Barkhausen Noise as a Reliable Tool for Sustainable Automotive Production

Abstract

This paper deals with the sustainable production of components in the automotive industry, with the focus on the nondestructive evaluation of components after plasma nitridation via the Barkhausen noise technique. This study investigates the influence of tool wear on surface state after turning, and the consecutive plasma nitriding process. Moreover, position in the nitriding chamber and the corresponding heterogeneity of components is investigated as well. The results of experiments indicate that an increasing mechanical and thermal load due to flank wear and the associated process dynamics negatively affects the heterogeneity of the surface state after turning, and consecutive nitriding in terms of Barkhausen noise emission. Moreover, it was found that the conditions in the chamber during the nitriding process vary and, especially near the venting system, the temperature is slightly lower, such that some components are found to be unacceptable as well. The study also unwraps the contribution of the diffusion and compound layers with respect to MBN and discusses the contribution of the MBN pulses of different frequencies. The pinning strength of nitrides is indicated with respect to their size and the related thickness of DWs. Finally, this study clearly demonstrates how the MBN technique can be employed for the monitoring nitrided components and the corresponding optimisation of manufacturing cycles.

1. Introduction

Environmental aspects of transport are widely investigated and discussed [1,2,3]. However, sustainable transportation should be also linked with the production phase. Sustainable automotive production requires avoiding monitoring techniques based on aggressive chemical etchants, such as those employed for the visualisation of thermal softening after grinding. These processes are considered as being risky with respect to human health as well as with respect to the environmental aspect (the disposal of chemical substances after use) [4]. Moreover, sustainable automotive production requires the identification of risky factors during components manufacturing, to avoid their premature failure [5]. Conventional destructive techniques based on metallographic observation, microhardness measurements, etc., damage the analysed body. On the other hand, these techniques provide information only about a limited number of produced pieces. For these reasons, a fast and reliable nondestructive technique could provide information about all samples without them being damaged. These data can be employed for process optimisation. Based on real industrial experience, it was found that the quality of components after the plasma nitriding of steels can vary markedly, and some pieces were detected as being unacceptable.

Nitriding is the well known heat and chemical process in which a hard surface lies on a tough deeper core. This process usually markedly improves resistance against cyclic loading, and resistance against wear and/or corrosion [6,7,8]. The high hardness of the near surface layer being nitrided is due to nitride’s very fine precipitates as a result of the limited solubility of N in α-iron. Nitrogen is produced by the decomposition of the ammonia NH₃ atmosphere. The nitrided surface is composed of the near surface nitriding layer (compound layer) containing ε and γ’ nitrides of low thickness (a few micrometres). On the other hand, the underlying diffusion layer is referred to as an α-iron matrix, containing a limited volume of dissolved N, face centred cubic γ’-nitride Fe₄N and hexahedral ε-nitride Fe_2–3N (sometimes orthorhombic ζ-nitride Fe₂N) [7]. The concept of components’ nitridation is widely used for a variety of components employed in the automotive industry. Contrasted with carburising, nitriding requires lower temperatures (about 550 °C or lower) and high hardness is directly generated during the chemical process [7]. For this reason, no phase alterations take place and the deformation of samples is only minor. Many studies have been carried out in this field, as mentioned above [5,6,7,8,9]. In addition, Marot et al. [10] reported about nitrogen transfer in steel during the plasma nitridation process. Flori et al. [11] analysed the influence of defects on the plasma nitridation process and found their remarkable role in the final state after nitriding.

The main disadvantage of plasma nitriding can be viewed as its quite complicated process control, since many variable input parameters can significantly affect the nitriding process, such as time duration, temperature in the chamber, current density, pulses duration, etc. Furthermore, it is also considered that the surface state before the nitriding process can also play a significant role in the final state after the nitriding process. The final state of any components after the nitriding process can be expressed in many terms, such as the thickness of the compound and diffusion layers, hardness of the diffusion layer and its profile, etc. The monitoring of components after plasma nitriding is usually long term and carried out in a destructive manner using light metallography, microhardness measurements, SEM or XRD techniques. Being so, the application of a suitable nondestructive and fast technique could be beneficial, to prevent the use of unacceptable components.

It is well known that domain walls (DWs) in ferromagnetic bodies, during their irreversible and discontinuous motion, encounter lattice imperfections, including precipitates of variable size, density and morphology [12,13,14,15,16,17]. For this reason, the electromagnetic pulses produced by DWs during their motion contain information about their microstructure. On the other hand, DWs’ alignment can be affected by a stress state, when tensile stresses align DWs towards the direction of the tensile stress, whereas compressive stresses align DWs perpendicularly against the direction of compressive stress [18,19,20].

The compound layer is composed of nonferromagnetic nitrides only. For this reason, magnetic Barkhausen noise (MBN) produced by nitrided components originates from the diffusion layer only [21], whereas the compound layer on the surface acts as the layer attenuating electromagnetic pulses propagating towards the free surface. Nitrides in the matrix pin hinder DWs in motion, thus decreasing the amplitude of electromagnetic pulses and the corresponding MBN. The concept in which the MBN technique is employed for the monitoring of components after the nitriding process is based on the contrast between the low MBN originating from the acceptable matrix (diffusion layer) having high hardness, and the corresponding high density of nitrides and the high MBN originating from an unacceptable matrix in which the density of nitrides is reduced together with the hardness [21,22]. This paper discusses the critical aspects of the plasma nitriding process in the automotive industry and provides brief reports in which the real components are investigated.

This study discusses the influence of tool flank wear during turning on the heterogeneity of the surface after plasma nitriding, as well as the influence of nonhomogenous conditions for the nitriding process in the chamber. Moreover, the study also unwraps the contribution of the diffusion and compound layers with respect to MBN and discusses the contribution of MBN pulses of different frequencies. The pinning strength of nitrides is indicated with respect to their size and the related thickness of DWs. Finally, this study clearly demonstrates how the MBN technique can be employed for monitoring nitrided components and the corresponding optimisation of manufacturing cycles.

2. Materials and Methods

The first test was carried out on the component as illustrated in Figure 1. This component was produced by the turning process and two components were investigated. The first one was turned by the insertion of low flank wear (VB) and second one turned by the insert of high VB. Further information about the matrix (its chemistry, initial microstructure) as well as information about turning process cannot be shared. MBN measurements in the dynamic regime were carried out by a Rollscan 350 device (Stresstech, Jyväskylä, Finland) and ViewScan software (frequency range of detected MBN pulses from 70 to 200 kHz, mag. frequency 125 Hz, mag. voltage 10 V, serial sensor S1-18-12-01). MBN was measured along the sample circuit in the position indicated by the arrow in Figure 1. MBN signal was sampled by 10 Hz (sample motion speed 15 mm·s⁻¹). Static MBN measurements in the chosen positions on both rings were carried out by the use of the same device, but MicroScan software (Stresstech, Jyväskylä, Finland) was employed (frequency range of detected MBN pulses 10 to 1000 kHz, mag. frequency 125 Hz, mag. voltage 10 V). MBN refers to effective (rms) value of acquired Barkhausen noise signal. Apart from MBN, also the number of detected pulses, MBN envelopes and the corresponding PP values were analysed. PP refers to the position of the envelope in which this envelope attains the maximum.

Plasma nitriding process was running at elevated temperatures without accelerated cooling rates. For this reason, it was considered that MBN in this particular case is mainly a function of microstructure and the height of surface irregularities. Microstructure is expressed in terms of the compound and diffusion layer thickness as well as microhardness profile of the diffusion layer. For these reasons, true interpretation of MBN signal is based on the aforementioned microstructure measurements (observations) as well as surface topography, following the flowchart indicated in Figure 2.

To reveal the microstructure, 20 mm long pieces were routinely prepared for metallographic observations (hot moulded, ground, polished and etched by 3% Nital for 10 s). All measurements were carried out under laboratory conditions. Microstructure observations were carried using the light microscope Neophot 2 (Carl Zeiss, Jena, Germany) in software Niss Elements (Nikon Instruments Inc., New York, NY, USA).

Microhardness (HV0.05) depth profiles were measured using an Innova Test 400TM (50 g for 10 s) (Innovatest, Maastricht, The Netherlands) on the cross sectional cuts after metallographic observations. Presented microhardness was obtained averaging the five repetitive measurements.

In order to analyse the contribution of surface height irregularities on MBN as a result of variable surface roughness, surface topography was studied using the laser confocal microscope Zeiss Axio applying the laser of wavelength 405 nm in ZEN software (Carl Zeiss Microscopy GmbH, Jena, Germany). The surface of the area of 3 × 3 mm was scanned using the function Z-stack when the height of 20 μm was cut into 50 levels. Band pass filter according the ISO standard (band pass filter) was employed for data filtration and surface reconstruction.

The second experiment was carried out directly under industrial conditions on the samples made of low alloyed steel of nominal yield strength of 355 MPa of cylindrical shape (outer diameter about 85 mm and length 200 mm). A total of 600 pieces were nitrided in the chamber and placed in 12 layers. Each layer contained 50 components distributed in three circles (outer, middle and inner), see Figure 3. This figure also depicts the position of the venting system during the nitridation process. Plasma nitriding process was performed at a temperature of approx. 525 °C for 8 h.

3. Results of Experiments

3.1. Influence of Flank Wear

Figure 4 illustrates the low MBN values in all positions for the components turned by the insert of low VB. Although MBN values fluctuate, MBN is kept below 200 mV. On the other hand, the component turned by the insertion of high VB exhibits a narrow region in which MBN exceeds 300 mV. Figure 4 also depicts three different positions in which detailed investigation of the microstructure, microhardness, as well as static MBN measurements were carried out. Position n.1 on the component turned by the insertion of low VB and nitrided consequently corresponds with the similar position, n.2, on the components turned by the insertion of high VB, which exhibits a similar MBN. Finally, the position n.3, which exhibits an MBN of 350 mV, was also investigated.

Figure 5 illustrates metallographic images in the aforementioned positions. The diffusion layer on the left column can be easily recognised as the dark region lying below the thin white layer, which represents the compound layer entirely composed of nitrides. Figure 5 clearly demonstrates no marked difference among the samples, with respect to the thickness of the diffusion layer. On the other hand, the thickness of the compound layer in position n.3 is not continuous, when contrasted with positions n.1 and 2. Moreover, its thickness is lower.

Microhardness measurements revealed that hardness in the near surface region is lower in position n.3 when compared with positions n.1 and 2, as a result of the reduced density of the nitrides in the diffusion layer, see Figure 6. Due to the very low thickness of the compound layer, microhardness is measured in the diffusion layer only. The softer diffusion layer is linked with a lower density of nitrides and corresponding lower pinning strength of the matrix. This aspect contributes to the longer free path of DWs’ motion and their earlier initiation, which, in turn, contributes to the higher MBN. The static MBN measurements prove the marked difference between positions n.1 and 3, see Figure 7. It should be also noticed that the MBN values in the dynamic records are significantly lower than those obtained from the static mode. The main reason should be viewed in the different frequency range of the MBN pulses recorded for the static and dynamic measurements, due to the different set up in the ViewScan (70–200 kHz) and MicroScan (10–1000 kHz) software.

It is known that the height of surface irregularities can contribute to a lower MBN [23]. Figure 8 illustrates that this height is much higher after turning by the insertion of high VB when compared with the insert of the low VB. The height of irregularities for the low VB oscillates in the range ±2 μm, whereas, in the case of the high VB, it is in the range ±6 μm. This indicates that this effect takes only a minor role in MBN, and the influence of the diffusion layer and the corresponding density of nitrides prevail.

Čížek [14] reported about the increased dislocation density of the surface after turning with a more developed VB. Increasing dislocation density should encourage nitrides to embed in the matrix, since dislocation cells are clustered by vacancies as preferential sites for nitrides [6]. As soon as the alteration in the machined surface exceeds a critical threshold, this effect seems to disappear and the rate of N diffusion in the steel matrix drops down (as is proven by micro hardness measurements).

The lower mechanic hardness (HV0.05) correlates with the lower magnetic hardness expressed in terms of PP and associated MBN envelopes. Figure 9 illustrates that an MBN envelope in the position n.1 is shifted towards the higher magnetic fields, as compared with position n.3. Expressed in other words, a stronger magnetic field is needed to unpin DWs in their positions due to the higher density of nitrides in the matrix, see also information in

Table 1. Measured MBN parameters.

	Dynamic MBN, mV *	Static MBN, mV *	PP, kA·m−1	Number of Pulses, -
Low VB, position n.1	182 ± 7	579 ± 15	0.93 ± 0.05	34,250
High VB, position n.2	170 ± 6	565 ± 18	1.01 ± 0.07	35,250
High VB, position n.3	353 ± 8	1226 ± 25	0.63 ± 0.05	30,500

* Note: dynamic measurements in the frequency range of MBN pulses 70–200 kHz; static measurements in the frequency range of MBN pulses 10–1000 kHz.

(PP refers to the maximum of the MBN envelope). The reduced number of nitrides in position n.3 also proves the lower number of detected MBN pulses that DWs encounter during their irreversible motion, see Table 1. It can be summarised that the higher MBN after the nitriding process can be also linked with the lower PP and the reduced number of MBN pulses.

Table 1 also indicates that MBN in the dynamic regime are remarkably lower that that obtained during the static regime, due to the remarkable difference in the frequency range of detected pulses. On the other hand, Table 1 also clearly proves that the higher MBN in the static regime directly correlates with the higher MBN values in the dynamic regime and vice versa. The FFT spectrum of the obtained MBN signals in the static regime illustrated in Figure 10 also demonstrates that the main difference between the static and dynamic MBN are due to the high frequency MBN pulses exceeding the threshold 200 kHz. The contribution of the low frequency MBN pulses is less pronounced and the electromagnetic pulses below 10 kHz represent mainly mechanical vibrations filtered by the employed software [24,25]. It is also worth mentioning that the obtained number of MBN pulses (see Table 1) does not directly refer to the number of DWs in motion or their repetitive interaction with nitrides, since DWs are clustered and their motion occurs in the form of avalanches [26,27].

3.2. Influence of Position in the Chamber

Figure 11 illustrates the distribution function of MBN based on the measurements of 600 pieces after plasma nitriding. This figure shows that the majority of pieces exhibit MBN in the range 100–250 mV. However, a few pieces produce MBN exceeding 300 mV, which is a similar value to that measured in chapter 3.1 in position n.3. All pieces were measured in the static mode but in the frequency range of 70–200 kHz. This frequency range is usually employed for the monitoring of components in real industrial conditions and exhibits enough sensitivity to alterations developed during the manufacturing processes. Detailed analysis, in which MBN is investigated as a function of position in the chamber, is depicted in Figure 12, Figure 13 and Figure 14. The angular dependence of MBN values is employed in order to follow the positioning of the samples in the chamber especially with respect to the position of the venting system, see also Figure 3. It should be noted that the angular distribution of MBN for samples in the lower section (below layer n.6) is nearly the same as that illustrated in Figure 12 for the inner circle. MBN does not exceed 250 mV and is mostly kept below 200 mV. On the other hand, the middle and the outer circles exhibit remarkable heterogeneity in MBN distribution, see Figure 13 and Figure 14. These figures also illustrate that the degree of heterogeneity increases from layer n.6 towards layer n.1. The nitridation process in this case is affected by the presence of the venting system, as shown in Figure 3 (its position is also indicated in Figure 12, Figure 13 and Figure 14). The temperature in this position of the chamber is slightly lower. The lower temperature decreases the rate of diffusion of N in steel, which, in turn, contributes to the lower density of nitrides [28] and higher MBN. The pinning strength of nitrides is very high, since the nitrides are very fine and their size is close to the DWs’ thickness [29,30]. It was also found that the thickness of the compound layer for the components emitting the higher MBN is more, when compared with those emitting the lower MBN. This finding indicates that the contribution of the compound layer, as the region attenuating electromagnetic pulses towards the free surface [31], is only minor and the density of nitrides in the diffusion layer prevails. It should be considered that the lower temperatures decelerate the diffusion of N to the deeper layers. For this reason, more free N can be found on the free surface, which produces the thicker compound layer [6].

4. Conclusions

This study demonstrates that the MBN technique is a promising tool for monitoring of components after the plasma nitriding process. The attenuation of DWs’ irreversible motion, as a result of their interference with nitrides embedded in the diffusion layer, can be easily linked with decreasing MBN. Conditions and processes in which an unacceptable surface state is produced after plasma nitriding can vary. However, the MBN technique should be capable of distinguishing between samples in order to reject components of an unacceptable state. The components emitting MBN of about 250 mV (in the dynamic regime or in the frequency range of MBN pulses from 70 to 200 kHz) can be considered as acceptable. Increasing MBN (especially above 300 mV) indicates the decreased density of nitrides in the diffusion layer and the corresponding lower hardness. In addition, higher PP, of about 1 kA·m⁻¹, and the number of detected pulses, about 35,0000, can be linked with an adequate surface state, whereas a decreasing number of MBN pulses and lower PP indicate decreased matrix hardness. MBN can be easily adapted in automated cycles or in robotic cells. Therefore, the satisfactory precision of multiple and/or repetitive measurements and high sensitivity of the monitoring process can be obtained. However, the critical threshold for the approval or rejection of components should be validated during the preliminary phase, in which surface state is linked with MBN and the extracted MBN features. The MBN technique can reveal risky factors in the automotive industry and the corresponding sustainability of manufacturing processes. A high number of nitrided components can be measured within quite a short time period, but the findings presented in this study should be initially considered.