Synthesis of Tertiary and Quaternary Amine Derivatives from Wood Resin as Chiral NMR Solvating Agents

Chiral tertiary and quaternary amine solvating agents for NMR spectroscopy were synthesized from the wood resin derivative (+)-dehydroabietylamine (2). The resolution of enantiomers of model compounds [Mosher’s acid (3) and its n-Bu4N salt (4)] (guests) by (+)-dehydroabietyl-N,N-dimethylmethanamine (5) and its ten different ammonium salts (hosts) was studied. The best results with 3 were obtained using 5 while with 4 the best enantiomeric resolution was obtained using (+)-dehydroabietyl-N,N-dimethylmethanaminium bis(trifluoromethane-sulfonimide) (6). The compounds 5 and 6 showed a 1:1 complexation behaviour between the host and guest. The capability of 5 and 6 to recognize the enantiomers of various α-substituted carboxylic acids and their n-Bu4N salts in enantiomeric excess (ee) determinations was demonstrated. A modification of the RES-TOCSY NMR pulse sequence is described, allowing the enhancement of enantiomeric discrimination when the resolution of multiplets is insufficient.


Titration
For titration, solutions containing the guest (2.0 mM) and host (46.6 mM) were prepared and 0.5 mL of guest solution was measured into an NMR tube. Guest was titrated by 2.5 or 5.0 μ doses of host solution. Both 1 H-and 19 F-NMR spectra were recorded.

Titration of 3 with 5
Concentration of guest (3) is presumed to remain constant during titration.

Titration of 4 with 6
Concentration of guest (4) is presumed to remain constant during titration.

Enantiomeric Excess Measurements
For ee determination studies, solutions of racemic 3 and S-3 (as well as 4 and S-4) were made (2.0 mM). Mixtures of enantiomerically enriched samples were prepared in an NMR tube (0.5 mL, 1.0 eq), CSA added (22.5 μ, 46.6 mM, 1.0 eq) and spectra recorded.

Resolution of racemic carboxylic acids
For chiral recognition ability determination study of racemic carboxylic acids, solutions of different racemic carboxylic acids (n-Bu4N salts of carboxylic acids in the case of 6) were prepared (2.0 mM). Experiment was performed by adding 22.5 μ of solution containing host (46.6 mM, 1.0 eq) to NMR tube containing 0.5 mL of guest (2.0 mM, 1.0 eq) and recording the spectra. 5   Table S7. Determination of magnitude of non-equivalences (Δδ) of five racemic carboxylic acids in the presence of 5, using 1 H-NMR (500 MHz) in CDCl3 at 27 °C.

No. Racemic Carboxylic Acid Actual Multiplet Spectra (If Split Is Detected) Δδ
10a

Modified RES-TOCSY, A Phase Sensitive Version
Recently, Lokesh et al., reported new homonuclear 2D technique, RES-TOCSY [1], for obtaining improved separation of enantiomer spectra when chiral resolving agents are utilized. RES-TOCSY is based on homonuclear decoupling in f1-dimension for selectively excited protons and for the whole coupled homonuclear spin system. Homonuclear decoupling is achieved by applying selective refocusing pulse followed by non-selective 180° pulse in between t1-period. The net effect is thus 0° pulse for selected protons and 180° pulse for other protons leading to homonuclear decoupling for the selected protons (and thus collapse of the multiplets into a single lines in f1-dimension) while retaining chemical shift evolution of the selected protons. Subsequent TOCSY-period transfers the magnetization through entire coupled homonuclear spin system. The original RES-TOCSY is a S13 magnitude mode experiment with two selective pulses (90° and 180°) and the coherence selection is performed by phase cycling. The modified RES-TOCSY pulse sequence utilized in this work is presented in Figure S19 and pulse sequence code in Varian syntax is shown in the end of this section. The presented pulse sequence is actually a simplified BASHD-TOCSY [2,3] where a single selective 180° pulse flanked by pulsed field gradient of opposite amplitudes is applied, followed by a non-selective 180° pulse in the middle of t1-period. Gradients with opposite polarities (g1) label the desired magnetization and, as with original RES-TOCSY, the net effect of the two pulses allows selective homonuclear decoupling in f1-dimesion. After t1-period TOCSY transfers the magnetization throughout the spin system and finally magnetization with gradient encoding is refocused during spin-echo period prior to the acquisition (AQ). Phase sensitive data is obtained with echo-anti echo method, where N-and P-type coherences are recorded separately by inverting the polarity of the refocusing gradient g2. Since all signals in the f1-dimension of the resulting spectrum are originating from the excitation bandwidth of the selective pulse, it is possible to utilize small spectral width in the indirectly detected dimension.  (tr1 and tr2). Pulsed field gradients are represented by rectangles marked with g1 and g2. Delay t1 represents incremented delay while delays τ1 and τ 2 include gradient pulse duration and following eddy current recovery delay. Phase cycle: ϕ1 = y, −y; ϕ2 = x, x, y, y; ϕ3 = x, x, x, x, y, y, y, y; ϕ4 = x, −x; ϕ5 = x; ϕR = x, −x, −x, x, −x, x, x, −x. The N-and P-type coherences are separately recorded by inverting the sign of the gradient g2. Axial peak displacement is obtained using States-TPPI method [5] by inverting the phases ϕ1 and ϕR on every second increment. Figure S18 show expansion of 2D spectrum recorded with pulse sequence presented in Figure S19 from 2.0 mM racemic N-acetyl-phenylalanine sample doped with CSA 5 with 1:1 host:guest ratio in CDCl3. The selective 180° pulse was applied at enantiomeric resonances at 4.67 ppm. The separation between the enantiomers Δδ = 6.1 Hz for the selectively inverted proton is present in each expansion (TOCSY transfer). Due to homonuclear decoupling, multiplets collapse into single lines in f1-dimension and thus allow good separation of the enantiomeric signals. Obviously, if the CSA induced Δδ between the enantiomer signals is of same magnitude as natural line width, baseline separation is naturally not possible even with homonuclear decoupling. Figure S21 shows expansion of the modified RES-TOCSY spectrum recorded from 2.0 mM ketoprofen sample doped with the same CSA as well as f2-traces extracted from locations marked by arrows. Since Δδ and linewidth in f1-dimension are of similar magnitude (~2.0 Hz), some cross talk between the traces is unavoidable. The data in Figure S19 was acquired by selectively inverting multiplet region at 1.55 ppm. Other acquisition and processing parameters were identical to data presented in Figure S20.