NOESY, ROESY and TOCSY: regioisomers, E and Z isomers, anomers, and the signals of one compound in a mixture
Share
NOESY, ROESY and TOCSY: regioisomers, E and Z isomers, anomers, and the signals of one compound in a mixture
The three experiments and their spectra, the structural questions they answer on 2 to 30 mg of dissolved sample, the accuracy of a distance measured by NOE, and how little a missing cross-peak proves
An accurate mass gives the formula of a compound, and COSY, HSQC and HMBC show how its atoms are connected (Part 3). Three questions often remain: which of two possible isomers is present, which signals of a crowded spectrum belong to one compound, and what shape the molecule takes in solution. NOESY and ROESY answer the first and the third, because they reveal which protons lie close together in space. TOCSY answers the second: it gathers the protons of one unbroken chain of couplings.
A column separates two isomers, and a mass spectrometer assigns the same formula to both; neither method identifies them. Until 2013 the configuration at the double bond of (Z)-endoxifen, a drug candidate against breast cancer, rested on empirical trends in chemical shifts; a ROESY spectrum of 11 to 12 mg of the substance established it with two cross-peaks.[1] In a study published in 2022, one NOESY cross-peak showed which nitrogen atom of a pyrazole carries the methyl group.[2] NOE measurements published in 1992 reversed the anomeric configuration previously assigned to six nucleosides.[3] A selective TOCSY experiment of about one minute extracted the amino acid signals of honey from a spectrum dominated by two sugars.[4] The experiments have limits. A missing cross-peak does not prove that two protons are far apart, and NMR alone could not confirm the (E)-isomer of endoxifen, because two of its signals overlap with others.[1]
Six terms used in this article. The nuclear Overhauser effect (NOE) is the change in the intensity of one signal when the magnetisation of a nearby nucleus is disturbed; it is transmitted through space and requires no bond. A cross-peak is a signal in a two-dimensional (2D) spectrum at the chemical shifts of two different protons; it shows that the experiment connects them. The mixing time is the part of the pulse sequence during which magnetisation passes from one proton to another. A spin system is a group of protons linked by an unbroken chain of couplings; Part 15 explains coupling. A spin-lock is radio-frequency irradiation, one long pulse or a train of pulses, applied throughout the mixing time of TOCSY and ROESY. In TOCSY it makes the couplings outweigh the differences in chemical shift, so that all protons of a spin system exchange magnetisation.[5] The correlation time is the period in which a molecule in solution rotates by about one radian; it increases with the molecular weight and with the viscosity of the solvent.[6, 7]
At a glance: the question, the experiment and its limit
| Question | Experiment and what the spectrum shows | Limit |
|---|---|---|
| Which of two regioisomers was formed? | NOESY or ROESY: a cross-peak between the new substituent and the group next to it | A missing cross-peak is weak evidence; HMBC supports the assignment |
| E or Z at a double bond without two neighbouring protons? | NOESY or ROESY: cross-peaks between groups on the same side of the double bond | The signals concerned must not overlap |
| α or β anomer of a nucleoside? | NOESY or selective 1D NOESY: NOEs between protons on the same face of the sugar ring | The H1′–H4′ distance differs by only about 0.5 Å between the anomers; several NOEs are compared |
| Which signals belong to one compound of a mixture? | Selective 1D TOCSY: the signals of one spin system, without those of the other compounds | One signal of the compound must be resolved; a molecule can contain several spin systems |
| Which protons belong to one sugar unit of a glycoside? | 1D or 2D TOCSY: all protons of the unit, reached from its anomeric proton | HMBC, not TOCSY, shows how the units are linked |
| How far apart are two protons? | NOESY or ROESY with a known reference distance: a mean error of about 3 % in a rigid molecule | A flexible molecule gives an average over its conformers |
| Is the molecular weight about 700 to 1200 g/mol? | ROESY in place of NOESY: cross-peaks of one sign for molecules of any size | TOCSY peaks appear as artefacts |
How the experiments work
COSY, TOCSY, NOESY and ROESY share one outline. A first pulse excites the protons, the evolution time t1 labels each of them with its chemical shift, a mixing step passes magnetisation between them, and the signal is recorded. Only the mixing step differs (Figure 1).
Figure 1. The four pulse sequences in their simplest form. Narrow rectangles are 90° pulses; the coloured bars mark the mixing step. Schematic: gradients and phase cycles are omitted. Durations from the sources: TOCSY;[5, 8] NOESY of a small molecule and the first trial for ROESY.[7]
TOCSY: a spin-lock in place of the second pulse
TOCSY (total correlation spectroscopy) originates from a paper of Braunschweiler and Ernst on coherence transfer by isotropic mixing, published in 1983.[9] A spin-lock replaces the second pulse of the COSY sequence.[5] While it is applied, magnetisation passes from a proton to its coupling partner, on to the next one, and so along the chain. The transfer stops at an ether oxygen or an amide bond, and wherever the coupling is small, as between an alkyl chain and an aromatic ring.[5] The part of a molecule between two such breaks is one spin system, and every compound in a mixture has its own.
NOESY: a waiting period between two pulses
In NOESY (nuclear Overhauser effect spectroscopy) the second pulse stores the magnetisation along the main field. During the mixing time that follows, the NOE passes part of it to protons closer than 4 to 5 Å, whatever the bonds between them.[10] The guide of Columbia University puts the limit at 4 Å for small molecules and at about 5 Å for large ones, and it recommends a mixing time of 0.5 to 1 s for the former.[7] The sequence of three pulses was published in 1979 by Jeener, Meier, Bachmann and Ernst, in Brussels and Zürich, as a method for the study of exchange processes.[11] A NOESY spectrum shows the cross-peaks of chemical exchange as well (Part 8).
ROESY: the NOE under a spin-lock
In ROESY (rotating-frame Overhauser effect spectroscopy) the NOE builds up under a weak spin-lock. Bothner-By and co-workers introduced the experiment in 1984, in the structure determination of a tetrasaccharide, and named it CAMELSPIN.[12] A year later Bax and Davis at the US National Institutes of Health proposed the present name and described the practical aspects of the 2D version.[13] The experiment suits molecules of intermediate size, whose NOE in a NOESY spectrum is close to zero.[13, 14] The section on the molecular weight gives the ranges. Table 1 compares the three mixing steps.
Table 1 · The three experiments compared
| Experiment | Mixing step | A cross-peak means |
|---|---|---|
| TOCSY | Spin-lock of 15 to 120 ms | The two protons belong to one spin system; they need not couple directly |
| NOESY | Waiting period of 0.5 to 1 s for a small molecule | The two protons are closer than 4 to 5 Å, or they exchange with each other (Part 8) |
| ROESY | Weak spin-lock of 0.1 to 0.5 s | As in NOESY, with one sign at every molecular weight |
Mixing steps and meaning of the cross-peaks from the sources cited in Figure 1.[5, 7]
How the spectra are read
Figure 2 shows thymidine, a nucleoside from our catalogue, with its 1H signals and the three 2D spectra as diagrams. The shifts are those of a public reference entry, measured in D2O at 400 MHz.[15]
Figure 2. Thymidine. Top left: the molecule in Haworth form with its protons; the reference entry gives one shift for both H2′ protons and one for the H5′ pair. Top right: the published positions of the 1H signals in D2O;[15] multiplets are not drawn. Bottom: COSY, TOCSY and NOESY as diagrams, with the signals evenly spaced in the order of their shifts. The cross-peaks are those expected from the structure and from our force-field models; they are not measured.
The diagonal and the cross-peaks
In all three 2D spectra both axes carry the 1H chemical shift. Every proton gives a peak on the diagonal, at its own shift on both axes. A correlation between two protons adds a pair of cross-peaks, mirrored across the diagonal.
The diagrams of Figure 2 apply this to thymidine. COSY gives four pairs of cross-peaks, one for each coupling along the sugar. TOCSY fills the whole block of the sugar protons: the row of H1′ contains H2′, H3′, H4′ and H5′. Neither spectrum connects the sugar with the base.
NOESY does. H6 of the base lies near the sugar, and the strength of its cross-peaks depends on the rotation about the bond that joins the two parts. H1′ gives a cross-peak with H4′ although the two do not couple: both lie on the same face of the ring, 3.3 Å apart in our models. Figure 6 shows a published ROESY spectrum read in this way.
The reach of TOCSY
The mixing time sets the reach of the transfer (Figure 3). A facility guide of Stanford University states that 20 ms is about right for one strong transfer, one medium and one weak, and that 80 to 100 ms suffices for five to eight.[8] Course material of the University of Wisconsin–Madison from 2014 is more cautious: 15 ms shows the first neighbours only, 80 ms three steps with a hint of a fourth, and 120 ms, the longest time it recommends, the whole chain.[5] A short mixing time gives a spectrum close to COSY; a long one shows the whole spin system.
Figure 3. Reach of the TOCSY transfer along a chain of coupled protons at five mixing times. Our drawing of the description in the course material.[5] The reach in a real molecule depends on the size of each coupling.
The size of an NOE
In a selective 1D NOESY spectrum an NOE peak is typically below 2 % of the excited signal and often as small as 0.01 %.[10] The intensity falls steeply with the distance (Figure 11). A strong peak therefore indicates a short distance, whereas a missing one has several possible causes (Table 6).
Selective 1D versions
Each experiment also exists in a one-dimensional form. A selective pulse excites one signal, and the spectrum shows only the protons that receive magnetisation from it: the spin system in 1D TOCSY, the neighbours in space in 1D NOESY.[7, 10] The 1D experiment is faster and answers one question per measurement; the 2D spectrum shows all correlations at once. A group at the University of Manchester states that a selective 1D measurement can often give the information needed for a structure in a fraction of the time of the 2D spectrum, down to about 1 %.[16] The selective 1D NOESY based on pulsed field gradients was reported by Stott, Keeler, Shaka and co-workers in 1995 and described in detail two years later;[17, 18] the measurements on 4-propylaniline discussed below used it.[19]
A selective pulse needs a signal that is separated from all others. In 2021 Kiraly and co-workers in Manchester presented an excitation named GEMSTONE, which selects one multiplet among overlapping ones in a single scan. Their example was the steroid 17β-oestradiol in DMSO-d6: the multiplets of three protons from different rings overlap between 1.7 and 1.9 ppm, a conventional selective pulse excites all of them, and the NOEs cannot be assigned. With GEMSTONE each proton gave its own 1D NOESY spectrum in 12 min.[16]
When NOESY or ROESY is the preferable method
For the three questions of this section the classical proof is X-ray diffraction, which requires a single crystal. The NOE experiment needs only a solution: a selective 1D NOESY can give a significant result in about ten minutes, and the 2D experiment takes from under one hour to several hours (Table 7).[10, 20]
Which nitrogen atom carries the new substituent
The synthesis of an N-substituted heterocycle with two ring nitrogen atoms often gives two regioisomers. Lusardi and co-workers at the University of Genoa prepared phenylaminopyrazoles for tests on cancer cell lines and reported in 2022 that an acrylate, heated with methylhydrazine without solvent at 80 °C for 1.5 h, gave two N-methyl isomers. Thin-layer chromatography showed two spots, and a column separated the compounds 8a and 8b.[2]
The routine data distinguish the isomers but do not identify them. Both give the protonated ion at m/z 261.1344. The 1H spectra, recorded in DMSO-d6 at 400 MHz, contain the same groups of signals at different shifts, with the N-methyl singlets at 3.35 and 3.48 ppm, and the retention times on a reversed-phase column are 13.70 and 20.03 min.[2]
The NOESY spectrum locates the methyl group (Figure 4). Isomer 8a gives a cross-peak between the N-methyl signal at 3.35 ppm and phenyl protons at 6.63 ppm, so the two groups are neighbours. Isomer 8b gives no such cross-peak. In the HMBC spectra the N-methyl protons of 8a correlate with the ring carbon at 142.14 ppm, which bears the phenylamino group. In 8b their partner is the carbon at 148.76 ppm, the one that carries the amino group.[2]
Figure 4. The two N-methylpyrazoles of the 2022 study. H marks the phenyl proton nearest to the methyl group. Gold outline: the two groups connected by the NOESY cross-peak of 8a. Blue lines: the HMBC correlations of the N-methyl protons. Dotted grey line: the short distance to the amino group in 8b. Structures redrawn from the names in the paper; shifts as published;[2] distances from our own force-field models.
Table 2 · What each method showed for the two isomers
| Method | Isomer 8a | Isomer 8b |
|---|---|---|
| Accurate mass of the protonated ion | m/z 261.1344 | m/z 261.1344 |
| Thin-layer chromatography, Rf | 0.11 | 0.30 |
| Retention time, reversed-phase column | 13.70 min | 20.03 min |
| N-methyl signal | 3.35 ppm | 3.48 ppm |
| NOESY | Cross-peak between 3.35 ppm (N-methyl) and 6.63 ppm (phenyl) | No cross-peak between these groups |
| HMBC from the N-methyl protons | To the ring carbon at 142.14 ppm, which carries the phenylamino group | To the ring carbon at 148.76 ppm, which carries the amino group |
| Fragments of the ion at m/z 215.0926 | No fragment at m/z 170.0600 | Fragment at m/z 170.0600 |
Values from the 2022 study.[2] The authors suggest that the additional fragment of 8b can form only where the N-methyl and amino groups are adjacent.
In our models the shortest distance between the N-methyl group and a phenyl proton is 2.7 Å in 8a and 3.8 Å in 8b. Averaged over all protons of the two groups, the NOE to the phenyl ring would be about seven times weaker in 8b, but not zero. The assignment of 8b therefore rests on the HMBC correlation, on the fragment ion of Table 2 and on the comparison with 8a, not on the missing cross-peak alone. Direct NOE evidence for 8b would be a cross-peak between the N-methyl and the amino protons, 2.6 Å apart in the model; the paper reports none. For a related pyrazole of the same study, NOESY signals connect the N-methyl group with both the phenyl ring and the NH proton.[2]
Indazoles raise the same question. Doganc and Göker at Ankara University reported in 2024 that the benzylation and alkylation of indazoles with potassium carbonate in DMF gave the N1 and N2 isomers in almost equal amounts; they assigned the structures by NOESY together with 1H–13C and 1H–15N HMBC.[21] In an N1-substituted indazole the new group lies next to H7 of the benzene ring, and after reaction at N2 beside H3. Part 5 describes the 15N experiment.
The ratio of the two isomers depends on the conditions. Alam and Keating at University College Cork reported in 2021 that methyl indazole-3-carboxylate and pentyl bromide gave the N1 and N2 products in a ratio of 1.4:1 with caesium carbonate in DMF, of more than 99:1 with sodium hydride in THF, and of 1:2.5 under Mitsunobu conditions with pentanol. Indazoles with a nitro or an ester group at C7 gave at least 96 % of the N2 isomer with sodium hydride. C3 of the ester carries no proton, so that an alkyl group on N2 has no NOE partner there (our remark); the authors assigned the products by HMBC. The NCH2 protons correlate with C7a in the N1 isomer and with C3 in its N2 counterpart.[22]
E or Z at a double bond without two neighbouring protons
When each carbon of a double bond carries one proton, the coupling constant between the pair usually gives the configuration (Part 15). A double bond with three or four substituents offers no such coupling. Endoxifen, an active metabolite of tamoxifen, has four. Its (Z)-isomer is potently anti-oestrogenic, and the (E)-isomer is an impurity of the drug substance that increases under certain storage conditions.[1]
A group at RTI International, MRIGlobal and the US National Cancer Institute characterised the drug substance for preclinical studies; the work appeared online in 2013. Earlier assignments of the two isomers were empirical: they relied on trends in the chemical shift of the OCH2 protons of related triarylethylenes.[1] The authors dissolved 11 to 12 mg of the hydrochloride in about 0.8 mL of DMSO-d6 and recorded 1H, 13C, HSQC, COSY and ROESY spectra at 500 MHz.
The assignment proceeded in four steps.[1]
- The 13C shifts and the HSQC spectrum identified the ethyl protons (H19 at 2.41 ppm, H20 at 0.85 ppm) and those of the OCH2 group (H5 at 4.11 ppm).
- The two rings with substituents in the para position give four doublets of similar shift. COSY paired them: H2 with H3, and H10 with H11.
- A ROESY cross-peak between H5 and the doublet at 6.64 ppm identified the latter as H3, and with it the ring that carries the side chain. H10 (6.97 ppm) and H11 therefore belong to the phenol ring.
- ROESY cross-peaks of H10 with H19 and H20 placed the ethyl group on the same side of the double bond as the phenol ring. This is the (Z)-configuration (Figures 5 and 6).
Figure 5. (Z)- and (E)-endoxifen. In the (Z)-isomer the dashed lines mark the two ROESY correlations of H10 (disc) with the ethyl group. In the (E)-isomer the ethyl group faces the other ring; the grey lines mark the contact with H2 expected there. Structures redrawn; shifts and correlations as published;[1] distances from our own force-field models.
Figure 6. The ROESY spectrum of (Z)-endoxifen hydrochloride as a diagram. Top left: the full spectrum. Top right: the aromatic protons against the aliphatic ones, enlarged. Bottom: the structure, numbered as in the paper, with the four correlations and the order in which they are read. Our drawing of the published shifts and correlations;[1] it is not the measured spectrum. The sizes of the peaks are schematic, and the signals of OH and NH above 9 ppm are omitted.
The paper describes the cross-peak of H10 with CH3 as somewhat stronger than that with CH2.[1] Our models agree: the shortest distance from H10 is 2.6 Å to the methyl and 3.3 Å to the methylene protons. In the (E)-isomer both distances are about 4.5 Å. Table 3 lists the contribution of each method.
Table 3 · What each method contributed for endoxifen
| Method | Result | What it left open |
|---|---|---|
| LC–MS with accurate mass | Two peaks, at 9.40 and 10.54 min, with the same formula, C25H27NO2 | The identity of each peak |
| HPLC with UV detection | Amounts of both isomers: the solution prepared for NMR contained 3.7 % of the (E)-isomer before the measurements and 7.4 % after them | The configuration of either peak |
| ROESY of the (Z)-isomer | H10 next to the ethyl group: (Z)-configuration | Nothing |
| ROESY of the (E)-isomer | No assignment: the signals of H2 and H10 overlap with others | The configuration; the authors identified the isomer by comparison with published NMR data and by its accurate mass |
Values from the study.[1] The authors concluded that the isomerisation during the measurements did not affect the NMR results.
α or β anomer of a nucleoside
Chemical syntheses of nucleosides often give both anomers. A review published in 2019 lists a fusion reaction from 1969 with yields of 14 and 16 % for the β- and α-anomers of 2′-deoxyguanosine, and a glycosylation from 1994 with an α:β ratio of 75:25.[23] The anomers are then separated, and the configuration of each fraction has to be assigned.
Rosemeyer, Tóth and Seela showed in 1989 that NOE difference spectroscopy assigns the anomeric configuration of ribo-, arabino-, 2′-deoxyribo- and 2′,3′-dideoxyribonucleosides: they saturated the signal of H1′ and measured the NOE at H4′, H3′ and H2′.[24] In 1992 the same method showed that the configurations published for three pairs of lumazine 2′-deoxyribonucleosides had to be reversed; crystal structures of two of the six compounds confirmed the revision.[3]
Kool, in a patent granted in 2000 and now assigned to the University of Rochester, used the same approach for the anomers of C-nucleosides in which naphthalene, phenanthrene, pyrene or a substituted benzene takes the place of the base. NOE difference spectra at 300 MHz gave two patterns, quoted here for the 1-naphthyl compounds. In the α-anomer, irradiation of one H2′ proton enhanced both H1′ and H3′, by 8 and 7 %; its geminal partner affected neither. In the β-anomer each H2′ proton enhanced one neighbour only, H1′ by 8 % or H3′ by 5 %, and irradiation of H1′ raised the H4′ signal by 6 %; the patent states, without data, that the α-anomer lacks this last enhancement. A crystal structure of the α-anomer confirmed its assignment.[25] Enhancements of this size are typical of the older difference experiment, which saturates a signal for several seconds; a selective 1D NOESY gives smaller ones.
Figure 7. The two anomers of a 2′-deoxynucleoside in Haworth form: substituents drawn upwards lie on one face of the ring, those pointing downwards on the other. Gold: H1′ and H4′. Blue: the short distances of the H2′ protons. Distances are our own calculation for thymidine and its α-anomer; the enhancements are those reported for the two 1-naphthyl C-nucleosides.[25]
Table 4 · Calculated distances in the two anomers of thymidine
| Proton pair | β-anomer | α-anomer |
|---|---|---|
| H1′ and H4′ | 3.0 to 3.4 Å | 3.8 to 4.0 Å |
| H2′β and H1′ | 3.1 Å | 2.4 Å |
| H2′β and H3′ | 2.4 Å | 2.4 Å |
| H2′α and H1′ | 2.4 Å | 3.1 Å |
| H2′α and H3′ | 2.7 Å | 3.1 Å |
Our own calculation with a force field: range for H1′ and H4′, median for the other pairs. H2′β lies on the face of H3′, H2′α on the opposite one. These are not measured values.
The calculation agrees with the pattern reported for the α-anomer (Figure 7, Table 4): one H2′ proton is 2.4 Å from both H1′ and H3′, and the other 3.1 Å. The step from 2.4 to 3.1 Å corresponds to a factor of about four in NOE intensity. For the β-anomer the agreement is partial: H2′α is 2.4 Å from H1′ but also 2.7 Å from H3′, which predicts a second, weaker enhancement that the patent does not report. The H1′–H4′ distance differs less between the anomers: 0.5 Å, or a factor of about two. A single enhancement is therefore a narrow basis, and the published procedures compare several.[24, 25]
When TOCSY is the preferable method
One compound in a mixture
In a selective 1D TOCSY experiment one resolved signal of a compound is excited, and the spin-lock passes its magnetisation to the other protons of the same spin system. The spectrum then contains that spin system only (Figure 8). The compound stays in the mixture: the experiment separates its signals, not the substance.
Figure 8. Principle of the selective 1D TOCSY experiment on a mixture of two compounds. The lower trace contains only the signals of compound A, including those that overlap with B in the full spectrum. Schematic; our drawing.
Sandusky and Raftery at Purdue University applied the experiment to commercial honeys in 2005. A very few species, α-glucose and fructose in particular, dominate the 1H spectrum of honey, so that a statistical comparison of samples reflects little else. Selective TOCSY gave access to minor components such as the amino acids, in an acquisition of about one minute, and allowed their quantification.[4]
Schievano, Tonoli and Rastrelli at the University of Padova extended the approach to sugars in 2017. A filter that selects one chemical shift, followed by a TOCSY step of 70 ms, gave a background-free signal for each of 22 sugars; four of them were below the limit of quantification in the honeys tested. The honey was dissolved at 240 mg/mL without further preparation, and one minor sugar took about 7 min at 500 MHz. The authors report a precision of 0.99 to 4.03 % (relative standard deviation), a bias of 0.4 to 4.2 % and limits of quantification of 0.03 to 0.4 g per 100 g of honey; the amounts were read from calibration curves.[26]
In 2024 Montgomery and co-workers combined the GEMSTONE excitation described above[16] with the removal of the multiplet structure, so that every signal of the chosen spin system appears as a singlet. The examples were mixtures of Cinchona alkaloids, used as catalysts in asymmetric synthesis, and of glucocorticoids.[27]
The sugar units of a glycoside
The ring protons of all sugar units of a glycoside resonate in one narrow region. The anomeric proton of each unit is usually resolved, and TOCSY from it reaches the other protons of the same ring and no further, because the oxygen of the glycosidic bond ends the transfer.
In 2017 Prakash and co-workers, at a beverage company and a contract research organisation, described rebaudioside IX, a steviol glycoside of Stevia rebaudiana with nine glucose units. They isolated about 4.5 mg from 18.3 g of a commercial extract and measured solutions in pyridine-d5 at 500 MHz, one of them about 2 mg in 180 µL.[28] The nine anomeric protons appear between 6.40 and 5.06 ppm (Figure 9). From each of them the authors traced the protons of the unit by a combination of COSY and 1D TOCSY; in units VIII and IX, H-4, H-5 and H-6 were assigned from the 1D TOCSY data.[28]
Figure 9. Rebaudioside IX. Top: the nine anomeric doublets, numbered as the glucose units I to IX of the paper, at their published shifts and coupling constants. Below: the protons of the units I and VII at their published shifts, drawn as the sub-spectra that a 1D TOCSY from each anomeric proton contains. Our drawing of published assignments;[28] line shapes and intensities are schematic.
In this study each experiment answered a different question.[28]
- 1D TOCSY, with COSY, assigned the protons of the nine glucose units.
- HMBC showed how the units are linked. A correlation from the anomeric proton of unit VII to C-6 of III established a 1→6 α-linkage, which the authors call rare for this class of compounds.
- NOESY gave the relative configuration of the diterpene core: correlations between H-14 and H-20, and between H-5 and both H-9 and H-18, place these groups on two faces of the ring system.
The configuration at each anomeric carbon followed from a coupling constant (Figure 10): 7.4 to 8.2 Hz for the eight β-linked units and 3.5 Hz for the single α-linked one.[28] The NOE distinguishes the two cases as well: in our models of the methyl glucosides, H1 is 2.4 to 2.7 Å from H3 and H5 in the β-anomer and 3.7 to 3.9 Å in its α counterpart.
Figure 10. A β- and an α-glucoside as chairs. In the β-anomer H1 and H2 are both axial, which gives the large coupling constant; in the α-anomer H1 is equatorial. Blue: H3 and H5, which share a face of the ring with H1 in the β-anomer. Coupling constants of rebaudioside IX as published;[28] angles and distances from our own force-field models of the two methyl glucosides. R is the rest of the molecule.
What only the NOE measures: distances and conformations in solution
The NOE between two protons falls with the sixth power of their distance.[10] A pair at 3 Å gives 4 % of the NOE within a CH2 group, whose protons are 1.76 Å apart; at 5 Å the value is 0.2 % (Figure 11; our calculation). Routine work sorts NOEs into three classes: strong for 2 to 3 Å, medium for 3 to 4 Å and weak above 4 Å.[19] The reference work on the effect is the monograph of Neuhaus and Williamson, which we name for further reading.[29]
Figure 11. Relative intensity of the NOE as a function of the distance, on a logarithmic scale. Our calculation from the sixth-power relation; the three classes[19] and the two distances between neighbouring protons[30] as published.
The same relation makes a measured distance insensitive to errors in intensity, because it follows from the sixth root of a ratio of two NOEs. An intensity error of 20 % shifts the distance by 3 %, and a factor of two by 12 % (our calculation).
A rigid molecule: strychnine
In 2011 Butts and co-workers at the University of Bristol tested the accuracy on strychnine, a rigid molecule with known geometry. Selective 1D NOESY at 500 MHz in CDCl3 and C6D6, with a mixing time of 500 ms, gave 55 distances of up to 4.5 Å from 123 NOE signals. They differed from the distances calculated by density functional theory by 3.3 % (0.09 Å) on average; the scale was set by the 1.76 Å between the protons of one CH2 group. A third sample was dissolved in DMSO-d6, and none was degassed.[30] The usual assumption for NOE distances is an error of about ±10 %.[19] The difference matters for stereochemistry: two neighbouring protons are 2.55 Å apart in a gauche and 3.0 Å in an anti arrangement.[30]
Figure 12. Strychnine, drawn with the hydrogen atoms of its five CH stereocentres, and the mean errors of its interproton distances from six experiments at 500 MHz. Our drawing of published values.[30] The value for 1D NOESY combines the measurements in CDCl3 and C6D6; the 2D NOESY values refer to CDCl3. Grey bars: the two conditions with the largest errors.
The accuracy depends on the conditions. 2D NOESY in CDCl3 gave a mean error of 3.8 % with a delay of 5 s between scans and of about 8.3 % with 1 s. In DMSO, the most viscous of the three solvents of the study, 1D NOESY gave 6.9 % at a mixing time of 500 ms and 4.5 % at 200 ms. 1D ROESY with a spin-lock of 200 ms gave 3.2 % in CDCl3 and 4.8 % in DMSO; at 500 ms in DMSO the error rose to 6.7 % (Figure 12).[30]
An accuracy of 3 % reveals details that the three classes hide. One distance across the seven-membered ring of strychnine was about 0.6 Å, or 15 %, shorter than the crystal structure and calculation predict. The same group traced this deviation to a second conformer of low population: 2.5 % according to the abstract of the report, about 2.2 % in a later paper. With that conformer included, the error of the distance fell from 15 to 3 %.[19, 31]
Distances of this accuracy have since served in stereochemical assignments. In 2012 Butts, Jones, Song and Simpson established the relative configuration of arugosin C, a flexible molecule, 39 years after the compound was first reported.[32] In the same year a group from Salerno, Naples, Perugia and Bristol combined interproton distances measured by NMR with calculated 13C shifts to determine the configuration of conicasterol F, a nuclear receptor ligand from the marine sponge Theonella swinhoei.[33]
A flexible molecule: 4-propylaniline
The propyl chain of 4-propylaniline rotates between one anti and two gauche conformers, and each NOE is an average over them. In 2011 the Bristol group measured three averaged distances with selective 1D NOESY on about 10 mg in CDCl3; each irradiation took 45 min. The sequence was the gradient-based one named above,[18] extended by a filter against zero-quantum signals that Thrippleton and Keeler had published in 2003.[34] A fourth distance, taken from the calculation, set the scale. The measured values agree with those calculated for the mixture of conformers, with a mean error of 2.7 % (Figure 13).[19]
Figure 13. 4-Propylaniline with the proton numbers used for the distances, and its conformers about the bond between the two CH2 groups (gold) as Newman projections. Populations, distances and the range of 40 to 65 % as published;[19] the drawing is ours.
The NOE data fitted best with about 55 % of the anti conformer; calculations gave 47 to 49 %. With distances accurate to 3 %, the data are compatible with 40 to 65 %; under the usual assumption of ±10 %, any population would fit.[19] In a flexible molecule the NOE therefore describes a population of conformers, not one structure.
NOESY or ROESY: the molecular weight sets the choice
The NOE in a NOESY spectrum changes sign with the correlation time. It is positive for small molecules, negative for large ones, and zero where the product of the angular spectrometer frequency and the correlation time is about 1.12.[14] In ROESY the effect keeps one sign throughout.[7] The upper part of Figure 14 shows the curve for the steady-state experiment, in which one signal is saturated. The NOE of a NOESY spectrum is smaller,[6] and it changes sign at the same value.
Figure 14. Top: the maximum steady-state NOE between two protons as a function of the product of angular spectrometer frequency and correlation time; our calculation with the textbook expression for an isolated proton pair, which reproduces the limits of +50 and −100 %[6, 35] and the zero at 1.12[14] given in the sources. Bottom: molecular weights at which five sources expect a weak or absent NOE in a NOESY spectrum; our drawing of published statements: Columbia University;[7] University of Chicago;[10] University of Cincinnati;[20] Bax and Grzesiek;[14] Australian National University.[6] The sources state neither field nor solvent for these values.
The zero lies at a correlation time of 0.45 ns at 400 MHz and of 0.30 ns at 600 MHz (our calculation). A higher field, a more viscous solvent and a lower temperature each move a given molecule towards the negative side. This probably explains why the sources place the range of weak NOE at different molecular weights (Figure 14, bottom); the explanation is ours. Table 5 gives the recommendations of the Columbia guide.
Table 5 · Choice of experiment and first mixing time by molecular weight
| Molecular weight and NOE | Recommended experiment | Mixing time |
|---|---|---|
| Below 600 g/mol: positive NOE | 2D NOESY with zero-quantum suppression | 0.5 to 1 s; first trial 0.5 s |
| 700 to 1200 g/mol: the NOE passes through zero | ROESY or EASY ROESY | 0.1 to 0.5 s; first trial 0.25 s |
| Above 1200 g/mol: negative NOE | Usually 2D NOESY | 0.05 to 0.2 s; first trial 0.1 s |
Guide of the NMR facility of Columbia University, 2018, which describes the limits as approximate.[7] The procedure for ROESY in the same guide starts with a spin-lock of 0.2 s.
The sources disagree on small molecules. The Columbia guide states that ROESY has only disadvantages below 600 g/mol.[7] A technical note of a contract development company recommends ROESY for all but large molecules,[35] and the endoxifen study applied it to a compound of 373 g/mol (free base).[1] The Columbia guide names four drawbacks: TOCSY peaks between coupled protons appear as artefacts, the sensitivity for large molecules is lower than in NOESY, the intensity of a cross-peak depends on its distance from the centre of the spectrum, and TOCSY transfer can partly cancel a ROESY cross-peak, which harms distance measurements.[7] Bax and Davis had described the first and the third in 1985, on a cyclic tripeptide at 270 MHz: the TOCSY-type peaks have the sign of the diagonal, opposite to that of the ROE signals, and a weak spin-lock field with a suitably placed transmitter frequency reduces them.[13] In 2009 Thiele, Petzold and Schleucher presented EASY ROESY, a version designed to give reliable cross-peak integrals without adjustment to the sample;[36] the Columbia guide offers it as an alternative and states that it minimises the TOCSY artefacts.[7] Part 8 lists the signs of the cross-peaks in both experiments.
What a spectrum does not prove
Each of the three experiments can show a cross-peak, or lack one, for a reason other than the structure (Table 6).
Table 6 · Observations that are easily over-interpreted
| Observation | Why it is not conclusive | Check |
|---|---|---|
| No NOESY cross-peak between two protons | The NOE is close to zero at intermediate molecular weight; dissolved oxygen reduces it; a short mixing time or a low concentration leaves it below the noise | ROESY; a second, independent correlation such as HMBC |
| A NOESY cross-peak in a flexible molecule | The NOE is an average over all conformers, and one with a short distance contributes most | Treat the result as a population, not as one geometry |
| A cross-peak between distant protons of a large molecule | At long mixing times the NOE is relayed through a third proton; in NOESY the relayed peak has the sign of a direct one | A shorter mixing time; ROESY, in which a relayed peak has the opposite sign |
| A cross-peak with the sign of the diagonal, small molecule | Chemical exchange gives such peaks in NOESY and ROESY; in ROESY, TOCSY artefacts have the same sign | See Part 8 |
| Cross-peaks between coupled protons in NOESY | They may be zero-quantum peaks, which have positive and negative parts | NOESY with zero-quantum suppression |
| No TOCSY cross-peak between two groups | An ether oxygen, an amide bond or a small coupling ends the transfer, so that one molecule can contain several spin systems | HMBC or NOESY across the break |
| Identical NOESY and TOCSY spectra of two samples | Enantiomers give identical spectra | A chiral auxiliary (Part 12) |
Rows 1, 3, 4 and 5 follow the Columbia guide;[7] row 2 follows the study of 4-propylaniline and the sixth-power relation;[19] row 6 follows the course material on TOCSY.[5] Row 7, the last cause named in row 1 and the checks in rows 1 and 6 are our reasoning. The suppression of row 5 in current use is the single-scan element of Thrippleton and Keeler.[16, 34]
Overlap is a further limit. An NOE between protons whose signals coincide with others cannot be assigned; for this reason the (E)-isomer of endoxifen was not confirmed by NMR alone.[1] The GEMSTONE excitation was developed for such cases.[16]
Practical points
Table 7 · Settings and times reported in the sources
| Quantity | Value | Source |
|---|---|---|
| TOCSY mixing time | 20 ms for about one strong step; 80 to 100 ms for five to eight steps; 120 ms at most; 70 ms in the method for sugars | Facility guide, course material, 2017 study[5, 8, 26] |
| NOESY mixing time, small molecule | 0.5 to 1 s; 0.5 s for the distances in strychnine; 0.3 s as the standard of one facility | Two facility guides, 2011 study[7, 20, 30] |
| ROESY spin-lock | 0.1 to 0.5 s; 0.2 s at the start of one procedure and for the distances in strychnine | Facility guide, 2011 study[7, 30] |
| Time for a selective 1D NOESY | About ten minutes for a significant result; 12 min per proton in the oestradiol example; 45 min per irradiation for accurate distances on 10 mg | Facility guide, 2021 and 2011 studies[10, 16, 19] |
| Time for a 2D spectrum | NOESY about 50 min and ROESY about 48 min as the standards of one facility; 2.5 to 12 h for NOESY at another | Two facility guides[10, 20] |
| Time for a selective 1D TOCSY | About one minute per spectrum of honey; about 7 min per minor sugar | 2005 and 2017 studies[4, 26] |
| Delay between scans, for distances | At least three times T1, the relaxation time explained in Part 9 | Facility guide[7] |
A selective 1D experiment or a 2D spectrum
The Columbia guide recommends a selective 1D NOESY for a concentrated sample of a small molecule when one or two NOEs are needed and the signals to be excited are well separated; otherwise it prefers the 2D spectrum.[7] The Chicago guide suggests the 2D spectrum when more than five or six protons are of interest.[10] For distance measurements the Bristol group recommends 1D rather than 2D data.[30]
Amount of sample
NOE signals are typically 0.1 to 1 % of regular peaks, so the experiment needs more material than a 1H spectrum.[10] The amounts in the sources cited here were 10 mg of carvone in 1 mL (the example of a guide),[10] 11 to 12 mg of endoxifen in 0.8 mL,[1] 30 mg of strychnine in 0.7 mL of CDCl3 and 6 to 7 mg in each of the two other solvents,[30] about 10 mg of 4-propylaniline in 0.7 mL[19] and, for the glycoside, about 2 mg in 180 µL or about 5 mg in 700 µL.[28] A normal 5 mm tube needs about 500 µL of solution; for smaller volumes the Columbia guide names Shigemi tubes.[7]
Dissolved oxygen
Oxygen is paramagnetic and reduces the NOE. The Columbia guide calls its removal extremely important for small molecules and prescribes three to four freeze–pump–thaw cycles, because bubbling argon through the solution is not sufficient.[7] The guide of the Australian National University recommends the same technique.[6] The Bristol group measured without degassing and concluded that it is not strictly necessary,[30] and the Chicago guide reserves it for careful distance measurements.[10] Without oxygen the relaxation times become longer, and the delay between scans has to grow with them; this remark is ours. A degassed sample must stay closed during the measurement; Part 6 describes tubes with a PTFE valve.
Delay between scans
A short delay between scans distorts the intensities: the mean error of the 2D NOESY distances in strychnine rose from 3.8 to about 8.3 % when the interval was cut from 5 s to 1 s.[30] Where the only question is whether a cross-peak is present, the standard delay of a facility is sufficient; this last statement is our reasoning.
Summary
- TOCSY connects all protons of one spin system; the transfer ends where the chain of couplings is broken, and its reach grows with the mixing time, up to 120 ms.
- NOESY and ROESY connect protons that are closer than 4 to 5 Å, whatever the bonds between them.
- A NOESY cross-peak showed which nitrogen atom of a pyrazole was methylated; the two isomers have the same accurate mass, and their retention times do not identify them.
- The N-alkylation of an indazole gave its two isomers in ratios from 1.4:1 to more than 99:1, depending on base and solvent; HMBC assigns them where the NOE has no partner.
- Two ROESY cross-peaks established the (Z)-configuration of endoxifen on 11 to 12 mg of substance; LC–MS gave both isomers the same formula, and HPLC measured their amounts.
- NOE measurements assign the α- or β-configuration of a nucleoside and reversed a published assignment in 1992. The H1′–H4′ distance differs by only about 0.5 Å between the anomers, so several NOEs are compared.
- Selective 1D TOCSY gives the signals of one compound in a mixture without separation; a published method covers 22 sugars in honey with a precision of 1 to 4 %.
- In a glycoside with nine glucose units, TOCSY assigned the protons of each ring, HMBC gave the linkages and NOESY the configuration of the core.
- In a rigid molecule the NOE gives distances with a mean error of about 3 %; in a flexible molecule the result is an average over the conformers.
- Between about 700 and 1200 g/mol the NOE in NOESY is weak or absent, and ROESY is used; the sources place this range differently.
- A missing cross-peak is weak evidence, and none of the three experiments distinguishes enantiomers.
NOESY, ROESY and TOCSY from NorrChemica
Within our NMR service NOESY, ROESY and TOCSY are quoted per sample. The packages are measured at 400 MHz; 600 MHz and above are available on request. Package VI (Characterisation: 1H, 13C, DEPT and gCOSY) and Package VII (Full elucidation: with gHSQC, gHMBC and a full assignment) are quoted in the same way, and Package HRMS-02 combines LC-HRMS with 1H and 13C NMR. A chemist assigns each spectrum, and the signed report includes the native and processed data files. Please state the question on the request form, for example which of two isomers the sample is. The service is available worldwide. Research use only.
- 5 mm NMR tubes with PTFE valvefor degassed samples
- 5 mm NMR tubes with constrictionfor flame sealing after degassing
- 3 mm NMR tubesfor small amounts of sample
- Shigemi NMR tubes, solvent-matchedfor small volumes
- Thymidinethe nucleoside of Figures 2 and 7; see also our nucleosides and nucleobases
NMR tubes made in Germany by Deutero and shipped from Finland.
References
- Elkins P, Coleman D, Burgess J, Gardner M, Hines J, Scott B, Kroenke M, Larson J, Lightner M, Turner G, White J, Liu P. Characterization of the isomeric configuration and impurities of (Z)-endoxifen by 2D NMR, high resolution LC–MS, and quantitative HPLC analysis. J. Pharm. Biomed. Anal. 2014, 88, 174–179. DOI
- Lusardi M, Profumo A, Rotolo C, Iervasi E, Rosano C, Spallarossa A, Ponassi M. Regioselective synthesis, structural characterization, and antiproliferative activity of novel tetra-substituted phenylaminopyrazole derivatives. Molecules 2022, 27, 5814. DOI
- Cao X, Pfleiderer W, Rosemeyer H, Seela F, Bannwarth W, Schönholzer P. Nucleosides. Part L. Structure of lumazine N1-(2′-deoxy-D-ribonucleosides): a revision of the anomeric configuration. Helv. Chim. Acta 1992, 75, 1267–1273. DOI
- Sandusky P, Raftery D. Use of selective TOCSY NMR experiments for quantifying minor components in complex mixtures: application to the metabonomics of amino acids in honey. Anal. Chem. 2005, 77, 2455–2463. DOI
- University of Wisconsin–Madison, Department of Chemistry. Chem 636, Homework 9: TOCSY and HSQC. Course material, updated 3 November 2014. Link
- Australian National University, NMR Centre. Guide to NOE experiments. Web page, undated. Link
- Decatur J. NOESY and ROESY. Columbia University, NMR Core Facility, user guide, version 5.1, 8 August 2018. Link
- Stanford University, Department of Chemistry, NMR Facility. TOCSY. Help document, undated. Link
- Braunschweiler L, Ernst RR. Coherence transfer by isotropic mixing: application to proton correlation spectroscopy. J. Magn. Reson. 1983, 53, 521–528. DOI
- Kurutz J. 1D NOESY made easy. University of Chicago, Chemistry Department, NMR Facility, 10 March 2021. Link
- Jeener J, Meier BH, Bachmann P, Ernst RR. Investigation of exchange processes by two-dimensional NMR spectroscopy. J. Chem. Phys. 1979, 71, 4546–4553. DOI
- Bothner-By AA, Stephens RL, Lee J, Warren CD, Jeanloz RW. Structure determination of a tetrasaccharide: transient nuclear Overhauser effects in the rotating frame. J. Am. Chem. Soc. 1984, 106, 811–813. DOI
- Bax A, Davis DG. Practical aspects of two-dimensional transverse NOE spectroscopy. J. Magn. Reson. 1985, 63, 207–213. DOI
- Bax A, Grzesiek S. ROESY. In: Encyclopedia of Magnetic Resonance. John Wiley & Sons, 2007. DOI
- Cui Q, Lewis I, Jofre F, Anderson ME, Markley JL. Thymidine. Biological Magnetic Resonance Data Bank, metabolomics entry bmse000244, deposited 23 February 2006. DOI
- Kiraly P, Kern N, Plesniak MP, Nilsson M, Procter DJ, Morris GA, Adams RW. Single-scan selective excitation of individual NMR signals in overlapping multiplets. Angew. Chem. Int. Ed. 2021, 60, 666–669. DOI
- Stott K, Stonehouse J, Keeler J, Hwang T-L, Shaka AJ. Excitation sculpting in high-resolution nuclear magnetic resonance spectroscopy: application to selective NOE experiments. J. Am. Chem. Soc. 1995, 117, 4199–4200. DOI
- Stott K, Keeler J, Van QN, Shaka AJ. One-dimensional NOE experiments using pulsed field gradients. J. Magn. Reson. 1997, 125, 302–324. DOI
- Jones CR, Butts CP, Harvey JN. Accuracy in determining interproton distances using nuclear Overhauser effect data from a flexible molecule. Beilstein J. Org. Chem. 2011, 7, 145–150. DOI
- University of Cincinnati, Department of Chemistry, NMR Facility. NMR experiment guide. Web page, undated. Link
- Doganc F, Göker H. Differentiation of regioisomeric N-alkylation of some indazoles and pyrazolopyridines by advanced NMR techniques. Magn. Reson. Chem. 2024, 62, 765–774. DOI
- Alam RM, Keating JJ. Regioselective N-alkylation of the 1H-indazole scaffold; ring substituent and N-alkylating reagent effects on regioisomeric distribution. Beilstein J. Org. Chem. 2021, 17, 1939–1951. DOI
- Ni G, Du Y, Tang F, Liu J, Zhao H, Chen Q. Review of α-nucleosides: from discovery, synthesis to properties and potential applications. RSC Adv. 2019, 9, 14302–14320. DOI
- Rosemeyer H, Tóth G, Seela F. Assignment of anomeric configuration of D-ribo-, arabino-, 2′-deoxyribo-, and 2′,3′-dideoxyribonucleosides by NOE difference spectroscopy. Nucleosides Nucleotides 1989, 8, 587–597. DOI
- Kool ET. Nucleoside analogs with polycyclic aromatic groups attached, methods of synthesis and uses therefor. US Patent 6,140,480, granted 31 October 2000; the same description appears in US Patent 6,218,108. Link
- Schievano E, Tonoli M, Rastrelli F. NMR quantification of carbohydrates in complex mixtures. A challenge on honey. Anal. Chem. 2017, 89, 13405–13414. DOI
- Montgomery JRD, Gates EL, Smith MJ, Taylor DA, Bradley JP, Berry DBG, Kiraly P, Nilsson M, Morris GA, Adams RW, Castañar L. Ultraselective, ultrahigh resolution 1D TOCSY. Chemistry–Methods 2024, 4, e202400013. DOI
- Prakash I, Ma G, Bunders C, Charan RD, Ramirez C, Devkota KP, Snyder TM. A novel diterpene glycoside with nine glucose units from Stevia rebaudiana Bertoni. Biomolecules 2017, 7, 10. DOI
- Neuhaus D, Williamson MP. The Nuclear Overhauser Effect in Structural and Conformational Analysis, 2nd ed. John Wiley & Sons, New York, 2000.
- Butts CP, Jones CR, Towers EC, Flynn JL, Appleby L, Barron NJ. Interproton distance determinations by NOE – surprising accuracy and precision in a rigid organic molecule. Org. Biomol. Chem. 2011, 9, 177–184. DOI
- Butts CP, Jones CR, Harvey JN. High precision NOEs as a probe for low level conformers – a second conformation of strychnine. Chem. Commun. 2011, 47, 1193–1195. DOI
- Butts CP, Jones CR, Song Z, Simpson TJ. Accurate NOE-distance determination enables the stereochemical assignment of a flexible molecule – arugosin C. Chem. Commun. 2012, 48, 9023–9025. DOI
- Chini MG, Jones CR, Zampella A, D’Auria MV, Renga B, Fiorucci S, Butts CP, Bifulco G. Quantitative NMR-derived interproton distances combined with quantum mechanical calculations of 13C chemical shifts in the stereochemical determination of conicasterol F, a nuclear receptor ligand from Theonella swinhoei. J. Org. Chem. 2012, 77, 1489–1496. DOI
- Thrippleton MJ, Keeler J. Elimination of zero-quantum interference in two-dimensional NMR spectra. Angew. Chem. Int. Ed. 2003, 42, 3938–3941. DOI
- Kornev A. Meet the family: NOE and selective excitation experiments. Technical note, Almac Sciences, Craigavon, online document, undated (file path of 2026). Link
- Thiele CM, Petzold K, Schleucher J. EASY ROESY: reliable cross-peak integration in adiabatic symmetrized ROESY. Chem. Eur. J. 2009, 15, 585–588. DOI
About this article. All sources were read on 18 November 2026. Read in the browser as full text, without the images of their figures: sources 1, 16, 19 and 22 in full; of source 2 the sections on the synthesis and on the NMR and mass spectra of the two isomers, with the experimental data; source 28 in full, apart from its three tables, which the page shows as images; of source 23 the passages on the syntheses quoted here. Source 30, a scanned copy of source 13 on a server of the US National Institutes of Health and the accepted manuscript of source 26 in the repository of the University of Padova were read through a retrieval tool that returns extracts and answers to questions about a document; the numbers taken from source 30 were queried several times with the same result. The guides and web documents 5, 6, 7, 8, 10, 20 and 35 and the encyclopedia chapter 14 were read through the same tool, and so was the data file of the database entry 15. Of sources 31, 32 and 36 we read the abstracts only. Sources 3, 4, 21, 24 and 27 are known to us from their abstracts or bibliographic records; for 9, 11, 12, 17, 18, 33 and 34 we had the latter alone, and what we state about these seven follows from their titles, from the sources that cite them (13, 16 and 19) and from textbook knowledge. The monograph 29 was not read; source 19 cites it six times. The description of the patent 25 was read through the retrieval tool; its two tables were not legible there. The bibliographic data of the journal articles were compared with the Crossref record of each article; the places of work of the authors of sources 11, 21 and 33 are taken from those records. Source 35 carries no date; the year in the reference is that of its file path. Source 2 gives the yields of the two isomers as 70 and 30 %, in opposite order in two places, and we do not quote them. The abstract of source 31 and the text of source 19 give different populations for the second conformer of strychnine, and we quote both. That source 27 uses the GEMSTONE excitation is stated in its abstract. Our own work: Figures 1, 3 and 8 as drawings, with the pulse sequences of Figure 1 as textbook knowledge; Figure 2 as a drawing of published shifts, with diagrams of the cross-peaks we expect and not of measured spectra, and with the weak coupling between H6 and the methyl group of thymine as general knowledge; all structures, drawn from the names of the compounds or, for strychnine, from the record of a public database; Figures 6 and 9 as diagrams of published shifts; the remark that the 1H spectra of the two pyrazoles do not identify them; the positions of the neighbouring protons in N1- and N2-substituted indazoles, and the remark that an alkyl group on N2 of the indazole ester has no NOE partner at C3; the statement that H10 of endoxifen is the ring proton next to the double bond, which follows from the 13C shift of its carbon in the paper; the assignment of the numbers H3, H5, H6 and H7 of 4-propylaniline to its protons, which follows from the text of the paper, whose labelled structure we did not see; the molecular weights of endoxifen and strychnine; the remark on the size of the enhancements in NOE difference spectra and the statement that the NOE of a NOESY spectrum changes sign at the same value as the steady-state NOE, as textbook knowledge; the comparison of our calculated distances with the enhancements of source 25; the viscosity of DMSO compared with chloroform and benzene, and Theonella swinhoei as a marine sponge, as general knowledge; the remark on the relaxation times of a degassed sample; the sixth-power calculations; the curve of Figure 14 and the correlation times at its zero; and all calculated distances and angles. For these, 300 conformers of each molecule were generated and minimised with the MMFF94 force field, and those within 3 kcal/mol of the lowest were evaluated; the result is a model, not a measurement.
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
- 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 (this article)
- 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™ · Content may not be reproduced without written permission