The effect of heat treatment parameters on the performance of X46Cr13 used in a magneto-mechanical sensor

Abstract

A manufacturer producing torque sensors noted variations in the magnetic performance of their product which they suspected to be the result of variations in the heat treatment process, including temperature differences between torque sensor shafts placed at different locations within the furnace. The hypothesis was justifiable considering the large heat treatment batch sizes. Performance of these sensors is determined by the ‘sensitivity’, i.e. the change in signal (mV) with applied torque (Nm), and the ‘rotational signal uniformity’ (RSU), i.e. the maximum change in signal registered as the shaft is rotated freely for one complete revolution. Relevant studies carried out to date are scarce and only include generic investigations on the effect of heat treatment and chemical composition on the sensitivity. These studies lack microstructural corroboration, ignore the effect of changes in carbon content and individual heat treatment parameters and do not simultaneously consider the magnetic and other properties important for this application. This gave scope to the present investigation. The major aim of this study was to investigate the relationship between the microstructure, magnetic and mechanical properties of martensitic stainless steel X46Cr13, and in the process establish the cause for the variation in torque sensor performance being claimed by the manufacturer. Following this, a secondary aim was to determine the heat treatment parameters that yield the best combination of properties. The study ensued by investigating the effect of heat treatment parameters, namely the austenitising and tempering temperatures and time, and double austenitising, on the microstructure, magnetic and mechanical properties/characteristics. Microstructural characterisation was carried out for each heat treatment condition and involved the use of optical microscopy, SEM with EDAX facility, and XRD with Relative Intensity Ratio (RIR) analysis to measure the austenite grain size, establish carbide type and quantity, and study matrix transformations. B-H curves were plotted for the whole range of microstructures when magnetised in the circumferential and axial directions. Bulk hardness, tensile, impact and rotation-bending fatigue testing, were also carried out. Measurements of the torque sensor sensitivity and RSU, were related to the microstructure and B-H curves. The three-way relationship was discussed and heat treatment parameters that would yield the best combination of magnetic and mechanical properties/characteristics required for torque sensor applications, were chosen. The material was supplied in the annealed state and consisted of a ferritic matrix with circa 8% M23C6 carbides. In samples tempered at 350°C/375°C, increasing the austenitising temperature from 980°C to 1055°C increased the prior austenite grain size from 14 – 25 µm, decreased the quantity of M23C6 carbides to around 3%, increased the carbon dissolved in the martensitic matrix and increased the bulk hardness from 51.8 to 55 HRc. Double austenitising at 1040°C/980°C led to the finest grain size (11 µm), a homogenous distribution of M23C6 carbides and a bulk hardness that approached that of samples austenitised at 980°C. Increasing the austenitising time from 45 minutes to 135 minutes had a minimal effect on the microstructure. In samples austenitised for 45 minutes at 1040°C, tempering at 350°C/375°C for 2 x 2 hours resulted in structures having a martensitic matrix, up to circa 5% partitioned austenite and M3C carbides. Increasing the tempering temperature to 500°C/525°C rid the structure of austenite and led to sensitisation, manifested by the formation of grain boundary carbides. Tempering at 600°C/625°C transformed the needle-like M3C carbides to M23C6 carbides and intensified the network of grain boundary carbides. The martensite-to-ferrite transformation occurred at 600°C/625°C when tempering for 2 x 2 hours and at 500°C/525°C when tempering for 2 x 6 hours. In general, quenched and tempered samples having a martensitic structure were more permanently magnetisable than structures in the annealed state. Changes in austenitising parameters had a limited effect on the B-H curve of X46Cr13. Despite the matrix remaining martensitic, increasing the tempering temperature from 350°C/375°C to 500°C/525°C rendered the material easier to magnetise circumferentially. This was attributed to a lower quantity of M3C carbides and localized areas of lower hardness. A similar change in B-H curve was observed as the tempering temperature was raised further to 600°C/625°C, resulting from the matrix transformation to ferrite and corresponding drop in hardness to 35 HRc. A change in B-H curve did not always result in a corresponding change in torque sensor sensitivity. In fact, the change in B-H curve following tempering at 500°C/525°C for 2 x 2 hours was not accompanied by a change in sensitivity. On the other hand, increasing the tempering temperature to 600°C/625°C for 2 x 2 hours, shifted both B-H curve and sensitivity. In this respect, the bulk hardness and the Young’s Modulus were found to be better indicators of changes in sensitivity than the B-H curve. Although a low RSU could be achieved with both ferrite and martensite structures, the RSU of martensitic samples was more susceptible to small inadvertent variations in the magnetisation procedures. Austenitising for 45 minutes at 1040°C, followed by tempering at 350/375°C, resulted in the highest yield, tensile and fatigue strength, but the poorest ductility and impact energy. Double austenitising marginally improved the fatigue strength and the ductility when compared to austenitising at 980°C. This was attributed to the better distribution of M23C6 carbides. Irrespective of the austenitising treatment, tempering at 500°C/525°C decreased the tensile and fatigue strength whilst improving the ductility. This was attributed to the formation of softer areas in the microstructure. In summary, it was shown that relatively high magnetic and mechanical performance can be achieved over a wide process window and that stringent control of the heat treatment parameters is necessary only when tempering close to the martensite-to-ferrite transformation temperatures. The sensitivity was found to be dependent on the matrix structure and variation is only to be expected when tempering close to transformation temperatures. The variation in RSU was attributed to magnetisation setting and this was noted to be more of a problem for martensitic samples. A ferritic structure yielded the best sensitivity and a low RSU but exhibited much poorer mechanical properties. Considering this, the best combination of properties/characteristics may be achieved when austenitising at 1040°C for 45 minutes and tempering at 350°C/375°C for 2 x 2 hours. Double-austenitising may be attractive if a small decrease in magnetic performance can be sacrificed for an improvement in ductility.