Cullity B.D. Introduction to Magnetic Materials. Second Edition (2008)

Fig. 13.9 Torque curves for a disk of cold-rolled iron at different values of the applied field. [H. J. Williams, Rev. Sci. Instrum., 8 (1937) p. 56.]

the L ¼ 0 axis. In Fig. 13.9 the torque in a field of 50 Oe is practically constant. At 170 Oe the overall anisotropy of the specimen has become evident, but there is still rotational hysteresis because the net area between the curve and axis is still negative. At 1500 Oe, however, the curve is symmetrical about the axis, no net torque is required to rotate the specimen, and Wr is zero. The specimen is then saturated and its Ms vector rotates reversibly with the field; the torque curve is simply an expression of the anisotropy, just as the curve of Fig. 7.15, for example.

There is a real difference between the alternating hysteresis loss Wh and the rotational hysteresis loss Wr. The value of Wh steadily increases with field until the specimen is saturated, but Wr goes through a maximum and becomes zero at saturation. Figure 13.10 illustrates this point for silicon steel. The reason for the difference is that a specimen rotated in a saturating field is always saturated, but a specimen subjected to alternating magnetization passes through the demagnetized state twice in each cycle of magnetization. The behavior of the rotational hysteresis as a function of the applied field can be characterized by a quantity called the rotational hysteresis integral, defined by I. S Jacobs and C. P. Bean [J. Appl. Phys., 28 (1957) p. 467] as

Although alternating and rotational hysteresis losses are conceptually different and separately measurable, both are due to the same fundamental processes, irreversible wall movements and irreversible rotations.

The conditions of the Epstein test correspond quite closely to the conditions in a working transformer. The major differences are that in a working transformer there are mechanical stresses acting on the magnetic laminations as a result of the need to hold everything firmly

452 SOFT MAGNETIC MATERIALS

Fig. 13.10 Dependence on 4pM of the rotational hysteresis loss Wr and the alternating hysteresis loss Wh in the rolling direction for grain-oriented 3.13% silicon steel. [F. Brailsford, J. Inst. Elect. Engrs, 84 (1939) p. 399.]

in place, and the corner joints are designed so that there are no gaps between the laminations in the transformer legs. The transformer losses predicted from the Epstein test results are always too low by a factor called the build factor or the construction factor, which is known only from experience.

Measurements of rotational loss are much less useful in practice, because in a motor or generator different parts of the steel laminations are subjected to different mixtures of alternating and rotational magnetization; furthermore the alternating flux is usually not sinusoidal and the rotating flux is not rotating at a constant rate. Therefore it is not possible to specify a set of test conditions that are generally applicable.

13.4ELECTRICAL STEEL

Three kinds of materials are used for the cores of most electrical machines: (1) low-carbon steel, (2) nonoriented silicon steel, and (3) grain-oriented silicon steel. These materials are usually called electrical steel or transformer steel; sometimes all of them are referred to simply as “iron,” regardless of composition. The magnetic quality and the price increase in the order listed.

Before we examine these three grades, some general remarks on texture (preferred orientation) are in order. This subject was introduced in Section 7.8, where we saw that a crystallographic texture could result in a polycrystalline specimen having an easy axis of magnetization. In an iron or iron-silicon crystal the easy axis is k100l and the hard axis is k111l. In a transformer, the magnetization direction is always parallel to the long edges of the laminations. If the sheet can be made with an fhklg k100l texture, there will be an easy k100l direction in the direction of magnetization, which will increase

permeability and decrease losses. This kind of texture can be made in silicon steel. The actual texture is f110g k001l, and such material is called “grain-oriented” steel.

In cores for rotating machines, on the other hand, the field is in the plane of the sheet, but the angle between the field and rolling direction is variable. Here there is no point in having the easy direction in the rolling direction, and a satisfactory texture would be f100g kuvwl, which keeps the hard k111l direction out of the plane of the sheet. A k100l fiber texture would be even better, i.e., a texture in which all grains had a k100l direction normal to the sheet surface and all possible rotational positions about this normal, because the sheet would then be isotropic in its own plane. A f100g k100l texture, known as the cube texture, has been made on a trial basis, but has never been produced in quantity in iron-silicon alloys.

Electrical steel for cores must be in the fully recrystallized, magnetically soft condition. There are two kinds of recrystallization, primary and secondary. These will be distinguished later on. It is enough to note here that magnetically desirable textures have been achieved only by secondary recrystallization and that this process requires a rather high annealing temperature. After these preliminaries we can turn to the materials themselves.

13.4.1Low-Carbon Steel

This was the original core material for transformers, motors, and generators, but it is limited today mainly to the cores of small motors where energy loss is not a major consideration. [An example is the household vacuum cleaner motor, which typically runs for short times and is well-cooled by the air flow. These motors are advertised on the basis of power consumption (“powerful 600 W motor”), not on the basis of power output. High power losses and low efficiency thus are turned into a sales advantage.]

This is basically the same material as that used for automobile bodies, washing machines, refrigerators, and the like, with the carbon content reduced as low as practical in normal steelmaking operations. Low-carbon sheet steel is one of the cheapest steel products made and is produced in large tonnages; the portion used for motor cores forms a small fraction of the total.

The carbon content is about 0.03 wt%. The core loss at 15 kilogauss, 60 Hz, 0.020 inch (0.5 mm) thickness is about 5–7 W/lb or 3 W/kg, which is nearly 10 times the loss for grain-oriented silicon steel. But core loss is of small importance to the manufacturer of small motors, since low cost rather than efficiency is the chief object. The market is very large. A typical home in an advanced country contains dozens of small electric motors, in large and small appliances, power tools, toys, clocks, etc., plus many more in the family car(s), e.g., heater, engine fan, seat adjusters, power roof, power windows, and retractable radio antenna.

The normal (primary) recrystallization texture of sheet steel is complex, not very strong, and magnetically unattractive. Attempts to develop strong and useful textures have not been successful. Strong textures usually result from secondary recrystallization, which requires high temperatures, and at temperatures no higher than 9108C, low-carbon steel undergoes a phase transformation. Any texture existing below this temperature is effectively destroyed by heating, and any texture existing above this temperature is destroyed by cooling.

Low-carbon steel has another class of magnetic applications which require only constant or slowly changing magnetization. These are regarded as dc applications, and they include lifting and holding magnets, laboratory electromagnets, and beam-bending magnets in high-energy particle accelerators (although for the latter use, superconducting magnets are now used). For dc uses, the steel need not be laminated, and castings may be used.

g ! d transformation is lowered until the two meet at about 2.5% Si, forming a closed “gamma loop.” As a result, an alloy containing more than about 2.5% Si is body-centered cubic at all temperatures up to the melting point. This means that (1) such an alloy may be recrystallized at any temperature without concern for phase changes, and (2) single crystals of such an alloy can be made by slow solidification from the liquid or by recrystallization at any desired temperature. These operations are difficult in the case of pure iron or at lower Si contents, because of the g ! a transformation on cooling to room temperature. The a solid solution of silicon in iron is often called silicon ferrite by metallurgists, and the magnetic Si–Fe alloys are often called silicon iron.

The preceding remarks apply to pure Fe–Si alloys. The presence of carbon widens the (a þ g) region, and only 0.07% C is enough to shift the nose of the gamma loop over to about 6% Si. In practice, the carbon content of silicon steel varies from about 0.03% (nonoriented) to less than 0.01% (oriented), and it would be reduced to even lower levels if it were economically feasible to do so. Iron carbide and nitride precipitates degrade the magnetic properties by interfering with wall motion, and the slow precipitation of carbides during service, called “aging,” can cause a substantial increase in core losses.

The addition of silicon to iron has the following beneficial effects on magnetic properties:

1.The electrical resistivity increases, causing a marked reduction in eddy currents and therefore in losses.

2.The magnetocrystalline anisotropy decreases, causing an increase in permeability.

3.The magnetostriction decreases, leading to smaller dimensional changes with magnetization and demagnetization, and to lower stress-sensitivity of magnetic properties.

On the debit side, silicon additions decrease the saturation induction and tend to make the alloy brittle, so that it becomes difficult to roll into sheet when the silicon content is much higher than 3%. Figure 13.12 shows the effect of silicon on these various properties; the percentage elongation in the tensile test is a measure of ductility.

Fig. 13.12 Data on iron-silicon alloys: magnetocrystalline anisotropy constant K1, saturation induction Bs, electrical resistivity r and percentage elongation for nonoriented, polycrystalline sheet, 0.0185 inch or 0.47 mm thick.

456 SOFT MAGNETIC MATERIALS

Core loss increases rapidly with the maximum induction reached in the cycle, and is normally not measured above 17 kG (1.7 T). Core loss also increases with sheet thickness, and there has been a clear trend to use thinner gages as a way to decreases core loss.

Nonoriented silicon steel sheet is made by hot rolling to near-final thickness, pickling in acid to remove the oxide scale, and cold rolling to final thickness, which gives the best surface finish and sheet flatness. It may be sold in this state, called semiprocessed, which is good for punching finished shapes with little deformation but requires a subsequent annealing treatment for best magnetic properties. Or it may be annealed at the steel mill to a fully processed state. The final annealing may be done in a decarburizing atmosphere to lower the carbon content.

The use of other alloying elements instead of, or in addition to, silicon has been investigated at various times and places. The most obvious candidate is aluminum, which affects the magnetic properties of iron very much as silicon does. The presence of aluminum oxide particles in aluminum-iron alloys increases the rate of wear of the punching dies used to stamp out motor and transformer laminations, and aluminum appears to offer no compensating advantages over silicon.

13.4.3Grain-Oriented Silicon Steel

This material was developed by the American metallurgist Norman Goss in 1933. He discovered that cold rolling a silicon iron with intermediate anneals, plus a final hightemperature anneal, produced sheet with much better magnetic properties in the rolling direction than hot-rolled sheet. This improvement was due to a magnetically favorable texture produced by secondary recrystallization during high-temperature annealing. Grain-oriented sheet went into commercial production about 1945, and its properties have since been continually improved. Material with a core loss about 0.5 W/lb or 1W/kg (at Bmax ¼ 17 kG or 1.7 T, power frequency, 0.009 inch or 0.23 mm thickness) is now routinely produced, and such sheet is the standard core material for large transformers.

Secondary recrystallization differs markedly from primary. Primary recrystallization occurs when a cold-worked metal is heated to a temperature at which new, strain-free grains can nucleate and grow throughout the cold-worked matrix. It is truly a recrystallization. Secondary recrystallization, on the other hand, is a particular kind of grain growth and is sometimes called discontinuous, exaggerated, or abnormal grain growth. It occurs in some, but not all, materials when (a) normal grain growth is inhibited, and (b) the material is annealed, usually for a long time, at a temperature much higher than that required for primary recrystallization. The result is the preferential growth of a relatively few grains at the expense of the others, leading to extremely large grains. The grains are no longer microscopic in size, with grain diameters of some tens of micrometers, but they are now visible to the naked eye, having diameters of the order of millimeters or even centimeters. The grain size has therefore increased by a factor of several hundred (Fig. 13.13), and such grains extend through the thickness of the sheet.

In order for secondary recrystallization to occur, something must prevent normal grain growth. This is usually a second-phase precipitate at the primary grain boundaries, which either disappears by solid state solution at high temperatures, or is removed by a chemical reaction at the sample surface.

The feature of secondary recrystallization that is of interest here is the fact that a secondaryrecrystallization texture is usually quite different from the primary-recrystallization texture. In silicon steel the primary texture is weak and complex, whereas the secondary texture

Fig. 13.13 Partial secondary recrystallization in silicon iron. A few large secondary grains are growing in a matrix of primary grains during a 9008C anneal. About actual size. [C. G. Dunn,

Cold Working of Metals, ASM (1949) p. 113.]

is a quite sharp single-component texture, namely, f110g k100l. This texture is illustrated in Fig. 13.14. In Fig. 13.14a is shown the stereographic projection of the principal directions in a single crystal, with the plane of the sheet taken as the projection plane; (011) is parallel to the sheet surface and [100] parallel to the rolling direction. The sketch in Fig. 13.14b of the orientation of the unit cell of each grain relative to the sheet shows why this is called the cube-on-edge texture. The schematic f100g pole figure in Fig. 13.14c shows where the k100l directions are concentrated for a large number of grains in polycrystalline sheet; the texture is sharper, i.e., has less scatter about the ideal orientation shown in Fig. 13.14a, the smaller the areas of the high-density regions on the pole figure. Pole figures are the result of X-ray diffraction measurements and are the most direct description of a texture. Useful information about texture in magnetic materials can also be obtained from torque curves, subject to the limitations mentioned in Section 7.8; Fig. 13.15 shows how closely the torque curve for a grain-oriented polycrystalline sheet can approach that of a single crystal.

The details of the manufacturing process for oriented silicon steel are proprietary, and presumably are not identical for all manufacturers. However, the important steps appear to be the following:

1.Heavy cold-rolling (more than 50% reduction in area), to final thickness, which may actually be done above room temperature.

2.Recrystallization and decarburization at about 8008C in moist hydrogen.

3.Annealing in dry hydrogen at 1100–12008C to produce the secondary recrystallization texture.

The particles inhibiting normal grain growth were believed to be MnS in the original grain-oriented material. Other compounds, such as AlN, have been used more recently. It is important that the particles be removed in the final heat treatment, since if they remained in the alloy they would act as domain wall pinning sites and increase the coercive field.

458 SOFT MAGNETIC MATERIALS

Fig. 13.14 Single-crystal projections (top), unit-cell orientations (center), and {100} pole figures (below). RD ¼ rolling direction; TD ¼ transverse direction.

Treatments that give the best crystallographic texture tend to produce very large grain sizes, and large grains tend to have relatively few domain walls. As seen previously, closely spaced domain walls are desirable to reduce eddy-current losses. Therefore various methods have been employed to provide nucleation sites for domain walls: mechanical

Fig. 13.15 Torque curves of a single crystal in the {110} plane and of a disk cut from a sheet of grain-oriented silicon steel having a strong cube-on-edge, {110} k100l texture. [C. D. Graham, Magnetism and Metallurgy, Volume 2, Academic Press (1969).]

scratches, chemical etches, spark erosion, and laser scribing. The best and most practical procedure seems to be laser scribing to produce a series of parallel lines arrayed perpendicular to the rolling direction of the sheet, spaced a few millimeters apart.

The cube-on-edge texture makes the easy k100l directions in all grains almost parallel to one another and to the rolling direction of the sheet. The magnetic properties in this direction are therefore excellent, and all properties quoted for grain-oriented sheet refer to this direction. Epstein strips for core-loss measurements of grain-oriented material are all cut parallel to the rolling direction of the sheet. In nonoriented sheet, on the other hand, properties are measured as an average for the rolling and transverse directions, and the Epstein square is built up of strips cut half in one direction and half in the other. The magnetic properties in the other directions in grain-oriented sheet are drastically inferior to those in the rolling direction; for example, the permeability at 15 kG (1.5 T) in the transverse direction is less than 2% of the permeability in the rolling direction. The reason is not far to seek: as Fig. 13.14a shows, both the hard k111l direction and the medium-hard k110l direction lie in the plane of the sheet, the former at 558 to the rolling direction and the latter at 908. The ductility also differs markedly: the percent elongation in tension is only 8% in the rolling direction compared to 28% in the transverse.

Because the magnetic properties are so directional, care must be taken in the construction of transformers to make the rolling direction of the sheet parallel, as far as possible, to the direction of flux travel. This is no problem in the largest transformers, because each leg is formed of separate pieces of sheet. This kind of construction is illustrated in Fig. 13.16a, where the striations indicate the rolling direction of each piece. (The largest transformers are called power transformers. They are located at the generating plant and at large substations, and have cores weighing up to 250 tons, limited mainly by considerations of transportation. Distribution transformers are much smaller, and are seen mounted on poles in residential neighborhoods.) In smaller transformers this construction is not economical; one solution is to make E–I laminations, as in Fig. 13.16b, where the core is made of two sets of sheets, cut in the form of these letters. Flux travel is in the easy direction except in the top leg, where it is in the transverse direction of the sheet; this leg may be widened to decrease the flux density and thus compensate for the lower permeability in

460 SOFT MAGNETIC MATERIALS

Fig. 13.16 Transformer cores. Open and closed circles represent primary and secondary windings; parallel lines denote rolling direction of sheet material.

this direction. (Both primary and secondary windings are placed on the central leg. These are only schematically indicated; the windings actually occupy most of the space between the legs.) Still another solution is to roll up a long length of strip on itself to form the tapewound or strip-wound core shown in Fig. 13.16c, with windings (not shown) placed on opposite sides. Here the entire flux path is in the easy (rolling) direction. The tapewound design presents problems in assembly: the magnetic strip must be wound through the prewound coils, or the coils must be wound on the prewound core, or the core must be cut in half to install the prewound coils.

The magnetic properties of sheet steel are degraded to some extent by mechanical deformation, so usually the sheets are given a stress-relief anneal at about 8008C in a protective atmosphere after any shearing or punching operations. During actual construction of a transformer, care must be exercised so minimize strains introduced into the sheets by handling and assembly.

ASTM International (formerly the American Society for Testing and Materials) has established a standard specification system for electrical steels. It is based on a minimum core loss value, and does not explicitly define the alloy content. Each grade of steel is identified by a number that consists of two digits, followed by a single letter, followed by three digits, such as 35H056. The first two digits specify the sheet thickness in units of 0.01 mm, so 35 means 0.35 mm thick material. The letter specifies a class of material, including the magnetic flux density at which the core loss is measured. The final three digits specify the maximum core loss, in watt/pound (with a decimal point added after the first digit), at 60 Hz. Only a limited range of thicknesses and loss values are regarded as standard, and for these a table gives equivalent thicknesses in inches, loss in watt/kg, and also loss at 50 Hz. The International Electrotechnical Commission (IEC) publishes a similar system as an international standard. Table 13.2 gives some representative values of core loss in electrical steels.

13.4.4Six Percent Silicon Steel

The upper limit of silicon content in steel produced by normal production techniques is about 3.25%. Higher silicon contents make the steel too brittle to be cold-rolled into sheet. However, a higher silicon content would increase electrical resistivity and decrease magnetic anisotropy, at the cost of somewhat lower magnetic saturation. Various methods have been investigated to produce silicon contents above 3.25%; usually 6%