pull or captodative effect(see also, Sec. 4.K.i). The effect arises from increased resonance, as in 41.

There is some evidence in favor²⁵⁷ of the captodative effect, some of it from ESR studies.²⁵⁸ However, there is also experimental²⁵⁹ and theoretical²⁶⁰ evidence against it. There is evidence that while FCH₂√ and F₂CH√ are more stable than CH₃√, the radical CF₃√ is less stable; that is, the presence of the third F destabilizes the radical.²⁶¹

Certain radicals with unpaired electron on a carbon are also very stable.²⁶² Radicals can be stabilized by intramolecular hydrogen bonding.²⁶³ Diphenylpicrylhydrazyl (42) is a solid that can be kept for years, and stable neutral azine radicals have been prepared.²⁶⁴ Nitroxide radicals were mentioned previously,²⁶⁵ and the commercially available TEMPO (2,2,6,6-tetramethylpiperidine-1-oxyl) free radical (43) is a stable nitroxyl radical used in chemical reactions (e.g., oxidations),²⁶⁶ or as a spin trap.²⁶⁷ Nitroxyl radical 44 is a nitroxide radical so stable that reactions can be performed on it (e.g., the Grignard reaction shown with 44; see Reaction 16-24) without affecting the unpaired electron²⁶⁸ (the same is true for some of the chlorinated triarylmethyl radicals mentioned above²⁶⁹). Several nitrogen-containing groups are known to stabilize radicals, and the most effective radical stabilization is via spin delocalization.²⁷⁰ A number of persistent N-tert-butoxy-1-aminopyrenyl radicals (e.g., 45) have been isolated as monomeric radical crystals (see 46, the X-ray crystal

structure of 45),²⁷¹ and monomeric N-alkoxyarylaminyls have been isolated.²⁷² α-Trichloromethylbenzyl(tert-butyl)aminoxyl (47) is extremely stable.²⁷³ In aqueous media it is stable for > 30 days, and in solution in an aromatic hydrocarbon solvent it has survived for >90 days.²⁷³ Although the stable nitroxide radicals have the α-carbon blocked to prevent radical formation there, stable nitroxide radicals are also known with hydrogen at the α-carbon,²⁷⁴ and long-lived vinyl nitroxide radicals are known.²⁷⁵ A stable organic radical lacking resonance stabilization has been prepared (48), and its X-ray crystal structure was obtained.²⁷⁶ Dissociation energies (D values) of R–H bonds provide a measure of the relative inherent stability of free radicals R.²⁷⁷ Table 5.4 lists such values. ^278279280281 The higher the D value, the less stable the radical. Bond-dissociation energies also have been reported for the C–H bond of alkenes and dienes²⁸² and for the C–H bond in radical precursors XYC–H, where X,Y can be H, alkyl, COOR, COR, SR, CN, NO₂, and so on.²⁸³ Bond dissociation energies for the C–O bond in hydroperoxide radicals (ROO√) have also been reported.²⁸⁴ However, note that basing radical stabilization energy on the difference between the bond dissociation energy (BDE) of CH₃–H, as a reference point, and of R–H has been observed to have shortcomings.²⁸⁵ The problem is that these values are only applicable to carbon-centered radicals, and the stabilization energies are not transferable and cannot be used to estimate BDE of R–R′, R–R, or any R–X compounds.²⁸²

Table 5.4 The D₂₉₈ Values for Some R–H Bonds^a,b

There are two possible structures for simple alkyl radicals.²⁸⁶ They might have sp² bonding, in which case the structure would be planar, with the odd electron in a p orbital, or the bonding might be sp³, which would make the structure pyramidal and place the odd electron in an sp³ orbital. The ESR spectra of √CH₃ and other simple alkyl radicals as well as other evidence indicate that these radicals have planar structures.²⁸⁷ This finding is in accord with the known loss of optical activity when a free radical is generated at a stereogenic carbon.²⁸⁸ In addition, electronic spectra of the CH₃ and CD₃ radicals (generated by flash photolysis) in the gas phase have definitely established that under these conditions the radicals are planar or near planar.²⁸⁹ The IR spectra of √CH₃ trapped in solid argon led to a similar conclusion.²⁹⁰ Despite the usual loss of optical activity noted above, asymmetric radicals can be prepared in some cases. For example, asymmetric nitroxide radicals are known.²⁹¹ An anomeric effect was observed in alkoxy radical 49, where the ratio of 49a/49b was 1:1.78.²⁹²

Evidence from studies on bridgehead compounds shows that although a planar configuration is more stable, pyramidal structures are not impossible. In contrast to the situation with carbocations, free radicals have often been generated at bridgeheads, although studies have shown that bridgehead free radicals are less rapidly formed than the corresponding open-chain radicals.²⁹³ In sum, the available evidence indicates that although simple alkyl free radicals prefer a planar, or near-planar shape, the energy difference between a planar and a pyramidal free radical is not great. However, free radicals in which the carbon is connected to atoms of high electronegativity (e.g., √CF₃), prefer a pyramidal shape;²⁹⁴ increasing the electronegativity increases the deviation from planarity.²⁹⁵ Cyclopropyl radicals are also pyramidal.²⁹⁶ Free radicals with resonance are definitely planar, although triphenylmethyl-type radicals are propeller shaped,²⁹⁷ like the analogous carbocations (Sec. 5.A.i). Radicals possessing simple alkyl substituents attached to the radical carbon (C√) that have C^sp3–C^sp3 bonds, and rotation about those bonds is possible. The internal rotation barrier for the tert-butyl radical (Me₃C√), for example, was estimated to be ~1.4 kcal mol⁻¹ 6 kJ mol⁻¹.²⁹⁸

A number of diradicals (also called biradicals) are known,²⁹⁹ and the thermodynamic stability of diradicals has been examined.³⁰⁰ Orbital phase theory has been applied to the development of a theoretical model of localized 1,3-diradicals, and used to predict the substitution effects on the spin preference and S–T gaps, and to design stable localized carbon-centered 1,3-diradicals.³⁰¹ When the unpaired electrons of a diradical are widely separated, for example, as in √CH₂CH₂CH₂CH₂√, the species behaves spectrally like two doublets. When they are close enough for interaction or can interact through an unsaturated system (as in trimethylenemethane),³⁰² they can have total spin numbers of +1, 0, or -1, since each electron could be either or . Spectroscopically they are called triplets,³⁰³ since each of the three possibilities is represented among the molecules and gives rise to its own spectral peak. In triplet molecules, the two unpaired electrons have the same spin. Not all diradicals have a triplet ground state. In 2,3-dimethylelecycohexane-1,4-diyl (50), the singlet and triplet states were found to be almost degenerate.³⁰⁴Diradicals (e.g., 51) are very stable with a triplet ground state.³⁰⁵ Diradicals are generally short-lived species. The lifetime of 52 was measured to be <0.1 ns and other diradicals were found to have lifetimes in the 4–316-ns range.³⁰⁶Diradical 53 [3,5-di-tert-butyl-3′-(N-tert-butyl-N-aminoxy)-4-oxybiphenyl] was found to have a lifetime of weeks even in the presence of oxygen, and survived brief heating in toluene up to ~60 °C.³⁰⁷ Radicals with both unpaired electrons on the same carbon are discussed under carbenes. 1,4-Biradicals are known, and α-carbonyl substituents increase the lifetime of the radical, and negative α-hyperconjugation (see Sec. 2.M) has been suggested as the cause.³⁰⁸

5.C.ii. The Generation and Fate of Free Radicals³⁰⁹

Free radicals are formed from molecules by breaking a bond so that each fragment keeps one electron.^310,311 The energy necessary to break the bond is supplied in one of two ways.

1. Thermal Cleavage. Subjection of any organic molecule to a high enough temperature in the gas phase results in the formation of free radicals. When the molecule contains bonds with D values or 20–40 kcal mol⁻¹ (80–170 kJ mol⁻¹), cleavage can be caused in the liquid phase. Two common examples are cleavage of diacyl peroxides to acyl radicals that decompose to alkyl radicals³¹² and cleavage of azo compounds to alkyl radicals³¹³

2. Photochemical Cleavage (see Sec. 7.A.v). Light energy of 600–300 nm is 48–96 kcal mol⁻¹ (200–400 kJ mol⁻¹), which is on the order of magnitude of covalent-bond energies. Typical examples are photochemical cleavage of alkyl halides in the presence of triethylamine,³¹⁴ alcohols in the presence of mercuric oxide and iodine,³¹⁵ alkyl 4-nitrobenzenesulfenates,³¹⁶ chlorine and of ketones:

Photolytic decomposition of N-hydroxypyridin-2-thione is a method that generates hydroxyl radicals.³¹⁷ The photochemistry of radicals and biradicals has been reviewed.³¹⁸

Radicals are also formed from other radicals, either by the reaction between a radical and a molecule (which must give another radical, since the total number of electrons is odd) or by cleavage of a radical³¹⁹ to give another radical, for example, the decomposition of benzoyl peroxide to give the benzoyl radical:

Radicals can also be formed by oxidation or reduction, including electrolytic methods.

Reactions of free radicals either give nonradical products (termination reactions) or lead to other radicals, which themselves must usually react further (propagation reactions). The most common termination reactions are simple coupling of similar or different radicals:

Another termination process is disproportionation:³²⁰

A propagation reaction is one in which a radical reacts to give at least one radical product, which continues the radical reaction sequence. There are four principal propagation reactions, of which the first two are most common:

1. Abstraction of Another Atom or Group, Usually a Hydrogen Atom (also see Chap 14).

A radical may abstract hydrogen atoms from a second molecule, or by an intramolecular process. A bromine radical (Br√) reacts with an alkane, for example, to give HBr and a carbon radical. This type of reaction is known as hydrogen-atom transfer. Water is an excellent hydrogen atom source for many reactions involving metals.³²¹ The reduction of a carbon radical with Bu₃SnH is an example of a hydrogen-transfer reaction (see Sec. 14.A.i). Indeed, carbon radicals react as hydrogen-bond acceptors.³²² Other atoms may be removed by a radical via atom-transfer reactions. A halogen atom can be transferred in some cases, including the transfer of an iodine atom from an aryl iodide to give an aryl radical.³²³ Solvent effects play a role in hydrogen-atom transfer (hydrogen abstraction), and hydrogen-bonding plays a role.³²⁴

2. Addition to a Multiple Bond (see Chap 15).

The radical formed from an alkene may add to the double bond of a second equivalent of alkene, and so on. This is one of the chief mechanisms for vinyl polymerization.

3. Decomposition. This process can be illustrated by the decomposition of the benzoxy radical (see 55).

4. Rearrangement.

This reaction is less common than rearrangement of carbocations, but it does occur (though not when R = alkyl or hydrogen; see Chapter 18). Perhaps the best-known rearrangement is that of cyclopropylcarbinyl radicals to a butenyl radical.³²⁵ The rate constant for this rapid ring opening has been measured in certain functionalized cyclopropylcarbinyl radicals by picosecond radical kinetics.³²⁶ Substituent effects on the kinetics of ring opening in substituted cyclopropylcarbinyl radicals have been studied.³²⁷ “The cyclopropylcarbinyl radical (56) has found an important application as a radical clock.³²⁸ Various radical processes can be clocked by the competition of direct reaction with the cyclopropylcarbinyl radical (k_t) and opening of that radical to the 1-buten-4-yl radical (k_r) followed by trapping. Relative rates (k_t/k_r) can be determined from yields of 4-X-1-butene and cyclopropylcarbinyl products as a function of the radical trap³²⁹ (X–Y) concentration. Absolute rate constants have been determined for a number of radicals with various radical traps by laser flash photolysis methods.³³⁰ From these absolute rate constants, reasonably accurate values of k_t can be estimated, and with the relative rate (k_t/k_r), a value for k_r can be calculated. From the calibrated radical-clock reaction rate (k_r), rates (k_t) of other competing reactions can be determined from relative rate data (k_t/k_r).”³²⁶ Other radical clocks are known.³³¹

Free radicals can also be oxidized to carbocations or reduced to carbanions.³³²

5.C.iii. Radical Ions³³³

Several types of radical anions are known with the unpaired electron or the charge or both on atoms other than carbon. Examples include semiquinones³³⁴ (57), acepentalenes (58),³³⁵ ketyls³³⁶ (59) and the radical anion of the isolable dialkylsilylene (60).³³⁷ Radical anions are formed by the reaction of carbene anions with chloromethanes.³³⁸ Reactions in which alkali metals are reducing agents often involve radical anion intermediates (Birch reduction, e.g., Reaction 15-13) that proceed via radical anion 61.

Several types of radical cation are also known.³³⁹ Typical examples include alkyl azulene cation radicals (62),³⁴⁰ trialkyl amine radical cations,³⁴¹ 1,2-bis(dialkylamino)benzenes radical cations (e.g., 63),³⁴² dimethylsulfonium cation radicals (Me₂S√⁺),³⁴³N-alkyl substituted imine cation radicals (Ph₂C=NEt√⁺),³⁴⁴ dibenzo[a,e]cyclooctene (64, a nonplanar cation radical),³⁴⁵ and [n.n]paracyclophane cation radicals.³⁴⁶ A twisted radical cation derived from bicyclo[2.2.2]oct-2-ene has been reported.³⁴⁷