Yunhua Li*,
Xin Yang,
Linhui Zhu,
Hua Zhang and
Binghui Chen
Department of Chemical and Biochemical Engineering, College of Chemistry and Chemical Engineering, National Engineering Laboratory for Green Chemical Productions of Alcohols-Ethers-Esters, Xiamen University, Xiamen 361005, PR China. E-mail: yunhuali@xmu.edu.cn
First published on 14th September 2015
This study investigates phenol hydrodeoxygenation on supported Ni2P, prepared via the sol–gel and TPR methods, and noble metal (Pd, Pt and Ru)–Ni2P catalysts, prepared from Ni2P by the partial in situ reduction of a noble metal precursor. 10% Ni2P/SiO2 had a relatively uniform distribution of Ni2P nanoparticles and a highly active activity for phenol hydrodeoxygenation. Phenol conversion increased with increasing reaction temperature and the main products on noble metal–Ni2P/SiO2 also changed from cyclohexanol at 453 K to cyclohexane at 493 K. In comparison with supported Ni2P or noble metal catalysts and their physical mixture, Pd–Ni2P/SiO2 presented the highest conversion activity and cyclohexane selectivity. Physicochemical characterization showed that the number of active sites on the catalysts increased and electron transfer occurred from Ni2P to the noble metal due to the intimate contact between Ni2P and the noble metal. A synergistic effect of the deoxygenation and carbonyl hydrogenation from Ni2P and the hydrodeoxygenation from Pd resulted in an improvement of the catalytic activity and differences in the selectivity of the catalysts.
Up to now, hydrodeoxygenation catalysts mainly include the traditional MoS2, noble metal or transition metal phosphide based catalysts.2–4 Among them, nickel phosphide has been prepared using the traditional temperature programmed reduction (TPR) method. However, there are still problems such as high preparation temperatures and the large and uncontrollable particle sizes of nickel phosphide with this method. For this reason, several other processes have been developed, such as the solution-phase method,5 the decomposition of metal hypophosphites6 and converting metal nanoparticles into metal phosphides.7
At the same time, a series of composite catalysts for the hydrodeoxygenation of guaiacol or phenol, the common model chemicals in biofuel, have also been studied.8,9 For example, Yan et al.10 catalyzed the hydrodeoxygenation of phenols using Ru nanoparticle catalysts combined with Brønsted acidic ionic liquids. Hong et al.11 prepared Pd/WOx/γ–Al2O3 for guaiacol hydrodeoxygenation. With respect to bimetallic catalysts containing a transition metal phosphide, Pd–Ni2P/C has presented a 1.5 times higher power density than commercial Pd for direct formic acid fuel cells.12 Ni2P could also enhance the activity and durability of the Pt anode catalyst in direct methanol fuel cells.13 In addition, some researchers also noticed that the programmed reduction temperature for phosphate decreased and the hydrodesulfurization activity increased with the addition of Pd or Pt into a Ni2P supported catalyst.14 However, these catalysts have usually been obtained by impregnating a passivated phosphide or phosphate with a noble metal precursor solution. It is difficult to ensure the directional loading of the noble metal and the formation of an intimate contact between the phosphide and noble metal due to the nonselective deposition of the noble metal on the support or phosphate. It is noteworthy that a transition metal phosphide requires further passivation to avoid the drastic oxidation of nickel phosphide when it is exposed to air. This means that phosphide has a certain reducibility.
Herein, supported nickel phosphide was firstly prepared from the sol–gel and the traditional TPR method. Thus, Ni2P was uniformly dispersed over SiO2. The as-prepared supported nickel phosphide was then impregnated with a noble metal precursor solution to obtain the noble metal–Ni2P catalyst. This preparation process not only favors the distribution of the noble metal and minimizes its usage due to the controllable distribution of reductant Ni2P, but also facilitates the formation of an intimate contact structure and intensifies the interaction between the noble metal and Ni2P. Catalytic HDO evaluation and catalyst characterizations were conducted on the supported Ni2P, noble metal and noble metal–Ni2P catalysts. The results showed that noble metal–Ni2P/SiO2 had higher catalytic performances for phenol hydrodeoxygenation, due to more active sites and electron transfer from Ni2P to Pd, than the supported Ni2P or noble metal catalysts. Based on these outcomes, the relationship between the structure and catalytic performance were also discussed. The results may provide valuable information for developing Ni2P or noble metal based catalysts to upgrade bio-crude-oil to feasible fuel.
The noble metal–Ni2P/SiO2 was prepared by an in situ reduction of the noble metal precursors. The detailed procedure is as follows. After the preparation of the fresh Ni2P/SiO2 in the U-tube quartz reactor was completed, the H2 flow was switched to a N2 atmosphere. This reactor filled with N2 was placed in an ultrasonic cleaner (KQ-250DE). A certain concentration of a noble metal precursor solution, such as PdCl2, H2PtCl6, or RuCl3, was injected using a syringe into the freshly prepared Ni2P/SiO2 in the U-tube quartz reactor under ultrasonic conditions. The concentration of the noble metal precursor was 0.01 g mL−1 and the corresponding liquid volume used was based on the nominal loading of the noble metal, 1%. After 40 min, the sample was washed, dried under H2 at 393 K for 2 h and passivated under 1.5% O2/N2 for 3 h.
For comparison, the supported pure noble metal catalysts were prepared using the impregnation method. The preparation of the support SiO2 was similar to that of Ni/SiO2 except for the addition of the nickel precursor. After impregnating with the noble metal precursor solution for 12 h, the samples were calcined under H2 at 723 K for 3 h.
The specific surface area was determined via the N2 adsorption–desorption isotherm method at 77 K using a Micromeritics ASAP 2020 system and calculated using the BET (Brunauer–Emmett–Teller) method. The pore structure parameters were calculated using the BJH (Barrett–Joiner–Halenda) method. Prior to the analyses, the samples were outgassed at 523 K for 3 h to eliminate volatile adsorbates on the surface.
CO chemisorption was performed (Micromeritics ASAP 2020) to provide an estimate of the number of active sites on the samples. Usually, 0.1 g of a passivated sample was loaded into a quartz reactor and re-reduced in a H2 flow (40 mL min−1) at 723 K for 2 h. After evacuating for 1 h at this temperature, the samples were cooled down to RT in a He flow (60 mL min−1). Subsequently, the first isotherm was taken. After the first isotherm was measured, the sample was evacuated for 2 h. The second isotherm was performed. The CO chemisorption obtained was calculated by subtracting the amount of adsorbed CO in the second isotherm.
To characterize the structure and composition of the catalysts, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDS) were performed using a FEI TECNAI F30. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Quantum 2000 Scanning ESCA Microprobe system with Al Kα radiation under ultrahigh vacuum. The binding energies were internally referenced to the C1s peak at 284.6 eV.
Transmission infrared spectra of the catalysts were collected in situ in a reactor cell placed in a Bruker Vertex 70 Fourier transform infrared (FTIR) spectrometer. The IR cell was equipped with CaF2 windows, connections for inlet and outlet flows, and thermocouples to monitor and control the temperature. Before introducing CO, the samples were reduced in H2 at 723 K for 3 h, then cooled to room temperature in a He flow and exposed to CO for 3 h to achieve saturation. The samples were then purged in an Ar flow for 30 min to remove the gaseous and weakly adsorbed CO species. The spectra were obtained in the absorbance mode and were presented after the subtraction of a background spectrum obtained on the freshly reduced samples.
![]() | ||
Fig. 1 Phenol conversion and product selectivities of the different Ni2P/SiO2 catalysts: (a and b) 453 K; (c and d) 493 K. |
![]() | ||
Fig. 2 TEM images and particle size distribution for the different Ni2P/SiO2 catalysts: (a) 5%, (b) 10% and (c) 15%. |
Cyclohexane, cyclohexanol and cyclohexanone are the main products over Ni2P/SiO2 at 453 K (Fig. 1). Cyclohexanol and cyclohexanone are intermediates for the phenol hydrodeoxygenation to cyclohexane.16 The higher reaction temperature favors the formation of more cyclohexane. Thus, the cyclohexane selectivity prominently rises and those of cyclohexanol and cyclohexanone dramatically drop, accompanied by an increase in phenol conversion when the reaction temperature increases from 453 K to 493 K.
![]() | ||
Fig. 3 Phenol conversion and product selectivities of the different noble metal–Ni2P/SiO2 catalysts: (a and b) 453 K; (c and d) 493 K. |
![]() | ||
Fig. 4 Phenol conversion and product selectivities of 1% Pd–10% Ni2P/SiO2 and the physical mixture of Pd/SiO2 and Ni2P/SiO2. |
Thus, a series of physicochemical characterizations were conducted to further unravel the root of the improvement of the catalytic performances of the noble metal–Ni2P/SiO2. Table 1 lists the BET surface area, pore volume, pore size and CO chemisorption for the different catalysts. As shown in Table 1, compared with the Si2O support, the BET surface area and pore volume of the 10% Ni2P/SiO2 decrease with the addition of the phosphorus precursor. This indicates that parts of the catalyst pores were blocked by the phosphorus source during the impregnation process. On the other hand, with the addition of a noble metal, the pore size and surface area increase for noble metal–Ni2P/SiO2. The increments of the pore sizes result from the consumption of part of the Ni2P. At the same time, only noble metal precursors in contact with dispersed Ni2P can convert into noble metal nanoparticles because of the Ni2P reducing the noble metal precursors during the preparation process. As a result, the increasing surface area is a compromise between the consumption of Ni2P and the high dispersion of the generated noble metal nanoparticles. Moreover, with the addition of the noble metal, the number of active sites also increases, except for 1% Ru–10% Ni2P/SiO2. If it is assumed that CO is linearly adsorbed on the nickel atoms of Ni2P, as was theoretically and experimentally shown by Nelson et al.18 and Layman and Bussell,19 respectively, and the same type of adsorption exists for the noble metal atoms, except for partially bridged CO over Pd and Ru, then increased CO chemisorption uptake implies that more active sites for hydrodeoxygenation are generated with the addition of a noble metal. As shown in Table 1, the trend for CO chemisorption uptake is 1% Pd–10% Ni2P/SiO2 > 1% Pt–10% Ni2P/SiO2 > 10% Ni2P/SiO2 ≈ 1% Ru–10% Ni2P/SiO2. This is roughly consistent with the activity result of the catalysts. Thus, the high hydrodeoxygenation capacity of Pd or Pt–Ni2P/SiO2 is partly attributed to the increase in the number of active sites.
Catalyst | Pore size (nm) | Pore volume (cm3 g−1) | BET surface area (m2 g−1) | CO uptake (μmol g−1) |
---|---|---|---|---|
SiO2–sol–gel | 9.30 | 0.98 | 421.00 | — |
10% Ni2P/SiO2 | 10.03 | 0.90 | 238.55 | 92.25 |
1% Pt–10% Ni2P/SiO2 | 11.60 | 0.85 | 293.63 | 109.51 |
1% Ru–10% Ni2P/SiO2 | 11.49 | 0.88 | 307.39 | 90.77 |
1% Pd–10% Ni2P/SiO2 | 15.62 | 1.00 | 257.12 | 121.32 |
Fig. 5 shows the XRD patterns of noble metal–Ni2P/SiO2. The XRD peaks of Ni2P(111), Pd(111), Pt(111) and Ru(111) are 40.714 (PDF#74-1385), 40.365 (PDF#87-0645), 39.796 (PDF#87-0646) and 40.772° (PDF#88-2333), respectively. The intensity of the Ni2P(111) diffraction peak is relatively weak, whereas the noble metal peak becomes prominent with the addition of noble metal. This implies that the generation of a noble metal nanoparticle is at the expense of Ni2P. The XRD patterns show that the Ni2P(111) peak shifts toward lower angles for Pd(111) and Pt(111) with the addition of the noble metal, but not for Ru–Ni2P/SiO2 with its weak diffraction. This result indicates that Pd (0.179 nm) or Pt (0.183 nm) with their larger atomic radii is doped into the Ni2P (Ni (0.124 nm)), or that the addition of Pd or Pt leads to lattice distortion of Ni2P.20 Furthermore, we believe that an interaction between the noble metals and Ni2P is formed.
HAADF-STEM mapping demonstrates the distributions of palladium, nickel and phosphorus in Fig. 6. The loading site of the phosphorous precursor is uncontrollable while the formation of Pd nanoparticles depends on Ni2P reducing the Pd precursor. As shown in Fig. 6, the distribution density of Pd is mainly consistent with that of Ni2P. This result further verifies that a controllable distribution of noble metal nanoparticles is achieved and the intimate contact structure between the noble metal and Ni2P is formed by Ni2P reducing the noble metal precursor. Combined with the CO chemisorption and XRD results, we speculate that the controllable distribution of the noble metal nanoparticle results in the intimate contact between noble metal and Ni2P, further leading to interactions between both of them and the increment in the number of active sites, compared with 10% Ni2P/SiO2.
Fig. 7 shows the XPS spectra of the different pure noble metal and composite catalysts. It has been noted that the signal to noise ratio is similar for the three couples of samples containing Pd, Pt and Ru, respectively, and that the intensity differences between the SiO2 and Ni2P/SiO2 supported samples are in the same direction. This implies that noble metal–Ni2P/SiO2 might have a stronger signal in the XPS spectra than the pure noble metal catalysts. In addition, Table S2† shows the surface atomic ratios with respect to Si, O and the noble metal of these catalysts. A comparison also showed that the composite catalysts have higher loading ratios of the noble metal on the surface than the SiO2 supported catalyst. This result might be related to Ni2P reducing the noble metal precursor into nanoparticles. The existence of reductant Ni2P contributes to the generation of more noble metal particles. On the other hand, the spectrum peak of Pd in 1% Pd–10% Ni2P/SiO2 shifts toward a lower binding energy compared to the pure noble metal catalysts. The XPS signal of the supported metal particles shifts to a higher binding energy with decreasing particle size.21 Meanwhile, the electrons transferring from Ni2P to the noble metal render the latter electron rich and move the corresponding signal toward a lower binding energy. Therefore, the Pd spectrum peak shifting toward the lower binding energy in Fig. 7 is a compromise, resulting from the smaller noble metal particles and electron transferring. However, this electron transferring is likely the reason for the high catalytic performance of Pd–Ni2P/SiO2. As proposed by Ueckert et al.,22 a metal site with a high electron density favors the formation of a π back bond between the aromatic ring and metal sites, which promotes the hydrogenation of the aromatic ring. Moreover, since the lowest unoccupied molecular orbital (LUMO) of C–O is antibonding and surfaces that are able to transfer electron density to this orbital facilitate the dissociation of the C–O bond,23 the increased electron density on Pd accounts for the higher hydrodeoxygenation catalytic activity of Pd–Ni2P/SiO2. For product selectivity, cyclohexanol increases while cyclohexanone decreases over the noble metal–Ni2P/SiO2 catalysts due to the improvement in the hydrogenation capacity in comparison with 10% Ni2P/SiO2 at 453 K. As mentioned above, the synergistic effect of the hydrodeoxygenation from Pd and the deoxygenation and carbonyl hydrogenation from Ni2P results in the promotion of the catalytic performance of the noble metal–Ni2P catalysts. Herein, this effect is mainly due to the interaction between Ni2P and the noble metal and the electron transfer from Ni2P to Pd on Pd–Ni2P/SiO2. As the reaction temperature rises from 453 K to 493 K, the main product changes from cyclohexanol to cyclohexane because of the intensifying deoxygenation at higher temperatures. In contrast, although electrons also transfer from Ni2P to Pt on the noble metal–Ni2P catalysts,11 1% Pt–10% Ni2P/SiO2 exhibits a lower catalytic selectivity as well as a higher conversion than Ni2P/SiO2. That is, the synergistic effect of Pt and Ni2P on the catalytic performance is less prominent than Pd–Ni2P/SiO2. Ru exists in the form of Ru4+ in 1% Ru–10% Ni2P/SiO2 because Ru is prone to oxidation again. Ru that has been oxidized and fewer active sites on 1% Ru–10% Ni2P/SiO2 account for a lower catalytic performance than of the other composite catalysts.
The electronegativity of an element represents the capability of an atom to draw an electron to itself. A larger electronegativity means that this atom has a greater ability to draw electrons. For Ni2P, Niδ+ (0 < δ < 2) covalently bonds with Pδ− (0 < δ < 1). Ni has the lower electronegativity (1.91) while the value for phosphorus is 2.19. There is a small electron transfer from Ni to P in nickel phosphide. In contrast, the electronegativities of Pd, Pt and Ru are 2.20, 2.28 and 2.20, respectively. A noble metal in contact with Ni2P probably accepts electrons from nickel due to the larger electronegativity of the noble metal.
CO-FTIR spectroscopy was also performed to further verify the electronic contribution of the supported catalysts (Fig. 8). The FTIR spectra of adsorbed CO shifts from 2077 cm−1 to higher wavenumbers, 2094, 2083 and 2079 cm−1 with the addition of Pd, Pt and Ru, respectively. According to the Blyholder model,24 this shift toward higher frequencies is indicative of the strengthening of the CO bond due to decreased donation of the metal d-electrons into the CO π* antibonding orbital by back-electron transferring.25 Since the presence of Pd gives rise to the electron donation from Ni2P to Pd, electron enrichment on Pd and electron deficiency on Ni2P occur, as also reported by the XPS data. Thus, metal d-electron transfer into the CO π* antibonding orbital decreases and the stretching vibration of the adsorbed CO shifts toward higher frequencies.
![]() | ||
Fig. 8 Infrared spectra of adsorbed CO on 10% Ni2P/SiO2, 1% Pd–10% Ni2P/SiO2, 1% Pt–10% Ni2P/SiO2 and 1% Ru–10% Ni2P/SiO2. |
In addition, the vibration frequency of CO adsorbed on the catalysts between 2077 and 2094 cm−1 also indicates that most of the CO is linearly adsorbed on noble metal–Ni2P/SiO2, in agreement with the assumption for the determination of active sites in CO chemisorption. However, we also notice that the intensity change of the CO stretching vibration with the addition of the noble metal is not consistent with the active site number by CO chemisorption. Utilizing CO-FTIR for quantification is inaccurate due to the effects of the loading of catalysts and the roughness of the tested surface. In comparison with CO-FTIR, CO chemisorption is a typical method to determine the active sites of catalysts. Therefore, herein CO-FTIR is performed to only confirm the electron transfer from Ni2P to the noble metal on the catalyst surface.
The evaluation and characterization results suggested that the key factors influencing the catalyst activity and the selectivities toward different products were the increased number of active sites and the electron transfer from Ni2P to Pd upon addition of the noble metal as well as the appropriate reaction temperature and time. Furthermore, the increment of the active sites and the electron transfer are from the reduction of the noble metal precursor on Ni2P/SiO2 and the interaction between Ni2P and the noble metal. Based on experimental results, a possible deoxygenation scheme over noble metal–Ni2P/SiO2 is also suggested as follows (Fig. 9). Pure Ni2P has a higher selectivity for cyclohexanone and cyclohexanol. In contrast, cyclohexanone can continue to hydrogenate into cyclohexanol over noble metal–Ni2P/SiO2 under the same reaction conditions due to the synergistic effect of Ni2P and Pd. When the reaction temperature further increases or the hydrodeoxygenation time is prolonged, cyclohexanol can continue to convert into cyclohexane over noble metal–Ni2P/SiO2. In terms of the Ni2P catalyst system, benzene, however, is generated rarely.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra11203f |
This journal is © The Royal Society of Chemistry 2015 |