Shangfeng Jiangab,
Baolian Yia,
Qing Zhaoab,
Hongmei Yu*a and
Zhigang Shao*a
aFuel Cell System and Engineering Laboratory, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian 116023, PR China. E-mail: hmyu@dicp.ac.cn; zhgshao@dicp.ac.cn
bUniversity of Chinese Academy of Sciences, Beijing 100039, PR China
First published on 16th February 2017
Ordered titanium dioxide nanorod arrays are prepared by a hydrothermal method and applied to the catalyst supports of the palladium–nickel catalysts (Pd2Ni3–TiO2) for formic acid oxidation. Due to the three-dimensional catalytic structure and alloying effect, the Pd2Ni3–TiO2 exhibits superior mass activity and long-term stability than those of Pd–TiO2 and commercial PdC catalysts. The present work opens a new approach to design catalysts with superior catalytic performance and durability for direct formic acid fuel cells with three-dimensional catalytic structure.
Previous studies have proven that controlling the sizes and morphologies of Pd-based catalysts is an effective way to improve the catalytic performance for FAO. Several Pd-based catalysts have been demonstrated, such as Pd nanowires,8 Pd nanorods,9 and the combination of Pd with transition metals (e.g., Ni, Co, Fe, Cu), Ag or P to form bimetallic compounds.10–16 Nevertheless, the catalytic structure and the support material also have an impact on the catalytic activity of the electrocatalysts.17 Metal oxides can be used as the support materials in fuel cells, due to their high corrosion resistance, reasonable surface area, and mechanical strength.18,19 Among them, Ti-based oxides has been applied as the catalyst support material of fuel cells. Subramanian et al. applied anodized TiO2 nanotubes (T_NT) as the Pt support for simultaneous FAO and photo-assisted FAO, which generates around 39-fold higher current density than that of the pure T_NT.20 Xing et al. investigated the fine-tuned pore structured TiO2–C as the Pt or Pd catalyst supports for the electro-oxidation of methanol or formic acid, which showed a much superior catalytic performance than that of the carbon support.21
Based on the above considerations, we present TiO2 NRs as the catalyst supports of Pd-based catalysts. To the best of our knowledge, the Pd-based alloy nanoparticles supported on the TiO2 NRs are rarely reported as the catalysts toward FAO. The advantages of this approach are listed: (1) high transport rate of HCOOH; (2) high electrochemical surface area (ECSA) and high utilization of electrocatalysts; (3) direct charge carrier transport pathways. More importantly, we combine high catalytic activity and advanced catalytic structure to show the highest electrochemical activity. Compared with the commercial PdC catalyst and pure Pd–TiO2 catalysts, the Pd2Ni3–TiO2 catalyst exhibited higher electrochemical activity and stability.
The XRD analysis was first applied to monitor the corresponding crystal phase of all the samples and is displayed in Fig. 1. The as-prepared products are mainly composed of rutile TiO2 phases according to JCPDS 21-1276. The Ti 2p XPS spectrum of the as-prepared sample is shown in Fig. S2.† The binding energies of Ti 2p1/2 and Ti 2p3/2 are 458.58 eV and 464.38 eV, respectively. Since the binding energy of Ti4+ is 458.5 eV, the XPS data prove the presence of TiO2. After deposition of the Pd and PdNi nanoparticles, the (111) peak for the Pd–TiO2 and PdNi–TiO2 is obviously in the XRD pattern (Fig. 1). Due to the low metal content in the products, the diffraction peaks of the (200), and (220) planes of the Pd-based catalysts are not specified in the XRD patterns.
FESEM images of Pd–TiO2 NRs and PdNi–TiO2 NRs grown on carbon paper are shown in Fig. 2a and b (inset). The oriented TiO2 nanorods grew on the surface of carbon fibres, and the mean diameters of the TiO2 nanorods are about 120 nm (Fig. S1†). After deposition of the Pd-based catalysts onto the TiO2 NRs, the resulting samples are heat-treated in 5% H2/Ar gas at 300 °C for 2 h. The morphology and structure of the as-prepared samples are not obviously changed compared with that of TiO2 NRs. The TEM image of a TiO2 nanorod shows that the surface of the TiO2 nanorod is smooth (Fig. 2c). The Pd nanoparticles and PdNi nanoparticles are dispersed on the TiO2 nanorod (Fig. 2d and e) and the size distributions of the Pd and PdNi nanoparticles are shown in Fig. 2d and e (inset) and S3.† Fig. 2d and e show that the mean sizes of the Pd and PdNi catalysts are about 3.52 nm and 4.24 nm, respectively. However, the sizes of some other PdNi particles are about 20 nm (Fig. S3†), which are larger than Pd nanoparticles. The average sizes of the Pd and PdNi catalysts according to XRD are 6.09 nm and 7.53 nm, respectively, which are larger than the particle sizes of the Pd-based catalysts. Since the general detection depth of XRD measurements is around 10 nm, the composition detected by XRD indicates both surface and bulk atoms of the Pd-based catalysts.
The HR-TEM image reveals that the PdNi particles are crystalline and the lattice structure is clear (Fig. 2f). The (111) lattice spacing of the PdNi nanoparticles was measured to be 0.219 nm, which is between that of Pd (0.223 nm) and Ni (0.20 nm), confirming the formation of PdNi alloy in the as-prepared samples.25 EDX analysis was also carried out to measure the composition of the PdNi–TiO2 catalysts. The spectrum confirms that Pd and Ni elements exist in the PdNi nanoparticles and their atomic ratio is 1:1.72 (Fig. S4†). In addition, the Pd and Ni content in the as-prepared samples is estimated by ICP-OES, and the results are 40.55 μg cm−2 and 14.4 μg cm−2, respectively. Thus, the atomic ratio of Pd and Ni in the PdNi–TiO2 catalysts is 1:1.55 according to the result of ICP-OES, which is consistent with the EDX data. Thus, the PdNi–TiO2 catalysts can be named as Pd2Ni3–TiO2 according to the data of ICP-OES. Meanwhile, the Pd loading of the Pd–TiO2 NRs is measured as 29.7 μg cm−2.
XPS was applied to analyze the chemical states and electronic states of Pd and Ni in the as-prepared samples. The Pd 3d spectra for the PdNi–TiO2 NRs and Pd–TiO2 NRs are shown in Fig. 3a. The Pd 3d XPS spectra of Pd–TiO2 NRs show the binding energies of Pd0 3d (3d5/2 = 335.26 eV; 3d3/2 = 340.53 eV, Fig. 3a), which are slightly higher than those of bulk Pd (3d5/2 = 334.90 eV; 3d3/2 = 340.15 eV).26 The slight shift of binding energies of Pd0 to higher ones may be attributed to the significant interaction between Pd nanoparticles and TiO2 supports. Meanwhile, after the formation of PdNi alloys, the Pd 3d binding energy of Pd2Ni3–TiO2 NRs is shifted to a more negative position than that of Pd–TiO2 NRs. This shift may be attributed to the difference of the electronegativity between Pd and Ni, resulting in a partial electron transfer from Ni to Pd, which would change the penetration of outer electron density of Pd.27 Fig. 3b shows that Ni(OH)2 and NiOOH with obvious shake-up peaks can be seen at higher binding energies than those of the high-energy band (Ni 2p3/2) and the low-energy band (Ni 2p1/2).25
Fig. 3 XPS spectra of (a) Pd 3d carried out on the Pd–TiO2 NRs and PdNi–TiO2 NRs and (b) Ni 2p carried out on the PdNi–TiO2 NRs. |
To investigate the electronic activity of the as-prepared Pd2Ni3–TiO2 NRs, we employed FAO as the model system, comparing with pure Pd–TiO2 NRs and commercial PdC. Typical Pd electrochemical behaviours of all the Pd-based electrocatalysts and the TiO2 sample were observed in 0.5 M H2SO4 (Fig. 4a and S5†). The ECSA was estimated according to the area of hydrogen adsorption peaks and the calculation equation is according to that previously reported.17,28,29 The CVs of Pd–TiO2, and Pd2Ni3–TiO2 are shown in Fig. 4a. The calculated ECSAs of Pd–TiO2, Pd2Ni3–TiO2, and commercial PdC are 53.6 m2 g−1, 43.6 m2 g−1, and 37.6 m2 g−1, respectively. Since some PdNi nanoparticles are larger than the Pd nanoparticles at the surface of the TiO2 nanorods, the ECSA of the Pd2Ni3–TiO2 NRs is smaller than that of the Pd–TiO2.
The electrocatalytic properties of the Pd2Ni3–TiO2 catalysts towards FAO were investigated in a 0.5 M H2SO4 + 0.5 M HCOOH mixture. The electrocatalytic properties of the Pd–TiO2 and the commercial PdC toward FAO were measured for comparison. Fig. 4b shows the CV curves of FAO of different catalysts. All the current densities of FAO of different samples were normalized by Pd loading (mass activity). The TiO2 samples show no catalytic performance in the 0.5 M H2SO4 and 0.5 M H2SO4 containing 0.5 M HCOOH mixture (Fig. 4a, b, S5a and S4b†). Meanwhile, after deposition of Pd nanoparticles, the as-prepared Pd–TiO2 catalyst exhibits high catalytic activity as its mass density is 371.87 mA mg−1, which is around 1.17-fold higher than that of commercial PdC (Fig. 4c). The enhancement of the onset potential of the as-prepared Pd–TiO2 catalysts compared with the commercial PdC for FAO might be explained by the formation of a three-dimensional catalytic structure. The performance will be further improved by adding Ni element to Pd. Many studies shows that the highest FAO performance of a PdNi catalyst is when the atomic ratio of Pd and Ni is around 1:1.10,25,30 After alloying with Ni, the catalytic activity of Pd2Ni3–TiO2 is further improved (753.42 mA mgPd−1, Fig. 4b), indicating the electronic structure modification after Pd alloying with Ni, as demonstrated in the XPS analysis, which has been observed in many other Pd-based alloys.31 The FAO reaction occurs over the surface atom layer of the Pd-based catalysts for the test samples. With a similar amount of Pd–TiO2, Pd2Ni3–TiO2 shows better catalytic activity toward the FAO than Pd–TiO2. Ni element does not catalyze the FAO reaction, thus the Ni must be improving the catalytic activity of the Pd element toward FAO in such a way that the FAO rate is increased consequently.32 One possible effect is that the Pd surface electronic structure is perturbed by Ni.33 The d-band center of the catalysts will be altered and the electronic structure of the Pd surface layer will be affected by the formation of the PdNi alloy.34 During the FAO reaction, the HCOOad (HCOO–Pd) is formed firstly through breaking the O–H bond of HCOOH.35 The possible reaction steps are listed as follows.35,36
HCOOH + Pd → HCOO–Pd + H+ + e− | (1) |
HCOO–Pd → Pd–H + CO2 | (2) |
Pd–H → Pd + H+ + e− | (3) |
After the Pd d-band center is shifted to an optimum value, the Pd-adsorbate bond strength will be optimized.32 We believe that the Pd d-band center is shifted to an optimum value after alloying with Ni, thus Pd2Ni3–TiO2 shows the best catalytic activity toward FAO. At the same time, the strong interactions between TiO2 and metals (strong metal–support interaction, SMSI) have been shown in previous research,37 thus the stability of the catalysts will be enhanced. Compared with the oxidation peak potential of formic acid in the literature,38 Pd2Ni3–TiO2 did not show excellent catalytic activity for FAO with a more positive oxidation peak potential, and the possible reason is the low electronic conductivity of TiO2 though oriented TiO2 NRs, which were hydrothermally grown on carbon fibers. Meanwhile, compared with the similar state-of-the-art Pd-based electrocatalysts, the as-prepared Pd2Ni3–TiO2 exhibits higher mass activity than most of them (Table S1†).
To further evaluate the surface poisoning, CA curves of Pd–TiO2, Pd2Ni3–TiO2, and commercial PdC were investigated in 0.5 M H2SO4 + 0.5 M HCOOH with the potential of 0.25 V for 4000 s during the measurements (Fig. 4d). It is obvious that the decrease in current density for the FAO on the Pd2Ni3–TiO2 NRs is slower than that of the Pd–TiO2 NRs and commercial PdC, which indicates its higher tolerance to the intermediaries during FAO. The improvement of the catalytic performance is possibly attributed to the alloying effect and the advanced catalytic structure.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ra00194k |
This journal is © The Royal Society of Chemistry 2017 |