Two sets of signals in the NMR spectrum: rotamers, diastereomers or an impurity, and how to prove which
Share
Two sets of signals in the NMR spectrum: rotamers, diastereomers or an impurity, and how to prove which
A second set of signals does not always mean a second compound. An exchange experiment at room temperature, and spectra at other temperatures, show which case applies
The spectrum of a purified compound sometimes shows every signal twice, for example in a ratio of 3:1. There are two explanations. Either the sample contains a second compound, which may be an impurity or a diastereomer, or it contains one compound in two forms that change into each other slowly, such as rotamers.
The difference matters, because a mixture has to be separated and a pair of rotamers cannot be. This article shows how to decide by NMR. The direct evidence is an exchange peak, which appears only between the signals of two forms that interconvert. The second tool is temperature, because temperature sets the rate of the interconversion and therefore determines what the spectrum shows.
Six terms used in this article. Rotamers are forms of one molecule that differ by rotation about a single bond, such as the C–N bond of an amide. Diastereomers are stereoisomers that are not mirror images, and they are different compounds. Chemical exchange is the passage of a nucleus between two environments, here between the two rotamers. Exchange is slow when each form gives its own signal, and fast when the two signals have merged into one average signal. Coalescence is the point between the two, at which the signals merge. The barrier (ΔG‡) is the free energy of activation of the process, in kcal/mol, and it determines the rate at each temperature.
At a glance: the observation and its cause
| Observation | Cause | Comment |
|---|---|---|
| Every signal is doubled, with exchange peaks between the two sets | One compound in two forms that interconvert: rotamers, or E and Z isomers of a conjugated double bond | Usual for amides and carbamates.[1, 2, 3, 4] |
| Every signal is doubled, with no exchange peaks between the two sets | Two compounds: diastereomers or an impurity. Or rotamers with a barrier above about 19 kcal/mol | See “When no exchange peak appears”. |
| Small extra signals, with exchange peaks to the main signals | A minor form of the same compound, not an impurity | Shown for a ratio of 17.6:1.[4] |
| Broad signals at room temperature | Exchange close to coalescence | The spectrum sharpens into one set on heating and into two sets on cooling. |
| One clean set at room temperature, two sets at low temperature | Rotamers in fast exchange | N-Aryl carbamates, at 12 kcal/mol and below.[3] |
| Two sets merge on heating | Rotamers, or two compounds whose signals coincide at that temperature | Not a proof on its own.[2] |
| An exchange peak between an NH or OH signal and water | A labile proton that exchanges with water | It is exchange, but not between rotamers.[5] |
| A peak with the sign of exchange between NH and the neighbouring CH | Scalar cross-relaxation, not exchange | Expected when the NH signal is broad. See the section on signals that imitate exchange.[6] |
| One set of signals, but a CH2 group in a molecule without a stereocentre shows two different protons | Slow rotation about an axis | The two forms are mirror images, so they have the same spectrum. See the section on atropisomers.[7] |
Why one compound gives two sets of signals
In an amide or a carbamate, rotation about the bond between the carbonyl carbon and the nitrogen is hindered, so the molecule exists as two rotamers with different chemical shifts.[3, 8] Whether the spectrum shows them separately depends on a comparison of two numbers: the rate of the rotation, in s−1, and the distance between the two signals, in Hz.
If one rotamer changes into the other much less often than this distance, each gives its own signal. If it changes much more often, only the average is observed, and one signal appears. The two signals merge when the rate constant reaches 2.22 times their distance, k = πΔν/√2, where k is the rate constant for the change of one rotamer into the other.[9, 10] Figure 1 shows the line shapes for a pair of signals 40 Hz apart.
Figure 1. Calculated signal of one proton in two rotamers of equal population, 40 Hz apart, at five rate constants of exchange. Each trace is scaled to its own height, and the natural line width is 1 Hz. The calculation is ours.
The distance is counted in Hz, so it differs from one pair of signals to the next within the same spectrum. A pair that is 10 Hz apart merges at a rate at which a pair 200 Hz apart is still well separated. For this reason a sample close to coalescence often shows some signals doubled and others only broad.
The rate follows from the barrier to rotation and the temperature. Table 1 gives the values at 25 °C, and Figure 2 shows some measured barriers on the same scale.
Table 1 · Barrier, rate constant and spectrum at 25 °C
| Barrier ΔG‡, kcal/mol | Rate constant at 25 °C | Half-life at 25 °C | Spectrum at 25 °C for two signals 40 Hz apart |
|---|---|---|---|
| 12 | 9,900 s−1 | 0.07 ms | One sharp signal at the average position |
| 14 | 340 s−1 | 2 ms | One broad signal, about 9 Hz wide |
| 15 | 63 s−1 | 11 ms | Two broad signals that overlap |
| 16 | 12 s−1 | 60 ms | Two signals, each broadened by about 4 Hz |
| 18 | 0.40 s−1 | 1.7 s | Two sharp signals |
| 20 | 0.014 s−1 | 51 s | Two sharp signals |
| 22 | 0.00046 s−1 | 25 min | Two sharp signals |
Barrier, rate and appearance of the spectrum at 25 °C, for two forms of equal population. Calculated with the Eyring equation and a transmission coefficient of 1. The calculation is ours.
Figure 2. The barrier to rotation and what the 1H spectrum shows at 25 °C for two signals 40 Hz apart. The limits between the regions are calculated, and the compounds are shown at their measured barriers.[3, 8, 11] The calculation is ours.
Barriers of about 15 to 16 kcal/mol cause most of the confusion, because the signals are then broad or doubled at room temperature, and a few degrees change their appearance.
The direct proof: an exchange peak
The NOESY experiment contains a waiting period, the mixing time, which is 0.5 to 1 s for a small molecule.[12] If a nucleus moves from one rotamer to the other during this period, the spectrum shows a cross-peak between its two signals. Used for this purpose, the experiment is called EXSY (exchange spectroscopy). The pulse sequence is the same.[5]
Exchange peaks and NOE peaks are distinguished by their sign. In a small molecule an exchange peak has the same sign as the diagonal, and an NOE peak has the opposite sign (Table 2).[5, 12, 13] Figure 3 shows the pattern that is expected for two rotamers.
Table 2 · Sign of cross-peaks relative to the diagonal
| Cross-peak | NOESY, small molecule | NOESY, large molecule | ROESY, any size |
|---|---|---|---|
| NOE or ROE (two protons close in space) | Opposite to the diagonal | Same as the diagonal | Opposite to the diagonal |
| Chemical exchange | Same as the diagonal | Same as the diagonal | Same as the diagonal |
Sign of cross-peaks relative to the diagonal.[12]
Figure 3. The NOESY spectrum of a compound with two rotamers, read as an exchange spectrum. Filled circles have the sign of the diagonal, and open circles the opposite sign. Schematic.
The result is read as follows. If each signal of the minor set has an exchange peak to its counterpart in the major set, the two sets belong to one compound. If two sets show no exchange peak between them, the two species do not interconvert during the mixing time.
The published test case
The use of this test for the present question was described in 2012.[1] An ester of N-Boc-N-methylvaline showed the proton next to the nitrogen as two signals of equal intensity, at 4.59 and 4.28 ppm in CDCl3. Two explanations were possible: two rotamers of the carbamate, or two diastereomers formed by epimerisation during the coupling reaction.
The signal at 4.59 ppm was excited selectively in a 1D gradient NOE experiment. A second signal appeared at 4.28 ppm with the same phase as the excited signal, whereas an NOE signal has the opposite phase. The two signals therefore belong to one compound.[1]
The authors then prepared the other diastereomer deliberately and measured a mixture of both (Figure 4). The mixture showed four signals. Excitation of signal I gave a second signal only at III, and excitation of signal II gave a second signal only at IV. Magnetisation was transferred between the two rotamers of each diastereomer, and not from one diastereomer to the other.[1]
Figure 4. The selective exchange experiment on a mixture of two diastereomers, each present as two rotamers. Drawn from the published chemical shifts and coupling constants.[1] Signal heights are not to scale. Schematic.
The spectra were recorded at 600 MHz with a selective Gaussian pulse of 100 ms and a mixing time of 1.2 s. The sample is not heated and nothing is added to it, so it can be recovered unchanged.[1]
A selective pulse can also excite a neighbouring signal directly. The same study gives a control for two signals only 0.06 ppm apart at 600 MHz, which is 36 Hz. After the first experiment, the second signal was excited. Its partner appeared again, but a third signal of the sample, at the same distance on the other side, did not. The pulse had therefore not excited the neighbour directly.[1]
The 2D experiment
The 2D experiment gives all pairs at once. Esters of a hydroxy β-lactam with N-Boc-proline were obtained as two diastereomers, and each diastereomer showed two rotamers. The exchange spectrum at 25 °C in CDCl3, with a mixing time of 0.5 s at 400 MHz, gave cross-peaks with the sign of the diagonal between the two rotamers of each diastereomer, and each diastereomer had its own set of cross-peaks.[2]
The selective 1D experiment can also separate the two kinds of information. In a published example, a mixing time of 0.2 s gave only the NOEs within the excited rotamer, and a mixing time of 2 s also gave the corresponding signal of the other rotamer, with the same sense as the excited signal.[13] The short experiment is therefore used to assign the signals of each rotamer, and the long one to detect the exchange.
What an exchange peak proves
An exchange peak proves that two species interconvert during the mixing time. It does not identify the kind of isomer. Rotamers are the common case. The E and Z isomers of an enaminone or an enamino ester are formally diastereomers, but conjugation lowers the barrier to rotation about the C=C bond, and the two isomers interconvert at room temperature.[4] One such compound showed the two isomers in a ratio of 2.5:1 in CDCl3, with exchange peaks in both the 1D and the 2D experiment. A mixture of cis- and trans-stilbene, measured as a control, showed none.[4]
The same study contains an example of the third explanation in the title. A compound had been reported as a single isomer, and small extra signals in its spectrum had been attributed to impurities. The 1D exchange experiment connected these signals to the main signals. They belong to the minor isomer, which is present in a ratio of 17.6:1 (Figure 5).[4]
Figure 5. A small extra signal that is not an impurity. In the upper trace the two signals are drawn in the published ratio of 17.6:1. In the lower trace the main signal is excited, and a second signal of the same phase appears at the position of the small one. Drawn from the published shifts and ratio.[4] Schematic.
Exchange by a chemical reaction gives the same kind of peak. E and Z isomers can also interconvert through protonation and deprotonation, and the authors consider this route unlikely for their compound, because the ratio was the same in CDCl3 treated with K2CO3.[4] The 2012 study likewise assumed that epimerisation in neutral chloroform is negligible during the experiment.[1]
When no exchange peak appears
The test needs an exchange that is slow enough to give two separate signals and fast enough to transfer magnetisation during the mixing time.[1] With a mixing time of about 1 s, this requires a rate constant of roughly 0.1 s−1 or more. This limit is our estimate: about 10% of the magnetisation is then transferred. At 25 °C a rate constant of 0.1 s−1 corresponds to a barrier of 18.8 kcal/mol (Figure 2). For a higher barrier the rotation is too slow, and the spectrum at 25 °C shows no exchange peak although the compound is a pair of rotamers.
A missing exchange peak is therefore not a proof of two compounds.
Warming the sample makes a slower exchange detectable: for a barrier of 20 kcal/mol, the calculated rate constant rises from 0.014 s−1 at 25 °C to 0.20 s−1 at 50 °C. A minor form also requires more scans or a higher concentration.[4]
NOESY and ROESY are among the experiments that our NMR service quotes per sample. A selective 1D NOESY experiment for exchange is available on quotation: write to contact@norrchemica.com.
Temperature: heating and cooling
Temperature is the second tool, and it is also a frequent cause of the confusion, because a sample near coalescence changes its appearance within a few degrees. A carbamate with a barrier of 16 kcal/mol and two signals 10 Hz apart reaches coalescence at 32 °C by calculation, so its spectrum is broad at an ordinary probe temperature.
Heating
On heating, the two signals of a pair of rotamers broaden, merge at the coalescence temperature Tc, and then sharpen into one signal. The coalescence temperature gives the barrier:[9]
ΔG‡ = 4.575 × 10−3 × Tc × [9.972 + log(Tc / Δν)] kcal/mol
Here Tc is in kelvin and Δν is the distance between the two signals in Hz, measured well below coalescence. As an illustration with chosen numbers, two signals 40 Hz apart that merge at 48 °C (321 K) give a rate constant of 89 s−1 at that temperature and a barrier of 16.0 kcal/mol. Figure 6 shows the calculated series of spectra for this case.
Figure 6. Calculated spectra of two rotamer signals 40 Hz apart, for a barrier of 16.0 kcal/mol, from 0 to 90 °C. Coalescence is reached at 89 s−1, which is 48.4 °C for this barrier. Each trace is scaled to its own height, and the natural line width is 1 Hz. The calculation is ours.
The equation is derived for two forms of equal population whose signals are not coupled to each other. A comparison with complete line-shape analysis showed how far it can be used outside this case.[10] For two equal signals more than 3 Hz apart, the rate constant was within 20% of the line-shape value. For unequal populations, the equation gave a good estimate of the larger of the two rate constants, which is that of the minor form changing into the major form, provided the signals are more than 4 Hz apart. The smaller rate constant is this value divided by the equilibrium constant. For two protons that are coupled to each other, Δν is replaced by √(Δν2 + 6J2), and this form is valid when Δν is larger than J.[10]
An error in the rate constant has a small effect on the barrier: an error of 25% corresponds to about 0.1 kcal/mol at 300 K. The temperature is the larger source of error, because ±2 K corresponds to 0.1 to 0.2 kcal/mol.[10] The distance Δν is a further source of error, because it is measured at low temperature although chemical shifts change with temperature.[9]
A more reliable value is obtained from line shapes fitted at several temperatures, or from exchange spectra. One study measured an amide barrier by both methods and found 16.4 ± 0.1 kcal/mol each time, in two solvents.[14]
The field strength also has an effect. The same two signals are further apart in Hz at higher field, so they merge at a higher temperature. For a barrier of 16 kcal/mol and a distance of 0.1 ppm, the calculated coalescence temperature is 48 °C at 400 MHz and 53 °C at 600 MHz.
What heating proves, and what it does not
A spectrum that merges at high temperature agrees with rotamers, but it is not a proof.
In the β-lactam case above, heating in DMSO-d6 at 400 MHz merged the rotamer signals near 60 °C and gave sharp signals at 80 to 90 °C. The signals of the two diastereomers then coincided as well, so the heated spectrum could no longer give their ratio, and the ratio was measured at 25 °C in CDCl3 (Figure 7).[2]
Figure 7. Two experiments on the same esters. The exchange spectrum at 25 °C keeps the two diastereomers apart, and the heated spectrum does not.[2] Schematic. The positions and heights of the signals are illustrative.
The opposite conclusion is also unsafe. Two sets that do not merge on heating may still be rotamers, because the calculated coalescence temperature for signals 40 Hz apart is about 125 °C for a barrier of 20 kcal/mol and about 164 °C for 22 kcal/mol.
Cooling
Cooling is the test for compounds that look clean. If the barrier is low, the spectrum at room temperature shows one averaged set, and the rotamers appear only at low temperature. The tert-butyl signal of an N-methyl-N-phenyl carbamate reaches coalescence at 250 K in CDCl3 at 500 MHz, which corresponds to 12.3 kcal/mol, and an N-(2-pyridyl) carbamate reaches it only at 203 K in THF-d8 (10.2 kcal/mol).[3]
A sample that is broad at room temperature can be heated or cooled, because it sharpens into one set on heating and into two sets on cooling.
Low temperature is also the method for metal complexes whose ligands change position quickly. Figure 8 shows a measured series from a published study of the hydrogenation chemistry of Wilkinson’s catalyst, in THF-d8 at 500 MHz.[15] At 298 K the rhodium(I) hydride of the reaction mixture gives one broad signal near −8.3 ppm. At 193 K the signal is a resolved multiplet at −8.47 ppm, assigned to HRh(PPh3)3, with couplings of 102 Hz to the phosphorus in the trans position, 20 Hz to the two phosphorus atoms in cis positions and 14 Hz to rhodium.[15]
The same series contains two further changes. A second hydride signal appears on cooling and is resolved at 193 K. It is assigned to the rhodium(III) trihydride that forms by addition of H2. The N-methyl signal of the guanidine base is one signal at 2.60 ppm at 298 K and two signals, at 2.64 and 2.51 ppm, at 193 K.[15]

Figure 8. Measured 1H spectra of one sample (500 MHz, THF-d8): the complete spectrum at 300 K, and the marked regions at ten temperatures from 298 K to 193 K. Reproduced from the Supporting Information of reference [15].
The 31P spectra of the same sample were recorded at six temperatures from 193 to 243 K and fitted by line-shape analysis (Figure 9). Table 3 gives the rate constants.[15] The exchange of the phosphine positions has ΔG‡ = 11.9 kcal/mol at 223 K, which is equal to the value reported earlier for this complex.[15] The rate constant at 223 K alone gives 11.8 kcal/mol by our calculation. Reading 31P spectra is the subject of part 2 of this series.

Figure 9. Measured 31P{1H} spectra at six temperatures (left), and the spectra simulated in the line-shape analysis (right). Reproduced from the Supporting Information of reference [15].
Table 3 · Rate constants from the line-shape analysis
| Temperature, K | 193 | 203 | 213 | 223 | 233 | 243 |
|---|---|---|---|---|---|---|
| Exchange of the phosphine positions in HRh(PPh3)3, s−1 | 1.0 | 1.5 | 2.0 | 12.5 | 20.0 | 75.0 |
| Interconversion of the rhodium(I) and the rhodium(III) hydride, s−1 | 1.5 | 3 | 5 | 14.0 | 40.0 | 150 |
Rate constants from the line-shape analysis of the 31P spectra.[15]
Three practical points follow from this study. The sample was under hydrogen in an NMR tube with a valve. It was left for 10 min at each temperature before the measurement. A salt precipitated from the solution on cooling.[15] Solubility at the lowest temperature should therefore be checked before a series is planned. NMR tubes with a PTFE valve are made for samples under a gas atmosphere.
A series of spectra at different temperatures is available on quotation: write to contact@norrchemica.com.
Which bond, which barrier
Measured barriers show what to expect before the experiment (Table 4 and Figure 10).
Table 4 · Measured barriers to rotation about the C–N bond
| Compound | ΔG‡, kcal/mol | Method and conditions | Source |
|---|---|---|---|
| N-Methylformamide | 22.0 and 20.7 | Line shapes of the neat liquid, values at 298 K, for the major and the minor rotamer. Ratio about 9:1 | [8] |
| N,N-Dimethylformamide | 20.2 to 20.7 | Neat liquid, values at 298 K, from earlier work quoted in the source | [8] |
| Formamide | 17.8 | In diglyme, from earlier work quoted in the source | [8] |
| Amide group of a substituted dimethylurea | 16.4 ± 0.1 | 2D EXSY and line-shape analysis, CDCl3 and tetrachloroethane-d2, 600 MHz | [14] |
| N-Alkyl carbamates | about 16 | General value stated in the source | [3] |
| tert-Butyl N-methyl-N-phenylcarbamate | 12.3 | Coalescence at 250 K, CDCl3, 500 MHz | [3] |
| An N-(2-pyridyl) carbamate | 10.2 | Coalescence at 203 K, THF-d8 | [3] |
| An N-(2-pyrimidyl) carbamate | below 9 | No coalescence down to 183 K, THF-d8 | [3] |
Measured barriers to rotation about the C–N bond of amides and carbamates.
Figure 10. Four structures with their measured barriers. The bond that rotates is marked in gold.
Three ranges follow from these values. Simple amides have barriers of about 20 to 22 kcal/mol, so they give two sharp sets at room temperature and do not merge below about 125 °C for signals 40 Hz apart. N-Alkyl carbamates have barriers of about 16 kcal/mol, so they give broad or doubled signals at room temperature and merge on moderate heating, as the Boc-proline esters above did near 60 °C.[2, 3] N-Aryl carbamates have barriers of about 12 kcal/mol and below, so they give one averaged set at room temperature.[3]
The source for the carbamates states that their barriers are 1 to 4 kcal/mol lower than those of structurally related amides.[3] The two rotamers are also seldom present in equal amounts: N-methylformamide shows them in a ratio of about 9:1.[8]
Signals that imitate exchange
A cross-peak with the sign of the diagonal is not always an exchange peak between two forms of the compound, and an NOE peak is not always a direct NOE. Table 5 lists the effects that are documented in the sources.
Table 5 · Cross-peaks that imitate exchange
| Effect | What it looks like | How to recognise it | Source |
|---|---|---|---|
| NOE relayed by exchange | NOE peaks from a proton of one rotamer to groups of the other rotamer | The peaks disappear at a short mixing time. A selective 1D NOESY with a mixing time of 0.2 s showed only the NOEs within one rotamer | [13] |
| Exchange with water | A peak with the sign of exchange between an NH or OH signal and the water signal | Expected for labile protons. It is exchange, but not between two forms of the compound | [5] |
| Scalar cross-relaxation | Strong peaks with the sign of exchange between an NH proton and the neighbouring CH protons | The NH signal in the 1D spectrum is broad, without visible coupling | [6] |
| TOCSY transfer in ROESY | Peaks with the sign of the diagonal between coupled protons that are close in chemical shift | They change with the position of the transmitter. The EASY-ROESY sequence reduces them | [12] |
| Zero-quantum peaks | Peaks between coupled protons with an up-and-down pattern | The pattern. A NOESY sequence with zero-quantum suppression removes them | [12] |
| Spin diffusion | Extra cross-peaks, mainly in large molecules | Weaker in ROESY, where indirect peaks often have the opposite phase | [12] |
Cross-peaks that can be mistaken for exchange peaks or for direct NOE peaks.
An NOE that crosses from one rotamer to the other
In the 2D NOESY spectrum of a compound with two rotamers, recorded with a mixing time of 0.9 s, a proton of one rotamer showed NOE peaks to groups of both rotamers, because the bond rotates during the mixing time (the dashed circles in Figure 3).[13] Such a peak does not prove that the two groups are close in space. A selective 1D NOESY with a mixing time of 0.2 s showed only the NOEs within one rotamer.[13]
A spurious peak with the sign of exchange has also been reported between protons of one isomer of a small enaminone. Its authors suggest spin diffusion as the cause.[4]
Larger molecules
The sign rule of Table 2 depends on the size of the molecule. One facility guide gives approximate limits: the NOE peak is opposite in sign to the diagonal below a molecular weight of about 600, close to zero between about 700 and 1200, and of the same sign as the diagonal above about 1200.[12] In the last case the sign in a NOESY spectrum no longer distinguishes exchange from NOE.
ROESY distinguishes the two at any size, and the guide prefers it for the middle range.[12] In a ROESY spectrum, however, TOCSY peaks have the sign of the diagonal, so a peak of that sign between two coupled protons is not evidence of exchange.[12]
A cross-peak between NH and CH that is not exchange
In 2,4-disubstituted thiazolidines and oxazolidines, the NOESY spectrum at 500 MHz, with a mixing time of 0.8 s, showed strong peaks with the sign of exchange between the amine NH proton and the neighbouring CH protons, which cannot exchange with each other (Figure 11).[6] The authors assign these peaks to scalar cross-relaxation. The NH proton leaves the nitrogen and returns to it, and this changes the coupling to the neighbouring protons on a time scale of milliseconds.[6]
Figure 11. The cross-peak between NH and the neighbouring CH protons that is not an exchange peak, and the 1D signal that indicates when to expect it.[6] Schematic.
Traces of acid switch the effect on. Some batches of CD2Cl2 gave the peaks and others did not, and a little deuterated acetic acid restored them. The peaks appeared after a work-up at pH 5 and not after a work-up at pH 7 or 9.[6] A higher concentration and a higher temperature increased them. At 193 K only the NOE peaks and the genuine exchange peaks between the two diastereomers of these compounds remained.[6] The effect was seen in CD2Cl2, CDCl3, C6D6 and toluene, and DMSO appeared to suppress it.[6]
The 1D spectrum shows whether the effect is likely. If the NH signal is a resolved multiplet, it is unlikely. If the NH signal is broad, without visible coupling, the peaks may appear.[6] The paper does not test a remedy. From its findings, a neutral and dry solvent, a lower concentration and a lower temperature should reduce the effect, but this is our reading.
Atropisomers: when the rotamer is a separate compound
If the barrier is high enough, the two rotamers can be separated and stored. They are then called atropisomers. The usual criterion is a half-life of interconversion above 1000 s at a given temperature.[7, 16] By our calculation, a half-life of 1000 s at 25 °C corresponds to a barrier of 21.8 kcal/mol, which is the start of region 5 in Figure 2.
Atropisomers are sorted into three classes by their barrier (Table 6). The two sources that we read give different upper limits and different temperatures.
Table 6 · The three classes of atropisomers
| Source | Class 1 | Class 2 | Class 3 | Temperature |
|---|---|---|---|---|
| [16] | Below about 20 kcal/mol, half-life under minutes | About 20 to 30 kcal/mol, hours to days | Above about 30 kcal/mol, years | 37 °C |
| [7] | Below 20 kcal/mol, half-life of racemisation under one minute | 20 to 28 kcal/mol | Above 28 kcal/mol | 25 °C |
The three classes of atropisomers, as each source gives them.
By our calculation, 28 kcal/mol at 25 °C corresponds to a half-life of 1.2 years, and 30 kcal/mol at 37 °C to 4.7 years. The two upper limits therefore describe a similar stability, each at its own temperature. A barrier of 20 kcal/mol corresponds to 51 s at 25 °C and to 13 s at 37 °C. These are half-lives of the rotation. Racemisation is twice as fast.[7]
What the spectrum shows
Two atropisomers that are mirror images of each other give one set of signals, so slow rotation about an axis does not double the spectrum. It is detected at a group with two halves, such as a benzyl CH2, an ethyl or an isopropyl group. The two halves give separate signals as long as the rotation is slow, and variable-temperature NMR follows their coalescence.[7]
If the molecule also contains a stereocentre, the two atropisomers are no longer mirror images. They are diastereomers, and each gives its own set of signals. This follows from the definitions, and it is the case in which two sets of signals show no exchange peak at room temperature although the two compounds interconvert within hours or days.
Which method measures which barrier
One protocol paper gives the range of each method (Figure 12):[7]
- Variable-temperature NMR: up to about 21 kcal/mol, for two signals 10 Hz apart and spectra up to 400 K. Spectra are recorded every 5 °C near coalescence, and line-shape analysis is named as the most precise option.
- EXSY: for systems in slow exchange.
- Dynamic HPLC: about 20 to 23 kcal/mol.
- Kinetics of racemisation: 22 kcal/mol and above.
The first limit agrees with the coalescence equation above: two signals 10 Hz apart that merge at 400 K give 21.2 kcal/mol by our calculation.
Figure 12. The classes of atropisomers at 25 °C, the range of each method, and four 1,1′-binaphthyls on the same scale.[7, 11] The half-lives are our calculation for the rotation.
The stable cases in the catalogue
The 2,2′-substituted 1,1′-binaphthyls are far above these limits (Table 7).[11]
Table 7 · Barriers of 1,1′-binaphthyls
| Compound | Barrier, kcal/mol | Kind of value | In the catalogue |
|---|---|---|---|
| 1,1′-Binaphthyl | 23.5 to 24.1 | Measured in several solvents, quoted in the source | Not listed |
| BINOL | 37.2 and 37.8 | Measured in naphthalene and in diphenyl ether, quoted in the source | (S)-BINOL |
| BINOL | 39.3 | Computed in the source | |
| BINAM | 42.2 | Computed in the source | (R)-BINAM, (S)-BINAM, racemic BINAM |
| BINAP | 49.4 | Computed in earlier work, quoted in the source | (R)-BINAP, (S)-BINAP |
Barriers to rotation about the 1,1′ bond of binaphthyls.[11] No measured barrier for BINAM or BINAP was found.
The source concludes that the size of the 2,2′ substituents controls the barrier, and that the solvent and the electronic character of the substituents have little effect.[11] By our calculation, 23.8 kcal/mol corresponds to a half-life of about 9 hours at 25 °C, and 37.8 kcal/mol to about 18 million years.
Tautomers and labile protons
Two tautomers in slow exchange also give two sets of signals. Acetylacetone is the usual example. Its keto and enol forms interconvert slowly, so each gives its own signals, and the keto form appears at 3.6 and 2.2 ppm in CDCl3.[17] Within the enol form the proton moves between the two oxygen atoms very fast, so the two methyl groups give one signal.[17] The same molecule therefore shows slow and fast exchange side by side.
Tautomers in fast exchange give one averaged set of signals, so there is no second set. They are identified by 15N shifts, coupling constants, isotope effects or spectra at low temperature.[17] Part 5 of this series has a section on this, with the NH-pyrazoles as the example.
A labile NH or OH proton can show an exchange peak to the water signal of the solvent, as in a published example in DMSO-d6.[5] This is chemical exchange, but with water and not between two forms of the compound.
A ratio or a purity when rotamers are present
For the ratio of two diastereomers, the signals of the two rotamers of each diastereomer are added first. In the β-lactam case this was done by quantitative 1H NMR with a 90° pulse and a delay of 60 s between scans, about five times T1. The result agreed with chiral HPLC of the final enantiomers within 3 to 4%.[2] The rotamer ratios of the two diastereomers were 1:1.14 and 1:1.[2]
The same applies to a purity figure. The integral of the compound must include both rotamers. If only the major rotamer is integrated, the purity is too low by the share of the minor rotamer.
Heating the sample until the rotamer signals merge is the alternative, but in the same case it removed the separation between the two diastereomers (Figure 7). Part 4 of this series describes how a purity figure is made by quantitative NMR.
Practical points
Which experiment
- A selective 1D NOESY suits a concentrated sample, one or two questions and well-separated signals. A 2D spectrum shows all correlations at once.[12]
- For two signals that are close together, the control described above shows whether the selective pulse excited only one of them.[1]
- A minor form needs more scans or a higher concentration.[4]
Mixing time
Table 8 · Mixing times in the sources
| Experiment | Mixing time | Source |
|---|---|---|
| NOESY of a small molecule, guideline | 0.5 to 1 s | [12] |
| Selective 1D gradient NOE, carbamate rotamers, 600 MHz | 1.2 s | [1] |
| 2D EXSY, Boc-proline esters, 400 MHz | 0.5 s | [2] |
| 2D NOESY, a compound with two rotamers | 0.9 s | [13] |
| Selective 1D NOESY, the same compound | 0.2 s for the NOEs within one rotamer, 2 s for the exchange | [13] |
| 2D NOESY, thiazolidines, 500 MHz | 0.8 s | [6] |
Mixing times in the sources. The guideline for molecules of medium size is 0.1 to 0.5 s, and for large molecules 0.05 to 0.2 s.[12]
Degassing
Dissolved oxygen reduces the NOE. One facility guide calls its removal very important for small molecules. It asks for three to four freeze–pump–thaw cycles in a tube that can be closed, and states that bubbling argon through the sample is not enough.[12] This matters when NOE peaks are needed, for example to assign the signals of each rotamer.
NMR tubes with a PTFE valve can be closed after degassing. Part 6 of this series describes the tubes for each experiment.
Heating
- The temperature is raised in steps of 10 °C. After each step the sample is left for about 5 minutes, and the shim is adjusted only after that.[18, 19]
- The compound must be stable at the highest temperature of the series.[18, 19]
- The temperature stays below the boiling point of the solvent, because a boiling solvent can spill into the probe.[18, 19] One guide lists these boiling points: CD2Cl2 39.75 °C, acetone-d6 56.5 °C, CDCl3 61 to 62 °C, toluene-d8 110.6 °C, DMF-d7 153 °C and DMSO-d6 189 °C.[18]
- The real temperature of the sample is calibrated with ethylene glycol for high temperatures and with methanol for low temperatures.[19]
Cooling
- The published series above used CDCl3 at 250 K and THF-d8 down to 183 K.[3, 15]
- The sample was left for 10 minutes at each temperature.[15]
- A solvent that freezes can crack the tube, and a component of the sample can precipitate.[15, 18, 19]
Field strength
A higher field is not an advantage for this question. The same pair of signals merges at a lower temperature at 400 MHz than at 600 MHz, which is 48 °C against 53 °C in the example above. Our NMR service measures at 400 MHz as standard, and at 600 MHz and above on request.
Summary
- Two sets of signals have three explanations: one compound in two forms that interconvert, two compounds, or a signal that imitates one of these.
- An exchange peak at room temperature is the direct proof of interconversion. It has the sign of the diagonal. It appears between the rotamers of one compound, and not between diastereomers that do not interconvert.[1, 2]
- A missing exchange peak is not a proof of two compounds, because the test needs a rate constant of roughly 0.1 s−1 or more. The experiment is then repeated at a higher temperature.
- Temperature is a cause and a tool. Barriers near 15 to 16 kcal/mol give broad or doubled signals at room temperature. Signals that merge on heating agree with rotamers but do not prove them, and signals that do not merge do not prove two compounds.[2]
- Simple amides have barriers of about 20 to 22 kcal/mol, N-alkyl carbamates about 16 kcal/mol, and N-aryl carbamates about 12 kcal/mol and below.[3, 8]
- Above about 22 kcal/mol the two rotamers are separate compounds. BINOL, BINAM and BINAP are far above this limit.[7, 11]
- For a ratio or a purity, the signals of both rotamers are added.[2]
Figure 13. The decision path from two sets of signals to the answer. The temperature of 50 °C is an example from our calculation.
Exchange and variable-temperature NMR from NorrChemica
NOESY and ROESY are quoted per sample. A selective 1D NOESY for exchange and a series of spectra at variable temperature are available on quotation: write to contact@norrchemica.com. Purity by quantitative NMR is Package V of the NMR service. The service is available worldwide for chemical and pharmaceutical R&D. Each report is signed and is supplied with the native data. Research use only.
- NMR tubes with PTFE valvedegassed samples, samples under gas
- NMR tubes for quantitative NMRratios and purity
- (S)-BINOLstable atropisomer
- (R)-BINAMstable atropisomer
- (S)-BINAMstable atropisomer
- (R)-BINAPstable atropisomer
- (S)-BINAPstable atropisomer
NMR tubes made in Germany by Deutero and shipped from Finland. See the NMR tube guide →
References
- Hu DX, Grice P, Ley SV. Rotamers or diastereomers? An overlooked NMR solution. J. Org. Chem. 2012, 77, 5198–5202. DOI
- McLoughlin EC, O’Brien JE, Trujillo C, Meegan MJ, O’Boyle NM. Application of 2D EXSY and qNMR spectroscopy for diastereomeric excess determination following chiral resolution of β-lactams. ChemistryOpen 2023, 12, e202200119. DOI
- Deetz MJ, Forbes CC, Jonas M, Malerich JP, Smith BD, Wiest O. Unusually low barrier to carbamate C–N rotation. J. Org. Chem. 2002, 67, 3949–3952.
- Hussaini SR, Kuta A, Pal A, Wang Z, Eastman MA, Duran R. Application of NMR spectroscopy for the detection of equilibrating E–Z diastereomers. ACS Omega 2020, 5, 24848–24853. DOI
- University of Ottawa NMR Facility Blog. What are those positive peaks in my NOESY spectrum? 24 July 2008. Link
- Panduwawala TD, Josa-Culleré L, Kuprov I, Odell B, Moloney MG, Claridge TDW. Scalar cross-relaxation detected in the NOESY spectra of oxazolidines and thiazolidines. J. Org. Chem. 2016, 81, 4142–4148. DOI
- Heeb J-P, Clayden J, Smith MD, Armstrong RJ. Interrogating the configurational stability of atropisomers. Nat. Protoc. 2023, 18, 2745–2771. DOI
- Neuman RC Jr, Jonas V, Anderson K, Barry R. Hindered rotation in N-methylformamide: a peptide-bond model system. Biochem. Biophys. Res. Commun. 1971, 44, 1156–1161.
- University of Colorado, NMR facility. Determining the energy of activation parameters from dynamic NMR experiments. Facility note, 2009. Link
- Kost D, Carlson EH, Raban M. The validity of approximate equations for kc in dynamic nuclear magnetic resonance. J. Chem. Soc. D 1971, 656–657. DOI
- Tkachenko NV, Scheiner S. Optical stability of 1,1′-binaphthyl derivatives. ACS Omega 2019, 4, 6044–6049. DOI
- Columbia University, NMR facility. NOESY and ROESY experiments. Facility guide. Link
- University of Ottawa NMR Facility Blog. Sorting out NOE’s for exchanging rotamers. 19 January 2012. Link
- Vassilev NG, Ivanov IC. A small change in structure, a big change in flexibility. Molecules 2023, 28, 8004. DOI
- Perea-Buceta JE, Fernández I, Heikkinen S, Axenov K, King AWT, Niemi T, Nieger M, Leskelä M, Repo T. Diverting hydrogenations with Wilkinson’s catalyst towards highly reactive rhodium(I) species. Angew. Chem. Int. Ed. 2015, 54, 14321–14325. DOI
- Faisca Phillips AM, Pombeiro AJL. Atropselective organocatalytic synthesis of chiral compounds containing nitrogen along the axis of chirality. Symmetry 2023, 15, 1261. DOI
- Hansen PE. Tautomerism detected by NMR. Encyclopedia (MDPI), entry 307, earlier revision. Link
- University of Kansas, NMR laboratory. High temperature NMR. Guide, 2022. Link
- Michigan State University, NMR facility. VT operation. Instruction. Link
About this article. Sources 1, 4 and 10 were read in full, and source 15 in its Supporting Information. Sources 2, 3, 6, 7, 8, 11 and 14 were read in their open-access or repository copies, and source 16 in part. Sources 5, 9, 12, 13, 18 and 19 are guides, notes and blog entries of university NMR facilities. They were used for practical parameters, such as sign rules, mixing times and temperature limits, that papers seldom state. Source 17 is an encyclopedia entry. The line shapes of Figures 1 and 6, the regions of Figure 2, and all rate constants, half-lives and coalescence temperatures without a reference are our own calculations with the Eyring equation and a transmission coefficient of 1. Figures 3, 4, 5, 7, 11 and 13 are schematics. Figures 8 and 9 are measured spectra, reproduced from the Supporting Information of source 15. The sources were read on 11 and 12 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 (this article)
- 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
- 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