Nuclear Instruments and Methods in Physics Research B 334 (2014) 43–47

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research B

journal homepage: www.elsevier .com/locate /n imb

Characterization of magnetic degradation mechanismin a high-neutron-flux environment

http://dx.doi.org/10.1016/j.nimb.2014.05.0020168-583X/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +1 614 247 8701; fax: +1 614 292 3163.E-mail address: Cao.152@osu.edu (L. Cao).

Adib Samin a, Jie Qiu a, Jason Hattrick-Simpers b, Liyang Dai-Hattrick b, Yuan F. Zheng c, Lei Cao a,⇑a Nuclear Engineering Program, Department of Mechanical and Aerospace, The Ohio State University, Columbus, OH 43210, USAb Department of Chemical Engineering, University of South Carolina, Columbia, SC 29208, USAc Department of Electrical and Computer Engineering, The Ohio State University, Columbus, OH 43210, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 17 February 2014Received in revised form 18 April 2014Accepted 11 May 2014

Keywords:Nd-Fe-B permanent magnetRadiation damageNeutronDegradation of magnets

Radiation-induced demagnetization of permanent magnets can result in the failure of magnet-baseddevices operating in high-radiation environments. To understand the mechanism underlying demagne-tization, Nd-Fe-B magnets were irradiated with fast and fast plus thermal neutrons at fluences of 1012,1013, 1014, and 1015 n/cm2, respectively. After irradiation, magnetic flux losses were shown to increasewith the fluence. Compared with samples irradiated only with fast neutrons, the samples exposed tothe fast plus thermal neutrons have higher magnetic flux losses, which is attributed to the thermal neu-tron capture reaction of boron. Hysteresis loops of the Nd-Fe-B magnets reveal a slightly increase in thecoercivity after irradiation. Full remagnetization of the samples after irradiation was possible, whichindicates that structural damage is unlikely an important factor in the demagnetization process at theselevels of neutron flux and fluence. Finally, we performed a preliminary Molecular Dynamic (MD)simulation on a cube of ions to obtain a better understanding of the thermal spike mechanism.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

During the past decade, there has been renewed interest in thephenomenon of radiation-induced demagnetization of permanentmagnets [1–3]. This interest has been driven by the need for betterradiation-resistant magnets in a variety of applications. The high-power ion propulsion engines used by NASA [4], cyclotrons [5],synchrotrons [6], and the high-radiation environments of both theproposed rare isotope accelerator (RIA) and neutrino factory (NF)are among the applications that require radiation-resistantmagnets [7,8]. In addition, the accident at the Fukushima DaiichiNuclear Power Plant, Japan, in 2011 prompted investigations intothe ability of robots to conduct sampling, recovery, and rescue mis-sions in intense-radiation environments [7]. The magnets that areused as components in the DC motors of these robots must alsobe radiation resistant; otherwise, robot joints and wheels will notmove even when the battery still provides the power. Therefore,to successfully design radiation-hardened magnets that can with-stand high levels of radiation without heavy shielding, it is essentialto develop a fundamental understanding of the underlying physicsbehind the radiation-induced demagnetization mechanism.

The type of radiation, total dose, and dose rate are importantfactors when considering demagnetization mechanisms. Previousresearch has indicated that degradation of the magnetic propertiesof Nd-Fe-B magnets irradiated by gamma rays is small [9–11].Therefore, the effect of neutron irradiation on Nd-Fe-B magnetsis expected to play a bigger role in the process of radiation-induceddemagnetization. Since the effects of neutron irradiation on Nd-Fe-B magnets are not well understood, the objective of the presentwork is to focus on the demagnetization of Ne-Fe-B magnets usingirradiation of fast and full-spectrum (fast plus thermal) neutronsfrom a nuclear reactor. The radiation dose rates at various sitesinside a reactor facility where robots are likely to enter are esti-mated from the Chernobyl reactor site data [12] and from a reviewof past criticality accidents (an uncontrolled nuclear chain reac-tion), from which the neutron fluence near the core or nearspent-fuel sites is conservatively estimated to be in the range of1014 to 1016 n/cm2 [13]. A previous study applied fast neutron flu-ence of 4 � 1012 n/cm2 to Nd-Fe-B magnets [14]. In this work, thefluence levels from 1012 to 1015 n/cm2 are reached and their effectsare discussed, which corresponds to an absorbed dose estimated atless than 100 Mrad. Although it is known that Sm–Co based mag-nets have better thermal stability and radiation resistance thanNd-Fe-B magnets, they are brittle, mechanically unreliable, andalso have lower energy products with higher material costs, which

44 A. Samin et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 43–47

is the reason that Nd-Fe-B magnet, instead of Sm–Co magnet, infocused in this paper.

2. Experimental

2.1. Sample

Commercially available Nd-Fe-B magnet samples manufacturedby Apex Magnets Company (M18 � 116DI) were used in the pres-ent work. The physical properties of this Nd-Fe-B magnet are listedin Table 1.

Each cylindrical sample has a diameter and thickness of 6.35and 1.6 mm, respectively. The direction of magnetization was par-allel to the cylinder axis.

2.2. Irradiation with neutrons

The irradiation experiments were conducted in the ‘‘rabbit’’ sys-tem at The Ohio State University Research Reactor (OSURR). The‘‘rabbit’’ system is a pneumatic facility designed to rapidly movesamples in and out of close proximity to the reactor core. Boththe fast neutrons (with an average energy of 1.98 MeV and a mostprobable energy of 0.73 MeV, which are conveniently referred as1-MeV neutrons) and fast plus thermal neutrons (with an energyless than 0.4 eV and a peak at 0.0253 eV) were utilized. A cadmiumbox (1 mm thickness) was used to filter out the thermal neutroncomponent. The temperature during irradiation was approxi-mately 80 �C. Table 2 shows the neutron fluences and therespective neutron flux, i.e., dose rate, achieved for each sample.

2.3. Measurement system

Before and after neutron irradiation, the open-circuit magneticflux was measured at room temperature using a commerciallyavailable Helmholtz coil and an integrating fluxmeter (LakeShoreModel 480). Observed flux values for irradiated samples were com-pared with the original values to determine relative flux losses.Hysteresis loop measurements were performed for each samplebefore and after irradiation using Physical Property MeasurementSystem with the Vibrating Sample Magnetometer option (PPMS–VSM). The magnetic domain structure was investigated using mag-netic force microscopy (MFM). MFM images were obtained usingthe tapping/lift mode of the Veeco Dimension 3000 Scanning ProbeMicroscope (SPM). The MFM probe used in this study was coatedwith a Co–Cr alloy.

3. Results

Fig. 1 shows the neutron-fluence dependence of the magneticflux loss for the Nd-Fe-B magnet irradiated by fast and fast plusthermal neutrons. The data were plotted against neutron fluenceon a logarithmic scale. Notice that the magnetic flux decreasedwith increasing neutron fluence, regardless of neutron energy. At1012 n/cm2 neutron fluence, the difference between demagnetiza-tion due to fast and fast plus thermal neutron is within statisticalinsignificance, indicating the thermal neutron contribution is neg-ligible at this level of fluence, which is in agreement with reference[2]. Irradiation by fast plus thermal neutrons resulted in a greatermagnetic flux loss compared to the fast-neutron irradiation at 1013

and 1014 n/cm2 neutron fluences, respectively, the explanation of

Table 1Physical properties of the Nd-Fe-B sample.

Grade name Material Residual magnetization, Br (T) Intrinsic coercive

N48 Nd2Fe14B 1.38 876

which is attributed to the boron consumption by thermal neutroncapture and more to the subsequent energetic particle (7Li and4He) production, which exacerbate the magnetic loss due to fastneutron alone. While the detailed discussion is in the followingsession, it is noted that the highest fast neutron fluence and theequivalent thermal neutron fluence in our study is about 6.25times and 272 times higher than that in reference [2], where aradioisotope Cf-252 was the source for fast neutrons and a polyeth-ylene was used as moderator between source and sample toprovide thermal-only neutrons.

Other factors such as different dose rate and different irradiationtemperature (room temperature in [2] versus 80 �C in our study)may explain the accelerated demagnetization with our higherneutron fluence applied. For example, it took 3 days toreach 3.3 � 1012 n/cm2 fluence whereas in our case it is 40 s for1012 n/cm2 neutron fluence. At 1015 n/cm2 fluence, the lossmechanism due to thermal neutron seems to be saturated.

The hysteresis loops for the periods before and after neutronirradiation are shown in Figs. 2 and 3, respectively. There is a clearindication from the enlarged details in Figs. 2 and 3 showing thatall 7 samples presented a larger coercivities, albeit minor, thanthe control sample (except for the one received 1013 fast neutronfluence). The difference in the change of coercivity among irradi-ated samples of the same type are not statistically significant,which could be due to the experimental uncertainty. It is alsounclear from Figs. 2 and 3 whether the increase in coercivity ismore pronounced when irradiating samples with fast plus thermalneutrons instead of only fast neutrons. Based on the trends of thedata in Figs. 3 and 1, the known effect of thermal neutrons fromthe literature, the sample receiving 1013 n/cm2 fast neutron fluencein Fig. 3 cast doubt on its sample preparation and measurementprocess. Nevertheless this data point is still preserved in the figureinstead of discarded as an outlier.

As previously mentioned, only small changes were found in thehysteresis curves before and after neutron irradiation, whichindicates that the neutron irradiation did not cause permanentmetallurgical structure changes. The evolution of magnetic micro-structure changes was investigated using MFM. Fig. 4(a) and (b)show the MFM images of magnet samples before and after1015 n/cm2 fast-neutron irradiation, respectively. There is a notice-able difference in the domain structure pattern. The domain pat-tern in the un-irradiated sample is corrugation and spikes, whichare the typical domain structures for sintered Nd-Fe-B magnets.These complex domain structures are formed owing to a balancebetween the magnetostatic energy and exchange energy [14,15].Although the corrugation and spike domains still existed afterthe neutron irradiation, plate domains appeared, as shown inFig. 4(b). A review of data from the corrugation and spike domainsindicates that their easy axes were close to or in the normal of themagnet surface, while the easy axes of the plate domains were par-allel or close to the magnet surface [16]. Specifically, the platedomain structure was formed by the nucleation and growth ofthe reversal domain [17]. This explanation accounts for the lossof magnetic flux, which reached nearly 80% after a 1015 n/cm2 neu-tron irradiation.

4. Discussion

When interacting with nuclei, fast neutrons typically undergoinelastic scattering: (n, n0)-type reactions. The atom that

force, Hcj (kA/m) Coercive force, Hcb (kA/m) Curie temperature, Tc (�C)

860 320

Table 2Neutron fluences and neutron flux for each sample.

Sample Fast neutron fluence (n/cm2)/flux (n/cm2-s) Thermal neutron fluence (n/cm2)/ /flux (n/cm2-s) Total neutron fluence (n/cm2)

1 2.0 � 1011/5.0 � 109 8.0 � 1011/2.0 � 1010 1.0 � 1012

2 2.0 � 1012/1.0 � 1010 8.0 � 1012/4.0 � 1010 1.0 � 1013

3 2.0 � 1013/1.0 � 1011 8.0 � 1013/4.0 � 1011 1.0 � 1014

4 2.0 � 1014/4.5 � 1011 8.0 � 1014/1.8 � 1012 1.0 � 1015

5 1.0 � 1012/5.1 � 109 1.0 � 1012

6 1.0 � 1013/1.0 � 1011 1.0 � 1013

7 1.0 � 1014/4.6 � 1011 1.0 � 1014

8 1.0 � 1015/4.6 � 1011 1.0 � 1015

Fig. 1. Effect of neutron irradiation on magnetic flux of the Nd-Fe-B magnet; thesolid lines show the general trend of the magnetic flux loss.

Fig. 2. Hysteresis loop for Nd-Fe-B samples before and after irradiation byfast + thermal neutrons with different neutron fluences.

Fig. 3. Hysteresis loop for the Nd-Fe-B samples before and after irradiation by fastneutrons with different neutron fluences.

A. Samin et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 43–47 45

experiences the collision becomes the primary knock-on atom(PKA) that initiates a cascade collision and enhances the thermalvibrations of the atoms in the material [18]. Cost [19] has roughlyestimated the primary knock-on events for Nd-Fe-B magnets irra-diated with a neutron fluence of approximately 1016 n/cm2. Theresults of their study indicates that the average number of dis-placements per atom is too low to modify the metallurgical struc-ture of the Nd-Fe-B magnet. This finding implies that themetallurgical structure of this magnet is not significantly damagedby neutron irradiation under the current level of neutron fluence.

Moreover, these results suggest that thermal spikes [2,19] aregenerated in the material at the point of collision, while the heatis diffused quickly via electrons (compared with phonons).Consequently, the temperature exceeds the Curie temperature inthe volume of the sample near the PKA site, and thermal agitationdisrupts the alignment of the magnetic moments in the affecteddomains. This disruption leads to a ferromagnetic-to-paramagneticphase transition in the affected volume. However, after the heathas dissipated, another phase transition occurs in the affected vol-ume, and the domains again become ferromagnetic. Therefore, theelectrons will be strongly correlated within a domain. Then thenucleation of reverse domains occurs to reduce the magnetostaticenergy, which induced the magnetic flux loss.

Our findings indicate that the magnetic flux decayed faster forthe fast plus thermal neutrons (Fig. 1). We believe this phenome-non results from the thermal neutron capture reaction. Boron hastwo stable isotopes: 10B and 11B. 10B has a large thermal capturecross section (3842 barns). The neutron reaction of 10B inducedby thermal neutrons is described by

10Bþ nth ! 11B! 7Liþ 4Heþ 2:8MeV

The demagnetization is unlikely due solely to the consumptionof boron by such reaction because, even after 1015 n/cm2, this reac-tion only creates the loss of B at 0.76 ppm. However, energeticcharged particles (mainly 840 keV 7Li and 1472 keV 4He) are cre-ated from this reaction, the stopping of which will create cascadedefects in Nd2Fe14B, which explains the observed greater flux losstrend for the fast plus thermal neutrons, when compared with thatof fast neutrons. These nuclei deposit their energies primarily byinteracting with electrons with a range about a few micrometersin the magnet, which ultimately supports the thermal spikemodel [20].

Fig. 4. Magnetic force image of the Nd-Fe-B sample (a) before and (b) after 1015-n/cm2 fast-neutron irradiation; the surface is perpendicular to the easy axis.

46 A. Samin et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 43–47

The observation of coercivity increase after irradiation may beexplained by domain wall pinning due to aforementioned defects.Such conclusion is consistent with previous studies [1,2,14], whichconcluded that neutron radiation damage may pin domain walls.

5. Simulation

To obtain a better understanding of the thermal spike mecha-nism, a preliminary MD simulation was performed on a cube ofiron, which was utilized to approximate a complicated magneticstructure. First, a Monte Carlo simulation was used to determinethe most probable PKA energy. Then, a SRIM simulation was per-formed to help determine an appropriate cell size. In this analysis,we assumed that the collisions were elastic. It was determined thatthe simulation box should have an edge length greater than 120 Å.In the MD simulation, we used a simulation box consisting of50 � 50 � 50 bcc unit cells (250,000 atoms) and the cascade wasinitiated at the center of the box. The PKA was given an initialdirection of {1 3 5} to avoid channeling effects. Furthermore thesimulation was performed using the Embedded Atom Model(EAM) potential for iron along with the Two Temperature modelwhich takes electronic effects into account [21]. The simulationbox was divided into 9 � 9 � 9 cells and each cell included about343 atoms. A plot of the atomic temperature for the central cellis shown in Fig. 5. It indicates a sharp rise in temperature and thena swift decay to equilibrium and reaches a temperature of 484 Kwithin 400 fs. This supports the idea of thermal spikes and givessome insight into the time scales involved in the process.

Fig. 5. The atomic temperature for the central 343 iron atoms of a 50 � 50 � 50 bccunit cell simulation box plotted against time to show the effect of neutronirradiation and the thermal spikes.

The magnetism of rare-earth metals originates from the mag-netic moment of the incompletely filled 4f shell, whose electronsare highly localized. Additionally, the LS coupling is sufficientlystrong in rare-earth atoms. Thus, the total angular momentumJ = L + S is a good quantum number for the 4f shell. Hence, theinteraction between the 4f electrons in rare-earth metals origi-nates from the Ruderman-Kittel-Kasuya-Yosida (RKKY) interac-tion, which is mediated by the mobile conduction electrons. Thistype of indirect interaction is developed within the framework ofsecond-order perturbation theory and leads to an effective Heisen-berg Hamiltonian [22].

6. Conclusion

Radiation-resistant magnets are needed for a wide variety ofapplications such as particle accelerators and in actuators ofrobots, collecting samples or conducting rescue missions after anuclear accident. To understand the mechanism governing radia-tion-induced demagnetization, Nd-Fe-B magnets were irradiatedwith fast neutrons and fast plus thermal neutrons at fluences of1012, 1013, 1014, and 1015 n/cm2, and resulting magnetic fluxeswere observed. The experiments indicated that the flux lossincreased for both cases as the neutron fluence increased, the causeof which is proposed as due to thermal spikes rather than structuredamage. Moreover, compared with fast-neutron-irradiated sam-ples, the samples irradiated with fast plus thermal neutrons exhib-ited a higher magnetic flux loss at the same fluence. This findingmay be attributed to the thermal neutron capture reaction ofboron. Furthermore, hysteresis loops for Nd-Fe-B magnets exhib-ited an increase in coercivity following irradiation. We furtherobserved a 100% recovery of the magnetic flux, which indicatesthat structural damage is unlikely to play an important role inthe demagnetization process at these levels of neutron flux and flu-ence. For the model we proposed an ab initio MD simulation thattracks the temporal evolution of atomic positions and defects,which will calculates the change in magnetization of the samplefor the updated positions. Moreover, Sm–Co magnets will also bestudied in conjunction with Nd-Fe-B.

Acknowledgments

We would like to thank the support of the staff at the Ohio StateUniversity Nuclear Reactor laboratory. We also thank ProfessorAlan Litsky for lending us the diamond slow saw. This research isbeing performed using funding received from the U.S. Defense

A. Samin et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 43–47 47

Threat Reduction Agency’s (DTRA) research Grant (HDTRA1-13-0012).

References

[1] C.H. Chen, J. Talnagi, J. Liu, P. Vor, A. Higgins, S. Liu, The effect of neutronirradiation on Nd-Fe-B and Sm2Co17-based high-temperature magnets, IEEETrans. Magn. 41 (2005).

[2] J. Alderman, P.K. Job, R.C. Martin, C.M. Simmons, G.D. Owen, Measurement ofradiation-induced demagnetization of NdFeB permanent magnets, Nucl. Instr.Methods Phys. Res. A 481 (2002) 1–3.

[3] N. Miyahara, T. Honma, T. Fujisawa, Irradiation effects of a 10 MeV neutronbeam on a Nd-Fe-B permanent magnet, Nucl. Instr. Methods Phys. Res. B 268(2010) 57–61.

[4] J.R. Anderson, J.S Snyder, J. L Van Noord, C. Soulas, Environmental testing ofNASA’s evolutionary xenon thruster prototype model 1 reworked ion engine, J.Propul. Power 25 (2009) 94–104.

[5] F. Tang, J. Li, L. Ren, J. Shi, The Rd international symposium onsuperconductivity Jiao, A conceptual design of high-temperaturesuperconducting isochronous cyclotron magnet, Physica C 471 (2011) 1419–1423.

[6] San Francisco C.A. Aug Conference: 8. International Conference on SynchrotronRadiation Instrumentation, Radiation damages on permanent magnets:challenges for the future light sources, AIP Conference Proceedings, 705(2004).

[7] A.F. Zeller, J.C. Dekamp, Radiation resistant magnet R&D at the NSCL, ParticleAccelerator Conference, 2003. PAC 2003. Proceedings of the, 2003, pp. 161–163Vol. 161.

[8] R. Anerella, M. Harrison, M. Sampson, W. Schmalzle, J. Ronningen, R. Zeller, A.Gupta, Accelerator and detector magnets – Magnets for accelerators –Radiation resistant HTS Quadrupoles for RIA, IEEE transactions on appliedsuperconductivity: a publication of the IEEE Superconductivity Committee., 15(2005) 1148.

[9] K. Liehr, M. Rodewald, W. Salzborn, E. Schlapp, M. Wall, B. Boockmann, Effectof g-radiation on Sm-Co- and Nd-Dy-Fe-B-magnets, J. Magnetism Magn. Mater.101 (1991) 345–346.

[10] X.K. Maruyama, H.B. Luna, N.J. Colella, J.S. Hobbs, R.S. Hornady, B. Kulke, J.V.Palomar, Bremsstrahlung radiation effects in rare earth permanent magnets,Nuclear Inst. Methods Phys. Res. A 23 (1) (1998) 19–25.

[11] S. Ohashi, K. Kobayashi, N. Okuda, Effects of electron-beam and g-rayirradiation on the magnetic flux of NdFeB and SmCo permanent magnets,Nucl. Instr. Methods Phys. Res. B 94 (1994) 227–230.

[12] G. Medvedev, JPRS Report, Soviet Union: Economic Affairs (‘‘ChernobylNotebook’’ By G. Medvedev, Published in Novy Mir, June 1989), DefenseTechnical Information Center, Ft. Belvoir, 1989.

[13] G.J. Vargo, Review paper – A brief history of nuclear criticality accidents inRussia-1953–1997, Health phys. 77 (1999) 505.

[14] R. D. Brown, J. R. Cost, Radiation-induced changes in magnetic properties ofNd-Fe-B permanent magnets, IEEE (Institute of Electrical and ElectronicsEngineers) Transactions on Magnetics, 25:4 (1989) 3117–3120.

[15] S. Makinen, M. Talvitie, M. Manninen, O -P. Kahkonen, Radiation damage inNd-Fe-B magnets: temperature and shape effects, J. Phys: Condens. Matter 4(1992) 1007–1014.

[16] Youhui. Gao, Jinghan. Zhu, Yuqing. Weng, Baoshan. Han, Domain structure inFe-implanted Nd2Fe14B magnets, Appl. Phys. Lett. 74 (1999) 1749–1751.

[17] Yi-Kun Yang Li Li Wei Zhen Liang Han Bao-Shan Fang, Magneticmicrostructures of a high coercivity Nd-Fe-B sintered magnet in remanentand incomplete thermal demagnetization states, MAGMA J. Magnetism Magn.Mater., 322 (2010) 3720–3723.

[18] K.H. Lieser, Nuclear and Radiochemistry: Fundamentals and Applications,Wiley-VCH, Berlin, New York, 1994.

[19] J.R. Cost, R.D. Brown, A.L. Giorgi, J.T. Stanley, Effects of neutron irradiation onNd-Fe-B magnetic properties, IEEE Trans. Magn. 24 (1988) 2016–2019.

[20] R. Lindstrom R.W. Klaffy, B. Maranville, B.J. Micklich, R. Shull, J.H. Vacca,Thermal Neutron Demagnetization of NdFeB Magnets, Proceedings of EPAC2006, Edinburgh, Scotland, (2006) 3589–3591.

[21] A.M. Rutherford, D.M. Duffy, Including the effects of electronic stopping andelectron-ion interactions in radiation damage simulations, J. Phys.: Condens.Matter 19 (2007).

[22] R. Anupuru, W. Nolting, Quantum theory of magnetism, Springer, Heidelberg,New York, 2009.