Israel
Cabasso
,
Mingyu
Li
and
Youxin
Yuan
*
The Michael Szwarc Polymer Research Institute, Chemistry Department, State University of New York-esf, Syracuse, NY 13210. E-mail: yxyuan@syr.edu
First published on 24th August 2012
Reported is the electrosynthesis of a novel liquid compound, 5-acetyl-2,9-decanedione (TK) [n35d = 1.4524 and Tm < −70 °C, Tb > 200 °C], the molecule has unique hierarchic “Y” structure with three carbonyl groups anchored at the end of each arm. The TK is formed though the electrochemical reformation of bio-derived levulinic acid in MeOH/H2O system. The possible mechanism of formation of the TK via a side reaction of Kolbe electrolysis has been proposed. Theoretical analysis of the TK coordination with lithium ions, forming bidentate Li+ (TK)Li-O1O2, or Li+ (TK)Li-O2O3, or tridentate Li+ (TK)Li-O1O2O3 complexes has been conducted. The results suggest that one TK can possess two Li+ ions, thus, it might be a competitive solvent for lithium ion batteries.
The 5-acetyl-2,9-decanedione being liquid, with low melting point and high boiling temperature, has a unique hierarchic “Y” structure with three carbonyl groups anchored at the end of each arm. The preliminary results of theoretical computation of solvation of the 5-acetyl-2,9-decanedione with lithium ion provide insight into the electronic structure and thermodynamic properties such as the standards enthalpy, entropy and Gibbs energy change, and suggest that TK could be a potential solvent in lithium-ion battery technology.
We use electrochemical synthesis to produce such an unconventional compound as TK because the synthesis method is a one-pot reaction that is inherently “clean”, and consists of an unlimited supply of electrons (via electricity) as a “reagent”.14,15 As such, the old Kolbe electroorganic reaction16 which is highly efficient for decarboxylating alkanoic acids due to its specificity and multilateral flexibility. The reaction is initiated by electro-oxidation of the carboxylate anion, that decomposes into an alkyl radical (i.e., one-electron oxidation), and CO2. Consequently, two radicals combine to form a dimer, or by losing another electron form a carbenium ion (i.e., two-electron oxidation). The relative ratio of radicals and carbenium species depends on the electrode materials, the polarity of the solvent, the chemical structure of carboxylate acids and other electrolytic parameters.17–19 Irrespective of the mechanism both intermediates yield longer carbon chain products through an alkyl radical coupling, or via esterification of the carbenium ions with carboxylate anions.
Scheme 1 Electrochemical reformation of levulinic acid. |
The entire product mixture was separated using silica gel (Silica Gel 60, EM Science) column and hexane/ethyl acetate (2:1 to 1:2 by volume) mixture as the eluent and analyzed by GC-MS. The yield was determined by GC/FID using naphthalene as the internal standard. Each product was identified using high-resolution mass spectrometry (HRMS, ESI-TOF) and NMR spectroscopy.
The geometries of all the stationary points including the Li+ complexes, Li+ and the ligands were optimized at restricted Hartree–Fock (RHF) level of theory with 6-31G basis set. All the stationary points were characterized by their positive harmonic vibrational frequencies as minima. The entropies, the enthalpies and the Gibbs free energies were calculated at RHF/6-31G as well. All the ab initio molecular orbital computation were performed with the GAUSSIAN 09 program package.20
The high-resolution mass spectrum found that the molecular mass 212.1410 Dalton is in agreement with the theoretical value of 212.1412 Dalton for the molecular formula of C12H20O3.
Major MS fragments (m/z) (Fig. 1): 194[M–H2O]+; 169[M–CH3CO]+; 141[CH3COCH2CH2CH2+CHCOCH3]; 127[CH3COCH2CH2+CHCOCH3]; 85[CH3COCH2CH2CH2+]; 71 [CH3COCH2CH2+]; 43[CH3CO+]. One and two dimensional 1H and 13C NMR experiments were used to deduce the structure of the proposed compound 5-acetyl-2,9-decanedione as follows: the 1D 1H NMR spectrum (Fig. 2) shows one methine, five methylenes, and three methyls. There are three methyl singlets that appear in the 2.08–2.11 ppm region which indicate three acetyl groups, other patterns were not first order and therefore analyzed by 2D COSY data.
Fig. 1 Mass spectrum of 5-acetyl-2,9-decanedione. |
Fig. 2 1H NMR spectrum of 5-acetyl-2,9-decanedione in CDCl3. |
13C NMR, δ (ppm) consists of three keto-carbons 211.71 (C11), 208.21(C9) and 207.89(C2); nine aliphatic carbons which can be attributed to one methine at 51.80 (C5), five methylene at 43.23(C8), 40.66 (C3), 30.52 (C6), 24.55 (C4), 21.23 (C7), and three methyls at 29.87 (C1), 29.91 (C10), 28.66 (C12) (Table 1). A detailed structure determination of 5-acetyl-2,9-decanedione was carried out by using 2D NMR spectroscopy (Fig. 3). The 1H–13C HSQC-DEPT edited NMR technique correlates the direct 1H–13C spin couplings in the molecule, as well as giving information on how many hydrogens are attached to a given carbon. This technique is very useful as the proposed structure has a chiral center at position C5 (δ 51.80 ppm in the 13C-NMR) that renders the neighboring methylene protons H-4 and H-6 diastereotopic, splitting the corresponding 1H-NMR signals into two paired multiplets; one pair at δ 1.66–1.73 ppm and δ 1.77–1.85 ppm (m, 2H, H-4) and the other pair at δ 1.31–1.39 ppm and δ 1.52–1.60 ppm (m, 2H, H-6). The 1H–1H COSY NMR confirms the H-5 and H-4, H-5 and H-6 couplings for the assumed structure. The linear six-carbon network C3 to C8 in the assumed structure were evidenced by couplings of the H-3 to H-4, H-4 to H-5, H-5 to H-6, H-6 to H-7, and H-7 to H-8 in the 1H-1H COSY spectra. These assignments were used to assign the 13C signals with the 2D-HSQC-DEPT data. Table 1 shows all assignments that were made from the 2D data. The remaining structural assignments are the three acetyl groups, the HMBC data confirmed the placement of these groups as follows: the CH3-1 correlates with the keto C2 and methylene C3; CH3-12 correlates with keto C11 and methine C5; and the CH3-10 correlates with the keto C9 and methylene C8. These three methyl group correlations support the connection to the carbonyl group as drawn in Scheme 1. Other long range HMBC correlations are listed in Table 1 which further supports the proposed structure.
Fig. 3 Characterization of 5-acetyl-2,9-decanedione with 2D-NMR in CDCl3: (a) 1H–13C HSQC-DEPT (black dots –CH and CH3; red dots –CH2); (b) 1H–1H COSY and (c) 1H–13C HMBC. |
Scheme 2 Proposed synthetic pathways toward 5-acetyl-2,9,-decanedione. |
No. | δ C | δ H (mult.) | HMBC |
---|---|---|---|
1 | 29.87 | 2.08(s) | C-2, 3 |
2 | 207.89 | ||
3 | 40.66 | 2.29–2.40(m) | C-2, 4, 5 |
4a | 24.55 | 1.66–1.73(m) | C-2, 3 ,5, 6, 11 |
4b | 1.77–1.85(m) | ||
5 | 51.80 | 2.39–2.47(m) | C-3, 4, 6, 7, 11 |
6a | 30.52 | 1.31–1.39(m) | C-4, 5, 7, 8, 11 |
6b | 1.52–1.60(m) | ||
7 | 21.23 | 1.43–1.52(m) | C-5, 6, 8, 9 |
8 | 43.23 | 2.37–2.42(m) | C-6, 7, 9 |
9 | 208.21 | ||
10 | 29.91 | 2.09(s) | C-8, 9 |
11 | 211.71 | ||
12 | 28.66 | 2.11(s) | C-5, 11 |
Scheme 2 illustrates the possible reaction pathway that leads to the formation of 5-acetyl-2,9-decanedione. The electrochemical reformation of levulinic acids starts with an initial discharge of levulinate ions at the anode, followed by decarboxylation forming 3-oxobutyl radicals. Subsequently, the recombination of the resulting radicals forms the Kolbe dimer (DK). However, 3-oxobutyl radicals can further undergo electro-oxidation at the anode, losing another electron to form the corresponding 3-oxobutyl carbenium ions. This reaction is favored in an aqueous medium due to higher ionization power21,22 and Brønsted base effect of H2O molecules. With the assistance of water molecules, a transient olefin product 3-buten-2-one (BO) would be formed through E1 elimination of a proton from the initial carbenium ion. Although other carbenium-derived products such as alcohol and ester are also possible, none of them were detected in this study. Most likely this is due to the electron-withdrawing induction effect of the keto group on the 3-oxobutyl cation. This induction increases the acidity of α-H and hence facilitates the E1 elimination products. The intermediate (BO) continuously reacts with the 3-oxobutyl radicals at the double bond from either the α or β position to form 5-acetyl-2,9-decanedione. It should be noted that the intermediate (BO) in the proposed synthetic pathways could not be isolated in the actual reaction mixture, apparently due to the fast reaction in radical addition step.
Thus, in order to verify the possibility of such mechanism, a commercially available product 3-buten-2-one was added during the electrochemical reformation of LA (with anhydrous MeOH as the solvent) and product distributions is illustrated in Scheme 3A. Aside from homo-coupling 2,7-octanedione (85.3% relative yield), the TK was found in a considerable yield (∼14.7%), as predicted by the proposed mechanism (Scheme 2). In fact, this reaction mechanism appears to be rather general and represents a novel approach for the synthesis of unique hierarchy “Y” structures (e.g., TK, Y-mer 1, and Y-mer 2) as we demonstrated by the electrochemical reformation of valeric acid (VA) in the presence of 3-buten-2-one (Scheme 3B), and by electrochemical co-reformation of LA and VA (Scheme 3C).
Scheme 3 Electrolysis of 3-buten-2-one with LA (A), VA (B) in MeOH, and LA with VA (C) in H2O. |
The solvent composition is shown to affect the yield and ratio of DK and TK. A list of the product compositions and yields as a function of various H2O/MeOH ratios is given in Table 2. In general, it is observed that high water content has a negative impact on the current efficiency and the yield of DK. This is most likely due to a competitive electro-oxidation of hydroxide ions [HO−] and/or H2O molecules at the anode. For example, the yield of DK decreased from about 70% to roughly 35%, and the current efficiency decreased from about 84% to about 35%, as the water content increased to 77 mol%. The formation of TK showed the opposite trend, the ratio of TK/DK increased in proportion to the water content.
[H2O]/[MeOH] (mol/mol) | Yield%a | TK/DKc | ||
---|---|---|---|---|
(DK) | (TK) | Total Yieldb | ||
a Yield was determined by GC/FID using naphthalene as the internal standard. b Current efficiency (ec %) in brackets. c Yield ratio of product (TK) to (DK). | ||||
MeOH | 68.3 | — | 68.3[83.3] | 0 |
20/80 | 68.2 | 5.1 | 73.3[83.6] | 0.07 |
35/65 | 57.3 | 8.6 | 65.9[48.1] | 0.15 |
50/50 | 42.5 | 11.0 | 53.5[39.3] | 0.26 |
70/30 | 41.9 | 14.4 | 56.3[35.3] | 0.39 |
77/23 | 34.9 | 20.3 | 55.2[34.2] | 0.58 |
84/16 | 32.7 | 21.6 | 54.2[35.3] | 0.64 |
90/10 | 32.8 | 20.6 | 53.3[35.3] | 0.66 |
95/5 | 33.1 | 22.1 | 55.2[33.4] | 0.68 |
H2O | 24.4 | 18.4 | 42.8[31.9] | 0.75 |
Fig. 4 Optimized structure of (a) TK; (b) Li+(TK)Li–O1O2; (c) Li+(TK)Li–O2O3; (d) Li+(TK)Li–O1O2O3; (e) (Li+)2(TK) Li–O1O2, Li–O3; (f) (Li+)2(TK)Li–O2O3, Li–O1; (g) Li+(TK)Li–O1O3(TK)′Li–O2′O3′; (h) Li+(TK)Li–O1O2(TK)′Li–O1′O2′. |
Species | Distance of Li-carbonyl (Å) | Distance of CO (Å) | ||||
---|---|---|---|---|---|---|
Li–O1 | Li-O2 | Li–O3 | CO1 | CO2 | CO3 | |
a Calculated from ab initio molecular orbital computation using RHF/6-31G. | ||||||
TK (4a) | - | - | - | 1.221 | 1.219 | 1.219 |
Li+(TK)Li–O1O2 (4b) | 1.807 | 1.826 | - | 1.238 | 1.237 | 1.219 |
Li+(TK)Li–O2O3(4c) | - | 1.807 | 1.814 | 1.218 | 1.241 | 1.238 |
Li+((TK)Li–O1O2O3(4d) | 1.890 | 1.860 | 1.878 | 1.230 | 1.230 | 1.232 |
(Li+)2(TK) Li–O1O2, Li–O3(4e) | 1.825 | 1.821 | 1.760 | 1.234 | 1.235 | 1.243 |
(Li+)2(TK) Li–O2O3, Li–O1(4f) | 1.769 | 1.826 | 1.813 | 1.245 | 1.237 | 1.237 |
Li+ (TK) Li–O1O3(TK)′Li–O2′O3′(4g) | 2.007 (-)′ | - (1.882)′ | 1.957 (1.983)′ | 1.230 (1.222)′ | 1.222 (1.233)′ | 1.229 (1.230)′ |
Li+ (TK) Li–O1O2(TK)′Li–O1′O2′(4h) | 1.925 (1.925)′ | 1.958 (1.958)′ | - (-)′ | 1.229 (1.229)′ | 1.229 (1.229)′ | 1.219 (1.219)′ |
The atomic net charge on the Li+ is reduced from 1.0 to 0.805, 0.795 and 0.759 for complexes 4b–4d, respectively (Table 3). The transfer of electron density to the Li+ through the carbonyl O atoms of the TK molecule suggests a charge relaxation of Li+ upon the solvation and a higher degree of covalent bonding between Li+ and the carbonyl O.1 This is also evidenced by the heat of formation ΔH0298.15 K increasing in the order of 4b (−356.75 kJ mol−1) < 4c (−364.33 kJ mol−1) < 4d (− 404.84 kJ mol−1), Table 4. The Gibbs free energy of TK-Li complexes 4b–4d is −318.78 kJ mol−1, −320.23 kJ mol−1 and −354.52 kJ mol−1, respectively. Due to the collective effect of the three binding sites per TK molecule, the ΔG0298.15 K of complexes 4b–4d is ∼100 kJ mol−1 higher than complexes of lithium ion with one molecule of PC, EC, or actone (ΔG0298.15 K for Li+(PC)2, (Li+(EC),25 and Li+(acetone)3 is −188.28 kJ mol−1, −212.34kJ mol−1, −164.01 kJ mol−1, respectively). Based on these results, it can be assumed that the coordination of TK to the lithium ion would be favored over PC or EC (in a mixed solvent system as often used in the Li-ion battery).
Reaction | ΔH°(298.15 K) (kJ mol−1) | ΔS°(298.15 K) (kJ mol−1) | ΔG°(298.15 K) (kJ mol−1) |
---|---|---|---|
Li+ + (TK) → Li+(TK)Li–O1O2 (4b) | −356.75 | −0.127 | −318.78 |
Li+ + (TK) → Li+(TK)Li–O2O3 (4c) | −364.33 | −0.148 | −320.23 |
Li+ + (TK) → Li+((TK)Li–O1O2O3(4d) | −404.84 | −0.169 | −354.52 |
Li+ + 4b → (Li+)2(TK) Li–O1O2, Li–O3(4e) | −54.50 | −0.106 | −22.30 |
Li+ + 4c → (Li+)2(TK) Li–O2O3, Li–O1(4f) | −32.01 | −0.104 | −0.98 |
4d+(TK)′ → Li+ (TK) Li–O1O3(TK)′Li–O2′O3′(4g) | −132.02 | −0.242 | −59.98 |
4b+(TK)′ → Li+ (TK) Li–O1O2(TK)′Li–O1′O2′(4h) | −178.27 | −0.157 | −131.51 |
Li+ + (EC) → Li+(EC) | −243.24 | −0.097 | −214.20 |
Li+ + (PC) → Li+(PC) | −249.95 | −0.100 | −220.20 |
Since TK possesses three carbonyl groups, it has the potential to bind more than one Li+. The free carbonyl O atoms on complexes 4b and 4c might solvate a second Li+ to form a (Li+)2(TK)1 complex such as (Li+)2(TK)Li–O1O2, Li–O3(4e), or (Li+)2(TK)Li–O2O3, Li–O1(4f). Upon binding a second Li+, the electron density on the complex is partially redistributed, which is demonstrated by the electron density loss of the Li+ and carbonyl O atoms in the O⋯Li⋯O structures (compared to the 4b or 4c structure), while the electron density gain of those of the second Li⋯O bond (e.g., the bond of Li⋯O3 in complex 4e (Li+)2(TK)Li–O1O2, Li–O3 compare to Li+). In these complexes, CO bonds in the bidentate structure are less polarized due to electron density loss of the lithium ion. Thus, the average distance of CO bonds contracts by ∼0.003 Å, however, the average distance of Li–O is slightly stretched by ∼0.007 Å for 4e and 0.009 Å for 4f.
In addition, the bond angle of Li⋯O3C in complex 4e is about 174.4°, in contrast to linear Li+(acetone) complexes (180°).23,26 Apparently, this due to electronic repulsion between the two Li+. The change in Gibbs free energy, ΔG, in reforming complex 4e from 4b is ∼−22.30 kJ mol−1, and reforming complex 4f from 4c is only about −0.98 kJ mol−1. The equilibrium constant of formation of 4e is ∼8.07 × 103, greater than the 4f (∼1.5), suggesting that the formation of complex 4e is a more favorable process than the formation complex 4f. It should be pointed out that solvating three lithium ions to form (Li+)3(TK)Li–O1, Li–O2, Li–O3 is thermodynamically unfavorable, since ΔG is greater than zero.
The solvation number of Li+ is also dependent on the steric features of the corresponding organic solvent. The favorable solvation number is reported to be four for PC or EC,2,3three for actone, two for diethylether23 and six for acetonitrile.27 In this study, we find that two TK molecules can be directly linked to a lithium ion using four carbonyl oxygen atoms, and another two tend to be liberated from the solvated complex. Thus, two possible formations are available, an asymmetrical tetrahedral complex of Li+ (TK) Li–O1O3(TK)′Li–O2′O3′ (Fig. 4g) in which Li+ is surrounded by carbonyl O1 and O3 from one TK, O2′ and O3′ from another TK′, and a symmetrical Li+ (TK) Li–O1O2(TK)′Li–O1′O2′ (Fig. 4h) where Li+ is surrounded by carbonyls O1and O2 from both TK molecules.
This journal is © The Royal Society of Chemistry 2012 |