RF Pulses and Pulse Sequence Physics

Version 1.0 — July 2026

1. Executive Summary

This page covers, in depth, the fourth of the eleven planned topic groups listed in Section 4.4 of the MRI Physics — Fundamentals and Principles master page: the physics of the radiofrequency pulses that make up every MRI pulse sequence. Four tightly linked topics are covered: how flip angle is chosen for optimal signal (the Ernst angle), the three broad categories of RF pulse design (selective, non-selective, and adiabatic), how refocusing pulses physically differ between spin-echo and gradient-echo sequence families, and the magnetization transfer phenomenon that arises from off-resonance RF irradiation of macromolecule-bound protons. Together, these determine essentially every aspect of how a pulse sequence’s RF chain is designed.

2. Relationship to the Master Page

The master page introduces RF excitation and the rotating reference frame only briefly (Section 3.2), and the companion child pages Fundamentals of Nuclear Magnetic Resonance (which introduces the resonance condition, flip angle, and the basic rotating-frame picture) and Relaxation Phenomena (which introduces T1/T2/T2* and the spin-echo refocusing mechanism) are both assumed as background here and not repeated. This page adds: the quantitative Ernst-angle optimisation of flip angle for steady-state sequences, the engineering distinction between non-selective, frequency-selective, and adiabatic RF pulses, a detailed physical comparison of spin-echo vs gradient-echo refocusing mechanisms, and the magnetization-transfer phenomenon as a distinct RF-pulse-driven contrast mechanism in its own right.

3. Flip Angle and the Ernst Angle

3.1 Flip Angle Recap and Steady-State Sequences

As introduced in the companion Fundamentals of Nuclear Magnetic Resonance child page, flip angle (α) is the angle through which an RF pulse rotates the net magnetization away from the z-axis, determined by the strength and duration of the applied B1 field. For a single excitation followed by full longitudinal recovery, a 90° flip angle maximises the immediately available transverse signal. However, the great majority of clinical pulse sequences do not wait for full T1 recovery between excitations — they operate in a repeated, steady-state regime governed by TR (repetition time) — and in this steady-state regime, a 90° flip angle is very rarely the flip angle that maximises signal.

3.2 The Ernst Angle

Richard Ernst’s 1966 theoretical treatment of pulsed Fourier-transform magnetic resonance derived the flip angle that maximises steady-state signal for a given tissue T1 and a given TR — now universally called the Ernst angle [1]:

cos(α_Ernst) = e^(−TR/T1)

This relationship captures a direct physical trade-off: for a fixed TR, tissues with longer T1 have a lower optimal (Ernst) flip angle, because less longitudinal magnetization has recovered by the time of the next excitation, and a smaller flip angle “spends” a correspondingly smaller (and therefore more sustainable) fraction of the available magnetization on each excitation. As TR becomes very long relative to T1, the Ernst angle approaches 90°, recovering the simple single-excitation case.

3.3 Practical Implications for Sequence Design

The Ernst angle relationship is the direct physical basis for flip-angle selection in essentially every steady-state gradient-echo sequence documented across MRIninja’s protocol pages — a fact discussed practically, without the underlying derivation, in the MRI Parameters cluster. It is worth noting explicitly, however, that the Ernst angle is derived specifically for a spoiled, steady-state, single-tissue signal-maximisation scenario: it does not by itself account for T2/T2* weighting considerations, multi-tissue contrast optimisation, specific absorption rate (SAR) constraints, or the different steady-state behaviour of non-spoiled (coherent, e.g. balanced SSFP-family) sequences, all of which can push a real protocol’s chosen flip angle away from the strict Ernst-angle value for good clinical reasons.

4. Selective, Non-Selective, and Adiabatic Pulses

4.1 Non-Selective Pulses

A non-selective (hard) RF pulse is designed to excite the entire sensitive volume of the receive coil roughly uniformly, regardless of spatial position — typically a short, high-bandwidth rectangular pulse. Non-selective pulses are used whenever spatial selectivity is not required at the excitation stage itself (for example, as part of certain spectroscopy sequences, or as refocusing pulses in sequences where slice selection has already been established by the preceding excitation pulse).

4.2 Frequency-Selective (Slice-Selective) Pulses

A frequency-selective (soft) RF pulse is deliberately shaped — most commonly with a sinc-like or similar smoothly varying amplitude envelope — so that its frequency spectrum is concentrated within a narrow bandwidth, rather than being broadband like a hard pulse. When applied simultaneously with a slice-select gradient (which makes local Larmor frequency a linear function of position along one axis, as introduced in the master page’s spatial-encoding discussion), a frequency-selective pulse excites only the thin slab of tissue whose local Larmor frequency falls within the pulse’s bandwidth — the physical basis of slice selection, used in the excitation pulse of essentially every 2D pulse sequence on MRIninja. The trade-off inherent to this approach is a fundamental time-bandwidth relationship: achieving a sharper, more rectangular slice profile (less “slice profile blur” at the slice edges) requires a longer pulse duration for a given desired slice thickness and bandwidth.

4.3 Adiabatic Pulses

Adiabatic pulses are a distinct third category: pulses that are simultaneously amplitude- and frequency- (or phase-) modulated, designed so that the effective magnetic field experienced by the spins (in the appropriate rotating/tilted frame) changes direction slowly enough, relative to the rate of spin precession around it, that the magnetization vector tracks the effective field’s changing direction throughout the pulse — the adiabatic condition. The landmark theoretical treatment by Silver, Joseph, and Hoult provided an exact analytical solution demonstrating how this class of pulse could achieve highly uniform spin inversion despite substantial variation in the actual B1 field strength delivered to different parts of the tissue [2]. This B1-insensitivity is adiabatic pulses’ central practical advantage: because they do not depend on the RF pulse achieving a precise nominal flip angle everywhere in the tissue, they perform reliably in situations where B1 field homogeneity is genuinely difficult to guarantee — most notably with surface/local receive-transmit coils, at high and ultra-high field strengths (where B1 inhomogeneity becomes a materially larger practical problem, as discussed in the planned High-Field Physics child page group), and in fat-suppression and inversion-recovery applications that specifically require reliable, uniform inversion.

5. Refocusing Pulses and Echo Formation

5.1 Spin-Echo Physics

As detailed in the companion Relaxation Phenomena child page, a 180° refocusing pulse applied partway through a dephasing period reverses each spin’s accumulated phase error due to static field inhomogeneity, causing the spins to rephase and form a measurable echo at a predictable time — the spin-echo mechanism originally described by Hahn in 1950 [3]. Because this refocusing pulse actively reverses the field-inhomogeneity (T2′) contribution to dephasing while leaving true molecular T2 dephasing untouched, spin-echo-family sequences (and their fast/turbo multi-echo variants) produce genuinely T2-weighted contrast, governed by true T2 rather than the faster T2*.

5.2 Gradient-Echo Physics

Gradient-echo sequences form an echo through an entirely different physical mechanism, with no RF refocusing pulse at all: after excitation, a dephasing gradient lobe is deliberately applied, followed by a rewinding gradient lobe of opposite polarity that reverses the gradient-induced dephasing specifically (and only that dephasing) — forming an echo at the point where the gradient-induced phase has been fully rewound. Critically, because no RF refocusing pulse is used, the static-field-inhomogeneity (T2′) contribution to dephasing is never actively reversed in a gradient-echo sequence — it continues to accumulate throughout the sequence, exactly as in a raw, unrefocused FID. Gradient-echo signal is therefore always governed by the faster T2* decay, not true T2, which is the direct physical reason gradient-echo-family sequences (introduced generically in the FLASH gradient-echo work of Haase and colleagues [4]) are described as T2*-weighted rather than T2-weighted, and are inherently more sensitive to susceptibility-related signal loss and artefact than spin-echo-family sequences acquired with comparable timing.

5.3 Spin-Echo vs Gradient-Echo — Physical Comparison

6. Magnetization Transfer

6.1 Physical Mechanism

Biological tissue water protons exist, physically, in two exchanging populations (pools): a “free” water pool, with a comparatively long T2, that is directly responsible for the conventional MRI signal, and a “restricted” or “bound” macromolecular pool (protons bound to large, immobile structures such as membranes and myelin), whose extremely short T2 makes its signal invisible to conventional MRI sequences entirely. Wolff and Balaban’s original 1989 description of magnetization transfer contrast demonstrated that applying an off-resonance RF pulse — tuned specifically to the broad resonance frequency range of the invisible, restricted macromolecular pool, well away from the free water resonance itself — selectively saturates that restricted pool’s magnetization, which is then partially transferred to the free water pool via cross-relaxation and chemical exchange, causing a measurable reduction in the ordinarily visible free-water MRI signal despite the off-resonance pulse never directly irradiating the free water protons at all [5].

6.2 The Magnetization Transfer Ratio and Tissue Specificity

The magnitude of this signal reduction — conventionally quantified as the magnetization transfer ratio (MTR), comparing signal with and without the off-resonance saturation pulse — varies substantially between tissues, in direct proportion to how large a restricted macromolecular pool a given tissue contains. Highly myelinated white matter shows a substantially larger MT effect than grey matter, which in turn shows a larger effect than cerebrospinal fluid (which contains essentially no restricted macromolecular pool) — the physical basis for magnetization transfer’s clinical use, elsewhere on MRIninja’s protocol pages, in contexts where macromolecular/myelin content is the object of clinical interest.

6.3 Magnetization Transfer as an Unintended Effect

Magnetization transfer is not only a deliberate contrast mechanism; it also occurs, usually unintentionally, whenever off-resonance RF energy is applied for other reasons — including, notably, the off-resonance fat-saturation pulses and certain multi-slice acquisition schemes discussed elsewhere on MRIninja — meaning some degree of incidental magnetization-transfer-related signal reduction is a normal, expected side effect of several completely unrelated technical choices, not solely of dedicated MT-weighted sequences.

7. Mathematical Formalism — RF Terms in the Bloch Equations and the Adiabatic Condition

The Bloch equations, introduced in the companion Fundamentals of Nuclear Magnetic Resonance child page, include explicit terms describing how the applied B1 field rotates the magnetization vector; flip angle (Section 3) is the direct, integrated consequence of these RF terms acting over the pulse duration. The adiabatic condition underlying adiabatic pulses (Section 4.3) can be expressed formally as a requirement that the rate of change of the effective field’s orientation remain much smaller than the instantaneous precession (nutation) frequency around that effective field throughout the pulse — a condition Silver, Joseph, and Hoult’s exact Bloch-Riccati-equation solution demonstrated could be satisfied robustly across a wide range of actual B1 amplitudes, which is the formal mathematical reason adiabatic pulses tolerate B1 inhomogeneity so much better than conventional constant-amplitude pulses [2].

8. MRI Technologist and Radiologist Pearls — Common Misconceptions

• “90° is always the best flip angle.” Only true for a single excitation with full T1 recovery beforehand; in any steady-state sequence, the Ernst angle (Section 3.2) — generally well below 90° for typical TR/T1 combinations — maximises signal instead.
• “Selective and adiabatic pulses are the same kind of thing.” Selectivity (Section 4.2) concerns spatial/frequency targeting; adiabaticity (Section 4.3) concerns robustness to B1 inhomogeneity — the two properties are largely independent, and adiabatic pulses can themselves be designed to be either selective or non-selective.
• “Gradient-echo and spin-echo sequences just use different hardware timing tricks to reach the ‘same’ echo.” They are governed by fundamentally different physics (Section 5.3): only the spin-echo mechanism actively compensates for field inhomogeneity, so the two families are never simply interchangeable timing variants of one another for a given desired contrast.
• “Magnetization transfer is an artefact to be avoided.” It is a genuine, physically real, and often clinically useful contrast mechanism (Section 6.2) in its own right, even though it also frequently occurs as an unintended side effect of unrelated RF choices (Section 6.3).

9. Practical and Clinical Relevance

Flip-angle and pulse-type decisions documented throughout MRIninja’s protocol and parameter pages are direct practical applications of this page’s physics. The choice between spin-echo-family (T2-true, higher SAR, slower) and gradient-echo-family (T2*-weighted, lower SAR, faster) sequences, discussed extensively in the MRI Sequences cluster, is fundamentally the Section 5 refocusing-mechanism distinction. Fat-suppression and inversion-recovery techniques’ reliance on adiabatic inversion pulses at higher field strengths, and off-isocentre or surface-coil applications generally, are direct practical consequences of Section 4.3. Incidental magnetization-transfer-related signal changes (Section 6.3) are a relevant, if often under-appreciated, consideration whenever comparing signal intensities across sequences that use different off-resonance RF content.

11. Evidence Gaps and Ongoing Debate

• Optimal flip-angle strategy beyond the simple Ernst-angle case. The classic Ernst angle (Section 3.2) is derived for a single-tissue, spoiled, steady-state scenario; optimal flip-angle selection for multi-tissue contrast optimisation, non-spoiled coherent steady-state sequences, and SAR-constrained high-field protocols remains a more complex, actively developed area without a single settled formula, and different vendors/institutions make different practical trade-offs.
• Adiabatic pulse SAR trade-offs at high field. Adiabatic pulses’ B1-insensitivity advantage (Section 4.3) is partly offset by their generally higher SAR cost relative to conventional pulses of comparable selectivity, a trade-off that becomes progressively more clinically material at higher field strengths — addressed more fully in the planned SAR, Bioeffects, and MRI Safety Physics child page group (master page Section 4.9) than here.
• Standardisation of quantitative magnetization transfer methodology. As noted in Section 10.2, quantitative (rather than simple-ratio) magnetization transfer modelling remains an active research area without full methodological standardisation across vendors and research groups, limiting direct cross-study and cross-site comparability of quantitative MT metrics.

12. Evidence-Based References

A. Guidelines / Consensus / Society Recommendations

No dedicated society guideline exists for foundational RF pulse and pulse-sequence 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.

C. Important Prospective / Original Studies

D. Technical MRI Papers

End of document — RF Pulses and Pulse Sequence Physics — Child Page under the MRIninja MRI Physics — Fundamentals and Principles master page — v1.0 — July 2026 Parent page: MRI Physics — Fundamentals and Principles