A Journey from Thermally Tunable Synthesis to Spectroscopy of Phenylmethanimine in Gas Phase and Solution

Abstract Phenylmethanimine is an aromatic imine with a twofold relevance in chemistry: organic synthesis and astrochemistry. To tackle both aspects, a multidisciplinary strategy has been exploited and a new, easily accessible synthetic approach to generate stable imine‐intermediates in the gas phase and in solution has been introduced. The combination of this formation pathway, based on the thermal decomposition of hydrobenzamide, with a state‐of‐the‐art computational characterization of phenylmethanimine laid the foundation for its first laboratory observation by means of rotational electric resonance spectroscopy. Both E and Z isomers have been accurately characterized, thus providing a reliable basis to guide future astronomical observations. A further characterization has been carried out by nuclear magnetic resonance spectroscopy, showing the feasibility of this synthetic approach in solution. The temperature dependence as well as possible mechanisms of the thermolysis process have been examined.


Introduction
In the last few decades, astrochemistry-the research field focused on the chemical composition and evolution of ordinary matter in space-has flourished dynamically:b orn as an iche sector,i tg rew up into af ield of broad interest. Its birth is relatively recent because, for many years, the interstellar medium (ISM)-the space between star systemsi nag alaxy-has been considered too hostile to bear any chemical complexity.H owever,w ith the rise of radio astronomy( in the 1960s), it became evident that the ISM harborsadiversec ollection of interesting polyatomic (both organic and inorganic)m olecules. [1,2] Since then, alargenumber of molecular specieshas been detected. [3] Among them, those having aprebiotic character are of particular interest because of their key role in postulated mechanisms leading to the emergency of life. Imines belong to this category.
While their astrobiological relevance has not yet unambiguously proven, imines are known to play major roles in many chemicalp rocesses. They have been recognized as crucial intermediates in organic synthesis, due to their extensive use in the preparation of N-containing compounds, [4] and as naturally or biologically active molecules. [5] Because of peculiar and extremec onditions, the chemical reactivity in space is very different from the terrestrial counterpart, and unstable species (even ions and radicals) can survive long. As ac onsequence, reactive species such as some NÀHi mines, whose isolationi s difficult on Earth, can be observed in the ISM, [6][7][8][9][10] with the prerequisite of as uccessful laboratory spectroscopy characterization. With the chemistry in the ISM being new and intriguing, severed ifficultiesa rise wheni nterstellarm olecules are to be produced by "terrestrial" organic proceduresf or their characterization,e .g.,b ym olecular spectroscopy.I ndeed, the instability of these species often requires that they are directly generated inside the spectrometer using harsh conditions and rather poorly predictable techniquess uch as pyrolysis or electric discharge. We have tried to overcome these difficultiesb yd eveloping ad ifferent and easier synthetic route, the thermally tunable formation of imines, which has been tested both in the gas phase and in solution.
The subjecto ft his work is the case study of phenylmethanimine (PMI), which is an on-standard reactivem oleculeo fp otentialastrochemical relevance. Due to the large amount of hy-drogen (in the form of H, H 2 ,a nd also H + /H 3 + )i nt he ISM, hydrogenation is an efficient process [11] and leads to the formation of saturated or partially saturated molecules. [12] On this ground, hydrogenation of benzonitrile (BN) in the ISM can yield PMI, an interesting imine in terms of chemical complexity and molecular evolution. This gained attention because of the recent detection of BN in the cold-core Taurus Molecular Cloud 1( TMC-1), which provided the unequivocal proof of the presenceo fb enzene in that cold environment, and-more generally-in the ISM. [13] In turn, benzene is the building block of polycyclic aromatic hydrocarbons (PAHs), which are recognized (but spectroscopically not proven) to be important constituents of small dust grains in interstellar clouds and are believed to play ak ey role in the chemical evolution in space from both ac atalytic and protective point of view. [14] Despite its appeal as ag ood approachf or imine synthesis in the ISM, on Earth, nitrile reduction using molecular hydrogen usually requires catalytic conditions which are not compatible with as ingle hydrogenation process. Catalytic processes directly lead to the saturated derivative, that is, the corresponding amine, or to am ixture of primary-, secondary-and even tertiary-amines via reactiono ft he NÀHi mine intermediates. [15,16] This is due to the high reactivity of the imine intermediate. [17][18][19][20][21] The NÀHi mine derivatives of aliphatic or aromatic aldehydes are known to be much less stable, ando nly in af ew cases they have been successfully isolatedo rc haracterized. [22,23] Indeed, the literature is richo fe xamples supporting the instability and reactivity of these imines. The chemical labile nature of NÀHi mines, and the difficulties relatedt ot heir isolation,r esulted in the employment of imines possessingv arious substituents on the nitrogen atom (N-substituted imines). Therefore, the activating or protecting groups introduced on the nitrogen atom need to be removed after the synthetic endeavor.B ased on the concept of atom economy,N -unsubstituted imines would be ideal,s ince deprotection steps, to obtain the final products, are not required. Unfortunately, unlike N-substituted imines, the approaches to N-unsubstituted imines are limited to (and, in many cases,w ere proposed as) unstable, not characterized intermediates. [24] The direct reaction with ammonia and aldehydes is not as uitable synthetic procedure for such imines.N ÀHi mines can be accessed by the controlled hydrolysis of N-metalloimines or silylimines. [25] Another simple accesst oN ÀHi mines involves availablea zides precursors with the removal of N 2 ands ubsequentm igration of hydrogen, under catalytic conditions. [26] Nevertheless, all these approaches are impracticable in order to generate NÀH imines for rotational spectroscopy investigations in gas phase. Indeed, such contexti ncreased the relevance of at hermally tunable approachf rom as imple precursor for the generation of NÀHi mines and their spectroscopici nvestigation. Such approaches avoid criticali ssues such as:( i) demanding experimental setups, that is, subtle control of reaction conditions;( ii) trapping procedures leadingt oi mine complexes;( iii)use of transition metal catalysts;( iv) unfeasible interface with spectroscopic techniques.
The last aspect is of particular interestt oo ur investigation since, in order to set up and validate our synthetic route, high-resolution spectroscopic techniques offer the highest specificity:r otationale lectric resonance( RER) spectroscopy in the gas phase and nuclear magnetic resonance (NMR) spectroscopy in solution. Since our strategyi sb ased on two key points, namely (i)the use of stable and affordable organic precursors and (ii)the generation of the unstablea nd reactive imines through simple conditions, it provides as uitable way to produce imines directly inside the spectrometer.
In this manuscript, we report af ull account of our successful endeavor,w hich has led -for the first time-to ac omplete and accurate spectroscopic characterization of PMI using RER, which is prerequisite for its identification in the ISM by means of radio astronomy.I tr equired am ultidisciplinary effort combining organic synthesis and molecular spectroscopy,s upported and guided by computational chemistry.

Spectroscopic characterization of PMI
To guide RER experiments, as tate-of-the-artq uantum-chemical (QC) characterization has been performed. Using the density functional theory (DFT,s ee Supporting Information, ford etails), ap reliminary scan of the potential energy surfaceh as been carriedo ut, which located two isomers (E and Z,r eferring to the relative position of the phenyl group with respect to the imine hydrogen, see Figure 1). Subsequently,a na ccurate structural determination has been obtained by resorting to the socalled "cheap" composite scheme [27] (hereafter denoted as ChS), whosed enomination refers to the limited computational cost in spite of its accuracy.
The ChS approach is expectedt op rovide results with an accuracy of about0 .001-0.002 for bond lengthsa nd 0.1-0.28 for angles. [27] This approach has also been used to derive accurate electronic energies, which are given in Figure 1. While structuralp arameters and ad etailed descriptiono ft he ChS model are provided in the SupportingI nformation, this Figure pointso ut that the E isomer is about 6kJmol À1 more stable than the Z species. TS E-Z has aq uasi-linear C-N-H geometry, with the isomerization process occurring in the molecular plane, similarly to what already observed for otheri mines such as ethanimine (ETI) [28] and C-cyanomethanimine( CCMI). [29] Althoughd ifferent computational schemes have been applied, a qualitative comparison between the relative energies is possible. As maller energy differenceb etweent he two isomersw as obtained for ETI and CCMI (2.8 kJ mol À1 and 2.0 kJ vmol À1 ,r espectively [28,29] ), while as imilar isomerization barrierw as found (about 116kJmol À1 for ETI, [28]~1 11 kJ mol À1 for CCMI, [29] and about 112kJmol À1 for PMI).
Equilibrium rotational constantsh ave been straightforwardly derived from the ChS equilibrium structure, while the required vibrational corrections have been evaluatedf rom DFT anharmonic force-field calculations (thereby using B3LYP, [30,31] see Supporting Information for details). First-order properties, such as dipole moment components and nuclear quadrupole coupling constants, have been computed using,i nt he framework of DFT,t he double-hybridB 2PLYP functional [32] (see Supporting Information for details). This level of theory has also been employed for deriving centrifugal distortion constants.
The gas-phasec haracterization of PMI has been performed in the 3-26 GHz range, using the COBRA-type (Coaxially Aligned Beam Resonator Arrangement) FourierT ransform Mi-crowaveS pectrometer (FTMW), described in detailse lsewhere. [33] As olenoid valve with reservoir,f illed with hydrobenzamide (HBA) and heated to 85 8C, was used to vaporize it and, at the same time, to thermally-decompose it, thus producing PMI in the gas phase. HBA was synthesized,f ollowing the procedure described in literature, [17,[34][35][36] by condensation of benzaldehyde and ammonium hydroxide solution and obtained as white solid, with its identity and purity being confirmed by NMR analysis( see the Supporting Information for details).
Given the high accuracyo ft he QC calculations performed (the relevant spectroscopicp arameters are collectedi n Ta ble 1), the computational prediction of the rotationals pec-trum was expected to match the accuracy [37] required for an unequivocal identification of the E and Z isomers of phenylmethanimine in the gas phase. Indeed, it was straightforwardt o successfully identify and assign more than one hundred transition frequencies for both PMI isomers. Their unique pattern (hyperfine structure), due to quadrupole coupling splitting originated by the presence of the 14 N nucleus (I = 1), is shown in Figure2 for both PMI isomers. Furthermore, we note that each rotational transition appearsa sad oublet due to the Doppler effect arising from as upersonic jet expansion. For the most stable E-PMI,t he spectrala nalysish as been extended in the 83-100 GHz range, thus allowing the improvement of the spectroscopic parameters, using af requency-modulation millimeter-wave (FM-mmW) spectrometer (see ref. [38] for ad etailed description). In this experiment,adifferentp roduction methodo fP MI has been employed, as reasoned below. Indeed, PMI was produced by meanso ff lash vacuum pyrolysis (FVP) [39,40] of asample of a-methylbenzylamine (890 8C).
The spectroscopic parameters derived from aw eighted nonlinear fit of the observed transitions, performed using the Pickett's CALPGMs uite of programs, [41] are reported in Table 1. An excellent agreement between the computed and experimental values can be noted, with an average error of 0.03 % on the effective rotational constants and am aximum discrepancy below 0.05 %. Only smalld eviations have also been found for the quartic centrifugal distortion and nuclear quadrupole coupling constants. Since rotational and quadrupole coupling constantsa re strongly tied to the molecular structure,t he very good agreementg ives further confidence in the reliability and accuracyo ft he computed geometries.
While the full list of the rotational transitions included in the fit is given in the Supporting Information, we state here that the root mean squaree rror (rms) is smaller than 5kHz, with a standardd eviation (s)s lightly below unity,t hat is, achieving experimental accuracy.

Thermal decomposition of HBA
The successful generation and characterization of PMI in the gas phase raised questionsa bout the mechanism of its formation and the possible extension of the methodology to the condensed phase (i.e. solution) and, more generally,t oo ther N-unsubstituted imines (thus avoiding more expensive and/or more chemically demanding procedures). For this reason, the thermalb ehavioro fH BA has been explored from room temperaturet o1 00-110 8C, both in the gas phase by means of RER and in solution by 1 H-NMR. In particular,t he J Ka,Kc = 5 0,5 ! 4 0,4 rotational transition of four molecular species, namely BA, BN, and the E and Z isomers of PMI, has been recorded with the COBRA-FTMWs pectrometer. To provideareliablec omparison between the intensityo ft he transitions, the field amplitude of the microwavee xcitation pulse has been scaled reciprocal to the value of dipole momentc omponent along the a-axis and a2 000a veraging signal acquisition has been chosen.
The recording has been started ten minutes after the temperatures et up. The recorded spectra (see Figure 3) reveal the presence of BA at all temperatures,w hile clearly distinguishable signals of PMI isomersa re observed only above 80 8C. BN lines appear at the same temperature. For both PMI isomers and BN, it is noted that rising the temperature increases the intensity of the transitions, thusi mproving the signal-to-noise ratio (S/N). Although not shown in Figure 3, the 12 321.0 MHz transition of the water dimer [42] was observed fore ach temperature increment.T his confirms the steady presence of water in the experimental apparatus,w hichv ery likelyp lays ak ey role in the investigated hydrolytic process.
An important note on the relativep opulations of the two PMI isomersi sd eserved. In principle, this information should be derivablef rom the spectra of Figure 3; however,t he E-PMI transition reported there lies very close to the strong transition of BA, thus clearly modifyingt he baseline of the E-PMI transition. As ac onsequence, the signal-to-noise ratios of Z-PMI and E-PMI lines cannot be quantitatively compared, thus preventing the derivation of the relative populations. We can qualitatively comment that, duringt he recording of the spectra, we noticedt hat the intensities of both isomers were comparable. This is probably due to ac ooperation of population distribu-tion andd ifferences in the electric dipole moment components.
As to the NMR characterization, according to the available literatured ata, PMIs hows two distinct sets (each with two doublets) of signals for the imino hydrogen, with different coupling constants (J = 16 Hz for the E isomer, J = 25 Hz for the Z one). [26] In our analysis, the sample was heated to the desired temperature and the 1 HNMR spectrum was acquired after 10 minutes to avoid thermalizationp rocesses during the acquisition (a completea ccount of the experiment is provided in the Supporting Information). The formationo fb oth BA and E-PMI was observed as ac onsequence of the step-by-step temperature rampf rom 25 8Ct o1 10 8C( see Figure 4). However,n o NMR spectroscopicc lues denoting the possible formation of BN or Z-PMI were observed, even when the experiments were performeda th ighert emperature.A sf ar as Z-PMI is concerned, however,o ur experimental resolution did not allow for unequivocally excluding its presence. This suggestst hat both experiments (RER and NMR) are characterizedb ye ither different hydrolytic mechanismso rt he availability of additional pathways in the case of gas-phase measurements. To confirm the absence of BN in solution after thermal treatmento fH BA, a  trace amount of the former was added to the NMR samples previously heated at 110 8Ca nd the spectrumw as again recorded. The analysis revealed the presence of new signals (doublet at 7.68 ppm, overlapping with aB At riplet;d oublet at 7.63 ppm;t riplet at 7.50 ppm, overlapping with aH BA multiplet), due to the presence of BN and previously absent. Hence, no BN is formed by hydrolysis of HBA in solution (see Supporting Information). This might suggest that its generation occurs through pathways that rely on metallicc atalytic surfaces, such as fast dehydrogenation of PMI. [43,44] Indeed, it has been provedt hat the metal catalysis can take place in the nozzle head in the RER experiment, [45] while this is not the case within the NMR tube. Figure 4s hows the changes-upon heating-of the diagnostic NMR signals (top) and the temperature-dependence of their normalized integral values with the respectivel inear regressions( bottom). Such analysisw as performed after an arbitrary assignment to the integral value of the residual peak of 1,1,2,2tetrachloroethane-D 2 (fixed to 10). The negative-slope traces correspond to the starting material, that is, HBA, whilet he positive-slope traces are attributable to forming species. As expected, the signalr elative to the two CHN protons of HBA (red trace) decreases twice as fast as the single-proton aminal signal (cyan trace).
Twos ignals (i.e. green and black traces) have almosts uperimposed trends in terms of both slope and normalized integral values, thus suggesting that the respective protons belong to the same species.I ndeed, they are compatible with the CHO and para-proton (triplet) of BA, respectively,t hus featuring a 1:1i ntegration relationship.T he other two signals (i.e. blue and gray traces) feature a2:1 integration relationship.Inparticular,t he singlet at 8.70 ppm (gray trace) is compatible with an iminic CHN proton, whilet he doublet at 7.77 ppm (blue trace) is compatible with the iminic ortho-protons. Furthermore, the grey trace has as lope which is similart ot hose observed for the aldehydic signals, thuss uggesting that the generation processes of the two species are somehow interconnected. The fact that BA appearst ob ep roduced before PMI suggests that the involved hydrolytic mechanism proceeds with the initial formation of BA, followed by the formation of PMI starting from other transient species.
Here, we propose ap utative hydrolytic pathway (Scheme1) which is compatible with the aforementioned observations and with previously proposed mechanisms. [46][47][48] As noted above,t he additional pathway leading to the formation of BN startingf rom PMI during the microwaves pectroscopic measurements is supposed to be ad ehydrogenation.I no ur opinion, the following key pointsn eed to be analyzed to explain the proposedm echanism:( i) the role of water;( ii)the role of the metallic surface in the COBRA-FTMW spectrometer with respect to the glass surfaces acting in the FM-mmW experiment. Within the former context, water is expected to play ac rucial role:t he hydrolytic pathway is supposed to proceed as long as water is available. In fact, the amounto fw ater detected by means of the low-frequency RER experiment as well as the residual water of the deuterated solvent is believed to be sufficient to observe the formation of the main species arising from HBA hydrolysis,t hat is, BA at room temperature and PMI at highertemperatures.
On the other hand, the FM-mmW experiment can provide furtherc lues for ad eeper understanding of the hydrolytic mechanism. At first, at entative generation of PMI in the gas phase has been carried out by thermolysis of HBA.
The solidw as placed in ag lass tube and heatedu pt o 100 8C, also ensuring au niform heatinga long the path to the absorption cell.W hile heatingt he sample up, ap ortion of the spectrum around8 5.5 GHz wass canned in the attempto fd etecting two strong transitions of E-PMI,a sp redicted by our low frequency measurements. No signal attributable to PMI (nor BN) was found.
Typically,t he whole glass apparatus of the FM-mmW spectrometeri sp umped continuously, thereby removing water,a lthoughi ts residual presence cannot be ruled out. However,n o metallics urfaces are availablei nt he instrument,t hus leading to the formulation of two hypotheses for the lack of formation of PMI in the FM-mmWe xperiment:( i) water is not availablei n as ufficient amount to hydrolyze HBA;( ii)the metal catalysis is mandatory to obtain PMI from HBA in gas phase.
To verify the reliability of spectral predictions and therefore rule out the possibility of false negatives, we adopted ad ifferent production method, that is, FVP.B yp yrolysis of two possible precursors of PMI (for av acuum dynamic preparation of PMI, see ref. [49]), that is,b enzylamine and a-methyl benzylamine, through dehydrogenation or elimination of CH 4 ,r espectively,asmall set of 27 transitions (belonging only to E-PMI) could be measured, with the latter precursor leading to the highest S/N of the spectra. Conversely,t he use of N-methyl benzylamine gave no signals ascribable to the presence of PMI. BN was found as ap yrolysis co-product,a sp roven by recording its rotational transitions. While the PMI signalr eached its maximum intensity by setting the furnacet emperature to 890 8C, the intensityo fB Nt ransitions kept increasing up to 1200 8C. This confirms the prevalence of BN at higher temperatures, in agreement with the low-frequency RER experiment.
Although at horough analysis of the mechanisms taking place in the FVP process is beyond the scope of this work, the FM-mmWe xperiment proved the reliability of the centimeterwave RER measurements and their extrapolation at higherf requencies, but it left some unexplored areas concerning PMI formation from HBA in the gas phase if no metallic surfaces are available. As mentioneda bove,s uch possibilities could be furthere xplored going beyond simple thermal conditions. Several attempts to identify the Z isomer have been carriedo ut, but no signal ascribable to it was found. This might suggest that only the E isomer is generated by ash vacuum pyrolysis, but we did not investigate further this aspecti nt he present study.

Conclusions
An easy and affordable approach based on hydrobenzamide thermolysis is presented, which ensures to obtainp henylmethanimine both in gas-phase and in solution as confirmed by rotational electric resonance (RER) and 1 Hn uclearm agnetic resonance (NMR) spectroscopy experiments, respectively.Adetailed structural and energetic description of phenylmethanimine has been carriedo ut by resorting to composite schemes for accurate results. This paved the wayf or the registration and analysis of the microwaves pectrumo fb oth E-a nd Z-phenylmethanimine, leadingt ot heir first laboratory identification. This work is ap rerequisite for the possible radio astronomical detection of theses pecies in the interstellar medium, relying on accurate rotationalr est frequencies.F irst, in view of the strong chemical connection between benzonitrile andp henylmethanimine, an astronomical search in the region Taurus Mo-lecular Cloud (TMC-1) is suggested. Finally,o wing to the thorough analysis of RER and NMR spectra at different temperatures, ap ossible mechanism of phenylmethaniminef ormation by thermalt uning of hydrobenzamide, in which water is thought to play ac rucial role, is also proposed.