**1. Introduction**

The proton exchange membrane fuel cell (PEMFC) is a kind of energy conversion device, which transforms chemical energy into electricity. Owing to its high power-density and zero-emission features, PEMFC has been considered as one of the most prospective power suppliers for vehicles, portable devices and distributed generation [1]. PEMFC is composed of a proton exchange membrane, catalysts, gas diffusion layers, and bipolar plates. The lifetime of PEMFC depends on the durability of individual components. One of the reasons for failure is corrosion of the bipolar plates. This may not only destroy the bipolar plate itself, it also produces ions, which can be detrimental to both membrane [2] and catalyst. Therefore, the corrosion resistance of the bipolar plates is one of the decisive factors influencing the lifetime of a PEMFC. The American Department of Energy (DOE) has established a series of required properties for bipolar plates and one crucial target is to improve corrosion resistance and conductivity [3,4]. Earlier studies concentrated on graphite materials and graphite-based composite bipolar plates [5–7], considering the high corrosion resistance of graphite. However, graphite is brittle and in order to prevent breaking, it is usually machined into quite thick bipolar plate, which increases weight and manufacturing costs of the PEMFC stack [8]. Along with the development of metal forming and welding techniques, metallic bipolar plates have become increasingly relevant for PEMFC due to the good mechanical strength. This is especially for automotive stacks where weight and volume are

critical parameters. In order to reduce cost, the plates should preferably be manufactured as coated inexpensive substrates [9–11]. Austenitic stainless steel (SS), especially 316L, is prone to passivation and can be a suitable bipolar plate material [12,13]. As for coatings, noble metals like Au have been investigated widely [14,15], but although most noble metallic bipolar plates established grea<sup>t</sup> corrosion resistance and conductivity, high prices have confined their applications. Compared with noble metals, the low-cost transition metals and transition metal nitrides [16] have showed good performances in PEMFC surroundings attracting significant interest. The niobized 304 SS was reported to a decrease of alloy's corrosion current density, meeting the DOE demand (<1 <sup>μ</sup>A·cm<sup>−</sup>2) [17]. In acid solution, the corrosion resistance of Ti still needs to be further improved [13]. Metal nitrides such as TiN and NbN have been deposited on different substrates establishing low interfacial contact resistance (ICR). However, after a transient polarization test, the TiN coated sample achieved higher current density than the base SS substrate material [18]. Similarly, the NbN coating's corrosion resistance at high potential also needs to be improved [19]. Considering the embarrassing situation that it is hard to obtain high conductivity and grea<sup>t</sup> corrosion resistance at the same time, the idea of mixing Ti, Nb, TiN and NbN properly to combine the best of each of the advantages is proposed. The alloyed TiNb and TiNbN have acceptable prices when applied as films. Additionally, a TiNbN coating can be expected to have attractive mechanical property and has been used in biomedical or mechanical areas [20,21]. TiNb alloys have also been studied earlier for their unique shape memory effect [22]. Aukland et al. [23] evaluated the chemical durability and surface resistances of Ti alloys with 3 at.% Nb and confirmed that the oxidized Ti-3Nb alloys had lower resistances than Ti oxides. However, to the best of our knowledge, there is still a lack of adequate literature investigating the corrosion behavior and conductivity of TiNb or TiNbN coated SS with low Ti content under simulated PEMFC working conditions. As for the deposition method, physical vapor deposition (PVD) provides firm, uniform, and dense coatings compared with electroplating [24]. Among all the PVD ways, magnetron sputtering is used extensively in industry due to its high stability and limited contamination. During sputtering, influenced by electrical and magnetic fields, argon ions in the vacuum chamber bombard the targets causing atoms or clusters to separate from the targets and finally be deposited on the substrate. Usually, coating particles form firm bonds with the base material.

In this study, TiNb and TiNbN coatings with low Ti content were deposited on 316L SS by magnetron sputtering to explore their performances in simulated PEMFC cathode environments. The microstructures, morphologies, and chemical compositions of the coatings were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive spectrometry (EDS), and X-ray photoelectron spectroscopy (XPS). Electrochemical experiments and ICR tests were conducted to evaluate the corrosion resistance and conductivity of coated and uncoated 316L SS.

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

#### *2.1. Preparation of Coatings*

The TiNb and TiNbN coatings were prepared by magnetron co-sputtering and reactive sputtering with a Process Equipment (KJLC, Kurt Company, US) using a radio frequency (RF) Ti target and a direct current (DC) Nb target. The substrate was 316L SS with a thickness of 0.2 mm. Before sputtering, the test specimens were ground with 800, 1500, and 2000 mesh sandpaper, polished with diamond polishing agents and ultrasonically cleaned in ethanol and acetone to remove the solid particles and grease. High purity (99.9999%) argon and nitrogen were used as working gases. The chamber was depressurized to below 7 mPa. During deposition, the substrate temperature was 523.15 K and the working pressure was in 0.5–0.8 Pa. To improve the adhesion between substrate and coating, a thin titanium layer was deposited at 7.7 W·cm<sup>−</sup><sup>2</sup> for 10 minutes followed by transition layers with different atomic ratios of Ti and Nb. Finally, the target powers of Ti and Nb were regulated to 4.4 W·cm<sup>−</sup><sup>2</sup> and held for 2 h to obtain the TiNb coating. The deposition parameters of the TiNbN coating were similar to that of TiNb with lower temperature (473.15 K). The gas flow volumetric ratio of reactive gas (nitrogen) and inert gas (argon) was set as 1:5. The thicknesses of the TiNb and the TiNbN coatings were about 660 and 2080 nm, respectively, according to coatings' deposition rates.

#### *2.2. Characterization of Coatings*

The structures of the TiNb and TiNbN coatings were determined by XRD (D8 Advance, Bruker, GER), using Cu K α radiation in glancing angle mode at 1◦. The scan speed was 2◦·min−<sup>1</sup> with a step size of 0.02◦. The surface morphologies of the coatings were characterized by SEM (Merlin, Zeiss Company, GER) equipped with EDS. To evaluate the chemical composition of the coatings, XPS (EscaLab 250XI, Thermo Fisher Scientific, UK) technique was used. The corrosion resistance of the bipolar plate material was tested by electrochemical methods including potentiodynamic polarization, electrochemical impedance spectroscopy (EIS) and potentiostatic polarization in 0.5 mol·L−<sup>1</sup> H2SO4 solutions saturated with O2 at room temperature, using an electrochemical workstation (CHI760E, Shanghai Chenhua Instruments Limited, CN). The sample size was 20 by 20 mm for the tests. The testing container was a three-electrode cell. Working electrodes were SS samples, sealed by circular rings with a hole of 1 cm<sup>2</sup> exposure area. To avoid the possible addition of adverse reactive ions from other electrodes, a saturated mercurous sulfate electrode (MSE) and a platinum foil served as reference electrode and counter electrode, respectively. Prior to tests, samples were immersed in electrolyte solutions for 30 min to ge<sup>t</sup> stable open circuit potential (OCP). The scan rate of potentiodynamic polarization was set as 2 mV·s<sup>−</sup>1. EIS was carried out at both OCP and 0.19 V vs MSE, within a 0.01–10<sup>5</sup> Hz frequency range and with a 10 mV potential amplitude. Besides electrochemical evaluation, ICR measurements were performed. Two pieces of Toray conductive carbon papers were sandwiched between SS sample (30 by 30 mm) and two Au coated copper plates like Wang's method [25]. An electrical current of 2A was applied via the copper plates. The ICR value between the coating and carbon paper was calculated from the voltage drop and the current with the resistances between the other contact interfaces measured by the similar method subtracted.

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

## *3.1. Structural Characterization*

To avoid the possible e ffects from rough SS substrates, TiNb and TiNbN surface films for structural characterization were deposited on well-polished silicon wafers. The XRD patterns are shown in Figure 1.

**Figure 1.** X-ray diffraction (XRD) patterns for TiNb and TiNbN coatings on silicon wafers. The hatched lines indicate peak positions of selected materials. The peaks of cubic Ti and Nb are showed by coinciding lines.

The vertical hatched lines represent peaks recorded in PDF cards No. 38-1155 (cubic NbN), No. 38-1420 (cubic TiN), No. 34-0370 (cubic Nb), and No. 44-1288 (cubic Ti). The di ffraction peaks of cubic Nb and Ti are very close and consequently indicated by the same lines. The TiNb film follows the same patterns suggesting that the alloy has the same cubic structure. It is reasonable to assume a solid

solution of Ti and Nb. The angles are shifted to lower values indicating some stress. For the TiNbN coating, the Bragg angles of diffraction peaks are between those of TiN and NbN, which denotes that TiN and NbN are likewise a solid solution [26]. Moreover, diffraction peaks arising from (111), (200), (220), (311), and (222) plane reflections are clearly identified, confirming the face centered cubic (fcc) crystalline structure.
