Plasma-Assisted Nitriding in Low-Frequency Inductively Coupled Plasma Enhanced with Ferromagnetic Cores

Introduction

Recent development in plasma nitriding technology has led to the transition from high pressure (~100–1000 Pa) DC glow discharge nitriding to low-pressure (~0.1–10 Pa), high-density plasma-assisted nitriding (PAN) concept [1]. The PAN concept is based on the uncoupling of plasma generation from a treated substrate, by using of an “external” plasma source and independent substrate biasing. In this case, the control of ion flux and ion energy on the substrate surface is much more flexible than for DC glow discharge nitriding. Also, using the external plasma source gives a possibility to produce low-pressure (<10 Pa), high-density (>10¹⁰ cm^‒3) plasma. In the low-pressure mode, the substrate sheath is not collisional, therefore ions gain the full energy corresponding to the substrate bias, while the high density of nitrogen ions and excited particles enhances the speed and quality of nitriding process.

A variety of PAN methods and arrangements is discussed in Ref. [1] (thermionic arc discharges, electron cyclotron resonance systems, thermionically assisted DC triode systems, etc.), but we are going to focus on inductively coupled plasma (ICP) as the source of ions and excited atoms for nitriding process. Radio-frequency (RF) ICP sources are an important class of high-density, low-pressure plasma sources, and well suited to a wide range of plasma applications. For example, in contrast to thermionic arc discharges, the life-time and the current strength of RF ICP sources are not limited with the processes of cathode erosion, allowing significantly increase the ion density in plasma. As a result, a high nitriding rate can be achieved in RF ICP [2].

Therefore, RF ICP sources seem to be a good choice for ion-plasma nitriding, but have a few disadvantages: high current frequency (typically, 13.56 MHz), insufficient magnetic coupling between ICP coil and plasma (k≈0.2–0.7) thus low power transfer efficiency from a power supply into plasma [3]. In some cases, the frequency of RF ICP can be decreased down to a few hundred kilohertz (for example, 500 kHz [2]), but it is still significantly higher than a frequency range of cheap and simple power supplies for induction heating (about 10–100 kHz), widely used in metalworking industry.

These disadvantages of the conventional RF ICP sources can be improved with enhancement of magnetic coupling between the ICP coil and plasma by ferromagnetic cores [3]. High magnetic flux density in the core allows decreasing the ICP frequency down to a very low frequency range (10 kHz [4], or even down to 60 Hz [5]) and significantly improving the power transfer efficiency. Such low-frequency (LF) ICP sources could be used in many areas of plasma processing, including PAN technology. The aim of this work is to investigate the process of plasma nitriding in the low-frequency, low-pressure inductively coupled plasma enhanced with ferromagnetic cores. As an object of the study, austenitic stainless steel has been chosen, which is known to be hard to nitride [6, 7]. In particular, AISI 321 stainless steel similar to AISI 304 but with a small addition of titanium preventing weld decay, which is widely used for welded structures, was used as the model object to reveal the features of PAN in LF ICP.

Experimental setup

A scheme of experimental setup is shown in Figure 1. Gas discharge chamber 1 is made of stainless steel water cooled sections with inner diameter of 230 mm. The sections are sealed and dielectrically separated with silicon rubber gaskets. The length of the chamber is 1 meter. A narrow (internal diameter of 40 mm) U-shaped part 2 of the discharge chamber, together with the main part 1, forms a toroidal path for discharge current. Ferrite cores 3 significantly enhance magnetic coupling between the inductive discharge and ICP coil 4, so a power supply for induction heating with a frequency range of 50–100 kHz 5 is used for discharge generation. The power supply is connected to the ICP coil through a matching network 6 (variable LC circuit). Rogowsky coil 7 is used to measure discharge current. Voltmeter 8 is used to measure discharge voltage and to determine electric field strength in the discharge chamber 1. MKS baratron 626a 9 is used to measure gas pressure in the discharge chamber. Double electric probe 10 is used to determine charged particles density. Stainless steel samples 11 with the size of 20×20×1.5 mm (AISI 321 stainless steel) are placed inside the discharge chamber. Sample temperature is measured with a thermocouple probe type k 12, with 3 mm stainless steel sealed sheath. The tip of the probe is tightly screwed into the sample (Figure 1), while the rest of the probe is hidden in a sample holder 13 (quartz tube). Sample bias is regulated with a DC power supply 14, connected to the discharge chamber and to the sample through the sample holder. To pump out the discharge chamber, a fore pump 15 is used.

Figure 1:

Experimental setup. 1 – Gas discharge chamber (internal diameter of 230 mm), 2 – U-shaped part of the discharge chamber (ID of 40 mm), 3 – ferrite cores, 4 – inductor (ICP coil), 5 – power supply (50–100 kHz, 500 V), 6 – matching network (variable LC circuit), 7 – current transformer (Rogowsky coil), 8 – voltmeter, 9 – MKS baratron 626a, 10 – double electric probe, 11 – stainless steel sample, 12 – thermocouple probe, 13 – sample holder (quartz tube); 14 – DC power supply, 15 – fore vacuum pump.

The nitriding process is divided into four stages: heating by gas discharge up to an intermediate temperature at a floating potential of the sample (about 5−7 min), heating by gas discharge and ion bombardment at the sample bias up to the process temperature (about 5 min), nitriding stage at the process temperature (15 or 55 min), cooling down to a room temperature at fore-vacuum pressure (a few hours).

To observe the morphology and thickness of the nitrided layers, cross-sections of samples are prepared using Struers Discotom-65 cutting machine and pressed in a polymer matrix using Buehler SimpliMet 1000 mounting press. After that, the cross-sections are polished using Buehler AutoMet 300 grinder-polisher, with a gradual reduction of abrasive particles size. Rough grinding is performed using abrasive wheels with the grain size of 120–50 mkm, fine polishing is performed using 9–0.5 mkm diamond emulsions. To reveal the structure of layers, chemical etching is performed in 30 % HCl, 30 % HNO3 and 40 % H2O solution. Finally, an optical microscope (Carl Zeiss Axio Observer A1m) is used to observe the morphology and thickness of the nitrided layers.

Microhardness of the nitrided layers is measured with a Wolpert micro vickers tester 402MVD with a load of 10 g. To determine the microhardness, a few measurements are done in a middle part of a layer, and then an averaged value of microhardness is calculated. The microstructure analysis of the sample surface is carried out by X-ray diffraction (ARL X’TRA).

Results and discussion

Plasma-assisted nitriding of the AISI 321 stainless steel samples has been performed in nitrogen LF ICP enhanced with ferrite cores, for nitrogen pressure of 7 Pa and discharge current of 7–60 A. Electric field strength of the discharge was in the range of 0.85–0.49 V/cm (a falling volt-ampere characteristic), therefore an average discharge power density was in the range of 14–70 mW/cm^‒3. The ion density determined with the double probe was varied in the range of 10¹⁰–10¹¹ cm^‒3. These values are comparable with the ion densities in low-pressure hot-cathode arc discharges [1], which were used for plasma-assisted nitriding of stainless steels AISI 321 [6] and AISI 304 [7]. Optical emission spectroscopy measurements of RF ICP plasma for comparable conditions (nitrogen pressure of 10 Pa, discharge power density of 160 mW/cm^‒3) indicate that only N₂⁺ ions are present in plasma [8], therefore we assume the same plasma composition is realized in the LF ICP.

DC sample bias of –500 V and –750 V was used to extract and accelerate the nitrogen ions to the surface of stainless steel sample. For a fixed value of sample bias, ion flux density and sample temperature were regulated by changing LF ICP current strength.

According to Ref. [9], the ion energy of 300 eV is enough for sputtering of a thin oxide layer forming on the hot steel surface even in the case of very low oxygen content in plasma (about 10^‒6 Pa). Such thin oxide layers significantly decrease the speed of nitrogen diffusion and, therefore, the speed of nitriding. However, to ensure the removal of oxide layer, we use the negative bias higher than –300 V.

Figures 2–5 show the optical micrographs of a cross-section of nitrided samples, treated for 20 min at DC sample bias of –500 V and sample temperatures of 470–590 °С, for various sample current densities. Also, averaged values of layer microhardness are presented in the figure captions.

Figure 2:

Cross-section of a nitrided sample, treatment time 20 min, nitriding temperature 470 °C, sample bias –500 V, sample current density 3.2 mA/cm², layer microhardness 7.7 GPa.

Figure 3:

Cross-section of a nitrided sample, treatment time 20 min, nitriding temperature 510 °C, sample bias –500 V, sample current density 2.0 mA/cm², layer microhardness 9.0 GPa.

Figure 4:

Cross-section of a nitrided sample, treatment time 20 min, nitriding temperature 565 °C, sample bias –500 V, sample current density 2.6 mA/cm², layer microhardness 9.1 GPa.

Figure 5:

Cross-section of a nitrided sample, treatment time 20 min, nitriding temperature 590 °C, sample bias –500 V, sample current density 3.3 mA/cm², layer microhardness 8.0 GPa.

Even for the short time of treatment, nitrided layers with the thickness of up to 40 μm and microhardness of up to 9 GPa are observed. For comparison, low-pressure (0.1 Pa) hot-cathode arc treatment [6] for the same time at nitriding temperature of 520 °C caused only insignificant (up to 2.5 GPa) hardening of the surface layer, and the nitrided layer thickness was only 9 μm. This difference may be caused by a lower sample bias used in [6] (–300 V) and a lower ion flux density on the sample surface respectively (unfortunately, authors didn’t present data on the sample current density). As shown in Ref. [8], a higher flux density improves the nitriding efficiency and microhardness of stainless steel. It is necessary to underline that short time nitriding of austenitic stainless steel using DC glow discharge treatment at a high pressure (500 Pa, NH₃) and nitriding temperatures of 520–560 °C produce only very thin nitrided layers of a few micrometers with microhardness of about 5 GPa [10].

Figures 6–8 show the optical micrographs of a cross-section of nitrided samples, treated for 60 min at DC sample bias of –750 V and nitriding temperatures of 490–625°С, for various sample current densities. For 1 hour time of treatment, nitrided layers with the thickness of up to 50 μm and microhardness of up to 13 GPa are observed.

Figure 6:

Cross-section of a nitrided sample, treatment time 60 min Nitriding temperature 490 °C, sample bias –750 V, sample current density 1.4 mA/cm², layer microhardness 13.6 GPa.

Figure 7:

Cross-section of a nitrided sample, treatment time 60 min Nitriding temperature 565 °C, sample bias –750 V, sample current density 2.4 mA/cm², layer microhardness 10.1 GPa.

Figure 8:

Cross-section of a nitrided sample, treatment time 60 min Nitriding temperature 625 °C, sample bias –750 V, sample current density 3.3 mA/cm², layer microhardness 12.0 GPa.

For comparison, hot-cathode arc treatment [6] for the same time at DC sample bias of –300 V and nitriding temperature of 520 °C formed a nitrided layer with the thickness of 24 μm and microhardness of about 9 GPa (stainless steel AISI 321). For stainless steel AISI 304 treated at nitriding temperature of 580 °C, nitrogen pressure of 0.6 Pa, sample bias of –700 V and sample current density of 3.2 mA/cm², nitrided layer with the thickness of 48 μm and microhardness of about 12 GPa was observed [7]. DC glow discharge nitriding during 1 hour at the nitriding temperature of 540 °C produces layers with the thickness of about 9 μm and microhardness of about 10 GPa [10].

Figure 9 shows X-ray diffraction patterns of the surface of treated samples, for 60 min treatment at 530 °C (1) and for 20 min treatment at 590 °C (2). In the case of 60 min treatment at 530 °C, X-ray diffraction pattern has peaks related to the nitrogen-expanded austenitic phase (γ_N), iron nitride Fe₄N, and very broad low-intensity peaks of chromium nitride CrN.

Figure 9:

X-ray diffraction patterns: 1 − 60 min, –750 V, 2.0 mA/cm², 530 °C. 2 − 20 min, –500 V, 3.3 mA/cm², 590 °C.

Detection of the nitrogen-expanded austenite at the elevated temperature of 530 °C is an unexpected result. Generally, it is supposed that the supersaturated solid solution of nitrogen in the expanded austenite lattice is formed at the treatment temperatures below 450 °C [11], while the higher temperatures lead to thermal decomposition of expanded austenite γ_N with chromium nitride precipitation [12]. However, nitrogen-expanded austenite formation at the elevated temperatures is also reported for short-time nitriding [10, 13]. The nitrided layer mainly composed of nitrogen-expanded austenite phase was formed on the surface of AISI 316 L austenitic stainless steel by plasma nitriding at 540 °C in short time (15−60 min) DC glow discharge nitriding [10]. In situ XRD measurements for low energy ion beam nitriding of austenitic stainless steel 316Ti at 550 °C [13] also revealed a fast formation of the expanded austenite layer at the initial stage (20–30 min) of the nitriding process, with the subsequent decomposition of γ_N phase. Therefore, we can assume that γ_N phase was not totally decomposed in the case of 60 min nitriding at 530 °C.

In the case of 20 min nitriding at 590 °C, X-ray diffraction pattern contains peaks related to the austenitic phase (γ) with a face centred cubic structure, peaks related to chromium nitride and a broad peak related to ferritic α-Fe phase. The absence of expanded austenite and the presence of γ-Fe, α-Fe and CrN indicates that the thermal decomposition of γ_N phase is almost completed [12]. Recent results on the dynamics of γ_N decomposition [14] show that it takes about 15 min for annealing of nitrided AISI 304 stainless steel at 575 °C, therefore the time of γ_N decomposition should be less than 15 min at 590 °C.

Comparing the low-pressure LF ICP nitriding with the treatment in low-pressure hot-cathode gas arcs [6, 7] and a high-pressure DC glow discharge nitriding [10], we observe a significant enhancement in the speed of nitriding of austenitic stainless steels for the case of plasma-assisted nitriding by the “external” low-pressure, high-density plasma sources. In the case of short-time treatment (20 min), the best results are achieved for the low-pressure LF ICP nitriding, probably due to high current density (up to 3.3 mA/cm²) on the sample surface. The same ion flux density was achieved in a low-pressure hot-cathode arc [7], but unfortunately they did not present data for the case of 20 min treatment. In the case of 60 min treatment, our results are comparable with the results presented in Ref. [7], while layer microhardness achieved in the work [6] is a bit smaller (9 GPa vs 10–13 GPa), probably due to a lower sample bias and ion flux density respectively.

Finally, it is necessary to underline that there is no physical limitation either on LF ICP current strength or discharge chamber diameter, the setup can be scaled up for plasma-assisted nitriding of objects significantly larger than the size of the samples.

Conclusions

The fast formation of nitrogen-expanded austenite at the elevated temperatures of 450−550 °C can be used to enhance the speed of plasma nitriding of stainless steels. To achieve a large thickness and high microhardness of the nitrided layers in short (<60 min) time, not only the elevated (>450 °C) temperature but also a high ion flux density is needed, while the ion energy should be high enough (>300 eV) for sputtering of a thin oxide layer limiting the speed of nitrogen diffusion. Ferromagnetic-enhanced inductively coupled discharges effectively generate low-pressure (0.1−10 Pa), high-density (10¹⁰−10¹² cm^‒3) nitrogen plasma allowing to realize the above-mentioned conditions and to reduce the time of the nitriding process.