---
title: Pure Magnets in Modern Research
url: https://www.emergentmind.com/topics/pure-magnets
type: topic
---

# Pure Magnets in Modern Research

“Pure Magnets” is not a single standardized category in contemporary magnetism. In current research usage, the term can denote at least four distinct but related ideas: a magnet whose field is stored by a bulk superconducting body rather than by a coil or ferromagnetic phase; a charge-insulating spin system in which the relevant degrees of freedom are purely magnetic excitations such as magnons or fractionalized Majorana modes; a homogeneous single-phase magnetic material whose functional response is intrinsic rather than composite-derived; and, in permanent-magnet science, a chemically simple benchmark magnet against which composite or defect-engineered systems are compared. Across these meanings, the unifying theme is that magnetic functionality is generated by the magnetic body or spin sector itself, rather than by auxiliary current-carrying windings, itinerant charge transport, or multiphase compensation architectures [1208.1854], [1902.07716], [2109.03557], [2310.02106].

## 1. Conceptual scope and competing meanings

In superconducting-magnet research, “pure magnets” refers to **bulk superconducting magnets**: solid superconducting bodies that, once magnetized, **retain a trapped magnetic field and function like stand-alone magnets** without continuous energization. In this sense, the “magnet” is neither a wound electromagnet nor a ferromagnetic permanent magnet; the bulk itself stores magnetic field through persistent supercurrents [1208.1854].

In quantum-magnetism and spintronics, the phrase is used differently. Kitaev materials and related Mott insulators are “pure” or insulating magnetic systems in the sense that electronic charge motion is frozen and the low-energy excitations are localized spins, magnons, or fractionalized quasiparticles rather than electrons. In such systems, topological transport and optically driven band engineering act directly on the spin sector [1902.07716], while chiral quantum spin liquids emerge from magnetic interactions, magnetic field, and Dzyaloshinskii-Moriya interaction in a spin-only setting [1910.09874]. A related usage appears in excitonic magnets, where the ordered state is “purely magnetic” in an unconventional way because magnetism comes from a condensate of spin-triplet excitons rather than from conventional ferromagnetic exchange alignment of local moments [2002.02131].

In materials chemistry and functional-alloy design, “pure” can also mean **single-phase** and **intrinsic**. The Mn\(_{1-x}\)Ni\(_x\)CoSi system is explicitly presented as a **pure metallic ZTE material**: a homogeneous single-phase metallic magnet whose zero-thermal-expansion response arises from its own spiral spin reorientation and magnetoelastic coupling rather than from composite compensation [2109.03557].

Permanent-magnet research introduces a further complication. Hard ferrites such as SrFe\(_{12}\)O\(_{19}\) and BaFe\(_{12}\)O\(_{19}\) are chemically simple, rare-earth-free ceramic magnets that function as robust baseline “pure magnets,” yet the highest-performing hard magnets are often not homogeneous at all. In Sm\(_2\)(CoFeCuZr)\(_{17}\)-class materials, coercivity is improved by controlled internal heterogeneity, and the paper on the “perfect defect” argues directly that the best hard-magnet behavior comes from deliberately organized defects and phase boundaries rather than from structural purity [2304.14958]. This suggests that “pure magnet” is often best understood as a contrast class whose meaning depends on whether the emphasis lies on the field source, the low-energy degrees of freedom, the phase purity, or the microstructural architecture.

## 2. Bulk superconducting magnets as stand-alone field sources

The most literal realization of a “pure magnet” is the **MgB\(_2\) bulk superconducting magnet**. The reported system consists of disk-shaped MgB\(_2\) bulks fabricated **using the in-situ technique from Mg and B powders**, then cooled below the superconducting transition and magnetized in an external field. After removal of the external field, **persistent supercurrents remain circulating inside the bulk**, trapping magnetic flux so that the superconducting body itself behaves as a magnet [1208.1854].

The materials argument for MgB\(_2\) is practical as well as physical. The paper highlights **transition temperature \(T_c \sim 40\) K**, operation in the **15–30 K** range with a **cryocooler**, **weak-link-free homogeneous current flow on a bulk scale**, **great flexibility in designing magnet shape**, **low-cost constituent materials** from Mg and B powders, and **light weight**, quoted as **1–2.6 g/cm\(^3\)** [1208.1854]. The bulk showed **homogeneous surface texture in macroscopic scale**, and the authors state that they did not observe any **cracks or domain structures**, features that would otherwise disturb bulk current flow and the trapped-field distribution [1208.1854].

The experimental geometry was central to the trapped-field interpretation. Measurements were made with a **transversal cryogenic Hall sensor**, either at the center of a single-disk surface or sandwiched between two identical disks. The paired-disk configuration was chosen “in order to measure a trapped field comparable to that of inside of the disk,” thereby approximating the interior field of a thicker bulk [1208.1854].

The reported trapped fields establish the operational reality of the bulk-superconducting-magnet concept.

| Configuration | Measurement condition | Trapped field |
|---|---|---|
| 20 mm single disk | 15 K, surface center | 1.8 T |
| 30 mm single disk | 15 K, surface center | 2.25 T |
| 20 mm disk pair | 15 K, between bulks | 2.8 T |
| 30 mm disk pair | 17.5 K, between bulks | above 3 T |

The same paper states that the obtained MgB\(_2\) bulks have **high critical current density of \(>10^5\ \mathrm{A/cm^2}\) at 15–30 K** [1208.1854]. The trapped field increases as temperature decreases, and the measured values increase with larger diameter and with the paired geometry. This is explicitly interpreted in the paper as consistent with the standard trapped-field picture: the field is supported by circulating critical currents, and geometry that better probes the interior yields larger observed values [1208.1854].

In this usage, a pure magnet is not merely a metaphor. The superconductor itself becomes the magnetized object, with no continuous energization and no ferromagnetic permanent-magnet phase. That stands as one of the clearest realizations of the phrase in the supplied literature [1208.1854].

## 3. Pure insulating magnets, Floquet magnons, and chiral spin liquids

A second major meaning of “Pure Magnets” concerns systems in which the relevant dynamics are those of localized spins in charge-insulating magnets. In the periodically driven Kitaev and Kitaev-Heisenberg models, off-resonant light can act directly on the spin sector through the **time-dependent Aharonov-Casher phase**, generating effective magnetic terms without itinerant charge carriers and **without any static magnetic field** [1902.07716]. The starting Hamiltonian is
\[
\mathcal H =2J_K\sum_{\langle ij\rangle_\gamma} S_i^\gamma S_j^\gamma +J_H\sum_{\langle ij\rangle}\vec S_i\cdot \vec S_j ,
\]
with the analysis focusing on the **pure Kitaev model** and the **ferromagnetic phase of the Kitaev-Heisenberg model** on the honeycomb lattice [1902.07716].

In the **off-resonant Floquet regime**, defined by
\[
\hbar\omega \gg A,
\]
the effective zeroth-order Hamiltonian maps onto an anisotropic static spin model plus a **photoinduced magnetic field along \([111]\)** [1902.07716]. The key real-space result is
\[
\mathcal H_{\rm eff}^{(0)} = \sum_{\langle ij\rangle_\gamma} J_\gamma(\mathcal E_0,\phi)\,S_i^\gamma S_j^\gamma + \sum_{\langle ij\rangle} J_{ij}(\mathcal E_0,\phi)\,\vec S_i\cdot\vec S_j + h(\mathcal E_0,\phi)\sum_i \left(S_i^x+S_i^y+S_i^z\right),
\]
with induced field
\[
h(\mathcal E_0,\phi) = (2J_K+3J_H)S \left[ 1-\frac{\mathscr J(\mathcal E_0,\phi)}{3} \right].
\]
This **photoinduced field along \([111]\)** stabilizes magnetic order and yields **Floquet topological magnons**, **chiral magnon edge modes**, and a tunable **thermal Hall effect** [1902.07716].

The topological characterization is explicit. The Chern number of each Floquet magnon band is
\[
\mathcal C_{\rm eff}^\alpha(\mathcal E_0,\phi) = \frac{1}{2\pi}\int_{\rm BZ} d^2k\, \Omega_\alpha^z(\vec k).
\]
For linearly polarized light \((\phi=0)\), the lowest magnon band has \(\mathcal C_{\rm eff}=+1\) in the ferromagnetic Kitaev-Heisenberg model and \(\mathcal C_{\rm eff}=-1\) at the antiferromagnetic Kitaev point for approximately
\[
0<\mathcal E_0\lesssim 1.35.
\]
For circular polarization \((\phi=\pi/2)\), the Chern number is nonzero whenever \(\mathcal E_0\neq 0\), indicating more robust topological behavior [1902.07716]. The paper also gives the magnon thermal Hall response in standard form:
\[
\kappa_{xy} = -k_B^2 T \sum_n \int_{\rm BZ}\frac{d^2k}{(2\pi)^2}\, c_2\!\left[n_B(\epsilon_{n\vec k})\right]\, \Omega_n^z(\vec k).
\]

The same spin-only theme appears in the study of **novel chiral quantum spin liquids in Kitaev magnets**. There, a pure Kitaev spin-exchange system under magnetic field or Dzyaloshinskii-Moriya interaction supports a sequence of topological phases with distinct Chern numbers and Majorana edge structures [1910.09874]. The Hamiltonian is written as
\[
H = H_K + H_{DM} + H_B,
\]
with
\[
H_K = 2\sum_{\langle ij\rangle,\gamma} K^\gamma S_i^\gamma S_j^\gamma.
\]
At weak field along \([111]\), the paper finds a gapped topological quantum spin liquid with
\[
\nu = +1,
\]
and a weak-field gap scaling
\[
\Delta \propto B^3
\]
before crossing over around \(B^*\sim 0.6\) to \(\Delta \propto B\) [1910.09874]. At stronger positive field, a topological transition occurs at
\[
B_c \simeq 1.43 \qquad (K=1,\ D=0,\ t=1),
\]
and the system enters a distinct gapped phase with
\[
\nu = -2
\]
for roughly
\[
B\in [1.43,1.64].
\]
The thermal Hall conductance is given as
\[
\frac{\kappa_{xy}}{T} = \frac{\pi}{12}\frac{k_B^2}{\hbar d}\,\nu,
\]
so the strong-field transition should appear as a jump and sign reversal of \(\kappa_{xy}/T\) [1910.09874].

These studies demonstrate that a “pure magnet” need not be a classical magnetized body at all. It may instead be a charge-insulating spin system whose operative degrees of freedom are magnons or fractionalized Majorana excitations, and whose topology, edge transport, and Hall response arise from the magnetic sector itself [1902.07716], [1910.09874].

## 4. Unconventional pure magnetic order and optically engineered compensated magnets

The term also extends to unconventional ordered states where magnetism is not reducible to ordinary ferromagnetism or antiferromagnetism. In the **doped excitonic magnet**, the ordered phase is a condensate of spin-triplet excitons in a two-orbital Hubbard model, with order parameter
\[
\Phi^\mathrm{t}=|\Phi^\mathrm{t}|e^{i\phi} =\frac{1}{L^2}\sum_{j,\sigma}\sigma \left\langle c_{j,\sigma}^\dagger f_{j,\sigma}\right\rangle .
\]
The paper emphasizes that this state is “purely magnetic” in an unconventional way: the magnetism comes from exciton condensation rather than from ordinary spin-polarized bands or static local moments [2002.02131].

A crucial ingredient is nearest-neighbor interorbital hopping with \(d\)-wave symmetry,
\[
\gamma(\bm k)=2V(\cos k_x-\cos k_y),
\]
combined with slight hole doping away from half filling [2002.02131]. In the excitonic phase with \(0\le \phi<\pi/2\), the system develops an anisotropic spin-dependent Fermi surface and a \(\bm k\)-space spin texture. In linear response, the paper predicts a **pure spin current** when an electric field is applied along a diagonal direction. The charge and spin currents are defined as
\[
\bm j_{\mathrm c}=\bm j_\uparrow+\bm j_\downarrow, \qquad \bm j_{\mathrm s}=\bm j_\uparrow-\bm j_\downarrow.
\]
The central transport result is that in the relevant geometry the transverse charge current cancels,
\[
j_{\mathrm c,\perp}\propto D_\perp(\uparrow)+D_\perp(\downarrow)=0,
\]
while the transverse spin current remains finite,
\[
j_{\mathrm s,\perp}\propto D_\perp(\uparrow)-D_\perp(\downarrow)\neq 0.
\]
The state thereby converts an applied electric field into a transverse pure spin current without spin-orbit coupling [2002.02131].

A different nonclassical extension appears in **Floquet odd-parity collinear magnets**. There, a conventional collinear antiferromagnet under circularly polarized light acquires **spin splitting in momentum space with odd parity** while retaining **zero net magnetization in real space** [2508.02542]. The symmetry criterion is explicit. If \([C_2T||E]\) is broken while \([C_2||P]\) is preserved, then
\[
\varepsilon(s,\mathbf{k})=\varepsilon(-s,-\mathbf{k}),
\]
which is precisely the condition for odd-parity spin splitting [2508.02542].

In the off-resonant high-frequency regime, the Floquet effective Hamiltonian is written as
\[
\begin{split}
H_{\mbox{\tiny eff}}(\mathbf{k})&=H_0(\mathbf{k})+M(\mathbf{k}),\\
M(\mathbf{k})&=\sum_{n\ge 1}\frac{[H_{-n},H_n]}{n\omega}+O\!\left(\frac{1}{\omega^2}\right).
\end{split}
\]
The paper shows that low-symmetry AFM lattices yield **\(p\)-wave** odd-parity collinear magnets, while honeycomb AFMs with threefold symmetry yield **\(f\)-wave** magnets [2508.02542]. In the honeycomb case, the induced term is
\[
H_F=i\delta \sum_{\langle\langle mn\rangle\rangle,\sigma} \nu_{mn} c_{m,\sigma}^{\dagger}c_{n,\sigma},
\]
with
\[
\delta=\sqrt{3}\frac{J_1(\tilde{A})^2}{\omega},
\]
a Haldane-like imaginary next-nearest-neighbor hopping [2508.02542]. The same work finds an antiferromagnetic Chern insulating phase with total Chern number
\[
\mathcal{C}=-2
\]
for high-intensity right-circularly polarized light and \(\mathcal{C}=+2\) for left-circular polarization [2508.02542]. In first-principles calculations for MnPSe\(_3\), the maximal spin splitting reaches
\[
40\ \text{meV},
\]
and the anomalous Hall conductivity can reach roughly
\[
6\frac{e^2}{h}
\]
when the Fermi level is tuned near \(-0.1\ \text{eV}\) [2508.02542].

Together, these works broaden the notion of pure magnetism beyond static moment alignment. A plausible implication is that “purity” in modern magnetic theory often refers to the autonomy of the magnetic sector: correlated particle-hole condensates, compensated collinear order, and Floquet-engineered spin splitting can all generate distinctly magnetic transport and topology without relying on conventional ferromagnetic order [2002.02131], [2508.02542].

## 5. Intrinsic single-phase magnets and rare-earth-free permanent-magnet baselines

A more materials-centered meaning of “pure magnet” appears in systems explicitly described as homogeneous single-phase magnets. The Mn\(_{1-x}\)Ni\(_x\)CoSi alloys,
\[
x = 0,\ 0.010,\ 0.015,\ 0.017,\ 0.020,\ 0.025,
\]
are presented as a **pure metallic ZTE material** in which the functional response is intrinsic to a single metallic magnetic phase [2109.03557]. The compounds are based on orthorhombic MnCoSi, space group **Pnma**, and are framed as **homogeneous single-phase materials** rather than multiphase composites [2109.03557].

The key physics is a temperature-driven coherent reorientation of a spiral magnetic texture. For \(x=0.020\), in situ neutron powder diffraction identifies a low-temperature **cycloidal spiral antiferromagnet** in the **\(bc\) plane**, with incommensurate propagation vector
\[
\mathbf{k} = (0,0,k_c),
\]
which transforms near **190 K** into a **helical spiral** in the **\(ab\) plane**, still propagating along \(c\) [2109.03557]. The neighboring-spin angle in a spiral chain satisfies
\[
\theta_{\text{Mn1-Mn3}} = 180^\circ k_c,
\]
and decreases with temperature. At **300 K** for \(x=0.020\),
\[
\theta_{\text{Mn1-Mn3}} \approx 23^\circ, \qquad \theta_{\text{Mn1-Mn2}} \approx 19^\circ,
\]
signaling progressive approach toward a ferromagnetic-like configuration [2109.03557].

This coherent spin rotation is strongly coupled to the lattice. The paper emphasizes a **large negative thermal expansion along the \(a\) axis**, and defines the effective coefficient as
\[
\alpha = \frac{1}{3}\frac{1}{V}\frac{dV}{dT}.
\]
For \(x=0.020\),
\[
\alpha = 6.9\times10^{-7}\ \text{K}^{-1},\qquad 10\text{–}190\ \text{K},
\]
and for \(x=0.025\),
\[
\alpha = 1.3\times10^{-7}\ \text{K}^{-1},\qquad 10\text{–}170\ \text{K},
\]
values stated to be about an order of magnitude smaller than Fe\(_{65}\)Ni\(_{35}\) Invar [2109.03557]. The mechanism is not simple order–disorder magnetovolume coupling; it is specifically **coherent spiral-spin rotation \(\Rightarrow\) continuous lattice motion \(\Rightarrow\) anisotropic compensation \(\Rightarrow\) ZTE** [2109.03557].

In permanent-magnet science, the simplest robust benchmark “pure magnets” are **hard ferrite ceramics**, especially **M-type hexaferrites** with nominal composition
\[
\mathrm{MFe_{12}O_{19}} \qquad (\mathrm{M = Sr^{2+}\ or\ Ba^{2+}}).
\]
The principal compounds are SrFe\(_{12}\)O\(_{19}\) and BaFe\(_{12}\)O\(_{19}\), which are already the most widely used permanent magnets by mass, accounting for about **85 wt.% of the permanent magnet market** [2310.02106]. They are attractive because they use **No rare-earth elements**, have **High Curie temperature**, \(>700\ \mathrm{K}\), **Low cost**, and **Good chemical stability** [2310.02106].

Their crystal structure is **hexagonal magnetoplumbite**, space group \(P6_3/mmc\), with approximate lattice parameters
\[
a \sim 5.9\ \text{\AA}, \qquad c \sim 23\ \text{\AA},
\]
and easy axis along the crystallographic \(c\)-axis [2310.02106]. The theoretical permanent-magnet ceiling quoted for a hypothetical fully dense and perfectly oriented hexaferrite is
\[
(BH)_{\max} = 45\ \mathrm{kJ/m^3},
\]
while commercial anisotropic ferrites reach
\[
33\text{–}42\ \mathrm{kJ/m^3},
\]
and some doped commercial ferrites reach up to
\[
44\ \mathrm{kJ/m^3}
\]
[2310.02106].

The review repeatedly distinguishes intrinsic from extrinsic limits. Strong uniaxial anisotropy makes M-type ferrites viable hard magnets, but practical performance depends on density, porosity, grain size, texture, and phase purity. The ideal alignment condition is
\[
M_r = M_s,
\]
and loop squareness is monitored via
\[
\frac{M_r}{M_s}.
\]
A central practical difficulty is that the processing needed to improve density and orientation often promotes grain growth, which lowers coercivity [2310.02106]. This makes hard ferrites exemplary “pure magnets” in a chemical sense, but not in the sense of being free from difficult microstructural optimization.

## 6. The limits of purity: composites, rare-earth compounds, and engineered defects

Permanent-magnet research shows most clearly that purity is not always the route to optimum performance. The SrFe\(_{12}\)O\(_{19}\)/Fe composite study is explicitly positioned at the boundary between improving a magnet class that already works industrially—pure strontium ferrite—and attempting to surpass it with a hard-soft composite [2309.10676]. The hard phase is SrFe\(_{12}\)O\(_{19}\), the soft phase is Fe, and the central result is not exchange-spring behavior but a small remanence increase in **exchange-decoupled** systems through **dipolar interactions** [2309.10676].

In anisotropic injection-molded magnets containing **10 vol% Fe**, the benchmark pure SFO bonded magnet has
\[
H_c = 219.6\ \mathrm{kA/m},\qquad J_R = 0.248 \pm 0.0025\ \mathrm{T}.
\]
The corresponding composite magnets with **50 nm Fe** or **1 \(\mu\)m Fe** reach
\[
J_R = 0.255 \pm 0.0025\ \mathrm{T},
\]
a remanence increase of **2.4%** relative to the otherwise identical pure ferrite magnet [2309.10676]. Yet the coercivity drops to
\[
144.8\ \mathrm{kA/m}
\]
for 50 nm Fe and
\[
143.6\ \mathrm{kA/m}
\]
for 1 \(\mu\)m Fe, while the 11 \(\mu\)m Fe composite yields only
\[
J_R = 0.250 \pm 0.0025\ \mathrm{T}
\]
with
\[
H_c = 184.7\ \mathrm{kA/m}
\]
[2309.10676]. The mechanism inferred from experiment and micromagnetics is dipolar rather than exchange: only about **4% of Fe spins** align with the hard phase at remanence in oriented powders [2309.10676]. The study therefore strengthens the case for pure ferrites as the dependable baseline whenever coercivity retention and microstructural simplicity matter more than a modest remanence gain.

At the opposite end of the performance spectrum lie **rare-earth permanent magnet compounds** such as Nd\(_2\)Fe\(_{14}\)B, Sm\(_2\)Fe\(_{17}\)N\(_3\), and NdFe\(_{12}\)N [1801.03455]. These are not pure elements but chemically ordered intermetallics whose large saturation magnetization, strong magnetocrystalline anisotropy, and high Curie temperature originate from the interaction between transition-metal 3d electrons and rare-earth 4f electrons [1801.03455]. The leading anisotropy term is written as
\[
E(\theta) \simeq K_1 \sin^2\theta ,
\]
with rare-earth contribution
\[
K_1 = -3 J (J-1) \alpha_J \langle r^2 \rangle A_2^0 n_R .
\]
The paper’s broader conclusion is that permanent-magnet strength at the electronic-structure level requires cooperative 3d–4f physics, suitable crystal fields or ligand hybridization, and sufficient 3d–4f exchange to preserve anisotropy at elevated temperature [1801.03455]. In this domain, “pure magnet” ceases to mean chemical simplicity and instead points toward intrinsic microscopic design principles.

The strongest challenge to any naïve notion of purity comes from Sm\(_2\)(CoFeCuZr)\(_{17}\)-class pinning magnets. The study on engineering the “perfect defect” argues that coercivity in these high-temperature magnets is governed by **domain-wall pinning** at a controlled hierarchy of phases and interfaces, not by homogeneity [2304.14958]. The material is a **production-grade sintered Sm(Co,Fe,Cu,Zr)\(_{7.7}\)** magnet of nominal composition
\[
\mathrm{Sm(Co_{0.65}Fe_{0.27}Cu_{0.08}Zr_{0.06})_{7.7}}.
\]
Its microstructure contains a **2:17 matrix phase**, a **1:5 cell boundary phase**, and **Z phase** lamellae, with grain sizes about **100–200 µm**, 2:17 cells around **200 nm**, 1:5 walls about **10–12 nm** thick, and Z platelets near **10–12 nm** [2304.14958].

A key result is that high coercivity correlates not with the largest amount of boundary phase, but with more favorable **Cu partitioning** and a distinct **1:5-like cover/wetting layer** at the Z–2:17 interface [2304.14958]. In the high-\(H_c\) region, the 1:5 phase has
\[
\text{Cu } 20.3 \pm 1.8\ \text{at.\%},\qquad \text{Co } 48.4 \pm 1.0\ \text{at.\%},
\]
whereas in the low-\(H_c\) region it has
\[
\text{Cu } 15.3 \pm 1.1\ \text{at.\%},\qquad \text{Co } 52.3 \pm 0.6\ \text{at.\%}
\]
[2304.14958]. Micromagnetic simulations with MuMax3 use the free-energy functional
\[
F = F_{\mathrm{ex}} + F_{\mathrm{ani}} + F_{\mathrm{dm}} + F_{\mathrm{ext}},
\]
and show that the presence of a 1:5 coating on Z platelets generally increases coercivity; increasing Cu in 1:5 from **18.8 to 28.0 at.%** raises the lower-bound coercive field from **0.8 T to 1.6 T** [2304.14958]. The central lesson is explicit: the best permanent magnets are not chemically pure, homogeneous ferromagnets, but materials with **distributed, chemically tuned, atomically structured pinning defects** [2304.14958].

This contrast clarifies the encyclopedia-level significance of the term. In some contexts, a pure magnet is literally a stand-alone superconducting bulk or a single-phase ferrite. In others, the phrase marks spin-only or compensated magnetic functionality. Yet in high-performance permanent magnets, purity in the homogeneous sense is often counterproductive. A plausible implication is that “pure magnet” is most useful as a descriptive category only when paired with the property under discussion: pure field source, pure spin sector, pure phase, or pure reference material.

Source: https://www.emergentmind.com/topics/pure-magnets