Carbon Nanotubes as Catalyst Supports

Carbon nanotubes (CNTs) have extraordinary electrical, thermal, mechanical, and chemical properties (Dresselhaus et al. 2001). With the extremely high specific surface areas that are comparable to carbon black (CB), they may become promising PEM fuel cell catalyst supports (Li et al. 2003; Wang et al. 2004) substantially improving the Pt catalyst activity via metal-support interactions (Arico et al. 2003). Recent studies have shown that Pt nanoparticles supported on multi-walled CNTs (Pt/MWCNT) have much higher catalytic activity for oxygen electroreduction than Pt nanoparticles supported on CB (Pt/CB) (Li et al. 2003).

When nanoparticles are deposited on supports, their physical and chemical properties strongly depend not only on their particle size, but also their surface structure and the characteristics of the metal/substrate interface. Nanoparticle-support interactions can be fully understood by studying the morphology and structures of the supported nanoparticles. Applied quantum chemical methods (Koch et al. 2001), based on solutions to the Schrödinger equation, can provide the quantum-chemical basis for such studies. Additional information about the dynamics of nanoparticle-support interactions, surface structure, dynamics and thermodynamics of supported nanoparticles can be obtained using Molecular Dynamics (MD) methods (Allen et al. 1987). These methods do not explicitly include the presence of electrons; instead, effective force fields (Table 18.1) are used to represent the electronic effects. The physical and thermodynamic properties are obtained using the laws of classical statistical mechanics, which provide rigorous mathematical expressions that correlate macroscopic properties to the distribution and motion of atoms in a N-body system.

To simulate supported Pt nanoparticle morphologies and structures, preliminary studies were conducted for a Pt nanoparticle containing 74 atoms (average size ~2.5 nm) supported on a carbon slab (CB), and on a multi-walled carbon nanotube (MWCNT) using MD methods. The MWCNT consists of three concentric carbon nanotubes having diameters of 2, 4, and 6 nm respectively, and the CB consists of two parallel carbon layers in a volume of 8.3 x 8.3 x 1.4 nm, both corresponding to a total of 9,527 and 4,500 carbon atoms respectively. Simulations are carried out at constant temperature (300K) using a Berendsen thermostat. The relaxation time is 0.4 ps, small enough for the thermodynamic equilibrium to be reached, where the simulated system approaches the canonical ensemble (NVT) at constant number of atoms N, volume V and temperature T, without periodic boundary conditions (Allen et al. 1987). The equations of motion are integrated using the Verlet leapfrog method with a time step of 0.001 ps. The simulation length is 600 ps with 400 ps equilibration steps and 200 ps for the production stage. In order to reduce the computational load, a static substrate was considered in these simulations by fixing the positions of the C atoms.

The many-body Sutton-Chen (Sutton et al. 1990) and the classical 12-6 Lennard-Jones (LJ) potentials (Cramer 2002) were used to describe metal-metal, and metal-substrate and carbon-carbon interactions respectively. Although developed for bulk metals, the SC model has been applied to optimise the structure of transition metal clusters with good agreement with experimental observations (Doye et al. 1998). The SC potential (Eq.18.4) describes reliably static and dynamic properties of transition and noble metals, such as bulk module and elastic constants (Sutton et al. 1990), as well as surface energies, stress tensor components, and surface relaxation of fcc metals (Todd et al. 1993), and has been successfully applied for the description of nanoclusters (Doye et al. 1998; Huang et al. 2002; Huang et al. 2003). The potential energy (Ut) in the SC model is (Sutton et al. 1990) given by Eq.18.4.

With the first term on the right hand side representing the pair-wise repulsive potential and the second term the metallic binding energy with the local electron density given by (Eq.18.5), where rij is the distance between atoms i and j, c is a dimensionless parameter, s is an energy parameter, and a is a length parameter.

The mixture parameters of the SC potential can be obtained from the Lorentz-Berthelot (Allen et al. 1987) mixing rules from those of the pure components, using the geometric and arithmetic means for the energy and length parameters.

Table 18.1. Force field parameters. Pt-Pt SC and C-C LJ parameters are taken from (Sutton et al. 1990) and (Bhetanabotla et al. 1990), respectively.

A

B

Field

s (eV)

a(Â)

a

n

m

c

Pt

Pt

SC

0.019833

3.92

10.0

8.0

34.408

C

C

LJ

0.002413

3.400

Pt

C

LJ

0.040922

2.936

Although accurate descriptions of metal-carbon substrate interactions are not currently available, the classical 12-6 Lennard-Jones (LJ) potential has been successfully used to describe the interactions of metallic clusters including platinum (Liem et al. 1995) and silver (Rafii-Tabar et al. 1997) with a graphite substrate. Hence, in these calculations the use of the LJ potential is well justified for the slab support. Even when the MWCNT support has a curved graphitic surface structure, its size (length and outer diameter) is large compared to the nanoparticle size thus minimizing curvature effects on the Pt nanoparticle morphology. Hence, using the LJ potential to describe metal-substrate interactions may be a good first approximation. Preliminary results at 300°K suggest that the Pt nanoparticle stable geometrical shape is faceted with well defined edges and angles for Pt/CB (Fig. 18.4(a)), while it is more spherical-like for Pt/MWCNT (Fig. 18.4(b)). These differences are attributed to a competition between metalmetal and metal-support interactions in both cases (Fig. 18.4(a-b)).

Fig. 18.4. Molecular Dynamics results at 300K corresponding to a 74-Pt atom nanoparticle supported on (a) carbon slab, and (b) MWCNT supports

Fig. 18.4. Molecular Dynamics results at 300K corresponding to a 74-Pt atom nanoparticle supported on (a) carbon slab, and (b) MWCNT supports

It is interesting to note that these preferred arrangements of Pt atoms form 5-layer and 4-layer stacks on the CB and MWCNT respectively in the direction perpendicular to the supports (later referred as the z direction). The preferred layered arrangements can be quantified by computing the Pt density profiles, p(z) (number of Pt atoms per unit length), for the nanoparticle on different supports as function of the distance to the substrate in the z direction (Fig. 18.5 and Fig. 18.6). These plots help in understanding how Pt atoms distribute inside the nanoparticle permitting the direct visualisation of its atomic structure.

Fig. 18.7 and Fig. 18.8 show the integrated Pt atomic population in the z direction starting from the layer next to the support to the complete nanoparticle (all layers). In addition, these figures also show the detailed layer-by-layer structure corresponding to the Pt layers parallel to the supports, starting from the layer next to it up to the topmost layer (L1), as well as the top view of the complete Pt nanoparticle.

The various Miller planes found on the overall nanoparticle surface suggest the type of crystallographic faces that are more exposed for catalysis applications (Fig. 18.9 and Fig. 18.10). Perfectly flat (111) and (100) fcc Miller planes are observed on the Pt/CB overall surface (Fig. 18.9). However, uneven (due to curvature effects) (111) and "mixed" (100)+(111) fcc Miller planes are found on the Pt/MWCNT surface (Fig. 18.10).

Different surface planes often display different catalytic activities. For instance, Pt (111) is known to be much more active than Pt (100) in the oxidation of CO and methanol, while Pt (100) is found to be more active for oxygen reduction. Platinum nanoparticles show (111) and (100) crystallographic flat or uneven planes on their overall surfaces irrespective of the nature of the carbon support.

"Mixture" (111) and (100) surfaces are found on the Pt nanoparticle supported on the MWCNT. However, it is still not known what catalytic effect this surface would have on the oxygen reduction reaction.

The results presented here aim to show clear differences on nanoparticles morphologies and structures depending upon the characteristics of the support. Nanoparticle surface characteristics, including exposed crystallographic phases, may play a key role on the catalytic activity and can be specially controlled and tailored by selecting a specific support.

0.5

Fig. 18.5. Density profiles for a 74-Pt nanoparticle at 300K as a function of the distance to the substrate, measured in the z-direction, perpendicular to the carbon slab substrate (z = 0 at the center of CB)

Fig. 18.5. Density profiles for a 74-Pt nanoparticle at 300K as a function of the distance to the substrate, measured in the z-direction, perpendicular to the carbon slab substrate (z = 0 at the center of CB)

jfM(nm)

Fig. 18.6. Density profiles for a 74-Pt nanoparticle at 300K as a function of the distance to the substrate, measured in the z-direction, perpendicular to the multi-walled carbon nanotube substrate (z = 0 at the center of MWCNT)

Fig. 18.7. Integrated Pt atomic population in the z direction for a 74-Pt nanoparticle supported on a carbon slab at 300 K. Snapshots show the detailed layer-by-layer in the z-direction (L1= topmost layer), as well as the top view of the complete nanoparticle (All layers)

Fig. 18.7. Integrated Pt atomic population in the z direction for a 74-Pt nanoparticle supported on a carbon slab at 300 K. Snapshots show the detailed layer-by-layer in the z-direction (L1= topmost layer), as well as the top view of the complete nanoparticle (All layers)

Fig. 18.8. Integrated Pt atomic population in the z direction for a 74-Pt nanoparticle supported on a multi-walled carbon nanotube at 300 K. Snapshots show the detailed layer-by-layer in the z-direction (L1= topmost layer), as well as the top view of the complete nanoparticle (All layers)

Fig. 18.8. Integrated Pt atomic population in the z direction for a 74-Pt nanoparticle supported on a multi-walled carbon nanotube at 300 K. Snapshots show the detailed layer-by-layer in the z-direction (L1= topmost layer), as well as the top view of the complete nanoparticle (All layers)

Fig. 18.9. Representative on a carbon slab

Miller planes found on the overall

Fig. 18.9. Representative on a carbon slab

Miller planes found on the overall

-Pt nanoparticle supported

Fig. 18.10. Representative Miller planes found on the overall 74-Pt nanoparticle supported on a multi-walled carbon nanotube

Was this article helpful?

0 0
Brain Blaster

Brain Blaster

Have you ever been envious of people who seem to have no end of clever ideas, who are able to think quickly in any situation, or who seem to have flawless memories? Could it be that they're just born smarter or quicker than the rest of us? Or are there some secrets that they might know that we don't?

Get My Free Ebook


Post a comment