- Email: [email protected]
Delta ferrite is ubiquitous in type 304 stainless steel: Consequences for magnetic characterization
Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials journal homepage: www.elsevier.com/locate/jmmm
Research articles
Delta ferrite is ubiquitous in type 304 stainless steel: Consequences for magnetic characterization C.D. Graham a,⇑, B.E. Lorenz b a b
Dept. of Materials Science, Univ. of Pennsylvania, Philadelphia PA, United States Dept. of Electrical Engineering, Widener Univ., Chester PA, United States
a r t i c l e
i n f o
Article history: Received 12 January 2018 Received in revised form 27 February 2018 Accepted 28 February 2018 Available online 1 March 2018 Keywords: 304 Stainless steel Delta ferrite Magnetic permeability
a b s t r a c t Using a vibrating-sample magnetometer with a maximum field of 20.5 kOe, we have measured over 50 samples of annealed 304 stainless steel, which is usually considered to be non-magnetic. In almost every case, we observe the presence of a small, usually less than 0.01, fraction of a ferromagnetic phase, which we believe to be equilibrium bcc delta ferrite. The consequences of this observation for the measurement and specification of the magnetic properties of annealed 304 stainless are discussed. Our measurements also establish the most likely value for the magnetic permeability of the fcc austenitic phase in 304 stainless steel austenite as 1.0033 ± 0.0003. Ó 2018 Elsevier B.V. All rights reserved.
1. Introduction According to standard reference sources, type 304 stainless steel in the annealed state is non-magnetic [1,2]. This actually means that it is paramagnetic, with magnetization M proportional to field H. The magnetic permeability m = B/H is variously given, usually either as 1.021 or in the range 1.002–1.004 [3]. We will return to this discrepancy later. Type 304 is considered to be paramagnetic because it is a single phase face-centered cubic (fcc) structure, known to metallurgists as austenite. The literature on welding of 304 stainless steel, however, is very much concerned with the presence of delta ferrite (d-ferrite) in the weld [4,5,6]. The use of the label ‘‘delta” ferrite implies this is an equilibrium body-centered cubic (bcc) phase that forms during solidification, analogous to the delta phase of pure iron that exists from the melting point of 1532 °C down to 1493 °C. Delta ferrite is to be distinguished from the bcc (or slightly tetragonal) phase that forms by a shear transformation when 304 stainless is plastically deformed below 80 °C [7]. This phase is known as martensite, and is usually labelled a’. The martensite in 304 stainless steel is often described as stress- or strain- or deformation-induced martensite. Deformation-induced martensite must have the same chemical composition as the austenite from which it forms, since it is produced by a diffusionless shear transformation. Delta ferrite, by contrast, is a thermodynamically stable phase that exists in
⇑ Corresponding author. E-mail address: [email protected] (C.D. Graham). https://doi.org/10.1016/j.jmmm.2018.02.092 0304-8853/Ó 2018 Elsevier B.V. All rights reserved.
equilibrium with austenite, and so has a different composition from the austenite. How much different is uncertain. Both martensite and delta ferrite are ferromagnetic, with saturation magnetization on the order of 100 emu/g or 10,000 gauss (1 T). We will return also to this subject later. Some delta-ferrite in a 304 stainless weld is regarded as desirable, to reduce the likelihood of crack formation during cooling. Therefore welding rod for 304 stainless is alloyed to contain a substantial content of delta ferrite. The presence of delta-ferrite apparently does not significantly affect the mechanical or corrosion-resistant properties of the alloy. We have found some suggestions in the literature that commercial 304 stainless steel may contain some delta ferrite in the annealed state. A publication from the British Stainless Steel Association [8] notes that Grades 304, 321, and 316 have ‘balanced’ compositions to enable them to be readily weldable. This is achieved by ensuring that in their normal annealed (softened) condition they contain a few percent of delta ferrite. The web site of Atlas Specialty Metals (Australia) [9] says It is common for wrought austenitic stainless steels to contain a very small amount of ferrite, but this is not sufficient to significantly affect magnetic performance except in very critical applications. The Metals Handbook volume on Metallography [10] includes a micrograph of annealed 302 stainless steel (the composition limits of 302 and 304 stainless overlap) with scattered circular patches of a second phase identified as ‘‘ferrite.” This is clearly an equilibrium phase, and not martensite. The ferrite appears to be present to about 1%, in the form of roughly spherical regions about 10 mm in diameter. A very recent paper from Iran [11] states as an accepted fact
16
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
that austenitic stainless steels contain delta ferrite and that it is regarded as undesirable; the paper investigates possible treatments to reduce or remove delta ferrite. 2. Experiment Over the course of several years, while investigating the formation of deformation-induced martensite and its reversion to austenite, we have measured about 50 samples of annealed and as-received (presumably annealed) 304 stainless steel, using a vibrating-sample magnetometer (VSM) [12] with a maximum applied field of 20.5 kOe. The samples were either disks about 6 mm in diameter and about 1 mm thick, magnetized along a diameter, or wires 6 mm long with a range of diameters up to 2 mm, magnetized along the length. The VSM was calibrated for disk samples with a disk of polycrystalline high-purity nickel, 0.25 inch (6.35 mm) in diameter and 0.050 inch (1.27 mm) thick. The wire sample calibration was a 6 mm long by 0.65 mm diameter highpurity polycrystalline iron wire. The samples came from a variety of sources. All were either asreceived and presumably annealed, or had been annealed at a temperature of at least 800 °C. Prior to annealing most of the samples had been deformed by varying amounts, by various methods, and at various temperatures. Some measurements were made on the same sample after different treatments. A few samples were type 304L or 304LN, indicating lower carbon, or lower carbon and higher nitrogen, than standard grade 304. One wire sample was type 302. In most cases the exact chemical composition is not known, and it may be important to note that the composition limits on 304 (and other) stainless steel are unusually wide. We make no claim that our samples are representative of anything; they are however, commercially-produced 304 stainless from several different sources, given a variety of deformation and annealing treatments. We report magnetization in emu/g (r) rather than emu/cm3 (M), since for small samples mass can be determined more accurately than volume. In almost every one of our 50 cases, the VSM result is like that in Fig. 1. We interpret this figure as showing the presence of a ferromagnetic phase (bcc delta ferrite), whose contribution to the magnetization of the sample is given by the intercept of the linear high-field data on the + or vertical axis, and a paramagnetic phase, fcc austenite, whose magnetic mass susceptibility is given by the slope of the linear high-field data. See Fig. 2. For the sample in Figs. 1 and 2, the magnetization resulting from the ferromagnetic phase gives a sample magnetization of 0.77 emu/g, and the mass susceptibility of the austenite is 33 10 6 emu/g-Oe. To convert this value to magnetic permeability (B/H), multiply by the density (8.0 g/cm3) and by 4p, which is very nearly a factor of 100, and add 1, to give permeability m = 1.0033.
Fig. 1. Typical VSM hysteresis loop of annealed 304 stainless steel.
Fig. 2. Data of Fig. 1 replotted to show paramagnetic component (solid line) and ferromagnetic component (data points). Solid line is a linear fit to the high-field data. The slope of this line gives the mass susceptibility of austenite. The points are the data points of Fig. 1 with the paramagnetic component subtracted. Note that in this sample, the paramagnetic and ferromagnetic magnetizations are nearly equal at a field of 20 kOe.
The values of ferromagnetic magnetization r in our samples ranged from 0 to 2 emu/g, in an apparently random way. See Fig. 3. The average value for 50 samples was 0.6 emu/g with a standard deviation of 0.65. A few values above 2 emu/g were eliminated from the data set on the grounds that the samples may have contained deformation-induced ferromagnetic martensite. It is not possible to distinguish martensite from delta ferrite simply on the basis of magnetic measurements. The value of the saturation magnetization of pure delta ferrite is unknown. Presumably it is similar to that of deformation-induced martensite, although (as noted above) the chemical compositions of the two phases are not expected to be identical. We are aware of just one value for the saturation magnetization of martensite in 304 stainless steel: 160 emu/g by Angel, dating to 1954 [7] There is also an interesting result from 1988 [13] in which a thin film was sputtered from a commercial 304 target to produce a fully bcc structure with a magnetization of 136 emu/g. The magnetization value is likely to be somewhat composition dependent, and, as noted above, the composition limits on type 304 stainless are unusually broad. It seems reasonable to take a value of 150 ± 25 emu/g for the saturation magnetization of delta ferrite. On that basis, a measured sample magnetization of 1 emu/g would imply about 0.7% delta ferrite content.
Fig. 3. Mass susceptibility vs. ferromagnetic magnetization r for 50 samples of Type 304 stainless steel. Susceptibility is generally near 30 10 6 emu/g-Oe and is uncorrelated with the magnetization.
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
Fig. 4. VSM data for a sample of annealed 304L stainless steel containing only paramagnetic austenite. The mass susceptibility is 29 10 6 emu/g-Oe. The composition of this material is known: C 0.017, N 0.08, Cr 18.94, Ni 8.85, Mn 1.32, Si 0.34, Cu 0.27, Mo 0.23, Ti 0.002, Fe bal, (all wt%).
17
Fig. 6. Low field hysteresis loop of an annealed 304 stainless steel. From the highfield data for this sample, the ferromagnetic magnetization is 0.77 emu/g and the mass susceptibility is 32 10 6 emu/g-Oe.
2.1. Low-field results
fields below 1 kOe on most of our samples. We did make more extensive measurements on some samples, and a typical result is shown in Fig. 6. Coercive fields were generally between 50 and 150 Oe, and remanent magnetizations much less than the extrapolated saturation magnetization, both with considerable variation from sample to sample. If the microstructure of our samples is like that shown in Ref. 10, it can be described as widely separated spheres of ferromagnetic delta ferrite in a paramagnetic austenite matrix. The ferromagnetic spheres are far enough apart to have minimal interaction, so each ferromagnetic sphere will experience only its own demagnetizing field. The demagnetizing field of a sphere is 4p M and if r is 140 emu/g, M is 1120 emu/cm3 and the demagne3 tizing field is about 4.6 kOe. This is in reasonable agreement with the observation that reaching a linear slope by saturating the ferromagnetic magnetization requires a field of 3–5 kOe. Coercive field values near 100 Oe are unusually high for an annealed metal. One possible explanation lies in the difference in thermal expansion coefficients between austenite and delta ferrite, which can lead to significant stresses on cooling of a two-phase sample. Differences in composition and annealing treatments, including cooling rates, could help to account for the variability in coercivity.
Since our primary interest was in quantities determined from high-field measurements, we took relatively few data points at
3. Conclusions
The only samples that showed no delta ferrite were obtained from Slovenia [14], and were made from material provided by an Italian supplier. Fig. 4 shows the VSM data from one of these samples. In contrast to the scatter in values of r, the magnetic susceptibility values given by the slope of the high-field linear magnetization data clustered strongly near 33 10 6 emu/g-Oe, with a few outlier points at lower and higher values. Fig. 5 shows the distribution of values. There are several possible explanations for the range of susceptibility values. There is likely to be a chemical composition dependence of the susceptibility; our VSM calibration standards were not perfect matches to the size and shapes of our samples; and there was also a scatter of ±2 or 3 10 6 emu/g-Oe in values for a single sample. We have found two published VSM hysteresis loops of annealed 304 stainless steel [14,15]. They are both consistent with our results, showing values of 0.50 and 0.79 emu/g for r and 29 and 26 10 6 emu/g-Oe for mass susceptibility.
1. It is possible, even likely, that annealed 304 stainless steel will contain up to about 2% of strongly ferromagnetic delta ferrite. 2. This accounts for the variation in reported permeability value for austenitic stainless steels. The presence of a ferromagnetic phase will lead to a measured permeability higher than the paramagnetic permeability of fcc austenite, and to a permeability that will vary with the applied field. 3. The most likely value for the susceptibility of the fcc austenite phase in 304 stainless steel is 33 ± 3 10 6 emu/g-Oe or 265 ± 25 emu/cm3-Oe, corresponding to a magnetic permeability of 1.0033 ± 0.0003. 4. Consequences
Fig. 5. Distribution of values of mass susceptibility for annealed 304 stainless steel.
The presence of a small fraction of delta ferrite is said to help prevent weld cracking and it appears to have no significant effect on the corrosion resistance or mechanical properties of 304 stainless. So for most purposes the delta ferrite is beneficial or
18
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
inconsequential. There are, however, applications where the magnetic response of the material needs to be minimized. Examples are scientific and medical equipment in which magnetic fields or charged-particle beams are present. The magnetic fields involved may be low (the earth’s field) or very high (in a superconducting magnet), or anything in between. There is a need for a measurement or test to determine whether a material is sufficiently non-magnetic for a particular use. Until 2014, ASTM International (originally the American Society for Testing Materials) specified three methods in Standard A 342/A 342 M-04 for determining ‘‘the Permeability of Feebly Magnetic Materials.” This document implicitly assumes that a feebly magnetic material can be described by a single magnetic permeability value. ASTM Method 1 is a point-by-point method using a fluxmeter, and so measures magnetic flux density B vs field H. The method originated before electronic integrators existed. It specifies a measurement at a single (undefined) magnetic field, and creates that field with an air-core solenoid which will be limited to a maximum field about 500 Oe. Method 2 is the classical Gouy method [16], and Method 3 is the Severn gage [17], although it is not identified by name. In 2014, the standard was revised. The Gouy method was dropped, the vibrating-sample magnetometer was added, as was a new commercial tester, the Magnetoscop [18], again not by name (although now the Severn gage is named in a footnote). The Standard now states that permeability is not necessarily a constant and shows a result like that of Fig. 1 above in the section on the VSM. We will consider each of these methods as they apply to a strongly paramagnetic material containing a small fraction of a ferromagnetic phase. We first note that although magnetic permeability m is always defined as m = B/H, its numerical value and its variation with field are very different in paramagnetic and ferromagnetic materials. In a paramagnet at ordinary temperatures, m has a value slightly greater than 1 (1.01 is a very high paramagnetic permeability) and is independent of field. In a ferromagnet, m is much greater than 1, and may reach 105 or higher in special cases. The value of m depends on the applied field, and special cases such as initial permeability and maximum permeability are defined. ASTM Method 1 measures the change in B for a known change in H; it can give a valid permeability value at a single field. It could be amended to require values for at least two fields different by a factor of two (or, using modern electronics, recording B vs H over a range of fields). This would at least permit a determination of whether the material contains a ferromagnetic component, and give a value for paramagnetic permeability if ferromagnetism is absent. Only if the same value of permeability is obtained at two different fields is the material shown to be purely paramagnetic. However, the maximum field obtainable from a solenoid is not sufficient to separate ferromagnetic from paramagnetic behavior. The same method could be used with an electromagnet rather than a solenoid to obtain much higher fields. The Gouy method, no longer as ASTM Method, requires that the material be paramagnetic, with M proportional to H, in order to give valid results. As with Method 1, measurements at different fields could determine whether or not ferromagnetism is present, and give a valid permeability if it is not. The VSM meets all the conditions for complete specification of the magnetic behavior. However, it does require the preparation of a small sample of defined geometry, and an electromagnet or superconducting magnet, and so is not well suited for routine measurements in a production environment.
The Severn gage and the Magnetoscop use a cylindrical permanent magnet that is placed in contact with a flat area of a sample or a finished part. The Severn gage compares the attractive force of the magnet to the sample with the force to one of a set of standards, and so can only assign a permeability value as falling between two standard values. The lowest permeability standard provided with the Severn gage is 1.01. The Magnetoscop measures the distortion of the magnetic field around the permanent magnet when it is placed in contact with the sample. According to the ASTM Standard, it claims a sensitivity of 0.00001 in permeability with an accuracy of ±5%. Both these probe-type testers are portable, quick and easy to use, can be used on finished parts, and provide a single permeability value (or a small range). However, they apply a magnetic field that varies widely both in magnitude and direction to a small and poorly-defined volume of the sample. Calibration is also a difficulty. Since there are no paramagnetic materials with permeabilities in the range of interest, the calibration standards are alloys or mechanical mixtures of small amounts of a ferromagnetic phase in a paramagnetic or diamagnetic matrix. The standards themselves do not have a permeability independent of field, and are themselves usually measured by ASTM Method 1 at some defined field such as 100 or 200 Oe [19]. All this means is that the permeability values provided by the probe testers are not on very solid scientific footing. They are also not interchangeable; that is, the two instruments will not give the same numerical result for a single sample. However, if they provide useful results and identify alloys that can function as required in magnetic environments, or serve as quality control standards, there is no reason not to continue to use them. It would be prudent to identify the values they produce as ‘‘Severn gage permeability” or ‘‘Magnetoscop permeability,” to indicate that they are not true permeabilities determined from a B/H curve. References [1] ASM Metals Handbook, 10th ed., 1992, vol. 1, p. 841. [2] William D. Callister, Jr. Materials Science and Engineering, an Introduction Wiley, 1985, p. 255. [3] Carpenter Technology Corp., Reading PA. (http://www.cartech.com) Technical Article ‘‘Magnetic Properties of Stainless Steels”. [4] F.C. Hull, Suppl. Weld. Res. J. (May 1973) 193s–203s. [5] T.A. Siewert, C.N. McCowan, D.L. Olson, Suppl. Weld. Res. J. Dec. (1988) p289s– 298s. [6] D.J. Kotecki, T.A. Siewert, Suppl. Weld. Res. J. (1992) 171Ss–178s. [7] J. Trygve Angel, Iron Steel Inst. 177 (1954) 5. [8] British Stainless Steel Assoc., Stainless Steel Advisory Service Informational Sheet No. 2.81, 1 May 2000. [9] Atlas Specialty Metals (Australia) (http://www.atlassteels.com.au) Tech Note No.11 (Dec. 2008). Also available at: http://www.kimballphysics.com/multiCF/Hardware/Technical Information. [10] ASM Metals Handbook, 9th ed. vol. 9, p. 287. [11] Mohammad Rezayat, Hamed Mirzadeh, Masih Namdar, Mohammad Habibi Parsa, Metall. Mater. Trans. A 47A (2016) 641–648. [12] S. Foner, Rev. Sci. Instrum. 30 (7) (1959) 548–557. [13] J.C. Childress, S.H. Liou, C.L. Chien, J. Appl. Phys. 64 (10) (1988) 6059–6061. [14] These samples were provided in a form suitable for our VSM through the cooperation of Dr. Vojteh Leskovsek of the Institute of Metals and Technology, Ljubljana, Slovenia, which we greatly appreciate. See V. Leskovsek et al., Metall. Mater. Trans. A, 45A, (2014), pp. 2819–2826. [15] J. Arpan Das, Magn. Mag. Mater. 36 (2013) 232–242. [16] G.L. Gouy, Compt. Rend. 109 (1889) p. 935. Or see B.D. Cullity, C.D. Graham, Introduction Magn. Mater. Wiley/IEEE (2009) p. 83. [17] Severn Engineering Co., Auburn AL USA (http://www.severnengineering.com). [18] Institut Dr Foerster GmbH, Reutlingen Germany (http:// www.foerstergroup.com). [19] Severn Engineering technical note ‘‘History of the Severn Engineering Permeability Standards.”
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials journal homepage: www.elsevier.com/locate/jmmm
Research articles
Delta ferrite is ubiquitous in type 304 stainless steel: Consequences for magnetic characterization C.D. Graham a,⇑, B.E. Lorenz b a b
Dept. of Materials Science, Univ. of Pennsylvania, Philadelphia PA, United States Dept. of Electrical Engineering, Widener Univ., Chester PA, United States
a r t i c l e
i n f o
Article history: Received 12 January 2018 Received in revised form 27 February 2018 Accepted 28 February 2018 Available online 1 March 2018 Keywords: 304 Stainless steel Delta ferrite Magnetic permeability
a b s t r a c t Using a vibrating-sample magnetometer with a maximum field of 20.5 kOe, we have measured over 50 samples of annealed 304 stainless steel, which is usually considered to be non-magnetic. In almost every case, we observe the presence of a small, usually less than 0.01, fraction of a ferromagnetic phase, which we believe to be equilibrium bcc delta ferrite. The consequences of this observation for the measurement and specification of the magnetic properties of annealed 304 stainless are discussed. Our measurements also establish the most likely value for the magnetic permeability of the fcc austenitic phase in 304 stainless steel austenite as 1.0033 ± 0.0003. Ó 2018 Elsevier B.V. All rights reserved.
1. Introduction According to standard reference sources, type 304 stainless steel in the annealed state is non-magnetic [1,2]. This actually means that it is paramagnetic, with magnetization M proportional to field H. The magnetic permeability m = B/H is variously given, usually either as 1.021 or in the range 1.002–1.004 [3]. We will return to this discrepancy later. Type 304 is considered to be paramagnetic because it is a single phase face-centered cubic (fcc) structure, known to metallurgists as austenite. The literature on welding of 304 stainless steel, however, is very much concerned with the presence of delta ferrite (d-ferrite) in the weld [4,5,6]. The use of the label ‘‘delta” ferrite implies this is an equilibrium body-centered cubic (bcc) phase that forms during solidification, analogous to the delta phase of pure iron that exists from the melting point of 1532 °C down to 1493 °C. Delta ferrite is to be distinguished from the bcc (or slightly tetragonal) phase that forms by a shear transformation when 304 stainless is plastically deformed below 80 °C [7]. This phase is known as martensite, and is usually labelled a’. The martensite in 304 stainless steel is often described as stress- or strain- or deformation-induced martensite. Deformation-induced martensite must have the same chemical composition as the austenite from which it forms, since it is produced by a diffusionless shear transformation. Delta ferrite, by contrast, is a thermodynamically stable phase that exists in
⇑ Corresponding author. E-mail address: [email protected] (C.D. Graham). https://doi.org/10.1016/j.jmmm.2018.02.092 0304-8853/Ó 2018 Elsevier B.V. All rights reserved.
equilibrium with austenite, and so has a different composition from the austenite. How much different is uncertain. Both martensite and delta ferrite are ferromagnetic, with saturation magnetization on the order of 100 emu/g or 10,000 gauss (1 T). We will return also to this subject later. Some delta-ferrite in a 304 stainless weld is regarded as desirable, to reduce the likelihood of crack formation during cooling. Therefore welding rod for 304 stainless is alloyed to contain a substantial content of delta ferrite. The presence of delta-ferrite apparently does not significantly affect the mechanical or corrosion-resistant properties of the alloy. We have found some suggestions in the literature that commercial 304 stainless steel may contain some delta ferrite in the annealed state. A publication from the British Stainless Steel Association [8] notes that Grades 304, 321, and 316 have ‘balanced’ compositions to enable them to be readily weldable. This is achieved by ensuring that in their normal annealed (softened) condition they contain a few percent of delta ferrite. The web site of Atlas Specialty Metals (Australia) [9] says It is common for wrought austenitic stainless steels to contain a very small amount of ferrite, but this is not sufficient to significantly affect magnetic performance except in very critical applications. The Metals Handbook volume on Metallography [10] includes a micrograph of annealed 302 stainless steel (the composition limits of 302 and 304 stainless overlap) with scattered circular patches of a second phase identified as ‘‘ferrite.” This is clearly an equilibrium phase, and not martensite. The ferrite appears to be present to about 1%, in the form of roughly spherical regions about 10 mm in diameter. A very recent paper from Iran [11] states as an accepted fact
16
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
that austenitic stainless steels contain delta ferrite and that it is regarded as undesirable; the paper investigates possible treatments to reduce or remove delta ferrite. 2. Experiment Over the course of several years, while investigating the formation of deformation-induced martensite and its reversion to austenite, we have measured about 50 samples of annealed and as-received (presumably annealed) 304 stainless steel, using a vibrating-sample magnetometer (VSM) [12] with a maximum applied field of 20.5 kOe. The samples were either disks about 6 mm in diameter and about 1 mm thick, magnetized along a diameter, or wires 6 mm long with a range of diameters up to 2 mm, magnetized along the length. The VSM was calibrated for disk samples with a disk of polycrystalline high-purity nickel, 0.25 inch (6.35 mm) in diameter and 0.050 inch (1.27 mm) thick. The wire sample calibration was a 6 mm long by 0.65 mm diameter highpurity polycrystalline iron wire. The samples came from a variety of sources. All were either asreceived and presumably annealed, or had been annealed at a temperature of at least 800 °C. Prior to annealing most of the samples had been deformed by varying amounts, by various methods, and at various temperatures. Some measurements were made on the same sample after different treatments. A few samples were type 304L or 304LN, indicating lower carbon, or lower carbon and higher nitrogen, than standard grade 304. One wire sample was type 302. In most cases the exact chemical composition is not known, and it may be important to note that the composition limits on 304 (and other) stainless steel are unusually wide. We make no claim that our samples are representative of anything; they are however, commercially-produced 304 stainless from several different sources, given a variety of deformation and annealing treatments. We report magnetization in emu/g (r) rather than emu/cm3 (M), since for small samples mass can be determined more accurately than volume. In almost every one of our 50 cases, the VSM result is like that in Fig. 1. We interpret this figure as showing the presence of a ferromagnetic phase (bcc delta ferrite), whose contribution to the magnetization of the sample is given by the intercept of the linear high-field data on the + or vertical axis, and a paramagnetic phase, fcc austenite, whose magnetic mass susceptibility is given by the slope of the linear high-field data. See Fig. 2. For the sample in Figs. 1 and 2, the magnetization resulting from the ferromagnetic phase gives a sample magnetization of 0.77 emu/g, and the mass susceptibility of the austenite is 33 10 6 emu/g-Oe. To convert this value to magnetic permeability (B/H), multiply by the density (8.0 g/cm3) and by 4p, which is very nearly a factor of 100, and add 1, to give permeability m = 1.0033.
Fig. 1. Typical VSM hysteresis loop of annealed 304 stainless steel.
Fig. 2. Data of Fig. 1 replotted to show paramagnetic component (solid line) and ferromagnetic component (data points). Solid line is a linear fit to the high-field data. The slope of this line gives the mass susceptibility of austenite. The points are the data points of Fig. 1 with the paramagnetic component subtracted. Note that in this sample, the paramagnetic and ferromagnetic magnetizations are nearly equal at a field of 20 kOe.
The values of ferromagnetic magnetization r in our samples ranged from 0 to 2 emu/g, in an apparently random way. See Fig. 3. The average value for 50 samples was 0.6 emu/g with a standard deviation of 0.65. A few values above 2 emu/g were eliminated from the data set on the grounds that the samples may have contained deformation-induced ferromagnetic martensite. It is not possible to distinguish martensite from delta ferrite simply on the basis of magnetic measurements. The value of the saturation magnetization of pure delta ferrite is unknown. Presumably it is similar to that of deformation-induced martensite, although (as noted above) the chemical compositions of the two phases are not expected to be identical. We are aware of just one value for the saturation magnetization of martensite in 304 stainless steel: 160 emu/g by Angel, dating to 1954 [7] There is also an interesting result from 1988 [13] in which a thin film was sputtered from a commercial 304 target to produce a fully bcc structure with a magnetization of 136 emu/g. The magnetization value is likely to be somewhat composition dependent, and, as noted above, the composition limits on type 304 stainless are unusually broad. It seems reasonable to take a value of 150 ± 25 emu/g for the saturation magnetization of delta ferrite. On that basis, a measured sample magnetization of 1 emu/g would imply about 0.7% delta ferrite content.
Fig. 3. Mass susceptibility vs. ferromagnetic magnetization r for 50 samples of Type 304 stainless steel. Susceptibility is generally near 30 10 6 emu/g-Oe and is uncorrelated with the magnetization.
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
Fig. 4. VSM data for a sample of annealed 304L stainless steel containing only paramagnetic austenite. The mass susceptibility is 29 10 6 emu/g-Oe. The composition of this material is known: C 0.017, N 0.08, Cr 18.94, Ni 8.85, Mn 1.32, Si 0.34, Cu 0.27, Mo 0.23, Ti 0.002, Fe bal, (all wt%).
17
Fig. 6. Low field hysteresis loop of an annealed 304 stainless steel. From the highfield data for this sample, the ferromagnetic magnetization is 0.77 emu/g and the mass susceptibility is 32 10 6 emu/g-Oe.
2.1. Low-field results
fields below 1 kOe on most of our samples. We did make more extensive measurements on some samples, and a typical result is shown in Fig. 6. Coercive fields were generally between 50 and 150 Oe, and remanent magnetizations much less than the extrapolated saturation magnetization, both with considerable variation from sample to sample. If the microstructure of our samples is like that shown in Ref. 10, it can be described as widely separated spheres of ferromagnetic delta ferrite in a paramagnetic austenite matrix. The ferromagnetic spheres are far enough apart to have minimal interaction, so each ferromagnetic sphere will experience only its own demagnetizing field. The demagnetizing field of a sphere is 4p M and if r is 140 emu/g, M is 1120 emu/cm3 and the demagne3 tizing field is about 4.6 kOe. This is in reasonable agreement with the observation that reaching a linear slope by saturating the ferromagnetic magnetization requires a field of 3–5 kOe. Coercive field values near 100 Oe are unusually high for an annealed metal. One possible explanation lies in the difference in thermal expansion coefficients between austenite and delta ferrite, which can lead to significant stresses on cooling of a two-phase sample. Differences in composition and annealing treatments, including cooling rates, could help to account for the variability in coercivity.
Since our primary interest was in quantities determined from high-field measurements, we took relatively few data points at
3. Conclusions
The only samples that showed no delta ferrite were obtained from Slovenia [14], and were made from material provided by an Italian supplier. Fig. 4 shows the VSM data from one of these samples. In contrast to the scatter in values of r, the magnetic susceptibility values given by the slope of the high-field linear magnetization data clustered strongly near 33 10 6 emu/g-Oe, with a few outlier points at lower and higher values. Fig. 5 shows the distribution of values. There are several possible explanations for the range of susceptibility values. There is likely to be a chemical composition dependence of the susceptibility; our VSM calibration standards were not perfect matches to the size and shapes of our samples; and there was also a scatter of ±2 or 3 10 6 emu/g-Oe in values for a single sample. We have found two published VSM hysteresis loops of annealed 304 stainless steel [14,15]. They are both consistent with our results, showing values of 0.50 and 0.79 emu/g for r and 29 and 26 10 6 emu/g-Oe for mass susceptibility.
1. It is possible, even likely, that annealed 304 stainless steel will contain up to about 2% of strongly ferromagnetic delta ferrite. 2. This accounts for the variation in reported permeability value for austenitic stainless steels. The presence of a ferromagnetic phase will lead to a measured permeability higher than the paramagnetic permeability of fcc austenite, and to a permeability that will vary with the applied field. 3. The most likely value for the susceptibility of the fcc austenite phase in 304 stainless steel is 33 ± 3 10 6 emu/g-Oe or 265 ± 25 emu/cm3-Oe, corresponding to a magnetic permeability of 1.0033 ± 0.0003. 4. Consequences
Fig. 5. Distribution of values of mass susceptibility for annealed 304 stainless steel.
The presence of a small fraction of delta ferrite is said to help prevent weld cracking and it appears to have no significant effect on the corrosion resistance or mechanical properties of 304 stainless. So for most purposes the delta ferrite is beneficial or
18
C.D. Graham, B.E. Lorenz / Journal of Magnetism and Magnetic Materials 458 (2018) 15–18
inconsequential. There are, however, applications where the magnetic response of the material needs to be minimized. Examples are scientific and medical equipment in which magnetic fields or charged-particle beams are present. The magnetic fields involved may be low (the earth’s field) or very high (in a superconducting magnet), or anything in between. There is a need for a measurement or test to determine whether a material is sufficiently non-magnetic for a particular use. Until 2014, ASTM International (originally the American Society for Testing Materials) specified three methods in Standard A 342/A 342 M-04 for determining ‘‘the Permeability of Feebly Magnetic Materials.” This document implicitly assumes that a feebly magnetic material can be described by a single magnetic permeability value. ASTM Method 1 is a point-by-point method using a fluxmeter, and so measures magnetic flux density B vs field H. The method originated before electronic integrators existed. It specifies a measurement at a single (undefined) magnetic field, and creates that field with an air-core solenoid which will be limited to a maximum field about 500 Oe. Method 2 is the classical Gouy method [16], and Method 3 is the Severn gage [17], although it is not identified by name. In 2014, the standard was revised. The Gouy method was dropped, the vibrating-sample magnetometer was added, as was a new commercial tester, the Magnetoscop [18], again not by name (although now the Severn gage is named in a footnote). The Standard now states that permeability is not necessarily a constant and shows a result like that of Fig. 1 above in the section on the VSM. We will consider each of these methods as they apply to a strongly paramagnetic material containing a small fraction of a ferromagnetic phase. We first note that although magnetic permeability m is always defined as m = B/H, its numerical value and its variation with field are very different in paramagnetic and ferromagnetic materials. In a paramagnet at ordinary temperatures, m has a value slightly greater than 1 (1.01 is a very high paramagnetic permeability) and is independent of field. In a ferromagnet, m is much greater than 1, and may reach 105 or higher in special cases. The value of m depends on the applied field, and special cases such as initial permeability and maximum permeability are defined. ASTM Method 1 measures the change in B for a known change in H; it can give a valid permeability value at a single field. It could be amended to require values for at least two fields different by a factor of two (or, using modern electronics, recording B vs H over a range of fields). This would at least permit a determination of whether the material contains a ferromagnetic component, and give a value for paramagnetic permeability if ferromagnetism is absent. Only if the same value of permeability is obtained at two different fields is the material shown to be purely paramagnetic. However, the maximum field obtainable from a solenoid is not sufficient to separate ferromagnetic from paramagnetic behavior. The same method could be used with an electromagnet rather than a solenoid to obtain much higher fields. The Gouy method, no longer as ASTM Method, requires that the material be paramagnetic, with M proportional to H, in order to give valid results. As with Method 1, measurements at different fields could determine whether or not ferromagnetism is present, and give a valid permeability if it is not. The VSM meets all the conditions for complete specification of the magnetic behavior. However, it does require the preparation of a small sample of defined geometry, and an electromagnet or superconducting magnet, and so is not well suited for routine measurements in a production environment.
The Severn gage and the Magnetoscop use a cylindrical permanent magnet that is placed in contact with a flat area of a sample or a finished part. The Severn gage compares the attractive force of the magnet to the sample with the force to one of a set of standards, and so can only assign a permeability value as falling between two standard values. The lowest permeability standard provided with the Severn gage is 1.01. The Magnetoscop measures the distortion of the magnetic field around the permanent magnet when it is placed in contact with the sample. According to the ASTM Standard, it claims a sensitivity of 0.00001 in permeability with an accuracy of ±5%. Both these probe-type testers are portable, quick and easy to use, can be used on finished parts, and provide a single permeability value (or a small range). However, they apply a magnetic field that varies widely both in magnitude and direction to a small and poorly-defined volume of the sample. Calibration is also a difficulty. Since there are no paramagnetic materials with permeabilities in the range of interest, the calibration standards are alloys or mechanical mixtures of small amounts of a ferromagnetic phase in a paramagnetic or diamagnetic matrix. The standards themselves do not have a permeability independent of field, and are themselves usually measured by ASTM Method 1 at some defined field such as 100 or 200 Oe [19]. All this means is that the permeability values provided by the probe testers are not on very solid scientific footing. They are also not interchangeable; that is, the two instruments will not give the same numerical result for a single sample. However, if they provide useful results and identify alloys that can function as required in magnetic environments, or serve as quality control standards, there is no reason not to continue to use them. It would be prudent to identify the values they produce as ‘‘Severn gage permeability” or ‘‘Magnetoscop permeability,” to indicate that they are not true permeabilities determined from a B/H curve. References [1] ASM Metals Handbook, 10th ed., 1992, vol. 1, p. 841. [2] William D. Callister, Jr. Materials Science and Engineering, an Introduction Wiley, 1985, p. 255. [3] Carpenter Technology Corp., Reading PA. (http://www.cartech.com) Technical Article ‘‘Magnetic Properties of Stainless Steels”. [4] F.C. Hull, Suppl. Weld. Res. J. (May 1973) 193s–203s. [5] T.A. Siewert, C.N. McCowan, D.L. Olson, Suppl. Weld. Res. J. Dec. (1988) p289s– 298s. [6] D.J. Kotecki, T.A. Siewert, Suppl. Weld. Res. J. (1992) 171Ss–178s. [7] J. Trygve Angel, Iron Steel Inst. 177 (1954) 5. [8] British Stainless Steel Assoc., Stainless Steel Advisory Service Informational Sheet No. 2.81, 1 May 2000. [9] Atlas Specialty Metals (Australia) (http://www.atlassteels.com.au) Tech Note No.11 (Dec. 2008). Also available at: http://www.kimballphysics.com/multiCF/Hardware/Technical Information. [10] ASM Metals Handbook, 9th ed. vol. 9, p. 287. [11] Mohammad Rezayat, Hamed Mirzadeh, Masih Namdar, Mohammad Habibi Parsa, Metall. Mater. Trans. A 47A (2016) 641–648. [12] S. Foner, Rev. Sci. Instrum. 30 (7) (1959) 548–557. [13] J.C. Childress, S.H. Liou, C.L. Chien, J. Appl. Phys. 64 (10) (1988) 6059–6061. [14] These samples were provided in a form suitable for our VSM through the cooperation of Dr. Vojteh Leskovsek of the Institute of Metals and Technology, Ljubljana, Slovenia, which we greatly appreciate. See V. Leskovsek et al., Metall. Mater. Trans. A, 45A, (2014), pp. 2819–2826. [15] J. Arpan Das, Magn. Mag. Mater. 36 (2013) 232–242. [16] G.L. Gouy, Compt. Rend. 109 (1889) p. 935. Or see B.D. Cullity, C.D. Graham, Introduction Magn. Mater. Wiley/IEEE (2009) p. 83. [17] Severn Engineering Co., Auburn AL USA (http://www.severnengineering.com). [18] Institut Dr Foerster GmbH, Reutlingen Germany (http:// www.foerstergroup.com). [19] Severn Engineering technical note ‘‘History of the Severn Engineering Permeability Standards.”