The spin Hamiltonian is rarely truly static
In a protein, liquid or flexible molecular system, distances, orientations and electronic structure fluctuate continuously. That turns fixed spin parameters into time series and makes molecular motion part of the spin-dynamics problem rather than a separate background effect.
Turn a trajectory of molecular structures into fluctuations of spin-Hamiltonian parameters.
Interpret autocorrelation functions and spectral densities in terms of dynamical timescales.
Decide whether motion is fast, resonant with a spin transition, or effectively quasi-static.
Time-dependent Hamiltonians
Replace one structure by a trajectory
If the molecular coordinates evolve as \(\mathbf R(t)\), then the spin Hamiltonian can inherit that motion:
Hyperfine tensors can change as spin density redistributes, exchange can change dramatically with donor–acceptor geometry, dipolar tensors rotate and change with distance, and anisotropic \(g\)-tensors move with the molecular frame.
Parameter trajectories
First separate the mean from the fluctuation
For any scalar interaction parameter \(A(t)\), define
The mean determines the static part of the Hamiltonian. The fluctuation \(\delta A(t)\) contains the motion that can broaden lines, dephase coherences or drive relaxation.
For tensors the same idea applies component by component—but only after the frames are handled consistently. Comparing tensor components from snapshots in unrelated molecular frames can manufacture artificial “fluctuations” that are purely rotational bookkeeping.
Correlation functions
How long does the system remember a fluctuation?
A correlation function measures memory, not merely amplitude
What it is: The mean-square size of the fluctuation around its average value.
What it changes: It sets how strongly a fluctuating Hamiltonian parameter can perturb the spin system.
What you observe: Broader parameter distributions and, together with timescale, stronger relaxation/dephasing.
What it is: A characteristic time over which the sign and magnitude of a fluctuation remain statistically related to their earlier values.
What it changes: It determines where the fluctuation power sits in frequency space.
What you observe: Whether motion appears motionally averaged, relaxation-efficient or quasi-static on the spin timescale.
The autocorrelation function is
A rapidly decaying \(C_A(t)\) means the fluctuations lose memory quickly. A slowly decaying or multi-exponential correlation function indicates persistent structural memory or several dynamical processes.
A common teaching model is a single exponential,
where \(\tau_c\) is the correlation time and \(\sigma^2=C(0)\) is the fluctuation variance.
Spectral density
Relaxation cares about frequency content, not just fluctuation size
The spectral density tells the spin which parts of molecular motion it can 'hear'
What it is: The frequency-domain distribution of fluctuation power obtained from the correlation function.
What it changes: Relaxation is efficient when the fluctuating interaction contains power near an energy-gap frequency of the spin system.
What you observe: Frequency- and field-dependent relaxation rates such as \(T_1^{-1}\) and contributions to \(T_2^{-1}\).
What it is: The condition that molecular motion and spin precession occur on comparable timescales, often summarized as \(\omega\tau_c\sim1\) for a simple exponential model.
What it changes: It maximizes spectral weight at that transition frequency for the single-timescale model.
What you observe: A relaxation maximum as field/frequency or molecular correlation time is varied.
The spectral density is the Fourier transform of the correlation function. Using the two-sided convention,
For the single-exponential correlation model,
This tells us something important: large fluctuations are not automatically efficient at relaxing a particular spin transition. The motion must also contain spectral weight near the relevant transition frequency.
Timescale matching and spectral weight
The current fluctuation timescale is close to the region of strongest spectral overlap.
The plotted quantity is the spectral density at one chosen angular frequency, normalized to its maximum as a function of \(\tau_c\). Real relaxation rates generally combine several spectral-density values with operator-specific prefactors.
Dynamic regimes
Fast and slow motion produce different physics
Fluctuations average rapidly. This is the regime behind motional narrowing and many Markovian relaxation models.
Spectral density at the transition frequency is large, so fluctuations can drive efficient relaxation.
The system begins to look like an ensemble of slowly changing or quasi-static Hamiltonians.
There is no single universal correlation time for a protein. Side-chain motion, global tumbling, loop rearrangements and conformational exchange can all live on different timescales and couple to different spin-Hamiltonian terms.
From MD to spin parameters
A trajectory is not yet a spin-dynamics model
How densely you need quantum-chemical calculations depends on how rapidly the parameters vary and how transferable the electronic-structure model is. Interpolating a slowly varying dipolar interaction is very different from learning an exchange coupling that changes exponentially with geometry.
Practical pitfalls
Correlation functions are easy to calculate badly
If the trajectory drifts between states, a single stationary correlation function may not be meaningful.
Slow tails in \(C(t)\) are especially sensitive to trajectory length and the number of independent samples.
Tensor components must be compared in a consistent molecular or laboratory frame.
Real biomolecular correlations are often multi-timescale and can contain oscillatory or non-exponential structure.
Selected reading
Examples from my work
Magnetosensitivity of Model Flavin–Tryptophan Radical Pairs in a Dynamic Protein Environment
Directly connects protein motion and fluctuating radical-pair interactions to spin observables.
J. Phys. Chem. B (2025) →Spin Dynamics of Radical Pairs Using the Stochastic Schrödinger Equation in MolSpin
An efficient route for propagating large spin systems under stochastic dynamics.
J. Chem. Theory Comput. (2024) →Conformational Switching Controls Biradical Spin Dynamics in Flavin–Tryptophan Dyads
Shows how distinct molecular conformations can control spin-dynamic pathways.
JACS (2026) →Multiscale modeling approaches in biomolecular physics
A broader framework for linking atomistic motion to electronic and quantum observables.
Advances in Physics: X (2026) →Key external literature
Where to read next
These are deliberately selected from outside my own work: foundational papers or reviews that are especially useful for this topic.
Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. Theory and range of validity
G. Lipari and A. Szabo · JACS (1982). A classic connection between molecular correlation times, spectral densities and relaxation observables.
Open DOI →On the Theory of Relaxation Processes
A. G. Redfield · IBM Journal of Research and Development (1957). The foundational route from fluctuating interactions to reduced spin relaxation dynamics.
Open DOI →