NMR Glossary

Expand All

An experiment used for the multiplicity editing of 1D carbon-13 spectra, such that CH and CH3 carbons have an opposite phase (i.e. opposite sign) relative to CH2 and quaternary carbons.

The DEPT or DEPTQ experiments (or even better, a multiplicity-edited HSQC) are generally to be preferred because these transfer magnetisation from 1H to 13C, leading to sensitivity gains.

Original literature:

The CLIP-COSY is a modern variant of the COSY experiment which provides high-resolution crosspeaks.

In a standard 2D COSY, peaks are a mixture of absorption and dispersion lineshapes—the dispersive part in particular causes long ‘tails’ which limit the achievable resolution. The DQF-COSY alleviates this to some extent by removing the dispersive contributions, but the remaining absorption components are still antiphase (a mixture of positive and negative parts). The CLIP-COSY goes one step further to ensure that all peaks also have the same phase (i.e. sign), hence the name ‘clean in-phase’.

This is accomplished through a combination of zero-quantum filters (ZQF) as well as a perfect echo, which transfers in-phase magnetisation to in-phase magnetisation.

Original literature:


One of the first 2D NMR experiments ever proposed, the COSY experiment is used to identify nuclei that share a scalar coupling (J-coupling). Each distinct spin yields a diagonal peak, and the presence of off-diagonal peaks (crosspeaks) in the spectrum reveals coupled partners.

COSY is most often used to analyse coupling relationships between protons, but may generally also be used to correlate other high-abundance spins such as 31P, 19F, and 11B.

Original literature:

A 1D experiment used for enhancing the sensitivity of carbon observation and for editing of 13C spectra.

Note that the DEPT experiment has largely been superseded by the multiplicity-edited HSQC, which provides more information with greater sensitivity and often in less time. 1D 13C spectra (including DEPT) are only useful for structural assignment if there is intractable peak overlap in the HSQC, which is rare.

The sensitivity gain arises because the experiment is started with proton excitation, with the proton magnetisation subsequently being transferred onto carbon. Since 1H has a fourfold larger gyromagnetic ratio compared to 13C, the initial population difference for 1H spins (loosely speaking, the difference between the number of spin-up and spin-down nuclei) is approximately four times larger, which translates into higher sensitivity.

The editing feature alters the sign of the carbon resonances according to the number of directly attached protons, allowing the identification of carbon multiplicities. The signs of the peaks may be controlled using the flip angle of the final proton pulse, which is usually appended to the experiment name. In the table below, 0 means not observed, + positive, and − negative, and the DEPTQ variant is included for completeness:

























These signs are not necessarily unique: it is possible to phase a DEPT spectrum (by 180°), such that all peaks are inverted, i.e. positive peaks become negative and vice versa. There is no general guarantee that the automatic phase correction routine you use will give the same signs as the table above. However, their relative signs can be guaranteed: thus, the CH and CH2 peaks will always have opposite signs in a DEPT-135. To eliminate any possibility of ambiguity, it is best to assign the multiplicity of a known peak in the 13C spectrum before inferring the multiplicities of the other peaks.

Original literature:

In a traditional DEPT experiment, quaternary and non-protonated carbons are not observed as they do not bear any protons: thus, proton magnetisation cannot be transferred onto those carbons (the polarisation transfer process in DEPT makes use of one-bond C–H couplings).

The DEPTQ experiment is a variant of DEPT, in which the signals of non-protonated carbons are retained with the same sign as that of CH2 groups. The polarisation transfer process is still used for protonated carbons, but because this is not applicable to quaternary carbons, the sensitivities of these carbons in the DEPTQ spectrum are lower.

Original literature:

The DIPSI pulse sequence element is a fixed series of pulses, which was originally developed for heteronuclear decoupling, but has now also been used widely to effect isotropic mixing: the transfer of (x, y-, and z-) magnetisation between spins which share scalar couplings.

This transfer can occur over multiple steps, leading to propagation of magnetisation through a spin system, and is most commonly used in the TOCSY experiment.

Several variants of the DIPSI sequence exist, and are labelled by numbers (DIPSI-n). DIPSI-2 is the version which is in use nowadays.

Original literature:

A pseudo-2D NMR experiment, where chemical shifts are presented on the horizontal axis and self-diffusion coefficients corresponding to each peak on the vertical axis (much like the indirect dimension in 2D NMR, hence ‘pseudo-2D’).

A DOSY experiment consists of a series of 1D diffusion experiments, recorded with different amplitudes of pulsed field gradients (PFGs). As per the Stejskal–Tanner equation, the intensities of each peak decrease as the gradient amplitude is increased, allowing the diffusion coefficients to be extracted from the decay pattern via one of many curve-fitting procedures. The peak linewidth in the diffusion dimension corresponds to the uncertainty in the diffusion coefficient.

The method can, for example, be employed to investigate molecular size, complexation phenomena, binding and aggregation.

Original literature:

A method for selective excitation of one resonance in a spectrum, which uses pulsed field gradients (PFGs) to suppress all other resonances.

Once the desired spin has been excited, the magnetisation may be used to carry out selective 1D NOESY or TOCSY experiments which reveal crosspeaks originating from only the target spin. This is substantially faster than running a full 2D NMR experiment and is the experiment of choice where only one particular crosspeak is of interest.

See also: PFG

Original literature:

A variation on the standard 2D COSY experiment, used for identifying scalar couplings between protons.

Standard COSY experiments, contain unfavourable phase properties which lead to large ‘tails’ in peaks and limited resolution in crowded areas. The double-quantum filter alters the phase properties of the detected NMR signal, which enables a phase-sensitive presentation with greater resolution, and thus aiding the analysis of crowded spectra.

The ‘double-quantum’ coherences referred to in the name can only be generated from two different mutually coupled spins, so another useful feature of the DQF-COSY is that it suppresses diagonal peaks which arise from singlets (such as water). Higher-order multiple-quantum filters are also possible—for example, a triple-quantum filter would suppress spin systems with only two spins—but are less widely used.

Coupling constants can in principle be extracted from DQF-COSY crosspeaks, but there are nowadays more targeted modern methods for that based on J-resolved / selective refocusing experiments.

See also: COSY, TOCSY

Original literature:

A method for the selective excitation or suppression of one specific resonance, or a group thereof, in a spectrum.

Excitation sculpting uses selective pulses (or a group of pulses) which are designed to only affect a small range of frequencies in the spectrum. The resonances falling within this range can either be suppressed or retained through the appropriate placement of pulsed field gradients around the selective pulse.

The major breakthrough provided by the excitation sculpting method is that any selective pulse element can be used, regardless of its phase properties. This is because the pulse element is applied twice, and the original paper (given below) contains an elegant mathematical demonstration that any phase errors incurred from the first application are refocused by the second.

The ES method can be used for water suppression in aqueous samples (although it has a notable drawback in that peaks near the water resonance are also very cleanly suppressed, which is not always desirable). By selecting for a specific resonance, instead of rejecting it, ES also forms the basis for selective 1D NOESY/TOCSY experiments.

Original literature:

EXSY is a 2D experiment used to identify spins which are undergoing chemical exchange, and can, in favourable cases, also be used to quantify the kinetic parameters.

Its appearance is similar to the basic COSY experiment in that both diagonal peaks and crosspeaks are present, but crosspeaks indicate an exchange process occurring between the correlated spins. Its use is limited to those systems in which the exchange kinetics are faster than, or comparable to, spin relaxation rates: loosely speaking, this can be explained by the fact that spins must carry information from their old chemical environment to their new environment before it is lost through relaxation.

EXSY is most often used in the study of fluxional inorganic or organometallic systems, but can also be applied to conformational exchange in organic systems, for example.

In its most basic form, the EXSY experiment is in fact identical to that of the NOESY experiment. Thus, cross-peaks may in principle also indicate through-space correlations between spins. This can, however, be distinguished easily for small molecules: NOE crosspeaks have opposite sign to the diagonal peaks, whereas exchange peaks have the same sign, and are typically much larger in intensity. (These are often called ‘negative’ and ‘positive’, with the implicit assumption that the spectrum is phased such that the diagonal peaks are positive.)

Original literature:

GARP is a pulse sequence element for heteronuclear decoupling, frequently used for 13C decoupling in HSQC experiments. This suppresses the effects of one-bond C–H couplings, meaning that each proton appears as a single peak instead of a doublet.

The GARP element itself is a series of pulses, applied repeatedly on the heteroatom with varying durations and alternating phases. The achievable bandwidth with GARP (i.e. the range of frequencies which can be decoupled) is very wide, which is what allows it to fulfil its role in 13C decoupling: other decoupling methods such as WALTZ do not cover the entire 13C spectral width perfectly.

The main drawback of GARP is the high duty cycle of the pulses being used. As a result, GARP decoupling cannot be applied for long durations, or spectrometer damage may result. This tradeoff is very general: in order to cover a wider bandwidth, a higher pulse power is required.

Original literature:

A 2D experiment used to identify long-range couplings between 1H and a heteronucleus, typically 13C.

‘Long-range’ generally refers to 2- or 3-bond couplings, though in rare cases even longer couplings may be observed (e.g. in unsaturated or structurally rigid systems). For 1H–13C HMBC in particular, 3-bond couplings tend to be more frequently observed and stronger in the spectrum.

The HMBC experiment is conceptually similar to the HMQC experiment. One key difference is that the couplings observed in HMBC are far smaller; consequently, the required evolution delays are longer (loosely speaking it takes a longer time to ‘see’ the couplings). In the HMBC experiment, the refocusing period at the end of the sequence is removed in order to avoid further relaxation losses, meaning that lineshapes in the direct dimension are highly mixed, and heteronuclear decoupling cannot be used at the end of the sequence. Magnitude-mode processing is typically used for the direct dimension of HMBC experiments (‘xf2m’ command in TopSpin).

Original literature:

A 2D experiment used to correlate directly bonded protons with heteronuclei, typically 13C or 15N.

For 13C correlations in small molecules, the HSQC experiment tends to be preferred. This is because the HMQC experiment allows proton–proton couplings to evolve during t1, leading to multiplet structure in the indirect dimension. This is often not resolved and is simply manifested as line broadening, but nevertheless leads to a reduction in peak height (i.e. sensitivity). The HSQC spectrum does not have this issue.

Original literature:

HOESY is the heteronuclear analogue of NOESY: it detects nuclear Overhauser effects between dissimilar nuclides (as opposed to the NOESY, which is used for proton–proton NOEs).

Both 1D and 2D HOESY methods exist, and one of the two nuclides is typically 1H. One issue is sensitivity: NOEs are a relatively weak phenomenon and when used with nuclei with low natural abundance (e.g. 1H–13C HOESY), the resulting poor sensitivity can limit the application of HOESY. On the other hand, 1H–19F HOESY can have sensitivity comparable to that of 1H–1H NOESY and so finds use in fluorine chemistry.

19F–1H NOEs were a previous research interest in Oxford: see e.g. Chem. Eur. J. 2012, 18 (41), 13133–13141.

Original literature:

A 2D experiment which correlates directly bonded protons and heteronuclei, i.e. reveals one-bond H–X correlations. The heteronucleus used is frequently 13C or 15N.

The correlations can be used to map known proton assignments onto their directly attached carbons. The 2D spectrum can also prove useful in the assignment of the proton spectrum itself by dispersing the proton resonances along the 13C dimension and so reducing proton multiplet overlap. It also provides a convenient way of identifying diastereotopic geminal protons (which are sometimes difficult to distinguish unambiguously, even in COSY) since only these will produce two correlations to the same carbon. Alongside COSY, this represents the front-line 2D technique of organic chemistry.

Note that a 1H–13C HSQC experiment is substantially more sensitive than a 1D 13C spectrum, and is much faster to acquire. The reason for this is twofold: firstly, 1H magnetisation is excited and transferred to 13C (instead of directly exciting 13C magnetisation), using the INEPT technique which is described elsewhere on this page. Secondly, after the 13C chemical shifts are allowed to evolve, this magnetisation is transferred back to 1H again. 1H has a larger gyromagnetic ratio which leads to improved detection sensitivity.

By insertion of a spin echo at the end of t1, multiplicity editing can be included in an HSQC experiment. This leads to CH and CH3 peaks having an opposite sign to CH2 peaks, which provides exactly the same information as in a DEPT-135 experiment.

Original literature:

An extension of the 2D HSQC experiment in which a TOCSY transfer between protons is added just prior to data acquisition. This relays the original proton–carbon correlation peak onto neighbouring protons within the same spin system, thus producing a 13C-dispersed TOCSY spectrum. Crosspeaks will be observed between 13C nuclei and 1H nuclei in the same spin system as the directly bonded 1H. This proves to be useful when analysing complex proton spectra for which the 2D TOCSY becomes too crowded for unambiguous interpretation.

Original literature:

A variant of the HSQC experiment that has been optimised for the detection of long-range couplings (as observed in HMBC), but is more specifically tailored for the measurement of the magnitudes of the coupling constants themselves. A number of variants of HSQMBC exist, including Clean In-Phase (CLIP) and Pure In-Phase (PIP) HSQMBC experiments. Most useful for measuring nJCH and nJNH coupling constants. 

Original literature:

An experiment which is superficially similar to HMBC, but specifically identifies two-bond correlations between 1H and 13C. This is therefore complementary to the HMBC experiment which typically emphasises three-bond correlations over two-bond correlations.

It is important, however, to note that the H2BC experiment does not genuinely measure two-bond H–C couplings. Instead, it uses a combination of 1JCH (HC–X–H) and 3JHH couplings (H–C–X–H) to map out two-bond C–H correlations (H–C–X–H). The use of 1JCH means that only two-bond correlations to protonated carbons are observed: unlike in the HMBC experiment, quaternary carbons cannot be observed.

Therefore, the H2BC experiment does not yield any extra information compared to a combination of HSQC + COSY. It does, however, provide improved dispersion of peaks compared to the COSY, which may be useful for extremely crowded spectra.

Original literature:

  • Nyberg, N. T.; Duus, J. Ø.; Sørensen, O. W. Heteronuclear Two-Bond Correlation: Suppressing Heteronuclear Three-Bond or Higher NMR Correlations while Enhancing Two-Bond Correlations Even for Vanishing 2JCH. J. Am. Chem. Soc. 2005, 127 (17), 6154–6155.

A 1D experiment used to enhance the sensitivity of nuclei with low magnetogyric ratios (γ), such as 15N or 13C, by transferring the greater population differences of a high-γ spin (such as 1H, 19F, or 31P) onto the heteronucleus. The transfer is most often performed from protons onto a directly bound heteronucleus, i.e. over a one-bond coupling.

The INEPT experiment first involves the excitation of the high-γ spin. Because the difference in energy between spin-up and spin-down states is proportional to γ, this spin will have a larger energy difference, and via the Boltzmann formula, a larger population difference (or polarisation). Then, the coupling between the two nuclei is allowed to evolve, which generates coherence which is antiphase with respect to the low-γ spin: this can then be transferred to the low-γ spin using a pair of 90° pulses.

Overall, this leads to a theoretical maximum sensitivity increase of γI/γS where γI and γS are respectively the magnetogyric ratios of the high- and low-γ nuclei.

While the 1D INEPT sequence is not commonly used on its own, it has found its way into a large number of NMR experiments as a means of enhancing sensitivity. For example, the HSQC experiment actually begins with an INEPT block: this is part of the reason why HSQC experiments are more sensitive than 1D 13C.

Original literature:


A family of 2D methods which separate chemical shifts and scalar (J) couplings into different dimensions. Can operate in homonuclear or heteronuclear modes and thus provides a means of measuring homonuclear or heteronuclear couplings in the J-dimension. The homonuclear experiment in particular is plagued by problems arising from strong-couplings, so is best performed at the highest available field strength.


A through-space phenomenon used in the study of 3D structure and conformation. It gives rise to changes in the intensities of NMR resonances of spins I when the spin population differences of neighbouring spins S are altered from their equilibrium values (by saturation or population inversion). Proton-proton NOEs are by far the mostly widely used in structure elucidation. Since the effect has a (non-linear) distance dependence, only protons "close" in space (within 4-5 Å) give rise to such changes and the NOE is thus an extremely useful probe of spatial proximity. The NOE is a spin relaxation phenomenon and has very different behaviour depending on molecular motion and, in particular molecular tumbling rates. Small molecules (<1000 Da) under typical solution conditions tumble rapidly and produce weak, positive proton NOEs that grow rather slowly whereas, in contrast, large molecules (> 3000 Da) tumble slowly in solution and so produce large, negative NOEs that grow quickly. Mid-size molecules (ca 1000-3000 Da) tumble at "intermediate " rates where the NOE crosses from the positive to the negative regime and thus can have vanishingly small NOEs. Under such circumstances conventional NOEs may not be observed and it is necessary to either alter solution conditions (eg temperature, solvent viscosity) to change the motional properties or use so-called rotating-frame NOE (ROE) measurements. ROEs are generated under rather different physical conditions but from the chemist's perspective the key feature is that they remain positive for any tumbling rate.




A 2D method used to map NOE correlations between protons within a molecule. Most popular with, and best suited to, the study of very large molecules such as bio-polymers, although it still has a place in small molecule work. The observed NOEs are termed "transient NOES" and should not be confused with the "steady-state NOEs" that are observed with the NOE difference experiment. The spectra have a layout similar to COSY but crosspeaks now indicate NOEs between the correlated protons. Positive NOEs (rapidly tumbling molecules) have opposite phase to the diagonal peaks whereas negative NOEs (slowly tumbling molecules) have the same phase as the diagonal (saturation transfer from chemical or conformational exchange also has the same phase as the diagonal and may be confused with negative NOEs).



Non-uniform sampling is a method of reducing experiment time in multidimensional (2D, 3D, ...) NMR experiments.

Unlike 1D experiments where a FID is recorded at one shot, 2D experiments are instead built up from a series of FIDs, each of which must be individually recorded. In order to be processed via the Fourier transform, these FIDs must be uniformly sampled: that is, the indirect-dimension evolution time (t1) must be increased in uniform steps between successive FIDs (or ‘increments’).

In an NUS spectrum, only a fraction of the requisite t1 increments are actually recorded: the remainder of the data are reconstructed using an algorithm such as compressed sensing. This allows 2D data to be acquired in a much shorter period of time, at the cost of signal-to-noise as well as some reconstruction artefacts.

The extent of NUS applied can be measured by the percentage of points which are actually recorded: a fully sampled spectrum would be 100%. This must be chosen carefully in order to avoid creating too many artefacts. Sparse spectra such as HSQC experiments can often be run with 25% NUS (there is ‘less information’ in these spectra to reconstruct), whereas crowded spectra such as HMBC do not do so well. Spectra with large dynamic ranges (i.e. very strong and very weak peaks) such as NOESY are also poorer candidates for NUS.

There is too much original literature to list here, but here is a review:


This is the application of a short (pulsed) magnetic field gradient across the NMR sample which momentarily destroys the magnetic field homogeneity within the sample. The effect is such that chemically similar spins that exist in different locations within the NMR sample precess with different frequencies during the pulse (in contrast to the usual requirement for high-resolution NMR spectra where, in a well shimmed magnet, these should all precess with identical frequencies). The net result of the pulse is that these spins are dispersed in the transverse plane (defocussed) and produced zero net magnetisation. This is the basis on which pulsed field gradients may be used to suppress unwanted resonances in a spectrum. Furthermore, appropriate combinations of these pulses can be employed to selectively refocus signals that we do wish to see in the final spectrum whilst leaving the unwanted resonances defocussed and thus unobservable. Thus, pulsed field gradients provide a means for signal selection in NMR experiments that provide clean, high-quality data sets often very quickly when sample concentrations are not limiting. Many modern NMR methods are thus referred to as "gradient-selected", gradient-enhanced" or simply "gradient" experiments.


A 2D experiment that measures NOEs in the "rotating-frame" and is used to map NOE correlations between protons, particularly for mid-sized molecules (1000-3000 Da) that have close-to-zero conventional NOEs. Again has a similar overall appearance to COSY, but cross-peaks (which have opposite phase to the diagonal regardless of molecular tumbling rates) now indicate ROEs between the correlated spins. The experiment is also prone to interference from TOCSY transfers (between J-coupled spins) and requires careful analysis.


See also: nOe, NOESY


An experiment used to detect the binding of small molecule ligands to macromolecules such as proteins. The technique relies on the transfer of magnetisation between the protein (whose proton resonances are saturated by direct irradiation) and the bound ligand via the proton-proton NOE, followed by the release of the ligand back into free solution where its proton spectrum is observed. The free ligand carries the negative NOE from when it was bound, so has reduced proton signal intensities. Subtraction of a control spectrum (recorded without protein saturation) reveals resonances of the binding ligands. Any molecule that does not bind should not appear in the resulting STD spectrum. The technique is often complimentary to waterLOGSY.

See also: water-LOGSY, nOe



A 2D homonuclear correlation experiment used to analyse scalar (J) coupling networks between protons. It has a similar appearance to the 2D COSY spectrum. However, COSY crosspeaks are limited to identifying directly coupled spins, that is, those spins that share a mutual J-coupling, A-B. TOCSY is able to “relay” magnetisation between spins, A-B-C-D.., and can therefore show correlations amongst spins that are not directly coupled (eg A-C and A-D) but exist within the same spin system. This proves useful in the analysis of crowded spectra where correlations from a single resolved proton may be used to trace the coupling network. Popular for the analysis of peptides and oligosaccharides where molecules are typically composed of discrete subunits (spin systems) ie. amino-acids or saccharide units.




The WALTZ (wideband, alternating-phase, low-power technique for residual splitting (!)) element is a cluster of pulses applied repeatedly to achieve spin-decoupling, typically of protons during observation of a heteroatom (an example of so called "heteronuclear decoupling"). Most commonly it is used for proton decoupling during the acquisition of carbon-13 spectra. The sequence produces very effective removal of couplings with little residual broadening of peaks, as required for high-resolution heteronuclear NMR measurements.


See also: GARP


An experiment used to detect the binding of small molecule ligands to macromolecules such as proteins. The technique relies on the transfer of magnetisation from irradiated water onto the protein and then onto the bound ligand via the proton-proton NOE, followed by the release of the ligand back into free solution where its proton spectrum is observed. The free ligand carries the negative NOE from when it was bound. The ligand in the free state will also receive a positive NOE directly from water, so it is often necessary to record the waterLOGSY spectrum in the absence of the protein and look for differences in peak intensity as indicative of binding. Any molecule that does not bind should not show such intensity changes in the presence of the protein. The technique is often complimentary to STD, and often finds use in fragment-based ligand screening.


See also: nOe, STD


A method for solvent suppression that employs pulsed field gradient spin-echoes to destroy the unwanted solvent resonance but retain all others. Most commonly employed in the study of biological molecules in 90%H2O/10%D2O where the water signal dominates all others. Other variations, including those based on excitation sculpting, are also widely employed.


See also: PFG