Gadolinium's big moment doesn't make a permanent magnet: a literature check

We looked at gadolinium because of its moment. At 7.55 μB per atom, elemental Gd has a saturation polarization of about 2.66 T. That is higher than iron (2.15 T) and higher than Fe₆₅Co₃₅ (2.45 T), the top of the Slater–Pauling curve. On paper, that is more magnetization than any magnet in commercial use.

It doesn't carry over to a permanent magnet. There are three independent reasons, and any one of them is enough. This post collects the sources and the measured numbers so the next person doesn't have to redo the search.

1. The moment disappears at room temperature

Gd orders ferromagnetically at 293 K. The 2.66 T figure is a 0 K value from single crystals (Nigh et al. 1963). Thin films reproduce it: 2.61 ± 0.26 T at 4 K, with the same 293 K Curie point (Scheunert et al. 2012). At motor operating temperatures, around 400 K, Gd is paramagnetic. Its coercivity in those films was 160 Oe, so it is also soft.

2. In Fe and Co compounds, Gd subtracts

To raise the ordering temperature you alloy Gd with Fe or Co. But heavy rare earths couple their spins antiparallel to the transition metal. Gd has no orbital moment, so its whole 7 μB spin moment opposes the Fe or Co moment. The result is a ferrimagnet.

• GdCo₅: net ≈ 5 × 1.8 − 7.3 ≈ 1.7 μB/f.u. at low temperature. YCo₅ has the same Co framework with no opposing rare-earth moment. The Gd–Co exchange field is about 235 T. Forcing the moments to cant takes 44–48 T at 1.4 K (Patrick et al. 2018), so no practical field will line them up.

• GdCo₅₋ₓNiₓ: the net moment cancels completely (compensation) for 1 ≤ x ≤ 3 (Tedstone et al. 2019).

• Gd₂Fe₁₄B: it has the highest Curie temperature in the R₂Fe₁₄B family (~661 K), but lower Ms and anisotropy field than Nd₂Fe₁₄B. Nd is a light rare earth and couples parallel to Fe. The best melt-spun Gd₂Fe₁₄B reached Jr = 0.51 T, Hc = 187 kA/m, and (BH)max = 33 kJ/m³ (J. Rare Earths 2023). Sintered NdFeB is about ten times that.

So adding Gd to an Fe or Co magnet lowers its magnetization.

3. Gd contributes almost no anisotropy

Gd³⁺ is 4f⁷, an S-state ion with L = 0. Crystal-field anisotropy needs an orbital moment, so Gd adds essentially none of the coercivity that Nd, Sm, Tb, and Dy provide. Elemental Gd's weak anisotropy is mostly dipolar, and its easy axis rotates below 240 K. In Gd–Co compounds, the anisotropy that exists comes from the Co sublattice. Room-temperature K₁ is 4.1 MJ/m³ for GdCo₅ and −0.27 MJ/m³ (easy plane) for Gd₂Co₁₇ (JMMM 1981).

The closest published version of the idea

Autieri et al., Sci. Rep. 2016 set out to beat FeCo by forcing Gd parallel to Fe. Thin Cr, Mn, or Sc spacer layers between Gd and Fe flip the interlayer coupling. It worked, but only below about 250 K, and only the few Gd layers nearest the interface coupled. It's a thin-film trick with a cryogenic ceiling, not a route to a bulk magnet.

Where Gd is genuinely useful

• Temperature stability. The Gd sublattice loses magnetization faster with temperature than Fe does, which partly offsets Fe's own decline. Melt-spun Gd₂Fe₁₄B has a remanence coefficient of only −0.066 %/K and a positive coercivity coefficient (+0.171 %/K). Small Gd additions to NdFeB trade remanence for flatter temperature curves. This is the most promising Gd direction for this team if we want high-temperature NdFeB.

• Cheap magnets without Nd. (Ce,Gd)-Fe-B magnets use Gd to repair Ce₂Fe₁₄B's low Curie temperature (J. Alloys Compd. 2018).

• Cryogenic and magnetocaloric uses, where the 293 K Curie point is an advantage.

Gd compounds can't be submitted to the RE-Free Permanent Magnet Leaderboard

Measured values

Every number above is in this dataset, with its source, temperature, and sample form:

Structures

Unmodified Materials Project GGA cells:

Don't trust Materials Project's convex hull for Gd–Co. It puts Gd₂Co₁₇ 0.72 eV/atom and GdCo₂ 1.63 eV/atom above the hull. Both are well-known experimental phases, so this is a GGA failure on the Gd 4f shell, not real instability (search run). Gd–Fe looks sane by comparison (search run).

What our simulation tools can and can't do here

• Prophet gets Gd wrong in both directions. It makes elemental Gd antiferromagnetic and couples Gd parallel to Co in GdCo₅, which inflates the magnetization about tenfold (benchmark post

Before any Gd simulation on Ouro means anything, run elemental Gd and GdCo₅ as anchors and check them against the dataset above.

If we want to confirm it in the lab

Anyone with an arc melter and a VSM (vibrating-sample magnetometer) can make the comparison in a day:

1. Arc-melt GdCo₅ and YCo₅ from the elements under argon. Flip and remelt 3–4 times, with about 2 wt% excess rare earth to make up for evaporation. Tedstone et al. found a 950 °C / 10-day anneal didn't improve phase purity, so as-cast is fine. Check with XRD for 2:7 and 2:17 impurities.

2. Measure M(H) to at least 7 T at 300 K and 10 K.

3. Expected: GdCo₅ well below YCo₅ at both temperatures, with the gap largest at 10 K where the Gd sublattice is fully ordered. If GdCo₅ comes out above YCo₅, something is wrong with the sample.

It's a cheap control, and it also gives us a measured anchor for testing whether any future Gd-capable model has the coupling sign right.

Bottom line

Gd's moment is real, but it only exists below room temperature. Every practical way to raise the Curie temperature pairs Gd with Fe or Co and flips its moment against them, and Gd provides none of the anisotropy a hard magnet needs. Gd is worth considering as a minor additive for thermal stability in NdFeB or Ce-Fe-B. It isn't a route to more magnetization. For raw magnetization, the right starting point is still the Fe–Co end of the Slater–Pauling curve with added anisotropy.