Introduction

Polymer electrolyte fuel cells (PEFCs) with high energy yield and low environmental impact has been focused by a huge number of researches, which are considered to be the most promising energy conversion technologies available today1. The cathode oxygen reduction reaction (ORR) of PEFCs acts as a critical role in controlling the performance2,3, for being always six or more times of magnitude slower than the anode hydrogen oxidation reaction4. Additionally, the ORR is also important in metal-air batteries5,6,7, water electrolysis, corrosion and so on8,9. Traditional cathode materials of fuel cells are used platinum as an active ORR catalyst, with the high price of this precious metal having a limitation on its economic usefulness. Apart from the high cost, Pt-based catalysts also suffer from limited durability10,11,12 and carbon monoxide, methanol or other small organic molecules deactivation3,13,14. Therefore, Pt replacement candidates that attracts the most attention have been reported widely recently including metal oxides15,16,17,18, carbides19,20, nitrides21,22, chalcogenides23,24 and nitrogen-doped carbons3,7,25,26. However, rare materials can do satisfy the requirement of activity and stability for ORR in PEFC condition referring to platinum and to realize manufacture in large quantities is also a problem in practical industry.

Titanium nitride (TiN) leads the electrocatalytic activity among various metal nitrides since its more active sites, stronger adsorption ability and much cheaper than Pt21,27. According to the market price, TiN reagent is ~$ 1.0 per gram compared with ~$ 95 per gram of Pt/C reagent, while the industrialized manufacture of TiN would further cut down the cost even less than $ 0.1 per gram but industrialized Pt hardly decrease its spending so much. Nevertheless, the reported TiN nanostructures with inferior ORR performance in accordance with Pt-based catalysts haven't reach the height for catalysis in the cathode reaction of PEFCs yet, even though the small enough particle size and the use of ideal supports like carbon black, carbon nanotube or graphene. Because the injected electrons to TiN would be difficult to finish the O-O bond dissociated easily, which bring about two-electron oxygen reduction process and the generation of undesirable H2O221,22,28,29. In general, transition metal nitride/carbide ought to play a crucial role in the practical applications of optical, electronic, magnetic and catalytic area, contrast with oxides and carbon materials, attributing to the high electrical conductivity, hardness and durability30,31. Thus, to solve the issue fundamentally, the architecture of electrocatalyst should be designed to improve the adsorption mode on nitride so that the electron could be injected more effective from the catalyst to oxygen.

Herein we use a facile strategy of chemical vapor deposition (CVD) to grow TiCN and TiN thin films in two-step on the nickel foam template. After milling, the Ni is removed by acid and a hierarchical TiN/TiCN structure is obtained. In addition, the electrocatalytic activity for oxygen reduction reaction is studied and evaluated. The catalyst displays a highly improved activity compared with TiN and TiCN in alkaline environment, which is found to be even more efficient than commercial 20% wt Pt/C, as well as also better durability and tolerance to methanol in the cathode reaction of PEFCs.

Results

The TNTCNHS were prepared via CVD process and the removal of template as illustrated in Figure 1. Above all, Ni foam was firstly coated on TiCN thin films via CVD, as soon as TiN thin films were deposited on TiCN to obtain TiN/TiCN/Ni thin films (TNTCNTF). As the photographs show, the Ni foam with yellow deposits became fragile so that it could be milling to uniform powder. Finally, chlorhydric acid was employed to dissolve the Ni template and TNTCNHS was obtained under ultrasonic wave. The color of the catalyst after dried had become black yellow and the dispersion was greatly improved in ethanol. The thickness of each layer could be easy to tune via controlling the airflow of the mixed gas and the multilayer deposits always keep uniform, dense, good crystallinity and closely integrated with the substrates32. Figure 2A and 2B are the scanning electron microscopy (SEM) images of the two sides of TNTCNHS respectively. As we see the TiN side shows ~100 nm diameter crystals array densely and smaller size TiCN nanoparticles (NPs) on the opposite displays porous structures which would supply high specific surface area for catalysis. The interface of the TiN and TiCN (TNTCNHS-1) is also observed in Figure 2C (TNTCNHS-2 with thicker layers is shown in the inset). The compact structures of two compositions are likely to make them score high in electron transfer and stability. The corresponding transmission electron microscopy (TEM) images and fast Fourier transform (FFT) patterns are in Figure 2D and 2E, which indicate the (200) of TiCN and (111) of TiN. Here we put forward a facile method to construct double-layer structures by using chemical vapor deposition.

Figure 1
figure 1

Schematic diagram of the synthesis of TNTCNHS catalysts.

Full size image
Figure 2
figure 2

SEM images of (A) TiCN side, (B) TiN side and (C) the interface of TiN and TiCN (TNTCNHS-1) and the inset shows the interface of TNTCNHS-2 (The scale bar is 200 nm). TEM images and coresponding FFT patterns of (D) TiCN and (E) TiN crystal on TNTCNHS.

Full size image

X-ray diffraction (XRD) pattern of the TNTCNHS is shown is Figure S1A which indicates our prepared catalyst is TiN/TiCN with good crystallization. However, the unit cell parameters of titanium nitride and titanium carbonitride are too closed, thus it could not be confirmed both of them two exist in the catalyst. Therefore, X-ray photoelectron spectroscopy (XPS) is utilized to determine the elements and chemical bonds. Figure S1B demonstrates a XPS survey and high revolution spectrums of Ti 2p, N 1 s and C 1 s in TNTCNHS with the element carbon as internal standard (B.E. = 284.6 eV). Obviously, the binding energy (B.E.) of Ti-N could be examined at Ti 2p3/2 (B.E. = 456.3 eV), Ti 2p1/2 (B.E. = 461.2, 464.0 eV) and N 1 s (B.E. = 397.3 eV), while the Ti-C and C-N bond are confirmed by the Ti 2p3/2 (B.E. = 455.1 eV), C 1 s (B.E. = 281.5 eV and 287.8 eV) and N 1 s (B.E. = 295.8 eV).

The ORR activity of various catalysts containing TNTCNHS, TiN/TiCN/Ni thin film (TNTCNTF), broken hierarchical structure of blending TiN/TiCN nanocomposites (TNTCNNC, see the synthesis method in Supporting Information and SEM images in Figure S2), commercial TiN nanoparticles (TNNP), TiCN nanoparticles (TCNNP) and Pt/C (20 wt%) are shown in Figure 3 and Figure S3–S8. Linear sweep voltammetry (LSV) with rotating disk electrode (RDE) studies of TNTCNHS was conducted at room temperature and in 0.1 M KOH electrolyte. Figure 3A illustrates the polarization curves of TNTCNHS-1 at different rotating speed in O2-saturated alkaline solution, indicating that the onset potential is ~0.93 V vs. RHE and the limiting diffusion current density increases with increasing rotation rate. The corresponding Koutecky-Levich (K-L) plots are shown in Figure 3B from 0.3 V to 0.6 V and the average number of electrons transferred (n) during the ORR are calculated to be ~4.0 for TiN/TiCN hierarchical structures. TNTCNHS-2 which owns smaller interface/mass ratio than TNTCNHS-1 is also revealed as a good ORR catalysts and the corresponding linear sweep voltammetric curves for RDE testing is displayed in Figure 3C. The average of transfer electron number is ~3.96 calculated from its K-L plots (Figure S3). The ORR polarization curves and K-L plots of other catalysts are also determined exhibited in Supporting Information. For more detail discussion, Tafel plots of the steady-state polarization curves were carried out to investigate oxygen reduction reaction rate. The corresponding Tafel slopes of TNTCNHS-1, TNTCNHS-2 and Pt/C are shown in Figure 3D, which present closed values indicating that TNTCNHS catalyst will lead to a fast increment of reaction rate with increasing overpotential as well as platinum33.

Figure 3
figure 3

(A) Polarization curves of TNTCNHS-1 at rotating speed of 400–2500 rpm in O2 – saturated 0.1 M KOH. (B) The corresponding K-L plots of TNTCNHS at 0.6 V–0.3 V. (C) Polarization curves of TNTCNHS-2 in O2 – saturated 0.1 M KOH. (D) Corresponding Tafel plots of TNTCNHS-1, TNTCNHS-2 and commercial Pt/C. The catalysts loading are all kept at ~0.1 mg cm−2.

Full size image

Summarized LSV curves for TNNP, TCNNP, TNTCNTF, TNTCNNC, Pt/C and TNTCNHS are shown in Figure 4A where our prepared TiN/TiCN hierarchical structures reveal a closed onset potential (ΔEonset = ~30 mV) and half-wave potential (ΔE1/2 = ~38 mV) to Pt/C, even the current density has overtaken platinum's obviously at the more negative potential than 0.7 V. Electrochemical impedance spectroscopy was also utilized to test the electron transfer resistance (Ret) of TNTCNHS, TNNP and TCNNP at 0.8 V where a mixed kinetic-diffusion control region for ORR existed (Figure S9). The Nyquist plots demonstrated the hierarchical structures show smaller Ret than single components of TiN and TiCN. Therefore, the TNTCNHS is the most effective catalysts in shuttling charges from electrode to solution and the catalytic activity for ORR can be remarkably promoted34. The kinetic current density of various catalysts was also calculated to compared the activity for ORR, which is consider as a convincing parameter in electrochemical reaction process. Figure 4B demonstrates the comparisons of kinetic current density and electron transfer number at 0.4 V vs. RHE, where our prepared TNTCNHS displays a great enhancement (17.8 mA cm−2) refer to TNNP (1.5 mA cm−2) and TCNNP (3.3 mA cm−2) as well as even obviously superior to Pt/C (14.8 mA cm−2).

Figure 4
figure 4

(A) Compared polarization curves of TNNP, TCNNP, TNTCNTF, TNTCNNC, TNTCNHS and Pt/C at 1600 rpm. The catalyst loadings for RDE are all kept at 0.1 mg cm−2. (B) Bar diagram of the calculated kinetic current and electron transfer number.

Full size image

The durability of the TNTCNHS and commercial Pt/C catalysts was also evaluated by using a chronoamperometric method at 0.7 V (Figure 5A). The TNTCNHS shows higher stability (holds ~95% current density) than that of Pt/C (holds ~75% current density) with only a little decline in its activity during a 35000 s testing period. Moreover, Methanol tolerance is an important requirement for cathode materials in direct methanol fuel cells (DMFC) and also a remarkable shortcoming of Pt-based catalysts. The inset figure shows almost no effect on the oxygen reduction activity of TNTCNHS although 1 M methanol was added to the electrolyte, while the current density of Pt/C dropped more than 15% immediately.

Figure 5
figure 5

(A) The chronoamperometric responses for the ORR on the TNTCNHS-1 and commercial Pt/C catalyst at 0.7 V. Inset shows the current response with 1 M methanol added. Rotating speed: 400 rpm. (B) Single cell testing of the TNTCNHS-1 and commercial Pt/C as cathode catalysts with a loading of 4 mg cm−2 and 2 mg cm−2, respectively. Anode catalysts are both platinum on carbon (mPt = ~0.4 mg cm−2).

Full size image

The single cell testing was carried out in this study shown in Figure 5B. A H2 - O2 alkaline fuel cell (AFC) was constructed to compare our prepared TNTCNHS and commercial Pt/C for cathode materials. As we see, the I–V curves reveal closed performance of two catalysts where the open circuit potential of TNTCNHS (0.96 V) has overtaken platinum (0.92 V) and the maximum power density are both ~100 mW cm−2 for Pt/C and ~82 mW cm−2 for TNTCNHS-1. Considering such excellent performance as AFC cathode catalyst, we believe it would be a promising non-precious material for the electrocatalysis of oxygen reduction reaction to take place Pt-based catalysts in practical applications.

Discussion

To investigate the improved ORR performance TNTCNHS-1 referring to TNTCNHS-2, we took SEM images of the cross-view (Figure 6A). The thickness of two catalysts are observed clearly which proves TNTCNHS-1 owns thinner TiN and TiCN layers (2 μm vs. 8 μm). While the mass loadings of the catalysts on the disk electrode are kept identical, it is obvious the hierarchical structures with thinner layers would expose more surface and interface. To find electrochemical active surface area (EASA), we measured the non-faradic charged - discharged current associated with electrochemical double layer capacity (Cdl) upon cyclic voltammetry35,36. Figure S10 shows larger Cdl of TNTCNHS-1 than TNTCNHS-2 indicating larger EASA. The high specific surface area of TiN or TiCN haven't ensured such superior activity that TNTCNHS exhibits, because neither TiN nor TiCN NPs can be as efficient electrocatalysts for ORR (see in Figure S4 and S5). By contrasted with TNTCNNC (interface is destroyed), the interface between TiN and TiCN layers probably makes the contribution to ORR (a scheme is shown in Figure 6B). TNTCNHS-1 holds almost four times larger area of the interface than TNTCNHS-2, which brings the enhancement of ΔJ = ~0.5 mA cm−2 at 0.4 V. A deep insight about the active interfaces in this study is carried out by the calculation in the following of the discussion.

Figure 6
figure 6

(A) SEM images of the cross-view of TNTCNHS-1 and TNTCNHS-2 before Ni template removed. (B) Schematic the active interfaces of TNTCNHS.

Full size image

It seems to understand why the hierarchical structures of TiN/TiCN have such a high activity is important; XPS of the sample was studied in Figure S11, which was collected after going through the process of oxygen reduction. Strong binding energy of Ti-O was examined at the high revolution spectrums of Ti 2p and O 1 s (458.0 eV, 531.2 eV and 529.5 eV). Thus, the titanium atoms in TiN and TiCN crystals are probably considered as the active sites in electrocatalytic reaction. In a typical ORR, firstly, oxygen molecules are adsorbed at the surface of the catalyst via Ti-O bond and then O-O bond is dissociated for further reduction to OH− in alkaline solution. If the O-O bond is not dissociated, undesirable HO2− is obtained27. The ORR performance, therefore, is majorly depended on two crucial steps that one is the sites for oxygen chemisorption and the other is the electron injection from catalyst to electroactive substance. The traditional adsorption is single oxygen atom binding with the active site atom on the surface of catalyst via chemical bond such as the path I and path II in the model of Figure 7A. However, this mode requires the active site atom has enough high positive charge density to restrain free molecular oxygen. On the other aspect, the O-O bond is stable so that it is hard to dissociate with low energy supplied. Hence, we proposed another adsorption mode at our high ORR activity catalyst of TNTCNHS that the two oxygen atoms in one molecule could be reacted with Ti atoms from double-side at the interface of the TiN and TiCN simultaneously (as the path III). The TEM image of the interface between TiN and TiCN is fortunately recorded which shows the distance ranged from ~0.2 nm to 1 nm in Figure 7B. As we know, the diameter of oxygen molecule is ~0.3 nm so that it would be reasonable for the proposed adsorption mode. To confirm the observed interface is indeed between the TiN and TiCN, we compare the FFT patterns of the two particles in Figure S12, which are fitting well with the results in Figure 2D and 2E, indicating both TiN and TiCN are presence. Another two modes for oxygen adsorption which were the TiN - O2 - TiN and TiCN - O2 - TiCN systems (path IV and path V) were also considered for the situations of TNNP or TCNNP catalyst, although most of the particles were separated from a related long distance so that the double-side adsorption almost couldn't take place. The adsorption energies (Ead) of the five modes were also obtained by density functional theory (DFT) calculations (Figure 7C, see the Supporting Information for the molecular models of Figure S13). Our proposal path III displays lower Ead (−4.09 eV) than other four indicating that this adsorption mode could be more efficient and easier for Ti-O bonds generation. In addition, the porous structure of the TNTCNHS also provides a number of tunnels for the transport of O2 and electrolyte to the interface. The charge density of the Ti atoms in TiN and TiCN are also calculated, which is 0.74 and 0.72 respectively. The gap of the electric charge at the two sides of active site could be positive for the dissociation of the adsorbed electronegative oxygen molecule because the electrons of O-O bond would sides of the catalyst to O2. As for TNNP and TCNNP, most adsorptions take place on single oxygen atom (path I and path II, Ead = −1.29 and −1.99 eV respectively) and only a few O2 might be absorbed by two Ti atoms (path IV and path V, Ead = −1.08 and −3.60 eV respectively). Moreover, these paths couldn't weaken the stability of O-O bond so that there aren't satisfied activities, concerning the onset potential, limit diffusion current and transfer electron number observed from Figure S4 and Figure S5. Because of limited region for the double-side adsorption, most of space is remained to transport the electrolyte, reactants and products between the external solution and the interface. Figure S14 reveals a schematic process for the entrance and exit of OH− and O2. Further study has started to investigate the interface via more sound approach and to extend this active specific area is also needed in the following work.

Figure 7
figure 7

(A) A model for the three adsorption paths of O2 on TNTCNHS. (B) The calculated adsorption energies of three possible adsorption paths on TNTCNHS and other two paths on TiN/TiN and TiCN/TiCN hierarchical structures (path IV and path V).

Full size image

In summary, we construct TiN/TiCN hierarchical structures via layer-by-layered CVD method. After removal of template, the obtained powder (TNTCNHS-1) displays an enhanced ORR activity in alkaline electrolyte, which exhibits a low onset potential of 0.93 V, high limit diffused current density of 5.1 mA cm−2 at 1600 rpm and smaller Tafel slope of 79 mV dec−1. The stability and single cell testing also reveal the more excellent performance of TNTCNHS-1 compared with commercial Pt/C. Moreover, the interface between TiN and TiCN layers is considered as an active region for oxygen adsorption. Such high performance cathode catalyst for AFC provides three adsorption modes for ORR which displays strong oxygen utilization. Further, TiN and TiCN also have good electric conductivity, excellent wear and high temperature resistance30,31,32, so it would be employed at both ambient and extreme conditions. Chemical vapor deposition have been a mature and large-area productive technique for the relatively low-cost manufacture with industrial-scale facilities, as well as the strategy for the construction of hierarchical structures is facile, rapid, cheap and universal. Thus, to build the active interfaces on double-layered architecture like TNTCNHS might enable more electrocatalysts to achieve a new class of activity that they can rarely reached in traditional constructions.

Methods

TiCN film was firstly coated on Ni foams (2 * 4 cm) via chemical vapor deposition. In a typical CVD process, TiCl4 (Kermel Chemical Reagent Company, China) was chosen as a Ti source with N2, CH4 and H2 (TiCl4: N2: CH4: H2 = 2:1:1:100, v/v) to pass into a CTI-C280A MT-CVD device system (Φ240 × H500, Chengdu Tool Research Institute Co. Ltd, China) at 700 ~ 900°C for 1 h. About 1200 pieces of Ni foam with total surface area of more than 10000 cm2 can be deposited simultaneously. And then TiN film was also coated on the same Ni foams in a similar procedure with TiCl4, N2 and H2 (1:1:25, v/v) for ~1 h. After cooled to room temperature, Ni foam coated with TiN/TiCN films was carried out and washed by distilled water and ethanol. Afterwards, the coated Ni was milled to powder and the TiN/TiCN thin films (TNTCNTF) were obtained. Then Ni foam template was removed by 2 M hydrochloric acid at 50°C for 15 h followed with ultrasonic dispersion to homogenous. Finally, the TiN/TiCN hierarchical structures catalyst was filtered, rinsed by distilled water and ethanol for three times and dried at 60°C in vacuum. The thickness of the TiCN and TiN layers can be adjusted by controlling the flow rate (ν) of the hydrogen (νTNTCNHS-1 = ~40% νTNTCNHS-2 = ~800 cm2 min−1) We also prepared the broken hierarchical structure of blending TiN/TiCN nanocomposites (TNTCNNC) with the same CVD process but coating on a smooth nickel plate. Next the Ni plate was directly immersed into 4 M hydrochloric acid at 60°C for 24 h without need of milling to powder. After ultrasonic dispersion to homogenous, the TNTCNNC was filtered, wash and dried as above described.