Hydrogen Atom Transfer (HAT)-Mediated Remote Desaturation Enabled by Fe/Cr-H Cooperative Catalysis

A highly selective, remote desaturation reaction, catalyzed by an iron/chromium cooperative system, generates unsaturated amides from abundant starting materials. The reaction appears to operate via a triple hydrogen atom transfer (HAT) cascade: (i) A 1,5-HAT from a remote C – H bond to an iron nitrenoid; (ii) a chromium-hydride-mediated HAT to N -center, (iii) the abstraction of a second H• from the remote organic radical by the chromium radical Cr• . Unsaturated amides and conjugated dienamides can be prepared with good regio-and stereoselectivity. The synthetic utility of this method is verified by late-stage desaturation of natural product derivatives and biologically important molecules.

of these radicals into carbon-carbon double bonds has been hampered by a lack of catalysts for that step.To address this challenge, we have turned our attention to the use of transition-metal radicals M• for the second H• abstraction (Fig. 1c, right). 26C(sp 3 )-H bonds that adjoin radicals are relatively weak, around 35 kcal/mol, 27 whereas metal-hydrogen bonds (M-H) range from 55 to 83 kcal/mol 26 -making H• abstraction by metal radicals M• thermodynamically favorable.
9][30][31][32] Our previous report on the epoxide hydrogenation has proven the feasibility of H• abstraction by Cr• from C(sp 3 )-H bonds  to radicals. 33r design 1,4,2-Dioxazol-5-ones A, easily made from carboxylic acids, have been used with Ru II porphyrins and copper salts in the synthesis of oxazoles and oxazolines. 34A Ru nitrenoid complex appears to be involved in the Ru/Cu reaction, which suggests that an iron nitrenoid complex can be generated from Fe II by the decarboxylation of S1. 23 (Results that use an Fe catalyst for that purpose are being reported separately by the Li group.)In place of a dioxazolone we can use an N-acyloxy amide like B, C, or D (Fig. 1d); 35 Arnold has already shown that Fe-acyl-nitrenoid intermediates are generated when acid is eliminated from such esters in the presence of Fe II .The nitrenoids generated in these ways should be able to remove H• from remote C(sp 3 )-H bonds and thus to make high-value unsaturated amidesimportant synthetic building blocks.
Herein, we report a triple HAT cascade that uses iron and chromium cooperative catalysis to carry out such remote C(sp 3 )-H desaturations (Fig. 1e).An intramolecular 1,5-HAT onto the N of the iron-nitrenoid I releases the carbon-centered radical II; the N-atom in II accepts a hydrogen atom from the chromium hydride Cr-H to form intermediate III.The chromium radical Cr• now abstracts an  hydrogen atom from III via the thermodynamically favorable HAT process just described, generating the final product and regenerating the chromium hydride (Cr-H).The Li group is separately reporting a different cooperative catalyst using iron and cobalt instead of iron and chromiumfor this desaturation reaction.

Results and discussion
To test our proposal, we tried several iron and chromium catalysts (Table 1).We found that a combination of iron II acetate, H-CrCp(CO)3, and acetic acid in 1,4-dioxane as solvent gave an optimal yield (entry 1, 85%) of the desaturated product 1 of 3-(3-phenylpropyl)-1,4,2dioxazol-5-one S1.Likewise, N-acyloxy amides (B, C, D) worked well (entry 2); all three provided products in yields up to 85%.(A weak acid is not needed when B, C, or D is used.)Other iron catalysts (entries 3-5) proved less effective at desaturation; apparently direct decomposition of S1 results in the saturated formamide 1'.Manganese, ruthenium, and iridium catalysts (entry 6), used to generate nitrenoid complexes in previous research, [36][37][38] proved ineffective at replacing Fe(OAc)2 in our desaturation chemistry.Variation of the Cp in the chromium hydride cocatalyst (entry 7) revealed that steric hindrance was important, with the more congested ligands Cp* and Cp # leading to a decrease in yield (to 31% and 52%, respectively).The [CrCp(CO)3] -anion proved even less effective at desaturation (entry 8, 10%).Acetic acid was indispensable when dioxazolones were employed as substrates (when it was omitted, entry 9, or decreased, entry 10, much lower yields were obtained).Perhaps acetic acid accelerates the dissociation of product from the iron complex.Stronger acids (entry 11) were not tolerated, presumably because they decompose S1.No improvement in the yield was obtained with other solvents, including tetrahydrofuran or 1,2-dichloroethane (entries 12 and 13).A lower yield was obtained when the reaction was conducted at 60 o C (entry 14), presumably because substrate S1 decomposes at that temperature.The desaturation reaction proved so sensitive to air that no product was formed from S1 in its presence (entry 15).
Methodologies for the preparation of conjugated dienamides have been even less available 7 than methods for the preparation of but-3-enamides.As mentioned above, radical processes are often encountered in pharmaceutical literature for the preparation of high-value bioactive compounds.Our HAT-based desaturation reaction has proven applicable to these targets (17-43, Fig. 2B), which are versatile synthetic intermediates found in various natural products and their metabolites. 40The required substrates for dienamide synthesis (17-43) are readily obtained from aldehydes or ketones via Wittig reactions (See SI for more details).Compatible functional groups are the electron-donating ones in 18-22, the bromines in 23 and 24, the electron-withdrawing ones in 25-29, and the sterically hindered ones in 20 and 27.In general conjugated dienamides are obtained in good to excellent yields (55-88%) with good E,E selectivity.Particularly good yields have been obtained for 2,2-difluoro-1,3-dioxole (30) and furanyl (31)-substituted and 6,6-disubstituted (32-35) products, and with dienamides (38-41)  containing alkyl substituents in distal positions.Terminal dienes like 42 and 43 are available in moderate yields, and cyclic structures like 36 and 37 in lower yieldspresumably as a result of steric hindrance.
We have tested our desaturation reaction on non-activated aliphatic substrates (like S44-S50) ranging from 96 to 101 kcal/mol. 41Disappointingly, only a little γ,δ-desaturated amide 44 is formed, along with 56% of the byproduct 44' (Fig. 2C) when 30 mol% of catalysts were used.An additional methyl substituent, which makes the C-H bond in the substrate S45 tertiary, gives an appreciably better yield (70%) of the γ,δ-unsaturated product 45.Linear aliphatic substrates are compatible with our standard conditions; the primary C-H bonds in methyl groups are favored, giving mostly γ,δ-desaturated products (46 and 47) in good yields, rather than β,γ-desaturated products.Cycloalkyl substrates are selectively desaturated at their δ C-H bonds, giving γ,δ-desaturated derivatives (48, 49) in useful regioselectivities (up to 4.8:1).The yield of 50 is decreased by steric hindrance.The yields of 45 and 50 were much higher in 1,2-dichloroethane than in 1,4-dioxane.
A carbon-carbon double bond is a privileged functional group, easily transformed into many other functional groups. 42,43The late-stage desaturation of pharmaceutics has enormous applicability in drug discovery, as illustrated in the transformation of citronellal and lithocholic acid in Fig. 2D.Our iron/chromium cooperative catalysts easily converted the citronellal derivative S51 to 51 in 63% yield, although the reaction was accompanied by some formation of the isomeric dienamide 51' (20%).We have also been able to convert the dioxazolone S52, derived from lithocholic acid, to the unsaturated steroid derivative 52.

Mechanistic investigations
Radical mechanisms have been established by earlier work on the H• transfer chemistry of oxo and nitrene iron complexes. 46,47Evidence in the present case is offered by the configuration of the gray double bond in the diene 17 (Fig. 3a).An E configuration in substrate S17 gives 17 that retains the E configuration of that (gray) double bond (Fig. 3a-1), whereas S17' with a mixture of the E and Z configurations gives 17 in which the gray double bond has entirely an E configuration (Fig. 3a-2)surely the result of the Z/E isomerization of an allyl radical intermediate. 22,48ditional evidence is offered by the formation of double bond isomers 39', 40', 56', and 57' along with the expected dienamides 39 and 40 (Fig. 2B), 56, and 57 (Fig. 3a-3/4).The structures of 39' and 40' are shown in Fig. 2B, while those of 56' and 57' are shown in Fig. 3a; all four are surely formed via allyl radical intermediates.(The structure of 56' has been confirmed (details in SI) by X-Ray crystallography.)The dienamide originally expected, 56, does not isomerize to 56' under the reaction conditions (Fig. 3a-3).Kinetic isotope effect (KIE) experiments (Fig. 3b) have been conducted to explore the rate-determining steps in these desaturation reactions.An intramolecular competition experiment was performed on S1-2,3-d2, with each carbon bearing a deuterium; a KIE value of 2.3 was observed for C3, while a KIE of 1.5 was observed for C2.In contrast, no KIE was observed for an intermolecular competition between S1 and S1-3,3-d2.These results show that C-H bond cleavage is irreversible and occurs after the rate-determining step, which must be the formation of the iron nitrenoid species (Fig. 3b-2). 49nally, we wanted to explore the extent of H• transfer from Cr to the reactive radicals II formed from unweakened C-H bondsfor example, the radical formed from S44 in Fig. 3c.The use of CpCr(CO)3D and AcOD did not give any deuterium incorporation into 44' (no 44'-d was detected by HRMS or 1 H NMR), which means pathway a (Fig. 3c) can be ruled out for the generation of 44'.Presumably the D• is transferred instead to the nitrenoid N (pathway b) and exchanged during workup in the presence of the excess acid.The same resultlack of D incorporation into 44'confirms that that cleavage of C-H bonds is irreversible, which is consistent with our conclusion from the KIE experiments.
On the basis of these experiments and previous reports, 23 we propose the catalytic cycle in Fig. 4 for the remote C(sp 3 )-H desaturation of amides.Dioxazolones, easily prepared from carboxylic acids, can be decarboxylated by Fe II

Conclusions
We have demonstrated the remote β,γ-/ γ,δ-desaturation of amides by iron and chromium hydride cooperative catalysis.The necessary nitrenoid precursors (dioxazolone or Nacyloxy substituted amides) are easily prepared from commercially available carboxylic acids.The mechanism seems to involve a triple HAT process: (i) A 1,5-HAT from a remote C-H bond to an iron nitrenoid (C to N); (ii) a chromium-hydride-mediated HAT to N-center (Cr to N), (iii) the abstraction of a second H• by the chromium radical Cr• (C to Cr).Most tertiary C-H bonds, and activated secondary C-H bonds, are subject to the desaturation reaction.Because the substrates are readily available, the metals used as catalysts are cheap, and the conditions are mild, we hope that the reaction can be applied to site-selective desaturation in drug derivatives and other biologically relevant compounds.
to give Fe IV -nitrenoid intermediates like I. By intramolecular 1,5-HAT from the  carbon to the nitrenoid nitrogen, these intermediates generate carbon-centered radicals II, which have -C-H bonds of only 35 kcal/mol.(Another possibility is an intermolecular MHAT between Cr-H and the nitrenoid I, whichafter removal of the N from the Fewould explain the observed byproduct V.) The transfer of an H• from Cr to the N of II, releasing •CrCp(CO)3, gives intermediate III.Removal of the  hydrogen from III by CpCr(CO)3•, and protic cleavage of the ligand from iron complex, gives the unsaturated product IV, and the iron catalyst is regenerated.