HMBC: Long-Range Heteronuclear Correlation

Introduction

COSY and TOCSY reveal proton networks, and HSQC pins each proton to its carbon, but molecules are full of atoms that interrupt these networks: carbonyl groups, quaternary carbons, ester oxygens and aromatic ring junctions. HMBC is the tool that reaches across those gaps. It shows correlations between protons and carbons two or three bonds away, which lets a chemist connect isolated fragments and place quaternary carbons, turning a list of pieces into a complete molecular skeleton.

Core explanation

HMBC detects carbon–proton pairs coupled through several bonds. Two-bond couplings ²J(CH) and three-bond couplings ³J(CH) usually lie between 0 and about 10 Hz, much smaller than the one-bond coupling of 125–170 Hz. The HMBC sequence therefore uses a long evolution delay, typically about 1/(2 × 8 Hz) ≈ 60 ms, to let this small coupling generate transferable coherence. A low-pass J filter at the start of the sequence suppresses one-bond correlations, so that HSQC-type peaks are largely removed from the spectrum.

Unlike HSQC, HMBC is not refocused with respect to proton–proton couplings, so peaks are recorded in magnitude mode and appear broader. The spectrum has ¹H in F2 and ¹³C in F1, just like HSQC, but each proton may show several peaks, one for each carbon within about three bonds. Critically, those carbons include quaternary carbons and carbonyls, which never appear in HSQC.

The interpretive power comes from bridging. Consider a methyl ester, CH₃–O–C(=O)–R. The OCH₃ protons have no proton neighbours, so COSY is silent. HMBC shows a three-bond correlation from the OCH₃ protons to the carbonyl carbon near 170 ppm, proving that the methoxy group is attached to that carbonyl through oxygen. In the same way, a methyl singlet on a quaternary carbon shows HMBC peaks to that quaternary carbon (two bonds) and to the carbons attached to it (three bonds). Methyl singlets are often called anchors, because their strong, sharp signals give reliable HMBC correlations that fix the surroundings of a quaternary centre.

The main difficulty is that HMBC does not state the number of bonds. ²J and ³J are of similar size, and ³J often exceeds ²J. ³J(CH) follows a Karplus-type dependence on dihedral angle, so it can be near zero for some geometries; ²J can be close to zero too. Aromatic rings are a classic case: ³J across the ring (about 7–8 Hz) is usually larger than ²J (about 1 Hz), so meta-carbons often show stronger peaks than ortho-carbons. Four-bond correlations can appear in conjugated or rigid systems. A missing HMBC peak is therefore weak evidence, and a single peak cannot distinguish two from three bonds without support from other data.

Incomplete suppression of one-bond couplings leaves residual ¹J artefacts: a pair of peaks split by roughly 140 Hz in F2, centred on a proton's true shift. Recognising these, and comparing with HSQC, prevents mistaking them for long-range correlations. Variants such as H2BC use proton–proton couplings to pick out two-bond correlations specifically, helping to resolve the bond-count ambiguity.

Step-by-step reasoning

1. Use COSY/TOCSY and HSQC to define fragments with assigned protons and carbons. 2. Find HMBC peaks from each fragment's protons to carbons outside it. 3. Place quaternary carbons by looking for protons from two fragments that both correlate to the same carbon. 4. Check that each proposed link needs no more than three bonds. 5. Test alternative structures: the correct one explains all key correlations.

Visual explanation

Sketch two fragments on either side of a C=O. Draw curved arrows from protons of both fragments to the carbonyl carbon, each arrow spanning two or three bonds. Where arrows from separate fragments converge on one carbon, the fragments must join there.

Real-world analogy

HMBC is like reading faint signatures on shared documents. Two groups of colleagues who never speak directly may both have signed the same memo, proving they work in the same department. The signatures show a link but not how many desks apart the signers sit.

Real-world example

In the structure elucidation of new natural products, HMBC is often the decisive experiment, because many such compounds contain several quaternary carbons and ester or ether linkages. Regulatory structure proofs for new drug molecules routinely include tables of key HMBC correlations.

Why?

Why does HMBC use a delay of tens of milliseconds while HSQC uses under 2 ms? The coherence transfer needs a delay of about 1/(2J). With J ≈ 8 Hz, that is roughly 60 ms; with ¹J ≈ 145 Hz it is only about 3.4 ms, halved in the INEPT design.

Common misconception

"An HMBC peak always means a three-bond relationship." Correlations over two bonds are common, four-bond peaks occur in conjugated systems, and residual one-bond artefacts can appear. The bond count must be inferred, not read off the spectrum.

Worked example

Question: A compound contains an isolated CH₃ singlet at 2.1 ppm (δC 30) and an ethyl fragment, OCH₂ at 4.1 ppm and CH₃ at 1.25 ppm. HMBC shows both the 2.1 ppm singlet and the 4.1 ppm quartet correlating to a carbon at 171 ppm absent from HSQC. Propose the structure.

Reasoning: The 171 ppm carbon is a carbonyl. The CH₃ at 2.1 ppm is two bonds from it (CH₃–C=O). The OCH₂ is three bonds away (CH₂–O–C=O), consistent with an ester oxygen.

Answer: Ethyl ethanoate, CH₃C(=O)OCH₂CH₃.

Quick check

1. Why can HMBC connect fragments that COSY cannot, even when no proton–proton coupling links them? Answer: HMBC detects protons coupled to carbons two or three bonds away, including quaternary and carbonyl carbons that interrupt proton coupling.

Exam focus

State that HMBC shows ²J and ³J carbon–proton correlations and suppresses ¹J. Explain its role in placing quaternary carbons and linking fragments across heteroatoms. Remember that bond count is ambiguous and missing peaks prove little.

Advanced insight

Because ³J(CH) depends on dihedral angle, quantitative long-range coupling measurements from HMBC-type experiments can inform conformation and relative stereochemistry. Computer-assisted structure elucidation programs take COSY, HSQC and HMBC correlations as constraints, generate every structure consistent with them, and rank candidates by predicted ¹³C shifts, helping to avoid human bias toward a favoured answer.

Summary

HMBC correlates protons with carbons two or three bonds away using delays tuned to small long-range couplings, and filters out one-bond peaks. It reveals quaternary and carbonyl carbons and links fragments across protonless atoms. Bond counts are ambiguous, peaks may be missing for geometric reasons, and residual ¹J artefacts must be recognised.

Practice questions

1. Why are methyl singlets especially useful in HMBC analysis? Answer: They give strong, sharp signals with reliable correlations to the carbon they are attached to and its neighbours, anchoring quaternary centres. 2. In a benzene ring, why might an HMBC peak to a meta carbon be stronger than one to an ortho carbon? Answer: Three-bond coupling across the ring is typically around 7–8 Hz, larger than the two-bond coupling of about 1 Hz. 3. A proton shows two peaks about 140 Hz apart in F2 at the shift of its own carbon. What are they? Answer: Residual one-bond correlations that escaped the low-pass J filter, not genuine long-range peaks. 4. How does HMBC show that a methoxy group belongs to an ester rather than an ether on an aromatic ring? Answer: Its protons correlate to a carbonyl carbon near 165–175 ppm rather than to an aromatic carbon near 155–160 ppm.