A unified equation for saturation magnetization and spin transport in weakly disordered ferromagnets Sumanta Mukherjee 1,∗ Abstract In this report, a unified description of the loss of saturation magnetization in the presence of a distribution of finite-size effects is provided for weakly disordered spin-1/2 ferromagnets. This description allows us to derive a unified form of the Bloch equation for these systems. We further extend this approach to obtain a unified expression for spin transport in such systems. While the Mermin–Wagner theorem ref-1 ; ref-2 states that an infinite 1d or 2d Heisenberg ferromagnetic system cannot exhibit long-range order at any finite temperature due to the high density of low-energy excitations—whose associated entropy suppresses ordering—a 3d Heisenberg system, being above the lower critical dimension, experiences reduced fluctuations and can therefore exhibit long-range order at finite temperature ref-3 . A vast number of studies ref-4 ; ref-5 ; ref-6 ; ref-7 ; ref-8 ; ref-9 ; ref-10 ; ref-11 ; ref-12 have been carried out on 1d and 2d Heisenberg systems to elucidate their magnetic behavior. In this context, it has been suggested that in finite Heisenberg systems, finite-size effects can open gaps in the low-energy magnon excitation spectrum ref-13 ; ref-14 , resulting in an exponential suppression of the density of states for low-energy magnons. Such Lifshitz tail-like suppression ref-15 ; ref-16 ; ref-17 ; ref-18 of the low-energy density of states may significantly alter the magnetization behavior of 1d or 2d Heisenberg systems, often leading to ordering behavior similar to that of a 3d ferromagnetic system ref-19 ; ref-20 . While any real system is finite, the presence of intrinsic disorder, such as vacancies, can further fragment the lattice, thereby enhancing finite-size effects ref-3 . Since such disorder is random, its introduction into an otherwise infinite lattice results in a distribution of segment lengths ref-3 . This distribution of lengths, in turn, leads to a corresponding distribution of energy gaps in the system ref-31 ; ref-21 . In this report, we present a comprehensive study that incorporates the distribution of energy gaps in conjunction with the magnon spectrum and describes the resulting magnetization behavior in such systems. We calculate the reduction in saturation magnetization due to magnon excitations in the presence of a gap distribution. Furthermore, we extend this analysis to 1d, 2d, and 3d systems, deriving a unified equation—generalizing the well-known Bloch equation ref-22 ; ref-23 for magnetization loss—that captures the behavior of weakly disordered spin-1/2 Heisenberg systems. We further extend the analysis to understand spin transport phenomena in such disordered systems, deriving a unified equation for spin current. We find that the final expression for the spin current in weakly disordered magnetic systems is structurally analogous to the Efros–Shklovskii law ref-24 for electronic conductivity in disordered systems. In a disordered system, the probability that a fragmented region of length L is free of disorder and can be treated as a clean region is given by ref-3 . P(L)=a\exp\left\{-aL^{\psi}\right\},\text{ where }a=-\ln(1-p) (1) Here, p denotes the dilution or strength of the disorder, whereas \psi represents the effective dimensionality ref-3 . The prefactor a outside the exponential may not be required; however, it is often introduced as a normalization constant, for example for \psi=1 . Omitting or retaining this constant outside the exponential does not affect the outcome of the derivation, as the calculation relies on numerical normalization. On the other hand, the presence of finite size effectively discretizes the magnon spectrum in any dimension, introducing gaps (\epsilon) that depend on the system length L as ref-13 ; ref-14 . \epsilon=\frac{cn^{2}}{L^{2}} (2) Here, c is a constant, and n is an integer, typically taking values 1,2,3, \ldots In the presence of a distribution of gaps arising from the distribution of lengths, one may combine these two equations ref-3 ; ref-21 to obtain the expression for the gap distribution for a fixed n as Figure 1: The variation of <\epsilon_{m}> for different values of the parameters a , c , and \psi is shown. P(\epsilon)=\frac{an\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\} (3) Considering different values of n , one may write the total probability as P_{t}(\epsilon)=\sum_{n=1}^{n}\frac{an\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\} (4) This expression is very similar to the Lifshitz tail-like exponential suppression of the density of states, as is generally observed in disordered systems ref-15 ; ref-16 ; ref-17 ; ref-18 . With this, one may define the mean of this gap distribution as <\epsilon_{m}>=\frac{\int\epsilon P_{t}(\epsilon)d\epsilon}{\int P_{t}(\epsilon)d\epsilon}=\frac{\sum_{n=1}^{n}\int\epsilon\frac{an\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon}{\sum_{n=1}^{n}\int\frac{an\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon} (5) Furthermore, in most real systems, the gap typically opens in the low-energy region of the magnon dispersion ref-13 ; ref-25 , while the high-energy part of the spectrum, though discrete, may be treated as effectively continuous ref-25 . Hence, the above expression can be written as <\epsilon_{m}>=\frac{\int_{1}^{n}\int_{n^{2}x}^{n^{2}y}\epsilon\frac{gan\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon dn}{\int_{1}^{n}\int_{n^{2}x}^{n^{2}y}\frac{gan\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon dn} (6) While, in a discretized magnon spectrum, it may not be appropriate to include the usual dimension-dependent power-law density of states factor, one may instead introduce a degeneracy factor g in the above expression. In a crude approximation, g is at least equal to two, accounting for the symmetry of magnon modes defined by k and -k . In the low-gap limit, this factor may be generalized to an energy-dependent form g(\epsilon) ; for example, g(\epsilon)\approx\epsilon^{-1/2} in a 1d system ref-22 . With this expression, the parameters a , c , and \psi can be treated on an equal basis, as all three enter the theory in an equivalent manner and determine the mean gap <\epsilon_{m}> of the system. Figure 1 shows the dependence of <\epsilon_{m}> on each of these parameters. With these expressions, and for a spin-1/2 system, one may calculate the magnetization (saturation) loss per spin at any finite temperature due to magnon excitations as ref-23 ; ref-25 . \displaystyle M_{loss}=\mu_{B}\frac{\intP_{t}(\epsilon)d\epsilon}{\int P_{t}(\epsilon)d\epsilon} (7) \displaystyle=\mu_{B}\frac{\int_{1}^{n}\int_{n^{2}x}^{n^{2}y}\frac{gan\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon dn}{\int_{1}^{n}\int_{n^{2}x}^{n^{2}y}\frac{gan\sqrt{c}}{2\epsilon^{3/2}}\exp\left\{-a\left(n\sqrt{\frac{c}{\epsilon}}\right)^{\psi}\right\}d\epsilon dn} Figure 2: (a)–(c) The calculated variation of the saturation magnetization ( M_{s} ), obtained by varying the parameters a , c , and \psi , respectively, at different temperatures and in a fixed low magnetic field, is shown. (d) shows representative fits of \ln(M_{\text{loss}}) using Equation 11 with F(H)=1 . Figure 3: (a) shows the typical values of the exponents \alpha and \beta obtained from fitting the M_{\text{loss}} data using Equation 11 for various values of <\epsilon_{m}> , obtained through controlled variation of the parameters a , c , and \psi . (b) A typical variation of M_{s} as a function of magnetic field is shown. (c) A representative fit of \ln(M_{\text{loss}}(H)) obtained using Equation 12 is shown. Where \mu_{B} is the Bohr magneton, and the occupation number of a mode , under the discrete condition with large <\epsilon_{m}> , may be approximated by a Maxwell–Boltzmann distribution ref-22 ; ref-26 of the form given by =\exp\left\{-\epsilon/k_{B}T\right\} (8) Where k_{B} is the Boltzmann constant and T is the temperature (note the symobol and n are used to define different quantities). One may also include the effect of an external magnetic field H ref-27 , which introduces additional modifications to the gap distribution via the approximate Zeeman shift ref-26 ; ref-27 within linear spin-wave theory. It can be shown that the magnetic field effectively enters through the integration limits above, as well as through the occupation factor, modifying it to become =\exp\left\{\epsilon^{\prime}/k_{B}T\right\} (9) Where \epsilon^{\prime}=\epsilon+\mu_{B}H for a spin-1/2 system ref-27 ; ref-28 . While one may consider nonlinear effects of the magnetic field on the magnon gap, the linear effect—often observed in experiments and theory ref-27 ; ref-28 —suffices for the present discussion. With this, we may calculate the changes in the saturation magnetization M_{s} at finite temperature under different values of <\epsilon_{m}> , controlled by the parameters a , c , and \psi . Such estimations are shown in Figures 2(a)–2(c) for the parameters a , c , and \psi , respectively. The exponential suppression of the low-energy density of states, resulting in an effective suppression of the magnetization loss, is clearly visible in these figures. The magnitude of the suppression increases as <\epsilon_{m}> increases. In various contexts, it has been shown that in the presence of disorder, the saturation magnetization often follows an expression slightly different from the usual Bloch equation and can be written as ref-23 ; ref-25 ; ref-29 : M_{s}=M_{0}\left[1-\frac{M_{loss}}{M_{0}}\right]=M_{0}\left[1-CT^{3/2}\exp\left\{-BT^{\beta}\right\}\right] (10) Where B and C are arbitrary constants. While it is difficult to obtain such an expression analytically within the present model, we have attempted to simulate the saturation magnetization using an expression of the form: M_{s}=M_{0}\left[1-\frac{M_{loss}}{M_{0}}\right]=M_{0}\left[1-CT^{\alpha}\exp\left\{-BT^{\beta}\right\}F(H)\right] (11) Where F(H) arises from the magnetic-field dependence of M_{loss} , and for very low values of H used in the calculations, it may be approximated as \sim 1 . Figure 2(d) shows one such representative fit of \ln(M_{loss}) using the above expression with F(H)=1 . In the same way, M_{loss} obtained for different values of <\epsilon_{m}> , by varying the parameters a , c , and \psi , was fitted using the above expression. Figure 3(a) shows the values of \alpha and \beta obtained from such fittings. It is interesting to note that, within the uncertainties, the values of \alpha and \beta can be approximately taken as 0.5 and -0.5, respectively. In order to determine the expression for F(H) , we calculate M_{loss} as a function of H at different temperatures for specific values of the parameters a , c , and \psi . Note that, to observe a substantial change in the presence of a magnetic field, one must choose a low value of <\epsilon_{m}> . The variation of M_{s} with H at a fixed temperature of approximately 50 K is shown in Figure 3(b) . Interestingly, we find that the variation of M_{loss} with H at any temperature between 5 K and 100 K can be fitted with an expression of the form: M_{loss}(H)=D\exp\left\{-\mu H/k_{B}T\right\} (12) D is an arbitrary constant. One such representative fit is shown in Figure 3(c) . By combining this behavior in the presence of weak disorder with the above expression, one may write the overall expression for M_{s} for a spin 1/2 system of any dimension in the weak-disorder limit as: M_{s}=M_{0}\left[1-AT^{0.5}\exp\left\{-BT^{-0.5}-\left(\mu H/k_{B}T\right)\right\}\right] (13) Where A and B are arbitrary constants. One may further extend this analogy to spin-transport phenomena ref-30 ; ref-31 in weakly disordered ferromagnetic systems. If one assumes that, in the weak-disorder limit, magnons can diffuse ref-31 or tunnel ref-32 through the boundaries created by these disorders, and that effective scattering by these disorders is reduced, then one may write the spin current along a particular direction due to magnon excitation using the following simplified expression ref-30 : j\propto\sum_{k}{v_{k}} (14) Where v_{k} is the group velocity of the magnon for a particular mode k , and is the population. Thus, one may follow the same derivation presented earlier and show that, if it can be approximated that, under a discretized energy spectrum, v is independent of , then one may derive a similar expression for the spin current in the presence of weak disorder and in the weak-magnetic-field regime as: j=A^{\prime}T^{0.5}\exp\left\{-B^{\prime}T^{-0.5}\right\} (15) Where A^{\prime} and B^{\prime} are arbitrary constants. Note that this expression, though derived in a very different way, is structurally very similar to the Efros–Shklovskii equation ref-24 , which describes electron conductivity in the presence of disorder. In conclusion, we have adopted a simplified approach and derived a modified Bloch equation to describe the loss of saturation magnetization in a weakly disordered spin-1/2 ferromagnetic system, which may be applied to systems of varying dimensionality. This approach further allows us to obtain an approximate expression for the spin current in such systems. I Acknoledgement The author thanks the Indian Institute of Science, Bangalore, for providing the facilities to conduct this work. Parts of the manuscript was proofread for grammatical accuracy using ChatGPT, an AI-based language tool. II Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request.