Fundamentals of Nuclear Magnetic Resonance

Version 1.0 — July 2026

1. Executive Summary

This page covers, in depth, the first of the eleven planned topic groups listed in Section 4.1 of the MRI Physics — Fundamentals and Principles master page: the basic physical chain of events that makes magnetic resonance possible at all. Four tightly linked phenomena are covered: nuclear spin and precession (with the Larmor equation), the origin of net magnetization in a strong static field (B0), how a radiofrequency pulse tips that magnetization away from equilibrium (best understood in the rotating reference frame), and the free induction decay (FID) signal this produces. Everything downstream in MRI — every sequence, every parameter, every image — is built on these four phenomena.

2. Relationship to the Master Page

The master page introduces these four topics briefly, at survey level, in Sections 3.1–3.2. This child page assumes that survey-level introduction as background and does not repeat it. What this page adds: the quantum-mechanical origin of nuclear spin, a fuller derivation and physical interpretation of the Larmor equation, the Boltzmann-statistics basis of net magnetization, a proper treatment of the rotating reference frame and the resonance condition, the physical origin and mathematical form of the FID envelope, and a brief presentation of the Bloch equations as the formal mathematical description that ties all of this together. Relaxation (T1/T2/T2*) is deliberately treated only to the minimum extent needed to explain the FID envelope’s decay — full treatment of relaxation is the subject of the separate, not-yet-published Relaxation Phenomena child page (Section 4.2 of the master page’s roadmap) and is not duplicated here.

3. Nuclear Spin, Magnetic Moment, and the Larmor Equation

3.1 Quantum Origin of Nuclear Spin

Spin is an intrinsic quantum-mechanical property of certain atomic nuclei, unrelated to any literal physical rotation. A nucleus with a non-zero spin quantum number (I ≠ 0) possesses an associated magnetic dipole moment (μ), because a spinning charge distribution behaves, to first approximation, like a small current loop. Whether a given nucleus has non-zero spin depends on its proton and neutron count: nuclei with an odd mass number have half-integer spin (hydrogen-1, the clinically dominant nucleus in MRI, has I = 1/2), nuclei with even mass number and even atomic number have zero spin (and are therefore invisible to NMR — the reason, for instance, that carbon-12 and oxygen-16, the most abundant isotopes of those elements, contribute nothing to the MR signal despite their biological abundance), and nuclei with even mass number but odd atomic number have integer spin. Hydrogen-1’s spin-1/2 nature, combined with its overwhelming natural abundance in water and lipid molecules throughout the body, is the entire practical basis for clinical proton MRI.

3.2 Magnetic Moment and Precession

In the absence of an external magnetic field, the magnetic moments of a population of spin-1/2 nuclei point in random directions with no preferred orientation, and no net magnetic effect is observable. When placed in a strong, static external magnetic field (conventionally labelled B0 and, by convention, aligned with the z-axis), each individual nuclear magnetic moment does not simply align with the field. Instead — behaving like a spinning gyroscope subjected to gravity — it precesses around the direction of B0 at a constant angle, tracing out a cone. This precessional motion, rather than being a curiosity, is the central mechanical fact that all of MRI physics is built on: it is what allows a static magnetic field to be converted into a well-defined, tunable oscillation frequency that can subsequently be manipulated and detected using radiofrequency engineering.

The first quantitative experimental description of this behaviour — using precisely the term “Larmor frequency” for the precession — was published by Rabi, Millman, Kusch, and Zacharias in 1939, using a molecular-beam apparatus rather than the bulk-sample induction methods introduced later by Bloch and by Purcell [1].

3.3 The Larmor Equation in Depth

The precise precession frequency (the Larmor frequency, ω₀, in angular units, or f₀ in Hz) is given by the Larmor equation:

ω₀ = γB0 (angular frequency form), or equivalently f₀ = γB0 / 2π = γ̄B0 (frequency form, where γ̄ = γ/2π is the gyromagnetic ratio expressed in Hz/T rather than rad·s⁻¹·T⁻¹)

Here, γ (the gyromagnetic ratio) is a physical constant specific to each nuclear species, reflecting the ratio between its magnetic moment and its angular momentum. The Larmor equation states, in words, a relationship of profound practical importance: the precession frequency of a given nucleus is directly and linearly proportional to the local magnetic field strength it experiences, with the nuclear species itself determining only the proportionality constant. Every major capability of clinical MRI — from basic signal detection, to RF-based spatial encoding via gradients (which work only because they locally perturb B0 and therefore locally perturb the Larmor frequency in a predictable, linear way), to chemical shift phenomena (which arise from tiny, chemically-determined local variations in the effective field a nucleus experiences) — is a direct consequence of this equation’s linearity.

3.4 Gyromagnetic Ratios Across Clinically Relevant Nuclei

At clinical field strengths, hydrogen-1’s Larmor frequency works out to approximately 63.87 MHz at 1.5 T and approximately 127.74 MHz at 3 T — the precise transmit/receive frequencies that every clinical MRI scanner’s RF chain is built around.

4. The B0 Field and Net Magnetization

4.1 Boltzmann Population Distribution

Even though each individual hydrogen nucleus precesses at the same Larmor frequency in a given field, quantum mechanics restricts a spin-1/2 nucleus to only two allowed energy states relative to B0 — conventionally described as “spin-up” (parallel to B0, lower energy) and “spin-down” (antiparallel to B0, higher energy). At thermal equilibrium, the relative population of these two states is governed by the Boltzmann distribution, which depends on the energy gap between the states (itself proportional to B0) and on absolute temperature. Because the thermal energy available at body temperature is vastly larger than this energy gap even at clinical field strengths, the excess population in the lower-energy state is extremely small — on the order of a few nuclei per million at 1.5 T — yet this tiny population excess, multiplied by the enormous number of hydrogen nuclei in even a small imaging voxel, is sufficient to produce a measurable net magnetization.

4.2 The Net Magnetization Vector (Mz)

The vector sum of all individual nuclear magnetic moments in a sample, at equilibrium, is called the net (or longitudinal) magnetization, denoted Mz, and it points along the same direction as B0. Because the individual moments’ transverse (xy-plane) components are randomly phased relative to one another at equilibrium (each precessing at the same frequency but with no fixed phase relationship to its neighbours), these transverse components cancel exactly, leaving only the longitudinal component. Mz, not any individual nuclear moment, is the physical quantity actually manipulated and ultimately detected in an MRI experiment.

4.3 Field-Strength Dependence

Because the Boltzmann population excess scales with B0, the equilibrium net magnetization Mz also scales approximately linearly with B0 (to first approximation, at the field strengths and temperatures relevant to clinical MRI) — this is the fundamental physical basis for the widely quoted “higher field strength gives higher signal-to-noise ratio” rule of thumb discussed practically in the MRI Parameters cluster and, in more depth, in the planned High-Field Physics child page group (master page Section 4.10).

5. RF Excitation and the Rotating Reference Frame

5.1 Why We Need a Rotating Frame

Described in the laboratory (stationary) reference frame, the net magnetization vector’s motion during RF excitation is a rapid spiral — precessing at the Larmor frequency (tens to hundreds of MHz) while simultaneously being tipped away from the z-axis by the applied RF field, a motion that is mathematically awkward to visualise and analyse directly. The standard solution, introduced from the earliest theoretical treatments of magnetic resonance, is to mathematically transform the description into a reference frame that itself rotates around B0 at the Larmor frequency. In this rotating frame, the rapid precessional component of the motion disappears from view entirely, and the magnetization’s response to an on-resonance RF pulse simplifies to a simple, slow rotation (nutation) about the axis along which the RF field (B1) is applied — vastly easier to reason about, and the standard way flip angle, pulse sequence diagrams, and pulse design are all taught and conceptualised in practice.

5.2 The Resonance Condition

An RF pulse tips the net magnetization away from equilibrium efficiently only when its frequency matches (or very closely matches) the Larmor frequency of the nucleus being targeted — the resonance condition that gives the technique its name. An RF pulse transmitted at a frequency far from the local Larmor frequency has negligible effect on the magnetization, a fact directly exploited in frequency-selective excitation (the physical basis of slice selection) and in fat-suppression techniques that rely on the small but reliable chemical-shift-driven frequency difference between fat and water protons.

5.3 Flip Angle and the B1 Field

The angle through which the net magnetization is rotated away from the z-axis by an RF pulse — the flip angle (α) — depends on the strength of the applied RF field (B1), its duration, and (for shaped, frequency-selective pulses) its detailed waveform. In the rotating frame, for a simple rectangular on-resonance pulse, the flip angle is given by α = γB1·τ, where τ is the pulse duration — a direct, practical illustration of the same gyromagnetic ratio (γ) that governs the Larmor equation also governing how efficiently a given nuclear species responds to RF excitation.

5.4 On-Resonance vs Off-Resonance Behaviour

When the applied RF frequency is not perfectly matched to the local Larmor frequency (off-resonance excitation), the effective rotation axis in the rotating frame tilts away from the simple B1-only axis, and the achieved flip angle deviates from the nominal on-resonance value in a predictable way described by the Bloch equations (Section 7). This is not merely a theoretical nicety: off-resonance effects are the physical origin of a number of practically important phenomena addressed elsewhere on MRIninja, including chemical-shift-related signal loss at slice edges and some categories of RF-related image artefact.

6. Free Induction Decay (FID)

6.1 Physical Origin of the FID Signal

Immediately after an RF pulse tips the net magnetization away from the z-axis, it possesses a transverse (xy-plane) component that — unlike the equilibrium state — is coherent: all the individual nuclear moments contributing to it are, immediately after the pulse, precessing together, in phase, at (approximately) the same frequency. A coherently precessing bulk magnetic moment induces a real, measurable, oscillating voltage in a nearby receiver coil, by straightforward electromagnetic induction (Faraday’s law) — this induced signal is the free induction decay, or FID, and it is the most fundamental detectable signal in all of magnetic resonance, first directly observed and described in the original 1946 papers of Bloch and of Purcell and colleagues [2,3].

6.2 The FID Envelope, T2*, and Field Inhomogeneity

The FID signal does not persist indefinitely: its amplitude decays over time, because the individual nuclear moments that were initially precessing in phase gradually lose their mutual phase coherence (dephase), for two combined physical reasons — true molecular-level spin-spin (T2) interactions, and, in any real (imperfectly homogeneous) magnet, additional dephasing caused by small local variations in the actual field each nucleus experiences. The combined, faster decay actually observed is characterised by the time constant T2*, always shorter than or equal to the true molecular T2. The FID envelope, to a good approximation in a reasonably homogeneous field, decays exponentially with time constant T2*. Erwin Hahn’s landmark 1950 description of the spin echo — a radiofrequency refocusing technique that reverses the field-inhomogeneity component of this dephasing while leaving the true T2 (molecular) dephasing untouched — remains the conceptual basis for how spin-echo-family pulse sequences recover a “true” T2-weighted signal despite the field imperfections that would otherwise dominate a raw FID measurement [4].

6.3 Why the FID Alone Cannot Form an Image

A raw FID signal carries no spatial information whatsoever: it is simply a single, time-varying voltage representing the coherently decaying sum of every excited nuclear moment in the entire receive coil’s sensitive volume, with no way to distinguish signal originating from one location from signal originating from another. Forming a spatially resolved image requires the additional, deliberate application of magnetic field gradients to make position-dependent Larmor frequency and phase differences that can subsequently be decoded — the subject of the planned Signal Localisation and k-Space child page group (master page Section 4.3), not covered further here.

7. Mathematical Formalism — The Bloch Equations

Felix Bloch’s original 1946 paper did not merely report the experimental discovery of nuclear induction; it also proposed a compact set of coupled differential equations — now universally known as the Bloch equations — that phenomenologically describe the time evolution of the net magnetization vector (Mx, My, Mz) under the combined influence of an external field, RF excitation, and relaxation, without requiring a full quantum-mechanical treatment [2]. In their simplest form, the Bloch equations combine three terms for each magnetization component: precession around the effective field, decay of the transverse components toward zero with time constant T2, and recovery of the longitudinal component toward its equilibrium value M0 with time constant T1. Nearly every quantitative statement made elsewhere on MRIninja about signal behaviour, sequence timing, or contrast mechanism can, in principle, be derived directly from the Bloch equations — they remain, more than seven decades after their original publication, the standard classical formalism underlying essentially all practical MRI pulse sequence design and analysis.

8. MRI Technologist and Radiologist Pearls — Common Misconceptions and Teaching Pitfalls

• “The nucleus physically spins like a top.” Nuclear spin is a genuine intrinsic quantum property, not literal rotation; the classical spinning-top analogy is a useful visualisation for precession but should not be over-interpreted as a literal mechanical description.
• “A higher gyromagnetic ratio always means a better MRI nucleus.” Sensitivity depends on the combined effect of the gyromagnetic ratio, natural abundance, and in-vivo concentration — hydrogen-1’s clinical dominance comes from the product of a reasonably high γ and an overwhelming biological abundance, not from having the single highest γ of any nucleus.
• “T2* and T2 are just two names for the same thing.” They are related but distinct: T2* always decays faster than true T2 because it also incorporates the (in principle avoidable, via techniques like spin-echo refocusing) contribution of macroscopic field inhomogeneity, not just the (unavoidable) molecular T2 process.
• “The FID is the image.” The FID is the raw time-domain signal from a single excitation with no spatial information; it must be manipulated through spatial encoding (gradients) and mathematically transformed (via the Fourier transform, applied to the full k-space dataset, not a single FID) before an image exists.
• “Resonance frequency is a property of the scanner, not the patient.” While the nominal Larmor frequency at a given field strength is a fixed scanner property, the effective local Larmor frequency experienced by any given nucleus in the body is subtly modulated by its precise chemical environment (chemical shift) and by any local field inhomogeneity — a distinction with direct downstream consequences for fat suppression, susceptibility artefact, and off-resonance behaviour discussed elsewhere on MRIninja.

9. Practical and Clinical Relevance

Although this page is deliberately non-clinical in its own scope, every practical decision documented elsewhere on MRIninja traces back to the physics described here. Field-strength selection (1.5 T vs 3 T) is, at its physical root, a Larmor-frequency and net-magnetization decision (Sections 3.3–3.4, 4.3). Fat-suppression technique selection (spectral vs STIR vs Dixon, discussed in the anatomical protocol pages and in the planned Chemical Shift child page group) depends directly on the resonance condition’s frequency selectivity (Section 5.2). Every reported T2*-related signal loss or susceptibility artefact (discussed throughout the anatomical protocol pages) is a direct, practical manifestation of the field-inhomogeneity contribution to FID/T2* decay described in Section 6.2.

11. Evidence Gaps and Ongoing Debate

• Pedagogical debate on classical vs quantum teaching approaches: MRI physics education materials vary considerably in whether they introduce nuclear spin behaviour primarily through the classical precessing-vector picture or through a more formal quantum treatment; no consensus standard curriculum exists across radiology/technologist training programmes, and this page has deliberately used the classical picture as primary, consistent with how the great majority of clinical MRI physics teaching is actually conducted.
• Precise boundary of quantum vs classical validity: while the classical Bloch-equation picture is well established as accurate for bulk clinical imaging (Section 10.1), the precise theoretical boundary conditions under which quantum effects become non-negligible for specialised techniques (e.g. certain spectroscopy and hyperpolarisation methods) remains an active area of physics research rather than settled clinical teaching material.

12. Evidence-Based References

A. Guidelines / Consensus / Society Recommendations

No dedicated society guideline exists for foundational nuclear magnetic resonance physics as such — this is textbook/landmark-paper physics rather than a clinical practice area subject to society guidance. Category A is therefore not populated for this child page.

D. Technical MRI Papers

End of document — Fundamentals of Nuclear Magnetic Resonance — Child Page under the MRIninja MRI Physics — Fundamentals and Principles master page — v1.0 — July 2026 Parent page: MRI Physics — Fundamentals and Principles