Enantiomeric excess by NMR: when NMR is the best method, which chiral auxiliary resolves the signals of the two enantiomers, and how little of the minor enantiomer can be detected
Share
Enantiomeric excess by NMR: when NMR is the best method, which chiral auxiliary resolves the signals of the two enantiomers, and how little of the minor enantiomer can be detected
Determination without a racemate, the three-component boronic acid assembly, isotopically chiral molecules, and the level at which a chiral column is still required
The enantiomeric composition of a newly prepared chiral compound usually has to be determined before a chiral HPLC method has been developed for it, and often before a racemic reference sample exists. NMR spectroscopy requires neither.
With an enantiopure derivatising agent, a chiral solvating agent or a chiral liquid-crystalline solvent, the two enantiomers give separate signals, and the ratio of the integrals is the enantiomer ratio. A primary amine can be analysed in this manner with two commercially available reagents and a 1H spectrum acquired in 5 minutes. The method has also been applied to compounds for which chiral chromatography failed or is regarded as impracticable: molecules whose chirality arises solely from deuterium substitution, and a lipid in which the two hydrocarbon chains attached to the stereocentre differ only at a considerable distance from it. Its main limitation is sensitivity. Published NMR determinations reach 0.5 % of the minor enantiomer with a solvating agent and a ratio of 1000:1 for diastereomers, whereas a validated chiral HPLC method quantifies 0.025 %; a specification limit of about 0.1 % therefore still requires a validated chiral separation.
Five terms used in this article. Enantiomeric excess (ee) is the difference between the mole fractions of the two enantiomers,[1] so that an enantiomer ratio of 95:5 corresponds to 90 % ee. A chiral derivatising agent is an enantiopure compound that is bonded covalently to the analyte, whereas a chiral solvating agent associates with it through non-covalent interactions;[2] both are called chiral auxiliaries here. The shift difference Δδ is the difference in chemical shift, in ppm, between the signals of the two enantiomers in the presence of the auxiliary. A scalemic sample is a mixture of enantiomers that is neither racemic nor enantiopure.[3]
At a glance: the analytical problem and the NMR method
| Analytical problem | Chiral column | NMR method |
|---|---|---|
| New compound; neither the racemate nor the opposite enantiomer available | The analysis generally requires one of the two | Imines formed with both enantiomers of a chiral aldehyde, 13C |
| Primary amine for which no separation has been developed | Method development for each compound | 2-Formylphenylboronic acid and BINOL in the NMR tube, 1H |
| Large numbers of samples from a reaction screen | Up to 60 minutes per run | 1H in 5 minutes; 19F in a chiral liquid crystal in about 10 seconds |
| Alcohol of unknown absolute configuration | A ratio, but no configuration without an authentic sample | The two esters of Mosher's acid (MTPA), 1H |
| Compound chiral solely through isotopic substitution | Regarded as impracticable; only a few partial resolutions are cited | 2H NMR in a chiral liquid crystal |
| Chains that differ only at a distance from the stereocentre | Standard separations completely ineffective for one amino-lipid | Solvating agent, 1H at 500 MHz |
| Specification limit near 0.1 % for the unwanted enantiomer | A validated method quantifies 0.025 % | Insufficient: the published limits are 0.5 % and, for diastereomers, 1000:1 |
When NMR is the preferable method
A chiral HPLC method has to be developed for each compound, and the analysis generally requires the racemate or the opposite enantiomer as a reference.[1, 4] With a derivatising agent that is available as both enantiomers, a determination by NMR requires no such reference.
A new compound for which no racemate exists
In 2015 a synthesis group at University College London reported a problem from its work on amide couplings mediated by the borate ester B(OCH2CF3)3.[4] The group frequently had to establish whether an α-amino amide had racemised during the coupling.
In many cases the enantiomer ratio was obtained by chiral HPLC of the tert-butoxycarbonyl (Boc) derivatives. For several amides, none of the columns available in the laboratory separated the enantiomers. Each analysis also required the racemate or the opposite enantiomer, and the authors describe the preparation of these reference samples for a large number of compounds as “extremely tedious”.
With the aldehyde-based derivatisation methods previously reported for amines, imine formation from these amino amides was slow and incomplete. The group then used 2-(tert-butyldimethylsilyloxy)propanal, the silyl ether of lactaldehyde, which is obtained in two steps from a lactate ester. Both enantiomers were prepared on a multigram scale, and the aldehyde did not racemise readily on storage for up to one month.
The aldehyde condenses with a primary amine in the NMR solvent at room temperature, and each enantiomer of the amine gives one imine (Figure 1). The 1H spectrum did not resolve the two diastereomeric imines well enough for integration. In the 13C spectrum of the amino amide shown in Figure 1, the signals of the two imine carbons near 170 ppm were separated by 0.09–0.21 ppm in the four solvents examined, and an acquisition of typically 15 minutes gave a sufficient signal-to-noise ratio. The two integrals can be compared only if the two carbons have similar relaxation times (Part 9).
Figure 1. Condensation of the (S)-aldehyde with the racemic N-propylamide of alanine to give two diastereomeric imines. Redrawn from Scheme 2 and Table 1 of [4]; the conditions are those of the graphical abstract of the same paper, and the two signals are schematic. TBS = tert-butyldimethylsilyl, Pr = propyl.
When a sample consists mainly of one enantiomer and no racemate is available, the position of the minor signal is not known in advance. The second enantiomer of the aldehyde locates it (Figure 2). With the (R)-aldehyde, the major enantiomer of the amine gives the mirror image of the imine that the minor enantiomer gives with the (S)-aldehyde. Mirror images have identical spectra, so the intense signal of the second spectrum appears at the chemical shift of the weak signal of the first.
Figure 2. Commercial (R)-1-phenylethylamine, analysed with each enantiomer of the aldehyde. The first descriptor refers to the carbon derived from the aldehyde, the second to that derived from the amine. The ratios and the separation of 0.139 ppm (in CD3OD) are taken from Table 2 of [4]. The spectra are drawn by us from these values; the source does not state which signal lies at the higher shift.
The authors analysed every amine with both enantiomers of the aldehyde and averaged the two ratios. The method was then tested blind: each of three researchers prepared three mixtures of the enantiomers of commercial amines by weight, and the other two analysed them (Figure 3). Fourteen of the eighteen results were within one percentage point of the gravimetric composition. The largest deviation, three points, occurred for a 94:6 mixture, which was found to be 91:9; the authors attribute the deviations at high ratios to the uncertainty of a small integral.
Figure 3. The blind test. Plotted by us from Table 3 of [4]; the two results for each mixture are drawn side by side. The authors corrected the gravimetric composition for the enantiomer ratio that they had determined for each commercial amine.
Three amides obtained from the coupling were analysed by both methods; the results differed by no more than 1.5 percentage points.
Three amides from the borate-mediated coupling, analysed by both methods
| N-Benzylamide of | Chiral HPLC of the Boc-protected amide | 13C NMR of the free amine with the aldehyde |
|---|---|---|
| Boc-methionine | 98 : 2 | 98.5 : 1.5 |
| Boc-alanine | > 99 : 1 | 99 : 1 |
| Boc-phenylalanine | 99 : 1 | 97.5 : 2.5 |
Enantiomer ratios from Scheme 4 of [4]. The Boc group was removed with trifluoroacetic acid before the NMR measurement.
The spectra must be recorded directly after the sample is prepared, because some of the imines isomerise to enamines on standing. With α-methylphenylalanine methyl ester, in which the α-carbon is fully substituted, the enamine formed during the condensation itself.[4]
A primary amine and two commercially available reagents
In 2006 Bull, James and co-workers introduced an analysis that uses two reagents, 2-formylphenylboronic acid (2-FPBA) and enantiopure 1,1′-bi-2-naphthol (BINOL).[5, 6] The amine condenses with the formyl group and the diol esterifies the boronic acid, so that each enantiomer of the amine gives one iminoboronate ester (Figure 4).[1] The original report covered 13 chiral primary amines, and the protocol published in 2008 takes less than 90 minutes.
Figure 4. The three-component assembly with α-methylbenzylamine (1-phenylethylamine, the amine of Figure 2). Redrawn from Schemes 1 and 2 of [1]. The first descriptor refers to the axis of BINOL, the second to the carbon of the amine; the axial configuration of BINOL is indicated by its label and is not drawn. The arrow from nitrogen to boron is a dative bond.
In the academic year 2015–2016, 113 second-year undergraduates at the University of Birmingham carried out this analysis as a laboratory exercise.[1] Each sample consisted of 0.3 mL of a 60 mM solution of α-methylbenzylamine and 0.3 mL of a solution containing 2-FPBA and (R)-BINOL, each at 50 mM. Both solutions were prepared in CDCl3 and dried over 4 Å molecular sieves. The students integrated the imine and benzylic signals of the two esters and, with a calibration line drawn from mixtures of known ee, obtained values within ±10 % of the true ee. The authors estimate the cost of materials for one analysis at 6.5 pence for BINOL, 4.4 pence for 2-FPBA and 27 pence for CDCl3, at the UK prices of one supplier.
The amine must be present in excess over the two reagents, and all solutions must be dry, because water hydrolyses the imine.[1] The group at University College London found the assembly satisfactory for simple branched amines such as α-alkylbenzylamines but could not apply it to its α-amino amides.[4] By 2023, protocols of this kind had been reported for diols, amino alcohols, diamines, amino esters, hydroxy acids, diacids and hydroxylamines.[7]
Both reagents are available from our catalogue: 2-formylphenylboronic acid and (S)-BINOL. If (S)-BINOL is used, the two esters are the mirror images of those shown in Figure 4, and the two signals exchange positions.
Large numbers of samples from a reaction screen
The screening of asymmetric reactions produces hundreds or thousands of samples per day. Chiral HPLC, which requires method development and has long run times, is poorly suited to this throughput.[1] Enantiomer ratios have been determined by NMR by undergraduate students and, at 500 MHz, by an automated system in an industrial laboratory.
Published NMR determinations of the enantiomer ratio: time, sample and instrument
| Measurement | Time and sample | Instrument and setting |
|---|---|---|
| 1H, primary amine in the three-component assembly[1] | 5 min; 0.6 mL of CDCl3 | 200 MHz or more is stated to be sufficient; undergraduate laboratory, University of Birmingham, 2015–2016 |
| 13C, primary amine as the imine of the lactaldehyde TBS ether[4] | typically 15 min | Synthesis laboratory, University College London, 2015; the article does not state the field |
| 1H, one analyte with each of 32 solvating agents[8] | 80 µL of a 10 mM solution per agent, of which 35 µL is measured in a 1.7 mm tube | 500 MHz with a liquid handler and a sample changer; research laboratory of a pharmaceutical company, 2016 |
| 31P, alcohol or amine with a phosphite from PCl3 and BINOL[2] | less than 5 min; 500 µL of deuterated solvent | Developed in 2009 for catalytic hydrosilylations and transfer hydrogenations; known to us only from a review of 2017 |
| 19F, flurbiprofen and efavirenz in a chiral liquid crystal[9] | about 10 s | 300 MHz with a standard probe, 2024 |
| Chiral HPLC, for comparison[1] | up to 60 min; about 60 mL of solvent | As quoted by the Birmingham authors |
An alcohol of unknown absolute configuration
A chromatogram gives the ratio of two enantiomers but does not show which of them is the major one unless a sample of known configuration is available. The esters of α-methoxy-α-(trifluoromethyl)phenylacetic acid (MTPA, Mosher's acid) give both the ratio and the configuration. Mosher and co-workers introduced the acid in 1969 as a reagent for alcohols and amines,[10] and its esters have since become the standard method for chiral alcohols.[4]
The alcohol is esterified in two separate portions, one with (R)-MTPA and the other with (S)-MTPA (Figure 5). The 1H chemical shifts of corresponding protons in the two esters are subtracted, and the signs of the differences give the configuration of the carbinol carbon.[11] The protocol of 2007 estimates 4–6 hours of work spread over one to two days.
Figure 5. The Mosher ester analysis of a secondary alcohol R1R2CHOH. Drawn by us after the abstract of [11]; the configuration shown for the alcohol is arbitrary.
The MTPA amides of amines are often difficult to analyse, because slow rotation about the amide C–N bond produces rotational isomers (Part 8).[4] The stereodescriptor of the reagent also changes on derivatisation: the acid chloride prepared from (R)-MTPA is designated S, because the COCl group takes precedence over CF3 in the sequence rules, whereas the CO2H and ester groups rank below it. (S)-MTPA chloride therefore gives the (R)-MTPA ester.
When a chiral column cannot separate the enantiomers
A chiral column separates two enantiomers only if its stationary phase binds one of them more strongly than the other. An NMR solvating agent does not have to meet this condition, because two complexes of equal stability can differ in geometry and give separate signals.[8] The argument comes from a group that screened solvating agents in a pharmaceutical research laboratory in 2016. They name two classes of compound for which chromatographic separation can be extremely difficult: compounds that are chiral through isotopic substitution, and compounds that are chiral through a small difference in the length of two alkyl chains.
Chirality through isotopic substitution
A carbon atom that carries one hydrogen and one deuterium atom can be a stereocentre (Figure 6). The replacement of hydrogen by deuterium in drug molecules, known as the deuterium switch, has a marked effect on efficacy and safety; according to a paper of 2026, its use is limited by the difficulty of controlling and analysing the site and the extent of deuteration.[12]
The ee of an isotopically chiral compound is generally assumed to equal that of its synthetic precursor; where it has been verified, this has mostly been done indirectly.[12] The same paper describes chiral chromatography as impracticable for such compounds and cites only a few cases of partial resolution, among them the separation reported by Pirkle and Gan in 1997.[13] According to its authors, two methods measure the ee directly in practice, rotational spectroscopy in the gas phase and NMR in a chiral liquid crystal; chiral lanthanide shift reagents have succeeded in isolated cases.
Figure 6. Molecules whose chirality arises solely from deuterium, and the chiral liquid crystal. Ethylbenzene-1-d is one of the known examples listed in [12]; the bianthryl is compound 10 of that paper, whose axial configuration is not drawn here. The structures are redrawn from the published ones; the doublets are schematic in position and splitting; positions 2 and 3 are the two that were differentiated.
The authors prepared a 9,9′-bianthryl with four deuterium atoms on one ring in six steps from commercial BINOL and dissolved it in a solution of poly-γ-benzyl-L-glutamate (PBLG) in chloroform. This polypeptide forms a chiral liquid crystal, which orients the dissolved molecules. Because the two enantiomers adopt different average orientations, the quadrupolar splittings of the deuterium signals can differ between them (quadrupolar nuclei are the subject of Part 7).
The 2H spectrum of the racemate, recorded at 9.4 T, contained six doublets, because two of the four deuterium positions, 2 and 3, gave separate doublets for the two enantiomers. Each enantiomerically enriched sample gave only four, from which the authors derived an ee of at least 97 %; the 13C spectrum was consistent with a value above 95 %.[12] Unlike derivatising and solvating agents, a liquid crystal requires no particular functional group in the analyte.[14] Solvating agents have also been applied to such compounds: in the screening of 2016, two isotopically chiral compounds were differentiated at least partially.[8]
Chains that differ only at a distance from the stereocentre
The screening study of 2016 includes an amino-lipid from work on lipid nanoparticles for the delivery of small interfering RNA.[8] Its stereocentre carries two long hydrocarbon chains, a nonyl group and a dienyl chain, which are identical over their first eight methylene groups (Figure 7); the authors note that the asymmetry becomes apparent only eight bonds from the stereocentre. They state that the standard methods of chromatographic enantioseparation were “completely ineffective” for this compound.
Figure 7. The amino-lipid, with its stereocentre marked, and the solvating agent that differentiated its signals. The structures are redrawn from Figure 8 of [8]; the axial configuration of the agent is indicated by its label. The signals are drawn by us at the published shifts (2.20 ppm without the agent; 2.69 and 2.65 ppm with it). The two chains differ from the ninth carbon atom onwards.
The automated screening found several agents that differentiated signals of the lipid, of which 1,1′-binaphthyl-2,2′-diyl hydrogen phosphate was the most effective. With the (S)-agent at 20 mM and the lipid at 10 mM in CDCl3 at 500 MHz, the N-methyl signal at 2.20 ppm was replaced by two signals of equal area at 2.69 and 2.65 ppm, which suggested a racemic sample. A two-dimensional ROESY experiment showed that the two signals belong to the two enantiomers and not to two inequivalent methyl groups of one molecule.[8]
The detection limit for the minor enantiomer
If x is the fraction of the minor enantiomer, ee = 1 − 2x. An error in x appears doubled in the ee: a minor fraction measured as 2 % when it is 1 % lowers the ee from 98 % to 96 %.
Enantiomer ratio, minor fraction and enantiomeric excess
| Enantiomer ratio | Minor enantiomer | ee |
|---|---|---|
| 90 : 10 | 10 % | 80 % |
| 95 : 5 | 5 % | 90 % |
| 99 : 1 | 1 % | 98 % |
| 99.5 : 0.5 | 0.5 % | 99 % |
| 99.9 : 0.1 | 0.1 % | 99.8 % |
Our calculation from the definition of ee.
A validated NMR method
In 2018 Lankhorst and co-workers validated a 13C determination of mandelonitrile against mixtures of known composition (Figure 8).[15] The solvating agent was 2,2,2-trifluoro-1-(9-anthryl)ethanol (TFAE, Pirkle's alcohol).[16] The 1H spectrum could not be used, because the aromatic signals of the analyte overlapped with those of the agent and the methine signal was not differentiated.
Figure 8. The determination of mandelonitrile with TFAE. Structures drawn by us from the compound names; conditions and values from [15]. The two signals are drawn by us at the published separation.
In the 13C spectrum, a pair of equivalent aromatic carbons gave one signal for each enantiomer. Both signals arise from the same carbon in the two enantiomers and have the same relaxation time (2.1 s) and the same Overhauser enhancement. The ratio of their integrals is therefore quantitative in a decoupled spectrum, although the delay between pulses was only 3 s. This does not hold for signals of different carbons (Part 9).
Published limits
Published limits and precision for the minor enantiomer
| Measurement | Instrument and sample | Result |
|---|---|---|
| 13C, mandelonitrile with TFAE, 2018[15] | 700 MHz, cryogenically cooled probe, 260 K, 256 scans in 20 minutes, about 4 mg | Limit of detection 0.5 % (99 % ee). Relative standard deviation 0.3 % at 23.8 % of the minor enantiomer and 1.4 % at 6.8 %, six replicates each |
| 1H, mandelonitrile with derivatised cyclodextrins in water[15] | Earlier method, cited in the paper of 2018 | Integration reliable only above about 5 % of one enantiomer |
| 1H, sulfiniminoboronic acid with BINOL as solvating agent, 2023[7] | 500 MHz, 1.5 equivalents of BINOL | Accurate ratios up to 99.5 : 0.5 |
| 19F, flurbiprofen and efavirenz in a chiral liquid crystal, 2024[9] | 300 MHz, standard probe | Standard deviation 0.5 % or less. Efavirenz of 90.5 % ee was found as 90.9 %; flurbiprofen of 88.7 % ee was found as 84.5 % |
| 1H, minor diastereomer compared with a 13C satellite, 2008[17] | Quantitative 1H spectrum | Diastereomer ratios up to 1000 : 1 |
| Chiral HPLC, starting material of dextromethorphan, 2018[18] | Validated method | Limit of detection 0.0075 %, limit of quantification 0.025 % of the unwanted enantiomer |
Each value applies to the compound and the conditions of its source. For flurbiprofen, the authors consider line-fitting errors for signals that are not separated to the baseline a possible cause of the deviation.
The 13C satellites as an internal reference
In a 1H spectrum, the 13C satellites of the major signal can lie in the same region as the signal of the minor diastereomer. The natural abundance of 13C is 1.1 %,[2] so each satellite has 0.55 % of the area of the parent signal, and a minor enantiomer of equal area corresponds to 98.9 % ee. In 2008 Claridge and co-workers used the satellite as an internal reference: in a quantitative spectrum (Part 4), the minor signal is compared with one satellite of the major signal.[17] A derivatised analyte is a mixture of diastereomers, and the comparison applies to it as well.
Figure 9. A signal of the major enantiomer with its two 13C satellites, and the signal of a minor enantiomer present at 0.5 %. Calculated by us for Lorentzian lines of equal width; the positions are schematic.
Where a chiral column is still required
Chiral liquid chromatography reaches lower limits than NMR. The HPLC method in the table quantifies 0.025 % of the unwanted enantiomer, and the same paper reports that the United States Pharmacopeia (USP 40) limits the unwanted enantiomer of dextromethorphan to 0.10 %, tested by HPLC.[18] That limit equals the lowest level reported for NMR in the sources read for this article, the ratio of 1000:1 measured against a satellite. A specification limit near 0.1 % therefore requires chiral chromatography. Yang and co-workers describe the NMR determination as orthogonal to chromatography,[8] so that each method can confirm the other.
Which chiral auxiliary resolves the signals
Two enantiomers have the same internuclear distances and bond angles, and their spectra are identical in an achiral medium. This is not true of diastereomers, whose spectra differ (Part 8). A derivatising agent converts the enantiomers into diastereomers by covalent bonding; a solvating agent forms diastereomeric complexes with them. Parker reviewed the subject in 1991, and Wenzel and Chisholm in 2011.[19, 20]
Derivatising agents
The diastereomer ratio equals the enantiomer ratio only if three conditions are met:
- The agent is enantiopure.
- Neither the analyte nor the agent racemises under the reaction conditions.
- The reaction proceeds to completion, or both enantiomers react at the same rate. If a chiral agent reacts more rapidly with one enantiomer (kinetic resolution) and the reaction is incomplete, that enantiomer is over-represented in the product.
The first condition follows from symmetry. The minor enantiomer of the agent converts the major enantiomer of the analyte into the mirror image of the minor diastereomer, which has the same spectrum. The observed ee is the product of the ee of the analyte and that of the agent.
An analyte of 99 % ee, analysed with an agent that is not enantiopure
| ee of the agent | ee found for the analyte | Minor enantiomer found |
|---|---|---|
| 100 % | 99.0 % | 0.50 % |
| 99.5 % | 98.5 % | 0.75 % |
| 99 % | 98.0 % | 1.0 % |
| 98 % | 97.0 % | 1.5 % |
| 96 % | 95.0 % | 2.5 % |
Our calculation for a complete reaction.
Mosher and co-workers found MTPA configurationally stable under prolonged reaction conditions and regarded the 19F signals of its CF3 group as an advantage over reagents that give proton signals only.[10] In the three-component assembly, enantiopure α-methylbenzylamine is used in place of BINOL when the analyte is a 1,2-, 1,3- or 1,4-diol.[21] Fluorinated 2-formylphenylboronic acids give assemblies that can be observed by 19F NMR; with sulfinamides, the 3-fluoro compound allows the same ratio to be determined from the 1H and the 19F spectrum.[2, 7]
Solvating agents
A solvating agent is added to the solution of the analyte, and the spectrum is recorded without any reaction or work-up. Pirkle reported one of the earliest examples, observed in a 19F spectrum, in 1966.[8, 22] It cannot be predicted which agent will work for a given compound. In the screening of 2016, 32 agents were tested against 16 analytes at 500 MHz, with the analyte at 10 mM and the agent at 20 mM in CDCl3. Most combinations gave no differentiation, but every analyte was differentiated, at least partially, by at least one agent.[8]
Solvating agents by functional group of the analyte
| Analyte | Result of the screening |
|---|---|
| Carboxylic acids[8] | The class with the largest number of effective agents. 1-(1-Naphthyl)ethylamine; quinine for naproxen |
| Amines[8, 23] | 1,1′-Binaphthyl-2,2′-diyl hydrogen phosphate. Also (R)-α-methoxyphenylacetic acid, (R)-O-acetylmandelic acid and (R)-2-methoxy-2-(1-naphthyl)propionic acid |
| Sulfoxides[8] | Whelk-O; quinine for omeprazole |
| Alcohols[8] | The class with the fewest effective agents. For three alcohols only one or two agents were effective |
| Agents effective for the largest number of analytes[8] | Whelk-O, Rh2(MTPA)4, quinine, the 1-(1-naphthyl)ethyl urea derivatives of valine and tert-leucine, dibenzoyltartaric acid |
Source 23, of 2018, compares 13 commercial agents for primary, secondary and tertiary amines in CDCl3. Earlier reports on the binaphthyl phosphate (1989) and on quinine (1988) are [24] and [25]. TFAE is widely used but did not perform well with the 16 analytes of [8].
Replacing the agent by its enantiomer reverses the order of the two signals.[8] A weak signal that reappears on the opposite side of the major signal after this exchange can be assigned to the minor enantiomer and not to an impurity.
Dependence on concentration and enantiomer ratio
In 2023 Bull, James and co-workers examined BINOL as a solvating agent for sulfiniminoboronic acids, which are formed from a sulfinamide and 2-FPBA (Figure 10).[7] With one equivalent of (R)-BINOL in CDCl3, the two imine signals of the racemate were partially resolved at 100 and 50 mM but coincided at 25 mM and below. The shift difference also varied with the enantiomer ratio, and with one equivalent its sign differed between a 90:10 and a 10:90 sample.
Figure 10. BINOL as a solvating agent for a sulfiniminoboronic acid. The formation of the acid is redrawn from Figure 1b of [7], with the (R)-configuration at sulfur shown; the separations are published values, and the signals are drawn by us from them.
A single signal is therefore no proof of enantiopurity. Fully resolved signals gave correct ratios whatever the size of the shift difference. The authors recommend repeating such determinations with both enantiomers of the agent at a fixed concentration and stoichiometry, and reporting the concentration of the NMR sample.[7]
Lanthanide shift reagents
Chiral lanthanide complexes such as Eu(hfc)3, a europium complex of a camphor-derived β-diketone,[26] were in common use when spectrometers operated at 200 MHz or below.[8] In the screening of 2016, run at 500 MHz with the agent at 20 mM, Eu(hfc)3 broadened many spectra so much that they could not be evaluated. For some analytes it still gave the largest shift differences of all the agents, and the authors suggest concentrations of 5 mM or less.
Nuclei that give singlets
In 1H spectra, multiplets often overlap,[2] and a differentiated pair of signals may coincide with signals of the agent.[15]
Fluorine. The 19F signal of many compounds is a singlet, the chemical-shift range is wide, and an agent that contains no fluorine gives no signals.[8] Although the fluorine atom of flurbiprofen is far from the stereocentre (Figure 11), the dirhodium complex Rh2(MTPA)4 resolved the 19F signals of the enantiomers completely at 471 MHz, and Whelk-O resolved them partially.
Figure 11. Flurbiprofen and efavirenz. Both were analysed by 19F NMR in a chiral liquid crystal [9], flurbiprofen also with solvating agents [8]. Structures drawn by us from the compound names; flurbiprofen is shown without configuration.
Phosphorus. 31P has a natural abundance of 100 %.[2] For carboxylic acids derivatised with phosphorus(III) reagents derived from C2-symmetric diamines, the review of 2017 cites shift differences of 0.02–2.43; no unit is given, and we take the values to be in ppm.
Carbon. In a proton-decoupled 13C spectrum, every carbon gives a singlet unless it is coupled to fluorine or phosphorus. Figures 1 and 8 show determinations of this kind.
Pure-shift 1H spectra. A pure-shift experiment reduces every multiplet to a singlet. In the screening of 2016, it resolved a partially overlapped signal of omeprazole into two singlets and gave the two N-methyl singlets of the amino-lipid shown in Figure 7.[8, 27]
Published shift differences between the signals of the two enantiomers
| Nucleus and method | Δδ in ppm | Δδ in Hz at 400 MHz |
|---|---|---|
| 1H, MTPA esters and amides[10] | 0.03–0.13 | 12–52 |
| 19F, MTPA esters and amides[10] | 0.11–0.71 | 41–267 |
| 13C, imines of the lactaldehyde TBS ether, twelve commercial amines[4] | 0.036–0.792 | 3.6–80 |
| 1H, binaphthyl phosphate, N-methyl signal of the amino-lipid[8] | 0.04 | 16 |
| 1H, BINOL as solvating agent, sulfiniminoboronic acid[7] | 0.007–0.036 | 3–14 |
| 1H, Eu(hfc)3, the two C-4 protons of 2-amino-4-phenylbutane[8] | 0.25 and 0.22 | 100 and 88 |
| 13C, TFAE, mandelonitrile[15] | 0.065 | 6.5 |
| 19F, 3-fluoro-2-formylphenylboronic acid assembly, 2-methylsuccinic acid[2] | 1.24 | 467 |
| 31P, phosphorus(III) reagents, carboxylic acids[2] | 0.02–2.43 (no unit in the source) | — |
The values in hertz are our conversion for a 400 MHz spectrometer: 400 MHz for 1H, 376.5 MHz for 19F and 100.6 MHz for 13C. The measurements themselves were made at other fields; the 13C value for mandelonitrile, for example, was measured at 700 MHz, where it is 11 Hz.
Two sources of error in the sample itself
Separate signals without an auxiliary
Chiral molecules that self-associate in solution can give separate signals for the two enantiomers of a scalemic sample without any auxiliary, at chemical shifts that depend on concentration. The effect is called self-induced diastereomeric anisochronism.[28] In a study of 2020, the large concentration-dependent shifts of an alcohol at first led the authors to consider whether they were observing atropisomers and not two enantiomers; they advise close attention to the measuring conditions, the concentration in particular. Other causes of doubled signals are described in Part 8.
A fraction that is not representative of the material
When a scalemic sample is fractionated under achiral conditions, the enantiomers can be distributed unequally among the fractions, and only the recombined fractions have the ee of the original sample. The effect, reviewed in 2018 (Figure 12),[3] has been observed in chromatography on achiral silica and in precipitation, crystallisation, sublimation, evaporation and distillation. The review names amides of chiral amines, amino acid esters, sulfoxides and trifluoromethyl compounds as structural classes that favour it, and its authors recommend that a test for the effect become standard practice.
Figure 12. Two examples of fractionation on achiral silica. Plotted by us from values quoted in [3]; structures drawn by us from the compound names, without configuration.
The ee should be determined on the whole material and not on a single chromatographic fraction.
Summary
- NMR gives the enantiomer ratio of a compound for which neither a chiral separation nor a racemate is available. Chiral HPLC generally requires the racemate or the opposite enantiomer; derivatisation with both enantiomers of an agent does not.
- Primary amines have been analysed as imines of a lactate-derived aldehyde by 13C NMR, typically in 15 minutes, and as iminoboronate esters formed with 2-formylphenylboronic acid and BINOL by 1H NMR in 5 minutes.
- The two MTPA esters of a secondary alcohol give both the enantiomer ratio and the absolute configuration.
- Chromatography is regarded as impracticable for isotopically chiral compounds, and the standard separations were completely ineffective for one amino-lipid. The ratios were obtained by 2H NMR in a chiral liquid crystal and by 1H NMR with a solvating agent, respectively.
- The published NMR limits for the minor enantiomer are 0.5 % with a solvating agent and 1000:1 for diastereomers measured against a 13C satellite. As a validated chiral HPLC method quantifies 0.025 %, a specification limit near 0.1 % requires a chiral column.
- A derivatisation gives the correct ratio only if the agent is enantiopure, neither component racemises, and the reaction is complete or free of kinetic resolution.
- The effective solvating agent cannot be predicted, and a single signal is no proof of enantiopurity, because the shift difference depends on concentration and on the enantiomer ratio.
- 19F, 31P and proton-decoupled 13C signals are singlets and are therefore unaffected by the overlap of 1H multiplets.
- Chromatography on achiral silica can alter the ee of a scalemic sample from one fraction to the next.
Reagents and NMR from NorrChemica
2-Formylphenylboronic acid and (S)-BINOL, the two reagents of the three-component analysis of primary amines, are available from our catalogue. Our NMR service records 1H, 13C, 19F, 31P and 11B spectra in solution at 400 MHz, priced per sample; 2H spectra and long acquisitions are quoted per sample. If you do not know which experiment answers your question, describe the question on the request form. The service is available worldwide for chemical and pharmaceutical R&D. Every report is signed and supplied with the native data. Research use only.
- 2-Formylphenylboronic acid (2-FPBA)aldehyde and boronic acid in one molecule
- (S)-1,1′-Bi-2-naphthol, (S)-BINOLenantiopure diol
- 5 mm NMR tubes for quantitative NMRfor the integral of a minor signal
- 5 mm precision NMR tubes, 300–800 MHz1H and 19F spectra
- 3 mm NMR tubesfor a few milligrams of sample
NMR tubes made in Germany by Deutero and shipped from Finland. See the NMR tube guide →
References
- Fossey JS, Anslyn EV, Brittain WDG, Bull SD, Chapin BM, Le Duff CS, James TD, Lees G, Lim S, Lloyd JAC, Manville CV, Payne DT, Roper KA. Rapid determination of enantiomeric excess via NMR spectroscopy: a research-informed experiment. J. Chem. Educ. 2017, 94, 79–84. DOI
- Silva MS. Recent advances in multinuclear NMR spectroscopy for chiral recognition of organic compounds. Molecules 2017, 22, 247. DOI
- Han J, Kitagawa O, Wzorek A, Klika KD, Soloshonok VA. The self-disproportionation of enantiomers (SDE): a menace or an opportunity? Chem. Sci. 2018, 9, 1718–1739. DOI
- Gibson SM, Lanigan RM, Benhamou L, Aliev AE, Sheppard TD. A lactate-derived chiral aldehyde for determining the enantiopurity of enantioenriched primary amines. Org. Biomol. Chem. 2015, 13, 9050–9054. DOI
- Pérez-Fuertes Y, Kelly AM, Johnson AL, Arimori S, Bull SD, James TD. Simple protocol for NMR analysis of the enantiomeric purity of primary amines. Org. Lett. 2006, 8, 609–612 (abstract). DOI
- Pérez-Fuertes Y, Kelly AM, Fossey JS, Powell ME, Bull SD, James TD. Simple protocols for NMR analysis of the enantiomeric purity of chiral primary amines. Nat. Protoc. 2008, 3, 210–214 (abstract). DOI
- Groleau RR, Chapman RSL, Lowe JP, Lyall CL, Kociok-Köhn G, James TD, Bull SD. BINOL as a chiral solvating agent for sulfiniminoboronic acids. Anal. Chem. 2023, 95, 16801–16809. DOI
- Yang L, Wenzel T, Williamson RT, Christensen M, Schafer W, Welch CJ. Expedited selection of NMR chiral solvating agents for determination of enantiopurity. ACS Cent. Sci. 2016, 2, 332–340. DOI
- Gouilleux B, Moussallieh F-M, Lesot P. Potential and performance of anisotropic 19F NMR for the enantiomeric analysis of fluorinated chiral active pharmaceutical ingredients. Analyst 2024, 149, 3204–3213. DOI
- Dale JA, Dull DL, Mosher HS. α-Methoxy-α-trifluoromethylphenylacetic acid, a versatile reagent for the determination of enantiomeric composition of alcohols and amines. J. Org. Chem. 1969, 34, 2543–2549 (first page). DOI
- Hoye TR, Jeffrey CS, Shao F. Mosher ester analysis for the determination of absolute configuration of stereogenic (chiral) carbinol carbons. Nat. Protoc. 2007, 2, 2451–2458 (abstract). DOI
- Rigoulet F, Dauvergne G, Naubron J-V, Lesot P, Aroulanda C, Coquerel Y. [1H]/[2H] discriminated bianthryl atropisotopomers: enantiospecific syntheses from BINOL and direct multi-spectroscopic analyses of their isotopic chirality. Chem. Sci. 2026, 17, 6513–6523. DOI
- Pirkle WH, Gan KZ. A direct chromatographic separation of enantiomers chiral by virtue of isotopic substitution. Tetrahedron: Asymmetry 1997, 8, 811–814. DOI
- Berdagué P, Herbert-Pucheta JE, Jha V, Panossian A, Leroux FR, Lesot P. Multi-nuclear NMR of axially chiral biaryls in polypeptide orienting solvents: spectral discriminations and enantiorecognition mechanisms. New J. Chem. 2015, 39, 9504–9517. DOI
- Lankhorst PP, van Rijn JHJ, Duchateau ALL. One-dimensional 13C NMR is a simple and highly quantitative method for enantiodiscrimination. Molecules 2018, 23, 1785. DOI
- Pirkle WH, Sikkenga DL, Pavlin MS. Nuclear magnetic resonance determination of enantiomeric composition and absolute configuration of γ-lactones using chiral 2,2,2-trifluoro-1-(9-anthryl)ethanol. J. Org. Chem. 1977, 42, 384–387. DOI
- Claridge TDW, Davies SG, Polywka MEC, Roberts PM, Russell AJ, Savory ED, Smith AD. “Pure by NMR”? Org. Lett. 2008, 10, 5433–5436 (abstract). DOI
- Anerao A, Solase V, More A, Pradhan N. Separation, quantification and control of enatiomers [sic] of the key starting material of dextromethorphan hydrobromide. J. Appl. Pharm. Sci. 2018, 8 (08), 032–038. DOI
- Parker D. NMR determination of enantiomeric purity. Chem. Rev. 1991, 91, 1441–1457. DOI
- Wenzel TJ, Chisholm CD. Using NMR spectroscopic methods to determine enantiomeric purity and assign absolute stereochemistry. Prog. Nucl. Magn. Reson. Spectrosc. 2011, 59, 1–63. DOI
- Kelly AM, Pérez-Fuertes Y, Fossey JS, Yeste SL, Bull SD, James TD. Simple protocols for NMR analysis of the enantiomeric purity of chiral diols. Nat. Protoc. 2008, 3, 215–219 (abstract). DOI
- Pirkle WH. The nonequivalence of physical properties of enantiomers in optically active solvents. Differences in nuclear magnetic resonance spectra. I. J. Am. Chem. Soc. 1966, 88, 1837. DOI
- Benedict B, Lietz CE, Wenzel TJ. Comparison of chiral NMR solvating agents for the enantiodifferentiation of amines. Tetrahedron 2018, 74, 4846–4856 (abstract). DOI
- Shapiro MJ, Archinal AE, Jarema MA. Chiral purity determination by 1H NMR spectroscopy. Novel use of 1,1′-binaphthyl-2,2′-diylphosphoric acid. J. Org. Chem. 1989, 54, 5826–5828. DOI
- Rosini C, Uccello-Barretta G, Pini D, Abete C, Salvadori P. Quinine: an inexpensive chiral solvating agent for the determination of enantiomeric composition of binaphthyl derivatives and alkylarylcarbinols by NMR spectroscopy. J. Org. Chem. 1988, 53, 4579–4581. DOI
- Fraser RR, Petit MA, Saunders JK. Determination of enantiomeric purity by an optically active nuclear magnetic resonance shift reagent of wide applicability. J. Chem. Soc. D 1971, 1450. DOI
- Castañar L, Pérez-Trujillo M, Nolis P, Monteagudo E, Virgili A, Parella T. Enantiodifferentiation through frequency-selective pure-shift 1H nuclear magnetic resonance spectroscopy. ChemPhysChem 2014, 15, 854–857. DOI
- Baumann A, Wzorek A, Soloshonok VA, Klika KD, Miller AK. Potentially mistaking enantiomers for different compounds due to the self-induced diastereomeric anisochronism (SIDA) phenomenon. Symmetry 2020, 12, 1106 (abstract). DOI
About this article. We read sources 1 and 4 in full in the browser, with their schemes and tables. Source 8 was read as the open-access text, and its Figure 8 with the passage on the amino-lipid was read again for the structures and the shifts. Of sources 2, 3, 7, 9, 12 and 15 we read the cited passages in the open-access text, and of 12 and 7 also the schemes from which Figures 6 and 10 are drawn. Source 18 was read through a retrieval tool only, which is not equivalent to reading the article itself. Of source 14 we read the abstract and the introduction. The 31P method with PCl3 and BINOL (Reiner, Naraschewski and Eppinger, 2009) and the individual 19F and 31P reagents are known to us only from the review 2; the review prints the name of the 3-fluoro boronic acid without “phenyl”, and we use the name given in 7. Sources 5, 6, 11, 17, 21, 23 and 28 were read as abstracts only and source 10 as its first page, so the description of the Mosher analysis is limited to what the abstract of 11 and the introduction of 4 state. Sources 13, 16, 19, 20, 22, 24, 25, 26 and 27 are bibliographic records, cited for the origin of a method or for further reading. The following are our own reasoning and arithmetic: the argument that the second enantiomer of an agent locates the minor signal; the counts taken from the blind test and from the comparison with HPLC in 4; the statement on (S)-BINOL; the statement that a chromatogram does not identify the major enantiomer; the descriptor of MTPA chloride, which we checked with a program for the priority rules; the two tables of ee values; the three conditions for a derivatisation; the statement on the signal that changes sides; the conversion of shift differences to hertz; the ee that corresponds to a 13C satellite; the remark on carbon signals coupled to fluorine or phosphorus; the application of the satellite comparison to derivatised samples; and the advice to analyse the whole material. That the amino-lipid comes from work on lipid nanoparticles is taken from the title of the reference that 8 cites for it. All structures are our drawings, made from the published schemes or from the compound names as stated in each caption. Figures 3 and 12 are plotted by us from published values, Figure 9 is our calculation, and the signals in Figures 1, 2, 6, 7, 8 and 10 are schematic or drawn from published values as stated. Our service is described from the service page of our shop. The sources were read on 14 and 15 November 2026.
NMR Series
- Part 1 — Reading 1H and 13C NMR
- Part 2 — Reading 19F, 31P and 11B NMR
- Part 3 — Reading 2D NMR: COSY, HSQC and HMBC
- Part 4 — Quantitative NMR: how a purity figure is made
- Part 5 — 15N NMR: which nitrogen is alkylated, protonated or coordinated?
- Part 6 — NMR tubes by experiment: which tube for which measurement
- Part 7 — Quadrupolar nuclei in NMR: broad lines and missing carbons
- Part 8 — Two sets of NMR signals: rotamers, diastereomers or an impurity
- Part 9 — NMR yield and isomer ratios: relaxation and the delay between pulses
- Part 10 — Silicon NMR: the glass signal, the negative NOE and M, D, T, Q units
- Part 11 — Solid-state NMR: magic-angle spinning, cross-polarisation and what a CP-MAS spectrum counts
- Part 12 — Enantiomeric excess by NMR: when NMR is the best method, the chiral auxiliaries and the detection limit for the minor enantiomer (this article)
- Part 13 — Predicted NMR shifts: how accurate ChemDraw, MestReNova and other predictors are
- Part 14 — Phosphorus NMR in biology: phospholipids, phosphorylated proteins, oligonucleotide drugs and phosphates in food
- Part 15 — Multiplets in ¹H NMR: coupling trees, second-order patterns and two signals from one CH₂
- Part 16 — NMR for cosmetics: adulterated essential oils, free formaldehyde, aloe vera, rancid lipsticks and mineral oil in lip products
- Part 17 — Polymers by NMR: composition, sequence, chain ends and branches, and what a spectrum of many chains can and cannot measure
- Part 18 — Benchtop NMR: what a permanent-magnet spectrometer does well, and when a sample needs high field
- Part 19 — NOESY, ROESY and TOCSY: regioisomers, E and Z isomers, anomers, and the signals of one compound in a mixture
- Part 20 — Rapid-injection NMR: starting a reaction inside the spectrometer, how fast the spectra can follow it, and the intermediates it has revealed
- Part 21 — Fluorine NMR for PFAS: total organofluorine, what targeted LC-MS/MS does not detect, and the detection limit
- Part 22 — Polysiloxanes by proton NMR: silicone oil in medicines, food and cosmetics, the reactive groups before curing, and the rubber network after it
© 2026 NorrChemica. All rights reserved.
Content may not be reproduced without written permission.