Why the magnet decides the motor
Most electric vehicles, many industrial drives and a large share of wind turbines run on permanent-magnet machines. The magnets in the rotor supply the field the motor works against, so their strength sets how much torque a given volume of motor can deliver, and their stability sets how hot the motor can run before it loses that strength. A designer can trade copper, steel and cooling against each other, but the magnet sets the ceiling.
Soft magnetic materials matter almost as much. The stator and rotor cores are stacks of thin electrical steel that carry the field and reverse it many times a second. Every reversal costs energy as heat, so the core alloy decides a large part of a motor’s losses. A program that only looks at the rotor is looking at half the machine.
MS Lattice treats both in one domain: hard magnets, which keep their field, and soft magnets, which carry it. The domain’s line says what it is for: more field, fewer rare earths.


Rare earths, heavy rare earths and supply risk
The strongest permanent magnets made are neodymium–iron–boron. Their performance comes from neodymium and, for grades that must work hot, from small additions of the heavy rare earths dysprosium and terbium, which hold coercivity up as the temperature rises. Rare-earth mining, and above all separation and magnet manufacture, are concentrated in very few places. Prices have swung sharply, and export controls are a standing concern for motor makers.
Engineers have answered in three broad ways, and each maps to material groups MS Lattice searches.
- Use less of the scarce elements: reduced-rare-earth NdFeB, with cerium and lanthanum doing more of the work, measured against high-performance NdFeB.
- Leave rare earths out: rare-earth-free MnAl, and hexaferrites, the lowest-cost magnets, where a modest energy product will do.
- Get more from the rest of the machine: iron–silicon steels, iron–cobalt with the highest saturation there is, and spinel ferrites for high-frequency power.
Cobalt is the second supply concern. It appears in some magnet grades and is the base of iron–cobalt alloys. A brief can leave it out with the Cobalt-free starting point, or cap how much a candidate may carry.
Temperature is the real specification
A magnet’s strength on a data sheet is usually quoted at room temperature. A traction motor’s rotor runs far hotter, and the property that suffers first is coercivity, the magnet’s resistance to being demagnetized. Under heavy load at high temperature, the opposing field from the stator can push a weak magnet past the point where it recovers, and the motor loses torque for good.
That is why MS Lattice’s targets for the domain include coercivity at operating temperature, not coercivity in general, and why its first starting point is Automotive, 150 °C: strength inside a running motor. The Curie temperature, where a magnet loses its magnetic order altogether, is a separate target. It sets an absolute ceiling, and a material whose Curie temperature sits close to the motor’s working range loses strength quickly as it warms.
So write temperature into the brief from the start. A candidate that looks excellent at room temperature and falls apart at 150 °C can cost a full synthesis campaign to discover. Held inside the search as a requirement, that limit removes it before your lab sees it.
A magnet is only as strong as it is at the hottest point of the motor’s duty cycle. Put that temperature in the brief.
What a magnet campaign aims for
The magnets domain lets a campaign aim for five things. You choose which a candidate must meet and which it should push as far as it can.
- Energy product: the stored magnetic energy that sets how much field a given volume of magnet supplies.
- Coercivity at operating temperature: resistance to demagnetization where the motor runs.
- Curie temperature: the ceiling above which magnetic order is lost.
- Saturation, for soft magnets: how much flux a core can carry.
- Rare-earth and cobalt content: the supply exposure each candidate carries.
Requirements are held inside the search, so a proposal that breaks a limit on heavy rare earths is turned away before a structure is built. Preferences are pushed as far as they go, never at a requirement’s expense. Three starting points cover common programs: Automotive, 150 °C; Cobalt-free, strength without the supply risk; and Maximum performance, the strongest the domain allows.
Open Discovery works from the target alone. Prototype-Guided Optimization starts from a material you already use, and the modes page gives the magnet case as its example: today’s automotive magnet, searched for variants that hold their strength at motor temperature.
Example briefs a motor team might write
A good brief names the application, the numbers that must hold and the elements you will not depend on. Four examples, each a separate campaign:
- A traction magnet with less heavy rare earthStart from Automotive, 150 °C. Require coercivity at 150 °C, cap heavy rare-earth content and push energy product as high as it goes. Reduced-rare-earth and high-performance NdFeB are ranked on one scale, so the cost of each substitution is visible.
- A rare-earth-free magnet for an auxiliary driveExclude rare earths and cobalt, and accept a lower energy product for a supply chain with nothing scarce in it. MnAl and hexaferrite candidates come back ranked.
- A core material for high-frequency powerAim the soft-magnet side at spinel ferrites, with saturation as the target and cost in view. Core loss is measured on made cores, so the shortlist starts that work.
- A high-saturation core for a power-dense machinePush saturation as far as it goes among iron–cobalt and iron–silicon alloys, and set the cobalt limit your program can carry.
Cooling with a magnetic field
The magnetocaloric effect is related physics put to another use. Some materials warm when a magnetic field is applied and cool when it is removed. Cycle a solid refrigerant through a field and carry the heat away with a fluid, and you have a refrigerator or heat pump with no refrigerant gas. The material problem looks a lot like the magnet problem: a strong effect, at the right temperature, without depending on scarce elements.
MS Lattice’s magnetocaloric domain searches five groups. Gadolinium solid solutions are the reference refrigerant. Gd₅(Si,Ge)₄ gives the giant effect of a first-order transition. La(Fe,Si)₁₃ gives a large effect while being mostly iron. (Mn,Fe)₂(P,Si) is first order with no rare earth. Ni–Mn–X Heuslers show the inverse effect of a magnetostructural transition.
The central trade-off is hysteresis. First-order materials give the largest effect but tend to lag between heating and cooling, which wastes work and limits how fast a device can cycle. A campaign can target the entropy change in a field, a transition near room temperature, low hysteresis, relative cooling power, adiabatic temperature change, and cost and supply risk. Its starting points include Room-temperature refrigerants, with the transition held between 250 and 350 K, and Low-hysteresis second-order refrigerants, for fast cycling.
The Heuslers link to one more domain: Ni–Mn–Ga martensites move in a magnetic field, and the shape-memory-alloys domain searches them as magnetic actuators.


What you receive, and what must still be tested
A magnets campaign hands back ranked hard- and soft-magnet candidates, from NdFeB to ferrites and iron alloys, with estimated energy product, coercivity at temperature, Curie temperature, saturation, and rare-earth and cobalt content. Each carries its uncertainty, the checks it passed and a standard structure file. A magnetocaloric campaign hands back ranked candidates with estimated transition temperature and order, entropy change at 2 T, adiabatic temperature change, hysteresis, cooling power and cost.
Four kinds of work stand between a shortlist and a motor:
- Microstructure: grain size, grain-boundary phases and alignment decide how much intrinsic coercivity a sintered or bonded magnet keeps.
- Processing: powder making, sintering and heat treatment change the result, and some phases, such as the magnetic one in MnAl, are metastable and must be kept from decomposing.
- Corrosion and coating: NdFeB corrodes without protection, and the coating has to survive the rotor’s environment.
- Motor performance: demagnetization under load at temperature, core losses and behavior over the duty cycle are measured in a built machine.
MS Lattice narrows a very large composition space to a shortlist worth making, and your lab finds out which candidates survive a real process. The magnets domain carries no proof of its own on the record, so treat its rankings as screening. The record does show the engine’s checks holding up in general: its first-principles stability check, which matters when a phase may be metastable, and its record of every candidate.
- 4 of 4stability verdicts right
- 73 of 73candidates rebuilt from their record
If you would rather not run the lab side yourself, Norck, powered by MS Lattice, offers discovery as a service in the scope you choose: Discovery; Discovery + Validation, with candidates synthesized and physically tested; Discovery + Validation + Production, a validated material, produced; or End to end, to your final, manufactured parts.
Questions teams ask
Can MS Lattice find a magnet stronger than NdFeB?
It ranks every candidate against your targets, NdFeB grades included, on one scale. Only making and measuring a new compound shows whether it beats NdFeB in a finished magnet.
Can we start from the magnet we use today?
Yes. Prototype-Guided Optimization searches around your structure and ranks every variant against the original.
Does the domain cover electrical steels and ferrites?
Yes. Iron–silicon, iron–cobalt and spinel ferrites sit on the soft side, hexaferrites on the hard side, alongside NdFeB and MnAl.
Are magnetocaloric materials part of the magnets domain?
No. They have a domain of their own, with targets for entropy change, transition temperature, hysteresis and cooling power.
What does a campaign cost?
Every plan shows its price before it launches. On the record, the blind benchmark ran 64 candidates for under 20 cents, about a third of a cent per candidate.


MatterSpace’s first domain engine
MatterSpace built MS Lattice for the scientists and engineers who discover and design new materials, starting from the properties their product needs.









