**Comparison of Mechanical and Antibacterial Properties of TiO2**/**Ag Ceramics and Ti6Al4V-TiO2**/**Ag Composite Materials Using Combined SLM-SPS Techniques**

#### **Ramin Rahmani 1,\*, Merilin Rosenberg 2,3, Angela Ivask <sup>3</sup> and Lauri Kollo <sup>1</sup>**


Received: 16 July 2019; Accepted: 7 August 2019; Published: 8 August 2019

**Abstract:** In present work, the combination of spark plasma sintering (SPS) and selective laser melting (SLM) techniques was introduced to produce composite materials where silver-doped titania (TiO2) ceramics were reinforced with ordered lattice structures of titanium alloy Ti6Al4V. The objective was to create bulk materials with an ordered hierarchical design that were expected to exhibit improved mechanical properties along with an antibacterial effect. The prepared composite materials were evaluated for structural integrity and mechanical properties as well as for antibacterial activity towards *Escherichia coli*. The developed titanium–silver/titania hybrids showed increased damage tolerance and ultimate strength when compared to ceramics without metal reinforcement. However, compared with titania/silver ceramics alone that exhibited significant antibacterial effect, titanium-reinforced ceramics showed significantly reduced antibacterial effect. Thus, to obtain antibacterial materials with increased strength, the composition of metal should either be modified, or covered with antibacterial ceramics. Our results indicated that the used method is a feasible route for adding ceramic reinforcement to 3D printed metal alloys.

**Keywords:** Ti6Al4V lattice structure; Ag-doped TiO2 anatase; spark plasma sintering; selective laser melting; additive manufacturing; antibacterial and photoactivity applications

#### **1. Introduction**

Titania (TiO2) is one of the very often used ceramic materials in medical, antibacterial, paint, varnish, and pigment applications. Two well-known mineral forms of TiO2 are anatase and rutile that have motivated interests in electrical conductivity and photocatalytic activity fields. Anatase TiO2 has been shown to exhibit photocatalytic and thus, self-cleaning, activity under ultraviolet illumination. Photodegradation capability of TiO2, especially under visible light conditions can be even further enhanced by depositing transition metal dopants like silver (Ag) [1,2]. For enhanced antibacterial effect of those materials, Ti-implants have been coated by TiO2-nanotubes and Ag-nanoparticles in dental and orthopedic applications [3]. The primary interest in TiO2 ceramics has been related to its use in thin films or as an additive [4]. The implementation of TiO2 as a bulk ceramic is mainly restricted due to its high brittleness, having fracture toughness in the order of 2–3 MPa·m1/<sup>2</sup> [5].

It is possible to reinforce ceramic materials with metals, to improve the mechanical properties of the ceramics. Titanium and its alloys are widely used as biomaterials due to their sufficient biocompatibility, light weight, and high mechanical strength [6]. Titanium has excellent physical properties, its corrosion resistance is utterly known for orthopedic, osteology, and dental applications [7] and it possesses the ability to be 3D printed into complex objects. Research studies have displayed that the powder of Ti6Al4V, a material that is finding increasing use in medical applications has a high potential as a starting material for selective laser melting (SLM), a method that can be used to create various 3D structures [8]. Titania that requires relatively low temperatures for full consolidation, may be successfully combined with 3D printed metals. Combining oxide ceramics with metals by common methods is however complicated due to the inherent incompatibility of the interphases. For example, fusing of wear resistant oxide ceramics on a titanium substrate for implants faces distinct challenges in terms of obtaining strong bonding between the ceramic and the metal [9].

Different methods can be applied for consolidating ceramics and composites based on these. When ultrafine- or nano-structures are desired, spark plasma sintering (SPS) is commonly used. Both nanostructured titania and titania based nanocomposites have been produced by SPS [10–12]. Due to very high heating and cooling rate that can preserve the nanostructure at nearly full density, silver-doped titania could be employed for medical applications where antimicrobial properties, chemical inertness, and wear resistance are needed and also for water purification. Combining with titanium could drastically increase the damage tolerance of the ceramic and also provide added application specific functions. For example, a device of combined TiO2/Ag and Cu/CoNiP was shown to effectively perform as magnetically rolling microrobots for water purification [13].

In this study a combination of selective laser melting and spark plasma sintering was introduced for the production of new versatile metal–ceramic hybrid structures to increase damage tolerance of the material. In the process, SLM was used to first 3D print the periodic titanium lattice structure followed by the embedding of titania–silver composite powder to the lattice and hot consolidation by SPS. Prepared materials were analyzed for mechanical properties under a compression test and antibacterial activity against *Escherichia coli* cells.

#### **2. Materials and Methods**

#### *2.1. Materials*

Titanium dioxide powders with BET surface area of 150 m2/g (Figure 1a) and flaky silver powder with the purity of 99.95% (Figure 1b) were purchased from ABCR GmbH and used to produce ceramic materials and as the matrix phase in composites. Titanium dioxide was first ultrasonically deagglomerated under isopropanol. Doping with 2.5wt% of silver was performed by using a bottle mixer at 15 RPM for 24 h. Yttria (Y2O3) stabilized polycrystalline zirconia (ZrO2) was manufactured in TOSOH corporation (TZ-3Y-E, Tokyo, Japan) and was used to produce control ceramic surfaces for antibacterial tests (Figure 1c). Gas atomized Ti6Al4V alloy powders having a particle size in the range from 15 to 45 μm (Figure 1d) were obtained from TLS Technik Spezialpulver GmbH and were used to print metal lattices.

#### *2.2. Specimens Preparation*

Ceramic materials of TiO2–2.5% Ag, pure TiO2 and ZrO2 (the latter two as controls for the antibacterial test) were produced using a spark plasma sintering (SPS) machine HP D10 from FCT GmbH. Sintering temperature of 750 ◦C, pressure of 75 MPa and holding time of 30 min were used to compact the composites. To produce metal–ceramic composite materials, first cylindrically shaped Ti6Al4V lattices (Figure 2, top row) were produced using selective laser melting (SLM50 metal additive manufacturing system from Realizer GmbH). The specimens had a diameter of 20 mm and height of 15 mm. The specimens exhibited diamond type porous lattices with unit cell size from 1 to 2 mm. To prepare metal–ceramic composites, porous titanium lattice was placed in a graphite mold and

ceramic powder of TiO2 supplemented with 2.5% Ag was embedded in the lattices (Figure 2, bottom row). SPS at 900 ◦C, 75 MPa and during 30 min was used to compact the composites (Figure 3). The height of compacted composite lattices was 3–4 mm and depended on the cell size of initial lattices. Lattices with 1 mm cell size were shrunk down to 4 mm, lattices with 1.5 mm cell size shrunk to 3.5 mm and lattices with 2 mm cell unit size shrunk to 3 mm. For the compressive test, identical SPS conditions were applied for 10 mm diameter and 25 mm height lattice structures.

**Figure 1.** SEM micrographs of (**a**) TiO2 anatase, (**b**) Ag, (**c**) ZrO2, and (**d**) Ti6Al4V powders.

**Figure 2. Top row**: Selective laser melting (SLM) manufactured Ti6Al4V lattice structures with unit cell size and volume fraction of 2 mm and 6%, 1.5 mm and 9%, 1 mm and 16%, respectively (dimensions of lattice structures are 20 mm diameter and 15 mm height); **Bottom row**: Spark plasma sintering (SPS) sintered TiO2–2.5% Ag embedded in the Ti6Al4V lattice structures.

**Figure 3.** SPS conditions for TiO2 powder embedded Ti6Al4V lattice structure.

#### *2.3. Mechanical and Microstructural Characterization*

Microstructure of the produced materials was examined with a scanning electron microscope (Zeiss EVO MA15, Oberkochen, Germany) equipped with energy dispersive spectroscopy (EDS). Compressive testing of the samples was performed on Instron 8516 servo-hydraulic test machine. The cylindrical samples with diameter of 10 mm and height of 8–9 mm (the height of 3D printed lattice structures was 25 mm which was shrunken to 8–9 mm after SPS) were loaded with a crosshead speed of 0.5 mm/min, according to standard ASTM E9/09.

#### *2.4. Antibacterial Assay*

A comparative antibacterial assay was carried out for TiO2–2.5% Ag, TiO2, and ZrO2 ceramics as well as for TiO2–2.5% Ag and Ti6Al4V lattice hybrid structures of 1, 1.5, and 2 mm cell sizes (Figure 2). The assay was carried out using an in-house protocol based on ISO 27447:2009 and ISO 22196:2007 standard methods [14] towards a model gram-negative bacterium *Escherichia coli* MG1655. Prior to all experiments, the specimens were sanded and polished, in order to remove potential contaminants and smoothen the surface, then sterilized by autoclaving at 121 ◦C for 15 min. The material samples were

reused for consecutive experiments. After each test, samples were thoroughly washed with water and 70% ethanol, drained, submerged in 80 mL deionized water and sonicated using Branson Digital Sonifier model 450 (max power 400 W) equipped with horn model 101-135-066R at 25% amplitude for 10 min before autoclaving. *E. coli* culture for inoculum suspension was collected from fresh nutrient agar (5 g/L meat extract, 10 g/L peptone, 5 g/L sodium chloride, 15 g/L agar powder in deionized water) plates incubated overnight at 30 ◦C, suspended in 500-fold diluted nutrient broth (3 g/L meat extract, 10 g/L peptone, 5 g/L sodium chloride in deionized water) and further diluted with the same medium to optical density of 0.01 at 600 nm. Sterile surface samples were placed on the bottom of sterile 6-well polystyrene plates, inoculated with 50 μL *Escherichia coli* MG1655 suspension and covered with 2 cm × 2 cm × 0.005 cm polyethylene film. Exposure medium was 1:500 diluted nutrient broth. Samples were in parallel either covered by 1.1 mm UVA-transmissive borosilicate glass sheet and exposed to 2–2.5 W/m<sup>2</sup> UVA at 315–400 nm spectral range (measured using Delta Ohm UVA probe) effective at the sample level or kept in the dark covered by a 6-well plate lid.

After 30 min and 4 h exposure bacteria were retrieved from samples by repeatedly pipetting 3 mL of neutralizing medium (soybean-casein digest broth with lecithin and polyoxyethylene sorbitan monooleate: 17 g/L casein peptone, 3 g/L soybean peptone, 5 g/L sodium chloride, 2.5 g/L disodium hydrogen phosphate, 2.5 g/L glucose, 1.0 g/L lecithin, 7 g/L nonionic surfactant in deionized water) over the surface, serially diluted in 2 mL volume of physiological saline and from each dilution 3 × 20 μL drop-plated on nutrient agar. Plates were incubated overnight at 30 ◦C after which colony forming units were counted. The experiment was repeated at least three times for each surface type and time point. Statistical analysis of test results were carried out in GraphPad Prism 7.04 software using one-way ANOVA analysis with Tukey's multiple comparisons test at 0.05 significance level. Due to highly variable and inconsistent results of lattice-embedded samples, these were excluded from statistical analysis at the 4 h time point.

#### **3. Results**

#### *3.1. Structural Study and Mechanical Properties of the Composite Materials and Ceramics*

The appearance of TiO2 and TiO2–2.5% Ag ceramic surfaces, and TiO2–2.5% Ag in Ti6Al4V composite materials is shown in Figure 4. The white color of TiO2 ceramic changed to gray when 2.5% Ag was added. However, lighter areas likely with lower Ag content can be seen in TiO2–2.5% Ag ceramic (Figure 4). Under SEM (Figure 5) adequate interphase cohesion with some porosity between ceramic and Ti6Al4V lattice rods was observed. A SEM image (Figure 5a) and digital photograph (Figure 2) show that the 1 mm lattice structure was less bent or distorted subjected to SPS conditions. Critical areas were the interphases between the lattice and ceramic. Although water tightness of the hybrids was achieved, still some porosity at the interphase remained (Figure 5b).

**Figure 4.** Digital photograph of (**a**) ZrO2, (**b**) pure TiO2 anatase, (**c**) TiO2–2.5% Ag, (**d**) composite structure with TiO2–2.5% Ag and Ti6Al4V lattice after sintering (the lattice unit cell size is 1 mm and diameter of samples are 20 mm).

**Figure 5.** SEM micrograph of TiO2–2.5% Ag and Ti6Al4V lattice structure (1 mm unit cell size) taken at ×50 magnification (**a**) and ×200 magnification (**b**).

The EDS elemental mapping results (Figure 6) showed presence of seven elements, namely, Ti 56.45%, O 39.36%, Ag 2.53%, V 0.56%, Al 0.48%, Cl 0.42%, and P 0.20%. The EDS spectrum illustrated acceptable distribution of TiO2 and Ag in composition. Also, rounded Ti6Al4V rods cross-section validated the resistance of 1 mm cell size printed lattice under compression (Figures 2 and 6). To reveal the damage tolerance characteristics of ceramic–metal composite materials compared to pure ceramics, compression tests were performed (Figure 7). When plain titania ceramic showed brittle fracture, then lattice composite specimens did not catastrophically fail until 25% of deformation. For samples with larger volume fractions of metal phase (1 and 1.5 mm unit cell sizes) in addition to the absence of critical failure, ultimate strength of the composites was significantly higher.

**Figure 6.** Energy dispersive spectroscopy (EDS) color mapping of TiO2–2.5% Ag embedded 1 mm cell size Ti6Al4V lattice structure.

**Figure 7.** Compressive test results for TiO2–Ag without and with different unit cell sizes of lattice structure. Height of lattices were 25 mm and diameter was 10 mm before SPS.

#### *3.2. Antibacterial Activity of the Surfaces*

Antibacterial activity of the composite and ceramic materials towards *E. coli* MG1655 was evaluated after 30 min and 4 h exposure (Figure 8) while using ceramic zirconia surface as a negative control. Among the tested materials, the highest antibacterial effect in dark conditions was observed for TiO2–2.5% Ag ceramics where >3 logs reduction in *E. coli* viability was observed already within 30 min compared to control (*p* < 0.01) and no viable bacteria detected at the detection limit of about 450 colony forming units (CFU) per surface. As ceramic TiO2 without added silver had no antibacterial effect in dark conditions (*p* > 0.05), we suggest that the effect seen for Ag-supplemented TiO2 ceramics was due to Ag ions released from the material. Results for lattice-embedded TiO2–2.5% Ag surfaces were not statistically different from control after 30 min (*p* > 0.05) and had too high variability to compare results with full ceramic materials after the 4 h time point. The fact that significantly less antibacterial effect was seen for lattice-embedded samples suggests that the release of Ag from those materials was much lower than from ceramic samples.

**Figure 8.** Viability of *Escherichia coli* MG1655 on ceramic and composite hybrid surfaces after 30 min and 4 h exposure in the dark and UV-A-illumination. Columns represent recovered viable bacteria as colony forming units (CFU). Mean and standard deviation of at least three independent values is shown on a logarithmic scale and only statistically significant differences (*p* < 0.05) marked on the graph (\* *p* < 0.05 and \*\* *p* < 0.01). Only the upper error bar is shown for samples with >100% SD. Lattice samples are excluded from statistical analysis at 4 h time points due to very high variability. Limit of detection at 458 CFU/surface marked in red.

Due to the photocatalytic nature of TiO2, the samples were assumed to exhibit UV-induced antibacterial effects. Indeed, 4 h exposure of bacteria to the ceramic TiO2 surface under UV-A decreased bacterial viability by 1.6 logs (*p* < 0.05) compared to the control surface. The efficacy of Ag-supplemented ceramic TiO2 surface under UV-A was higher than that of ceramic TiO2 surface but significantly lower than the efficacy of TiO2–2.5% Ag surface in dark conditions. This is likely because UV exposure has been shown to significantly decrease Ag solubility and subsequently, antibacterial activity as has been previously shown for photo-inducible Ag complemented ZnO surfaces [14].

Metal reinforced TiO2–2.5% Ag ceramics showed significantly lower antibacterial effect than what was seen for ceramic surfaces. After 30 min under UV-A, TiO2–2.5% Ag composite surfaces did not exhibit significant antibacterial effects compared with ZrO2 control. After 4 h UV-A exposure, the composite surfaces yielded results that had too high variability to statistically compare them with control surfaces or full ceramic materials. However, according to the general picture, the composite surfaces exhibited slight antibacterial effect as compared to ZrO2 control. These results showed that while including titanium lattice to TiO2–2.5% Ag ceramic material increased the damage tolerance of the material, it significantly decreased the antibacterial effect of the ceramic material.

#### **4. Discussion**

Bonding of brittle ceramics to structural metals in assemblies has often remained a challenge, requiring designing for bolting or specific soldering alloys. The results described in this work represent a new approach for bonding, using additively manufactured lattice structure in the interphase of metal and ceramic. The composite structure where TiO2–2.5% Ag was bonded with titanium showed not only increased damage tolerance but also increased ultimate compressive strength when compared to unreinforced ceramics. To explain the increased strength and damage tolerance, ceramic and metal–ceramic samples were subjected to a compressive test (Figure 7). The results showed large fractured pieces in the case of ceramics (Figure 9a), whereas ceramics in the composite sample was fractured into sub-micrometric particles (Figure 9b). For highly brittle material such as TiO2–Ag, a larger content of energy is absorbed in crack initiation for hybrid composites.

**Figure 9.** Appearance of (**a**) TiO2, (**b**) composite structure with TiO2–Ti6Al4V hybrid after compressive testing (sample diameter was 10 mm and unit cell size was 1 mm).

The proposed failure mechanism of composite hybrids is visualized in Figure 10. Three modes of deformation could be distinguished. During the first mode, at applied strain up to 2–3 percent, energy was absorbed in rearrangement and elastic deformation of the metallic lattice. The elastic modulus of the composite hybrid (≈50–80 GPa) was significantly lower than the elastic modulus of both separated constituents, Ti6Al4V (≈110–120 GPa) and TiO2 (≈230–280 GPa). In the middle region of the elastic part of the compressive loading curve a cracking sound was observed. This was due to the removal of fractures of ceramic elements exposed to the surface (schematically shown in Figure 10, Mode 2). Until the yielding point, surface exposed ceramic elements were gradually removed, and this remained as the main deformation mechanism during plastic deformation of the hybrid, during Mode 3. Additionally, the interior ceramic elements were fractured as shown in Figure 10, Mode 3. The opposing force from metallic lattice (indicated by black arrows in Figure 10) induced back-pressure and fractured ceramic pieces got embedded into a ductile metal lattice. This interaction increased the damage tolerance of the composite hybrid under compressive loading.

**Figure 10.** Schematic showing fracture mechanisms of metal–ceramic hybrids under compressive loading. Mode 1: rearrangement and elastic deformation of metal lattice; Mode 2: fracturing of ceramic surface elements; Mode 3: fracturing of the interior ceramic elements and embedding of these in ductile metal lattice.

Consequently, three modes of failure of the hybrid composite were differentiated during compressive loading. The increased damage tolerance and compressive strength were attributed to higher input energy needed to fracture the ceramics and interaction between fractured ceramic pieces and ductile metal lattice inside the material. As it was seen on the hybrid specimens after compressive testing (Figure 9b), the ceramic material was removed preferentially at the perimeter of the cylindrical sample. Further strengthening could be achieved if hybrid composite would be surrounded by an additional metallic layer so that ceramic would not be exposed on the outer surface.

Metal 3D printing enables the production of lattice structure objects with different shapes and internal mesostructures. The composite, therefore, can be designed according to existing mechanical loads. Functionally grading in different directions, and integrating solid printed metals with ceramic–metal hybrids could be realized. Furthermore, a metal lattice could be designed so that it acts as a heating element when an electric current is directed through the material. The heating would further enhance the antimicrobial effect of the ceramic.

The compressive test result of TiO2 ceramic and Ti6Al4V–TiO2 lattice composite was captured (Figure 10) and showed the benefits of lattice structure perfectly. Finite element analysis prepares an estimation of metal–ceramic composite materials strength produced by combining SLM-SPS technique subjected impact, abrasion or compression loading [15–17]. Numerical simulation illustrated that fracture will occur at around 500 MPa for pure ceramic, while buckling for metallic lattice will start from 1200 MPa (Figure 11). Damage tolerance of Ti6Al4V–TiO2 composite depends on the densification of ceramic (SLM parameters) and metal–ceramic bonding phase (SPS parameters).

**Figure 11.** Compressive strength modelling of (**a**) TiO2 ceramic, (**b**) Ti6Al4V lattice structure. Simulation conditions and dimensions are identical for the ceramic and lattice structures.

The present work used SPS as a consolidation method, which can produce only simple, cylindrical shapes. Using other hot consolidation methods as hot isostatic pressing or hot forging, 3D shaped hybrid composites could also be manufactured. The ability to produce composites with complex shape could provide new solutions for a number of applications in the field of metal–ceramic hybrids.

This study was unique as there are no prior studies reporting on antibacterial effects of hybrid composite metal–ceramic materials. However, studies on antibacterial effects of Ag-containing ceramic materials have been previously published [18]. In general, those studies have shown the relationship between the amount of silver in the ceramic surface and antibacterial activity [19] and the importance of segregation and agglomeration of silver on the surface for improving antibacterial efficacy [20–22]. Similar observation was also done in this study but only in dark conditions. Under UV-A, the effect of added Ag to ceramics had significantly smaller effect than in dark conditions. We suggest that this was due to Ag ions which drive the antibacterial effect of Ag–ceramic surfaces [23–25] being reduced back to elemental Ag onto the surfaces [14]. Compared with TiO2–Ag ceramic material, the

antimicrobial effect of hybrid surfaces was drastically reduced (Figure 8). This change could not be only explained by reduced area of antimicrobial TiO2–Ag surface in composite material as titanium metal lattice occupied 15% to 25% of the surface, depending on the sample. In almost all cases, both in dark and under ultraviolet exposure the antimicrobial effect of composite surfaces was orders of magnitude lower when compared to the fully ceramic surface. The reason for this was not clear, but it could be assumed that there was a combination of direct surface contact and soluble silver toxicity in effect, both dependent on silver exposure at the material surface. These results indicated that if the hybrid composite needed to be exposed on the surface, the metal composition would also need to be chemically modified for an antimicrobial effect. Otherwise, we suggest that in order to preserve the antibacterial activity of the composite material and reduce variability in antibacterial activity results, a thin layer of pure ceramic material should be added to the surface of the composite.

#### **5. Conclusions**

An approach to produce titanium/silver-doped titania composites was introduced, by combining of SLM and SPS techniques. The metallic lattice structures were 3D printed, embedded with TiO2–Ag ceramic powder and consolidated by SPS. Compression strength and damage tolerance of the composites were shown to increase significantly when compared to TiO2–Ag ceramics. No collapsing of the composites was seen at up to 25% of deformation in the compressive test. Adding metallic lattice to ceramic silver-doped titania material however decreased the antibacterial effect compared with ceramics only, significantly. Thus, we suggest that the composition of metal that is used to produce the lattice should be chemically modified or a thin layer of pure ceramic material should be added to the surface of the composite.

**Author Contributions:** For research articles with several authors, a short paragraph specifying their individual contributions must be provided. The following statements should be used "conceptualization, R.R. and M.R.; methodology, L.K. and A.I.; software, R.R.; validation, M.R., A.I. and L.K.; formal analysis, R.R.; investigation, R.R. and M.R.; resources, L.K. and A.I.; data curation, M.R.; writing—original draft preparation, R.R.; writing—review and editing, R.R., M.R., A.I. and L.K.; visualization, A.I.; supervision, L.K.; project administration, A.I.; funding acquisition, L.K.".

**Funding:** This research was funded by the Estonian Ministry of Education and Research (IUT 19-29; PUT 748; IUT 23-5; base funding provided to R&D institutions B56 and SS427; M-ERA.NET DURACER project ETAG18012); The European Regional Fund, project number 2014-2020.4.01.16-0183 (Smart Industry Centre) and ERDF project TK134.

**Acknowledgments:** The authors would like to thank Mart Viljus for the help with EDS mapping.

**Conflicts of Interest:** The authors declare that there are no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Controlling Nitrogen Dose Amount in Atmospheric-Pressure Plasma Jet Nitriding**

#### **Ryuta Ichiki 1,\*, Masayuki Kono 1, Yuka Kanbara 1, Takeru Okada 2, Tatsuro Onomoto 3, Kosuke Tachibana 1, Takashi Furuki <sup>1</sup> and Seiji Kanazawa <sup>1</sup>**


Received: 25 April 2019; Accepted: 24 June 2019; Published: 25 June 2019

**Abstract:** A unique nitriding technique with the use of an atmospheric-pressure pulsed-arc plasma jet has been developed to offer a non-vacuum, easy-to-operate process of nitrogen doping to metal surfaces. This technique, however, suffered from a problem of excess nitrogen supply due to the high pressure results in undesirable formation of voids and iron nitrides in the treated metal surface. To overcome this problem, we have first established a method to control the nitrogen dose amount supplied to the steel surface in the relevant nitriding technique. When the hydrogen fraction in the operating gas of nitrogen/hydrogen gas mixture increased from 1% up to 5%, the nitrogen density of the treated steel surface drastically decreased. As a result, the formation of voids were suppressed successfully. The controllability of the nitrogen dose amount is likely attributable to the density of NH radicals existing in the plume of the pulsed-arc plasma jet.

**Keywords:** plasma nitriding; atmospheric-pressure plasma; nitrogen dose amount; hydrogen fraction; void

#### **1. Research Background**

Plasma nitriding is a surface technology to dope nitrogen atoms into metal surfaces via plasma chemical reactions to improve wear resistance, and fatigue strength, etc., of materials [1–16]. Plasma nitriding is now one of the essential surface treatments used in industry, especially in the automobile industry and die/mold fabrication. Conventional plasma nitriding uses low-pressure DC (or pulsed DC) plasmas in the abnormal glow discharge mode, where the batch process with a large vacuum furnace meets the purpose of mass production. In addition, a number of low-pressure plasma modes have recently been applied to nitriding treatment, e.g., active screen plasmas [4–6], electron cyclotron resonance plasmas [2,7], and radio-frequency plasmas [8], etc.

As another technological seed, nitriding methods using atmospheric-pressure plasmas have been developed, where the disuse of vacuum equipment makes the process much quicker and easier-to-operate. Two types of atmospheric-pressure plasmas are utilized to nitriding, namely the pulsed-arc (PA) plasma jet [17–22] and the dielectric barrier discharge (DBD) [23,24]. The PA plasma-jet nitriding has proved to be available to die steel [17,18,22], austenitic stainless steel [20], and titanium alloy [19,21], where the jet plume is sprayed onto the sample surface to thermally diffuse nitrogen atoms into it. Note that the nitrogen/hydrogen gas mixture is used as the operating gas. The plasma-jet nitriding will offer a drastically economical method to us compared with conventional plasma nitriding, especially when high-mix low-volume production is targeted.

The plasma-jet nitriding, however, is a relatively new, still developing technology. Thus, its controllability and reliability has to be improved further for practical application. For example, we had no methods to control the nitrogen dose amount from the jet plume to the metal surface, while such a method has been completed for conventional nitriding in which the nitrogen dose is well-controlled by adjusting the nitriding potential [25]. Due to the lack of dose controllability, the plasma-jet nitriding suffers from a problem of excess nitrogen supply due to the high pressure results in undesirable formation of voids and iron nitrides (the compound layer) attributed to nitrogen gas precipitation in the treated metal surface.

In this paper, a newly developed method is detailed to control nitrogen dose amount in plasma-jet nitriding to overcome the problem of excess nitrogen supply. A brief introduction of the method is as follows. The operating gas to generate the plasma jet is a nitrogen/hydrogen gas mixture. The optical emission spectroscopy proved that NH radical emission is dominant from the jet plume. In general, the NH emission intensity tends to decrease with increasing hydrogen fraction in the operating gas, *f* H2. If NH is the key radical for plasma-jet nitriding and if the decreasing tendency of NH emission with *f* H2 indicates decreasing NH density in the jet plume, we could decrease the nitrogen dose amount to metal surface by increasing *f* H2. Following this assumption, we addressed to control the nitrogen dose amount by changing *f* H2 in this study.

Prior to explaining our research, let us summarize here key species in various plasma nitriding techniques. As for low-pressure plasma nitriding methods, a comprehensive and systematic understanding of key species is not present. For example, several papers suggest the importance of ion species such as N2<sup>+</sup> [2], N<sup>+</sup> [9], and NH*<sup>x</sup>* <sup>+</sup> [10]. On the other hand, Matsumoto et al. proposed that neutral species govern the rate-limiting step [11]. For the radical nitriding, one of the low-pressure plasma nitriding methods using NH3 and H2 gas, NH radicals are considered to play a key role [12,13]. Besides, some other papers mention the importance of NH radicals for plasma nitriding [8,14]. Moreover, in gas nitriding, NH3 dissociates on an iron surface to NH2, NH2 dissociates to NH, and NH dissociates to N and H, in order, indicating that the presence of NH is essential for gas nitriding [25]. As described above, a number of studies regard NH radicals as species effective to nitriding. Thus, our expectation that NH is key for plasma-jet nitriding is not peculiar.

#### **2. Experimental Procedure**

Experiments were performed with the PA plasma jet system shown in Figure 1a. The jet nozzle was composed of coaxial cylindrical electrodes. The grounded external electrode measured 35 mm in inner diameter. The discharge gap was approximately 20 mm. The nitrogen/hydrogen gas mixture was introduced into the nozzle at the total flow rate of 20 slm, where the hydrogen fraction, and the ratio of hydrogen flow rate to total flow rate, was *f* H2. The low-frequency voltage pulse (5 kV in height and 21 kHz in repetition) was applied to the inner electrode, resulting in the maximum of the discharge current of ca. 1.2 A. The afterglow of the generated PA plasma was spewed out from the orifice and was 4 mm in diameter, forming the jet plume containing NH radicals.

**Figure 1.** Pulsed-arc plasma jet. (**a**) Schematic of jet nozzle. (**b**) Photograph of plasma jet.

For the spectroscopic experiment, the jet nozzle was fitted with a quartz pipe (30 mm in diameter and 500 mm in length) at the tip as shown in Figure 1b to observe the jet plume generating in the operating gas. The optical emission of NH (A3Π–X3Σ−) of 336 nm was detected with a spectrometer (Shamrock SR-500i, Andor, Belfast, UK). We collected the light emitted from the jet plume at the distance of 10 mm from the nozzle tip.

For the nitriding experiment, the jet nozzle was inserted into a cylindrically shaped cover made of quartz as shown in Figure 2 to purge residual oxygen from the treatment atmosphere. The height and diameter of the quartz cover was 85 and 124 mm, respectively. The gas was exhausted through the 1-mm gap between the jet nozzle and the quartz cover. The experimental system was put in a simple booth (1.0 <sup>×</sup> 1.2 <sup>×</sup> 1.8 (height) m3) surrounded by a vinyl curtain to lead the exhaust gas to the gas-treatment equipment. Prior to generating the plasma jet, residual oxygen inside the cover was gas-purged by the operating gas introduced through the nozzle.

**Figure 2.** Experimental Setup with the quartz cover to purge residual oxygen. (**a**) Schematic. (**b**) Photograph of plasma-jet nitriding.

The steel to be treated was cold roll steel JIS SPCC. The composition was as follows: 0.02% C, 0.09% Mn, 0.017% P, 0.004% S, and the balance was Fe. The sample dimension was 25 <sup>×</sup> <sup>25</sup> <sup>×</sup> 1.2 mm3. The hardness of base material was ca. 150 HV. The surface was mirror finished with alumina powder (1 μm) and degreased in an ultrasonic acetone bath.

To make the effects of nitrogen dose amount as conspicuous as possible, the surface temperature was set into the range of 1000 to 1100 K during nitrogen doping and the doped sample was immediately quenched to invoke iron-nitrogen martensite transformation. Such nitro-quenching treatment was

known to form voids, which can be readily observed with a microscope, in the surface when excess nitrogen was doped [26]. In addition, the formation of iron-nitrogen martensite indicated to us the answer as to whether or not a non-trivial amount of nitrogen had been doped even when the nitrogen dose amount was intently reduced to suppress the formation of voids. The surface temperature of ca. 1000–1100 K was maintained by the plasma-jet spraying itself, where the distance between the nozzle tip to the surface was set to 7 mm. The treatment temperature was measured by spraying the jet plume to a dummy sample with a thermocouple on the surface. The doping duration was 900 to 1800 s. The doped steel was quenched by water cooling, where tap water was poured onto the opposite surface through the hole bored in the center of the sample stage.

The doped nitrogen concentration in the steel surface was detected by an electron probe micro analyzer (EPMA, JXA-8200SP, JEOL, Tokyo, Japan). The void formation and the metallographic structure were observed with an optical microscope (VHX-5000, KEYENCE, Osaka, Japan) to a cross-section of doped steel surface. The sample surface was etched in nital solution (3%) for observing the metallographic structure. The formation of iron nitrides was detected by X-ray diffraction (XRD, SmartLab, Rigaku, Tokyo, Japan) using Co Kα radiation (λ = 0.179 nm). The hardness profile of the cross-section was measured with a Vickers microhardness tester (FM-300, FUTURE-TECH, Kawasaki, Japan), where the indenter load was 0.098 N and the loading time was 10 s.

#### **3. Results and Discussions**

#### *3.1. NH Emission Intensity*

Figure 3 shows the *f* H2 dependence of the emission intensity of NH radicals from the jet plume. The NH emission intensity increased with increasing *f* H2 up to 0.25%. On the contrary, the NH emission intensity turned to decrease with increasing *f* H2 over 0.25% up to 5%. The decreasing tendency of the emission intensity suggested the likely possibility that the density of NH radical existing in the jet plume decreased with increasing *f* H2 in the range more than 0.25%. Following this suggestion, we increased *f* H2 for the purpose of decreasing the nitrogen dose amount. Incidentally, the minimum *f* H2 was set to 1% in this study because by our previous work, *f* H2 was less than 1% proved to result in oxidization of the steel surface due to the lack of reduction performance against residual oxygen [18].

**Figure 3.** Emission intensity of NH radicals (336 nm) in jet plume vs H2 fraction in the operating gas.

#### *3.2. Nitrogen Density of Treated Steel Surface*

Figure 4a,b shows the cross-sectional mapping and depth profile of nitrogen concentration, respectively, where the observation point was at the center of plasma-jet spraying. For *f* H2 of 1%, nitrogen was considerably condensed in the surface. In the vicinity of the outermost surface, the nitrogen concentration reached ca. 12 at% and monotonically decreased in the depth direction. The concentration gradient was a typical characteristic in such a diffusion treatment. On the other hand, the nitrogen concentration in the vicinity of the surface became obviously less for *f* H2 of 2.5%

even though the gradient tendency was analogous. The maximum concentration was merely 4 at% in this case. Moreover, further decreases in nitrogen concentration was seen for *f* H2 of 5%, where the maximum value was reduced down to 2 at%. In summary, the nitrogen concentration in the doped steel surface monotonically decreased with increasing *f* H2, while it maintained the gradient tendency in the depth direction. This result indicates that the nitrogen dose amount was successfully controlled by changing *f* H2.

**Figure 4.** Distribution of nitrogen concentration for several *f* H2. The doping duration was 1800 s. (**a**) Two-dimensional mapping of sample cross-section in the vicinity of treated surface. (**b**) One-dimensional depth profile.

#### *3.3. Formation of Voids*

Figure 5 shows cross-sectional micrographs of doped steels observed in the vicinity of the center of plasma-jet spraying. For *f* H2 of 1%, a number of black dots were seen, which corresponded to the voids due to excess nitrogen doping. The existence of such voids would have made the material surface extremely brittle. On the other hand, the number of voids tended to decrease with increasing *f* H2 and as a consequence, they become invisible in the optical microscopic scale for *f* H2 of 5%. This result indicates that increasing *f* H2 can suppress the void formation in the steel surface. From the tendency of nitrogen concentration described in Section 3.2, it follows that decreasing the nitrogen dose amount was the cause of the suppression of void formation.

**Figure 5.** Micrographs of sample cross-section. The dots appearing in the vicinity of surface correspond to voids. The doping duration was 900 s.

#### *3.4. Formation of Compound Layer*

Figure 6 shows XRD spectra of the treated steel surface, where the observation point was the sample surface in the vicinity of the center of plasma-jet spraying. For *f* H2 of 1%, the treated surface obviously contained iron nitrides, namely Fe4N (γ' phase) and Fe2–3N (ε phase). On the other hand, the spectral intensities of the iron nitrides suddenly decreased with increasing *f* H2 and as a consequence, they became less than the detection limit for *f* H2 over 2%. This result indicated that increasing *f* H2 reduced the formation of iron nitrides in the steel surface. From the tendency of nitrogen concentration described in Section 3.2, it follows that the formation of iron nitrides was suppressed owing to a decreasing nitrogen dose amount.

**Figure 6.** XRD spectra of treated steel surface in the vicinity of the center of plasma-jet spraying.

In addition, the XRD peaks of the retained austenite (γ phase) were clearly seen. The formation of the retained austenite was attributed to the austenitic transformation over the critical temperature A1 and an excess solution of nitrogen. The formation of retained austenite can be regarded as another negative effect of excess nitrogen supply as well as the voids and iron nitrides. However, Figure 6 exhibits that the peak intensity of γ tended to decrease with increasing *f* H2, indicating that the amount of retained austenite was reduced. Moreover, the γ peak shifts toward high theta with *f* H2, that resulted from decreasing the lattice constant due to decreasing dissolved nitrogen concentration. From the relationship between the austenitic lattice constant *a* and the dissolved nitrogen concentration *X*<sup>N</sup> in atomic percentage (*a*/nm = 0.3564 + 0.00077*X*N) [26], we obtained the dependence of *X*<sup>N</sup> on *f* H2 as shown in Figure 7. For *f* H2 = 1%, *X*<sup>N</sup> = 9.6 at%. Increasing *f* H2 monotonically decreased *X*<sup>N</sup> and for *f* H2 = 4%, *X*<sup>N</sup> was reduced down to 0.26 at%. These characteristics of retained austenite are additional evidence for the controllability of nitrogen dose amount.

**Figure 7.** Nitrogen concentration in retained austenite calculated from the XRD spectral shift.

#### *3.5. Hardness Profile*

Figure 8 shows the two-dimensional hardness profiles of the cross-section of treated steels. The horizontal axis is the surface position of sample, the origin of which corresponds to the center of plasma-jet spraying. The vertical axis is the depth from surface. Here the micro-Vickers hardness was measured at intervals of 2 mm in the horizontal direction and 10 μm in the vertical direction. The hardness is displayed by gray scale.

**Figure 8.** Two-dimensional hardness profiles of sample cross-section. The micro-Vickers hardness is displayed by gray scale. The doping duration was 900 s.

We see that for every *f* H2, the hardness beneath the center of plasma-jet spraying was increased significantly beyond the original hardness. Here the area of ca. 5 mm in diameter was locally hardened. The local hardening was most likely due to the limited heating ability of the plasma jet only to a narrow area, not due to local nitrogen supply because of the following fact. It has already been proved that nitrogen can be supplied to the circular area as large as 20 mm in diameter by identical plasma-jet spraying, where the steel sample was heated up to ca. 800 K with the assistance of an external heater [18]. The highest hardness was 815, 755, 815, 822, and 606 HV for *f* H2 = 1%, 2%, 3%, 4%, and 5%, respectively. Such hardness cannot be obtained without nitrogen doping because the original carbon content was too low in the sample to invoke iron-carbon martensite transformation. Although the hardness was relatively low when *f* H2 = 5%, the drastic increase in hardness proved that the nitrogen dose amount was still appropriate.

Figure 9 shows the depth profile of hardness averaged within the range of ±2.5 mm of the surface position. The error bar corresponds to the standard deviation of each of the three data sets. We can see the typical trend of hardness gradient for every *f* H2. Note that for *f* H2 = 1%, obvious softening occurs in the outermost surface within the depth profile from 10 to 30 μm. (This is clearly seen also in Figure 8.) This softening was possibly due to the considerable amount of retained austenite. For *f* H2 = 2%, the softening effect became much weaker and for more *f* H2 it was not seen any more. The dependence of the softening effect on *f* H2 is consistent with the peak intensity of γ shown in Figure 6. The outermost hardness of *f* H2 = 3% and 4% reached the largest value of ca. 800 HV. For *f* H2 = 5% it became lesser, likely due to a lower nitrogen dose amount.

**Figure 9.** Depth profile of hardness of sample cross-section. The error bars correspond to the standard deviation.

#### *3.6. Metallographic Structure*

Figure 10 shows the metallographic micrographs of sample cross-section in the vicinity of the surface. Here the martensite layer and the compound layer are denoted by m and c, respectively. For *f* H2 = 1%, the compound layer was clearly seen as a discontinuous thin layer. However, no discontinuous layer appeared for more *f* H2, being consistent with the behavior of γ' and ε peaks shown in Figure 6.

**Figure 10.** Metallographic micrograph of sample cross-section. The arrow pairs denoted by m and c specify the vertical range of the martensite layer and the compound layer, respectively.

The thickness of the martensite layer depended on *f* H2. That is, the thickness increased with increasing *f* H2 up to 2%, and turned to a decrease from 2% to 4%, and then kept constant for more *f* H2. We consider that the thickness change was caused by a shift of the surface temperature. The temperature of jet plume tended to increase with increasing *f* H2 owing to the high thermal conductivity of hydrogen, transporting the thermal energy from the pulsed arc to the jet plume. The sample temperature (measured with a dummy sample) was ca. 1000 K for *f* H2 = 1% but increased up to ca. 1100 K for more *f* H2. From this fact, it follows that the increase in the thickness from *f* H2 = 1% to 2% was caused by the enhanced thermal diffusion and the subsequent decrease was caused by the decrease in nitrogen dose amount. The change in the layer thickness can be seen also in Figures 8 and 9.

#### **4. Conclusions**

To overcome the problem of excess nitrogen supply in the PA plasma-jet nitriding, a controlling method of nitrogen dose amount was proposed on the basis of NH radical emission and was addressed to experimentally performed. Consequently, we have demonstrated that the nitrogen dose amount to the steel surface can be controlled by changing the hydrogen fraction in the operating gas. As a result of nitrogen dose control, undesirable formation of voids and iron nitrides were successfully suppressed, while a nitrogen dose enough to invoke martensite transformation was simultaneously achieved.

This achievement means that we have first obtained the technique to control "the nitriding potential" even in the new plasma-jet nitriding. We believe that the upgraded controllability presented here was of great help for future practical applications of the plasma-jet nitriding, especially for applications to high-mix low-volume production of mechanical and medical fabrications. Note that although the treatable area in this study seems too small for practical use, we can practically treat the circular area of at least 20 mm in diameter for the ordinary nitriding temperature at ca. 800 K [18].

**Author Contributions:** Conceptualization, all; data curation, R.I., M.K. and Y.K., methodology, R.I., M.K. and Y.K.; investigation, R.I., M.K., Y.K., T.O. (Takeru Okada), T.O. (Tatsuro Onomoto), K.T., T.F. and S.K.; writing—original draft preparation, R.I.; writing—review and editing, K.T. and S.K.; visualization, M.K., Y.K., T.O. (Takeru Okada), T.O. (Tatsuro Onomoto) and T.F.; supervision, R.I.; project administration, R.I.; funding acquisition, R.I.

**Funding:** This work was supported by JSPS KAKENHI Grant Number 15K17482.

**Acknowledgments:** We wish to acknowledge valuable discussions with Masahiro Okumiya, Toyota Technological Institute, and Nobuyuki Kanayama, Santier Giken Co., Ltd. We are grateful to Masaki Sonoda, Oita Industrial Research Institute, for their technical assistance.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Self-Lubricating PEO–PTFE Composite Coating on Titanium**

#### **Limei Ren 1,2,3, Tengchao Wang 1,2, Zhaoxiang Chen 2,3,\*, Yunyu Li 2,3 and Lihe Qian <sup>4</sup>**


Received: 19 December 2018; Accepted: 30 January 2019; Published: 1 February 2019

**Abstract:** A self-lubricating plasma electrolytic oxidation–polytetrafluoroethylene (PEO–PTFE) composite coating was successfully fabricated on the surface of commercially pure titanium by a multiple-step method of plasma electrolytic oxidation, dipping and sintering treatment. The microstructure and tribological properties of the PEO–PTFE composite coating were investigated and compared with the PEO TiO2 coating and the PTFE coating on titanium. Results show that most of the micro-pores of the PEO TiO2 coating were filled by PTFE and the surface roughness of PEO–PTFE composite coating was lower than that of the PEO TiO2 coating. Furthermore, the PEO–PTFE composite coating shows excellent tribological properties with low friction coefficient and low wear rate. This study provides an insight for guiding the design of self-lubricating and wear-resistant PEO composite coatings.

**Keywords:** self-lubricating; composite coating; titanium; plasma electrolytic oxidation (PEO); polytetrafluoroethylene (PTFE)

#### **1. Introduction**

Titanium and its alloys are lightweight structural metals, exhibiting high strength-to-weight ratio and excellent corrosion resistance. Due to these advantages, they are widely used in many industries, especially the automotive, aerospace and shipping industries [1–3]. However, titanium alloys generally show low surface hardness and poor tribological properties, characterized by severe abrasive wear and adhesive wear, which seriously restricted their applications in field of machinery [4,5]. Thus, developing proper surface modification techniques to improve the tribological properties is a crucial step to expand the application scopes of titanium alloy. At present, the commonly used methods to enhance surface performance of titanium alloys mainly include physical vapor deposition [6], chemical vapor deposition [7], ion implantation [8], thermal spraying [9], plasma electrolytic deposition [10], plasma electrolytic oxidation (PEO) [11], etc. Among these techniques, PEO is a simple and environment-friendly process with rapid deposition of anodic oxide coating on the titanium surface. Moreover, PEO coatings show increased surface hardness and excellent wear resistance. In spite of these advantages, PEO coatings usually take on high surface roughness and porosity with high friction coefficient, which has limited the extensive industrial application of the PEO technique [12]. Therefore, improving the tribological properties of PEO coating, is the key to

enlarge its application range. In the previous study, we have explored depositing diamond-like carbon (DLC) on the PEO coating by using the unbalanced magnetron sputtering technique. Although results show that the TiO2/DLC composite coating exhibits improved tribological properties with low friction coefficient and low wear rate, DLC coating deposition is an expensive and time-consuming process [13]. Furthermore, this technique is not favorable to producing coatings on complex-shaped specimens.

In general, using liquid lubricants can effectively reduce the friction coefficient of the PEO coatings and decrease the wear between the friction pairs. However, in some extreme conditions, such as high temperature and high vacuum, liquid lubricant would volatilize and cause lubrication failure. Solid lubricants commonly used in mechanical lubrication include graphite, hexagonal boron nitride (hBN), molybdenum disulfide (MoS2), polytetrafluoroethlene (PTFE), etc [4,11,14]. PTFE has good chemical stability and thermal stability, with excellent lubricating property in a relatively wide temperature range and almost all of the ambient atmosphere [15]. In recent years, some researchers have attempted to disperse PTFE particles into the electrolyte and directly incorporated these self-lubricating particles into the anodic coating during the PEO process, but the localized high temperature during the PEO process may cause the decomposition of PTFE. Consequently, it is difficult to control the content and distribution of PTFE in the composite coatings by this technique.

In this paper, with the aim of improving the tribological properties of the PEO coating, especially the lubricating property, the PEO coating was used as the substrate and then the solid lubricant PTFE was deposited directly on the surface of the PEO coating by dipping and sintering treatment to fabricate a PEO–PTFE composite coating. The surface of the PEO coating is characterized with lots of micro-pores and micro-cracks which ensured the probability to deposit the small sized solid lubricant into these micro-pores on the PEO coating. The microstructure and tribological properties of the PEO–PTFE composite coating were investigated using scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), Raman, surface profiler and ball-on-plate tribometer test and then compared with the PEO TiO2 coating and the Ti-PTFE coating.

#### **2. Materials and Methods**

#### *2.1. Preparation of Coatings*

The commercially pure titanium was used as the substrate for preparing the PEO coating, and the titanium plate was cut into samples with a size of 25 × 70 × 2 mm. The chemical composition of the titanium substrate were Fe ≤ 0.30 %, Si ≤ 0.15 %, O ≤ 0.20 %, C ≤ 0.10 %, N ≤ 0.05 %, H ≤ 0.015 % and Ti balance. In preparation for PEO process, all samples were ground with SiC abrasive papers (from 150-grit to 1500-grit) and cleaned ultrasonically with 95% alcohol for 30 min. Then, all samples were cleaned with deionized water and air dried. A pulsed asymmetric bipolar AC power supply (MAO120HD-III, Xi'an University of Technology, China) was employed for the PEO process. The PEO of titanium was carried out under the mode of constant voltage and the parameters setting of PEO is listed in Table 1. The electrolyte used in the PEO process consisted of Na2SiO3 (15 g/L), Na3PO4 (10 g/L) and NaOH (1 g/L). During the PEO treatment, titanium samples and stainless steel plates were used as the anode and cathode, respectively. A circulating cooling system was used to keep the temperature of the electrolyte below 40 ◦C. After the PEO treatment, the coated samples were cleaned with deionized water and dried at ambient temperature.

**Table 1.** The parameters setting of plasma electrolytic oxidation power supply.


After the PEO treatment, the samples were subjected to dipping and sintering treatment to fabricate the PEO–PTFE composite coating. The PEO samples were immersed in PTFE dispersion (Shanghai Aladdin Bio-Chem Tech Co. LTD, Shanghai, China) heated to 50 ◦C and held for 20 min in a thermostat water bath, and then put into a muffle furnace (TESE, RXF1400-5-12, Shanghai, China). Figure 1 shows the process of sintering treatment for PEO–PTFE composite coating. Firstly, all coated samples placed in the muffle furnace were heated from room temperature to 150 ◦C with a heating rate of 5 ◦C/min, keeping the temperature at 150 ◦C for 20 min. After that, the temperature was increased to 350 ◦C with a heating rate of 5 ◦C/min, holding at the temperature of 350 ◦C for 20 min. Finally, all samples were cooled to room temperature inside the furnace. Following the same procedure as mentioned above, the PTFE coating directly deposited on the titanium was fabricated.

**Figure 1.** Sintering treatment process.

#### *2.2. Characterization of Coatings*

The surface morphologies of the coatings were studied by using scanning electron microscopy (SEM, SIGMA-500, ZEISS, Jena, Germany), equipped with EDS. The EDS was used to analyze the elementary composition of the PEO–PTFE composite coating. Raman spectra of the coatings were obtained with an Nd-YAG solid state laser (wavelength 532.0 nm) through the 50x objective lens of a Raman microspectrometer (Horiba, Kyoto, Japan). The surface roughness of the coatings was examined by a confocal microscope (Conscan Profilometer, Anton Paar Tritec SA, Peseux, Switzerland). The tribological properties of coatings were measured at room temperature under dry sliding condition by using a ball-on-plate reciprocating tribometer (Tribometer, Anton-Paar Tritec SA, Peseux, Switzerland). And the GCr15 stainless steel ball, with a diameter of 3 mm, was used as grinding material in the test. The friction coefficient and wear rate of the coatings was measured at different loads (2~8 N) with a sliding distance of 100 m. After the friction and wear tests, the wear tracks were measured by a surface profiler (MarSurf, Mahr, Göttingen, Germany). The wear rate of samples was calculated according to the formula: *K* = (*S*· *l*)/(*F*· *L*) [16], where *S* is the cross-sectional area of wear track (mm2), *l* is the length of wear track (mm), *F* is the applied load (N), *L* is the sliding distance (m).

#### **3. Results and Discussion**

#### *3.1. Surface Morphology and Compositions*

Figure 2a–d illustrates the morphologies of titanium substrate, Ti-PTFE coating, PEO TiO2 coating and PEO–PTFE composite coating. The SEM image of titanium substrate ground by sandpapers is shown in Figure 2a, from which lots of plough marks can be observed. This phenomenon reflects poor wear resistance of the commercially pure titanium. Figure 2b displays the morphology of the PTFE coating deposited directly on the surface of titanium substrate. It can be seen that the PTFE coating did not cover the substrate very well and peeled off at some areas, indicating the poor bonding between the PTFE coating and the relatively smooth titanium substrate. The typical surface morphology of PEO TiO2 coating fabricated in the silicate-phosphate electrolytic solution is presented in Figure 2c. It can be seen that the PEO TiO2 coating has a porous and rough surface structure, characterized with a large number of crater-like micro-pores and micro-protrusions. These micro-pores were formed by the plasma micro-arc discharges, which resulted in local high temperature and high pressure, causing the molten materials to erupt from the micro-arc discharge channels. Then, the erupted molten oxides solidified and accumulated around the micro-pores, leading to the formation of micro-protrusions. Although these surface structures led to a high surface roughness, this provided the possibility of depositing small sized solid lubricant into micro-pores and around micro-protrusions. In addition, the cross-sectional morphology of PEO TiO2 coating is also shown in Figure 2c. It can be seen that the oxide coating adhered tightly to the substrate and the average thickness of the oxide coating is around 10 μm.

**Figure 2.** Morphologies of (**a**) titanium substrate; (**b**) Ti-polytetrafluoroethylene (PTFE) coating; (**c**) plasma electrolytic oxidation (PEO) TiO2 coating; (**d**) PEO–PTFE composite coating.

Figure 2d shows the surface morphology of PEO–PTFE composite coating. Compared with Figure 2c, it can be observed that a great deal of micro-pores on the PEO TiO2 coating had been filled with PTFE and the porosity of the PEO TiO2 coating decreased significantly. During the process of sintering treatment, the sample was heated to 350 ◦C, which exceeded the melting point of PTFE materials [17]. As the PTFE has a dynamic viscosity in the melting state, the melted PTFE filled the pores of PEO and formed a continuous composite PEO–PTFE coating [18–20]. Obviously, the rough and porous PEO surface enhanced the crosslinking and bonding performance of PTFE materials to the bottom of the TiO2 coating.

The EDS elemental maps of the PEO–PTFE composite coating is shown in Figure 3, which are obtained from the EDS scanning of Figure 2d. F and C elements were detected in the PEO–PTFE composite coating, indicating that PTFE materials deposited successfully on the PEO coating. From the distribution of F and C elements, it can be found that the most area of the PEO coating was covered by PTFE materials, except for the locations where the PEO oxides highly accumulated. This phenomenon is because of the fluidity of PTFE materials during the dipping and sintering processing. The mechanical interlinking formed between the porous PEO coating and the continuous PTFE materials after

the sintering treatment. Such mechanical interlinking ensured a good bonding performance of the PEO–PTFE composite coating. In contrast to the single PEO TiO2 coating, this PEO–PTFE composite coating maintained the good wear-resistance of the ceramic PEO component while its PTFE component played the role as the friction-reducing lubricant. As reported in a recent study, a self-lubricating PEO coating was fabricated on AZ91 magnesium alloy via the in-situ incorporation of PTFE particles [21]. Results showed that the in-situ PTFE incorporation into the growing PEO coating resulted in non-uniform PTFE distribution and some PTFE-enriched ridge-like protrusions were formed on the coating surface. When the incorporation time was sufficient, the PTFE-enriched protrusions formed can act as lubricant reservoirs, leading to a low and stable friction coefficient. The present PEO–PTFE composite coating aimed to achieve uniform PTFE distribution by using the dipping and sintering treatments, ensuring even and sufficient PTFE lubricant supply during the friction and wear test.

**Figure 3.** Energy-dispersive X-ray spectroscopy (EDS) elemental maps of PEO–PTFE composite coating.

The Raman spectra of PEO TiO2 coating and PEO–PTFE composite coating are shown in Figure 4. Recorded Raman spectra were made in the frequency range from 100 to 1500 cm−1. It can be seen from the Raman spectrum of PEO TiO2 coating that bands shift located at 129 (Eg) and 498 (A1g) cm−<sup>1</sup> positions originate from different modes of anatase TiO2 [22]. Bands shift located at 423 (Eg) and 591 (A1g) cm−<sup>1</sup> positions originate from different modes of rutile TiO2. It can be inferred from Raman spectra of the PEO coating that there are two TiO2 crystal structures with anatase and rutile on the surface of the coating. According to the Raman spectrum of PEO–PTFE composite coating, one can see that some characteristic peaks of PTFE can be detected. The Raman peaks at 271, 364, 711 and 1360 cm−<sup>1</sup> are related to the different vibrational modes of CF2 groups, whereas bands shift located at 1218 and 1282 cm−<sup>1</sup> positions come from the bands C-C and CF, respectively [23]. The Raman result is consistent with the previous EDS analysis result, indicating the successful deposition of PTFE on the PEO coating.

**Figure 4.** Raman spectra of the PEO TiO2 coating and PEO–PTFE composite coating.

Figure 5 presents the surface roughness (Ra) of the titanium substrate, Ti-PTFE coating, PEO TiO2 coating, and PEO–PTFE composite coating. Compared with the titanium substrate, the roughness of Ti-PTFE coating slightly increased. This is because the Ti-PTFE coating deposited directly on the smooth titanium substrate was easy to peel off (see Figure 2b), causing a rougher surface than the titanium substrate. The roughness of the PEO TiO2 coating turned out to be the highest due to the existence of micro-pores and micro-protrusions on its surface (see Figure 2c). PTFE polymer materials deposited into the porous TiO2 coating (see Figure 2d) and formed the composite coating. Therefore, the roughness of the PEO–PTFE composite coating decreased significantly compared with the PEO TiO2 coating.

**Figure 5.** Surface roughness (Ra) of titanium substrate, Ti-PTFE coating, PEO TiO2 coating and PEO–PTFE composite coating.

#### *3.2. Tribological Behaviors*

The tribological tests were performed at room temperature using a ball-on-plate reciprocating tribometer. Figure 6a shows the relationship between the friction coefficient and sliding distance of the titanium substrate, Ti-PTFE coating, PEO TiO2 coating and PEO–PTFE composite coating. From the friction coefficient curve of titanium substrate, it can be seen that the value remained about 0.5 after a slight fluctuation at the initial stage of the test. In the case of PEO TiO2 coating, this curve exhibited significant fluctuation before the sliding distance of about 20 m. After that, the friction coefficient gradually decreased to 0.65 and remained stable until the end of the test. This phenomenon was in connection with the unique structure of PEO TiO2 coating. The PEO coating typically has a rough and porous outer layer and a dense inner layer [13]. The porous outer layer featured with high surface roughness and varied internal microstructure, which contributed to the high and unstable friction coefficient at the initial stage of the test. With the gradual wearing-off of the porous outer layer, the GCr15 ball began to contact and slide against the dense inner layer of PEO TiO2 coating, resulting in the decreased and relatively stable friction coefficient.

**Figure 6.** Friction coefficient of different samples against GCr15 after a sliding distance of 100 m (**a**) titanium substrate, Ti-PTFE coating, PEO TiO2 coating and PEO–PTFE composite coating under the load of 4 N; (**b**) PEO–PTFE composite coating under different loads.

Comparing the friction coefficient curves of the Ti-PTFE coating and the PEO–PTFE composite coating in Figure 6a, it can be seen that the friction coefficient of PEO–PTFE composite coating remained stable at 0.1 during the whole process of the test. For the Ti-PTFE coating, the friction coefficient was about 0.1 from the beginning until the sliding distance of 55 m and then increased sharply to about 0.4, which was close to the coefficient of titanium substrate. This result indicated that the Ti-PTFE coating had been worn off after a certain sliding distance. Afterwards, the friction behavior occurred between the grinding ball GCr15 and titanium substrate. As shown in Figure 2b, the PTFE coating exhibited poor adhesion to the titanium substrate. Therefore, it was easily damaged and peeled off in the process of reciprocating friction and wear. In the case of the PEO–PTFE composite coating, its long-term low friction coefficient proved that depositing of PTFE materials can effectively improve the friction property of PEO TiO2 coating. Also, it is demonstrated that the self-lubricating friction behavior occurred between the grinding pairs.

Figure 6b shows the relationship between the friction coefficient and sliding distance of the PEO–PTFE coating under different loads. It can be observed that when the load was 2 N and 4 N, the friction coefficient was stable at about 0.1. Under the higher load of 6 N and 8 N, the friction coefficient of the self-lubricating PEO–PTFE composite coating increased slightly with the increase of the sliding distance, but was still lower than 0.15. The friction coefficient curves of the self-lubricating PEO–PTFE composite coating under different loads consistently appeared to be low and stable, indicating that the self-lubricating PEO–PTFE composite coating has a good load-bearing capacity. Recently, Wang et al. constructed a lubricant composite coating on Ti6Al4V alloy using micro-arc oxidation and grafting hydrophilic polymer [24]. Results showed that the composite coating exhibited the low friction coefficient and favorable wear resistance in water under a low contact stress of 1.52 MPa. It was explained that the hydrophilic polymer formed a hydrated lubricating layer through the interaction with water and the TiO2 ceramic layer provided the resistance to wear.

Figure 7a–c presents the SEM images of wear tracks of the Ti-PTFE coating, PEO TiO2 coating and the PEO–PTFE composite coating after a sliding distance of 100 m under the normal load of 4 N. The cross-sectional profiles of wear tracks are shown in Figure 7d. From the wear track profiles and worn morphologies of the coated samples, it can be noticed that the wear track of Ti-PTFE coating was narrower and deeper than the PEO TiO2 coating. Due to the loose adhesion of PTFE coating to titanium substrate, the PTFE coating was easily peeled off and damaged. After the PTFE coating was worn out, the titanium substrate was exposed to the grinding ball GCr15. Therefore, the poor wear resistance of titanium substrate resulted in a deeper wear track. For the PEO TiO2 coating, it can be seen that although the width of the wear track was large, the overall wear track was shallow. The ceramic PEO TiO2 coating has the advantage of high hardness and excellent wear resistance. Consequently, a lot of wear and tear of the GCr15 counterpart occurred with the proceeding wear test. Therefore,

the contact areas between the grinding pairs increased, leading to the increase of the wear track width. For the PEO–PTFE composite coating, it can be seen from Figure 7c that the wear track was obviously shallower and narrower than both the Ti-PTFE coating and PEO TiO2 coating. This is because the deposition of PTFE polymer materials on the PEO TiO2 coating significantly improved the tribological properties of the PEO TiO2 coating. Firstly, the deposition PTFE materials in the PEO micro-pores or around the PEO micro-protrusions resulted in decreased the surface roughness and thus increased contact area and decreased contact stress between the friction pair, which avoided the rapid fracture and wearing-off of PEO outer layer to some extent. Secondly, the existence of self-lubricating PTFE component in the composite coating decreased the friction force and kept the GCr15 counterpart from severe wear.

**Figure 7.** Scanning electron microscope (SEM) images and cross-sectional profiles of wear tracks after a sliding distance of 100 m under the 4 N load of the (**a**) Ti-PTFE coating; (**b**) PEO TiO2 coating; (**c**) PEO–PTFE composite coating; and (**d**) cross-sectional profiles of wear tracks.

Figure 8 shows the micro worn morphologies of the PEO TiO2 coating and the self-lubricating PEO–PTFE composite coating after a sliding distance of 100 m under the 4 N loading. The element composition of wear tracks were examined by EDS and the results are listed in Table 2. For the worn track of PEO TiO2 coating (Figure 8a), it can be seen that there were still a great deal micro-pores after the tribo-test, indicating high porosity both in its surface and interior. The existence of micro-pores and debris greatly weakened the strength and toughness of the PEO TiO2 coating. It can be observed that there were two kinds of typical worn surface morphologies of the PEO TiO2 coating, as marked by point 1 and point 2 in Figure 8a. The first worn morphology (see point 1 in Figure 8a) was relatively smooth. A lot of Fe and O elements as well as a small quality of Cr element were detected (see Table 2), indicating that materials were transferred from the GCr15 ball to the PEO coating surface during the wear process. The second worn morphology (see point 2 in Figure 8a) was relatively rough. The brittle and porous surface layer of the PEO coating was crushed under the normal loading and formed lots of oxide debris at the initial stage of the tribo-test, which caused abrasive wear between the PEO coating

and GCr15 ball during the subsequent tribo-test. Therefore, lots of O, Ti and Si elements and small amounts of Fe elements derived from GCr15 ball were detected (see Table 2).

**Figure 8.** Micro worn morphologies after a sliding distance of 100 m under the load of 4 N: (**a**) PEO TiO2 coating; (**b**) the self-lubricating PEO–PTFE composite coating.

**Table 2.** Element composition of the PEO TiO2 coating and PEO–PTFE composite coating worn track after a sliding distance of 100 m under 4 N in Figure 8.


As shown in Figure 8b, the worn surface morphology of the self-lubricating PEO–PTFE composite coating was obviously different from the PEO TiO2 coating. There were only a small number of micro-pores being observed on the worn surface, indicating a compact worn surface of the self-lubricating PEO–PTFE composite formed by repeated crushing of PTFE polymer materials and PEO TiO2 coating. In addition, a lot of F and C elements (see point 3 in Figure 8b and Table 2) were detected on the worn surface, demonstrating the formation of self-lubricating film. As PTFE polymer materials deposited into the PEO TiO2 coating, the porous surface was covered by the self-lubricating materials. The micro-pores were sealed by the solid lubricant. In the process of friction, a lubricating film would be formed on the friction contact surface. As a result, the sliding pairs in the contact surface were changed from the steel ball versus the composite coating to PTFE transfer film on the surface of the steel ball versus PTFE materials. Because of the poor adhesion between the PTFE transfer film and the steel ball, the transfer film usually fell off under the obstacle of the abrasive particle [25]. But the self-lubricating materials stored in the micro-pores provided continuous supply for the formation of the self-lubricating film. As the sliding distance increased, the PEO TiO2 component played a role as a wear-resistant reinforcing phase and supported the self-lubricating film.

Figure 9 shows the wear rate of all samples against the counterpart GCr15 ball after the sliding test under the normal load of 4 N and a sliding distance of 100 m. During the process of sliding, the titanium substrate sample suffered from severe wear and tear, mainly due to its low hardness. Therefore, the wear rate value of the titanium substrate is large, which is 99.27 × <sup>10</sup>−<sup>5</sup> mm3·N−1·m<sup>−</sup>1. For the Ti-PTFE sample, the average wear rate was 28.39 × <sup>10</sup>−<sup>5</sup> mm3·N−1·m−1, smaller than that of the titanium substrate sample, because there was self-lubricating friction behavior at the beginning stage of the tribo-test (see Figure 6a). When the PTFE coating was worn off and the self-lubricating film failed, the bottom titanium substrate began to be subject to wear. From Figure 9, it can be seen that the wear rate of the PEO TiO2 coating (7.02 × <sup>10</sup>−<sup>5</sup> mm3·N−1·m<sup>−</sup>1) was smaller than that of both the titanium substrate and Ti-PTFE samples. The PEO TiO2 coating sample has higher hardness than Ti and Ti-PTFE samples, leading to the decrease of wear rate. The wear rate of the self-lubricating PEO–PTFE composite coating was 1.34 × <sup>10</sup>−<sup>5</sup> mm3·N−1·m−1, which is the smallest among all the

samples. As the PTFE was integrating into the PEO coating and filling into the micro-pores, the high hardness of PEO coating and the good lubrication of PTFE combined to increase the wear resistance and decrease the friction coefficient.

**Figure 9.** Wear rate of the titanium substrate, Ti-PTFE, PEO TiO2 coating and the self-lubricating PEO–PTFE composite coating under the load of 4 N and the sliding distance of 100 m.

#### **4. Conclusions**

The self-lubricating PEO–PTFE composite coating was successfully fabricated on the titanium surface by the combined methods of plasma electrolytic oxidation, dipping and sintering. Most micro-pores of the PEO TiO2 coating were effectively filled with PTFE and the surface roughness of the PEO TiO2 coating decreased significantly. The fabricated PEO–PTFE composite coating, integrating the advantages of wear resistance of the PEO TiO2 coating and the self-lubrication of PTFE polymer materials, exhibited a low friction coefficient and wear rate. The friction coefficient of the PEO–PTFE composite coating was much smaller than that of PEO TiO2 coating and remained stable (around 0.1) under different loads. The PEO–PTFE composite coating was more durable than the single PTFE coating and its wear rate was about 5 times lower than that of the PEO TiO2 coating.

**Author Contributions:** Z.C. and L.R. conceived and designed the experiments; T.W. and Y.L. performed the experiments; L.R. and T.W. analyzed the data and wrote the paper; Z.C. supervised the work; L.Q. contributed analysis tools.

**Funding:** This research was supported by Natural Science Foundation (E2016203270) and Returned Overseas Chinese Talents Foundation (CL201726) of Hebei Province, and the Fundamental Research Foundation (020000904) and Doctoral Foundation (B942) of Yanshan University.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
