22nd International Symposium on Plasma Chemistry July 5-10, 2015; Antwerp, Belgium Iron functionalization of graphene nanoflakes in thermal plasma conditions for catalyst applications U. Legrand, N.Y. Mendoza-Gonzalez, J.-L. Meunier and D. Berk Plasma Processing Laboratory (PPL), McGill University, Chemical Engineering Montreal, Quebec, Canada Abstract: Graphene nanoflakes (GNFs), a stack of 5 to 20 layers of graphene sheets, are the product of methane decomposition by a thermal plasma. The GNFs were functionalized with nitrogen to support atomic iron and create catalytic sites. The possibility to evaporate iron particle has been proven by modelling. The iron functionalization step is realized within the reactor, by the partial evaporation of iron powder injected in the argon plasma. Keywords: graphene nanoflakes, thermal plasma, iron, catalytic sites, evaporation model Introduction Decreasing the price of catalyst for polymer membrane electrolyte fuel cells (PEMFC), and replacing the platinum based catalyst by a corrosive resistant and nonnoble catalyst are major issues to allow PEMFC to reach larger markets. The non-noble catalyst which has been proposed in 2011 by Pristavita [1][2] is based on the implementation of catalytic sites on well crystallized carbon matrix, the graphene nanoflakes (GNFs) . The catalytic sites are formed by an atom of iron linked to nitrogen atoms, and are inspired by the work of Jasinski, where he observed a high activity of iron phthalocyanine in the oxygen reduction reaction. While it has been shown that nitrogen can be efficiently added to GNFs within the thermal plasma system resulting in low (~2 at%) and high (~16 at%) functionalization levels [3], the present study is focusing on the optimization of the iron functionalization step within the reactor. With this objective in mind, a modelling study of iron particle evaporation, and experimental optimization of the iron particles feeding within the reactor to get nitrogen and iron functionalized GNFs (N/Fe-GNFs) are presented. heat-mass transfer calculations are based on the force balance on the particle and on the convective heat and mass transfer from the particle, using the local continuous phase conditions as the particle moves through the flow. The following assumptions are considered for the discrete phase model: particles are spheres and the fluid is an ideal gas, the gas-particle system is diluted, the thermophoresis effect is considered as an external force on the transport of particles. This effect is responsible for small particles in a temperature gradient being driven from high to low temperature regions. Finally, gravity and turbulence have no effect and there is no particle radiation interaction. Because in the discrete phase model the particle phase is regarded as a source of mass, momentum, and energy to the gaseous phase, the following equations are used to calculate the set of heat and mass transfer laws involved. The trajectory of a discrete phase particle is calculated by integrating the force balance on the particle as shown in Eq. 1 This force balance equates the particle inertia with the forces acting on the particle. (Eq. 1) Methods A. Particle evaporation model In the synthesis of GNF by ICP plasma, solid micron sized iron particles are injected into the plasma core to be evaporated. The choice of using iron powders instead of some liquid-based iron precursors follows a stringent requirement on the purity of the resulting Fe/N-GNF structure for fuel cell catalyst applications. The modeling of the evaporation of the raw iron particles is solved with a discrete phase model using the PSI-Cell approach [4] in a Lagrangian frame of reference. The calculation method implies that raw particles are considered as a second phase composed of spherical particles dispersed in the continuous phase. The effects of the particles are considered negligible, which means that the volume fraction has not to exceed 10-12%. The particles trajectories are computed using the equations of the movement of a rigid particle in a fluid. The trajectory and P-II-7-14 ππ ππ’π 1 1 = πΆπ· π οΏ½ πππ2 οΏ½ οΏ½π’ β π’π οΏ½οΏ½π’ β π’π οΏ½ + πΉπ‘β + ππ πΉπ (π‘) ππ 2 4 where πΆπ· is the drag coefficient that depends on the particle form and the relative Reynolds number, ππ the particle mass, π’π the particle velocity, π’ the fluid phase velocity, π the fluid density, ππ the particle diameter, and finally πΉπ (π‘) is the Brownian force per unit mass which is considered important for submicron particles and πΉπ‘β the thermophoretic force. The thermophoresis effect πΉπ‘β is included in the particle force balance according to [5]. The heat transfer to the particle is determined through an energy balance as shown in Eq. 2. (Eq. 2) ππ πΆπ πππ πππ = βπ πππ2 οΏ½π β ππ οΏ½ + β ππ ππ πΏ 1 where πΆπ is the particle heat capacity (J/kg K), ππ the particle temperature (K), βπ the convective heat transfer coefficient (W/m2 K), and T the temperature of ππ π continuous phase (K). The factor is the rate of ππ evaporation (kg/s) and βπΏ is the latent heat. The rate of vaporization is governed by concentration gradients, with the flux of droplet vapour into the gas phase related to the gradient of the vapour concentration between the droplet surface and the bulk gas. For the discrete phase model, the iron raw particles are specified as having a logarithmic distribution with diameters summarized in Table 1. Also, conditions like position, velocity, size and initial temperature of individual particles as shown in Table 1. Table 1: Properties of iron in solid and gaseous states, and properties of the injected particles in the reactor. Properties of iron particles Density [kg/m3] 7874 . Cp [J/mol K] 25.10 Thermal conductivity 80.4 [W/m.K] Latent heat [kJ/kg] 272 Boiling point [K] 3134 Volatile component 100 fraction [%] 1728K: 1 Pa 2346K: 1kPa Saturation vapour 1890K: 10 Pa 2679K: 10kPa pressure 2091K: 100 Pa 3132K: 100kPa Properties of iron gas 1000K: 22.5 4000K: 29.7 Cp [J/mol.K] 2000K: 23.2 5000K: 34.0 3000K: 26.2 6000K: 39.9 Thermal conductivity 79.45 [W/m.K] Molecular weight 55.845 Injection properties for the discrete phase model Temperature [K] 300 Mass flow rate 0.006 [g/min] Max. Diameter [m] 3x10-4 Min. Diameter [m] 1.1x10-5 B. Experimental procedure GNFs are grown following Pristavita et al. procedure [1], with methane as a carbon source that is decomposed within an argon thermal plasma. GNFs are formed as small nuclei through homogeneous nucleation, and essentially grow laterally in a sheet-like geometry within 2 a very narrow growth window in the plasma decay zone [2][6]. The nanoparticles are deposited on the walls and the end plate downstream of the reactor. It is to be noted the experimental reactor is totally axisymmetric with respect to the flow pattern, including for the exit flow that extends radially from the endplate. This is providing a very good match with the modelled geometry, and an experimental flow pattern having a true stagnation point flow geometry. A small amount of nitrogen is added during the growth, leading to nitrogen content up to 2 at% on the surface of the GNFs. As indicated above, iron is added in the form of iron powders using a PFV100-VM-NO powder feeder from Tekna. A mixture of two iron powders is fed in the reactor. The first has a size distribution between 10 and 300 µm, while the second powder shows a smaller distribution between 1 and 10 µm. It has been observed that the smallest particles were difficult to feed continuously, while the biggest particles were difficult to evaporate. Mixing the two powders results in taking advantage of both aspects, i.e. a continuous feeding for a good rate of evaporation. The iron particles are carried by an argon flow of 5 slpm. The iron powder is added simultaneously with the methane and the nitrogen in order to process the growth of the GNFs, the nitrogen functionalization and the iron implementation in the same time. An equilibrium thermodynamic evaluation using the software Factsage shows that iron vapours are present along with the growing GNFs and atomic nitrogen in the temperature range 3000-5000 K, indicating that it is thermodynamically possible to produce the catalyst in only one step. Scanning Electron Microscopy (SEM) was made using a FEI Inspect F-50 FE-SEM. X-ray Photoelectron Spectroscopy (XPS) were obtained using a VG Scientific ESCALAB MK II operating at a pressure of 10-9 torr, and using an aluminium x-ray source. Results and discussion A. Modeling results The evaporation of iron particles is first analyzed using the present model. A very low amount of iron particles is injected in the ICP torch (0.006 g/min) in the presence of argon plasma at 55.2 kPa and 10 kW effective power. The resulting temperature (K) and mass fraction or iron vapour (kg fe /kg Ar ) profiles can be observed in Figure 1. The left side of the figure corresponds to the temperature profiles, where we can observe that the highest temperature is 10500 K located within the torch zone and which is actually the plasma core. This temperature further decreases once the plasma gas enters the reactor zone due to cooled walls. Relatively smooth temperatures gradients are observed which are characteristic of the present conical reactor design. The isotherm line corresponding to 1811 K is depicted in the figure as an indication of the melting temperature of iron. The mass fraction of iron vapour is observed on the right side of the Figure 1. The range of mass fraction of iron vapour from P-II-7-14 1x10-6 to 5x10-4 clearly depicts the species diffusion in the reactor. The iron vapour created in the torch is further transported by diffusion with the highest amount at the exit of the torch and along the axis region of the reactor. The largest volume of the reactor is showing a mass fraction of 1x10-6, this being sufficient to promote the iron functionalization in the zone of graphene growth (35005000 K). Since the melting point of iron particles is 1811 K, we expect to have the formation of iron particles at the bottom end of the reactor. the analysed sample does not interfere with the determination of the iron bonded to nitrogen; the iron from iron particles being mainly in oxidized form while the binding energy for Fe-N is different than Fe-O. The composition of the overall sample is given in Table 2. Fig. 2. SEM picture of Fe particles fed within the reactor. Fig. 1. Distribution of a) temperature and b) mass fraction of iron vapour inside the conical reactor, with a plasma power of 10 kW and a pressure of 55.2 kPa. The formation of these particles should however be minimum since the amount of iron precursor is very low. Iron particles are not desirable for the process but are not problematic as such, these can also be removed in a subsequent step. B. Structure and composition of N/Fe-GNFs The iron particles to be injected during the growth of N/Fe-GNFs have been collected at the exit of the powder feeder without plasma heating, and observed by SEM (Figure 2). The smallest particles with a diameter less than 10 µm are agglomerated around biggest particles, suggesting large particles are efficiently carrying the smallest one. A SEM image of the N/Fe-GNFs produced is shown in Figure 3. The GNFs look homogeneous, and it can be noted that no amorphous carbon or polymer are present in the sample. It has been observed that some partially evaporated particles remain in the N/Fe-GNFs samples collected. XPS was used to determine the atomic composition of the N/Fe-GNFs surface. The presence of iron particles in P-II-7-14 Fig. 3. SEM picture of N/Fe-GNFs. Table 2: Atomic composition at the surface of the N/FeGNFs. Element At% C 94.5 O 3.6 N 0.4 Fe 1.5 A closer look at the iron peak is made through high resolution analysis and deconvolution using GaussianLorentzian curve fitting to determine the different states of this element (Figure 5). Most of the iron is bonded with oxygen, corresponding to the oxide layer at the surface of the partially evaporated iron particles. This oxidation readily occurs upon removal of the samples from the reactor. A small percentage of the whole measured iron 3 (13.3%) has a lower binding energy (~707.9 eV) which corresponds to Fe-N binding. There is hence some probability that atomic iron has effectively been implemented on the surface of the GNFs through nitrogen coordination, a requirement for catalytically active sites distributed at the atomic level. Fig. 4. XPS survey scan of N/Fe-GNFs. Fig. 5. High resolution and deconvolution of iron peak of the N/Fe-GNFs. 4 Conclusion The modeling results showed that a sufficient amount of iron vapour is available in the graphene nucleation zone within the reactor. Equilibrium thermodynamic evaluations indicate the synthesis of graphene and their functionalization with iron can be made in the same onestep process. Experimentally, a fraction of the iron vapour has been implemented on GNFs through Fe-N bonds. Further work needs to be done to improve the evaporation of the iron particles, and increase the amount of iron able to bond to GNFs. For the non-noble metal catalyst applications, measurements of the activity and stability over the time need to be performed on this N/Fe-GNF structure. Acknowledgements We acknowledge Dr. J. Lefebvre (Ecole Polytechnique de Montreal) for her help on the XPS analysis. We acknowledge the funding contributions from the McGill Engineering Doctoral Award (MEDA), the Fonds de Recherche Nature et Technologie du Quebec (FRNTQ) and from Natural Science and Engineering Research Council (NSERC) of Canada. References [1] R. Pristavita et al., Plasma Chem Plasma Process, 30: 267-279, 2010. [2] R. Pristavita et al., Plasma Chem Plasma Process, 31: 393-403, 2011. [3] D. Binny et al., IEEE Nano, 2012. [4] P. Proulx et al., Plasma Chem. Plasma Proc. Vol. 7, 29-53, 1987. [5] L. Talbot et al., J. Fluid. Mech. Vol. 101 (4), p. 737758, 1980. [6] J.-L. Meunier et al., Plasma Chem Plasma Process 34 : 505-521, 2014. P-II-7-14
© Copyright 2026 Paperzz