Silicon Steel

What is Silicon Steel

Silicon steel is a soft magnetic alloy of iron and silicon, supplied as thin cold-rolled sheet or strip and stacked into laminated cores. Its purpose is narrow and demanding: to be magnetized and demagnetized millions of times per day by alternating current while wasting as little energy as possible. The two performance metrics that define a grade are core loss, the watts dissipated per kilogram under AC magnetization, and permeability, how readily the material reaches a given flux density. Everything in the metallurgy, the silicon level, the grain structure, the thickness, and the surface coating, serves those two numbers.

The alloy works because of two physics levers. First, silicon raises the electrical resistivity of iron, and higher resistivity throttles the eddy currents that circulate inside the laminations when flux changes, since eddy-current loss scales with the square of both frequency and sheet thickness and inversely with resistivity. Second, the sheet is rolled thin (commonly 0.18 to 0.65 mm) and coated with an electrically insulating film so that the core behaves as many isolated thin sheets rather than one solid block, again cutting eddy currents. Hysteresis loss, the energy lost reorienting magnetic domains each cycle, is reduced separately by purity, grain size, and crystallographic texture.

The industrial history begins in 1900, when Robert Hadfield in Sheffield demonstrated that adding silicon to iron sharply lowered magnetic loss, giving the first silicon electrical steels. In 1933 the American metallurgist Norman Goss developed grain-oriented silicon steel, in which a special cold-rolling and annealing sequence aligns the crystal grains so the easy magnetization axis runs along the rolling direction. Cold-rolled grain-oriented production scaled commercially through the 1940s and 1950s, and high-permeability Hi-B grades using aluminum-nitride grain-growth inhibitors arrived in the 1960s and 1970s. Later refinements, laser and mechanical domain refinement, thinner gauges, and high-silicon 6.5% sheet, continued to push loss downward.

In application scale, silicon steel is one of the most-produced engineered magnetic materials in the world, with global electrical-steel output in the tens of millions of tonnes per year. It sits inside the iron core of almost every power transformer, distribution transformer, industrial and appliance motor, generator, and electromagnetic actuator. China Baowu (Baosteel) is the single largest producer; Nippon Steel, JFE Steel, POSCO, Cleveland-Cliffs, ThyssenKrupp, ArcelorMittal, and voestalpine are the other major makers. The drive for higher motor efficiency standards and for electric-vehicle traction motors has made low-loss thin-gauge non-oriented steel one of the fastest-growing segments.

It is worth separating silicon steel from neighboring magnetic materials. It is a soft magnetic material, easily magnetized and demagnetized, the opposite of the permanent (hard) magnets used as field sources. Compared with amorphous metal ribbon, which has even lower loss but lower saturation and a poorer stacking factor, silicon steel offers a better balance of cost, saturation flux density, and manufacturability, which is why it remains the workhorse core material at line frequency. At much higher frequencies, soft ferrites and nanocrystalline ribbon take over, because silicon steel laminations become impractically thin and lossy above roughly a few kilohertz.

The product is sold mainly as cold-rolled coil and slit strip, then processed by the user into laminations: wound strip cores for some transformers, or sheared and stamped sheets stacked into the E-I, C, toroidal, or radial-segment geometries used in transformers and machines. Because the core is built from many thin insulated sheets rather than a solid block, the buyer is really purchasing a controlled combination of alloy chemistry, gauge, texture, and coating. The grade name on the purchase order, as the next chapters show, compresses most of that into a single code.

Grain-Oriented vs Non-Oriented

The single most important classification of silicon steel is grain-oriented versus non-oriented. The two are not interchangeable: they are optimized for opposite flux conditions, and using one where the other belongs costs efficiency and money. The distinction comes from the crystallographic texture, the statistical alignment of grain orientations within the sheet, which is controlled during cold rolling and annealing.

Grain-oriented electrical steel is processed through secondary recrystallization so that a very small fraction of {110}<001> Goss grains grow at the expense of all others, producing a sheet whose easy magnetization direction, the <001> cube edge, lies along the rolling direction. Magnetized along that axis it shows exceptionally low loss and high permeability, but its properties degrade sharply at angles away from rolling. This makes it ideal for transformer cores, where the laminations are wound or stacked so the flux always runs in the grain direction. It is almost always cold-rolled, hence the trade name CRGO. High-permeability Hi-B grades sharpen the Goss texture further for still lower loss.

Non-oriented electrical steel is annealed to a near-random in-plane grain distribution, so its magnetic behavior is approximately the same in every direction within the sheet plane. That isotropy is exactly what a rotating machine needs, because the flux in a motor or generator stator and rotor sweeps continuously around the iron rather than following one fixed line. Non-oriented grades are graded at 1.5 T, carry a wide range of silicon (low-silicon grades for cheap appliance motors, higher-silicon thin gauges for efficient and high-speed machines), and are supplied either fully-processed (ready to use) or semi-processed (the user performs a final anneal to develop properties).

A practical consequence for buyers: if a vendor offers a single grade for both a transformer and a motor, something is wrong. Transformer projects call out grain-oriented or Hi-B grades by their 1.7 T loss; motor projects call out non-oriented grades by their 1.5 T loss and often by magnetic polarization at a fixed field. The rest of this guide treats the two families separately wherever their numbers diverge.

Grades and Grade Naming

Silicon steel is one of the few materials where the grade name itself encodes the key specification. Several naming systems coexist, but they all communicate two things: how lossy the steel is and how thick it is. Learning to read them lets an engineer compare an EN grade, an AISI M-grade, and a JIS or GB grade without a conversion chart.

The EN system is the most transparent. A name like M270-35A breaks into M for magnetic material, 270 as one hundred times the guaranteed maximum specific total loss in W/kg (so 2.70 W/kg), 35 as one hundred times the nominal thickness in millimeters (0.35 mm), and a letter for the type: A for conventional non-oriented (EN 10106), D and E for semi-processed, S for conventional grain-oriented and P for high-permeability grain-oriented (EN 10107). Non-oriented loss is at 1.5 T; grain-oriented loss is at 1.7 T. Lower leading numbers always mean lower loss and, generally, higher price.

The AISI M-grade system, standardized in ASTM A876 for grain-oriented steel, labels grades M-2 through M-6 roughly by thickness and loss class, with M-2 the thinnest and lowest-loss and M-6 the thickest. Cleveland-Cliffs, the originator of the system, supplies M-3 at 0.23 mm, M-4 at 0.27 mm, M-5 at 0.30 mm, and M-6 at 0.35 mm, each carrying an ASTM core-loss type code such as 23G045, 27G051, 30G058, and 35G066. ASTM A677 covers the equivalent fully-processed non-oriented grades. The older designations are still common on transformer drawings, so engineers must map them to the EN or measured loss values.

The JIS and GB systems used across Asia put thickness first. In 50W470 (JIS C2552, mirrored by GB/T 2521), 50 is the thickness in hundredths of a millimeter (0.50 mm), W marks watt-loss-graded non-oriented steel, and 470 is one hundred times the maximum loss at 1.5 T 50 Hz (4.70 W/kg). Grain-oriented JIS grades under C2553 such as 23ZH85 encode 0.23 mm, the ZH high-permeability class, and a loss index. The GB/T 2521 family aligns closely with the JIS scheme, which is why a 50W470 coil from a Chinese, Japanese, or Korean mill is broadly comparable.

Magnetic and Physical Properties

Behind the grade name sit the physical properties that the metallurgy controls. The dominant lever is silicon content, which trades off three things at once: it cuts loss by raising resistivity, but it lowers saturation flux density and makes the steel harder and more brittle to roll. The table below shows how key properties shift with silicon level.

Silicon and resistivity. Each percent of silicon raises resistivity by roughly 12 to 13 microohm-cm, climbing from about 25 microohm-cm in near-pure iron to roughly 48 microohm-cm at 3% silicon and about 82 microohm-cm at 6.5% silicon. Since classical eddy-current loss is inversely proportional to resistivity, this is the primary reason to alloy with silicon. The same silicon also reduces the magnetocrystalline anisotropy and the magnetostriction of iron, which lowers hysteresis loss and audible transformer noise.

The saturation trade-off. Silicon dilutes the magnetic iron atoms, so saturation polarization falls from about 2.15 T for pure iron to roughly 2.03 T at 3% silicon and about 1.80 T near 6.5% silicon. A lower saturation means a designer must use more iron, or accept lower peak flux, to carry a given total flux. This is why motor designers who want maximum torque density sometimes choose lower-silicon, higher-saturation grades and accept higher loss, while transformer designers favor higher-silicon, lower-loss grades. Density also falls with silicon, from about 7.85 to 7.49 g/cm³, which matters when core mass is multiplied across large transformers.

The formability ceiling. Silicon makes iron hard and brittle. Above roughly 3.5% silicon the sheet becomes difficult to cold roll without cracking, which is why ordinary commercial grades stay at or below about 3.5%. The very-low-loss 6.5% silicon material, prized for its near-zero magnetostriction and excellent high-frequency behavior, cannot be conventionally rolled to that composition and is instead made by chemical vapor deposition siliconizing of a lower-silicon base sheet or by special rapid-solidification routes, which keeps it niche and expensive.

Grain structure and texture. For grain-oriented steel the magnetic prize is the sharp {110}<001> Goss texture; the closer the cube edges align to the rolling direction, the lower the loss along that axis. Hi-B grades use grain-growth inhibitors such as aluminum nitride to perfect this texture during secondary recrystallization. For non-oriented steel, larger and more uniform grains lower hysteresis loss, but very large grains raise eddy-current loss, so an optimum grain size exists for each gauge and frequency. Final magnetic properties are also sensitive to residual stress, which is why stress-relief annealing after cutting is so important.

Key Specification Parameters

A silicon steel datasheet lists more numbers than most buyers need. The parameters that actually drive selection are core loss, magnetic polarization at a fixed field, thickness, stacking factor, insulation coating, and a short list of mechanical and dimensional tolerances. Each is explained below, with the grain-oriented and non-oriented conventions noted where they differ.

Core loss (iron loss) is the headline specification, the power dissipated per kilogram under sinusoidal AC magnetization, in W/kg. It is always quoted at a stated peak flux density and frequency: P1.5/50 (also written W15/50) for non-oriented steel at 1.5 T and 50 Hz, and P1.7/50 (W17/50) for grain-oriented steel at 1.7 T and 50 Hz. Core loss sums three parts, hysteresis loss (proportional to frequency), classical eddy-current loss (proportional to frequency squared and thickness squared), and anomalous or excess loss; this is why thinner gauges and higher silicon win at higher frequency. It is measured on Epstein frame strips per IEC 60404-2 or ASTM A343, or by single-sheet tester per IEC 60404-3.

Magnetic polarization at fixed field describes permeability indirectly and is critical for motor steel. Non-oriented grades quote the minimum polarization J at field strengths of 2500, 5000, and 10000 A/m, for example J at 5000 A/m of roughly 1.60 to 1.65 T for a typical grade. A higher value means the steel reaches working flux with less magnetizing current, lowering copper loss in a motor. Grain-oriented grades instead often quote the induction at 800 A/m (B800), with Hi-B grades reaching about 1.88 to 1.95 T versus around 1.80 to 1.85 T for conventional grades.

Thickness (gauge) sets the eddy-current ceiling and is part of the grade name. Standard gauges are 0.50, 0.65, and 0.35 mm for general motor steel, 0.20 to 0.35 mm for high-efficiency and high-speed motors, and 0.18 to 0.30 mm for grain-oriented transformer steel. Thinner is lower-loss but more expensive per tonne, harder to handle, and gives a lower stacking factor.

Stacking factor is the ratio of active iron cross-section to physical stack cross-section, typically 0.95 to 0.97 for silicon steel. It captures the volume lost to coatings and to thickness and flatness variation. A high stacking factor lets the designer reach target flux in a smaller core.

Insulation coating provides interlaminar resistance, corrosion protection, and stamping lubrication, classified by ASTM A976 and IEC 60404-1-1 into types such as:

• C-2 (forsterite glass film): the magnesium-silicate layer formed on grain-oriented steel during high-temperature annealing; survives stress-relief anneal.
• C-3 / C-4 (organic / semi-organic): common on non-oriented steel; good punchability but not weldable or anneal-stable in all cases.
• C-5 (inorganic phosphate tension coating): CARLITE-type top coat on grain-oriented steel that adds tensile stress to lower loss and noise; anneal-stable.
• C-6 (inorganic-organic): balances weldability, punchability, and anneal tolerance for motor laminations.

Anisotropy matters when matching steel to the build. Grain-oriented steel is deliberately anisotropic: its loss and permeability are far better along the rolling direction than across it, so transformer laminations must be cut so the flux follows the grain, and mitered corner joints are used to keep the flux near the easy axis at the joints. Non-oriented steel is specified with low anisotropy, but even good grades show a few percent difference between the rolling and transverse directions, which designers of high-efficiency motors account for when orienting stamped segments.

Mechanical and dimensional specs round out the sheet: tensile and yield strength (higher-strength non-oriented grades exist for high-speed rotor rims that must resist centrifugal stress), hardness, edge camber, flatness (waviness), burr height after slitting, and width and length tolerance. For semi-processed grades the buyer must also confirm the required final-anneal cycle to develop the rated properties. Magnetostriction, the tiny dimensional change of the steel under magnetization, is a further consideration for transformers because it is the root cause of the audible 100 or 120 Hz hum; lower-magnetostriction Hi-B grades and tension coatings are specified where acoustic noise limits apply.

Selection Decision Factors

To turn this knowledge into a specific coil, follow the decision sequence below. Most silicon steel mistakes come not from one wrong number but from deciding gauge and loss class before fixing the application and frequency. These eight steps work as a fixed RFQ template.

1. Application and flux path: First decide transformer (one-directional flux, choose grain-oriented or Hi-B) versus rotating machine (multi-directional flux, choose non-oriented). This single choice fixes the whole grade family.
2. Operating frequency: 50 or 60 Hz line frequency tolerates 0.35 to 0.50 mm gauges; high-speed motors, EV traction, and converters running at hundreds of hertz or more demand 0.20 to 0.35 mm or even thinner to control eddy-current loss.
3. Loss budget versus saturation: Set the target core loss (W/kg at the grading point) against the required peak flux. Lower-loss grades trade away saturation, so confirm the chosen grade reaches your working flux density with acceptable magnetizing current.
4. Grade and thickness: Translate the loss and flux targets into a grade name (for example M-3 / M120-23S for a distribution transformer, M270-35A or 50W470 for a general motor) and confirm the gauge is stocked in the width you need.
5. Processing state: Choose fully-processed (ready to stack, no anneal) for small high-volume parts, or semi-processed if you can run a final anneal and want lower material cost. For large laminations and wound cores, plan a stress-relief anneal after cutting.
6. Coating class: Match the coating (C-class) to the build process: weldable, bondable, punch-friendly, or anneal-stable. Confirm compatibility with transformer oil if relevant and with the assembly method (welding, interlocking, gluing).
7. Standards and certification: Specify the governing standard (EN 10106 / EN 10107, ASTM A677 / A876, JIS C2552 / C2553, GB/T 2521, IEC 60404 series), the mill test certificate with Epstein or single-sheet loss data, and any RoHS or efficiency-regulation compliance.
8. Total cost of ownership: Compare purchase price per tonne against lifetime energy loss. In a transformer that runs for decades, a lower-loss grade that costs more upfront often pays back many times over in saved no-load loss; in a low-duty appliance motor, the cheapest adequate grade usually wins.

A second overlooked factor is edge quality and build effect. Cutting, punching, and shearing introduce plastic strain and residual stress near the cut edge that degrades magnetic properties and raises core loss, an effect that grows as parts get smaller and edge length per unit area increases. The gap between the catalog loss of an Epstein strip and the loss of a finished core is captured by the building factor, often 1.1 to 1.3 for transformers and higher for stamped motor stacks. Stress-relief annealing, typically a controlled cycle around 750 to 820 degrees Celsius in a non-oxidizing or slightly decarburizing atmosphere, recovers much of the lost performance, so plan for it on wound cores and large laminations. Fully-processed grades can usually be run without annealing for small high-volume parts, while semi-processed grades require a user final anneal to develop their rated properties.

One last commonly overlooked dimension is supply and serviceability: gauge and width availability, minimum order quantity, slitting and edge-quality consistency from coil to coil, lead time, and second-source qualification. A grade that is theoretically optimal but only single-sourced with a long lead time is a production risk. Major mills such as Baowu (Baosteel), Nippon Steel, JFE Steel, POSCO, Cleveland-Cliffs, ThyssenKrupp, ArcelorMittal, and voestalpine maintain broad grade ranges and global distribution, and many regional service centers slit and stress-relief anneal to order, which makes dual-sourcing on common grades practical.