Coupling Constants (J) and Structural Information

Vicinal coupling constants remain a first-line, non-destructive tool for solving stereochemistry in synthetic and natural-product chemistry, precisely because they require no crystal and no isotopic label — just a well-resolved one-dimensional proton spectrum. In carbohydrate chemistry, $^3J_{HH}$ values between ring protons are used routinely to establish whether a sugar adopts a particular chair conformation and whether substituents are axial or equatorial, feeding directly into the assignment of anomeric configuration ($\alpha$ vs $\beta$ glycosides) in oligosaccharide synthesis and glycobiology. In peptide and protein NMR, the backbone $^3J_{HN\alpha}$ coupling constant (between the amide NH and the $\alpha$-CH) is Karplus-related to the backbone dihedral angle $\phi$ and is one of the classical restraints used, alongside NOE distances, in solution-state protein structure determination — a direct descendant of the same physical relationship developed here, scaled up to a macromolecule. In medicinal chemistry and total synthesis, $J$-based cis/*trans* alkene assignment is a routine quality-control step: a Wittig, Horner–Wadsworth–Emmons, or cross-metathesis reaction's geometric outcome is confirmed by measuring the vinyl coupling constant rather than relying solely on chromatographic behavior or NOE, since $J$ requires no additional experiment beyond the standard proton spectrum already recorded for every intermediate. Coupling-constant analysis also underlies structure elucidation of complex natural products (macrolides, polyketides, terpenoids) where dozens of adjacent stereocenters must be assigned; Karplus-type relationships, cross-validated with NOE data and increasingly with computed (DFT-predicted) coupling constants, remain central to that workflow. The same contact-interaction physics extends into $^{1}J_{CH}$ measurements used to probe hybridization changes in strained rings and into hetero-coupling constants ($^3J_{HP}$, $^3J_{HF}$) used analogously in organophosphorus and fluorine chemistry. Methodologically, $J$-analysis is the direct precursor to the coupling-network mapping performed in intro-2d-nmr-cosy, and the subtleties of when $J$ and chemical-shift differences begin to interfere (second-order effects, "roofing," and strongly coupled spin systems) are treated fully in second-order-and-complex-splitting.

Ask the tutor: How do I extract two different vicinal coupling constants from a proton that appears as a "dd" when the two J values are close in size and the lines start to overlap?

Start with what a coupling constant is not. It is not a chemical shift difference dressed up in different units; a chemical shift difference in Hz scales linearly with the spectrometer's field $B_0$, because shift is measured in ppm (a ratio to the Larmor frequency) and then converted to Hz by multiplying by the operating frequency. A coupling constant, by contrast, is measured directly and only in Hz, and that number is the same whether you are sitting in a 300 MHz or a 900 MHz magnet. This single experimental fact — do the line spacings change when you switch magnets? — is the cleanest diagnostic chemists have for telling shift-difference-driven splitting apart from true scalar coupling, and it is the reason $J$ can be tabulated as a molecular constant independent of instrument. Why is $J$ field-independent while $\Delta\delta$ is not? The chemical shift arises from the external field $B_0$ inducing local electron currents that shield or deshield a nucleus; double the field, double the induced shielding effect in absolute frequency terms. Coupling, however, is transmitted entirely through the bonding electrons connecting two nuclei — it is an internal magnetic communication that exists even in a zero external field (it shows up, for instance, in the internal splittings of NMR spectra recorded at very different fields, and is what determines fine structure in zero-field NMR and in ESR hyperfine coupling to nuclei). The mechanism is called the Fermi contact interaction: a nuclear spin polarizes the spin of the electron(s) immediately overlapping it (this requires finite electron density at the nucleus, hence s-orbital character), and that polarization is relayed, bond to bond, until it reaches the second nucleus, whose energy is thereby perturbed depending on whether its spin is parallel or antiparallel to the first. Because the relay happens through bonding orbitals, the efficiency of transmission depends on three structural variables that make $J$ so richly diagnostic: the number of intervening bonds $n$ (giving $^nJ$ notation: $^1J$, $^2J$, $^3J$, and rarely observable $^4J$ or $^5J$ through conjugated/rigid systems), the s-character and hybridization of the intervening carbons (sp carbon transmits more efficiently than sp³), and — critically for stereochemistry — the dihedral angle between the C–H bonds for three-bond (vicinal) coupling. This last dependence is the Karplus relation, and it is the single most exploited relationship in conformational NMR analysis.

The Karplus curve is not a fitted curiosity; it follows from how the C–H bonding orbitals overlap with the connecting C–C $\sigma$-bond as a function of dihedral angle. When the two C–H bonds are eclipsed ($\phi = 0°$) or anti ($\phi = 180°$), their orbitals align favorably with the C–C $\sigma$ framework, maximizing the coupling pathway; at $\phi \approx 90°$, the orbitals are nearly orthogonal to the relay pathway, and coupling nearly vanishes.

The consequence for a working structural chemist is direct: measure $J$, read off (via the Karplus curve, or a table calibrated against known geometries) the dihedral angle, and thereby determine whether two substituents are syn or anti, whether a ring proton is axial or equatorial, or whether an alkene is cis or trans.

The empirical Karplus relation for vicinal proton–proton coupling is written as a function of the H–C–C–H dihedral angle $\phi$:

where:

• $\phi$ is the dihedral (torsion) angle between the two C–H bonds, measured $0°$–$180°$
• $A, B, C$ are empirical constants (commonly $A \approx 7$–$9$ Hz, $B \approx -1$ to $-1.5$ Hz, $C \approx 5$–$6$ Hz for unsubstituted sp³–sp³ systems; substituent electronegativity and ring strain shift these)
• $^{3}J_{HH}(\phi)$ is the resulting vicinal coupling constant in Hz

Reading it: the $\cos^2\phi$ term dominates and is symmetric about $\phi = 90°$, guaranteeing that both $\phi \approx 0°$ (eclipsed/syn) and $\phi \approx 180°$ (anti) give large $J$, while $\phi \approx 90°$ gives a near-zero coupling; the linear $B\cos\phi$ term breaks the exact symmetry between the $0°$ and $180°$ maxima, making the anti value slightly larger than the syn value in most calibrations. The second governing relationship is simply the statement of field-independence, which is what makes $J$ usable as a structural constant in the first place:

where:

• $\Delta\nu_{\text{obs}}$ is the observed peak spacing measured directly in Hz on the recorded spectrum
• $B_0$ is the external static magnetic field of the spectrometer

Reading it: if you record the same multiplet at two different field strengths and convert everything to Hz, the coupling-derived spacings are identical, while any spacing arising from a chemical shift difference (which scales with $B_0$ through the Larmor relation $\nu = \gamma B_0/2\pi$) will change proportionally between the two spectra. Finally, the number of bonds separating the coupled nuclei sets the typical range and sign of $J$, giving the empirical hierarchy used constantly in spectral assignment:

where each superscript $n$ in $^{n}J$ counts the bonds in the shortest path linking the two coupled nuclei, and the inequality reflects the exponential-like fall-off of contact-interaction efficiency with the number of intervening $\sigma$-bonds through which spin polarization must relay.

Example 1 (Foundational) — Distinguishing shift-difference splitting from true coupling using two field strengths. A two-proton pattern is recorded at $300$ MHz and shows two doublets, each with an internal line spacing of $10.5$ Hz, and the two doublet centers separated by $150$ Hz. The same sample is then run at $600$ MHz. Step 1. At $600$ MHz, re-measure both features. Suppose the internal doublet spacing is again $10.5$ Hz, but the separation between doublet centers is now $300$ Hz. Step 2. Apply the diagnostic: any spacing that doubled when the field doubled must be a chemical shift difference (converted to Hz, it scales with the spectrometer frequency, since $\Delta\nu = \Delta\delta \times \nu_0/10^6$). Any spacing that stayed fixed at $10.5$ Hz is a true coupling constant $J$. Step 3. Conclude: $J_{HH} = 10.5$ Hz for this pair (field-independent), while $\Delta\delta = 150\text{ Hz}/300\text{ MHz} = 0.5$ ppm at the lower field, consistently reproduced as $300\text{ Hz}/600\text{ MHz} = 0.5$ ppm at the higher field. Limiting-case check. If the compound were achiral and the two protons were homotopic (chemically and magnetically equivalent, e.g., the two protons of a $CH_2$ in a symmetric environment with no coupling partner distinguishing them), $\Delta\delta \to 0$ and the "doublet" pattern would collapse to a singlet regardless of any nonzero $J$ between equivalent nuclei, since coupling between chemically equivalent nuclei produces no observable splitting — this is the standard equivalence caveat and is fully consistent with the field-independence argument, since $J$ between equivalent spins never appears in the spectrum at all.

Example 2 (Applied) — Assigning axial/equatorial protons in a cyclohexane using Karplus-derived $J$ values. A rigid, conformationally locked cyclohexane derivative (e.g., a trans-4-tert-butylcyclohexanol, where the bulky tert-butyl group locks the ring in one chair) shows the carbinol proton $H_1$ as a triplet-of-triplets pattern. Two coupling constants are extracted: $J_1 = 10.8$ Hz and $J_2 = 4.2$ Hz. Step 1. Recall the geometric dihedral angles in a chair cyclohexane: axial–axial (ax–ax) vicinal protons have $\phi \approx 180°$ (anti), diaxial–diequatorial or axial–equatorial protons have $\phi \approx 60°$, and equatorial–equatorial protons also sit near $\phi \approx 60°$. Step 2. Apply the Karplus relation qualitatively: $\phi \approx 180°$ predicts large $J$ (typically $8$–$13$ Hz), while $\phi \approx 60°$ predicts small $J$ (typically $2$–$5$ Hz). Step 3. Match the measured values: $J_1 = 10.8$ Hz is consistent with an axial–axial (anti) relationship, meaning $H_1$ itself must be axial (an equatorial proton has no anti-periplanar ring-proton partner available, since all its vicinal neighbors sit at $\approx 60°$ dihedral angles to it). $J_2 = 4.2$ Hz is consistent with the axial–equatorial (or equatorial–equatorial) gauche relationship. Step 4. Since $H_1$ shows one large ($10.8$ Hz, to the other axial proton on each flanking carbon) and one small (to the equatorial proton on each flanking carbon) coupling, repeated symmetrically on both sides (hence "triplet of triplets" — coupling to two equivalent axial neighbors giving $J_1$, and two equivalent equatorial neighbors giving $J_2$), the substituent on $C_1$ (the OH) must be equatorial, because $H_1$ occupies the axial position. Limiting-case check. If the ring were instead conformationally mobile (unlocked by a bulky group, undergoing fast chair–chair interconversion on the NMR timescale), the observed $J$ would be a population-weighted average, $J_{obs} = x_A J_A + (1-x_A) J_B$ over the two chair populations, collapsing toward an intermediate value (often $6$–$8$ Hz) rather than cleanly resolving into a large/small pair — this is exactly why locked or strongly biased systems (like 4-tert-butylcyclohexanes) are the standard textbook substrates for this analysis.

Example 3 (Exam/competition-level) — Using coupling-constant and multiplicity data together to assign a trisubstituted alkene geometry and a vicinal stereocenter relationship simultaneously. A synthetic intermediate, methyl (E or Z)-3-methylpent-2-enoate fragment analog, shows a vinylic proton $H_3$ as a quartet with $J = 7.1$ Hz (coupling to an adjacent $CH_3$, four bonds away through the allylic system is normally small, so this must instead be a homoallylic $^4J$ or, more likely in the real molecule, coupling to the methyl on the same carbon framework via allylic $^4J_{allylic}$ — here we take it as the diagnostic long-range allylic coupling, since $^{3}J$ to a directly bonded vinylic H is absent in a trisubstituted, tetra-substituted-adjacent alkene) alongside a distinguishing NOE-independent chemical clue: an authentic $^{4}J_{allylic}$ (H–C=C–C–H, "W" or allylic coupling) of $1.3$ Hz to the vinylic methyl, plus comparison against a model disubstituted alkene fragment in the same synthesis with unambiguous $^{3}J_{trans} = 15.6$ Hz. Step 1. Recognize that for a trisubstituted alkene (three non-hydrogen substituents, one vinylic H), there is no vicinal $H$–$C{=}C$–$H$ coupling available at all, because only one sp² carbon bears a hydrogen; geometric assignment must instead come from long-range allylic coupling ($^{4}J$, typically $0$–$3$ Hz) and/or from NOE — but the question restricts us to $J$-based reasoning, so we use the magnitude and known angular dependence of $^{4}J_{allylic}$. Step 2. Recall that $^{4}J_{allylic}$ through the fragment $H$–$C{=}C$–$CH_3$ also has a Karplus-like angular dependence, but referenced to the dihedral angle between the allylic $C$–$H$ (of the methyl) and the $\pi$-system: coupling is largest ($\approx 1$–$3$ Hz) when the allylic C–H bond is aligned perpendicular to the alkene plane (maximizing hyperconjugative overlap of the allylic $\sigma(C$–$H)$ with the $\pi^*$ orbital, which is the same overlap channel that governs the relay), and smallest when the allylic C–H lies in-plane. Step 3. Use the measured $^{4}J_{allylic} = 1.3$ Hz, which is a "cis-allylic" coupling in the diagnostic literature sense (allylic coupling between substituents on the same side of the double bond is typically somewhat larger in magnitude than between substituents on opposite sides, because the "W" or planar-zigzag pathway achieves better through-bond alignment when the two coupled protons/groups are cis rather than trans across the double bond) — combined with knowledge of the synthetic route (a Wittig or HWE olefination known to be $E$-selective for this substrate class), the $1.3$ Hz value corroborates the assignment as consistent with the vinylic $H$ and vinylic $CH_3$ being cis to one another across the double bond (i.e., $H_3$ and the methyl are cis, fixing the full substitution pattern once combined with knowledge of which groups are attached where). Step 4. Cross-check using the internal reference $^{3}J_{trans} = 15.6$ Hz measured on a disubstituted model fragment elsewhere in the same synthetic sequence: this confirms the spectrometer, sample, and calibration are all consistent with standard trans vicinal alkene couplings in the 12–18 Hz range, giving confidence that the smaller, long-range $1.3$ Hz value is being correctly assigned as $^{4}J$ rather than a mismeasured $^{3}J$. Limiting-case check. If the allylic C–H bond in the methyl group could rotate completely freely (as it does, since a methyl rotates rapidly relative to the NMR timescale even though the alkene itself does not), the observed $^{4}J_{allylic}$ is itself a rotationally averaged quantity over all three C–H bonds of the methyl group, $J_{obs} = \tfrac{1}{3}\sum_{i=1}^{3} J(\theta_i)$, which is why allylic couplings, while diagnostic, are always smaller in magnitude and noisier as a stereochemical tool than the rigid, non-averaged vicinal alkene coupling used in Example (from the second derivation) — a caution any exam-level answer should explicitly flag.

The Karplus relation is named for Martin Karplus, a theoretical chemist who, in 1959, derived the angular dependence of vicinal proton–proton coupling from valence-bond theory calculations, years before NMR instruments routinely resolved fine coupling in complex natural products. Karplus's original derivation used a molecular-orbital / valence-bond treatment of the ethane fragment, calculating how the Fermi contact term varied with the H–C–C–H dihedral angle under a set of simplifying assumptions about the intervening $\sigma$-bond framework — essentially the same overlap-based argument walked through mechanistically above, but carried out with the (then very new) tools of quantum chemistry rather than curved arrows. He later refined the parametrization repeatedly (the constants $A$, $B$, $C$ are not universal; Karplus himself, and many groups since, have published substituent-corrected versions), emphasizing that the relation is a physically-motivated trend, not an exact universal formula, and that electronegative substituents, ring strain, and bond-angle distortions all shift the calibration constants measurably. Karplus went on to a broader career in theoretical and computational chemistry, ultimately sharing the 2013 Nobel Prize in Chemistry (with Michael Levitt and Arieh Warshel) for developing multiscale models for complex chemical systems — work that grew directly out of the same instinct for connecting quantum-mechanical detail to observable macroscopic (or in this case, spectroscopic) behavior that produced the coupling-constant relation decades earlier.