Barkhausen Noise Analysis (BNA) is a nondestructive technique based on the inductive measurement of a noise-like signal generated in ferromagnetic material when introduced to an external magnetic field.
While the physical phenomenon of magnetic Barkhausen Noise was first observed and explained in 1919, industrial application of it as a measurement technique was uncommon before the 1980s. Today, BNA is a trusted nondestructive method for stress analysis, material characterization and the detection of heat-treatment and other thermally induced material defects.
Barkhausen Noise Analysis
BNA is a phenomenon that exists only in ferromagnetic materials, limiting the method to applications involving iron and most steels (excluding austenitic stainless steels) as well as cobalt and nickel alloys. Ferromagnetic materials are unique in that they can form permanent magnets. Furthermore, ferromagnetic materials are composed of magnetic domains, a direct result of the exchange interaction. Magnetic domains are volumes of material in which all magnetic dipoles align in the same direction. The boundaries, or walls, that separate domains give rise to the Barkhausen Effect.
As illustrated in Figure 1, when a ferromagnetic material is subject to an external magnetic field, the domains that are parallel to the direction of an applied magnetic field grow at the expense of neighboring domains. As these domains grow and shrink, domain-wall movement is hindered by imperfections in the crystal lattice, including precipitates, inclusions, dislocations, etc. These “pinning sites” result in discontinuities in the magnetization curve where the net magnetization of the sample does not change until the applied field strength is increased enough to de-pin the wall.
An abrupt step, or “Barkhausen jump,” then occurs in the magnetization curve. Domain walls then proceed to move through the material until encountering another pinning site. This process continues until the entire volume is magnetically aligned and has reached “saturation,” the point where any further increase in applied field strength does not affect the net magnetization of the sample.
Measuring Barkhausen Noise
Barkhausen Noise is the signal that results from measuring Barkhausen jumps using an inductive pickup. Each jump results in a measured pulse. A large number of pulses result in a burst with a noise-like structure. A sensor similar to the one shown in Figure 2 is typically used to apply an alternating magnetic field, commonly in sinusoidal waveform, via magnetizing poles that contact the sample surface. Using modern equipment, the strength and frequency of the applied magnetization can be adjusted to yield the highest sensitivity to the defect of interest. A centrally located inductive pickup is used to measure the rate of change of magnetization (Barkhausen jumps) in the magnetized sample volume.
The Barkhausen signal is collected, filtered and amplified by a Barkhausen Noise analyzer, and several parameters can then be calculated. The most common parameter utilized is the signal root-mean-square (RMS), which is sometimes referred to as the magneto-elastic parameter (MP). Barkhausen Noise RMS can be calculated in real-time, and the relative differences in the parameter are commonly used to detect near-surface (0.01-0.2 mm) defects such as decarburization, surface hardness or grinding (re-temper) burn. Evaluation of material characteristics at greater depths (such as case-depth analysis) typically on the order of millimeters, generally requires a slightly more involved analysis in which several Barkhausen Noise parameters are collected during a static measurement. In these cases, some post-processing may be required.
Barkhausen Signal Sensitivity
Several factors drive variation in the Barkhausen signal. The two most relevant in regard to industrial applications are material stress and microstructure. Simply put, positive (tensile) stress – residual or applied – creates an “easy access” for magnetization parallel to the direction of stress; hence, less resistance to domain-wall movement resulting in greater Barkhausen Noise RMS values. Conversely, negative or compressive residual stresses result in lower measured RMS values. As illustrated in Figure 3, within the elastic regime a linear relationship exists between stress and Barkhausen Noise RMS. This advantageous relationship has helped make BNA a valuable tool for determining in-situ loading, assessing shot- and shock-peen coverage, and detecting other stress-inducing processes such as grinding burn.
Of greater importance to heat-treatment applications is the sensitivity to sample microstructure. Changes in microstructure affect bulk magnetic properties, which, in turn, affect the Barkhausen Noise signal. Martensite, for example, is significantly more difficult to magnetize than ferrite, resulting in a considerably lower signal. Furthermore, the presence of carbides and nitrides also decrease the BN signal. In general, the relationship of Barkhausen Noise RMS to microstructure can be broadly expressed as a sensitivity to hardness. In most cases, there is a nearly linear correlation between RMS and hardness where an increase in hardness leads to a decrease in the BN signal and vice versa. It is important to note that these relationships hold true for iron and most ferromagnetic steels.
The competitive nature of – and increasingly stringent requirements placed on – modern manufacturers has resulted in a demand for more effective processes, lower cycle times and in many cases 100% part conformance. Traditional heat-treatment process verification and quality-control methods are generally time- and cost-inefficient.
For example, traditional case-depth analysis commonly involves destructive sectioning and subsequent microscopy or hardness profiles, which is a process requiring costly resources, significant operator expertise and effort, and additional costs related to scrap components. Detection of decarburization or soft spots is another issue traditionally requiring labor-intensive inspections. Commonly, the process involves a gross or localized acid etch and subsequent visual inspection of the component surface, which results in a relatively subjective inspection that exposes operators to potentially hazardous chemicals. BNA offers a completely nondestructive alternative that is free of consumables and capable of 100% part inspection.
Detection of Soft Spots
A manufacturer of carburized automotive camshafts was encountering a relatively high occurrence of soft spots (decarburization) on the surface of hardened cam lobes. At that time, their inspection was visual after an etch procedure to reveal soft microstructures. In this case, the manufacturer was motivated to find an alternative method to improve inspection cycle times.
A blind study was carried out to assess the feasibility and capability of BNA. Several camshafts and numerous lobes were measured. Barkhausen data was acquired during a single rotation with the sensor pickup centrally located on each lobe similar to Figure 4. The preliminary results acquired from six separate lobes are shown graphically in Figure 5.
The lobes having uniform surface hardness show a fairly consistent MP value of approximately 25 mV, while a lobe having a suspected area of decarb, at approximately 250 degrees (70% of rotation), spikes the signal to nearly 45 MP. This sensitivity was common for several other suspected soft spots within the sample set, and a straightforward rejection limit of 35 MP was chosen. A subsequent etch inspection of all samples confirmed that BNA achieved 100% detection with no false positives.
The decision to replace the traditional etch inspection with automated BNA in this case was quite simple. Utilizing a semiautomated inspection stand, a measurement time of less than three seconds was achieved for each lobe, greatly reducing the overall inspection cycle. Secondary benefits included the reduced use of nitric acid and improved traceability and repeatability of the inspection.
As previously mentioned, the verification of case depth is typically accomplished via destructive methods. A recently developed approach utilizing BNA provides a completely nondestructive alternative. The following example highlights the effectiveness of the method and explains some of the differences as compared to traditional Barkhausen measurements.
A manufacturer supplying the mining and power-generation industries, eager to investigate the effectiveness of BNA, created a spectrum of samples ranging in case depth from no case to a case depth of 3.8 mm in steps of 0.2 mm. The hardened cylindrical samples were measured at four circumferential locations, and each measurement was repeated four times for a total of 16 measurements per sample.
To achieve greater depth sensitivities, measurements were made using a magnetizing frequency of 0.5 Hz, a significantly lower frequency than the traditionally applied 125 Hz. A total of 10 Barkhausen bursts were collected at each location, requiring measurements lasting 20 seconds. Numerous parameters were collected during each measurement and then correlated to the expected case depths for all samples. As with many NDT methods, the sensitivity and accuracy of the approach is generally a case-by-case basis. Several variables can affect the Barkhausen response, and an initial correlation to traditional methods is typically required.
In this case, a strong exponential correlation (R2=0.97) was found between the median measured RMS and the expected case depth. As shown in Figure 6a, however, sensitivity was lost at depths greater than 1.5 mm. Fortunately, a second parameter, pulse count, which as the name suggests is a measure of the quantity of Barkhausen events during a magnetizing cycle, provides a strong linear correlation (R2=0.95) to case depths from 1 mm to 3.8 mm, as shown in Figure 6b. Together, the two parameters allow for an accurate estimation of case depth with levels of uncertainty similar to that of traditional methods.
Barkhausen Noise is a nondestructive magnetic method sensitive to changes in material stress and microstructure. The latter allows the method to be confidently utilized in process and quality-control applications including but not limited to heat-treatment defect detection and case-depth analysis. As with most NDT methods, the accuracy/sensitivity can be a case-by-case basis. In most applications the method requires an initial correlation to traditional methods. Upon completion of preliminary correlation and equipment setup, measurements require minimal operator expertise, eliminate the need for hazardous consumables and result in a reduction in the costs associated with labor-intensive inspections and scrap components.
For more information: Contact Wade Gubbels, American Stress Technologies, Inc., 540 Alpha Drive, Pittsburgh, PA 15238; tel: 412-784-8400; e-mail: firstname.lastname@example.org or email@example.com; web: www.astresstech.com