Niobium Nitride Cavitation Erosion Resistance: An Approach on the Gas Mixture Influence in Plasma Nitrided Niobium Surfaces

Abstract

This work shows an approach on the role of the gas mixture used in the pulsed DC plasma nitriding aiming to enhance the niobium cavitation erosion resistance through the formation of niobium nitride on the treated surfaces. For this purpose, nitriding was carried out at 1353 K (1080 °C) for 2 h, under a pressure of 1.2 kPa (9 Torr), and a 5 × 10⁻⁶ Nm³s⁻¹ (300 sccm) flow rate for three distinct gas mixtures, namely 30% N₂ + 50% H₂ + 20% Ar, 50% N₂ + 30% H₂ + 20% Ar, and 70% N₂ + 10% H₂ + 20% Ar. Surfaces were comparatively characterized before and after nitriding through scanning electron microscopy (SEM), X-ray diffractometry, 3D roughness, and nanoindentation hardness measurements. The cavitation erosion test was carried out in accordance with ASTM G32-09, obtaining the cumulative mass loss (CML) curve and the average (AER) and maximum (MER) erosion rate of the tested surfaces. Surfaces showed multiphase layers mainly constituted of ε-NbN and β-Nb₂N nitride phases, for the three distinct gas mixture conditions investigated. A CML of 25.0, 20.2, and 34.6 mg, and an AER of 1.56, 1.27, and 2.16 mg h⁻¹ was determined to the 960 min (16 h) cavitation erosion testing time, for NbN surfaces obtained at the 30% N₂, 50% N₂, and 70% N₂ gas mixture, respectively. In this case, the nominal incubation period (NIP) was 600, 650, 550 min, and the maximum erosion rate (MER) was 4.2, 3.4, and 5.1 mg h⁻¹, respectively. Finally, the enhancement of the cavitation erosion resistance, based on the NIP of the NbN surfaces, regarding the Nb substrates (with NIP of ≈100 min), was up ≈6 times, on average, thus significantly improving the cavitation erosion resistance of the niobium.

1. Introduction

Studies on the cavitation erosion behavior of engineering ceramics have presented a significant growing interest in the last few years, mainly in applications in power generation plants, such as evidenced in materials for valves and seals with use in boiler feed pumps at nuclear power systems, as an example, as well as in the aircraft and aerospace industries [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27].

Such interest is due to some typical characteristics of ceramics, mainly regarding their mechanical properties. According to Preece (1979), as indicated in ref. [1], due to their high hardness and excellent wear resistance, ceramics can be considered strong candidates for applications that require high cavitation erosion resistance.

The cavitation erosion behavior of a great number of different structural ceramics (namely, aluminas, zirconias, silicon carbides, silicon carbide–boron carbide composites, boron carbide, and sialon) was investigated in detail by Tomlinson and Matthews (1994), as shown in ref. [1]. In such a work, measured incubation periods varying from 15 min for reaction-bonded silicon carbide to 150 min for 99.9% alumina, and maximum erosion rates varying from 0.11 mg h⁻¹ for sialon to 8–25 mg h⁻¹ for 97.5% alumina were determined for specimens eroded for up to 30 h in distilled water at 18 °C, using an ultrasonic vibratory apparatus operating at 20 kHz, for a peak-to-peak amplitude of 50 μm. It was observed that the incubation period critically depends on the surface finishing, and concluded that the appropriate material selection and a suitable surface preparation can increase the incubation period by over 1000%, which is crucial for the use of materials in valves and seals [1]. It was also highlighted in conclusions that ceramics with high resistance to cavitation erosion (in terms of both the incubation period and erosion rate) would have ultra-fine microstructure, no porosity, high fracture toughness, and a smooth surface finish [1].

Similarly, ceramic phases showing high cavitation erosion resistance can also be presented on the surfaces of metallic parts, which explains the significant R&D efforts during the last few decades, aiming to produce such type of mechanical components. In this case, the application of plasma-assisted surface treatments carried out in metallic substrates has deserved special attention at manufacturing practices. Such practices comprise several coating techniques, resulting in the production of ceramic-phase formed surfaces, such as physical vapor deposition, chemical vapor deposition, magnetron sputtering, high-power impulse magnetron sputtering, double cathode glow discharge, cathodic arc physical vapor deposition [11,12,13,14,15,16,17,18,19,20,21,22,23,24], plasma spray [24,25,26,27], and even thermochemical treatments like plasma nitriding [28,29,30]. It should be noted that the plasma nitriding technique does not constitute in a coating (deposition) process, which is characteristically a diffusion-controlled process, thus the kinetics of the layer growth depends on the process (treatment) temperature and time [30]. Note that in ref. [22], NbN nanoceramic coatings deposited on Ti-6Al-4V substrates exhibited extremely low mass loss rates, almost independent of exposure time, and not exceeding 0.3 mg cm⁻² after cavitation exposure testing as long as 15 h.

The cavitation phenomenon was initially studied in ship propellers by L. Rayleigh, in 1917 [28]; thus, for over a century, it has deserved special attention in the engineering field [29]. Interestingly, a significant new boost in R&D related to the problem of cavitation erosion has recently been achieved by the aircraft industry, in an indirect way. It was shown [24] that the damage mechanism at the water droplet erosion, occurring in turbofan engines (widely used in commercial aircraft nowadays), is similar to the one observed in the cavitation erosion beginning stage, namely in the incubation period. Such a finding acts as a driving force for the research and development of new high-temperature cavitation erosion–resistant materials. Therefore, the cavitation erosion behavior of niobium substrates, and the modification of niobium surfaces by plasma thermochemical treatments, have also been investigated [30,31,32,33,34], keeping in mind that this typical refractory metal is a very stable nitride/carbide former.

It should be emphasized that metallic niobium substrates showing niobium nitride ceramic phase surfaces were successfully obtained through plasma nitriding treatment, as shown in refs. [30,31]. As can be observed in ref. [30], the niobium plasma nitriding kinetics was investigated for relatively low temperature and short time conditions (on the 1253–1453 K and 1–4 h ranges, using a gas mixture of 90% N₂ + 10% H₂). Results confirmed that the niobium plasma nitriding is a diffusion-controlled process. The obtained nitride (compound) layer is multiphase, being constituted of ε-NbN, γ-Nb₄N₃, and β-Nb₂N nitrides, and the compound and diffusion layer growth activation energy was 97.11 and 120.11 kJ mol⁻¹, respectively. The hardness distribution all over the nitrided layer (i.e., the transition between the compound and diffusion layer) is very sharp, varying from 18.5 GPa (≈1850 HV) at the surface (compound) layer to decreasing values varying from ≈480 to ≈100 HV along the diffusion layer, depending on the studied condition. The Sa and Sz roughness indicated increasing measured values for longer treatment times and higher temperatures, with the surface morphology showing island-like atom clusters. Apparently crack-free compound layers were also yielded, and such a result highlights the importance of the use of relatively low treatment temperatures (below the N-Nb system eutectoid temperature that occurs at 1523 K), and relatively short treatment times.

Promising results were also previously presented by the authors in ref. [33], by first comparing the results of the cavitation erosion behavior of plasma nitrided and non-nitrided niobium surfaces. In such a work, nitriding treatment was conducted at 1080 °C, for a 90% N₂ + 10% H₂ gas mixture, using a flow rate of 5 × 10⁻⁶ Sm³s⁻¹, and pressure of 1200 Pa (9 Torr), for a total time of 4 h comprised by two treatment steps of 2 h each. It is worth mentioning that, comparatively, the two main differences in terms of the nitriding procedure investigated here were the use of Argon in the gas mixture (as discussed ahead), and the realization of the nitriding treatment in a single thermal cycle for a shorter time (only 2 h).

Regarding the niobium nitride mechanical properties, nanocrystalline NbN films produced by the magnetron sputtering technique have shown high hardness (36.6 GPa) and high elastic modulus (457 GPa) values, and good elastic recovery (71%) [35]. These results are clearly different from those observed for pure niobium substrates, as expected, which apparently show unlimited ductility, hardness of the order of only 1 GPa, and an elastic modulus of ≈100 GPa, as observed in refs. [30,31,34,36,37].

So, in this work, an approach on the role of the N₂-H₂-Ar gas mixtures used in pulsed DC plasma nitriding aiming to enhance the niobium cavitation erosion resistance was carried out. Note that DC plasma under abnormal glow discharge conditions [30,31,32,33,38,39,40,41,42], mainly due its green and ecological-friendly characteristics, nowadays tends to present increasing use of the materials processing applications, which is the main focus of the present investigation (i.e., the cavitation erosion behavior of niobium nitride–based surfaces successfully obtained via pulsed DC plasma nitriding). It should be mentioned that this work is the first of a set of three, which approach in detail the subject of the NbN cavitation erosion behavior obtained via plasma nitriding (the other two works comprise an approach on the nitriding time influence, and an approach on the nitriding temperature influence, to be presented in the future).

2. Materials and Methods

Niobium specimens were prepared from 90% reduction cold-worked commercially pure niobium plates. Before the nitriding treatments, samples were cut using wire electrical discharge machining to dimensions of 30 × 20 × 4 mm³ and were annealed to a hardness of 80 HV. The annealing was performed in a vacuum furnace at 1273 K (1000 °C) for a time of 60 min, under 1.33 Pa (10⁻² Torr) pressure.

Annealed niobium samples were ground with 220, 320, 400, 600, and 1200 SiC sand-paper. The final finishing step of the testing surfaces comprised polishing using 1.0 µm diamond paste, and finally polishing with 0.05 μm Al₂O₃ suspension. Chemical etching via sample immersion was used to reveal the microstructure of the as-annealed niobium substrates, which showed typical equiaxed-grains (see Figure 1), and the obtained nitrided layers and its thicknesses, in the polished substrate cross sections. The microstructures were revealed using the etchant ASTM 160 SWAB 15Ss, composed of 40 mL HF, 30 mL H₂SO₄, 10 mL HNO₃, and 100 mL H₂O, during 15 s application at the polished surface. The final aspect of a typical polished niobium sample, before performing the plasma nitriding treatment, is shown in Figure 1. Also shown in Figure 1 is an in situ view of the niobium sample plasma nitriding.

Plasma nitriding was conducted at 1353 K (1080 °C) for 2 h, under a pressure of 1.2 kPa (9 Torr), and a 5 × 10⁻⁶ Nm³s⁻¹ (300 sccm) flow rate for three distinct gas mixtures, namely 30% N₂ + 50% H₂ + 20% Ar, 50% N₂ + 30% H₂ + 20% Ar, and 70% N₂ + 10% H₂ + 20% Ar. The cathode voltage, and the square waveform period of the pulsed DC power supply were kept constant at 670 ± 15 V, and 240 μs, respectively. The treatment temperature was controlled by changing the switched-on time or the duty cycle of the pulse (ton), in the glow discharge pulsed power supply. The sample temperature was measured using a 1.5 mm diameter type-K chromel-alumel thermocouple inserted to a depth of 15 mm inside the sample. Figure 2 shows the schematic representation of the plasma nitriding system used in this work.

The plasma treatment of the samples was divided in two steps: (1) cleaning in H₂ glow discharge at 873 K (600 °C), for 10 min, and 798 Pa (6 Torr) pressure, carried out before nitriding; and (2) nitriding, after plasma cleaning, adjusting the gas pressure to the specified value, and heating the sample via plasma species bombardment to the specified nitriding temperature, at a heating rate of about 0.32 K s⁻¹ (°C s⁻¹).

The cavitation erosion experiments were based on the ASTM G 32-10 standard [43], with the cavitation erosion testing performed up to 960 min (16 h). The cumulative mass loss (CML) as a function of the cavitation time, the average erosion rate (AER), the incubation period (IP), the nominal incubation period (NIP), the transition from the incubation period to the acceleration stage, and the maximum erosion rate (MER) for the studied NbN and Nb surfaces were determined. Note that for the times used here (up to 16 h), the performed testing has reached the transition between the acceleration and the maximum erosion rate (MER) stage, which means that the MER stage has only been initiated, thus the results for the MER considered in this work are only approximate, in fact. It is to be remembered that the NIP is geometrically obtained by extending the straight of the MER line up to the x-axis of the CML curve graph (see Figure 8, ahead), assumed as straight that one as obtained by adjusting the last three measurement points (in this case, for the testing times of 720, 840, and 960 min), for each CML curve. This assumption was considered here, as longer testing times would be necessary to correctly define the true straight MER line for each case (for additional details, see ref. [34]).

For the cavitation erosion testing, an ultrasonic generator KLN type 500 was used for an indirect method test condition, with the frequency and the peak-to-peak amplitude specified to 20 ± 2 kHz and 50 ± 1.0 μm, respectively. The reproducibility of the results was confirmed by the repetition of each performed experiment at least once, since the dispersion of the measurement values were shown to be very small after a first characterization, which was repeated twice. The temperature of the distilled water bath was maintained at 25 ± 2 °C. The surface of the test sample was immersed in 12 ± 4 mm of distilled water, and a distance of 0.5 ± 0.005 mm was maintained from the tip of the sonotrode to the sample. The mass loss of the test samples was carried out using a 0.1 mg resolution balance, with each mass measurement performed three times to ensure the results reproducibility. The vibration frequency was 20 ± 0.2 kHz. The sonotrode tip was manufactured from AISI 304 austenitic stainless steel with a diameter of 15.9 mm. Figure 3 shows the cavitation erosion testing experimental apparatus used here.

The nitrided samples were characterized by scanning electron microscopy (SEM) using both the secondary electrons (SE) and backscattered electrons (BSE) detectors, energy-dispersive x-ray microprobe analysis (EDS) and X-ray diffractometry (XRD). Samples were prepared using the conventional metallographic procedure. The morphology and the phases presented in the treated surfaces were determined from SEM images and XRD measurements, respectively.

The SEM characterizations were carried out by means of two microscopes: one SEM TESCAN—VEGA3 LMU equipped with an EDS Oxford probe and software AZ Tech, and the other one an SEM ZEISS—EVO-MA15 equipment.

The XRD patterns were obtained using a Shimadzu XRD 7000 X-ray diffractometer, with CuKα radiation, in Bragg–Brentano configuration, for a 2θ angle range of 30–120°, a step of 0.02° s⁻¹, and a 5 s step time. Characterization of the studied surfaces comprising 3D roughness and nanoindentation hardness measurements was also carried out.

The nanoindentation marks, performed using a Berkovitch nanoindenter, were analyzed aiming to confirm the brittle character of the nitrided surfaces. For the nanoindentation characterizations, the XP-MTS System equipment was used. The procedure adopted to obtain the average (and the respective dispersion, based on the standard deviation) hardness and elastic modulus values as a function of the contact depth, comprised the determination of a measurement matrix of 25 (5 × 5) indentations with 100 μm spacing, and 12 charge–discharge cycles, using loads up to 400 mN (40 gf), and 10 s loading times.

The Sa, Sq, Ssk, Sku, Sz, Sv, Sp, and Sds area roughness (or surface texture) parameters were measured using a Talysurf CCI—Lite Non-contact 3D Profiler equipment, from Taylor Hobson, showing a resolution of 0.01 nm in the height ordinate. The average and standard deviation values of such surface parameters were determined from six measurements per sample, with three characterizations on each sample side. The seven former parameters—namely the Sa, Sq, Ssk, Sku, Sz, Sv, and Sp—are considered height parameters, which comprise, along the height ordinate, the statistical distribution of height values. The latter one (i.e., the Sds) is a feature parameter. Based on the ISO 25178 standard [44], the 3D roughness parameters, determined here, can be defined as follows: (1) Sa is the arithmetic mean or the absolute values in the measured area for peaks height and valleys depth; (2) Sq is the root-mean-square height (or the root-mean-square value of height ordinate values within the reference surface); (3) Ssk is the ‘Skewness’, which assumes a value of zero (Ssk = 0) for a symmetrical height distribution in relation to the average line, regarding both the peaks height and valleys depth; it is positive (Ssk > 0) in relation to the average line for surfaces with high peaks and/or filled valleys; and it is negative (Ssk < 0) in relation to the average line for surfaces with deep valleys and/or removed peaks; (4) Sku is the ‘Kurtosis’, which assumes a value of 3 (Sku = 3) when the surface roughness height distribution is reflected in a normal probability density curve; for Sku > 3, spiked is the form of the curve height distribution in relation to the average line (that means the occurrence of many high peaks and many deep valleys); and for Sku < 3, the form of the surface roughness height distribution is squashed (meaning the occurrence of few high peaks and few deep valleys); (5) Sz is the average of the maximum height of the surface (or the height between the highest peak and the deepest valley); (6) Sp is the maximum peak height (or the height between the highest peak and the medium plane); (7) Sv is the maximum valley depth (or the deep between the deepest valley and the medium plane); and (8) Sds is the density of summits or peaks (expressed in the unit of ‘(peaks) mm⁻²’ or simply ‘mm⁻²’), assuming that a determined point is considered a peak if it is higher than its eight neighbors. In practice, a larger number means the more points of contact with other objects.

3. Results

Figure 4a,b comparatively show the aspect of a sample with its typical surface finishing attained after polishing (before the treatment), and after nitriding for the 70% N₂ + 10% H₂ + 20% Ar gas mixture, respectively. The roughness increment observed in Figure 4b, in regard to Figure 4a, is confirmed in the

Table 1. The measured Sa, Sq, Ssk, Sku, Sz, Sv, Sp, and Sds roughness parameters comparatively showing results for the non-nitrided (as-polished, indicated as ‘Nb’) and respective plasma nitrided (indicated as ‘NbN’) conditions. (The measure dispersion over the respective average values is based on the standard deviation of the measurements, as indicated in parentheses.).

Roughness	Sa (µm)		Sq (µm)		Ssk		Sku
Condition	Nb	NbN	Nb	NbN	Nb	NbN	Nb	NbN
30% N2	0.09 ± 0.007	0.16 ± 0.015	0.12 ± 0.009	0.20 ± 0.01	−1.03 ± 0.06	−0.93 ± 0.15	4.95 ± 0.17	4.55 ± 0.50
50% N2	0.10 ± 0.005	0.15 ± 0.009	0.13 ± 0.006	0.21 ± 0.009	−1.40 ± 0.14	−1.06 ± 0.14	6.10 ± 0.60	5.10 ± 0.30
70% N2	0.09 ± 0.005	0.14 ± 0.009	0.12 ± 0.007	0.20 ± 0.009	−1.03 ± 0.05	−0.27 ± 0.12	4.83 ± 0.01	3.90 ± 0.20
Roughness	Sz (µm)		Sv (µm)		Sp (µm)		Sds (mm−2)
Condition	Nb	NbN	Nb	NbN	Nb	NbN	Nb	NbN
30% N2	0.90 ± 0.08	1.58 ± 0.07	0.60 ± 0.03	0.92 ± 0.02	0.290 ± 0.04	0.66 ± 0.05	12,485 ± 540	16,250 ± 780
50% N2	0.94 ± 0.04	1.52 ± 0.12	0.70 ± 0.04	0.86 ± 0.08	0.270 ± 0.08	0.57 ± 0.10	10,640 ± 430	16,210 ± 1180
70% N2	0.87 ± 0.06)	1.49 ± 0.06	0.60 ± 0.03	0.77 ± 0.02	0.270 ± 0.02	0.72 ± 0.06	12,410 ± 280	16,710 ± 260

results. In a generic way, all the gas mixture conditions have led to the increment of the Sa, Sq, Sz, Sv, and Sp roughness parameters, when compared to the beginning condition of the niobium sample surface, before nitriding. As an example, an Sa roughness of 0.16 ± 0.015, 0.15 ± 0.009, and 0.14 ± 0.009 µm (being the measurement dispersion based on the standard deviation) was observed for samples nitrided with gas mixtures containing 30% N₂, 50% N₂, and 70% N₂, respectively, corresponding to values ≈1.78, 1.50, and 1.55 times higher than that of the respective non-nitrided niobium condition (which, initially, showed Sa roughness values of 0.09 ± 0.007, 0.10 ± 0.005, and 0.09 ± 0.005 µm). On the other hand, the negative skewness (Ssk) determined for all the initial polishing conditions of the Nb surfaces agree well with the typical finishing attained after polishing, as shown in Figure 4a, which was also evidenced for all as-polished Nb surfaces.

In addition, note that all the NbN surfaces showed for the Ssk roughness values a deviation tendency in direction to zero, thus a tendency to achieve a symmetrical height distribution of the peaks and valleys, in relation to the average line, with the skewness change from −1.03 (0.05) to −0.27 (0.12) for the NbN surface nitrided with the 70% N₂ gas mixture being more pronounced in this case. Regarding the kurtosis (Sku) of the studied surfaces, both the Nb and NbN surfaces, showing an Sku higher than 3, would be in agreement with the condition of many high peaks and deep valleys. It is worth mentioning that such a result for kurtosis could also be directly related to the sample surface preparation procedure before nitriding, considering the presence of pit-like marks in the Nb surface, as a result of the polishing step (see Figure 4a). Note that very similar results were obtained for the density of summits (Sds), for the three distinct NbN surfaces, which showed 16,250 ± 780, 16,210 ± 1180, and 16,710 ± 260 (peaks) mm⁻² for samples nitrided with gas mixtures containing 30% N₂, 50% N₂, and 70% N₂, respectively. Finally, generically, these results show that the changes in the pulsed DC duty cycle (results not shown) aiming to attain the specified nitriding temperature by varying the N₂/H₂ ratios in the studied gas mixtures do not significantly influence the resulting surface morphologies obtained here.

Figure 5a–d show the XRD patterns measured on the surfaces nitrided for gas mixtures containing 30% N₂, 50% N₂, and 70% N₂, besides a comparative graph of the obtained patterns, respectively. In order to characterize the phases presented on the nitrided surfaces, the JCPDS files for Nb (card no. 02 1108), ε-NbN phase (card no. 25 1361), and β-Nb₂N phase (card no. 39 1398) were used.

Results (Figure 5a–d) confirm the presence of both the ε-NbN and β-Nb₂N phases in the nitrided surfaces. The non-indexed diffraction peaks here should be referred to the γ-Nb₄N₃ phase, keeping in mind the results previously published in ref. [30].

Figure 6a,b show the nanoindentation results determined for the studied surfaces, showing the hardness, and elastic modulus profile as a function of the contact depth, respectively. The results indicate higher maximum hardness values of ≈14 (±3) GPa at a depth of ≈400 nm, for both the 30% N₂, and 50% N₂ gas mixture nitrided surfaces, while only ≈8 (±1) GPa for the 70% N₂ gas mixture nitrided surfaces (see Figure 6a). On the other hand, as indicated in Figure 6b, elastic modulus values of up to ≈460 (±240) GPa at a depth of approximately 100 nm for the surface nitrided to the 70% N₂-containing gas mixture were determined. Comparatively, the measured hardness of the niobium samples was ≈1 GPa (or ≈100 HV), disregarding the work-hardening effect due to the surface polishing step, evidenced here for contact depths of the indenter up to the first ≈800 nm. In this case, the Nb elastic modulus was also ≈100 GPa, agreeing well with the results observed for pure niobium in refs. [30,31,34,36,37].

Figure 7a–c show nanoindentation marks performed with the Berkovitch nanoindenter in the nitrided surfaces for 30% N₂, 50% N₂, and 70% N₂ gas mixture, respectively. The cracks evidenced for the three distinct nitrided surface conditions confirm the brittle character of all nitride-constituted surfaces.

Figure 8 shows the cumulative mass loss (CML) curve as a function of the cavitation time. In addition, from results shown in Figure 8,

Table 2. Values obtained from the Figure 8 results for the CML of the nitrided sample conditions, expressed in ‘mg’, and the respective estimated values for the AER, expressed in ‘mg h⁻¹’.

CML (mg)	Cavitation Time (min)
	0	20	40	60	80	100	120	140	160	180
30% N2	0	0	0	0	0	0	0.1	0.1	0.1	0.1
50% N2	0	0	0	0	0	0	0	0.2	0.2	0.2
70% N2	0	0	0	0	0	0	0	0	0	0
CML (mg)	Cavitation time (min)
	200	220	240	360	480	600	720	840	960	-
30% N2	0.1	0.2	0.2	0.3	1.3	4.2	8.2	15.3	25.0	-
50% N2	0.2	0.2	0.2	0.2	1.2	3.2	6.8	13.6	20.3	-
70% N2	0	0.1	0.1	0.2	1.4	6.2	14.2	23.3	34.6	-
AER (mg h−1)	Cavitation time (min)
	0	20	40	60	80	100	120	140	160	180
30% N2	0	0	0	0	0	0	0.05	0.043	0.037	0.033
50% N2	0	0	0	0	0	0	0	0.086	0.075	0.067
70% N2	0	0	0	0	0	0	0	0	0	0
AER (mg h−1)	Cavitation time (min)
	200	220	240	360	480	600	720	840	960	-
30% N2	0.03	0.054	0.05	0.05	0.162	0.42	0.683	1.093	1.562	-
50% N2	0.06	0.054	0.05	0.033	0.15	0.32	0.567	0.971	1.269	-
70% N2	0	0.027	0.025	0.033	0.175	0.62	1.183	1.664	2.162	-

shows in detail the values obtained for the CML curve, expressed in ‘mg’, and for the average erosion rate (AER), expressed in ‘mg h⁻¹’. A CML of 25.0, 20.2, and 34.6 mg, and an AER of 1.56, 1.27, and 2.16 mg h⁻¹ was determined to the 960 min (16 h) cavitation erosion testing time, for the NbN surfaces obtained at the 30% N₂, 50% N₂, and 70% N₂ gas mixture, respectively.

According to the results in Table 2, considering that for the IP determination the erosion rate is zero or negligible, compared with later stages (in accordance with ASTM G32-10), and assuming here as negligible a maximum mass loss of 0.3 mg, a similar incubation period (IP) of ≈360 min (6 h) was observed for the three studied surface conditions. In this case, a CML of 0.3, 0.2, and 0.2 mg was determined. It is also important to observe that for the 480 min (8 h) testing time, thus initiating the acceleration stage, again, a very similar result was observed for the CML, which showed values of 1.3, 1.2, and 1.4 mg, respectively for the 30% N₂, 50% N₂, and 70% N₂ gas mixture. In addition, the nominal incubation period (NIP) was ≈600 (10 h), 650 (10.8 h), 550 (9.2 h) min, and the maximum erosion rate (MER) was 4.2, 3.4, and 5.1 mg h⁻¹, respectively. Regarding the Nb substrate surface, an NIP of ≈100 min, and an MER of 15 mg h⁻¹ was determined.

Figure 9 and Figure 10 show the morphology of the 600 min cavitation eroded surface for a same region of a sample nitrided with 30% N₂ + 50% H₂ + 20% Ar gas mixture, and the morphology of the 720 min cavitation eroded surface for a same region of a sample nitrided with 70% N₂ + 10% H₂ + 20% Ar gas mixture, respectively, for different magnitudes. It should be noted that for the 600 min cavitation erosion time of Figure 9, the acceleration stage evidenced for the surface nitrided with the 30% N₂ + 50% H₂ + 20% Ar gas mixture had just initiated. Conversely, for the 720 min cavitation erosion time of Figure 10, the acceleration stage evidenced for the surface nitrided with the 70% N₂ + 10% H₂ + 20% Ar gas mixture had just terminated.

4. Discussion

In this work, special attention was given to determine the main surface characteristics that can influence the cavitation erosion behavior of the studied surfaces. Therefore, this section is focused on the discussion of the plasma–surface interaction and the mechanisms leading the surface roughness, the phases constitution, and the mechanical properties of the Nb surfaces to be altered, and, finally, the cavitation erosion behavior of NbN surfaces obtained via plasma nitriding for different N₂ + H₂ + 20% Ar-containing gas mixtures in niobium substrates.

Firstly, regarding the plasma–surface interaction and the main changes on the studied surfaces, the rougher appearance observed for the nitrided surfaces (Figure 4b), also confirmed by the roughness measurements (Table 1), as expected, is due to the mechanism of the nitrided layer formation in the plasma nitriding process, which involves sputtering and redeposition (via backscattering phenomenon [38]) of atoms (Nb and N) at the relatively high processing pressure (9 Torr or 1200 Pa) used in the present treatments, associated with the diffusion mechanism of nitrogen into the Nb substrate. It is very well established that the higher the pressure, the more significant is the backscattering effect, leading to the cluster’s formation between nitrogen and niobium atoms, thus forming islands on the nitrided surface (according to Chapman, 1980 [23]). This mechanism is confirmed from the Table 1 results, which show a generic tendency of roughness increment by comparing the Sa, Sq, Sz, Sv, and Sp values, obtained for both the Nb and NbN surfaces, in the studied conditions. In addition, by critically analyzing the roughness results from a statistics point of view, practically no significant difference was evidenced as a function of the gas mixture composition. In this case, the roughness parameter that better summarizes such a similarity of results would be the Sds roughness, since the density of summits at the NbN surfaces was practically identical for the three studied gas mixture compositions.

On the other hand, the hardness and elastic modulus values presented in Figure 6a,b would be considered an artifact, mainly due to two aspects: the surface roughness attained in the present plasma nitriding process and the NbN-based layer brittle character.

For the former case, two aspects deserve attention. Firstly, note that in the ≈450–650 nm contact depth range (see Figure 6a), increasing hardness values with very high measurement dispersion are observed (showing maximum values of ≈14 GPa, for both the 30% N₂ and 50% N₂ nitriding conditions), and ≈8 GPa (for surfaces nitrided with 70% N₂). Secondly, it should also be noted that in the Figure 6b results, the largest elastic modulus value of ≈460 GPa was obtained for surfaces nitrided with 70% N₂, agreeing very well with the ref. [35] results for NbN films, as previously presented. Thus, comparatively with the two other studied conditions, such highest elastic modulus to the sample nitrided with 70% N₂ does not agree with its respective smallest apparent hardness value of only ≈8 GPa. So, it would be assumed that all NbN surfaces obtained here should present similar real values of hardness and elastic modulus.

For the latter case, it is known that the very low hardness of the pure Nb substrate strongly affects the mechanical behavior of the surface plus subsurface system (or, in other words, the system formed by a hard layer developed on a soft substrate). In this case, under indentation, the cracks formed around the Berkovich indentations (see Figure 7a–c) confirm the brittle character of such a system formed by a ceramic (NbN) layer built on a metallic (Nb) substrate. So, it can be stated that the present system shows a very low toughness surface, despite its supposed good resilience, which is expected for the NbN-based layer.

It should be remembered that the existence of the softer β-Nb₂N phase (see ref. [35]) associated with the ε-NbN phase in the nitrided layer, as observed in the Figure 5 XRD pattern results (confirming the obtainment of a multiphase layer, mainly constituted of the ε-NbN and β-Nb₂N phase) would also reduce its hardness as a whole.

Another aspect to be discussed is how effective would be the gas mixture used in the plasma nitriding process in changing the Nb surface characteristics and properties. Since the argon content was fixed at 20 vol.%, only the N₂/H₂ ratio in the studied gas mixtures was effectively modified, showing values of 0.6 (for 30% N₂ + 50% H₂ + 20% Ar), 1.7 (for 50% N₂ + 30% H₂ + 20% Ar), and 7.0 (for 70% N₂ + 10% H₂ + 20% Ar). The main role of the H₂ gas is acting as a strong oxide reducer, keeping in mind that the niobium oxide (usually Nb₂O₅) is thermodynamically stable, which could be presented in the metallic surfaces exposed to the plasma (see ref. [41,42]), and helping in the glow discharge stabilization due to its relatively low ionization potential [38,39,40]. Argon was added due to its relatively higher atomic mass, being important in activating the sputtering mechanism and surface diffusion, as well as in promoting the cathode heating by momentum transfer, when argon-based plasma species (ions, excited, and neutrals) collide onto the cathode (substrate) surface (see ref. [38,39,40,41,42]). It deserves to be remembered that the sputtering yield is directly proportional to the mass of the impingent atom (see ref. [38,39,40]) bombarding the cathode surface. Keeping in mind such aspects, a nitride layer thickness of 2.46 ± 0.13, 2.62 ± 0.13, and 2.66 ± 0.22 μm was obtained for gas mixtures containing 70% N₂, 50% N₂, and 30% N₂, respectively (results not shown). In a statistics point of view, again there is no significant difference for the thicknesses obtained here.

The same can be stated regarding the effect of the gas mixtures on the cavitation erosion resistance studied here, since generically, very slight influence was observed. Based on the NIP results of the NbN surfaces, when compared with Nb surfaces, an enhancement for the cavitation erosion resistance observed for the NbN surfaces successfully produced by plasma nitriding in Nb substrates varied in the range of ≈6.1–7.2 times, at least for the conditions considered here. Note that the results are substantially better than the first ones of ref. [33], obtained by the same authors. For comparison purposes, by applying the same criteria used here in that work, an IP of ≈100 min (relative to a CML of 0.3 mg), and an NIP of ≈400 min (6.7 h) was observed for the nitrided surfaces of ref. [33]. In addition, for the 480 min testing time, a CML of 13.3 mg was determined for the niobium surface of ref. [33]. This means that, comparatively to the ref. [33] results, here, the IP of the NbN surfaces was improved in the order of 3.6 times, and for the 480 min (8 h) testing time, the CML was reduced in the order of one magnitude (10 times).

Finally, in brief, in terms of the damage mechanism, the energy provided by the impacts of microjets and the shock waves produced by the collapse of micro bubbles on the NbN surfaces would be mainly spent to promote the fracture in the form of very small debris (nanometric in size) at the surface, as evidenced for the eroded surfaces in the higher magnitude SEM images (50 kX) in Figure 9 and Figure 10. Note that for the present characterization, no surface microcracks are observed in both Figure 9 and Figure 10, at least for the same aspect such as observed in Figure 7a–c. Considering the obtainment of multiphase NbN layers, the boundaries between phases could be preferential points to initiate fracture of such debris. In addition, debris formation is observed in the early cavitation stages on the NbN surfaces, resulting in significant craters formation mainly in the transition of the incubation period and acceleration stage, thus in the beginning of the acceleration stage. The fracture of debris allied to the effect of cyclic loadings promoted by the cavitation impacts (fatigue) would also contribute to the growing of small craters and the high level of damage all over the tested surface, as soon the transition to the acceleration stage in initiated. It is worth mentioning that such mechanism is similar to that found in ref. [9], which studied in detail the silicon nitride behavior under cavitation erosion. Comparatively, the high covalent/ionic character of the atomic binding in the NbN phases prevents any plastic deformation, in this case, explaining its brittle character, differently of the Nb surfaces, which promptly show intense plastic deformation, as expected for metallic materials, thus resulting in work-hardening, according to ref. [34].

5. Conclusions

This work presented an approach on the gas mixture influence in plasma nitrided niobium surfaces aiming to investigate the niobium nitride cavitation erosion resistance. The main conclusions can be listed as follows:

• Pulsed DC plasma nitriding process using N₂-H₂-Ar gas mixtures aiming to enhance the Nb cavitation erosion resistance through the formation of NbN surfaces was successfully carried out.
• The 3D roughness parameters are strong enough to characterize the surface morphology changes for Nb substrates subjected to pulsed DC plasma nitriding treatment.
• Comparatively, the NbN surfaces obtained for the 70% N₂ gas mixture showed a slightly shorter NIP, and a higher MER, indicating a tendency of a worse behavior.
• The brittle character of the NbN surfaces was evidenced by the occurrence of cracks around the Berkovich indentations, which is typical for ceramic materials.
• NbN surfaces (hard, with no or very little ductility), produced on Nb substrate bulks (soft, with very high ductility), show good resilience, but very low toughness.
• The energy provided by the cavitation efforts on the NbN surfaces is predominantly spent to produce very small debris, which fracture is supposedly initiated in the boundaries between the different nitride phases at the surface. This phenomenon, allied to the effect of fatigue, contributes to the growing of small craters, thus a high level of damage, leading the transition from the incubation period to the acceleration stage to occur all over the tested surface.
• The enhancement of the cavitation erosion resistance, based on the NIP of the NbN surfaces, regarding the Nb substrates, was ≈6 times, on average.