1Reading a 2D spectrum

A 2D spectrum is a contour plot. A frequency axis runs along each edge — here ¹H on one axis, and ¹H or ¹³C on the other. Signal intensity is drawn as contour lines, in the manner of a topographic map. Each peak occupies the point where two frequencies meet, so its position corresponds to a pair of nuclei rather than to one.

Two types of peak occur. In a homonuclear experiment, where the same nucleus is on both axes (COSY), a diagonal extends from corner to corner at the points where the two frequencies are equal. The diagonal reproduces the 1D spectrum and adds no new information; it functions as a reference. The structural information is in the off-diagonal cross-peaks.

A cross-peak at (δ_A, δ_B) indicates that the nucleus at δ_A and the nucleus at δ_B are related. The experiment defines the relationship: scalar coupling through bonds (COSY), a one-bond C–H connection (HSQC), or a two- to three-bond C–H connection (HMBC). Tracing a cross-peak to each axis gives the two frequencies, and therefore the two atoms, that are related.

This is why a 2D spectrum resolves what a crowded 1D spectrum cannot: signals that overlap on one axis are separated on the second, and the cross-peaks indicate the connections between them.

2COSY homonuclear ¹H–¹H correlation

The same ¹H spectrum appears on both axes. On the diagonal, where the two frequencies are equal, each proton correlates with itself; the diagonal reproduces the 1D ¹H spectrum and functions as a reference. The structural information is off the diagonal. A cross-peak connects two protons that share a scalar coupling, in most cases the vicinal (³J) coupling between protons on adjacent carbons. Tracing a cross-peak to each axis gives the two coupled protons. A COSY therefore identifies which protons are coupling partners and groups the spectrum into sets of mutually coupled protons.

What it shows. The ester O–CH₂ (~4.1 ppm) and the terminal CH₃ (~1.2 ppm) share a cross-peak, from the adjacent CH₂ and CH₃ of the ethyl group. The N1–CH₂ (~4.4 ppm) has no cross-peak: it lies between the ring nitrogen and the ester carbonyl, with no proton on an adjacent carbon. A weaker long-range (⁴J) correlation connects the ring C6–H (~7.5 ppm) to the 5-CH₃ (~1.7 ppm). The exchangeable N–H (~11.3 ppm) gives no correlation. The absence of a cross-peak for the N1–CH₂ is resolved in §4: COSY does not locate it, HMBC does.

3HSQC one-bond ¹H–¹³C correlation

The ¹H spectrum is on one axis, the ¹³C spectrum on the other. There is no diagonal, because the two axes are different nuclei. Each cross-peak connects a proton to the carbon to which it is directly bonded (a one-bond, ¹J, C–H correlation). An HSQC assigns every protonated carbon and pairs each C–H signal in the ¹H spectrum with its carbon in the ¹³C spectrum. Carbons without an attached hydrogen produce no cross-peak.

What it shows. Each C–H is assigned: the ring C6–H at ~7.5/142 ppm, the N1–CH₂ at ~4.4/48 ppm, the ester O–CH₂ at ~4.1/61 ppm, the 5-CH₃ at ~1.7/12 ppm, and the ester CH₃ at ~1.2/14 ppm. The two CH₂ groups, close in the ¹H spectrum, are separated on the carbon axis (48 and 61 ppm), as are the two CH₃ groups (12 and 14 ppm). The N–H, the three carbonyl carbons (ester, thymine C2 and C4) and the quaternary C5 produce no cross-peak, since none has a directly bonded hydrogen. These carbons are assigned by HMBC in §4.

A multiplicity-edited HSQC also encodes the number of attached hydrogens in the phase of each correlation: CH and CH₃ appear with one sign, CH₂ with the opposite. The two methylene groups are therefore distinguished from the methyls by phase alone, independently of their shifts. This depends on correct phasing of the spectrum; NorrChemica processes and phases each spectrum by hand rather than by an automated routine.

4HMBC long-range ¹H–¹³C correlation

The axes are the same as in HSQC, ¹H against ¹³C, but the experiment detects correlations over two and three bonds rather than one. It therefore correlates protons with carbons that are not directly bonded to them, including the carbonyl and quaternary carbons absent from the HSQC. HMBC establishes connectivity across atoms that bear no hydrogen, such as a carbonyl carbon or a substituted nitrogen.

What it shows. The ester O–CH₂ (~4.1 ppm) and the ester CH₃ (~1.2 ppm) both correlate with the ester carbonyl at ~168 ppm, which defines the ethyl ester. The N1–CH₂ (~4.4 ppm) correlates with the ester carbonyl and, across the ring nitrogen, with the two ring carbons bonded to N1: C2 (~150 ppm) and C6 (~142 ppm). These correlations establish that the acetate group is attached at N1. The N1–CH₂ produced no cross-peak in COSY, since it has no proton on an adjacent carbon; HMBC assigns its position through the two- and three-bond correlations to the neighbouring carbons.

HMBC intensities are not uniform. The experiment is optimised for an average long-range coupling, so a correlation whose coupling is far from that value may be weak or absent, and a two-bond correlation is not always stronger than a three-bond one. Long-range couplings through a π system can give correlations over more than three bonds. The presence of a cross-peak establishes a connection; its absence does not exclude one.

5Other 2D experiments

COSY, HSQC and HMBC are sufficient for most routine structural work, and this guide covers those three. Other 2D experiments address specific questions. TOCSY correlates all protons within a coupled spin system, not only those on adjacent carbons, and is useful where spin systems overlap. NOESY and ROESY correlate protons that are close in space rather than through bonds, and are used to determine stereochemistry and conformation. These are outside the scope of this guide, but NorrChemica runs them on request as part of the NMR analysis service. The references below (Facey; Claridge) treat them in detail.