1 / f 3 / 2 1/f^{3/2} Power Spectrum at the Phonon Bottleneck
Steven T. Bramwell · 2025
Read 1 / f 3 / 2 1/f^{3/2} Power Spectrum at the Phonon Bottleneck by Steven T. Bramwell (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-19159.txt.
Introduction
Steven Bramwell explained in twenty twenty-five a long-standing puzzle in how materials relax. Many systems respond to a push with a power-law tail whose exponent sits oddly below two, a pattern seen for decades without a settled cause. Bramwell studies the phonon bottleneck, a classic model where spins in a material can only shed energy through a limited channel of lattice vibrations.
When that channel is strongly choked, the vibrations settle into a near-steady state held up by slow spin relaxation. Bramwell computes the response to an applied field exactly in this setting. A spectral density with exponent three halves emerges, holding over a widening range of frequencies as the bottleneck tightens. He discusses which real systems might show this effect and compares it with other cases of anomalous relaxation.
Catalog record
| CREATOR | Steven T. Bramwell |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 18 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-19159
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
1 / f 3 / 2 1/f^{3/2} Power Spectrum at the Phonon Bottleneck Steven T. Bramwell Abstract The common observation of a ‘ 1 / f α 1/f^{\alpha} ’ power spectrum with α < 2 \alpha<2 constitutes one of the enduring mysteries of condensed matter physics. Here it is shown that a 1 / f α 1/f^{\alpha} power spectrum, with α = 3 / 2 \alpha=3/2 , can arise when an ensemble of two–level systems is coupled to a heat bath by means of a system of Bosonic quasiparticles. The model considered is the classic model of Faughnan and Strandberg of the phonon bottleneck, and the anomalous relaxation is associated with an approximate non-equilibrium steady state of the phonons maintained by slow spin relaxation. It is shown that a frequency-dependent susceptibility can be defined in the steady state and that the prediction α = 3 / 2 \alpha=3/2 , or the equivalent stretched exponential relaxation with exponent β = 2 − α = 1 / 2 \beta=2-\alpha=1/2 , is consistent with existing experimental data. These results give an insight into the origins of anomalous relaxation in condensed matter and the practical behaviour of two-level systems. Exponential, or Debye, relaxation has a power spectrum that tends to 1/f^{2} at high frequency f . However many real condensed matter systems – conducting DuttaHorn ; Weissman , dielectric Jonscher ; Phillips ; NonDebye and magnetic Gunnarsson ; Ocio ; Sangregorio ; Thomas ; Hallen ; Billington – show power law behavior \sim 1/f^{\alpha} with an exponent \alpha that lies in the range 1\leq\alpha<2 . While a number of particular mechanisms for this behavior have been established (see e.g. Refs. Gunnarsson ; Ocio ; Sangregorio ; Thomas ; Hallen for the magnetic case), in general it remains a mystery as to why 1/f^{\alpha} behavior is so widespread. A common explanation DuttaHorn is that the power law is constructed from a wide range of exponential timescales, but in physical terms the range of timescales required (many decades) is usually unrealistic. An exception might be where the time scales are associated with an underlying broad range of length scales, as occurs in a critical or fractal system, but such instances cannot account for the majority of observations. This leads one to suspect that there are more fundamental causes at play, arising in non-equilibrium physics and quantum dynamics. In this context, simple and realistic models and experiments that produce well-defined exponents \alpha are of evident interest to signpost the way to fundamental causes. This paper reconsiders the canonical model of the ‘phonon bottleneck’ due to Faughnan and Strandberg FS . The model exemplifies a simple non-equilibrium, and essentially quantum mechanical, model that describes an ensemble of two-level systems (spins) connected to a heat bath by an intervening system of Bosonic quasiparticles (phonons). In a certain parameter regime the model describes a non-equilibrium steady state, and it is shown here that the steady state is associated with almost perfect 1/f^{3/2} power spectrum. This result, derived below, is illustrated in Fig. 1 where the exact solution for |\chi^{\prime\prime}|/\omega (the ratio of the imaginary part of the magnetic susceptibility to the angular frequency, a measure of the power spectrum power ) is compared to the approximation : \frac{|\chi^{\prime\prime}|}{\omega}=\frac{\chi_{T}}{\sigma^{-1}+\sqrt{2\sigma}\,\omega^{3/2}} (1) (where \chi_{T} is the isothermal susceptibility), and a 1/f^{3/2} power spectrum (given f=\omega/2\pi ) is observed over many decades of frequency. Figure 1: Exact power spectrum (blue line) compared with Eq. 1 (magenta dashed line). Here \sigma=10^{5} and a power law region with \alpha=3/2 extends over 1/\sigma<\omega<\sigma . The two curves are almost coincident on a log–log scale, illustrating a rare instance where a ‘ 1/f^{\alpha} ’ power spectrum can be calculated analytically (here \alpha=3/2 and f=\omega/2\pi ). The model considers a dilute system of two-state spins that relax by directly emitting or absorbing phonons in a narrow band around the resonant frequency. Owing to ‘quantum freezing’ of phonon modes at low temperatures, the phonon specific heat becomes too small to maintain thermal equilibrium, hence the ‘bottleneck’. The phonons are treated quantum mechanically Strandberg , leading to two coupled equations of motion. \frac{dv}{dt}=-v+\sigma u(1-v) (2) \frac{du}{dt}=a\left[(v-u)-\sigma u(1-v)\right] (3) Here the variable v is defined as v=\frac{(N_{i}^{0}-N_{j}^{0})-(N_{i}-N_{j})}{(N_{i}^{0}-N_{j}^{0})} (4) where N_{i} is the number of spins in the excited state, N_{j} is the number of spins in the ground state, the superscripts represent equilibrium values and N_{i}+N_{j}=N_{i}^{0}+N_{j}^{0}=N , the number of spins. If i denotes spin up, and j denotes spin down, then the magnetic moment \mathcal{M}=N_{i}-N_{j} and the equilibrium magnetic moment is \mathcal{M}^{0}=N_{i}^{0}-N_{j}^{0} . It follows that v=(\mathcal{M}^{0}-\mathcal{M})/\mathcal{M}^{0}, a normalised measure of the deviation of magnetic moment, or magnetization, from its equilibrium value. For example, if a field is suddenly applied to an equlibrium unmagnetized system, then v will evolve from v=1 to v=0 as the new equilibrium is established at \mathcal{M}=\mathcal{M}^{0} . The variable u is defined as u=\frac{1}{\sigma}\frac{p-p^{0}}{p^{0}+\frac{1}{2}} (5) where the p ’s are mean phonon occupation numbers at the frequency corresponding to the level splitting. Finally the two parameters are defined as follows. The parameter a=\tau_{\rm s}/\tau_{\rm p} where \tau_{\rm s} is a spin relaxation time and \tau_{p} is a phonon relaxation time. The parameter \sigma is defined by \sigma=\frac{c_{s}/\tau_{s}}{c_{p}/\tau_{p}}. (6) where the c ’s are specific heats. A large value of \sigma indicates a relatively small phonon specific heat: hence it is a measure of the strength of the ‘phonon bottleneck’. Note that, in this treatment, \sigma and a are dimensionless, so the time t and angular frequency \omega are also dimensionless, defined with respect to some reference time scale. The situation of interest arises where \sigma,a\gg 1 . The phonons are strongly bottlenecked but the spins relax relatively slowly, continually maintaining a far from equilibrium phonon population. This leads to an approximate steady state where the rate of change of the phonon population can be neglected. Thus |du/dt|\ll a , and the right hand side of Eq. 3 may be set to zero. This reduces Eqs. 2 , 3 to a single equation depending only on \sigma : \frac{dv}{dt}=\frac{-v}{1+\sigma(1-v)}=-u. (7) At short times v\rightarrow 1 , but for sufficiently large \sigma , there appears to be a region at 1-v\gg 1/\sigma where this equation approximates: v\approx 1-\sqrt{2t/\sigma}\approx e^{-\sqrt{2t/\sigma}}. (8) Correspondingly, the Fourier transform of this function at high frequency should approximate v(\omega)\approx\frac{1}{\sqrt{2\sigma}\omega^{3/2}}, (9) suggesting an equation analagous to Eq. 1 for v(\omega) . These ideas are born out by numerical solution of the coupled equations. Using the example of a=10^{2},\sigma=10^{5} , Fig. 2 (upper) illustrates the solution for v(t) compared to 1-\sqrt{2t/\sigma} and to e^{-\sqrt{2t/\sigma}} , which are very close approximations at small times, and the (exact) inverse Fourier transform Fourier of the right hand side of Eq. 1 , which captures the whole curve. Fig. 2 (lower) is the corresponding numerical Fourier transform on a log-log scale, confirming a slope very close to 3/2 . Fig. 3 illustrates the corresponding u(t) compared to \frac{d}{dt}e^{-\sqrt{2t/\sigma}} (see Eqs. 7 and 8 ) which is a close approximation at short time after transient effects have died out. Figure 2: Relaxation of the magnetization difference: numerical solution to Eqs. 2 , 3 , for a=10^{2},\sigma=10^{5} . (a) the function v(t) (blue line) compared to 1-\sqrt{2t/\sigma} (gray dotted line), e^{-\sqrt{2t/\sigma}} (gray