Reactions of Trimethylaluminium: Modelling the Chemical Degradation of Synthetic Lubricants

Abstract In investigating and seeking to mimic the reactivity of trimethylaluminium (TMA) with synthetic, ester‐based lubricating oils, the reaction of methyl propionate 1 was explored with 1, 2 and 3 equivalents of the organoaluminium reagent. Spectroscopic analysis points to the formation of the adduct 1(TMA) accompanied only by the low level 1:1 production of Me2AlOCEtMe2 2 and Me2AlOMe 3 when an equimolar amount of TMA is applied. The deployment of excess TMA favours reaction to give 2 and 3 over 1(TMA) adduct formation and spectroscopy reveals that in hydrocarbon solution substitution product 2 traps unreacted TMA to yield 2(TMA). The 1H NMR spectroscopic observation of two Al−Me signals not attributable to free TMA and in the ratio 1:4 suggests the formation of a previously only postulated, symmetrical metallacycle in Me4Al2(μ2‐Me)(μ2‐OCEtMe2). In the presence of 3, 2(TMA) undergoes thermally induced exchange to yield Me4Al2(μ2‐OMe)(μ2‐OCEtMe2) 4 and TMA. The reaction of methyl phenylacetate 5 with TMA allows isolation of the crystalline product Me2AlOCBnMe2(TMA) 6(TMA), which allows the first observation of the Me4Al2(μ2‐Me)(μ2‐OR) motif in the solid state. Distances of 2.133(3) Å (Al−Mebridging) and 1.951 Å (mean Al−Meterminal) are recorded. The abstraction of TMA from 6(TMA) by the introduction of Et2O has yielded 6, which exists as a dimer.


Introduction
Ideas on the atmosphericr eaction of chlorofluorocarbons (CFCs) have existed for more than 40 years [1] and are well documented. [2] Althoughl egislation has been implemented aimed at eliminating their use, [3] the effects of substitute refrigerants such as perfluorocarbons (PFCs) and hydrofluorocarbons (HFCs) have been the subject of subsequent scrutiny [4] and regulation. [5] More specifically,w ith emissions from (automobile) air conditioning units representing ag rowing climate control concern [6] action has been initiated [7] to avoid the use of refrigerants with ag lobal warming potential > 150 (GWP = 100 year warming potentialo fo ne kg of ag as relative to one kg CO 2 ). [8] This has had the effect of phasing out greenhouse gases such as R-134a( 1,1,1,2-tetrafluoroethane, GWP = > 1000). [8] However,i llicit HFC use remains ap roblem,w ith R-40 (chloromethane;G WP = 13) havingb een used as ac ounterfeit refrigerant. [9] This raisesi ssues of reactivity with aluminium components in refrigeration units. Although the reactiono fa lkyl chlorides with aluminium under the influence of an aluminium halidec atalyst is well established, [10] it is known that reaction also proceedsi nt he absenceo fc atalyst. [10b, 11] In this vein,i n our hands the autocatalyticf ormationo ft rialkylaluminium and (catalyst) AlCl 3 from an alkyl chloride-aluminium mixture has been initiatedb yh eat only. [12] The products of reaction be-tweenR -40 itself and aluminium include trimethylaluminium (TMA), whichi sp otentially reactivew ith respectt oo ther chemicals present.T hese include proprietary compound oils (e.g. RL 32H) [13] formulated for use in conjunction with HFC refrigerants. They comprise synthetic polyolesters (POEs) which, in RL 32H itself, have ap entaerythritol core. [14] Although the interaction of organoaluminium compounds with esters has been studied [15] the specifics of the mechanism remain surprisingly obscure and, in particular,r eaction intermediates are incompletelyu nderstood. AlEt 3 has been reacted with esters in an equimolar ratio to give 'ate complexes that rearrange to ketonesa nd aldehydes. [16] Studies using TMA have exploredt he formationo fd onor-acceptor complexes and their derivatization with excess aluminium reagent. [17] Moreover,t he deployment of excess TMA at high temperature has incurred doublem ethylation and tertiary alcohol formation. However, mechanistic insights were limited to the alternative use of Me 2 AlCl. [18] Although the formation of hemialkoxides has been postulated based on the derivatization of ketonesa nd aldehydesu sing alkylaluminiums, [19] these species have not hitherto been recordedinester-based systems. Meanwhile,t he use of excessT MA has been reported in the alkylation of acetates, [20] and the ketonization of heteroaromatic estersu sing 1equivalent of TMA has been reported. [21,22] Reaction selectivity has been investigated, with TMA used at low temperature [23] in the stereoselective reduction of cyclic ke-tones [24] to give neoliacinic acid. [25] The reactionw as done in the presence of ancillary ester groups,w ith competing transesterification provingc ontrollable. [26] The expected by-product of ester reaction with TMA, Me 2 AlOR, has been the subject of extensive study. [27] However, this has tended to focus not upon its synthesis as ab y-product of ketonization reactions but rather on the oxophilic derivatization of AlÀCbonds [28] by moisture [29] or oxygen. [30] From astructural point of view,a luminium organooxide formation [31] and di-/trimerization is welle stablished, [32] for example, the simple aluminium alkoxide Me 2 AlOMeh as been shown to be trimeric. [33,34] In this work we model the reaction of TMA with synthetic POEs and elucidate intermediates along the reaction pathway between TMA and esters in general for the first time. Structure and stability are monitored for intermediate complexes and solution data clarify the reaction stoichiometry.

Results and Discussion
The ability of alkylaluminiumc ompounds to be autocatalytically generated through the action of alkyl chlorides on aluminium metal has led us to seek to model the potential reactivity of lubricant oils used in industrial refrigeration units with respect to TMA. Reactions involving as imple aliphatic ester were undertaken whereby TMA in toluenew as initially added dropwise to methyl propionate 1 (1:1) under aN 2 atmosphere at À78 8C. Though this system failed to readily produce isolable products,the observation of ap ale-green colour upon heating, which disappeared when left to cool to room temperature, suggested the interaction of ester and TMA and led to further investigations. Accordingly,a ne xcesso fT MA (see the Supporting Information, FigureS1) was added to 1 (3:1 TMA:1)u nder N 2 at À78 8C. After reaching room temperature the solution was stirred for 2hours, whereupon the NMR spectra of an aliquot werec ollected. 1 HNMR spectroscopy and COSY suggested the formation of two species (Figure 1, top), with 13 CNMR spectroscopy confirming the complete absence not only of ester but of C=Og roups from each species( see Figure S3 in the Supporting Information). These data suggest that 2:1r eaction of TMA with ester has occurred, one equivalent of TMA expelling methoxide to induce the formationo fareactive EtMeC=Oi ntermediate alongside Me 2 AlOMe 3 (d H = 3.06, À0.59 ppm) [32] before as econd equivalent of TMA has reacted with the ketone to give the dimethylaluminium alkoxide Me 2 AlOCEtMe 2 2.I ntegrals of peaks at d H = 0.61 (2)a nd 3.06 ppm (3)s uggest the two products to be present in a1 :1 ratio. Lastly,t he observation that signals at d H = À0.47 and À0.59 ppm reveal relative integralso f2:1 leads us to speculate that 2 traps the final (unreacted) equivalent of TMA present to give Me 2 AlOCEtMe 2 (TMA) 2(TMA) (Scheme 1). Based on these spectroscopic data we attribute what would be an unusual 4membered Al 2 OC metallacycle to this adduct, in whichf our Me groups are equivalent, with the terminal groups (Me t )r esonat- ing at d H = À0.47 ppm while the unique bridging group (Me b ) resonates at d H = 0.09 ppm. This view is reinforced by 13 CNMR spectroscopy (Supporting Information, Figure S3), whichr eveals as harp signal at d C = À4.1 ppm due to the bridging methyl in 2(TMA) and broad signals for terminal AlMe groups at d C = À6.7 and À10.7 ppm in 2(TMA) and 3,r espectively. Lastly,i ti sc onsistent with 27 Al NMR spectroscopic evidence, which reveals ab road signal at d Al = 156.2 ppm (Supporting Information,F igure S3) attributable to 4-coordinate aluminium. [35] To further clarify the co-formationo fp utative 2(TMA) and 3, the same synthetic process was repeated using 2:1a nd 1:1 TMA:1 ratios (Figure 1, middle and bottom). The 1 HNMR spectrum of the last of these systems is dominated by the formation of ac omplex between 1 and TMA ( Figure 1, bottom), with signals from unreacted 1 at d H = 3.32 (s), 1.99 (q) and 0.93 (t) ppm (c.f. Supporting Information Figure S2) movedt od H = 2.98 (m), 1.97 (m) and0 .65 (m) ppm, whereas coordinated TMA is revealed downfield (d H = À0.33 ppm) of free TMA (d H = À0.35 ppm, Figure S1). 13 CNMR spectroscopy reveals retention of am odified ester in 1(TMA) (d C = 181.4 ppm, FigureS4; c.f. d C = 173.9 ppm in 1,F igure S2 in the Supporting Information) and the presence of coordinated TMA (d C = À7.7 ppm). However,c onsistentw ith previousr eports, [20,26] negligible conversion of the complex 1(TMA) into addition product is observed in this system.I nc ontrast, the 2:1T MA:1 system reveals not only the interaction of 1 with TMA to yield adduct 1(TMA) buta lso the development of two further species ( Figure 1, middle). Hence, signals attributable to 1(TMA) are noted in the 1 HNMR spectruma td H = 2.94, 1.95, 0.63 and À0.32 ppm and 13 CNMR spectroscopy reveals retention of the ester function at d C = 181.8 ppm (Supporting Information, Figure S5). However,t his adduct is now less populousi ns olution than two other distinct species. One of these presentss ignals at d H = 3.08 and À0.60 ppm and is attributable to evolvingM e 2 AlOMe 3,whereas the other is consistentw ith the alkoxide 2.T he high field region of the spectrum demonstrates 1 HNMR resonances at d H = 0.09 and À0.49 ppm (3 Ha nd 12 H, respectively), neither of which correspond to unreacted TMA. This reinforced the view already expressed (Figure 1, top) that an adduct, 2(TMA), exists in solution,t hat high field signals are attributable to one bridging (AlMe b )a nd four terminal (AlMe t )A lMe groups, respectively,i naMe 2 Al(m 2 -Me)AlMe 2 fragment and that 2(TMA) is symmetrical, metallacyclic Me 4 Al 2 (m 2 -Me)(m 2 -OCEtMe 2 ) (Scheme 1). In as imilar vein, 13 CNMR spectroscopy showed resonances at d C = À4.6 and À7.7ppm due to AlMe b and AlMe t ,r espectively,t he latter representing the superposition of signals attributable to both 1(TMA) and 2(TMA). Meanwhile, 3 was now clearly shown by the presence of as ignal at d C = À11.1 ppm. Both 2(TMA) and 3 were retained in the 3:1 system,w ith 1(TMA) now completely absenta nd as mall amount of unreacted TMA identified at d C = À0.36 ppm (Figure 1, top and Supporting Information Figure S3). 27 Al NMR spectroscopy evidenced the trend from 1(TMA) towards the formation of 2(TMA) and 3 by the gradual replacement of ad ominant signal at d Al = 185.0 ppm in the 1:1s ystem (carbonyl-bonded 4-coordinate Al) with as ignal at d Al = 156.2 ppm in the 3:1s ystem (alkoxide-bonded 4-coordinate Al;S upporting Information FiguresS3-S5).
Spectroscopy points to an Me 4 Al 2 (m 2 -Me)(m 2 -OCEtMe 2 )f ormulation based on as ymmetrical OAl 2 Cm etallacycle for 2(TMA). However, although this is similar to motifs previously proposed, [19] the thermal stabilityo fs uch am otif has not hitherto been reported. With this in mind, the reactionm ixture resulting from the introduction of TMA in toluenet o1 in a3:1 ratio (spectroscopically characterized as ostensibly a1 :1 mixture of 2(TMA) and 3,F igure 1, top) was heated to reflux for 4hours. NMR spectroscopic analysiso fa liquotso btained after t = 0, 1, 2, 3a nd 4hours revealed ag radualt hermalr earrangement (  1ðTMAÞþ2T MA ! 2ðTMAÞþ3 ð1Þ The use of 1:1 1:TMA withouth eatingr esultsi nv ery limited reaction, with only traces of 2(TMA) and 3 existing alongside (dominant) 1(TMA) (d H = 2.98 ppm, Figure 1, bottom). Even heatingf ails to completely consume 1 and insteada round 50 %u nreacted 1 can clearly be seen after 2hours (d H = 3.32 ppm, Figure 3, bottom). This is explained by viewing 1 as reactingw ith 3equivalents of TMA to yield 2(TMA)a nd 3, which then undergoes thermal exchange to give 4 + TMA. This latterly generated TMA can then react with remaining 1,e ventually converting half the available 1 into 4.I nt he 1:2 1:TMA system,t he greater amount of TMA present aids the formation of 2(TMA) +  Repeated attempts to isolate crystalline products of reaction between methyl propionate 1 and TMA proved fruitless on account of al ow melting point and led to the replacement of 1 with methyl phenylacetate 5 (Supporting Information, Figure S7) in an attempt to crystallographically verify the identities/structures of ester decomposition products. Hence, TMA in toluene was added dropwise to 5 (1:1, 2:1o r3:1 TMA:5). NMR spectroscopic analysiso ft he resulting mixture revealed similar behaviour to that noted for the methyl propionate system, with the formation of initial adduct 5(TMA) in the presence of 1equivalent of TMA followed by reaction to give 6(TMA) and 3 in the presence of more than 1equivalent of TMA (Scheme 1 and Supporting Information Figures S8-S10). As with 2(TMA), the capture of excess TMA by 6 could be inferred from the 1 HNMR spectroscopic observation of Al-bonded Me groups at high field (d H = À0.42 ppm (Me t )a nd d H = 0.13 ppm (Me b )) in a4 :1 ratio alongside retentiono ft he singlet at d H = À0.59 ppm due to 3 (see above). 13 CNMR spectroscopy reinforced the co-presenceo f3 alongside 6(TMA) through the observation of ab road high field resonance at d C = À11.1 ppm (3)a longside signals at d C = À4.5 (6(TMA), Me b )a nd À7.0 ppm (6(TMA), Me t ). For the 3:1T MA:5 combination, the liquid remaining after reaction was reduced in volume and stored at 4 8Cf or 1day to produce colourless crystals that analyzed as am ixture of Me 2 AlOCBnMe 2 (TMA) (6(TMA);B n= CH 2 Ph) and 3. It was now possible to confirm the identity of 6(TMA) as Me 4 Al 2 (m 2 -Me)(m 2 -OCBnMe 2 ), with X-ray diffraction establishing the symmetry of the Al 2 OC ring formed by the capture of TMA and the presence of the expectedt erminal (Me t )a nd m 2 -bridging (Me b )m ethyl groups ( Figure 4). The result is the observation of two distinct classes of AlÀMe interaction;A l ÀMe b 2.133(3) ,A l ÀMe t 1.951 (mean). Theo nly previous report of which we are aware of diffraction data for the symmetrical metallacyclic motif reported herein lies with the electron diffractiona nalysiso ft he hemialkoxide Me 2 AlOtBu(TMA) in the gas phase (AlÀMe b 2.103(10) ,A l ÀMe t 1.948(7) (mean)). [36,37] In the solid state the Al 2 OC motif haso nly very rarely been recorded, with as earch of the Cambridge Crystallographic Database returning justs even results. Of these, only five show trapped TMA, demonstratingt he highly unusualn ature of this phenomenon. The nearest analogues of 6(TMA) are based on asymmetric bis(oxyphenyl) structures of type I ( Figure 5) demonstrated by tetraaluminium bis(bis(oxyphenyl)methyl)anthracene and -dibenzofuran complexes [38] and the bisaluminiumd erivativeo fa1,1'-bis-2,2'oxynaphthyl ligand. [39] At ype I motif has been recorded once also in heterobimetallic Al-Tichemistry. [40] The repeated observation that both 2(TMA) and 6(TMA) form alongside Me 2 AlOMe 3 and that crystalline 6(TMA) is isolated contaminated by 3 led to attempts to separate the components.E fforts here took two forms. In one set of experiments, the solventm ixture was modifiedp ost-synthetically. Hence, the recrystallization of 6(TMA) (leading to the crystal structure shown in Figure 4) gave ac rystalline materialthat analyzed by 1 HNMR spectroscopy as an approximately1 :1 mixture of 6(TMA) and 3 (see Experimental Section, co-synthesisof 6(TMA) and 3,M ethod 1). This ratio accurately reflectedt hat of the two products generatedi nt he reaction, which point was simply evidencedb ya nalysing an aliquot of the reactionm ixture (see Experimental Section, spectroscopic characterization of 5 + 3T MA). In contrast, the introductiono fh exane prior to recrystallization vastly improved the purity with which crystalline 6(TMA) could be isolated (10:1 6(TMA):3 by 1 HNMR spectroscopy; see Experimental Section, co-synthesis of 6(TMA) and 3,M ethod 2). An alternative approachi nvolved attempting to solvate one component of the 3/6(TMA) mixtureu sing aL ewis base. With this in mind, TMA in toluene was added to methyl phenylacetate in a3 :1 ratio under N 2 at À78 8C. Removal of toluene was followed by the additiono fe xcess Et 2 O. This resulted in the precipitationo fawhite solid, which wasrecrystallised by heating to give as olution and then storing at room temperature to produce colourless blocks. 1 HNMR spectroscopic analysis suggested the presence of Ph but not of Et 2 O and high field signals previously attributed to TMA were absent.T hese data suggest the abstraction of TMA as an ether solvate, [41] leadingt ot he crystallization of 6.T his was confirmed crystallographically by the observation of as imple dimer based on an (AlO) 2 core of atypecommon in aluminium organooxide chemistry (Figure 6). [27b, 28a]

Conclusion
In summary,t he autocatalytic nature of the reaction between alkylchlorides and aluminium to give AlCl 3 and TMA has led us to study the reactions of TMA with model esters that mimic  synthetic lubricants of the type used in industrial refrigeration units. Reaction has been found to be heavily dependent on stoichiometry.H ence, the treatment of either methylp ropionate or methyl phenylacetate with 1equivalent of TMA gave predominantly the corresponding ester-TMA adducts 1(TMA) or 5(TMA) with only nominal reaction occurring to give a1 :1 mixture of Me 4 Al 2 (m 2 -Me)(m 2 -OCRMe 2 )( R = Et 2(TMA) or Bn 6(TMA)) and Me 2 AlOMe 3.I ne ither case reaction was encouraged by adding more TMA, with full conversion occurring for a3:1 TMA:ester ratio. Spectroscopy clarified the trapping of 2 and 6 by TMA, suggesting the structures of the resulting adducts to be based on symmetrical OAl 2 Cm etallacycles. For 2(TMA), the presence of concurrently formed 3 induced thermal exchange to yield am ore stable metallacycle in Me 4 Al 2 (m 2 -OMe)(m 2 -OCEtMe 2 ) 4.T he ability to isolate 6(TMA) from am ixture of reactionp roducts provedh ighly solventd ependent, with by-product Me 2 AlOMe 3 largely retained in hexane solution whereas 6(TMA) crystallized, allowing confirmation of the rare OAl 2 Ch eterocycle.E fforts are now underway to extend this study to the use of more complex diesters and pentaerythritol-based esters, the latter representing ac lose analogueo f bona fide POEs.

Experimental Section General synthetic and analytical details
Reactions and manipulations were carried out under dry nitrogen, using double manifold and glove-box methods. Solvents were distilled off sodium-potassium amalgam (Et 2 O, hexane) immediately before use. Methyl propionate (99 %) and methyl phenylacetate (> 99 %) were purchased from Sigma-Aldrich and stored over molecular sieve (4 ). TMA (2.0 m in toluene) was purchased from Sigma-Aldrich and used as received. Elemental analysis was carried out on aP erkinElmer 240 Elemental Analyser.N MR data were collected on aB ruker Avance III HD 400 MHz Smart Probe FT NMR spectrometer (400.130 MHz for 1 H, 100.613 MHz for 13 C, 104.261 for 27 Al). Spectra were obtained at 25 8C. For 1 Ha nd 13 C, chemical shifts are internally referenced to deuterated solvent and calculated relative to TMS. For 27 Al, an external reference was used (1 m AlCl 3 (H 2 O) 6 in D 2 O). Chemical shifts are expressed in d ppm. The following abbreviations are used:b r = broad, m = multiplet, q = quartet, s = singlet, sh = shoulder,t= triplet.

Crystallographic details
Crystals were transferred from the mother liquor to ad rop of perfluoropolyether oil mounted upon am icroscope slide under cold nitrogen gas. [42] Suitable crystals were attached to the goniometer head via aMicroLoop TM ,which was then centred on the diffractometer.D ata were collected on aB ruker D8 Quest (Cu-Ka, l = 1.54184 ), equipped with an Oxford Cryosystems low-temperature device. Structures were solved using SHELXT,w ith refinement, based on F 2 ,b yf ull-matrix least squares. [43] Non-hydrogen atoms were refined anisotropically and ar iding model with idealized geometry was employed for the refinement of H-atoms. For 6 2 one BnMe 2 Cg roup was modelled as disordered, though separate positions for the phenyl group could not be refined satisfactorily.T he occupancy was refined, with restraints placed upon both the 1,2and 1,3-distances and upon the anisotropic atomic displacement parameters. CCDC 1504652 and 1504653 contain the supplementary crystallographic data for this paper.T hese data are provided free of charge by The Cambridge Crystallographic Data Centre.

Spectroscopic characterization of BnC(O)OMe 5 + + TMA reaction mixtures
As for 1 + TMA but using methyl phenylacetate 5 (0.42 mL, 3mmol) to give 5(TMA), 6(TMA) and 3.A na liquot (0.1 mL) was mixed with [D 6 ]benzene (0.7 mL) and analyzed by NMR spectroscopy. Co-synthesis and characterization of BnMe 2 COAlMe 2 (TMA) 6(TMA) and 3 Method 1) TMA (4.5 mL, 9mmol, 2.0 m in toluene) was added dropwise to methyl phenylacetate (0.42 mL, 3mmol) under aN 2 atmosphere at À78 8Ca nd allowed to reach room temperature. The resulting solution was stirred and generated heat. After 2h ours the solution was placed under vacuum to remove the toluene. The remaining liquid was stored at 4 8Cf or 1day,p roducing colourless crystals of 6(TMA) and 3.C ombined yield of 6(TMA) and 3:9 10 mg (83 %o ft he total mass expected);m .p. Method 2) As for Method 1b ut after stirring the reaction mixture for 2h ours the solution was placed under vacuum to remove the toluene. The remaining liquid was treated with hexane (1 mL) and the resulting solution stored at À20 8Cf or 1day,p roducing as mall quantity of colourless crystals. Combined yield of 6(TMA) and 3: Synthesis and characterization of BnMe 2 COAlMe 2 6 TMA (4.5 mL, 9mmol, 2.0 m in toluene) was added dropwise to methyl phenylacetate (0.42 mL, 3mmol) under aN 2 atmosphere at À78 8Cb efore being allowed to attain room temperature. The resulting solution was stirred and generated heat. After 2h ours the solution was placed under vacuum to remove the toluene. The remaining liquid was treated with Et 2 O( 3mL) to give aw hite precipitate that dissolved upon gentle heating. Colourless prismatic crystals formed as the mixture cooled to room temperature and over ap eriod of 1day produced al arge crop of 6.