A Facile Direct Route to N‐(Un)substituted Lactams by Cycloamination of Oxocarboxylic Acids without External Hydrogen

Abstract Lactams are privileged in bioactive natural products and pharmaceutical agents and widely featured in functional materials. This study presents a novel versatile approach to the direct synthesis of lactams from oxocarboxylic acids without catalyst or external hydrogen. The method involves the in situ release of formic acid from formamides induced by water to facilitate efficient cycloamination. Water also suppresses the formation of byproducts. This unconventional pathway is elucidated by a combination of model experiments and density functional theory calculations, whereby cyclic imines (5‐methyl‐3,4‐dihydro‐2‐pyrrolone and its tautomeric structures) are found to be favorable intermediates toward lactam formation, in contrast to the conventional approach encompassing cascade reductive amination and cyclization. This sustainable and simple protocol is broadly applicable for the efficient production of various N‐unsubstituted and N‐substituted lactams.


Introduction
Lactams are cyclic amides, which form ac lass of core structural motifs privileged in av ariety of bioactive natural products and pharmaceutical agents, [1] and feature widely in functional materials and catalysts. [2] Pyrrolidones are au biquitous subgroup of lactams, which are of great and longstandingi nterest in the preparation of five-membered nitrogen-containing heterocycles. [2a, 3] The regioselective establishment of CÀCa nd CÀN bonds is ap owerful strategy to build N-heterocyclic compounds. An attractive approachtosecondary lactams is catalytic cyclizationo fp rimary amides with intramolecular alkenes in the presence of metal catalysts via aza-Heck or aza-Wacker mechanisms, in spite of the need to use, respectively,h ighly electrophilic or nucleophilic nitrogen species. [4] Lactams can also be obtained in other ways, such as by combiningh omoenolates with acid-activated imines, [5] intramolecular hydrocarbamoylation of allylic formamides, [6] Michael addition-proton transfer-enol lactonization of acyl chlorides with dicarbonyls, [7] intramolecular alkenylation of acyclicb romoalkenes, [8] and cycloamination of prenyl carbamates and ureas. [9] All of these approaches require specific catalysts and/ora dditives, as wella s organic solvents, to afford satisfactory yields.T herefore, it would be desirable to develop ac atalyst-and solvent-free methodf or the construction of valuablel actam skeletons from sustainable and low-costsubstrates.
Oxocarboxylic acids contain one or more aldehydic or keto groups in ac arboxylic acid and are widely available compounds. Ap rime examplei sl evulinic acid (LA), whichc an be easily obtained from renewable lignocellulosic biomass. [10] The keto-acid LA can be convertedw ith primary amines into Nsubstituted lactams or pyrrolidones by cascade reductive amination and cyclization catalyzed by transition metals (Scheme 1). [11] In connection with this, various strategies have recently been proposed to design and prepare catalysts with uniquef unctionalities for simplifying the overall process, such as the use of low-pressure hydrogen, or hydrogen-donating co-reactants such as formic acid (HCOOH) and hydrosilanes under mild conditions. [11][12][13] Significant progress has been made in the selective production of N-substituted lactamsa nd pyrrolidones from LA (see the Supporting Information, Ta ble S1), but in all cases the use of an organic solvent, ane xternal hydrogen source, and/or additives are key to obtain good results.
Leuckart reductive amination is aw ell-known reaction for catalyst-free upgrading of aldehydes and ketones to formamides by using HCOOHa sahydrogen donor,a lthough it is Lactams are privileged in bioactive natural products and pharmaceutical agents and widely featured in functional materials. This study presents an ovel versatile approach to the direct synthesis of lactams from oxocarboxylic acids without catalyst or externalh ydrogen. The method involves the in situ release of formic acid from formamidesi nduced by water to facilitate efficient cycloamination.W ater also suppresses the formation of byproducts. This unconventional pathway is elucidatedb ya combination of model experimentsa nd density functional theory calculations, whereby cyclic imines( 5-methyl-3,4-dihydro-2-pyrrolone and its tautomeric structures) are found to be favorable intermediates toward lactam formation, in contrast to the conventionala pproach encompassing cascade reductive amination and cyclization. This sustainable and simple protocol is broadly applicable for the efficient productiono fv arious Nunsubstituteda nd N-substituted lactams. hampered by low selectivity and high required reaction temperatures (ca. 180 8C). [14] Efforts to improvet he product selectivity and reactionr ate include the use of basic additives (e.g., NEt 3 )o rmicrowaveh eatingt oa ctivateH COOHa nd minimize the negative effect of its acidity during reductivea mination. [15] Herein, we propose an ovel and generic strategy involving in situ-controlled release of HCOOHf rom N-formyls peciesw ith deionized water for cyclic diamination of LA and other ketoacids. Our synthesis protocol provides access to various N-(un)substituted lactams without catalysto re xternalh ydrogen and requires no organic solvent(Scheme 1).

Results and Discussion
We exploredt he use of formamide (H 2 NCHO) and HCOOHa s nitrogen and hydrogen sources, respectively,f or the cycloamination of LA (without catalyst or solvent) and obtained am oderate 5-methyl-2-pyrrolidinone (MPD) yield of 83 %a fter reaction for 1.5 ha t1 60 8C( Ta ble 1, entry 1). From ag reen chemistry point of view,t he tenfolde xcesso fH COOH resultsi nu ndesired waste with respect to process mass intensity (PMI = 8.7; Ta ble 1, entry 1) and E-factor (5.0;T able S2) parameters. Without HCOOH, the MPD yield decreased to 41 %a t1 60 8C (Table 1, entry 2), implying the possibilityo fi ns itu releaseo f HCOOHf rom H 2 NCHO or other N-formyl intermediates during the reaction. To explore this opportunity,s everalc ontrole xperiments were conducted. First, LA can be directly cyclized to form angelica lactones (ALs) by dehydration. [16] Second, H 2 NCHO can be hydrolyzed in water with conversions up to around7 0% after 1.5 ha t1 60 8Ca nd were then constant upon prolonged reactiont o4h. These findings demonstrate that water obtained by LA dehydration can in principle promote the hydrolysis of H 2 NCHO and relevant N-formyl intermediates to releaseH COOH. As expected, the addition of water (30 equiv) into the reactions ystemsw ith and withoutH COOH ( Table 1, entries 3a nd 4) significantly acceleratest he cycloamination reaction, providing comparable MPD yields of 92 and 94 %. It is worth notingt hat the reaction withoutH COOH exhibits superior green chemistry metrics, including ah igh reaction mass efficiency (RME),alow E-factor, and good atom economy compared to the case with HCOOH( Ta ble S2). The PMI of the presentp rotocol that avoidst he use of HCOOHi s 8.5 (Table 1, entry 4), which is much lower than the values be-Scheme1.Cycloamination strategies for the synthesis of N-(un)substituted lactams.  [5][6][7][8][9][10][11]. [11, 12d, 13c, 17] These features render our synthetic methodh ighly promising for the sustainable production of lactams in good yield. Ta ble 1s hows that systemsd epending on metal catalysts require much longer reaction times (up to 72 h) to achieve moderate yields of either substituted or unsubstitutedl actams (entries [8][9][10][11], despite relativelyl ow reaction temperatures (25-130 8C). [11] Our method showss uperior performance in terms of the reaction rate with and without HCOOH( 1227 and 470 mmol h À1 ;T able 1, entries 3a nd 4) while affordingg ood lactam yields that are comparable to previously reported results (entries3-11). The significantly higher MPD formation rate with HCOOHi ndicatest hat hydrogen transfer may be the rate-determining step. Therefore, we hypothesized that optimizing the reaction conditions towards faster in situ releaseo f HCOOHf rom H 2 NCHO and the involved N-formyl speciesa ssisted by water can improve the reactionp erformance without HCOOH. The use of 6equivalents of H 2 NCHO and3 0equivalents of water resulted in as atisfactory MPD yield of around 90 %f rom LA at 160 8Ca fter 2h (Figures S1 and S2). The reaction temperature and time strongly impact the MPD formation rate and product distribution ( Figure S3). Specifically,u nreduced 5-methyl-3,4-dihydro-2-pyrrolone (MDPY) and nitrogenfree g-valerolactone (GVL) were observed as byproducts in the novel reaction system ( Figures S4 and S5), alongside previously reported 4-formamidopentanoic acid( FPAC) and 4-formamidopentanamide (FPAM) intermediates. [11b, 12d, 18] These findings suggest that the reducing and CÀNc oupling functionalities of the reactions ystem are responsible for the product distribution, which lies between typical catalyst-mediated reductive amination and our novel approach free from catalysta nd external hydrogen.
As followsf rom Ta ble 1, the thermalr eaction system of LA and H 2 NCHO with HCOOH (1107 mmol h À1 )i nsteado fw ater (470 mmol h À1 )r esults in as uperior MPD formation rate (entry 1 vs. 4), highlighting the significance of hydrogen transfer in the overall cycloamination reaction. Although both types of reactions proceed via the key intermediates FPAC and FPAM, the reactionw ith excessH COOH affords another cyclic compound N-formyl 5-methyl-2-pyrrolidinone (FMP) as am ainb yproduct ( Figure 1). We also observe that the presence of HCOOHs ubstantially decreases the time to completely convert LA (20 min), in comparison with the water-assisteds ystem (120 min). Notably,t he formation of the relatively stable FMP results in al ow MPD yield (Figure 1). Water suppresses the formation of FMP by hydrolysis, thus improving the MPD yield ( Table 1, entry 1v s. 3). Thus, we can infer that water is not only favorable for in situ release of HCOOH by hydrolysis of H 2 NCHO and N-formyl intermediates, but also avoids the formation of undesired species like FMP that are chemically stable under anhydrous conditions.
The formation of GVL ( Figures 1B and S3), which can be obtained from ALs by hydrogenation with HCOOH, [19] suggests that ALs could be the intermediates leading to the N-heterocycles MDPY and MPD. To explore this possibility, a-AL was thermally treated with H 2 NCHO in water or neat conditions, where-by 44 %a nd < 6% MPD yields together with MDPY in roughly 8% and < 1% yields,r espectively,w ere obtained after heating at 160 8Cf or 1h.T hese differences can be attributedt ot he reversibility of the reaction between LA and ALs, [16] with the presence of water shiftingt he equilibrium to the LA side. LA is the speciest hat reacts to form MPD andM DPY.Akinetic study of the conversion of LA into MPD with normal and deuterated water reveals as econdary hydrogen isotope effect with k H /k D = 0.94 ( Figure S6). The equilibrium deuterium isotope effect suggests reversible processes in the overall conversion of LA into MPD, [20] especially those involved with water.
Considering the possible activating role of the acidic carboxyl group (-COOH)o fL Ai nt he synthesis of MPD by cycloamination, ethyl levulinate (EL) was used as as ubstrate instead of LA, affording MPD in around9 0% yield ( Figure S7).Akinetic study showedc omparable reactionr ate constant (k LA /k EL =  .05) for cycloaminationo fL A( 0.0519 min À1 )a nd EL (0.0493 min À1 ;F igure S8), demonstratingt hat the acidity of the substrate has am inor effect on the reaction outcome. Wea lso investigated the order of amination and amidation steps in the conversion of LA into MPD by examining the reactivity of 2pentanone and pentanoic acid under otherwise similarc onditions ( Figure S9). The resultsc learly show that amination is much faster than amidation independent of the presence of HCOOH. Another salient detail of this experiment was that the conversionso f2 -pentanone and pentanoic acid were both lowered in water in comparison with the HCOOH-mediated reaction ( Figure S9), further revealing that the rate-determining step in the cycloamination without externalh ydrogen source is likely the in situ release of HCOOHf rom H 2 NCHO and involved N-formyl speciesassisted by water.
The hydrogen transfer process and the presence of ALs and MDPY tautomeric structures were investigated by deuterium labeling experiments (Figures S10-S12). Thel arge body of reaction data allowedu st oi nfer the reaction pathways forL A cycloamination with H 2 NCHO and water ( Figure 2A). Initially, H 2 NCHO is reversibly hydrolyzed to NH 3 and HCOOHu nder hydrothermal conditions with ar eportedf ree energy of activation of around1 46 kJ mol À1 , [21] whereas LA is able to undergo either CÀNc oupling with H 2 NCHO to afford 4-(formylimino)pentanoic acid (IM-1) via IM-0 or intramoleculard ehydration to give ALs with tautomeric structures, respectively,o wing to insufficient HCOOH derived from H 2 NCHO hydrolysis in the early stage. With the assistance of the in situ-formed HCOOH, al imited amount of ALs is hydrogenatedt oG VL ( Figure 2A). As the dominantr eactionp athway,I M-1 is more prone to release HCOOHa nd undergo intramolecular cyclization, giving af avorable precursor MDPY with relevant tautomeric structures that are gradually transformed into MPD by hydrogen transfer duringp rolonged reaction. In parallel, IM-1 may undergo cascade transfer hydrogenation( FPAC), amidation (FPAM), and cyclization to afford the MPD product, despite the low rates of these reactions.
To examinew hether conversion of LA into MPD proceeding through MDPY is kineticallya nd thermodynamically favorable, density functional theory (DFT) calculations were performed at the B3LYP/6-311 + G(2s,2p) level in conjunction with polarizable continuumm odel (PCM) to simulate the solvent effect (Figures 2a nd S13). LA can react with H 2 NCHO to obtain intermediate IM-0 by overcoming an activation free energy barrier (G a )o f1 71 kJ mol À1 ,f ollowed by conversion to IM-1 through releasing one water molecule (G a = 90 kJ mol À1 ). From IM-1 to the product MPD via the precursor MDPY involving the release of HCOOH, the highest activation energy barrier is calculated to be 177 kJ mol À1 ,i ng ood agreement with the experimental activation energy of 162 kJ mol À1 ( Figure S14). In contrast, the pathways via the precursor FPAC and FPAM toward MPD require 32 kJ mol À1 and 63 kJ mol À1 higher activation energy,r espectively,c learly demonstrating that the pathway involving MDPY is preferred. Notably,F PAMf ormation requires ac omparable activation energyo f1 80 kJ mol À1 ,i ndicative of the possibility of affording FPAM from IM-1 via FPAC, as identified by GC-MS( Figure S4), which is also consistent with the experimental results ( Figures 1a nd S3). In view of gaseous NH 3 and HCOOHb eing simultaneously formed in ac ertain amount by thermalh ydrolysis of H 2 NCHO,t he initial amination of LA with NH 3 instead of H 2 NCHO wasa lso takeni nto consideration (Figure 2). The energy barrier for the reaction between LA and NH 3 to afford MDPY successively via IM-2 and IM-3 was calculated to be 155 kJ mol À1 ,w hich is slightly lower than the latter (177 kJ mol À1 ). It is worth noting that the used reactants LA (b.p. % 245 8C) and H 2 NCHO (b.p. % 210 8C) are both in the liquid phase during the reactionp rocess at at emperature of 160 8C. Therefore, we may expect that amination occursb etweenthesecomponents as well as between liquid LA and NH 3 (gas), considering mass transfer limitations of gas-liquidphase reactions. [22] In addition, AL formation hinders LA dehydration (Figure 2a nd S13), as proven by the poor GVL yield. These resultse laborate that cyclic imines (MDPY with its tautomeric structures) derived from LA by cycloamination are preferentially converted into the lactam by subsequent hydrogen transfer,w hich is totally different to the typical cascade reductive amination and cyclization process.
Encouraged by the prominent performance of the developed reaction system,t he substrate scope with respecttofunctional group tolerance wasfurther examined (Table 2). Apart from H 2 NCHO (Table 2, entry 1), ammonium formate (HCOONH 4 )w as found to exhibit much higherr eactivity (entry 2), leading to shorter reactiont ime (3 h) and higher MPD yield (97 %). In comparison with H 2 NCHO,t he increasing hydrolysisa bility of ammonium formate can contribute to the superior performance, also proving that the releaseo fH COOH from thermalt reatment of H 2 NCHO andi nvolved N-formyl speciesi st he ratedetermining step. However,L Ap romoted by a1 :1 mixture of aqueous HCOOH and NH 3 only gave MPD in al ow yield of around 70 %u nder identical conditions, whereas increased amounts of GVL and formylated/condensed coproducts were formed. Even worse, when primary amines (e.g.,p ropylamine,c yclohexylamine,a niline, 3-methylaniline, 4-methylaniline, and benzylamine) and HCOOH were used as substrates fort he cascade aqueous reactionp rocesses, only 10-50 %y ields of N-substituted lactams were obtained under optimized conditions. These results indicatet hat the in situ releaseo fH COOH from H 2 NCHO,H COONH 4 ,o ro ther N-formyls pecies assistedb yw ater is more favorable for suppressing the occurrence of side reactions to ensure relatively high selectivity toward MPD. When the 4-substituent of the oxocarboxylic acid substrate is changed from methyl to phenyl, 4-chlorophenyl, 4-fluorophenyl, and 2-thienyl groups, the corresponding unsubstituted lactamsa re obtained in good yields (81-93 %; Ta ble 2, entries3-6). Alongside the 1,4-dicarbonyl substrates, ar ange of 1,5-and 1,6-dicarbonyl compoundsc an be also subjected to cycloamination, affording moderate to good yields (68-92%)o fs ixand seven-membered lactams,r espectively (Table 2, entries 7-10). Importantly,u nsubstituted benzolactams (62-80 %) can be obtained from o-phenyl dicarbonyl compounds with H 2 NCHO assisted by water (Table 2, entries 11 and 12). In addition to H 2 NCHO,a series of other formamidest hat can be simply syn- thesized by catalyst-free N-formylation of amines with CO 2 [23] were also employed as both nitrogen and hydrogen sources, and relevant N-substituted lactamsc an be obtainedw ith satisfactory yields (75-98 %; Ta ble2,e ntries 13-22). To evaluatet he potential practical implementation of our new method, the efficiency of the catalyst-and external hydrogen-free system in the cycloamination of LA with H 2 NCHO and water was examined in ac ontinuous-flow microreactor. Under the optimized reaction conditions (LA/H 2 NCHO/H 2 Om olar ratio = 1:6:30, T = 160 8C), ag ood MPD yield of 87 %w as obtained witharesidence time of 20 min. In comparison with the established batch procedure (Table 1, entry 4), the flow reactor shortenst he reaction duration, with superior reactionr ate and comparable lactam yield.

Conclusions
In summary,af acile versatile approacht od irectly synthesize both N-unsubstituteda nd N-substituted lactamsu nder catalyst-and externalh ydrogen-free conditions has been developed involvingi ns itu release of HCOOHf rom formamidesp romoted by water for efficient cycloamination. Experimental and computational resultss how that cyclic imines are the key species toward the lactam by undergoing subsequent hydrogen transfer.T his route is completely different to typical catalytic reactionp athways, involving reductive amination followed by cyclization. The novel benign and versatile protocol exhibits good universality and applicabilityi nl actam synthesis. We envision it to find broader application by giving access to various nitrogen-containing compounds, especially high-value N-heterocycles.

Reaction procedures
All experiments were carried out in aT eflon-lined stainless steel autoclave (inner volume 15 mL), placed in an oil bath that was preheated to the desired reaction temperature (120-180 8C). In at ypical reaction procedure, LA or keto-acid (2 mmol), H 2 NCHO (12 mmol), or formamides (6 equiv), and deionized water (60 mmol, 30 equiv) were added into the autoclave and the reaction duration was recorded as the autoclave was placed into the oil bath. After as pecific reaction time, the autoclave was taken out of the oil bath and immediately cooled-down to ambient temperature with tap-water. Upon completion, the autoclave was opened, and deionized water (or methanol) was added to the reaction mixture to dilute it to 25 mL before it was subsequently analyzed by HPLC (or GC). Each experiment was separately conducted and repeated 2o r3times. The obtained conversions and yields are average data of 2o r3 individual experiments, with standard deviation (s)i n the range of 0.5-4.6 %. Structures were confirmed by 1 H and 13 CNMR spectroscopy (JEOL-ECX 500 NMR spectrometer,CDCl 3 ), GC-MS, and HRMS.
For product separation from the reaction mixture, CH 2 Cl 2 or ethyl acetate can be used as an effective extractant. Typically,d eionized water (3-5 mL) was added into the mixture after the reaction and the product lactam could be isolated by extraction with CH 2 Cl 2 or ethyl acetate (3 5mL). Solvent was removed from the resulting combined extract by evaporation under reduced pressure to give the product lactam.
For continuous-flow reactions, aL abtrix Start microreactor system (Chemtrix BV,N L) with ag lass micro reactor (type 3227, volume: 19.5 mL) was utilized. Initially,L A, H 2 NCHO, and deionized water in am olar ratio of 1:6:30 were evenly mixed and added into af lask. The resulting solution was pumped into the microreactor (rate: 25 mLmin À1 )u nder cooling until the reactor was full, whereupon the reactor temperature was raised to 160 8Ca nd the solution flow rate was set at 1.5 mLmin À1 .A fter running for 1.5 h, sampling at timed intervals was conducted for GC analysis.

Computational methods
All DFT calculations were carried out using the hybrid functional B3LYP [24] as implemented in Gaussian 09 D.01 software. [25] The allelectron 6-311 + G(d,p) basis set was used for all atoms. The polarized continuum model (PCM) [26] with standard parameters for water solvent (e = 78.3) was used to account for bulk solvent effects during geometry optimization and searching of transition states. Frequency analysis was performed to ensure that each transition state has only one imaginary frequency in the direction of the reaction coordinate. All relative energies discussed herein are referred to Gibbs free energies considered the zero-point energy (ZPE) correction at 453 K.