Alpha‐Carbonic Acid Revisited: Carbonic Acid Monomethyl Ester as a Solid and its Conformational Isomerism in the Gas Phase

Abstract In this work, earlier studies reporting α‐H2CO3 are revised. The cryo‐technique pioneered by Hage, Hallbrucker, and Mayer (HHM) is adapted to supposedly prepare carbonic acid from KHCO3. In methanolic solution, methylation of the salt is found, which upon acidification transforms to the monomethyl ester of carbonic acid (CAME, HO‐CO‐OCH3). Infrared spectroscopy data both of the solid at 210 K and of the evaporated molecules trapped and isolated in argon matrix at 10 K are presented. The interpretation of the observed bands on the basis of carbonic acid [as suggested originally by HHM in their publications from 1993–1997 and taken over by Winkel et al., J. Am. Chem. Soc. 2007 and Bernard et al., Angew. Chem. Int. Ed. 2011] is inferior compared with the interpretation on the basis of CAME. The assignment relies on isotope substitution experiments, including deuteration of the OH‐ and CH3‐ groups as well as 12C and 13C isotope exchange and on variation of the solvents in both preparation steps. The interpretation of the single molecule spectroscopy experiments is aided by a comprehensive calculation of high‐level ab initio frequencies for gas‐phase molecules and clusters in the harmonic approximation. This analysis provides evidence for the existence of not only single CAME molecules but also CAME dimers and water complexes in the argon matrix. Furthermore, different conformational CAME isomers are identified, where conformational isomerism is triggered in experiments through UV irradiation. In contrast to earlier studies, this analysis allows explanation of almost every single band of the complex spectra in the range between 4000 and 600 cm−1.


Introduction
The reactivity of carbonic acid (H 2 CO 3 ,C A) towards its ester derivatives has been of interest for more than ac entury.S tarting from the basic formation studiesb yH empel and Seidel at the end of the 19th century, [1] nowadayst he interest is more focused towards, for example,i ts importance in biochemistry [2] or food chemistry. [3] Singly esterified carbonic acids are known as hemiesterso fc arbonic acid (HECAs), [2] and their salts are knowna sm onoalkyl carbonates (MACs). [4] The molecule investigatedi nt his work is the methyl hemiester of carbonic acid, which we will refer to as CAME (carbonic acid monomethyl ester) in the following. In biochemistry, [2] and in food chemistry, that is, carbonated alcoholicbeverages, [3] the focus is on detection andr eactiono fv ery low concentrations of carbonic ester derivatives in aqueouss olution. [5] MACs or HECAs are studied from small alkyl esters to quite complex esters, for example, with sugars. [4] The biological relevance emphasizes the need for simple synthesis and characterization of HECAs as ap ure substance. Most synthesis routes have temperatures below 273 Ki n common.P ure CAMEw as first synthesized by Hempel and Seidel [1] in 1898 (by reaction of aqueous CO 2 with methanol) as as olid that melts between À57 8Ca nd À60 8C. In 1972, Gattow and Behrendt [6] reported the formation of CAMEb yu sing nonaqueousc hemistry,n amely the reactiono fN aOCH 3 dissolved in methanolw ith CO 2 .T his hemiester was described as 'a colorlessm aterialt hat melts at À36 8C' and was characterized by using infrareds pectroscopy.I n2 006, Dibenedetto et al. [7] statedt hat the isolation of monoalkyl derivativeso fH 2 CO 3 is 'not trivial'.T hey observed traces of aqueous CAMEa troom temperature by forming NaOC(O)OCH 3 (from the reactiono f sodium methoxide with CO 2 )a nd subsequenta cidification. Their characterization methodo fc hoice was NMR spectroscopy.
In May 2014, one of us, Jürgen Bernard,s tated in his Ph.D. thesis that CAME can be synthesized and isolated as as olid by reactiono fK HCO 3 with absolutem ethanol followed by acidification at cryo-conditions. [8] Av ery similar( but not identical) preparation route was used by Hage, Hallbrucker,a nd Mayer (HHM) in 1993. [9] They assigned the resulting solid substance as ap olymorph of H 2 CO 3 on the basis of IR spectroscopy and termedi t'alpha-carbonic acid' (a-H 2 CO 3 ). Based on this pioneering work, HHM, [9][10] Winkel et al., [11] andBernard et al. [12] assumed in later work that dissolution of KHCO 3 in methanolf ollowed by acidification leads to a-H 2 CO 3 .I nc ontrast, 'beta-carbonic acid' (b-H 2 CO 3 )w as obtained by HHM by replacing methanol with water as as olvent and by high energy irradiation of CO 2 /H 2 O [13] mixtureso rH -implantation [13b, 14] -leading to the claim of polymorphism for H 2 CO 3 .T he interpretationo ft he formation of b-H 2 CO 3 remainsu ncontested. The reinterpretation of all earlier work on a-H 2 CO 3 and the polymorphism of H 2 CO 3 is outlined in the present work, in accordance with the first claim provided in the Ph.D. thesis of Bernard. [8] The revised interpretation is based on detaileda nalysiso fI Rs pectrao ft he solid at cryo-conditions and single-molecule IR spectra recorded after sublimation and matrixi solation. These spectra suggest the presence of CAMEm olecules rather than H 2 CO 3 molecules as originally envisaged. [12] In his Ph.D. thesis, Bernard investigated not only CAME but also CAEE-the monoethyl ester of carbonic acid. [8] The infraredd ata of solid CAEE as wella si ndividual CAEE molecules isolated in matrix can be found in ref. [15].T he reinterpretation of the matrix spectra was also suggested by Reisenauer et al. [16] in September 2014. In contrast to the presentc ryo-study,R eisenauer et al. have studied pyrolysis of dialkyl carbonates at about 1000 K. After isolating the pyrolysis products in argon matrix at 8K,t hey identified a product identical to the one found by Bernard. [8] Both Bernard and Reisenauer et al. identified thisp roduct to be carbonic acid monomethyl ester (CAME, HO 2 COCH 3 )r ather than H 2 CO 3 . Bernardt rapped the gas phase of the sublimed pure solid, while the matrix spectra presentedb yR eisenauer et al. [16] contain pyrolysis byproducts such as isobutene, thereby obscuring some spectral ranges. In contrast to our work, Reisenauere tal. also did not provide solid-state spectra to back up their claim that the solid-state spectra reported by HHMn eed reinterpretation.
In the presents tudy,t he re-evaluation of a-H 2 CO 3 as CAME is built on four pillars:( i) variation of solvents during both preparation steps;( ii)isotopics hifts in the solid-state spectra by substitution of the CH 3 with aC D 3 group and matching with calculated spectra of speciesc onnected by hydrogen bonds;( iii)complete assignmento fp ractically all bands between 4000 and 600 cm À1 of matrix isolation IR spectra by considering different CAME conformers, butalso CAME dimers and water complexes;a nd (iv) forced conversion of conformers by irradiation experiments of the molecules trapped in the matrix. The assignment for the single moleculest rapped in matrix (item (iii)) is guided by harmonic frequency calculations at the MP2/aug-cc-pVTZl evel of electronic structure theory and isotopic labeling, leading to the following CAMEi sotopomers: HO 2 COCD 3 (CD 3 -CAME), DO 2 COCH 3 (OD-CAME), and HO 2 13 COCH 3 ( 13 C-CAME). This strategy puts us in ap ositiont o identify minor species present in the matrix besides the two CAME conformers identified by Reisenauer et al. [16] To learnm ore about the chemistry of the methyl group in the process leadingt ot he pure solid, the cryo-technique as employed by HHM [9][10] was adapted. Specifically,t he solvent was evaporated two times rather than just once. HHM were depositing micrometer-thick sandwiches of alternating glassy layers of acid (e.g.,H Cl) and base (e.g.,K HCO 3 )a t7 8K.T his sandwich was heatedb yH HM to induce devitrification (transformation to the supercooled liquid), diffusion, andp rotonation, after whicht he solvent was evaporated. In our work, KHCO 3 was dissolved in methanol, deposited at approximately 80 K, then immediatelyh eated to remove the solvent for the first time. In the next step, the precipitate was cooled to approximately 80 K, and al ayer of glassy acid wasd eposited on top. Heating for as econd time induces devitrification, diffusion, anda cid-base reaction. After this, the solventw as evaporated again.E vaporatingt wice (rather than once by HHM) allowsf or systematic variation of the solvent in the first and second evaporation steps.B yu sing water,m ethanol, or ethanol and combinationso ft hese for the two evaporation steps, we reveal that the solvent used for the dissolution of the salt in the first step is decisive as to whether H 2 CO 3 , [10b] CAME, or CAEE [15] is obtained as the product.
IR bands pertaining to the methyl group in CAME are generally weak. To confidently assign the bands, we not only rely on the absolutec alculated frequencies themselves, but also on shifts of bands upon isotope substitution. Furthermore, the band assignment is also guided by matrix irradiation experiments. Upon irradiation of the matrix with UV light, the trapped speciesc an becomee xcited and internal rotationo r intramolecular bond cleavage is caused. As the excited molecule or its fragments cannot escape from the cage, different conformers are formed owing to relaxation or recombination. In difference spectra,i ti sp ossible to find bands that arise from the same conformationals pecies. The conformer that is formed upon UV irradiation will have bands pointing upwards, whereas the conformer that is depletedw ill show bands pointing downwards in the difference spectra.F inally,w ecompare the matrix spectra obtained here with spectra obtained in earlier work after sublimation of the monoethyl hemiester of carbonic acid (CAEE) [15] and b-H 2 CO 3 .T hese strategies allow for an assignment of practically all observed signals in the matrix isolation IR experiments and ac lear distinction of bandsa rising from different monomer conformations and dimers. Based on these procedures, we are even able to determine the ratio of different monomer conformers in the matrix.

Ab initio calculations
The structure of CAME in its solid state is amorphous and thus, not known.The solid-state spectra reportedc annotsatisfactorily be explained with the help of ab initio calculations owing to the lack of aw ell-defined crystal structure. Accordingly,t he assignmentf or the solid is not based on calculations. Instead, it is providedw ith the aid of differences pectroscopy and CD 3isotopes ubstitution experiments.
In contrast, matrix isolation spectroscopy is as ingle-molecule technique. As isolatedm olecules are trappedi na ni nert matrix,s uch spectra are in general very close to the gas-phase spectra. [17] The full width at half maximum (FWHM) of bands in matrix spectra are orders of magnitude smaller than FWHM of bands in solid-state spectra.T hus, line spectrao fi ndividual molecules calculatedb ya binitio methods in the gas phase are very useful to interpret and assign the observed bands. The static electric field exertedb yt he inert noble gas argon in matrix isolation spectroscopy is orders of magnitude smaller than the crystal field. Thus, ab initio quantum chemical calculations can directly be used to guide the band assignment of the matrix isolation spectra.O wing to the size of the molecule, we have to rely on the well-established harmonic approximation. Although the calculated spectra in general match the measured matrix spectra well, there are some small discrepancies even after applying ac orrection factor to the calculated frequencies. Thism ay arise as ar esult of anharmonicities and mode-modec oupling effects, which are not included in the harmonic approximation and may cause-depending on the specific mode-ar ed-or blueshift. Furthermore, the matrix cage causes as lights hifto ft he bands owing to the cage geometry and in somec ases am atrix splitting of bands as a result of different cage sites or symmetry reduction of the isolated molecules. [18]

Structures and stabilitieso fCAME monomers
To assess the conformational space and to estimate the kinetics, the potential energy surfacef or the torsional movement of the methyl group and the terminal hydrogen atom of the hydroxyl group was calculated with ab initio wave function methods (MP2/aug-cc-pVTZ) as depicted in Figure 1. Figure 1a shows the low-energy conformationso ft he CAME molecule and their relative electronic energies.T he nomenclature for these conformational isomers is based on an analogy to the nomenclature of 1,3-butadiene by using the descriptors s-cis and s-trans for the conformation around the single bonds 1-2 and 2-4. Structures Ia nd II are within 6.0 kJ mol À1 ,w hereas structure III is slightly higheri ne nergy (+ 14.7 kJ mol À1 ) and structure IV is energetically ratheru nfavorable (+ 46.7 kJ mol À1 ). Explicitly-correlated coupled cluster singlepoint calculations (CCSD(T)-F12/cc-pVTZ-F12)o nt he MP2/augcc-pVTZr e-optimized structures when molecular symmetryi s taken into accountc onfirmed the resultsf rom the potential energy surfaces can and yieldede nergieso fs tructure I: 0.0 kJ mol À1 ,s tructure II:5 .8 kJ mol À1 ,a nd structure III: 14.7 kJ mol À1 .A st hese values are very similart ot he MP2 ones, convergenceinthe electronic structure can be assumed.
The experimentall ow-temperature conditions make it highly unlikely to detect structures with ar elative energy highert han 15-20 kJ mol À1 compared with the globale nergy minimum (structure I)-see also calculations about conversion to structure Id uring the flight time of the preparation of the matrix, below.Thus, structure IV is from now on neglected.
Similarly,t he barriers to rotation for the methyl group and the hydroxyl group as estimated from the potential energy surface (PES) in Figure 1b are rather substantial, for example, approximately 40 kJ mol À1 for the conversion from structure It o structure II and even higher for the conversion to structure III. The remarkable conformational stability and the high torsional Figure 1. a) Energy minima and stereo-nomenclature for conformational isomers of CAME according to MP2/aug-cc-pVTZ. Energies are electronic energies without zero-pointenergy corrections. Atom colors:g ray = C, white = H, red = O. b) Potentiale nergy surface for torsional movement of the methyl group (x axis) and of the terminal hydrogen atom (y axis) as calculated at the MP2/aug-cc-pVTZ level of theory. barriero fC AME can be rationalized by am inimization of electrostatic and closed-shell repulsion between the oxygen lone pairs and the carbonyl double bond, which is best realized in the conformation 1-2 s-trans and 2-4 s-trans of structure I. Rotation of the terminal hydrogen or the methyl group to a cis conformation is associated with as ubstantial energy penalty owing to the close vicinity of the oxygen lone pair and the carbonyl double bond, which experience mutual electrostatic and closed-shell repulsion.

Constitution, stability,a nd interactioninCAME dimers
Although for carbonic acid (CA), no dimers were foundi nt he matrix, [19] the situation may be different for CAME. To access whether dimers are viable structures, likelyt oo ccur in the matrix, we studied the structures and stabilities of various CAME dimer conformations.
From the three low-energy conformationso ft he CAMEm onomer,six potentialCAMEdimers could be constructed.
The compositions of the possible dimers are: Each dimeri sa ssembled through two hydrogen bonds between the OH···O = Co ft he respective two monomers ( Figure 2). All structures were fully optimized with MP2/ augcc-pVTZa nd subsequent CCSD(T)-F12/cc-pVTZ-F12 single-point calculations, which yieldedr elative energies as depicted in Ta ble 1. The relative stabilities of these dimers vary by up to 40 kJ mol À1 .R elative free energiesa re very similar to the electronic energies (see Table 1). To further assess the characteristics and energetics of these dimers, we calculated dimerization energies andd imerizationf ree energies, that is, how much energy is released when the dimer is formed from two monomers. Interestingly, dimer 3d isplays the most negative and thus, most favorable dimerization energy of all six dimers, namely À88.6 kJ mol À1 .D imer 2s hows ad imerizatione nergy of À81.5 kJ mol À1 and dimer 1 À75.9 kJ mol À1 ,w hereas all other structures showh igherb ut still favorable relative energies. When considering dimerization free energies at 210 K, only the formation of dimers 1-3 is exergonic, whereas formation of dimers 4-6 is endergonic. Again, the dimerization is most favored for dimer 3( À16.7 kJ mol À1 ), ab it less favored for dimer 2( À10.5 kJ mol À1 )a nd dimer 1( À3.6 kJ mol À1 ), these energies are more favorable than the available thermale nergy at 210 Ki nt he classic approximationu sing R·T,w hich is 1.7 kJ mol À1 .T os hed light on the interaction and the hydrogen-bond strength in these dimers, we investigated the interaction energies, that is, the energy gain owing to the interaction of the two monomer fragments (at the geometry of the dimer complex). In contrast, the dimerizatione nergy is the interaction energy plus the energy that is required to distort the optimized monomers to the dimer geometry.A gain, dimer 3 showst he most favorable interaction energy (À111.6 kJ mol À1 ) and forms, thus, the strongest hydrogen bonds, followed by dimer 2( À99.5 kJ mol À1 )a nd dimer 1( À90.3 kJ mol À1 ). This is furthers upported by structural analyses, where dimer 3s hows the shortest O···H bond of 1.559 versus1 .596 and 1.586 , respectively,i nd imer 2a nd versus 1.619 in dimer 1.
To furtherj udge whether these structures are likely to occur, we comparet he dimerization energies to the energy gain as a result of decomposition into its components CO 2 and methanol. For the decompositiono ft he CAME dimers 1-3, we obtain zero-point corrected energies between À11.7 kJ mol À1 for dimer 1, À26.1 kJ mol À1 for dimer 2, and À40.6 kJ mol À1 for dimer 3, compared with zero-point corrected dimerization energies of À69.4 kJ mol À1 (dimer 1), À75.2 kJ mol À1 ,a nd À82.8 kJ mol À1 (dimer 3). In contrastf or the CA dimer,a ne arlier study found that the dimerization has almost the same energy as its decomposition into CO 2 and H 2 Oc onsidering zero-point energy corrected values, which are both about 67 kJ mol À1 . [20] Excluding entropic contributions, the decomposition of CAME dimers is significantly less favorable than decomposition of the CA dimer.
Althoughd imer 3i sb uilt from two CAME monomers in the less favorable structure III conformation (1-2 s-trans,2 -4 s-cis), its high abundance in the solid (see section 2.3) can be rationalized as follows:( i) the crystal field, that is, the environment in the solid, may affect the conformational preference ands hift the relative stability towards structure III;( ii)conversion between dimer 1a nd dimer 3m ay occurb ys ynchronousd ouble proton transfer of the two protons involved in the dimer bonds or by rotationo fb otht erminal groups. Calculation of rate constantsf or doublep roton transfer in CAME dimers, including the possibility of quantumt unneling,i sb eyond the scope of this work. However,t he protone xchange in CAME dimers can be compared with CA dimers;f or example, formic acid or benzoica cid dimers, which show calculated rate constants of k % 10 9 -10 10 s À1 at 300 Ka nd % 10 5 s À1 at 30 K. [21] Concerted proton transfer of benzoic acid at room temperature has been determined to exhibit an activation energy of approximately 5.4 kJ mol À1 ,w hich is lowered to an apparent activation energy of approximately 0.8 kJ mol À1 at temperatures below 50 Ko wing to quantum tunneling. The barrier for the formic acid dimer is about8kJ mol À1 higher. [21] These comparisons suggestt hat double proton exchange might play ar ole in the gas phase at about2 10 Ko ri nt he matrix at approximately 10 K. However,t unneling splittings associatedw ith this are not observed in the spectra, suggesting that double proton transfer is too slow at 10 K. Based on our computational studies and considerations, dimers 1-3 are likely to occur in the matrix, whereas all other species are thermodynamically not favored and unlikely to be formed. Therefore, only dimers 1-3 will be considered for the spectral assignment. Also, conformational tunneling for s-trans/s-cis rotamerization is too slow in CAME to be observed experimentally in the form of tunneling splittings. [22]

Structure and stability of CAME-watercomplexes
The presence of water vapor in the atmosphere andw ater as an impurity in the solvents may cause contaminationo ft he matrixw ithw ater itself and CAME-water complexes. We studied several conformations for each of the two low-energy conformationso ft he CAME monomer with one or two additional water molecules in various positions. Four CAME-water complexesw ith the following compositionw ere found to be stable and constituteenergy minima: The structure of these complexes is displayed in Figure 2, whereas the relative electronic energies are listed in Table 1. Complex 4w as discarded for furthera nalysis owing to its high relative energy.O ther water complexes,f or example, with structure III seem unlikely to occur owing to unfavorable stability.W ater complexes 1a nd 3c onsist of structures Ia nd II, where one water molecule forms two hydrogen bonds, one to the hydroxyl group and one to the carbonyl oxygen atom, giving rise to ad istortedc yclic arrangement. In water complex 2a nd complex 4, two water molecules form ac yclic structure with hydrogen bonds to the hydroxyl and the carbonyl O. For each water molecule, the oxygen ando ne hydrogen atom participate in the hydrogen-bond network, whereas the other Ha tomp oints outwards. Notably,t he water molecules are not in plane with the mirror plane of CAME but are out of plane. Table 1. Relative electronic energies of the threel ow-energyc onformers of the CAME dimers (dimer 1-6), CAME monomers( structure I, II, III) as well as water complexes ((H 2 O) x -complex 1-4). Structuresw ere fully optimizedw ith MP2/aug-cc-pVTZ exploiting the molecular symmetry, coupled cluster calculations (CCSD(T)-F12/cc-pVTZ-F12)a re single points on the MP2/aug-cc-pVTZ optimized structures. Energies are giveni n kJ mol À1 .F ree energies are calculated for T = 210 Ka nd p = 2 10 À5 mbar.
[a] Owing to unfavorable energy,f requencies not calculated. Chem

Calculated IR spectra
Infrareds pectra of the most abundantm onomers( structures I to III) and dimers (dimers 1t o3 )w ere calculated in the gas phase by employing ab initio wave function methods (MP2/ aug-cc-pVTZ). All frequencies obtained within the harmonic approximation were scaled by 0.98 as this ensures the least average deviation between experiment and calculation at wavenumbersb elow 2000 cm À1 (see tables in the following sections). The resulting deviationo ft heory and matrix isolation spectroscopy experiment below 2000 cm À1 is 4-8 cm À1 for the monomers and 13-22 and 25 cm À1 for dimers and water complexes. In general,the deviation between calculated and experimentalfrequencies above 2000 cm À1 ,e specially forX-H modes (X = C, O) is higher, in the present case 50-150 cm À1 ,b ecause of pronounced anharmonicities and strong normal mode coupling. [23] Calculated frequencies are plotted together with the experimental data as line spectra (see figuresi nt he following sections). To aid the assignment, spectra of isotopically labeled CAME species completeo ur analyses. They comprise 13 C-CAME, CD 3 -CAME, and OD-CAME.F or the spectral assignment, dimers 1, 2, and 3w ere also calculated as isotopically labeled molecules for CH 3 /CD 3 ,O D/OH, and 13 C/ 12 Cs ubstitution in the CAME molecule. As will be discussed in detail in the following chapter,t he OD-CAMEand 13 C-CAME experiments show impurities of unlabeled CAME.T hus, isotopically mixed dimers need to be considered as well (OD-OH dimers and 13 C-12 Cd imers, no impurities are found in the CD 3 -CAMEs pectrum). This results in additional calculated spectroscopic data for one mixed dimer 1, two mixed dimers 2, ando ne mixed dimer 3( for details see Quantumc hemical setup in the Experimental Section). In the following matrix isolation spectroscopy figures, the intensities of all calculated dimer modes are displayed with onetenth of the initially calculated intensity,w hich is required for an appropriate match with the experimental data. For isotopically mixed dimers, the intensities of the calculated normal modes are scaled by 1/20.F inally,I Rs pectrao ft hree CAMEwater complexes were calculated. However,n oi sotopically labeled water complexes are shown in this work, as this would go beyond the scope of the discussion. The peak intensities of these complexes are displayed with one-tenth of the initially calculated intensity (see Figure 6), which resultsi na na ppropriate match with the experimental data.

Experiments on the pure solid state:V ariation of the solvents
The preparation of CAME under cryo-conditions as ap ure solid and subsequentm atrix isolation was briefly described previously in reference [12],b ut here we want to provideashort discussion of the reactionp athway and the stabilityo ft he isolated product.I nt he present work, we divide the preparation into two steps:s tep (1) esterification of KHCO 3 ,f ormation of the hemiesters alt K[O 2 COCH 3 ]i ns olvent 1a nd step (2) protonation to CAME in solvent 2. This is illustrated and discussed in detail in the Supporting Information in Figure S1 'reaction pathway' and the corresponding FTIR spectrao fs olid K[O 2 COCH 3 ]/HO 2 COCH 3 in Figure S2. Table S1 (in the Supporting Information) lists the observed IR frequencies of the hemiester and its potassium salt, providing ac omparison of the K[O 2 COCH 3 ]s pectrum with the spectrum in the work of Behrendt et al. [24] and ar einterpretation of the modes of solid CAME compared with the former 'alpha-carbonica cid' assignment of HHM. [10b] The newly assigned modes of CAME are highlighted in red in Ta ble S1 (in the Supporting Information) and Ta ble 2. To underline our band assignment in Ta ble S1, Figure S3 (in the Supporting Information) shows the spectrum of solid CAME compared with the MP2/aug-cc-pVTZ calculated gas phase vibrational bands. Strong couplingb etween molecules and the crystal field severely broadens and shifts all bands in the spectra of solid CAME. Still, plotting the calculated in vacuo spectra together with the experimentalF TIR spectrum of solid CAME in Figure S3 b( comparison to calculated line spectra of CAME dimers, in the Supporting Information) strongly supports the reassignment provided in Ta ble S1 (in the Supporting Information) and Table 2. We are aware of the fact that isolated dimers also do not accountf or the crystal field properly.H owever,t he cyclic dimer motif is energetically favorable and the improved match of the dimer spectra with the solid-state spectra of CAME suggestsd imers as basic building blocks of the solid.
Ta rgeted variation of solvents during preparation in steps (1) and (2) clearly illustrates the reactionp athway and the stability of the monomethyl ester of carbonic acid.

Variation of solvents in step (1)
Using different solvents for the dissolution of KHCO 3 with the same experimental procedure, that is, by using water,m ethanol, or ethanol as the solventf or step (1) and subsequent uniform protonation with HCl in water in step (2), leads to the formation of b-H 2 CO 3 ,C AME, and CAEE, [15] respectively.A cid-catalyzed hydrolysis, however,d oes not take place under cryo-conditions as shown previously in detail for carbonic acid ethyl ester (CAEE). [15] Figure 3p rovides ac omparison of the spectra of solid CAME with solid b-H 2 CO 3 and CAEE after the exact same preparation procedure for all three solids with the only exception of varying the solvent in preparation step (1). Ta ble 2l ists the bands of all three speciesi ncluding their vibrational assignment. This direct comparison demonstrates that the formation and isolation of the hemiesters (CAME and CAEE)i ss uccessful with no hydrolysis to H 2 CO 3 occurring.
Ac lear distinction of the spectra of KHCO 3 (by using water in the first step) and K[O 2 COCH 3 ]( by using methanoli nt he first step) is possible, which is supported by comparison with the work of Nakamotoe tal. [25] (see Figure S4 and Ta ble S2 in the Supporting Information). Note that detailed discussions of b-H 2 CO 3 and CAEE,i ncluding also matrix isolation,c an be found in our earlierw ork. [12,15,19] Chem. Eur.J.2020, 26 (2) Althoughv ariation of the solventi ns tep (1) has an impact on the reactionp roduct, the variation of the solventi ns tep (2) has no impact.I nF igure 4, spectra of b-H 2 CO 3 and CAME are shown, which were recorded after acidification and solvent evaporation. No matter which acidic solution( aqueousH Br, methanolic HCl, or ethanolicH Cl) was used in step (2), the protonation of KHCO 3 leads to the same type of spectrum originating from b-H 2 CO 3 (Figure 4a-c).
Similarly,p rotonation of K[O 2 COCH 3 ]l eads to CAMEr egardless of whether aqueous, methanolic, or ethanolicH Cl is used as the solvent. All spectra in Figure 4d-f correspond to the CAME spectrums hown in Figure S2 b( in the Supporting Information).

Experiments on the pure solid state:Isotope labeling
To distinguishu nequivocally between the interpretationa sa-H 2 CO 3 and CAME, isotopically labeled solvents were used in both preparation steps as well as isotopically labeled KHCO 3 to produce HO 2 COCD 3 ,D O 2 COCH 3 ,a nd HO 2 13 COCH 3 .Aprecise as- [a] n s and n as :s ymmetric and asymmetric stretching modes; d ip and d oop :i n-plane ando ut-of-plane bending modes; d s and d as :s ymmetric and asymmetric bendingm odes.  signmento fa ll isotopically labeled CAME molecules is shown for the matrix isolation spectra,b ut for the FTIR spectra of the solid, CH 3 /CD 3 exchange is presented as an example, where CD 3 OH was used as the solvent for KHCO 3 instead of CH 3 OH. Ad iscussion of the spectral shifts of the solid precursor salts of CAME andC D 3 -CAME is provided in the Supporting Information together with Figure S5. Figure 5s hows the spectra of solid CAME (a) and CD 3 -CAME (b) togetherw ith the calculated line spectra of the dimers.T he color code in Figure5 and Ta ble 3i su sed to visualize the peak shift/splitting upon CH 3 / CD 3 exchange:m odes that are pure CH 3 /CD 3 modes are labeled in red, bands that involveC H 3 /CD 3 modesc oupled to other modesa re labeled in orange, and gray indicatesm odes that are not affected by isotopic labeling. Ta ble3shows acomparisonoft he frequencies and its isotopically labeled analogue including the H/D ratio, perfectly matching the assignment and the shifts predicted by calculatedm odes. In general, substitutiono fadeuterium atom for ah ydrogen atom redshifts the pure stretching modes by af actor of approximately p 2. . Am ode at 610 cm À1 ,w hich appears for CD 3 -CAME, can be assigned as a d(CD 3 )m ode. Typical bands that are unaffected by isotopic labeling are, for example, n(OH), n(C=O), or d oop (CO 3 ). These band shifts/splittings induced by using d 3 -MeOH instead of MeOH clearly demonstrate the presence of the methyl group in the product and its origin from the solvent.
Note that according to the criteria established by Winkel et al., [11] CD 3 -CAMEi sr ather amorphous whereas CAME is mainly crystalline. This can also be recognized by comparing the FWHM of the bands.T hus,s trictly speaking the H/D ratios listed in Table 3n ot only include the shifts induced by the isotopic labeling (CH 3 to CD 3 ), but also small shifts related to the crystallization,w hich only took place for CAME (Figure 5a), but not for CD 3 -CAME (Figure 5b).
The new assignment of solid-state spectra supported by calculated line spectra and isotopicl abeling in this work rules out the interpretation of the spectrum in Figure S2 b( in the Supporting Information) on the basis of a-H 2 CO 3 .T he high degree of similarity of the spectrum in Figure S2 b( in the Supporting Information) and the spectrum reported by HHM in their Figure 7i nr ef.
[10b] suggests that their interpretation on the basis of a-H 2 CO 3 is incorrect. In particular,t he presence of the bands assigned as CH modes in Ta ble S1 (most notably bands at 1447 cm À1 and 1200 cm À1 ,i nt he Supporting Information) clearlys peaks in favor of their product being CAME as well. This suggests that the sandwich technique, skipping the first evaporation of the solvent, used by HHM also involves K[O 2 COCH 3 ]a sa ni ntermediate in solution.T he fact that all modesp ertaining to the methyl group are of low intensitye xplains why HHM had overlooked its presence and rather considered the presence of disordered carbonic acid or impurities as the origin for these weak bands.

Matrix isolation: Trapping in argon
Solid CAME is evaporated at 210 Ki nt he matrix isolation setup,a nd the molecules above the solid are trapped in an Ar matrixa t1 0K.T he following figures show the results for CAME as well as its isotopically labeled isotopologues:C D 3 -CAME, OD-CAME,a nd 13 C-CAME.I na ddition, difference spectra before  and after UV irradiation of the molecules trapped in the matrix below 10 Ka re shown to corroborate the assignment in section 2.5. The assignment of the experimental spectra is supported by MP2/aug-cc-pVTZ calculatedl ine spectra, displayed together with the matrix isolation spectra, where as caling factor of 0.98 was used. Af ull assignment is possible by considering not only monomer structures but also dimers and water complexes (see also the discussion in section2.1). Figure 6s hows the matrix spectrum of CAME in the region of 4000-600 cm À1 .A part from CAME monomers andd imers, other species identified in the spectrum are H 2 Oa nd CO 2 .

CAME-monomers, dimers, and watercomplexes
These are labeled with *a nd #i nF igure 6a.T hey may either be products of CAME decomposition or enter the matrix through the transfer procedure and/orl eaks in the chamber. The experimentsu sing 13 Cs ubstitution (see below) indicate that CO 2 in fact originates from the decomposition pathway.I n addition, at race amount of methanoli si dentified as an impurity based on the observation of the n(C-O) mode at 1034/ 1029 cm À1 and very weak n(CÀH) modes at 2956, 2929, 2921, 2913, 2909, 2848, and 2055 (2 1034) cm À1 .T hisa ssignment is verifiedb ys eparatem atrix isolation experiments with pure MeOH in Ar (not shown here) andb yc omparison with the literature. [26] For ab etter overview,t he assignmento ft he matrix isolation spectrum of CAME is presented in three steps. For this reason,  Figure S5 a expt., Figure S5 bH /D ratio assign. [24] expt. Figure5a expt. Figure 5b H/D ratio assign. [  ,w ho isolated CAME through ah igh vacuum flash pyrolysis process. [16] Whereas the assignment of monomer structure Ia nd II is based on the observation of all significant bands in the calculation,t he presence of structure III is indicated solely based on the most intenseb and at 1797 cm À1 ,c orresponding to its n(C= O) mode. This assignment is doubtful as other normalm odes, for example, those arising from CAME-water complexes, might be at the origin of the band (see discussion below). In other Figure 6. Matrix isolation spectraa nd MP2/aug-cc-pVTZ calculated spectra of CAME. a) Monomer I = blue, monomer II = red, dimers and water complexes = gray.b )Dimer 1 = blue, dimer 2 = red, dimer 3 = orange, monomersand water complexes = gray.c)Water complex 1 = blue, water complex 2 = red, water complex 3 = orange,m onomersand dimers = gray.F or the calculatedline spectra, a6:1 mixture of monomer Iand II was assumed. words, either structure III is absent in the matrix or if the 1797 cm À1 band originates from it, then it is less abundantb y af actor of at least 10. Bands that are unexplained by CAME monomer conformers are compared with CAME dimer bands in Figure 6b.L ine spectra of three possible dimers, 1, 2, and 3, in Figure 6a re displayed in color with the same intensity (a tenth of the calculated values). The presence of these dimers allows for explanation of the bands in the region between 3050 and 2550 cm À1 and broad bands at approximately 1720, 1480, 1310, and 1090 cm À1 .Adetailed assignment of all dimer signals is shown in Ta ble 5. The most prominent dimer modes are the following: n(OH) (+ n(CH 3 )) at 3017 (dimer 1), 3005 (1), 2929 (2), and 2829 (3) cm À1 , n(C=O) at 1722 (3), 1720 (2), and 1708 (1) cm À1 , d ip (OH) + d ip (CO 3 ) + d(CH 3 )a t1 486 (1, 2, and 3), d ip (OH) and/or d ip (OH) + n(C-OCH 3 )a t1 312 (2, 1) cm À1 ,a nd n(O-CH 3 /O-CD 3 )a t 1092/1079 (1, 2, and 3) cm À1 .T he calculated OH stretching modes of all dimers are strongly shifted owing to the challenges accompanied with the calculation of hydrogen bonds (see also discussion in section2.1.3). Overall, the ratio of dimers to monomers is about1 :9 based on the observed intensities. Assessing the fractionso fd imers 1, 2, and 3i ndividually is not possible because the two most intense bands appear as a broad band rathert han three well-separated peaks.
Even after the assignment of modes to CAMEm onomers and dimers, several bands remainu nexplained. Thus, CAMEwater clusters are considered in Figure 6c,n amely two clusters containing one water molecule ando ne cluster containing two water molecules. Similar to the CAME dimers, the calculated OH stretchingf requencies are also shifted to higher wavenumbers for the CAME-water complexes owing to intermolecular hydrogen bonding. The spectral region above 2550 cm À1 is hard to assign to individual clusters (the only band assigned here is the n(OH) mode of complex 2a t2956 cm À1 ), but with- Table 4. Assignment of IR frequencies of monomer Iand II of CAME,CD 3 -CAME,O D-CAME, and 13 C-CAME (all values in cm À1 ). [ [a] Bold = validated by UV.F actor calc. = 0.98. n = stretching mode, d = bending mode,i ndex s = symmetric, as = asymmetric,i p = in plane, oop = out of plane, sh = shoulder.T ypical isotopic shifts are labeled in red. X th-exp >2000 cm À1 = average deviationt heory-experiment > 2000 cm À1 ;X th-exp <2000 cm À1 = average deviation theory-experiment < 2000 cm À1 .
[b] CAME impurity in OD-CAME.

Isotopologues
CD 3 -CAME( HO 2 COCD 3 )w asp reparedb y using HO-CD 3 as the solventi ns tep (1) of the preparation. 13 C-CAME (HO 2 13 COCH 3 ) was prepared by dissolving KH 13 CO 3 in step (1), and OD-CAME (DO 2 COCH 3 )w as generated by acidifying the salt in step (2) with DCl. Figure 7s eparates the spectrum into four spectral ranges. Each spectralr ange consists of four panels:( a) pure CAME, (b) CD 3 -CAME, (c) OD-CAME, and (d) 13 C-CAME. The matrix isolation experimentso ft hese labeled species (Figure 7) also show the same ratio of 6:1o fm onomer structure I/II. The calculated line spectrum for structure Ia nd structure II in this ratio is again indicated by blue and red lines in all panelsi n Figure 7. Again, the intensity of the dimers is at enth-corroborating the 9:1m onomer/dimer ratio. Dimersa re indicated by gray lines. The interpretationo ft he matrix spectra in Figure 7 reveals impurities of unlabeled CAME in the OD-CAMEa nd 13 C-CAME spectra.Ar atio of 1:1f or OH/OD-CAME and 1:14 for 12 C/ 13 C-CAME is deduced from the intensity ratios of bands shifted upon substitution. Line spectra of the unlabeled monomer structuresare included in Figure7cand dwith the respective intensities.A ll bands found in the pure CAME spectrum as well as in the OD-and 13 C-CAME spectra as an impurity are labeled with *a nd #i nT able4,T ableS3( in the Supporting Information),a nd Ta ble 5. Mixed OD-OH and 13 C-12 Cd imers need to be considered as well. Energetically plausible dimers (same considerations as for all molecules shown, see section2.1.1) are displayed with at wentieth of the calculatedi ntensity.P ossible mixed impurity dimers to be found in the matrix isolation spectrum are dimers 1a nd 3a nd two distinguishable dimer2 complements. Other impurities from water,c arbon dioxide, and methanola re labeled, but do not interferew ith the hemiester bands. No bands of d 3 -methanol (HO-CD 3 )a re found in the spectrum of CD 3 -CAME in Figure 7b and the only detectable peak of methanol in Figure 7c and di sasignal at 1034/ 1029 cm À1 ,representing the n(C-O) mode.
All bands assigned to monomerIandI Ia re listed in Table 4, including H/D shifts and 12 C/ 13 Cs hifts. Ta ble5listsa ll bands of the dimer structures 1, 2, and3 ,i ncluding isotopic shifts. Bands assigned to CH/CD andO H/OD modes shift with at ypical factor of 1.33-1.40, whereas 12 C/ 13 Cs ubstitution shifts the bands by af actor 1.02-1.03. Otherm odes are coupled vibrations of isotopically labeled and unlabeled parts of the molecule and, thus, these signals "are split" or "disappear" and new peaks are observed, which cannot be associatedw ith respec-tive signals in the CAMEs pectrum (see also discussion in sections 2.3 and 2.6.2). It is nevertheless possible to assign these peaks thanks to the excellent accordance with the calculated spectra.
Ad etailed description andr elevant statements that can be made about the spectra shown in Figure 7a re discussed in the Supporting Information. Ta bles S4 and S5 (in the Supporting Information) list all band assignment of mixed OD-OH and 13 C- 12 Cdimers.
To sum up this discussion of the isotopically labeled experiments, an excellent assignment of all spectra could be reached,w hich is in accordance with calculated spectra both in terms of band positions and isotopics hifts. From this interpretation, it is concluded that the cryo-preparation solely leads to formation of carbonic acid monomethyl ester by the proposed mechanism in two steps. It is possible to assign almost all peaks of the full spectral range between 4000a nd 600 cm À1 including dimers and water complexes of the hemiester.A ll shifts induced by isotopic labeling are plausible and match predictionsf rom the calculated data. Impurities of unlabeled CAME in the OD-a nd 13 C-CAME experiments do not compromise the analysis, but rather consideration of mixed dimers confirms the actual peak assignment of the CAMEm atrix isolation spectra.
For all matrix isolation experimentsi ns ections 2.4 and 2.5 discussed together with calculated line spectra considering monomers, dimers, andw ater complexes, av ery comprehensive assignment of almost all signals is possible. Ah andfulo f peaks remain after this assignment, whicha re without exception of low intensity and are mainly found in the region above 2000 cm À1 -the region of various OH and CH 3 modes,especially of dimers and complexes. Complexes that were not considered are, for example, monomer + methanol, methanol + water, dimers including monomer II and water complexes with monomer structure III.

Matrix isolation: UV irradiation
UV irradiation and subsequenta nalysis by using difference spectra (between experiments prior to and after irradiation) was performed to aid the assignment, similar to the case of matrix isolated carbonic acid. [19] UV irradiationc auses isomerization, specifically from monomer structure It os tructure II. In the present case, the energy transmitted by ultraviolet light induces the rotationo ft he CÀOH bond by 1808 with ab arriero f 42.5 kJ mol À1 (see Figure 1b). The barrierb etween monomer structure Iand III is justslightly higher( 45.5 kJ mol À1 for the rotation of the CÀOCH 3 bond by 1808)b ut structure III cannotb e identified after UV irradiation. This might be because the minimum of monomer structure III is 8.8 kJ mol À1 highert han that for structure II and, thus, the back reaction to structure Ih as a lower barrier, making it too fast to observe structure III in the subsequentI Rm easurement. Furthermore, rotationo ft he CÀ OCH 3 group in the argon cage might in fact have ah igher barrier than the one indicated in Figure 1b from in vacuo calculations. It is conceivable that rotation around the CÀOH bond is easier within the cage than that aroundt he CÀOCH 3 bond.
Other than monomer isomerization, UV irradiation does not cause any additional changes-dimers and water complexes remain unaffected. That is, UV irradiation is ideally suitable to discriminate between the two monomeri somersa nd to identify bands that are not caused by either of the two monomer conformers. In the differences pectra in Figure 8, bands pertaining to structure II point upwards andb ands pertaining to structureI point downwards. Bands of other speciesd on ot contributet ot he differences pectra, that is, they show ad ifference of zero. UV irradiation also does not trigger decomposition of the molecules captured in the matrix to CO 2 ,w ater,a nd MeOH. Figure 8s hows the differences pectrao fC AME (a) and 13 C-CAME (b) after 10 min UV irradiation. Bands of monomer structure Ia nd II that are identified by these additional experiments are printed in bold in Ta ble 4. Bands pointing downward per- Figure 7. Matrix isolation spectraa nd MP2/aug-cc-pVTZ calculated spectra of a) CAME,b )CD 3 -CAME,c )OD-CAME, and d) 13 C-CAME. For the calculated line spectra,a6:1m ixture of structures I(blue) and II (red)w as assumed. Isotopic monomeri mpurities were considered with ar atio of OD/OH = 1:1a nd 13 C/ 12 C = 14:1 ( 12 C-CAMEi sl abeled in orange in d)). Dimers and waterc omplexes are labeled in graya nd are scaled down to intensity/10.O D/OH impurity dimers in c) are labeled in orange with an intensity/20 and 13 C/ 12 Ci mpurity dimersi nd )are indicated by dashed gray linesw ith an intensity/20.Bands corresponding to CO 2 andH 2 Oa re labeled with *and #. Calculated frequencies are scaled by af actoro f0 .98.
In Figure8b, 13 C-CAME signals of monomer Ip ointingdownward are observed at 3610, 1735/1733, 1447, 1362, 1175, 896, and 770 cm À1 .S ignals of monomer II pointing upward are detected at 1780/1776, 1452, 1182, and 794 cm À1 .I na ddition to the 13 Cs pecies, also 12 Cm onomers appear in the difference spectrum:m onomer I( downward bands) is detecteda t1 830/ 1826, 1445, and 1327 cm À1 ,a nd monomer II (upward bands) induces signals at 3602, 1791/1787, 1443, 1309, 1172, 1070, 890, and7 62 cm À1 .W ith the exception of the weak n(CH 3 ) modes, all signals assigned to monomerso f 13 C-CAMEi nt he previouse xperiments are confirmed after UV irradiation. Band positions in the UV irradiation experiments match excellently (difference less than 1cm À1 )w ith the ones assigned in Ta ble 4-only for ac ouple of bands there is as hift of 1-2cm À1 .A nu nexplained weak band that appears in the UV irradiation spectra is the band pointing downward at 1797cm À1 . This suggestst hat it mighta rise from structure III rearranging to structureI.H owever,n oo ther bands pertaining to structure III can be identified. In addition, there are two weak bands pointingu pward (1268 and 1339 cm À1 )a nd one band pointing downward (1312 cm À1 ), which cannot be explained based on conformational changes. In Figure 8b,o nly the band at 1242 cm À1 remains unexplained.

Comparisonwith CA and CAEE
FigureS6( in the Supporting Information) shows ad irect comparisono ft he FTIR spectra of matrix isolation experiments of CAME (a), carbonic acid (CA, b), [19] and carbonic acid monoethyl ester (CAEE, c). [15] All three spectra are based on an identical preparation as describedi nt he Experimental Section. The only difference is the choice of the solventu sed in step (1) of the preparation:a )methanol, b) water,a nd c) ethanol. No evidence of non-esterified carbonic acid, which is referredt oa sb-H 2 CO 3 in the literature, [10c, 19] is detected in the FTIR spectra of the solid and in the matrix isolation experiment.
In Figure S6 (in the Supporting Information), impurities of water and carbon dioxideare marked in red. The OH stretching mode,w hich appears at nearly the same wavenumbers for all three species, is colored in blue. The apparent OH and CH vibrations of CAMEa nd CAEE in the spectral region above 2550 cm À1 are highlighted in orange. Most importantly,t wo distinct bands (n(C=O) and d ip (C-OH)) of carbonic acid are marked in green. It is clear that absolutely no signs of these bandsa t1 792/1789 and 1136 cm À1 are observedi nt he spectra of CAME and CAEE. Thati s, the originala ssignment given in Figure 8. Difference spectra after matrix isolation prior to and after UV irradiation. MP2/aug-cc-pVTZ calculated line spectraofC AME and 13 C-CAMEa re included:a)CAME,monomer 1 = blue, monomer 2 = red, and dimers1,2 ,a nd 3 = gray.b) 13 C-CAME,monomer 1 = blue, monomer 2 = red and 13 C-dimers 1, 2, and 3 = gray; 12 Cimpurities of monomer Ia nd II are labeled with dashed lines. The calculated line spectrao fmonomers are shownw ith maximum intensity and lines of dimers are shownw ith intensity/10.Bands corresponding to CO 2 and H 2 Oare labeledw ith *a nd #. Calculated frequencies are scaled by af actor of 0.98.
Chem. Eur.J.2020, 26,285 -305 www.chemeurj.org reference [12] on the basis of carbonic acid needs to be corrected, which is done in Ta ble 6. Reassignments are labeled in red. The modes in reference [12]c oncerning the two OH groups in carbonic acid need to be reassigned as the monomethyl ester of carbonic acidp rovides one OH and an O-CH 3 group. This includes the n s/as (OH) and n as (C(OH) 2 )m odes, whereas the latter is a d s (CH 3 )v ibration of the ester.S ignals around1 270 cm À1 (CAME)w ith af ormer assignment as d ip (COH) belong to water complex 1a nd bandsa t8 08 (CAME) and 784 ( 13 C-CAME) cm À1 are reassignedt od oop (OH) of water complex3and d oop (CO 3 )o fd imers 1, 2, and 3. This rectification is complemented by the very comprehensive assignment of nearly all other bands of the whole FTIR spectra of CAME and its isotopically labeled equivalents,i ncluding dimers and water complexes.B yc omparing CAME and CAEE, we can conclude that also for CAEE some dimers can be trapped in the Ar matrix, for example, broad peaks of n(C=O) or d ip (OH) modes around1 720a nd 1310 cm À1 but dimer bands are much more dominant in the CAME spectrum than in the CAEE spectrum. The correct matrix bands of carbonic acid after sublimation of solid b-CA are given in reference [19].

CH 3 modes of CAME-A retrospectived iscussion
Similar to the discussion of the FTIR spectra of solid CAME in Figure 5, Figure S7 (in the Supporting Information) also shows an alternative representation of calculated line spectra to demonstrate the shifts induced by CH 3 /CD 3 substitution. The calculated line spectra shown include monomer structure Ia nd II (int/6) and dimer structures 1, 2, and 3( int/10) analogous to Figure 6b ut using ad ifferent color scheme:a ll peaks that do not exhibit aC H 3 /CD 3 mode are gray,f or example, n(OH) or n(C=O) at 3724 and 1778 cm À1 (monomer I, wavenumber calculated 0.98), d(CH 3 )o rn(CH 3 )v ibrations are labeled in red, for example, around3 100a nd 1480 cm À1 and bands originating from CH 3 /CD 3 -coupled modes are colored in orange. For a properdisplay,the green line spectra in Figure S7 a(in the Supporting Information) represent water complexes of CAME, but they are not considered for CD 3 -CAME.
It is remarkable that the calculated vibrations of CD 3 -CAME in FigureS7b (in the Supporting Information) marked in orange and red have very low intensities,b eing hardly detected in the experiment.R ed and orange monomer peaks are already very weak in the non-labeled CAMEs pectrum. The sharp and highly resolved matrix isolation spectra and the consideration and calculation of dimers and water complexes allow a much more comprehensive assignment compared with the analysiso ft he FTIR spectra of the solid. This wasacrucial factor for the misinterpretation by HHM and led-together with the low knowns olubility andr eactivity of KHCO 3 in alcohols-to the outdatedc onclusion of different carbonic acid monomers.
Ta ble 4a nd Table 5l ist all assignable peaks of monomer and dimer structures of CAME and CD 3 -CAMEi ncludingH /D shifts but in this context, we want to pick out some characteristic examplest oi llustratet he challenging spectral appearance in the case of CH 3 /CD 3 exchange.
Typical pure CH 3 /CD 3 modes of monomer structure It hat shift with aH /D factor of approximately 1.3-1.4 (see Ta ble 4) that are found as very weaksignals in the experimental spectra are n as (CH 3 /CD 3 )a nd n s (CH 3 /CD 3 ). The n(C-OCH 3 ) + d(CH 3 )m ode of monomer I, detected at 1193/1189 cm À1 for CAME,s plits and bands at 1115a nd 905 cm À1 can be assigned as pure n(C-OCH 3 )a nd d(CH 3 )m odes for CD 3 -CAME.
In the direct comparison of the CAME and CD 3 -CAME matrix isolation spectra in Table 5, no dimer peaks with at ypical H/D shift are found. However,t he successful isotopic labelingi s provenb yt he overall change of the band positions, especially by change/disappearance of CH 3 /CD 3 -coupled modes.
Typical examples are d ip (OH) + d ip (CO 3 ) + d(CH 3 )m odes of dimers 1, 2, and 3around 1500 cm À1 that are split. The coupled modes are detected at 1486 cm À1 in the CAME spectrum and Table 6. Rectification of the matrix isolationband assignment in Bernarde tal.,2011. [a] [12] CAME OD-CAME 13 C-CAMEsc reference [12] new assign. Ar [12] Ar Ar [12] Ar Ar [12] 3 -CAME matrix isolation experiment. n(CH 3 ) + n(OH)a nd n(CH 3 )m odes of dimer 1 at 3163/3158 and 3038/3132 cm À1 (wavenumber calculated 0.98), which are distinct signals in the CAME spectrumd isappear.U pon CH 3 /CD 3 exchange, the CH 3 -part is decoupled and one pure theoreticallyw eak n(OH)r emains in the original wavenumber region at 3160 cm À1 ." New" but weak signals of n(CH 3 )a rise theoretically at 2348, 2326, and 2175 cm À1 .T hese peaks are not detected/resolved in the CD 3 -CAME matrix isolation experiment and have very low intensitiesi nt he calculated spectra as well.

Evaluation of the composition in matrixand solid spectra
Whereasm onomer bands dominate the matrix isolation spectrum, dimer bands are ab etter match fort he broad bands in the solid-state spectrum.F or example, the characteristic n(C= O) and d ip (OH)m odes of monomer Ia nd II are clearly resolved in the spectrum of the Ar matrix above 1750 and at 1182 cm À1 but for the solid-state spectra the broad characteristic signals around1 720, 1480, and 1310 cm À1 resultsf rom n(C=O), The matrix experiments can be explained based on ar atio of 6:1b etween structure Ia nd II, possibly with traces of structure III. The remaining bands can be explained very well based on the presence of cyclic dimers. Specifically,d imers composed of two structure III monomers are identified. This suggestst hat such building blocks might be present in the CAMEpolymorph before sublimation. Based on our thermodynamic calculations, CAME dimers are much more likelyt oo ccur in the matrix than CA dimers,w here no dimers are found. [19] Of course, the CAME dimers could also exist because of favorable kinetics, that is, a low reaction barrier,f or their formation and unfavorable kinetics, that is, ah igh reactionb arrier, for the decomposition into its components.
The 6:1r atio of monomer structure Ia nd structure II deviates from the ratio in thermodynamic equilibrium on the basis of the PES depicted in Figure 1b.T hermodynamically,aratio K of 22:1 would be expected, utilizing the relationship DG = RTln K with as ublimationt emperature of 210 Ka nd the calculated relative free energy difference of 5.4 kJ mol À1 between structure Ia nd II in equilibrium. An equivalent consideration for structure III with af ree energy differenceo f1 4.5 kJ mol À1 leads to ah igh ratio, which indicates that it is unrealistic to detect any signals of structure III in thermodynamic equilibrium. This discrepancy to the experimental ratios was already discussed in detail for as imilar situation for the matrix experiments of the monoethyl ester of carbonic acidi nr eference [15].T he difference might be caused by the rather short flight time of gas-phase molecules from the surface of solid CAME at 210 Kt ob eing trapped in the Ar matrix at 10 K. In our setup, this flight time is about 0.5 ms. Thus, the monomer ratio might be controlled kinetically,n ot thermodynamically. An analogousc alculation as in ref. [15] (see the Supporting Information) yields a6:1 ratio of structure Ia nd II in the matrix, which indicates an originalr atio of 1:2s ublimating from the crystal.

Conclusion
The cryo-preparation and rapid quenching techniquec omplemented with FTIR spectroscopy developed by HHM more than 20 years ago [9] has proven to be av ery suitable tool to prepare and characterize metastable, short-lived intermediates, in par-ticularH 2 CO 3 and its derivatives. Al arge body of significant work, especially on solid H 2 CO 3 ,h as been published, including studies on the polymorphism of H 2 CO 3 .T wo polymorphs of H 2 CO 3 are described in the literature, namely b-H 2 CO 3 [9, 10c, d] isolated from aqueous solutions and a-H 2 CO 3 [10a, b, d] isolated from methanolic solutions.F or both polymorphs, the amorphous phase, the crystalline phase, and the transition wered escribed. [11] Furthermore, the conversion from b-H 2 CO 3 to a-H 2 CO 3 by dissolving b-H 2 CO 3 in MeOH/HCl was reported. [10a] The reassignmento fahemiester rather than ac arbonica cid polymorph in the case of 'a-carbonic acid' was originally proposed in the Ph.D. thesis of our co-author Jürgen Bernard. [8] A similarc onclusion wasm ade later by Reisenauere tal. [16] based on ac omparison of matrix isolation spectra of am ixture of isobutene/CAME [16] and matrix isolation spectrao fa-H 2 CO 3 . [12] In the present study,t he rectification of the assignment of FTIR spectra of solid and matrix isolated formerly termed' a-H 2 CO 3 ' is built on four pillars:v ariationo fs olvents during different preparation steps, isotopics hifts in the solid-state spectra, nearly complete clarificationo fa ll bands between 4000-600 cm À1 of matrix isolation IR spectra supported by MP2/ aug-cc-pVTZcalculations, and isotopic labeling and forced conversion of conformers by irradiation experiments of the molecules trapped in the matrix. We used as imilarp reparation technique as HHM by dissolving KHCO 3 in absolute methanol followed by cryo-preparation steps and acidification. The resultingp roduct is the monomethyl ester of carbonic acid (CAME).
The variation of solvents during preparation proves the high reproducibility,p urity,a nd stability of either carbonic acid, CAME, or CAEE.I ti sd ecisive which solvent is used in the first preparation step, that is, in water b-H 2 CO 3 forms,i nM eOH CAME forms, and in EtOH CAEE forms. [15] Acid-catalyzed hydrolysis and formation of CA do not take place under thesec onditions.
By using CD 3 -labeled MeOH as as olvent, the FTIR spectra of the solid product reveal that the O-CD 3 group is transferred from MeOH to the salt and ultimately also to the product, which is CD 3 -CAMEb ut not a-H 2 CO 3 .T he presence of the methyl group in the product is evidenced by H/D ratios of 1.3-1.4 of the related bands in the spectra.T hese bands are now reassigned [10b] as CH/CD modes. We find no evidenceo fC Ai n the solid-state spectra of CAME.
IR spectra obtained after evaporating the solid at 210 Ka nd trapping the vapor in an argon matrixa t1 0Kcan also be reassignedo nt he basis of CAME in contrast to the former assignment as carbonic acid monomers and dimers. [12] The assignment relies on isotope substitution experiments, including

Experimental Section Preparation experimenti nt he solid state
The preparation of the starting material was done from methanolic solution of potassium bicarbonate (KHCO 3 ,S igma-Aldrich, > 99.5 %). Alkali bicarbonates and carbonates are barely soluble in methanol, whereas they can easily be dissolved in water.K HCO 3 was stirred in CH 3 OH (Sigma-Aldrich;m ethanol CHROMASOLV,f or HPLC, ! 99.9 %) or CD 3 OH (Sigma-Aldrich;m ethanol-D3, 99.8 atom %D ). KHCO 3 has ap K a of 10.25 and CH 3 OH has ap K a of 15.5. [7,29] Complementary experiments were done by using doubly distilled, deionized H 2 Oo ra bsolute ethanol as solvents. The solutions were nebulized in N 2 carrier gas by means of an air brush pistol (Harder &S teenbeck;m odel grafo or infinity) and introduced into av acuum chamber ( % 10 À7 mbar) through an aperture (500 mm). Upon impact of the aerosol on acryoplate at liquid nitrogen temperature (T = 78 K), alayer of glassy solution forms. The solution droplets (> 10 mmi nd iameter) are immobilized almost instantaneously at cooling rates up to 10 5 Ks À1 . [30] IR transparent windows (cesium iodide, CsI, or silicon, Si, windows) serve as the cryoplate. [9, 10d] After deposition of the bicarbonate solution, the cryoplate was heated in vacuo to 290 K, which results in evaporation of the solvent and as olid precipitate remaining on the cryoplate. This step was not part of the protocol employed by HHM. [9][10] The solid precipitate was later protonated by depositing al ayer of glassy 1.5 m HCl (diluted from Supelco methanolic HCl 3N or hydrochloric acid HCl 37 %) solution, either in water,m ethanol, or ethanol at T % 80 Ka nd subsequent heating. Heating triggers diffusive mixing of the acid with the base, acid-base reaction, and finally evaporation of solvent. The reactions were monitored in situ by Fourier transform infrared (FTIR) spectroscopy by using aV arian Excalibur 3100, in which the beam of light passes through optical windows (KBr) into the vacuum chamber,t hrough the thin film sample and out of the vacuum chamber to the detector.F TIR spectra were recorded with ar esolution of 4cm À1 and by accumulating 100 scans. The chamber was pumped to ab ase pressure of 10 À7 mbar by using an oilfree scroll pump (Varian Triscroll) and at urbomolecular pump (Leybold Turbovac 361). To keep the base pressure after the injection of the nebulized solutions in nitrogen as carrier gas low,acryopump (Leybold RW 6000 compressor unit and RGD 1245 cold head) was located inside the vacuum chamber and kept at 11 K. At this temperature, the carrier gas condenses as as olid on the cryopump.

Experimenti nt he gas phase
After preparation of the pure solid in the laboratory in Innsbruck, the cryoplate containing the CAME film was removed from the vacuum chamber,i mmersed in liquid nitrogen, and transported to Vienna for matrix isolation experiments. In Vienna, an ultrahighvacuum chamber was used, which was previously employed for successfully isolating reactive species such as halogen oxides [31] or carbonic acid. [32] FTIR spectra were recorded by using aB ruker Vertex 80v,w hich offers an evacuated optical path (2 mbar) and a resolution of 0.2 cm À1 at which 1024 scans were accumulated. The details of the matrix isolation procedure can be found in reference [12]. UV irradiation experiments were performed by using an Hg(Xe) arc light source operated at 300 W( solar simulator AM1.5G). While irradiating for 10 min, the matrix temperature was kept below 10 K. Subsequent to UV irradiation, FTIR spectra were collected and analyzed as difference spectra compared with measurements prior to irradiation. The sharp bands observed in matrix isolation spectra with av ery high resolution are perfect for monitoring isotope shifts and ad irect comparison with calculated gas-phase spectra.
Just like for mass-spectrometric techniques, evaporation of the solid is required for matrix isolation spectroscopy.H owever,i onization and ionization-induced fragmentation are not an issue in the matrix isolation technique as the neutral molecules are landed in the matrix, by contrast to mass-spectrometric techniques. In other words, as carbonic acid and carbonic acid esters very readily fragment upon ionization, both of them show fragments at the same m/z ratio, and this technique is not suitable for discriminating between CAME and carbonic acid. Also, the presence or absence of the methyl group in the mass spectrum cannot be reliably used to discriminate between the two as traces of methanol may be present in carbonic acid, for example, as inclusion. By contrast, the vibrational bands are shifted between the two molecules, and so matrix isolation spectroscopy is the better analytic technique to assess the purity of the sample. X-ray diffraction would in principle be suitable as well. However,t he thin film nature of the samples and their in situ formation in av acuum chamber do not allow for aready investigation by diffraction.

Quantum chemicals etup
The low-energy conformations of carbonic acid methyl ester (CAME) were determined by second-order Møller-Plesset perturbation theory (MP2) [33] by using augmented correlation consistent basis sets by Dunning and co-workers of triple-zeta quality (augcc-pVTZ). [34] To estimate the kinetics of conformational interconversion, ar elaxed potential energy surface (PES) scan was performed for the two dihedral angels that need to be rotated to interconvert the four conformers. The O = C-O-H and O = C-O-CH 3 dihedrals were scanned at 158 intervals from 08 to 3458 and 08 to 1808,r espectively.At otal of 312 points were calculated at MP2/aug-cc-pVTZ level of theory and the resulting PES is shown in Figure 1b. The PES energies do not include zero-point corrections, which were shown to be negligible previously for CAEE [15] and are also negligible here. All of these calculations were done in C 1 symmetry and performed by using Gaussian 09 Rev.C01. [35] The three low-energy monomer conformers were combined to construct six dimer structures, which were structure optimized by using MP2/aug-cc-pVTZ. For these calculations, molecular symmetries had to be exploited. For consistency,t he three low-energy monomers of CAME were re-optimized with MP2/aug-cc-pVTZ when applying their molecular symmetry group. These calculations were performed with Turbomole 7.1.1. [36] Energy differences between the fully symmetric and C 1 symmetric molecules are negligible. Dimerization energies and interaction energies were calculated on the MP2/aug-cc-pVTZ optimized structures as single points by using the explicitly correlated coupled cluster variant CCSD(T)-F12 [37] with density fitting-as implemented in Molpro 2015.1 [38] in combination with at riple-zeta basis-set (cc-pVTZ-F12). [39] Thermal as well as zero-point energy corrections to obtain dimerization and interaction free energies at temperatures between 180 and 220 Kw ere obtained by MP2/aug-cc-pVTZ. An umber of CAMEwater complexes with one or more water molecules at different positions were subjected to quantum chemical structure optimization. However,o nly those up to 20 kJ mol À1 were used for the further studies. Normal modes and infrared intensities for annotating the measured spectra were obtained for three distinct conformations of CAME, for three CAME-water clusters, and for three low-energy CAME dimers (dimers 1-3) by using the harmonic approximation at the MP2/aug-cc-pVTZ level of theory.F or each conformation, only real frequencies were obtained, confirming all investigated structures to be minimum energy conformations. Thermal and zeropoint energy corrections were calculated at 210 K( the experimental temperature) and p = 2 10 À5 mbar,w ith frequencies scaled by afactor of 0.9792 as suggested by Kesharwani et al. [40] For each of the three CAME low-energy conformers, isotopic shifts for the three isotopically labeled variants 13 C-CAME, CD 3 -CAME, and OD-CAME were extracted by transformation of the reduced masses in the mass-weighted Hessian matrix, which is computed at the MP2/aug-cc-pVTZ level of theory by numerical second derivatives. For the three low-energy conformations of the CAME dimers, shifts for all singly and double isotopically labeled species were obtained, resulting in the following variants:d imer 1a nd dimer 3: 13 C-13 C, 13 C-12 C, CD 3 -CD 3 ,C D 3 -CH 3 ,O D-OD, OD-OH (six isotopically labeled species for each dimer);d imer 2: 13 C-13 C, 13 C-12 C, 12 C-13 C, CD 3 -CD 3 ,C D 3 -CH 3 ,C H 3 -CD 3 ,O D-OD, OD-OH, OH-OD (nine isotopically labeled species). Intensities of the isotopically labeled species were scaled according to their experimentally determined abundance in the sample. All these calculations were performed with Molpro 2015.1. [38] This extensive analysis allowed us to identify the various convoluted signals observed in the experimentally obtained spectra. Vibrational spectra were scaled by the factor of 0.98 to match the experimentally observed C-O stretch mode. According to the "Computational Chemistry Comparison and Benchmark DataBase", the best scaling factor to be used for vibrations calculated by using MP2/ aug-cc-pVTZ is 0.953 AE 0.033 as determined from ac omparison of 358 vibrations in 117m olecules. [32] Structures were visualized with VMD. [41]