EEEEEEEEEEC STABEEEE’E S“‘EEJ'E SF Smiles FREEMANENE EEAEZE “ER 3 “ or ‘EZ‘xG E) .zeqt’ee 0E M. 5 If t- W mr t?‘,,n1fir-'{\ 19,3 . 4.4?! 30.5%; SJ «11!.ngqu-Li RGEE in 5. Parker 1973 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE EEEEEEEEEEE 3 1293 006235 ABSTRACT Magnetic Stability Studies of SmC05 Permanent Magnets By R. J. Parker Sintered samarium cobalt permanent magnet specimens were prepared for the purpose of measuring magnetic stability parameters. Samarium cobalt magnets are in pilot production and are being considered for use in equipments having temperature changes from -196°C to 300°C. Irreversible loss, reversi- ble coefficient and viscosity changes at constant temperature were measured for two different geometries. Specimens held at constant temperature for times up to 150 days were evaluated for evidence of structural change. The results indicate that samarium cobalt magnets exhibit a reversible temperature coefficient about twice that of Alnico 5. The irreversible loss is a strong function of the goemetry of the specimen and is several times as great as for an Alnico 5 magnet. There is evidence of a continuing structural change at the +300°C exposure. It is concluded that samarium cobalt magnets can be useful up to +200°C in equipments requiring a high order of magnetic stability if the irreversible loss is removed by cycling and the reversible coefficient is compensated for by temperature sensitive thermomagnetic shunt material. MAGNETIC STABILITY STUDIES OF SmC05 PERMANENT MAGNETS BY Rollin J'." Parker ~ A THESIS Submitted To Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Metallurgy Mechanics and Materials Science College of Engineering 1973 W ACKNOWLEDGMENTS The author would like to thank Dr. M. G. Benz and Dr. D. L. Martin of the General Electric Corporate Research Center for their interest and suggestions in this work. Also acknowledgment is made to the Air Force Materials Laboratory, Wright Patterson Air Force Base who supported in part this work under contract F 33615-70 -C01098 entitled "Manufacturing Methods and Technology for Processing Cobalt -Samarium Magnets. " Iwould like to express my appreciation and thanks to Dr. R. Summitt of Michigan State University for his suggestions and guidance during this research project. ii TABLE OF CONTENTS List of Tables List of Figures Symbols and Nomenclature Introduction and Theory Description of Experimental Procedure Experimental Results Discussion and Conclusions Bibliography iii Page iv vii 24 36 43 47 LIST OF TABLES TABLE 1. Summary of Reversible and Irreversible Losses for SmC05 Summary of Magnetization Losses Due to Structural Change Summary of Changes in Magnetization Due to Viscosity Effects iv Pa ge 37 38 41 Figure 10 11 12 13 14 15 16 17 LIST OF FIGURES The Hysteresis Loop The Theoretical Effect of Packing Fine Particles Permanent Changes in Alnico 5 Intrinsic Coercive Force Irreversible and Reversible Changes Temperature Dependence of Spontaneous Magnetization Temperature Dependence of Saturation Magnetization Temperature Dependence of Coercive Force Change of Magnetization with Time for Various Fixed Fields Change of Magnetization Due to Mechanical Impact Effect of Stress on Vicalloy Effect of Demagnetizing Field and Recoil of Magnetization Characteristics of SmC05 Demagnetization Phase Diagram Sm -Co SmC05 Process Flow Diagram Unit Magnetic Properties of Test Material Hysteresisograph Block Diagram of Hysteresisograph Elements Page 12 12 14 14 15 18 18 19 21 23 25 26 28 29 30 Figure l8 19 20 21 22 23 Basic Elements of Torque Magnetometer Magnetometer and Permanent Magnet Field Supply Coil Fixture and I. D. V. M. Superconducting Solenoid Display of Magnetization Loss at Various Time- Temperature Values Comparison of Alnico 5 and SmC05 with Respect to Time Adjustment of Magnetization ‘vi Page 32 32 33 33 4O 42 ci (BH) max ‘LIST OF SYMBOLS AND NOMENCLATURE or Coercive Force (Oersteds): The magnetizing force required to bring the induction to zero in a magnetic material. or Intrinsic Coercive Force (Oersteds): The magnetizing force required to bring to zero the intrinsic induction or intrinsic magnetization in a magnetic material. The intrinsic induction often is referred to as the magnetization. or Remanence (gauss): The magnetic induction corresponding to zero magnetizing force. or Intrinsic Saturation (gauss): The maximum intrinsic induction or magnetization possible in a material MS also is used in liter- ature . or Maximum Energy Product (gauss-oersteds X 106) (M. G. 0.): The value of the product of demagnetizing force and induction. This is represented by the largest rectangle which can be drawn within the demagnetization curve. vii B/H or Load Line: This is the coefficient of self demagnetization or the ratio of induction to demagnetizing force. This ratio is a function of the magnet geometry and magnetic circuit parameters. Magnets that are long and slender have high values of B/H. Short magnets of large cross section have low values of B/H. or Magnetizing Force (Oersteds): This force can be established with ampere turns or it may be established as a result of the free poles of a permanent magnet. viii INTRODUCTION AND THEORY Permanent magnets are key components in a variety of industrial and consumer devices and equipments. They are evaluated in terms of volumetric efficiency (BH)max, resistance to demagnetization (coercive force), and ability to maintain constant magnetic flux over long periods of time. Some magnet applications are relatively insensitive to magnetization changes, and, in such use, magnetization changes of the order of a few percent are of no consequence. In calibrated instruments, long term stability and constancy of flux with respect to temperature variation of the order of one part in 103 have been commonly available for the last quarter century. More recently the use of permanent magnets in gyros and accelerometers for space missions have required higher levels of stability and necessitated new studies of permanent magnet stability. For a gyro to be useful, the accuracy of the torque must be known to one part in 105 and it must be constant for many months, hence, the accuracy of our deep space probes is heavily dependent on the long term stability of permanent magnetic fields. Early steel permanent magnets which relied on an inclusion mechanism for coercive force, were subject to structural change with time (aging) and perma- nent magnets were viewed with considerable suspiCion and mistrust prior to the development of Alnico type permanent magnets in 1940. Alnico permanent magnets show no aging or change of structure with time up to approximately 400°C, whereas ferrite permanent magnets developed in the early fifties are very stable with time up to approximately 200°C. The excellent stability of modern magnets puts them at a considerable cost and weight advantage over an electromagnet with regulated power supply in applications where a very precise field is required. In the past ten years, permanent magnet investigators have concentrated on rare earth permanent magnets. Certain intermetallic rare earth transition metal compounds possess the favorable combination of both high intrinsic magnetization and high crystalline anisotropy. Particularly, SmCoS magnets with outstanding unit properties (20 M. G. O. and 15 kOe) currently now are in pilot production and therefore it is desirable to characterize their stability with respect to other materials and to the general background of theory of permanent magnet stability. Origin of F erro Magnetism More than 60 years ago Pierre Weiss“) postulated two essential require- ments to explain the observed behavior of ferromagnetic materials. He believed the atoms themselves possess an inherent magnetic moment, and that there must exist between atoms a cooperative "molecular" force system which aligns their magnetic moments. Weiss pointed out that the "molecular" field would be effective only in local regions of a crystal and that over large volumes other regions with different directions of magnetic alignment would occur. Thus a bar magnet is said to be "unmagnetized" when no external field can be detected. Internally, however, the material is magnetized to saturation at all points; it is the subdivision into local regions or "domains" with mutually cancelling directions of magnetization which is responsible for the absence of external field. Application of a weak ex- ternal field will orient the already magnetized domains of the unmagnetized bar, thus giving rise to strong external fields. Weiss' domain theory satisfactorily accounts for the behavior of early ferromagnetic materials and today it is still the corner stone of an enlarged and more sophisticated theory of magnetic materials. 3 This early theory failed to explain why only certain atoms possess an inherent magnetic moment and why only certain crystal structures give rise to a "molecular field. " As our understanding of atomic physics increased(2) it became known that the atomic magnetic moment of certain elements is associ- ated with a net unbalance of electron spin. Later quantum concepts accounted for the existence of the "molecular field" in some materials. This field is electrostatic and aligns the unbalanced electron spins and their associated magnetic moments. These exchange forces also interact with the crystal lattice and make certain directions within the crystal the easy or preferred direction of magnetization. On the atomic scale, one can visualize an unbalance of electron spin leading to a net magnetic moment in some elements and interatomic exchange forces aligning these magnetic moments to produce domain regions oriented along certain preferred crystal directions. (crystalline an'Botropy). An interesting part of the theory of ferromagnetic materials relates to the features of the regions that divide adjacent domains. The direction of magneti- zation occurring between adjacent domains does not take place abruptly in one atomic plane, but extends over hundreds of atomic planes. This boundary region comes about because the exchange forces are attempting to align the electron spins, and misalignment between atoms can be reduced by having more atoms in the boundary region. On the other hand, crystal anisotropy acts to minimize the number of atoms involved in the boundary, inasmuch as, the moments of these atoms are forced away from an easy crystal direction. Consequently, the width of a domain boundary or wall region is a compromise between opposing demands 4 of exchange and crystal anisotropy forces. The Magnetization Curve and Hysteresis Domain theory accounts for three regions of a typical magnetization curve in terms of three different processes: first, domain boundary stretching; second, the growth of favorably oriented domains by continuous boundary move- ment; and third, rotation of the magnetization of the domain. These regions are illustrated in Figure l. The initial region can be attributed to elastic stretching of boundaries around imperfections, a reversible process. The steep irreversible portion of the curve takes place as the domain boundaries break away from the imperfections and move through the material, allowing domains favorably oriented with respect to the magnetizing field to grow at the expense of their less favorably oriented neighbors. The upper portion of the curve is a reversible region occurring as the fully grown domains rotate their magnetization against the forces of strain and crystal anisotropy into close alignment with the applied field. Saturation magnetization Bis is approached as the field magnitude is increased. When the magnetizing field is removed, the anisotropy forces again prevail and rotate the domains back into easy directions of magnetization with the magnetic induction being reduced to the residual point 3r This latter process can be suppressed by orienting the crystals along easy directions of magnetization. By applying a reverse field the boundaries can be returned to their original positions. The magnitude of the field to accomplish this is the intrinsic coercive force Hci' the measure of a magnetic material's resistance to demagnetization. This brief physical picture of the magnetization process indicates the im- portance of understanding and controlling domain boundary movement. In the world Induction B or Bi V / III Domain Rotation 1' A II Domain (Emmax / fiundary ovement ll .v E I Domain Boundar Stretching Ci c Magnetizing Field H Figure l - The hysteresis loop. 6 of commercial magnetic materials, two extremes of controlling domain wall movement and coercive force are identifiable: (a) in transformer steels domain boundaries must respond to very low fields to yield low hysteresis loss and efficient transformers; (b) for good permanent magnets, domain wall motion must be made very difficult so that Hci will be high and magnets can operate in strong demagnetizing fields. Origin of Coercive Force Early steel permanent magnets developed their coercive force by the introduction of structural inhomogeneities into the material. Domainboundary motion was impeded due to the energy required to pass a boundary through a non- magnetic inclusion. In the early steel magnets the nonmagnetic inclusions were precipitated carbides and these magnets developed only a small resistance to demagnetization. It is believed that this form of domain boundary impedance can account for coercivities only of the order of a few hundred oersteds. In 1935 F renkel and Dorfman(3) suggested that instead of merely impeding domain boundary motion it would be more fruitful to try to completely eliminate easily moved boundaries. By preparing particles with diameters equal or less than the width of a domain boundary, the sample would be too small to contain a boundary. This is the concept of the "single domain" particle, whose interesting properties are the key to developing and understanding modern high coercive force permanent magnets. In bulk materials, the boundary energy is small compared with the external field energy, but this situation reverses at a sufficiently small particle diameter, resulting from the fact that the magnetostatic or pole energy of the particle is a volume effect proportional to the cube of the particle radius, whereas the domain wall energy is associated with an area proportional to the square of the radius. As the particle diameter decreases the volume magnetostatic energy falls off more rapidly than the domain surface energy. Below a critical particle size there is less energy associated with an external field than with a domain boundary; this is the critical diameter below which single domain particles exist. For iron, this diameter is approximately 100 A; for barium ferrite, it is approximately 10, 000 A. The single domain concept has lead to the preparation of materials whose magnetization can be changed only by the simultaneous rotation of atomic magnetic moments. This can be made a much more difficult process than simple domain boundary motion. With no boundaries which can move,the coercive force is determined by the anisotropy forces which oppose domain magnetization rotation. TWO forces of particular significance are crystal anisotropy and shape anisotropy. Crystal anisotropy directs the magnetization of a fine particle along an easy crystallo- graphic direction. In order to rotate the particle's magnetization, an applied field must provide enough energy to rotate the magnetization through a difficult crystal direction. The resulting intrinsic coercive force for favorably oriented particles is given by W (1) where Hci is the coercive force (oersteds), K the crystal anisotropy constant (ergs/ cm3), and MS is the magnetization (gauss). By substitution of the appropri- ate values in Equation (1), coercivities for several materials are: iron, 500 Oe; cobalt, 6000 Oe; barium ferrite, 16 000 Oe and cobalt samarium, 300 000 Oe. 8 Shape anisotropy arises from the interaction of dipole -dipole effects in various shaped particles. By preparing elongated single domain particles one finds that the preferred direction of magnetization is along the major axis. To rotate the magnetization across its width requires additional energy. It can be shown(2)(4) for a long cylindrical particle that: I_Ici = (Na " Nb) Ms (2) where Na and Nb are functions of particle geometry and MS is the saturation magnetization (gauss). The above equation assumes coherent rotation of all of the magnetization vectors within the cylinder, but in practice, coherent rotation does not occur. Instead lower energy modes of reversal such as curling and fanning prevail. (5) Alignment and Compacting Single Domain Particles Particle alignment is significant in permanent magnet technology because particles randomly oriented have their magnetization vectors rotated into align- ment only at saturation; when the saturation field is removed, the vectors are rotated by the anisotropy forces back into any easy axis of magnetization. Theoretical predictions(6) would indicate that both the residual induction and the coercive force can be doubled by alignment and the maximum energy product (BH)max would be improved by a factor of four. Particle packing has one obvious effect: packing particles closely together directly increases the induction of the magnet volume. If particles are in contact with one another, there is the possibility of reducing coercive force because of loss of single domain behavior. If this is overcome, fine particle magnets have their optimum energy product at the greatest possible packing as shown in Figure 2(a). 1.5.D-ph.x.'m .. vb... . Inn-3!. . n . . . p 11:. ”ambit-II?! Hci' oersteds Br' Gauss 1.0 1.0 C? o __, . . 0.8 0.8 '1 Q 00” ranH c1 - 0.6 0.6 "" °. :2" .8 \o a a .8 0.4 0.4 g :13 II: (BiH)max./Hci»° Br, 1'0 $5: 0.2 0.2 0 0 0 0.2 0.4 0.6 0.8 1.0 (a) Packing Fraction (BH)max, m. g. o. '0 0.2 0.4 0.6 0.8 1.0 (b) Packing Fraction Figure 2 - The theoretical effect of packing on fine particles with (a) crystal anisotropy and (b) shape anisotropy. (Luborsky and Parker9). b'llhr .u... n _ :5 .3 I'll - .Inr. . .. v 10 This is the observed case for particles deriving their coercive force from crystal anisotropy. For the case of magnets based on shape anisotropy an additional relation- ship must be considered. As elongated particles are packed closer together the coercive force will diminish. Neel“) suggested from magnetostatic energy con- siderations that the relationship between Hci of an elongated single domain particle at infinite dilution and its coercive force Hci in a compact or assembly of particles is given by Hci’ compact = Hci’ particle (1 -P) (3) where P is the packing fraction of the particles; this relationship has been confirmed for the case of dilute compacts. Since the Hci decreases and the induction increases as elongated particles are packed together, the optimum energy product occurs at an intermediate packing fraction as shown in Figure 2 (b). In "real life" permanent magnets, coercive force is often due to the contri- bution of both shape and crystal anisotropy. Alnico and Lodex® permanent magnets rely for their coercive force primarily on the formation of elongated iron-cobalt regions in a spacing matrix which is less magnetic or, in the case of Lodex, non- magnetic. Barium ferrite and rare earth cobalt magnets utilize crystal anisotropy for the development of very large values of Hci' Description of Stability Parameters In the study of permanent magnet stability it is useful to classify magneti- zation changes as to cause and nature. The following categories and definitions of magnetization change are relevant to the problem. ® Registered trade mark of General Electric Company 11 I. Irreversible Changes (A) Irreversible changes resulting from a permanent change in the structural aspects of the magnets. Examples of such changes which are time -temperature sensitive are: (1) (2) (3) (4) Growth of precipitate phase Oxidation Annealing effects Increase in proportion of an ordered phase. Remagnetization does not restore the original state of magnetization after this kind of change. In the literature this type of change often is referred to as aging since there is time dependence. Permanent changes in Alnico 5 after exposure to various temperature is shown in Figure 3. The temperature at which changes in properties first become noticeable corresponds to the beginning of structural changes and corresponds closely to the maximum service temperature. (B) Irreversible changes resulting from a change in the magnetic state. In this type of change after the removal of the disturbing influence the magnetization does not return to its original value. Examples of such changes are: (1) (2) (3) Ambient temperature changes Statistical local temperature fluctuations leading to viscosity effects. Magnetic field -induced changes such as a change in magnetic circuit reluctance or an external field. In this kind of change magnetization may be fully restored by remagnetization. 700 12 600 500- 400 300 200 100 l-lCi Oersteds 200 400 600 800 Temperature °C Figure 3 - Permanent changes in l-lC of Alnico 5 after one hour exposure to each temperature. (Luborsky and Parker9). II. Reversible Changes The reversible changes in the magnetic properties of permanent magnet materials as a function of temperature originate in the change in spon- taneous magnetization. These changes tend to obey the same temperature law as does saturation magnetization. Reversible changes are functions of temperature and are in no way time dependent. They disappear complete- ly without need for remagnetization when the permanent magnet is returned to its initial temperature. The distinction between irreversible and reversi- ble changes is shown in Figure 4. wn> Magnetization Tr T Temperature Figure 4 - Irreversible and reversible changes. III. 13 If A is the level of magnetization at a reference temperature, e. g. room temperature (Tr) then AC is the irreversible loss after exposure to temperature T with measurement made after the specimen has been returned to room temperature. Irreversible loss is usually expressed as a percentage of the reference temperature level. The change in magnetization CB represents the reversible change and is usually ex- pressed as an average coefficient over a limited temperature range in percent change/degree. Temperature Effects (A) Effects of Temperature on Bis and Hci The properties of a magnet vary with temperature in a manner which is predictable in many cases. The shape of the Bis vs T curves may be calculated from theory, provided there is available a detailed knowledge of the crystal and magnetic structure of the magnetic phase. Examples for iron, cobalt, nickel and 27. 5 Cu-Ni alloy are shown in Figure 5. In many cases such information is not yet available. Instead, direct measurements of Bis vs T have been made. Examples for many permanent magnet materials are shown in Figure 6. The change in Br with T, however, depends on the extensive properties of the system, i. e. , particle size, sample shape, and thus demagneti- zing fields and domain structures. These changes with temperature must be determined experimentally for the specific material and con- figuration of interest. Br or Bis Kilogauss 14 0. 2— T/ea'o 0.2 0.4 0:6 0.8 1.0 Figure 5 - Temperature dependence of spontaneous magnetization of iron, +; cubic cobalt, D ; nickel, O; and 27. 5 Cu-Ni, x. 0 is the Curie temperature and 6's and 6'“, the saturation magnetization per gram at T and at T = O°K. (Luborsky and Parker9). 16 Alnico 5 12 ————————————— "‘~~‘\ Lodex (50% 10 \ Vicalloy II N \PaCklflg Of Fe \ \\Co) \\ Alnico 3 \ \\\ \ __ Alnico "8", Br “ \ \\ \ Barium Ferrite \ | \ I I I | I -200 0 200 400 l 000 Temperature °C Figure 6 - Temperature dependence of saturation magnetization or remanence. “It; us .‘uaou8.. Hci Oersteds l 000 900 800 . 700 600 500 400 300 200 100 15 The changes in Hci with temperature may be understood and predicted from a knowledge of the change with temperature of the anisotropy K and magnetization. This assumes a knowledge also of the physical origin of all of the anisotropies contributing to the Hci as discussed in the section on permanent magnet theory. Experimental results are shown for many magnets in Figure 7. K“ LOdBX Barium Ferrite 35% Co Steel -200 0 200 400 600 800 Temperature , °C Figure 7 - Temperature dependence of coercive force. (Luborsky and Parker9). (B) 16 Time Effects at Constant Temperature The time adjustment of magnets at constant temperature is generally referred to in the literature as magnetic viscosity. The terminology must be used in the broad context that viscosity describes a material property which yields steadily before a constant stress. The domain regions of magnetization are in a static self imposed demagnetizing field. Additionally, they are in a field which fluctuates in time. The time dependence of this superimposed field was suggested by L. Neel. (4) This field is dependent on the logarithm of time. At a domain site the fluctuating field could arise from such external ambient conditions as stray magnetic fields and temperature changes, and from internal temperature fluctuations. The temperature fluctuations are seen as field changes because of the intrinsic magnetization temperature dependence. A freshly magnetized magnet has a certain number of domains whose magnetization vectors are in positions that might be called "metastable. " An energy change can initiate a transition to a more stable position. These events would involve a discontinuous orientation change. The adjustment of magnetization is essentially an activation process. Because this adjustment is thermally activated, it is accelerated at increased temperatures, so that quick stabilization can be achieved by raising a magnet's temperature for a short time. Since the changes result from magnetization reversals, they can be anticipated and stabilization effected by subjecting the magnet to alternating fields sufficient to demagnetize to the extent of the loss IV. 17 which would occur during the time period of interest. Street and (7) Woolley has shown that the time temperature dependence is given by M(t) = Const. + AN Ms k T log t (4) = Const. + S log t Where S = A N MS k T. N is the number of domain regions of magnetization MS per unit volume; is the constant probability energy density E and k is Boltzmann's constant. T is the constant absolute temperature. Since these factors are relatively independent of temperature except near TC; S is nearly directed proportional to T. However, N and A will depend on the level of self demagnetization of the sample. The results of experiments are in general agreement with equation (4) as shown in Figure 8. Mechanical Effects Mechanical shock and vibration add energy to a permanent magnet to decrease the magnetization in the same manner as discussed for the case of thermal energy. The only difference is that energy imparted thermally to the magnet is precisely kT while the energy imparted mechanically is usually not known. Thus, repetitive shocks or continual vibration should decrease the magneti- zation by the same logarithmic relations as derived for thermal energy effects but where the time variable is replaced, for example, by number of impacts. Stability-impact relationships are shown in Figure 9. The curvature of the results for the 2 Co, 4 Cr steel may be due to the fact that both the large normal room temperature time decrease and the decrease due to the impacts may be present. Little work has been done regarding Magnetization Loss, Percent 18 200 160 120 80 40 M, Gauss 30 60 300 600 3000 6000 Resting Periodt, Sec. Figure 8 - Change of magnetization with time for various fixed values of reverse field (Street and Woolley7). ¥ 35% Co Steel Tungsten Steel 2% Co, 4% Cr Steel Number of impacts less 1, N-l Figure 9 - Flux loss in bars dropped 1 meter onto hardwood floors. (D. Hadfieldll) l9 stabilization to minimize mechanical effects because it is seldom found necessary after thermal and field stabilization. There is limited infor- mation that suggests that both thermal and alternating field exposure will minimize but not entirely eliminate the change in magnetization due to shock. Some magnets subjected to tension or compression show large changes in properties. This is especially true of Vicalloy as shown in Figure 10. The changes are probably due to the contribution the stress makes to the total anisotropy of the system as discussed in the previous section. B, Gauss B,Gauss l -800 —600 -400 -200 H, De Figure 10 - Effect of stress on Vicalloy. Applied stress of (a) O; (b) 51; (c) 101; (d) 152; (e) 203; (f) 254; and (g) 305 kg/mmz, (Shur, Luzhinskaya, and Shubinalz). 20 V. Magnetic Field Effects The level of flux supplied by a permanent magnet can be irreversibly changed by exposing the magnet to a field. The effect of the external field while acting on the magnet and the final effect after its removal can be predicted. Intrinsic and normal demagnetization curves are shown in Figure 11 (a). Upon exposure to an external field, AH, the operating point moves down the curve and the change in intrinsic properties can be projected to the normal curve which shows a change AB occurring while the field is applied. The recovery along an interior loop when the field is removed is shown in Figure 11 (b). A B is, of course, smaller than in Figure 11 (a). Subsequent exposure to fields of lesser strength give small changes within the previously established minor loop, and upon removal no change in magnetization results. A given reverse field AH will cause different changes in A B depending on the permeance line. The loss will be greater when a magnet is operating in the steepest part of the demagnetization curve. Instead of exposing the permanent magnet to a field A H, the same effective stabilization will result if the air gap is increased and decreased so that the same minor hysteresis loop is estab- lished. For example, in the removal of a magnet from a magnetizing yoke and its subsequent transfer into an instrument gap and return path, stabilization is achieved. In the analysis of permanent magnet stability it is necessary to consider the role played by magnetic circuit loading and operating load lines. SmCo5 magnets are unusual in that Hci is much larger than Br“ Under these conditions, 21 B/H \ B/H + 1 = Bi/H D w I l t.— I | -—-l Magnetization B1 or Induction B Demagnetization Field (H) Figure 11( a)- Effects of a demagnetization field. /H \\B/H + 1 = Bi/H Magnetization (Bi) or Induction (B) Demagnetizing Field (H) Figure 11 (b) - Recoil of magnetization 22 the intrinsic magnetization curve and the normal induction curves differ greatly. This great difference is shown in Figure 12. The induction curve differs at every point by -H. With Hci>> Br the induction curve tends toward a straight line with HC approaching Br as a limit. The permanent magnet as normally used is subjected to the self demagnetization influence of its poles and operates in the second quadrant of the hysteresis loop. The extent of the self demagnetization is controlled by the magnet geometry and its magnetic return path parameters. For optimum design, often the self demagnetization is controlled so that the product of -H and resulting induction B are maximized. This is the maximum energy concept. In electro- magnets, the magnetization is a result of setting up a magnetic potential, a permanent magnet is inherently different. The magnetization results from the orientation of atomic moments and the field potential -H is a byproduct of the magnetization. If the magnetization curve is changed for example by thermal energy input, then the resulting change in useful induction (B) will depend on the load line (B/ H) and the level of self demagnetization involved. It is thus possible by geometry change to have a change in magnetization reflected into the induction curve by varying amounts. 23 Load Line ._.—-—-‘—-—--—' V I Net Flux Density B for External Circuit (BH)max Hci Hc H— -H ’I ‘ >- Useful Potential d B=Bi+(-H) l..._._. 13...- _._._.‘ Figure 12 - Characteristics of SmC05 demagnetization. EXPERIMENTAL PROCEDURE Sample Preparation Samarium cobalt magnet samples for magnetic stability measurements were prepared in the General Electric Magnetic Materials Development Laboratory using equipments and techniques developed for pilot production of samarium cobalt permanent magnets. The intermetallic compound SmC05 was formed by melting 99. 9% pure cobalt and samarium in a vacuum induction furnace. The resulting ingot was crushed in a jaw crusher and the powder was then reduced to the 5 to 10 micron size range in a fluid energy jet mill under nitrogen gas. Two compositions are involved. One of 33% samarium content and the other of a 60% samarium-rich phase which is used as a liquid phase additive. The two compositions are mixed and blended to yield a composite composition of 37% samarium and 63% cobalt. Figure 13 shows the phase diagram for Sm-Co. The powder was then placed in a rubber -lined container and the particles were oriented in the 60 KOe field of a super conducting solenoid. The particles were mechanically restrained while in the field by lightly pressing in on the covers of the container. The assembly of oriented crystals was then hydropressed at a pressure of 200 000 p. s. i. to a density of 80% of theoretical. After removal from the hydropress, the billet was then sintered in a stream of argon for 30 minutes at 1120°C. This final densifi- cation step raised the density to approximately 90% of theoretical. This procedure is shown in block diagram form in Figure 14. For the determination of irreversible and reversible coefficient data, one typical bar of SmC05 1/2" diameter and 2" long with orientation parallel to the 24 25 Sm mz '3.le " L ' ‘ II . 5M... .7 54.840-45.314... . _ 4 . . .E.‘ 4fil 4 “.1... W uni-05-: W 4.. .4. .3...Lu:.zg.t....H. .27.. 4 ”24% . 4.1....1431534 . ....m. .. . ‘ ......m- ...... i .1. 1...... E .1. m 1 W1“ ego . :4 . . . ._ .. . . ..... ....9 .. .... .4. .. ... .. _. .. 5 + 4 .... .... L414 Lift? .mtmhin 1 ...mm 1.4 H. 0.3.16.3- 7E4 u. :4”... :43. x43... .3 .. C .43“ .. 4 r: o 4. . ..., . w 411 .Mo 9... x? 54m :7. .1 M _ “.8 ........ mt . .4413. ...Mxflt . 4 4 - -14 ._ W. 1.47: ..fi 41.32.40 “ .414 :2; .H ,. 1m. 1: “my 17 m 0 q. 4.4 - Wm. .- :44» 4 444134-144 s a... .m” H“... . M. .4.” 3.40 O .4... 4. ..4 ..H 44 9.4 46 o/ . 11.41.11”. 1+ 1.. 1...... m.m..._4m:..r..-.1... _ 4 .4; 3.4 4 x. x. S . . .fi 4 ..x O T. O: 4 4E 1.. a 41 o _.1..1m :15 ....... .... .. . . . : ... . H 1.1 .. .4 m ... C . .0 . w G . :4.. 4 :4.. .34 ..... 114.”. .42).»..4- .1. 341.. u 4. . 4 4. 4. x . 4 .34 a . E 14 4 .EE . M. . n. .. ... 4 . n. w w m ... .H ........ 2-..... .14 4 .440 3 O 0‘ E _ . ..- 4 ...:....43.3 1.3., 4m £631.1I10140111T01ILIJOquuihii . vumerttfiagvgfn...an........w..- .Smu 0.13.. «Entr04... 5414.! . a . . .. 4 . .. wit.fiiv..§g..gu'od. r'r I ill I 1.10.7... 1.. 1;) 141.11%. IIErhI. 4 _ 4 E... (O 4.44m .00+Aoov..11 .. . 3.4“: 4JLMLH 124.... 4.1....“ '.'1:;.. ..... 5...- Co ‘ . ,...... .. .... 4 O O 0 w 0 m 0 o O O 9 8 7 6 5 3 1200 .... MOO “* 000 o o ATOM °/o 5m Phase diagram Sm -Co Item, I: (1968), 'P. 323 Figure 13 Co Sm V Y Vacuum Cast 26 Jaw —""' Crushed Align in 60, 000 Field ~1———— Blend Pulverize —* and Jet Mill 12 )4 Size Base Alloy Int—- 60% Rich Material V Hydropress Sinter Final Grind Figure 14 - Flow diagram of C053m manufacturing process. 27 2" length was used (reference No. 17). This bar was fully saturated in the super conductor and a second quadrant hysteresis demagnetization curve was obtained. The properties for this bar are plotted in Figure 15. Since this bar was uniform and exhibited typical unit properties, it was cut with a diamond wheel into the final desired specimen geometries. TVs/o geometries giving B/H values of approximately -0. 4 and -l. 0 were obtained by forming short cylinders. Speci- mens 17A and 188 were cut from one end of the bar. Specimen 17E was cut from the center and specimens 17H and 171 were cut from the opposite end. Sample magnets for the determination of viscosity changes and for the measurement of structural stability were cut from sintered rods which were centerless ground to 0. 515" diameter. After hysteresisograph measurements indicated the ingots were of typical unit properties, seventeen specimens were sliced from the long rods. Each specimen being .062" long with the magnetization being parallel to the axis of these short cylinders. This geometry specimen has a B/H load line of 0. 2. Eleven of these specimens were used in the structural long term stability measurements program and the remaining six were used to obtain the viscosity data. Description of Measuring Equipment Some rather specialized measurement equipments were involved in this measurements program. For the determination of the demagnetization character- istics of the specimens a model MH-S Walker Hysteresisograph was used on the long rods prior to cutting into final specimen size. The three main components of the hysteresisograph are shown in Figure 16. The electromagnet power supply is on the right, a 10" water cooled electromagnet is in the center of the picture 12 28 11 10 9 8 7 6 5 4 3 2 1 -H - KOe Figure 15 - Demagnetization characteristic of SmC05 rod used to cut test specimens from. B-KG 29 Figure 16 - Hysteresisograph and the control console and X-Y recorder are on the left side of the picture. A block diagram of the elements of the hysteresisograph is shown in Figure 17. A Hall effect probe measures the applied field H and a coil surrounding the test specimen reads Bi or level of intrinsic magnetization. This coil is wound so that the total induction (B) has the H component automatically subtracted from it to yield a voltage output proportional to intrinsic magnetization (Bi)' The output from the Hall probe and the integrated signal from the B1 coil are both amplified and the signals are displayed on meters and used to drive the X-Y recorder. As the field is changed by the control of current in the electromagnet the relation- ship between Bi and H .are displayed on the X-Y recorder. To measure percent change in magnetization at other than room temperature a torque magnetometer was used. The arrangement of the torque magnetometer El tromagnet Hall Probe Flux Meter / B Coil W Control Specimen Power Supply Figure 17 - Block diagram of hysteresisograph elements. X-Y Recorder 31 elements and the permanent magnet field supply are illustrated in Figure 18. The samples are rotated in a fixed field of 100 oersteds. The magnetic moment per unit volume of specimen gives the magnetization directly without the need for search coils and calibration constants. The millivolt output of the transducer is proportional to the torque produced. This output is read on a Hewlett Packard integrating digital voltmeter (IDVM). A photograph of the torque magnetometer H“ is shown in Figure 19. To determine changes of magnetization where the measurements are compared only at room temperature a search coil fixture is used. A precision '4 coil form in which a spring loaded sample holder can be moved a precise distance away from the coil is used. The voltage output from this fixture is measured with the IDVM. A photograph of this equipment is shown in Figure 20. For magnetization of specimens both the electromagnet of the Walker hysteresisograph and a General Electric 60 kilo oersted super conducting solenoid were used. The part of the measurements program for determination of irreversi- ble loss and reversible temperature coefficient used the General Electric super conductor in which fields of 60 kilo oersteds were measured. For reason of convenience in the later part of the program, in which structural stability and viscosity effects were measured, the Walker electromagnet was used. Although the electromagnet field is limited to 35 kilo oersteds, manufacturing experience has indicated that there is negligible difference in level of magnetization achieved with the two equipments. A photograph of the laboratory super conducting solenoid and its control equipment is shown in Figure 21. 32 Transducer ‘5. I II 7 7 ’ I [ IDVM J_ I (Fixed Support L '~ Tungsten Torsion l 7/ r / Support u / I r Be er Sample Holder Permanent Magnet Field Figure 18 - Basic Elements of torque magnetometer Figure 19 - Magnetometer and Permanent Magnet Field Supply 33 Figure 20 - Coil fixture and I. D.V. M. . Figure 21 -‘ Super conducting solenoid. 34 Measurement Procedure for Irreversible Loss and Reversible Coefficient Each of the five specimens for this part of the program was carefully measured dimensionally. The specimens were marked with an electric pencil so that identity could be preserved. The group of specimens were stacked together and placed in the super conducting solenoid for magnetization at 60 kilo oersteds. The magnets were carefully centered in the region of maximum field uniformity. At room temperature each specimen was placed in the sample holder of the torque magnetometer and the maximum torque was obtained by rotation of the specimen slowly through a position where the field and axis of the specimen magnetization were at 90° to each other. Four torque readings were averaged for the final millivolt reading representing torque. The samples were next inserted into a beaker of liquid nitrogen (-l96°C). The arrangement of the sample holder with respect to the field supply allowed the sample holder and magnet to be covered in the beaker. Torque readings at this low temperature point were measured and averaged as previously described for the room tempera- ture data. The magnets were then returned to room temperature and the torque readings repeated. For obtaining the high temperature exposure a beaker of oil was used in which each sample was rotated. An immersion heater and thermometer were used to maintain the oil temperature over the period of time required to make the measurements. The temperature of the oil was maintained at 150°C. For the final set of torque measurements, the magnets were returned to room temperature and the torque measurements repeated. 35 Measurement Procedure for Long Term Structural Stability Each previously described specimen was measured dimensionally and numbered with an electric pencil. The specimens were weighed and the physical density calculated and recorded. Each specimen was individually magnetized in the Walker electromagnet at 35 kOe. Using the precision search coil and IDVM a reference level of magnetization was obtained for each of the eleven specimens. The recorded data is the average of three readings on each specimen. Experience with this coil and IDVM with respect to repeatability of readings r-i indicates an accuracy of i 0. 1%. After the starting measurements the speci- mens were placed in three groups on aluminum heat sinks and placed in three separate furnaces at 100°C, 200°C, and 300°C in air atmosphere. After 24 hours, 7 days, 30 days and 150 days the magnets were removed from the furnaces and allowed to cool slowly to room temperature. Each specimen was then fully magnetized in the Walker equipment under precisely the conditions used to obtain the original magnetization reference. The precision coil and IDVM were used to measure the change in magnetization after each interval of time -temperature exposure. Procedure for Determinini Viscosity Changes The six specimens for this test are identical in size to those used in the structural test. Each specimen is magnetized in the Walker electromagnet at 35 kCe. The precision coil and IDVM were used to establish the reference magnetization level 24 hours after magnetization. Each specimen was read without remagnetization at the 48 hour and 120 day interval. Again, the recorded data is the average of three readings on each specimen. EXPERIMENTAL RESULTS In Table l a summary of reversible and irreversible losses is shown. The millivolt readings in this table are proportional to the torque (dyne cm) per oersted. Since we are only concerned with percent change, the data was not converted into absolute magnetization units. The data indicate, after the low temperature exposure, the irreversible loss is extremely small and could partially be the result of viscosity changes that would normally occur over the two hour time span involved in these cycles. The irreversible loss after the +150°C cycle shows two distinct levels of change as a function of specimen geometry. This is indeed significant and will be interpreted in the discussion. The reversible coefficient is expressed in percent change per degree. The difference in millivolt readings between the two temperatures expressed as a percentage of the room temperature reading divided by the degree span yields the reversible coefficient. The data indicate two distinct values with the low temperature region giving a coefficient considerably lower than the above room temperature range. This results from the non linear change of magnetic moment with respect to temperature in these compounds. The structural stability test data is presented in Table 2. The physical density (Dt) is expressed as a percentage of the theoretical density for SmC05 taken as 8. 6 gm/cm3. The percent change in magnetization is referenced to the starting level for each time -temperature exposure. At the 300°C exposure, three of the four samples were lost during the testing. These samples cracked severely and C(llld not be used in the precision coil test fixture. The results of 36 37 Table 1 - Summary of Reversible and Irreversible Losses Diameter (in) 0. Length (in) B/H -. r at +23°C 411. at -196°C 434. Torque at +23°C 410. (dyne -cm 1 per Oe) at +150°C 344. t at +23°C 363. % Irreversible change in magnetization at +23°C after exposure to: -196°C -0. " " +150°C -11. Reversible temp. coer. --% magnetization change per °C over range +23° to -196°C +. Reversible temp. coef. -—% magnetization change per °C over range +23° to +150°C -. 535 .195 958 9 28 042 0. 529 .095 188.9 198.2 188. 1 154. 6 162. 9 -13.3 17E 0.529 0. .200 -1. 00 - 411.8 189. 433. 4 198. 410. 8 189. 351. 6 151. 371. 3 159. -0. 24 -0 -9. 8 -15. +. 024 + -, 024 Avg - - 042 - 17H 171 532 0. 527 .095 . 204 . 416 —1. 03 6 436. 6 9 463. 6 0 435. 3 4 375. 4 8 393. 3 . 30 -0. 30 8 -9. 6 022 +. 028 042 -. 038 38 H.0H o.m H.m w .3 v.0 - o.o n.v m.~ o.N o.m H.N o.m o.N m.m m.m m.m m.H m.H m.~ w.o w.o w.o B.H n.H h.a m.o m.o ,m.o Q omH Q on Q N. omwouoofl o5 m.N H.v m.~ m.~ m.m o vflvm mmo.o oxoum oxoum who.a moo.H owo.o no~.~ oo~.~ wmo.H mwo.H £3 o2 >8 wmo.H moo.H www.o mmo.o woo.a hmo.a who.a who.a hoo.a noo.H wmm.o wmo.o ao~.~ noH.H oo~.~ HoH.H wmo.a mmo.~ Hwo.q owo.~ 93D Om 9an n >m: NPc: oxoum Ono.a mwm.o who.a mwo.a mho.a wom.o VHH.H noH.H wmo.a mwo.a .HE wm >8 vmo.o woo.a BNO.H VOH.H noo.a hwo.a mom.o vNH.H wo~.~ ooo.H owo.H gum >6 c.0w oom m N.Ho oom ma o.No com Va H.Ho oom Ma H.Ho OON NH O.Ho CON HM H.0m com OH o.oo com a m.oo OOH m H.0o OOH N o.oo OOH H HQ 05 Do @809 .02 638mm omcmnu Hausuosbm 8 «ED mowcwnu :oUmNUofiwg/H Ho >Hm885m . m 2an 39 these tests are displayed in graphical form in Figure 22. Each point is the average of the change for the total number of samples involved. In the case of the 300°C exposure for 150 days, the data point represents the single surviving sample. The results of the time adjustment of magnetization at room temperature are shown in Table 3. Percentage change referenced to the starting level was calculated and plotted in Figure 23. The time adjustment of Alnico 5 is shown also for comparison. This data on Alnico 5 is from the central research facility representing the magnet industry in Great Britain. (10) 105 100 95 90 85 80 Percent Magnetization Retained 75 4O 300°C 1 10 100 Time Days Figure 22 - Structural Change at Constant Temperature 1 000 41 Table 3 - Summary of Changes in Magnetization Due to Viscosity Effects. Magnet No. Start mV 48 Hr. mV 120 Day mV % Decrease (48 hr.) (120 D.) 4 1. 074 1. 073 1. 070 . 095 . 377 7 1.114 1.113 1.110 .083 .271 8 1.110 1.109 1.106 . 090 .360 16 1.079 1.078 1.074 .093 .466 17 1.119 1.118 1.113 .085 .505 18 1.044 1.043 1.038 .094 .475 Avg. % Decrease -------------------------------- . 090 . 409 42 0 SmCo5 (B/H = 0. 2) a . 25 .g \\ (B/H = 26) Cd .3 ‘\. a) \0.‘ Alnico 5 (British PMA) 50 \., § - 50 \‘x 5 B/H = 5.0 Q) U) C1 Cd 4: o 8 .75 U H Q) 04 1.00 1. 25 1 10 100 Days 1000 10 000 Figure 23 - Comparison of Alnico 5 and SmC05 with respect to time adjustment of magnetization. (British PMAlO). DISCUSSIONS AND CONCLUSIONS The reversible temperature coefficient for the range 23°C to 150°C for the average of the samples is -. 041%/°C. This value is about double that for Alnico 5 but still only one fifth that for barium ferrite. For applications where the flux output must be constant SmC05 will have to be used in conjunction with thermal shunts that change their permeability with temperature. The higher reversible coefficient only means that slightly more flux from the magnet must be in the shunt or regulation path and hence the useful flux available is less in . the useful path. The higher coefficient tends to place SmC05 at a disadvantage " in applications requiring constant flux output over a wide temperature range. In the temperature range +23°C to -l96°C the average reversible coefficient is . 024%/°C. This value is approximately the same as Alnico 5 in this temperature range. It is of int erest to note that with Alnico 5, Hci is much less than Br and consequently the magnetization varies as does the effective load line over a long bar of Alnico 5. Due to the relative relationships between B and -H useful Alnico magnets are long and slender. Under such conditions we find the reversible coefficient to vary from point to point along a bar. It is possible to choose magnetic geometry so that an over all reversible coefficient of zero is achieved. With SmC05 magnets, the magnetization is uniform over the short specimens that will find most usage. Hence, it will be impossible to control the effective reversible coefficient by choice of geometry. For specimens with uniform magnetization, this coefficient is an intrinsic property for a given composition. The reversible changes of magnetization with temperature have their origin in the same electronic -atomic structure changes which cause the spontaneous 43 44 magnetization of any ferromagnetic material to generally decrease with increasing temperature until it becomes nonmagnetic at the curie temperature. In general, reversible changes become more marked as the curie temperature is approached. Some of the rare earth elements are exceptions to the above and actually do show an increase in magnetic moment as temperature is increased. It may well be possible to compensate at the atomic level and develop a permanent magnet with zero or positive reversible coefficient. The irreversible loss of magnetization with temperature has been found to be a strong function of where the magnet specimen operates on its demagnetization curve. In general, permanent magnets which derive their intrinsic coercive force from crystal anisotropy exhibit a strong temperature dependence with respect to property change and SmCo5 is not an exception. Our data clearly indicate that as a magnet is more heavily self-demagnetized (lower B/H ratio), it will exhibit larger values of irreversible loss. The two levels of self -demagnet- ization used, B/H = 0. 4 and 1. 0 represent two levels of sensitivity with respect to the change of orientation with temperature. This work has clarified the importance of the level of Hci and the order or shape of the intrinsic demagneti- zation curve. Or, in other words, the nature of how the magnetization changes with respect to adverse forms of energy imput is very important. It is indicated that we need to increase the level of Hci so that it can diminish with increased temperature without a reflection of this change being detected on the induction curve which sets the flux output to a magnetic circuit in a device. By making Hci of the order of twice HC and achieving a well ordered system of magnetization vectors, the irreversible loss could be greatly reduced for exposures to 300°C. 45 At approximately 420°C, it has been shown(8) that SmC05 loses essentially all of its crystal anisotropy and from this point to the curie temperature of approximately 700°C, it is only feebly magnetic. The squareness of the intrinsic demagnetization curve has great influence on the change of magneti- zation with temperature. Ideally we would like to have all of the magnetization vectors associated with individual domain regions flip over at the same level of field. Due to a distribution of surface features which nucleates magnetization reversal sites, this is not realized in magnets made to date. At this time, we can only identify the need to study process features which could lead to particles having a narrower intrinsic coercive force distribution. The most significant data from this investigation is the evidence that structur- al change is occurring above 200°C within a very few days. At 200°C the change was found to stabilize after seven days. This problem could be corrected by adding a temperature cycle at the end of the manufacturing process to accelerate the small structural changes and force them to occur prior to assembly in devices. The continuous change observed at 300°C suggests that at this temper- ature level changes have occurred in the microstructure or composition which have permanently lowered Hci' Martin and Benz(8) have suggested oxidation may be a factor and have shown improved structural stability can be achieved by sintering to higher densities. Due to the narrow composition tolerance necessary to obtain SmC05 any time -temperature induced shifts in the phase field could certainly have significant influence on the intrinsic magnetization and coercive force. It is clear that additional work is required to determine if oxidation and compositional shifts are indeed the reason for the change. Most of 46 the development effort on these magnets has focused on improving the room temperature properties. It would seem reasonable to suggest that much more attention be given to improving the structural stability at elevated temperatures even if the room temperature properties suffer in such a shift of emphasis. Measurements made in this program indicate that SmCos permanent magnets can be used up to 200°C with good long term stability. Although the reversible coefficient and irreversible loss are higher than for Alnico 5, these magnets when cycled to remove irreversible loss and temperature compensated to remove reversible changes can be used successfully in equipments and devices requiring constant flux over wide temperature ranges. BIBLI OGRAP HY 10. 11. 12. BIBLIOGRAPHY P. Weiss, J. Phys. 6, 661 - 90 (1908). For more detailed discussion see for example, C. Kittel, Rev. Modern Physics, 21, 541 (1949); E. P. Wohlfarth, Adv. Physics, 8, 87 (1959). J. Frenkel and J. Dorfman, Nature 126, 274 (1935). L. Neel, Ann. University, Grenoble 22, 299 (1946). A. Aharoni, J. Appl. Phys. , 30, 705 (1959). R. J. Parker and R. Studders, "Permanent Magnets and Their Application, " John Wiley 8: Sons, N. Y. (1962), Chapter 2. R. Street and]. Woolley, Proc. Phys. Soc. A 62, 562 (1949). Magnetization Changes for Cobalt -Rare Earth Magnet Alloys When Heated to 650°C, by D. L. Martin and M. Benz, General Electric T.I.S. Report, 71 -C- 186. F. E. Luborsky and R. J. Parker, General Electric T.I.S. Report, 66C252, (1966). Stability of Permanent Magnets, Bulletin (2), Permanent Magnet Association Sheffield, England, (1964). D. Hadfield, "Permanent Magnets and Magnetism," John Wiley 8: Sons, N.Y., (1962). Shur, Luzhinskaya and Shubina, Inst. Fiz. Metal. , Akad. Nauk SSSR, 20, 111, (1958). .47 MICHIGAN STRTE UNIV. LIBRQRIES lllllll lllllll lllllllllll "II "III lllllllllllllll 9 3 54 312 30062 55