# Fileset

[Adv Eng Mater - 2025 - Durgun - Toward Maximum Utilization of Heavy Rare Earths in Sintered Nd Fe B Magnets by Grain.pdf](https://mdr.nims.go.jp/filesets/67344548-129c-4115-808e-c16b1bca1d66/download)

## Creator

Abdullatif Durgun, [Dominik Ohmer](https://orcid.org/0000-0002-7129-331X), Matthias Katter, [Andreas Thul](https://orcid.org/0000-0001-6930-3999), [Simon Steentjes](https://orcid.org/0000-0002-3346-4560), [Hossein Sepehri Amin](https://orcid.org/0000-0002-7856-7897), [Oliver Gutfleisch](https://orcid.org/0000-0001-8021-3839), [Imants Dirba](https://orcid.org/0000-0002-5335-2152)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Toward Maximum Utilization of Heavy Rare Earths in Sintered Nd–Fe–B Magnets by Grain Boundary Diffusion Source and Application Area Optimization](https://mdr.nims.go.jp/datasets/047bfb9a-8038-4d85-880c-55721aa6d289)

## Fulltext

Toward Maximum Utilization of Heavy Rare Earths in Sintered Nd–Fe–B Magnets by Grain Boundary Diffusion Source and Application Area OptimizationToward Maximum Utilization of Heavy Rare Earths inSintered Nd–Fe–B Magnets by Grain Boundary DiffusionSource and Application Area OptimizationAbdullatif Durgun,* Dominik Ohmer, Matthias Katter, Andreas Thul, Simon Steentjes,Hossein Sepehri Amin, Oliver Gutfleisch, and Imants Dirba*1. IntroductionReduction in usage of heavy rare earth (HRE) elements for coer-civity enhancement in NdFeB-based magnets has been in theresearch focus for the permanent magnetcommunity during the last decade.[1–5]First-generation magnets were producedby direct HRE alloying[6,7] resulting in asignificant reduction in remanence andhigh HRE consumption. The second gen-eration utilizes the HREs more effectively,hardening only the outer shells of thegrains via the grain boundary diffusionprocess (GBDP),[8–14] the two-alloy,[15,16]HRE powder addition,[17,18] or dual-main-phase approaches.[19–26] This resultsin preserved remanence and lower HREconsumption. Here, we investigate a thirdconcept–further HRE reduction by strate-gic selective hardening of areas in themagnet that are most susceptible todemagnetization. Using finite elementmagnetostatics simulation on an internalpermanent magnet synchronous tractionmotor we aim to identify the areas in aNd–Fe–B permanent magnet that are mosthighly susceptible to demagnetization.GBDP optimization is carried out to findthe best diffusion source composition andparticle size. Furthermore, we demonstratethat HRE utilization can be maximized by strategic local coercivityenhancement in areas in the magnet, such as corners or edges,that are experiencing the highest demagnetization fields in theactual motor applications.A. Durgun, O. Gutfleisch, I. DirbaFunctional Materials, Institute of Materials ScienceTechnical University of Darmstadt64287 Darmstadt, GermanyE-mail: abdullatif.durgun@tu-darmstadt.de;imants.dirba@tu-darmstadt.deThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/adem.202501145.© 2025 The Author(s). Advanced Engineering Materials published byWiley-VCH GmbH. This is an open access article under the terms ofthe Creative Commons Attribution License, which permits use,distribution and reproduction in any medium, provided the originalwork is properly cited.DOI: 10.1002/adem.202501145D. Ohmer, M. KatterVacuumschmelze GmbH & Co. KG63412 Hanau, GermanyA. Thul, S. SteentjesInstitute of Electrical Machines (IEM)RWTH Aachen University52062 Aachen, GermanyH. S. AminNational Institute for Materials ScienceTsukuba 305-0047, JapanReduction in the utilization of resource critical heavy rare earth (HRE) ele-ments such as Dy and Tb in NdFeB-based magnets is crucial for cost-effectivecoercivity enhancement at the high operating temperature in E-motors.The grain boundary diffusion process (GBDP) is optimized and aim tomaximize HRE utilization by selective magnetic hardening of areas in themagnet such as corners or edges that are highly susceptible to demagneti-zation, as demonstrated by finite element magnetostatics simulation on aninternal permanent magnet synchronous traction motor for electric vehicles.This becomes especially important considering the advent of additivemanufacturing, which has the potential to realize such tailored approacheswith local magnetic hardening based on specific application requirements.Commonly industrially used HRE source TbHx, as well as complex multi-component Tb-containing alloys such as Tb10Pr60(Cu,Al,Ga)30 are investigatedon commercial grade NdFeB-based sintered magnets. Highly efficient Tbutilization with a normalized coercivity increase of 3866 kA/m/wt% Tb(4.86 T/wt% Tb) is achieved with only a minor reduction in remanence fromaround 1.45 T in the initial magnet to 1.43 T after GBDP, paving the waytoward HRE-balanced high-performance magnets for sustainable electricmotor applications.RESEARCH ARTICLEEditor’s Choice www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (1 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbHmailto:abdullatif.durgun@tu-darmstadt.demailto:imants.dirba@tu-darmstadt.dehttps://doi.org/10.1002/adem.202501145http://creativecommons.org/licenses/by/4.0/http://www.aem-journal.comhttp://crossmark.crossref.org/dialog/?doi=10.1002%2Fadem.202501145&domain=pdf&date_stamp=2025-09-262. Results and Discussion2.1. Grain Boundary Diffusion Process Alloy OptimizationFirst task toward maximum utilization of HREs is to identify themost effective HRE source. Initially, we had chosen to comparedifferent materials reported in the literature quantitatively by cal-culating the respective coercivity enhancements per amount ofHRE, i.e., in units of kA/m/wt% HRE (or alternatively T/wt%HRE). In our opinion, this is the best merit since comparing onlythe final coercivities can be misleading due to different startingvalues and the amount of HRE also needs to be implementedbecause Hc will increase with HRE content, as clearly demon-strated later in the manuscript. A large amount of data on theGBDP for rare earth permanent magnets has been summarizedin the review article by Liu et al.[27] which is particularly useful forthis task. Unfortunately, a lot of articles report only the coercivityenhancement without mentioning the actual HRE amount andtherefore could not be used for this comparison. We suggest thatthis should change in the future, since inmagnet manufacturing,the total mass of HRE used is the number that matters instead ofHRE concentration measured locally on a selected area of thesample using an electron microscope. Consequently, we hadto find an alternative way for comparing the various HRE sourcesused and the results are summarized in Figure 1a where thecoercivity enhancements are plotted as a function of HRE or lightrare earth (LRE) wt% in the GBDP source. We have chosen todistinguish between Dy and Tb as HRE sources as well asbetween various Nd–Fe–B magnet manufacturing routes(HDDR, sintering, melt-spun) since they lead to different micro-structures and therefore a direct comparison of the resultantHRE utilization could be misleading. First, on the far-right sidein Figure 1a, simple elements (Dy, Tb), and conventionally usedbinary hydrides/fluorides (e.g., DyH3, TbH3)[10,28–34] followed bybinary low-melting alloys (e.g., Tb70Cu30, Dy70Cu30)[35–37] andLRE-based diffusion sources such as Nd70Cu30, N80Ga15Cu5,or Nd90Al10[16,38–41] are plotted. On the left side, results fromcomplex quaternary and quinary HRE-containing alloys suchas Tb–Pr–Cu–Al–Ga[37,42–45] are shown. Two main conclusionsare drawn from surveying the literature: 1) the highest absolutecoercivity enhancements are reported for diffusion of pure met-als or binary hydrides/fluorides; 2) the lowest HRE concentra-tions are achieved by using complex quaternary and quinaryGBDP alloys. Therefore, to identify the most effective and sus-tainable diffusion sources that minimize the use of critical rareearth elements Dy and Tb, the coercivity enhancement needs tobe normalized with respect to the HRE amount. The results inFigure 1b show that the complex quaternary and quinarycompositions clearly stand out with the highest coercivityenhancement observed for Tb10Pr60(Cu,Al,Ga)30.[42] Therefore,these were chosen for the GBDP studies conducted in thepresent work.GBDP source alloy optimization was carried out to find the bestcomposition and particle size. GBDP with application of thesource alloy on top and bottom planes perpendicular to the easyaxis of a sintered magnet sample with a right parallelepiped shapewas systematically investigated. A source alloy ingot directly afterinduction melting was crushed into a powder and sieved to differ-ent particle size fractions (0–63; 80–163; 163–250 μm). Resultswere compared with ribbons obtained after a subsequent meltspinning. Figure 2a shows that coercivity increases consistentlywith powder particle size, reaching the highest value in the caseof direct application of melt-spun ribbons without additionalgrinding. The reason for this behavior is the high affinity ofRE-based alloys toward oxygen, which leads tomore oxidation withincreasing surface area (smaller particle size) even though powde-rizing was done inside an Ar-filled glovebox. However, to investi-gate this effect in more detail, oxygen pickup across the differentprocessing steps would need to be quantified. Accordingly, melt-spun ribbons were chosen for the further experiments in thiswork. Next, compositional optimization was done for Tb- andDy-based alloys with the resultant hysteresis loops shown inFigure 2b. As expected, Tb-based alloys outperform Dy-based onesdue to the much higher anisotropy field (17.5MAm�1 vs11.9MAm�1) of the Tb2Fe14B phase compared to Dy2Fe14B.[46]The best magnetic properties were achieved for Tb10Pr60(Cu,Al,Ga)30 (Tb13.65Pr72.59Cu5.46Al2.32Ga5.99 in wt%), most likely dueto the higher Pr content compared to higher Cu, Al, Ga in theFigure 1. a) Selected literature values for coercivity enhancement by HRE/LRE diffusion in various types of NdFeB magnets. b) The correspondingcoercivity enhancement normalized to the RE content showing the efficiency of the different sources.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (2 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comTb10Pr40(Cu,Al,Ga)50 composition. The exact chemical composi-tion measured by inductively coupled plasma optical emissionspectroscopy (ICP-OES) is Tb13.81Pr72.40Cu5.57Al2.52Ga5.70 (in wt%)as given in Table S2, Supporting Information. The differenceis even more pronounced when the quinary GBDP alloys arecompared to conventionally used TbHx and DyHx. The weightof all applied materials was calculated to ensure similar HREusage. Since Tb10Pr60(Cu,Al,Ga)30 contains 13.65% Tb, whichequals 0.4 wt%, this amount was used in the case of the binaryhydrides. TbHx and DyHx result in coercivity enhancementΔHcof 568 and 348 kAm�1, respectively compared to 701 kAm�1for Tb10Pr60(Cu,Al,Ga)30 using the same HRE content.Therefore, this composition was chosen for further studiesin this work.To investigate the melting behavior during the GBDP process,differential thermal analysis (DTA) measurements were done upto 900 °C. DTA heating curve for Tb10Pr60(Cu,Al,Ga)30 alloy inFigure 3a reveals multiple endothermic melting events withmain peaks at 468 and 485 °C which is close to the secondGBDP annealing temperature of 500 °C. Scanning electronmicroscope (SEM) backscattered electron (BSE) image inFigure 3b shows large grains separated by fine two-phase eutecticintercellular mixture for the as-cast alloy. Chemical compositions(atomic %) determined by energy dispersive x-ray spectroscopy(EDX) are Tb8.1Pr56.8Cu2.4Al18.6Ga14.0 for the gray matrix phaseand Tb4.7Pr39.8Cu9Al3.1Ga1.1O42.4 for the fine dark regions.Oxygen in the Pr-rich phase originates from SEM sample prepa-ration since conventional polishing wheel in ambient atmo-sphere was used for surface preparation. X-ray diffractionmeasurement was conducted on the melt-spun ribbons as anattempt to identify the phases and match with the available phasediagrams. The diffractogram is shown in Figure S1, SupportingFigure 2. Optimization of GBDP source alloy composition and application form. a) the effect of particle size and b) the effect of composition. 3.0 wt% ofGBDP alloy (melt-spun ribbons) and 0.4 wt% HRE hydrides (powder) were used for equal HRE usage.Figure 3. a) DTA heating curve for Tb10Pr60(Cu,Al,Ga)30 composition showing multiple endothermic melting events. b) SEM BSE image of the eutecticmicrostructure for the as-cast alloy.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (3 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comInformation. Due to the high quenching rate of the alloy duringmelt spinning, no clear crystalline diffraction peaks could be dis-tinguished and therefore phase identification was not possible.Considering the importance of the grain boundary magnetism,melting point and wettability in coercivity development, theseresults illustrate the potential in search for optimum multicom-ponent alloys at/near eutectic compositions with suitable prop-erties using computational methods, since for simple binaryand ternary alloys equilibrium phase diagrams can be foundin handbooks which is not the case for higher order (e.g., qui-nary) constituent systems.2.2. Grain Boundary Diffusion Process Optimal HREApplication AmountAs next, a study was conducted to investigate the optimalHRE application amount. 0.3–3.0 wt% of the magnet massTb10Pr60(Cu,Al,Ga)30 melt-spun ribbons corresponding to0.041–0.408 wt% Tb, respectively were applied on top andbottom c-planes of the samples. As expected, coercivity increaseswith the applied HRE amount from 1010 kAm�1 in theinitial magnet to 1710 kAm�1 after 3.0 wt% GBDP alloy(0.408 wt% Tb) diffusion treatment, as shown in Figure 4a.Figure 4. Demagnetization curves a) and summary of the resultant remanence and coercivity values, b) after GBDP with 0.3–3.0 wt% Tb10Pr60(Cu,Al,Ga)30 melt-spun ribbons corresponding to 0.041–0.408 wt% Tb applied on top and bottom c-planes as illustrated in the inset. The shaded area in (a)indicates the statistical deviation in remanence of the initial commercial sintered magnet samples. The dashed line represents blank initial magnetexposed to the same heat treatment as a reference.Figure 5. A 2D FE-Simulated flux density distribution for a typical internal permanent magnet synchronous traction motor from electric vehicles in theshort-circuit operating point. a) overview of the geometry and b) magnified section showing that the level of demagnetization is not distributed homo-genously with the highest demagnetization occurring at the edges of the inner magnets.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (4 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comIt has to be noted that GBDP efficiency depends strongly on theinitial sintered magnet. A different commercial grade magnetwith a grain size of 5 μm (in contrast to 8.3 μm grain size forthe magnet used in the rest of this work) shows larger coercivityenhancement, from up 1100 to 1990 kAm�1 for the same HREamount. Remanence remains nearly unchanged, considering thespread in the initial commercial magnets properties, which isillustrated by the shaded range. The obtained Jr and Hc valuesare plotted as a function of GBDP alloy and Tb amount inFigure 4b and listed in the Table S1, Supporting Information.Minimal Jr reduction is observed considering the error bar, onaverage, from around 1.43 to 1.41 T. The most important obser-vation is that at first (0.3 wt%), significant Hc enhancement isachieved with minimal HRE usage. This results in a remarkablecoercivity enhancement of 3866 kA/m/wt% Tb (4.86 T/wt% Tb)for 0.6 wt% GBDP alloy, which is far superior to the convention-ally used TbHx and demonstrates the potential for using multi-component grain boundary diffusion sources in production ofsustainable high-performance magnets. The slope decreasesfor higher amounts and most likely would reach a plato at thefar-right side. This implies that after reaching a certain threshold,the magnetic hardening and further HRE utilization become lessefficient. A possible explanation is that the exchange length forNd2Fe14B is only few a nm and thus decoupling of adjacentgrains can be achieved with little amount of grain boundary dif-fusion. The subsequently introduced HRE diffuses into thematrix grains, forming a thick (Nd,Tb)2Fe14B shell (see SEMimage in Figure 7), which therefore leads to comparably low fur-ther improvement in coercivity.In addition, an attempt was made to enhance coercivitywithout using HRE by replacing Tb with Pr, resultingin Pr0.7(CuAlGa)0.3 GBDP alloy composition. In this case, a227 kAm�1 Hc enhancement was obtained.2.3. Local Strategic Magnetic HardeningThe results in the foregoing section described HRE diffusionsource optimization when applied to the top and bottom c-planesof the magnets. As next, a strategy to improve the HRE utilizationby selective application targeting specific parts of the magnet thatare most susceptible to demagnetization is presented. Finiteelement (FE) simulation was used for modeling the spatialdistribution of magnetization and demagnetization field config-uration. Detailed information about the model can be found inthe supplementary information. Permanent magnet synchro-nous motors used for traction drives typically have to cover alarge rotor speed operating range. The rotating magnetic fluxgenerated by the permanent magnets is linked to the statorand therefore induces a voltage in the stator winding, whichis proportional to the rotor speed. Due to the large speed range,the induced voltage is usually higher than the maximum voltageprovided by the supplying inverter. As a result, the stator currentshave to be controlled in such a way that the stator windings pro-duce a magnetic field opposed to the rotor magnetization above acertain rotor speed level. This operation region is called the fieldweakening region. Aside from the intended field weakening tolimit the induced voltage, field weakening can also occur unin-tentionally due to transient or short-circuit stator currents. As anexample, to illustrate the demagnetization caused by field weak-ening, a typical internal permanent magnet synchronous tractionFigure 6. Sketch illustrating HRE application at different locations of the magnet: a) middle of c-planes, b) uniform c-planes, c) edges, and d) corners. Thecorresponding demagnetization curves after GBDP of 0.3 wt% and 0.6 wt% of Tb10Pr60(Cu,Al,Ga)30 are shown in e). The shaded area in (e) indicates thestatistical deviation in coercivity (�50 kAm�1) after preparation and characterization of multiple samples.Figure 7. Sketch illustrating coercivity mapping after HRE application onc-planes corners. a) 3D view shows cutting in six cubes for individual localM(H) measurements. b) Color map represents the resultant coercivityenhancement (ΔHc) mapping constructed from individual measurementsillustrated in (a).www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (5 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.commotor for electric vehicles (EVs) is chosen. In this design,magnets are embedded in the soft magnetic rotor core. Eachmagnetic pole of the rotor consists of three magnets: a pair oflarger magnets forming a V-shape and an additional smallermagnet closer to the rotor surface. A typical worst-casedemagnetizing scenario is the so called ideal short-circuit oper-ating point, where the rotor flux linked to the stator is zero. A 2DFE-simulation of this operating point is shown in Figure 5a.Here, the flux lines and the flux density amplitude are shown.In order to visualize the flux density distribution inside the mag-nets, the color value range has to be limited. Thus, the maximumflux density in the soft magnetic parts, which is not of interesthere, is not depictable. It can be seen that due to the opposingfield generated by the stator windings, there is almost no mag-netic flux linkage between rotor and stator, i.e., the induced volt-age is zero.A detailed view of the flux density inside the magnets is shownin Figure 5b. Due to stray flux inside the rotor core, the magnetsthemselves are not completely demagnetized. The level ofdemagnetization is not distributed homogenously: the highestdemagnetization occurs on the edges of the lower magnets,where the flux density is reduced to �0.8 T. Yet in large innerareas of these magnets, the flux density is only reduced to�1 T. Further, the smaller magnets closer to the rotor surfaceare less demagnetized compared to the two larger magnets.The position dependency is similar to the larger magnets, butFigure 8. SEM BSE images and EDX elemental composition line scan for a GBDP sample with 0.6 wt% of Tb10Pr60(Cu,Al,Ga)30 applied on corners.a) Overview showing the residual GBDP alloy on the surface, b) an area near surface with pronounced core-shell structure, c) an area in the middle part ofthe magnet away from the diffusion source showing no core-shell structure, and d) EDX elemental line scan across the Tb-rich shell as indicated in (b).Table 1. Coercivity mapping for 0.6 wt% Tb10Pr60(Cu,Al,Ga)30 and TbHxHRE sources applied on c-plane corners as shown in Figure 6.Position Tb10Pr60(Cu,Al,Ga)30Hc [kA m�1]ΔHc [kA m�1] TbHxHc [kA m�1]ΔHc [kA m�1]1 1598 587 1365 3542 1521 510 1136 1253 1592 581 1316 3054 1503 492 1149 1385 1441 430 1079 686 1486 475 1120 109Average 1513 502 1210 199Initial 1011 – 1011 –www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (6 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comless pronounced regarding the amplitudes. Therefore, hardeningthe areas in the magnet where the highest demagnetizationoccurs[47,48] is a viable strategy for saving HRE[49] and can be usedfor the production of magnets with predefined properties in agiven region.[50]For this reason, we have studied different GBDP source appli-cation areas as shown in Figure 6b–d. 0.3 wt% and 0.6 wt%Tb10Pr60(Cu,Al,Ga)30 melt-spun ribbons were applied a) onthe middle of top and bottom c-planes, b) uniformly coveringc-planes, c) edges, and d) corners. The resultant demagnetizationcurves in Figure 6e do not suggest significant differences for0.3 wt%, all the samples show comparable coercivity increaseafter GBDP treatment. Due to the low-melting point, the diffu-sion alloy becomes liquid during the heat treatment and diffusesover the entire application area. In the case of 0.6 wt%, slightlybetter performance is observed for uniform c-planes and middleGBDP application. However, this could also partially result fromincreased ribbons layer thickness due to the limited area availableon corners and edges, as well as spreading over the edges ontothe side surfaces.To shed light on the HRE diffusion and local magnetic hard-ening, spatial coercivity mapping was done for 0.6 wt%Tb10Pr60(Cu,Al,Ga)30 and TbHx HRE sources applied on corners(Figure 6d). 3.1� 10� 10mm magnet after GBDP treatmentwas sliced perpendicular to c-axis and the top layer with a thick-ness of about 3.1mm was cut into smaller rectangular regionsas shown in Figure 7. Each obtained piece was measuredseparately to investigate the local coercivity enhancement. Theresults are summarized in Table 1 and graphically illustratedin Figure 7. Corners (1 and 3) clearly show the highest coercivityreaching nearly 1600 kA/m corresponding to 587 kAm�1enhancement compared to the initial magnet. As expected fora diffusion-governed process, coercivity decrease is proportionalto the distance from the diffusion source (corners), resulting in aslight decrease for sample 2 and a significant decrease for sample5. Coercivity enhancement of the entire sample is 513 kAm�1.As expected, the results are far worse for TbHx diffusion, whereΔHc reaches only 354 kAm�1 at the corners, drops more thantwice, to 125 kAm�1 in between, and 5 times, to 68 kAm�1, fur-ther away for sample number 5. These results show that forTbHx, not only is the HRE utilization less efficient, but alsothe diffusion is more sluggish, resulting in shorter diffusiondepth for conventional HRE sources.[9] TbHx does not melt dur-ing the heat treatment, therefore diffusion is too slow, and it isnot reaching the center of the magnet.SEM BSE images and EDX elemental composition line scanfor a GBDP sample with 0.6 wt% of Tb10Pr60(Cu,Al,Ga)30 appliedon corners are shown in Figure 8. a) Low-magnification overviewshows thin residual GBDP alloy on the sample surface with athickness under 20 μm. b) Area near the surface, right belowthe residual diffusion source layer, is represented by a pro-nounced core-shell structure . The thickness of the Tb-rich shellreaches several μm close to the GBDP alloy application anddecreases gradually with distance inside the magnet. c) Areain the middle part of the magnet, away from the diffusion sourceshows no core-shell structure and looks similar to an untreatedmagnet. d) EDX elemental line scan across the Tb-rich shellreveals a reduction in Nd and an increase in Tb. In addition,a slight enhancement of Pr and Al is visible.Scanning transmission electron microscopy-energy-dispersiveX-ray spectroscopy (STEM-EDS) elemental maps for differentelements near the magnet surface (area in Figure 8b) whereGBDP alloy was applied are shown in Figure 9. RE oxides(Nd, Pr, Tb)x can be identified along with metallic Nd, andNd–Cu and Nd–Ga-rich phases at the triple junctions. We haveused the STEM-EDS line profile to evaluate the rare-earth contentFigure 9. STEM-EDS elemental maps for a) Nd, Tb, Pr and b) Fe, Al, Cu, Ga, O. Elemental concentrations line profiles across the indicated arrow arepresented in c).www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (7 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comin the shell region since it mainly defines the local magneticanisotropy field at the interfaces. Three regions can be distin-guished from the elemental concentrations across the line scanindicated by the arrow and are presented in Figure 9c. The triplejunction phase (I) is rich in Nd, Pr, Ga, and Cu. In the grainboundary region (II), Nd content is only about 3 at.% with4 at.% Pr and 2 at.% Tb. Subsequently, in region (III), normalNd2Fe14B matrix concentrations are reached. Grain boundariesare enriched mainly in Tb and Pr whereas the diffused GBDPalloy elemental concentrations in the matrix region (III) arenearly zero. This explains well from the magnetic propertiesobserved high HRE utilization efficiency since Tb is not diffusingin the main phase to a significant degree but remains at the grainboundaries and shell region of 2:14:1 grains. Ga and Al arereported to infiltrate along the grain boundary phase, improvingwetting and promoting Tb diffusion.[51,52]Since traction motors in EVs operate at elevated temperatures,we have measured the temperature dependance of coercivity upto 180 °C to compare the various GBDP sources in Figure 10. Therespective temperature coefficient β values are given in parenthe-ses. As expected from the room temperatureM(H) results shownin Figure 2b, the best performance is observed for 3.0 wt%Tb10Pr60(Cu,Al,Ga)30 alloy corresponding to 0.4 wt% Tb (totalamount applied on both sides). At 120 °C a coercivity of771 kAm�1 is maintained, which, considering recent emphasison lowering the operating temperatures of traction motors, canalready be sufficient. For example, operating temperatures below70 °C have been achieved in additively manufactured air-cooledelectrical machine rotor for an automotive application.[53]Moreover, GBDP efficiency depends strongly on the initialsintered magnet. A different commercial grade magnet with agrain size of 5 μm shows larger coercivity enhancementfor the same HRE amount, resulting in an impressive1014 kAm�1 coercivity at 120 °C and 733 kAm�1 at 150 °C withusing only 0.4 wt% Tb.It has to be noted that completely eliminating HREs is animportant strategy; however, this case demonstrates that withan acceptable HRE amount, which is comparable to trace ele-ments added to commercial sintered magnets (e.g., Al, Cu,Ga), good high temperature performance can be reached. Dyand Tb amount optimization needs to be viewed keeping the rareearth (in)balance problem[54] in mind. This is illustrated inFigure 11 showing rare earth elements production in tons/yearas well as percentages in relation to Nd. Being very abundant,LREs Ce and La need to be significantly overproduced, whichmotivates the search for additional applications, such as moder-ate performance magnets. In contrast, only a fraction of HREs Tband Dy are produced owing to their low abundance, only 1.82%in the case of Tb. This work has shown that balanced HRE con-tents comparable to those actually produced can be sufficient forthe production of high-performance Nd–Fe–B magnets.Figure 11. Production of various rare earth elements in quantity (tons/year) and relative to Nd (%).Figure 10. Temperature stability of coercivity for different GBDP sourcesapplied on c-planes. The respective β values are given in the parentheses.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (8 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.com3. ConclusionThe effects of adding a commonly industrially used GBDP HREsource, Tb hydride, as well as complex multicomponent alloys asdiffusion source, on the coercivity increase of commercial gradeNdFeB-based sintered magnets were investigated. Coercivityincreases with powder particle size of the alloy diffusion source,reaching the highest value in the case of direct application ofmillimeter-size melt-spun ribbons on the magnet surface.This results in a remarkable coercivity enhancement of3866 kA/m/wt% Tb (4.86 T/wt% Tb) for 0.6 wt% Tb10Pr60(Cu,Al,Ga)30 alloy which is far superior to the conventionally usedTbH3 and demonstrates the potential for using multicomponentgrain boundary diffusion sources in production of sustainablehigh-performance magnets. For Tb hydride, not only is theHRE utilization less efficient, but also the diffusion is more slug-gish resulting in shorter diffusion depth. In addition, an attemptwas made to maximize HRE utilization by selective hardening ofareas in the magnet that are susceptible to demagnetization, suchas corners or edges. Using FE magnetostatics simulation on aninternal permanent magnet synchronous traction motor for EVs,we could show that the level of demagnetization is not distrib-uted homogenously, with the highest demagnetization occurringat the edges and corners of the inner magnets. It would be inter-esting to validate the simulation results against experimental datafor local demagnetization field measurements.The work demonstrates that with a small HRE amount com-parable to trace elements added to commercial sintered magnets(e.g., Al, Cu, Ga), good high temperature performance can bereached. Diffusing 0.4 wt% Tb results in coercivity of1014 kAm�1 at 120 °C, showing that balanced HRE contentscomparable to those actually produced can be sufficient andtherefore complete elimination of HREs may not be strictlynecessary.4. Experimental SectionGBDP HRE Source Preparation: Quinary Tb or Dy containing alloyswere prepared by induction melting and melt spinning (20m s�1 wheelspeed, 1 mm orifice, 1 mm distance from the wheel) starting from pureelements (99.99%). Prior to application, thickness of the melt-spun rib-bons was measured using a micrometer ensuring reproducible and pre-cise control over GBDP HRE source content on magnet surfaces. Theinduction melting was conducted under a high-purity argon atmosphere(99.999 %) after evacuating the chamber thrice to a base pressure below5·10�2 mbar and purging with argon. For melt spinning, the chamber wasevacuated to a base pressure below 5·10�5 mbar followed by filling withargon (99.999%). For the particle size study, the as-cast ingots wereground to powders using an agate mortar and pestle, followed by sievingto separate the respective particle size fractions (0–63; 80–163;163–250 μm). Tb and Dy hydrides were prepared by exposing pure ele-ments to hydrogen (100 bar, room temperature, 24 h) and subsequentjet milling (Picojet, Hosokawa Alpine). The powders were jet-milled undera high-purity argon atmosphere (99.999%) to prevent oxidation. Millingwas performed with a gas pressure of 8 bar in three cycles, each separatedby a 2-minute interval.The classifier wheel speed was set up to 30 000 rpm. The particle size ofthe milled powders was examined by SEM and confirmed to be below5 μm.GBDP Experiments: Commercial-grade sintered magnets provided byVACUUMSCHMELZE GmbH & Co. KG. were used for GBDP experiments.A mixture of polyvinyl butyral (PVB) and ethanol was used as a binder tocover the targeted parts of the magnets with the HRE source. The PVBbinder was first dissolved in ethanol and stirred for 30 min. A few dropsof this solution were then applied onto the cleaned magnet surfaces.Subsequently, melt-spun ribbons were placed on top. The samples wereleft to dry in air at ambient temperature for 30min. Mass of the initialmagnet and mass of the HRE source were measured to calculate theGBDP efficiency. The samples were loaded into a quartz tube and heattreatments at 900 °C for 8 h followed by 500 °C for 4 h were done undervacuum (10�5 mbar) for diffusion. After the heat treatment, the sampleswere quenched in water under vacuum and then allowed to cool to roomtemperature.Characterization: M(H) measurements were carried out at a pulsed-field magnetometer from HyMPulse (Metis Instruments) in open circuitconditions. The M(H) measurements were performed at 21 °C using amaximum applied magnetic field of 6.85 T. Demagnetization correctionswere done according to the shape and dimensions of the samples. Foreach selected area of the magnets, more than 10 samples were preparedandmeasured. The reported coercivity values correspond to the average ofthese measurements with an error of �50 kAm�1. Microstructural char-acterization was performed using a Tescan VEGA3-SBH SEM equippedwith an Octane Plus EDS detector for elemental analysis. The compositionof the samples was quantified using ICP-OES with an iCAP PRO XP Duofrom Thermo Scientific, employing a standard calibration curve method.For the calibration curve, four standard solutions in the concentrationrange between 50 and 200 ppm for Pr, 26 and 110 ppm for Tb, 1 and10 ppm for Dy, Co, Nb, B, and Al, and 6 and 30 ppm for Cu and Ga,as well as one blank solution, were prepared. The Fe concentrationwas calculated by subtracting the sum of the ppm of all measured ele-ments from the mass of the sample. Both samples were measured twiceusing 50mg of powder digested in 5mL of 32% HCl and 1mL of 65%HNO3 at 210 °C for 15min using the microwave digestion unitEthos.lab from MLS-MWS Laboratory Solutions. For the subsequentICP-OES analysis, the solution was diluted by a factor of 4 with 3%HNO3. DTA for melting temperature measurements was done usingDSC 404 F1 Pegasus (Netzsch). STEM was conducted using FEI Titan80-200 with a probe aberration corrector. TEM specimens were preparedby the lift-out method using a focused ion beam system Helios G4-UXDualBeam (FEI).Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsThis work is financially supported by the Deutsche Forschungsgemeinschaft(DFG, German Research Foundation), Project ID No. 405553726, TRR 270and by the German Ministry of Education and Research in the framework ofthe project “Scale2PM” (Skalierung der 2-Pulvermethode zur Herstellungvon Permanentmagneten mit reduziertem Gehalt kritischer Elemente),grant number 03VP10552. A. Durgun gratefully acknowledges the financialsupport of the Study Abroad Postgraduate Education Scholarship (YLSY)awarded by the Republic of Türkiye Ministry of National Education. Theauthors thank Dr. Franziska Scheibel for conducting ICP-OES measure-ments as well as careful data analysis and fruitful discussions.Open Access funding enabled and organized by Projekt DEAL.Conflict of InterestThe authors declare no conflict of interest.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (9 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.aem-journal.comData Availability StatementThe data that support the findings of this study are available from thecorresponding author upon reasonable request.Keywordscritical materials, diffusion, dysprosium, magnets, NdFeB, rare earthmetals, terbiumReceived: April 28, 2025Revised: September 1, 2025Published online:[1] H. Sepehri-Amin, S. Hirosawa, K. Hono, Handbook Of MagneticMaterials, Elsevier 2018, 269, https://doi.org/10.1016/bs.hmm.2018.08.003.[2] O. Gutfleisch, M. A. Willard, E. Brück, C. H. Chen, S. G. Sankar,J. P. Liu, Adv. Mater 2011, 23, 821.[3] S. Hirosawa, M. Nishino, S. Miyashita, Adv. Nat. Sci: Nanosci.Nanotechnol. 2017, 8, 013002.[4] K. Hono, H. Sepehri-Amin, Scr. Mater. 2018, 151, 6.[5] K. P. Skokov, O. Gutfleisch, Scr. Mater. 2018, 154, 289.[6] M. Sagawa, S. Fujimura, H. Yamamoto, Y. Matsuura, S. Hirosawa,J. Appl. Phys. 1985, 57, 4094.[7] M. Sagawa, S. Hirosawa, K. Tokuhara, H. Yamamoto, S. Fujimura,Y. Tsubokawa, R. Shimizu, J. Appl. Phys. 1987, 61, 3559.[8] H. Nakamura, K. Hirota, M. Shimao, T. Minowa, M. Honshima, IEEETrans. Magn. 2005, 41, 3844.[9] T. Helbig, A. Abel, S. Sawatzki, T. Gröb, I. Dirba, S. Ener,O. Gutfleisch, J. Alloys Compd. 2024, 992, 174490.[10] K. Loewe, D. Benke, C. Kübel, T. Lienig, K. P. Skokov, O. Gutfleisch,Acta Mater. 2017, 124, 421.[11] H. Sepehri-Amin, T. Ohkubo, K. Hono, Acta Mater. 2013, 61, 1982.[12] F. Li, J. Li, S. Ur Rehman, L. Zhang, Y. Hu, M. Yang, X. Yu, S. Zhong,T. Liang, J. Alloys Compd. 2022, 899, 163270.[13] S. Lee, G. Kim, K.-S. Lee, S. Kim, T.-H. Kim, S. Lee, D.-H. Kim, W. Lee,J.-G. Lee, Acta Mater. 2025, 285, 120660.[14] L. Zhao, J. He, W. Li, X. Liu, J. Zhang, L. Wen, Z. Zhang, J. Hu,J. Zhang, X. Liao, K. Xu, W. Fan, W. Song, H. Yu, X. Zhong,Z. Liu, X. Zhang, Adv. Funct. Mater. 2022, 32, 2109529.[15] M. Honshima, K. Ohashi, JMEP 1994, 3, 218.[16] I. Dirba, P. Pattur, I. Soldatov, E. Adabifiroozjaei, L. Molina-Luna,O. Gutfleisch, J. Alloys Compd. 2023, 930, 167411.[17] S.-E. Park, T.-H. Kim, S.-R. Lee, S. Namkung, T.-S. Jang, J. Appl. Phys.2012, 111, 07A707.[18] J. Zuo, D. Zhang, M. Zhu, Z. Deng, X. Song, X. Wu, Z. Li, J. AlloysCompd. 2024, 1005, 176047.[19] Y. Zhao, R. Han, X. Duan, A. Li, H. Chen, S. Dong, H. Feng, M. Zhu,G. C. Hadjipanayis, W. Li, J. Alloys Compd. 2024, 1002, 174588.[20] M. Zhu, R. Han, W. Li, S. Huang, D. Zheng, L. Song, X. Shi, IEEETrans. Magn. 2015, 51, 1.[21] W. Fu, S. Guo, C. Lin, R. Chen, X. Liu, D. Lee, A. Yan, IEEE Trans.Magn. 2013, 49, 3258.[22] K. Opelt, T. Ahmad, O. Diehl, M. Schönfeldt, E. Brouwer, I. Vogel,J. D. Rossa, J. Gassmann, S. Ener, O. Gutfleisch, Adv. Eng. Mater.2021, 23, 2100459.[23] T. Helbig, K. Loewe, S. Sawatzki, M. Yi, B.-X. Xu, O. Gutfleisch, ActaMater. 2017, 127, 498.[24] K. Opelt, C.-C. Lin, M. Schönfeldt, J. Gassmann, S. Yoon,O. Gutfleisch, Acta Materialia 2024, 270, 119871.[25] Z. Jia, Y.-H. Li, X.-T. Yang, S. Cao, G.-F. Ding, S. Guo, X.-D. Fan,Y.-H. Xie, Z.-W. Xiong, R.-J. Chen, A.-R. Yan, Rare Met. 2025, 44, 1267.[26] C. Lin, K. Opelt, J. D. Rossa, F. Maccari, S. Yoon, J. Gassmann,O. Gutfleisch, Adv. Eng. Mater. 2025, 27, 2500054.[27] Z. Liu, J. He, R. V. Ramanujan, Review, Mater. Design 2021, 209,110004.[28] W.-Q. Liu, C. Chang, M. Yue, J.-S. Yang, D.-T. Zhang, J.-X. Zhang,Y.-Q. Liu, Rare Met. 2017, 36, 718.[29] W. Ji, W. Liu, M. Yue, D. Zhang, J. Zhang, Physica B: CondensedMatter. 2015, 476, 147.[30] N. Watanabe, M. Itakura, N. Kuwano, D. Li, S. Suzuki, K. Machida,Mater. Trans. 2007, 48, 915.[31] M. Soderžnik, K. Rožman, S. Kobe, P. McGuiness, Intermetallics 2012,23, 158.[32] S. Sawatzki, I. Dirba, L. Schultz, O. Gutfleisch, J. Appl. Phys. 2013, 114,133902.[33] S. Sawatzki, I. Dirba, H. Wendrock, L. Schultz, O. Gutfleisch, J. Magn.Magn. Mater. 2014, 358, 163.[34] T.-H. Kim, T. T. Sasaki, T. Ohkubo, Y. Takada, A. Kato, Y. Kaneko,K. Hono, Acta Mater. 2019, 172, 139.[35] N. Oono, M. Sagawa, R. Kasada, H. Matsui, A. Kimura, J. Magn.Magn. Mater. 2011, 323, 297.[36] M. Tang, X. Bao, K. Lu, L. Sun, J. Li, X. Gao, Scr. Mater. 2016, 117, 60.[37] K. Lu, X. Bao, M. Tang, G. Chen, X. Mu, J. Li, X. Gao, Scr. Mater. 2017,138, 83.[38] H. Sepehri-Amin, I. Dirba, X. Tang, T. Ohkubo, T. Schrefl,O. Gutfleisch, K. Hono, Acta Mater. 2019, 175, 276.[39] L. Jin, Z.-H. Jin, J.-H. Zhu, G.-F. Ding, B. Zheng, S. Guo, R.-J. Chen,A.-R. Yan, X.-C. Liu, Chin. Phys. B 2021, 30, 027503.[40] L. Liu, H. Sepehri-Amin, T. Ohkubo, M. Yano, A. Kato, T. Shoji,K. Hono, J. Alloys Compd. 2016, 666, 432.[41] Z. H. Kautsar, H. Sepehri-Amin, X. Tang, N. Kulesh, T. T. Sasaki,T. Ohkubo, K. Hono, J. Alloys Compd. 2025, 1010, 177738.[42] T. Wu, S. Cao, M. Kou, Y. Xie, G. Ding, S. Guo, B. Zheng, R. Chen,M. Zhong, A. Yan, J. Alloys Compd. 2023, 934, 167888.[43] G. Chen, X. Bao, K. Lu, X. Lv, Y. Ding, M. Zhang, C. Wang, X. Gao,J. Magn. Magn. Mater. 2019, 477, 17.[44] J. Pan, S. Cao, Y. Li, Q. Li, G. Ding, B. Zheng, S. Guo, X. Fan, R. Chen,A. Yan, J. Alloys Compd. 2024, 984, 174003.[45] T. Zhou, J. Chen, Q. Wang, W. Pan, Q. Huang, R. Liu, M. Li, G. Xie,J. Alloys Compd. 2023, 937, 168368.[46] S. Hirosawa, Y. Matsuura, H. Yamamoto, S. Fujimura, M. Sagawa,H. Yamauchi, J. Appl. Phys. 1986, 59, 873.[47] J. M. D. Coey, Engineering 2020, 6, 119.[48] W. Li, J. Cheng, X. Lu, Y. Li, D. Zhao, W. Pei, R. Pei, AIP Adv. 2024, 14,035233.[49] W. Zhang, Z. Wang, Y. Luo, X. Sun, X. Bai, H. Peng, W. Yan, D. Yu,J. Magn. Magn. Mater. 2024, 592, 171768.[50] C. Huber, C. Abert, F. Bruckner, M. Groenefeld, S. Schuschnigg,I. Teliban, C. Vogler, G. Wautischer, R. Windl, D. Suess, Sci. Rep2017, 7, 9419.[51] R. Song, H. Feng, H. Zhao, F. Liu, J. Magn. Magn. Mater. 2023, 575,170744.[52] L. Jin, G. Ding, J. Zhu, Z. Jin, B. Zheng, S. Guo, R. Chen, A. Yan, X. Liu,J. Alloys Compd. 2021, 870, 159375.[53] M. Bieber, M. Haase, F. Tasche, A. Zibart, B. Ponick, in: 2023 IEEEInternational Electric Machines & Drives Conference (IEMDC), 2023.[54] K. Binnemans, P. T. Jones, T. Müller, L. Yurramendi, J. Sustain. Metall.2018, 4, 126.www.advancedsciencenews.com www.aem-journal.comAdv. Eng. Mater. 2025, 2501145 2501145 (10 of 10) © 2025 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adem.202501145 by National Institute For, Wiley Online Library on [28/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttps://doi.org/10.1016/bs.hmm.2018.08.003https://doi.org/10.1016/bs.hmm.2018.08.003http://www.advancedsciencenews.comhttp://www.aem-journal.com Toward Maximum Utilization of Heavy Rare Earths in Sintered Nd-Fe-B Magnets by Grain Boundary Diffusion Source and Application Area Optimization 1. Introduction 2. Results and Discussion 2.1. Grain Boundary Diffusion Process Alloy Optimization 2.2. Grain Boundary Diffusion Process Optimal HRE Application Amount 2.3. Local Strategic Magnetic Hardening 3. Conclusion 4. Experimental Section Supporting Information Acknowledgements