Perla B. Balbuena’s Research

My research aims to understand and predict thermodynamic, transport, and kinetic properties of materials, using state-of-the-art first principles computational chemistry and physics methods.  We focus on bulk and nanomaterials used as catalysts and electrolytes in power sources devices, such as lithium-ion batteries and fuel cells. We also investigate the growth of metal nanoparticles inside macromolecules, the catalyzed growth of single-wall carbon nanotubes, and the design of new materials for H2 and CH4 storage.  Our most recent work includes the investigation of reaction mechanisms of chain polymerization reactions in benign solvents, and the analysis of biomaterials with specific functions such as ion carriers and oxygen reduction catalysts.

 

1. Catalysis on metal nanoparticles

Metal nanoparticles supported on a carbon structure and embedded in a hydrated Nafion membrane provide the catalytic sites for the oxygen reduction reaction in low-temperature fuel cells.  The importance of understanding this reaction stems from its slow kinetics, and from the high cost of the Pt catalysts currently used.   Our research focuses on several aspects: 1) understanding of the oxygen reduction reaction mechanism on Pt clusters and surfaces using density functional theory, and ab initio molecular dynamics; 2) determining adsorption of intermediate species such as O, OH, H2O and H2O2 on clusters and surfaces made of Pt and Pt-alloys; 3) DFT analyses of the free energies of the elementary steps and elaboration of a set of thermodynamics guidelines for the design of alternative bimetallic catalysts; 4) proton transport at the catalyst-membrane interface.

 

Snapshots from ab initio MD simulations of the first reduction step on Pt(111) for initial distances dOH = 2.5 Å and dOS = 3 Å, at a high degree of proton solvation: (H3O)+(H2O)3 (H2O)6. First row: Initial configuration (left), at 0.7 ps (center), and at 0.8 ps (right).  Second row: At 1.1, 1.2, 1.6, and 2 ps.  Chemisorption takes place followed by O2 decomposition, whereas OOH formation was not observed during 5 ps. (P. B. Balbuena, Y. Wang, E. J. Lamas, S. R. Calvo, L. A. Agapito, J. M. Seminario, “Reactivity of bimetallic nanoclusters toward the oxygen reduction in acid medium.”  In “Device and Materials Modelling for the PEM Fuel Cell”, S. Paddison , Editor, Springer, in press.

 

2. Catalyzed growth of single-wall carbon nanotubes

We have developed reactive force fields to investigate aspects of the chemical vapor deposition process where a model precursor gas (CO) is catalyzed on the surface of metal nanoparticles yielding carbon atoms that combine forming various structures and eventually grow into a SWCNT. Classical molecular dynamics techniques were coordinated with experiments on model catalysts; our main findings have been reported in four journal articles.  

The modeling work consisted on the development and implementation of classical molecular dynamics (MD) simulations to follow the time evolution of the nanotube growth process over a catalyst particle using reactive force fields  for the description of C-C, C-metal, and metal-metal interactions.  The force field parameters for the metal-C interactions were obtained from density functional theory calculations in small clusters.   Simulations were carried out assuming that the reaction takes place instantaneously and irreversibly on the catalyst surface.  Two cases were analyzed: a catalyst particle floating on a gas phase of the precursor, and a catalyst particle deposited on a substrate.   In the first case the main growth steps were identified as: 1) dissolution of C inside the catalytic particle until saturation, 2) precipitation of C atoms on the surface initially forming short chains, 3) the short chains connect with each other forming quasi-sp3 structures (fullerene-type), 4) a cap separates from the surface and initiates the nanotube growth.  The simulations on a supported catalyst basically follow those stages, as shown in Figure 1.  However, the initial stages of growth are very sensitive to the strength of the metal-substrate interactions.  Specifically, strong substrate/cluster interactions favor a more rigid cluster structure which modifies the initial stage of C dissolution.  Even though C still dissolves into the cluster, the amount is lower than that obtained when the metal cluster is completely unrestrained.  This work suggested that the rate determining step of the nanotubes growth process is the nucleation step, leading to the formation of a stable cap over the surface of the catalyst.  For a catalyst floating in the vapor of a precursor gas, the cap structure and its attachment to the cluster are critical to the chirality of the nanotubes formed.  However, in processes where the cluster is supported on a substrate, the chemical nature of the substrate and the attachment of the metal cluster to the substrate are also crucial, because the interfacial contact metal cluster/substrate is one of the factors that influences the cluster morphology and atomic mobility, which in turn determines the type of carbon structure formed on the surface. 

Our most recent work has been oriented to determine the reactivity properties of carbon nanotubes of a given chirality, as a function of the growth process, i.e., of the number of atoms in the nanotube. Specifically we have analyzed two arm-chair and one chiral cap-ended single-wall carbon nanotubes of various finite lengths to determine their reactivity towards the simplest growth reaction, the adsorption/reaction of a C2 radical with the open end of the growing SWCNT.

 

3. Physical properties of metal nanoparticles

Metal nanoclusters exhibit unusual chemical and physical properties different from those of the bulk material or of the atoms, and have a number of fascinating potential applications in heterogeneous catalysis, micro- and nanoelectronics and opto-electronics devices. Therefore, the understanding of their thermal, structural and dynamical properties is a topic of intense interest from the scientific and technological viewpoint. One of the important factors of the nanosize regime is the presence of a large percentage of surface atoms.  In addition, when nanoclusters are deposited on surfaces, their physical and chemical properties are strongly dependent not only on their particle size and chemical composition, but also on the structure of the surface and that of the metal/substrate interface.

Text Box: • Nanosize transition metal clusters melt at much lower temperature than the bulk metal; the melting temperature depends on cluster size, shape, and chemical composition. • Wetting behavior (island formation on the substrate) can be induced by thermal treatments or by substrate chemical and physical modifications. • Bimetallic systems that exhibit surface segregation behavior melt in two stages: surface diffusion followed by total melting.

An improved understanding of heterogeneous catalysis is emerging thanks both to significant advances in surface science techniques as well as to the insights provided by the application of first-principles theoretical methods.   It is crucial to be able to elucidate and predict the structure and composition of materials used as catalysts, and it is now well recognized that chemical, thermal, and mechanical treatments may significantly affect the structure of the exposed faces, and therefore their catalytic activity.  A detailed understanding of the melting process of metal nanoclusters is one important factor to understand their special behavior. It is known that the solid-liquid transition in nanoclusters differs from that in bulk materials, and many studies have addressed the melting point change variation with the nanocluster size.  At low temperatures, the cluster structure is solid-like, and as the temperature increases the structure acquires liquid features, passing through an intermediate state, called dynamic coexistence, where the structure fluctuates between liquid and solid behavior. Recent experiments and molecular simulations have concluded that the nanoclusters melting temperature depends on their size, shape, composition, and in most cases is lower than that of bulk melting. As the cluster size is decreased, the melting temperature generally decreases. The phenomenon can be understood considering that a large percentage of atoms residing on the surface of cluster are weakly bonded and less constrained in their thermal motion.  However, quantum effects in clusters below a certain size can be important yielding a non-monotonic behavior for the variation of the melting temperature vs. cluster size at small sizes.

Molecular dynamics techniques provide useful insights that contribute to the understanding of complex microscopic phenomena such as mixing and growth in bimetallic systems, and the statistical thermodynamics of small systems on which such simulations are based has been discussed extensively.   We employ MD simulations to study the structure and dynamics of several mono and bimetallic systems, including Cu-Ni and Pt-Au nanoclusters with various compositions and cluster sizes, both in vacuum and deposited on a graphite surface. The many-body Sutton-Chen potential is used for the metal-metal interactions, and Lennard-Jones potentials describe the interaction between nanocluster and substrate.  Deformation parameters are defined to investigate shape changes in the nanoclusters, and changes in the structural and dynamical properties are discussed as a function of temperature. The diffusivity of the nanocluster as a whole in the solid and liquid phases is analyzed. 

 

4. Ion complexation inside macromolecules

Dendrimers are envisioned as promising materials in many areas of science and technology with applications spanning design of drug delivery systems, light harvesting, electronic devices, and catalysis.1-5  Poly(amido-amine) (PAMAM) dendrimers have proven successful template agents for several ions, metal atoms, and clusters, including Pt, Pd, Au, Ag, Cu.6-11  As ion complexation precedes the formation of embedded metallic nanoclusters, a clear understanding of ion complexation and that of the influence of the factors that affect the complexation process – such as pH, counterion, ligand exchange, and charge transfer- are needed.  Due to their own peculiarities each ionic complexation requires individual study; this fact hampers the understanding of currently known systems and the search for new precursor-dendrimer systems for metal nanoparticle synthesis.

We use classical MD and quantum DFT to analyze ion complexation inside dendrimers. 

a) Full-solvent MD final configuration A2 : Cu(II) is coordinated with two amide O, one tertiary N in EDA-core and one water O in equatorial position. Branch folding places one amide O in axial position (the other is occupied by one water O). For clarity, water molecules other than those in the first coordination sphere of the ion are not shown. b) 38-atom-fragment-Cu(II)-(H2O)3 closed DFT (optimized structure) : Cu(II) coordinates in a similar fashion to (a) except that the branch folding effect is obviously not reproduced; another water O substitutes the amide O of the dendrimer branch. Here the equatorial water and one of the axial waters form hydrogen bonds R’-CH2-(OH)---HOH---(OH)-CH2-R’ to both hydroxyl O atoms with lengths 2.12 Å and 1.99 Å respectively. These H-bonds may explain its enhanced stability compared to the “open” configuration. c) Full-solvent MD final configuration A3 : Cu(II) is coordinated to one amide O, one hydroxyl O and four water O. d) 38-atom-fragment-Cu(II)-(H2O)4 (optimized structure) : Coordination to one amide O, one hydroxyl O and two water O in equatorial plane and  one water in axial position. One of the equatorial waters forms H-bond –dashed lines - as R’- C=O---HOH---(OH)-CH2-R at 1.81 Å (with carbonyl O) and 1.81 Å (with hydroxyl O). The fifth water is in quasi-axial position (bond distance larger than typical Cu-O axial bond distance) and forms H-bonds with one amide O at 1.78 Å and with the equatorial water at 1.60 Å.

 


DFT optimized configurations. a) 38-atom-fragment -Cu(II): Cooperative effect of two amide O to the branching N is evident.  b) 38-atom-fragment -Cu(II)-(H2O)6: Coordination to one amide O and five water O (3 equatorial and 2 axial). Water interacts with dendrimer atoms forming H-bonds: b1. R-CH2-(OH)---H, length: 1.83 Å. b2. The sixth water molecule sits in the second coordination shell and forms two hydrogen bonds –dashed lines- R’-C=O---HOH---(OH)-CH2-R in the adjacent branch at 1.81 Å (amide O) and 1.94 Å (hydroxyl O).  c) 38-atom-fragment -Cu(II)-3H2O open configuration: Coordination to one branching N, two amide O and one water O  in equatorial plane and two water O in axial positions.

Snapshots obtained after 600ps full-solvent MD trajectories.  For clarity, water molecules beyond the ionic first coordination shell are not shown.  a) Coordination of one Cu(II) to an amide O , an hydroxyl O and four water O and another Cu(II) to one amide O and five water O in the opposite part of the dendrimer that remains open.  Two branches and their environment can at least hold one Cu(II). b) Coordination of both Cu(II) to two amide O, one hydroxyl O and three water O atoms. Again the flat configuration enables the attachment of two branches to each ion. The green atoms are Cl- counterions.

This work highlights the importance of water in Cu(II) stabilization during ionic complexation to dendrimers and suggests that amide oxygen atoms and possibly in a lesser extent hydroxyl O atoms are likely to be the preferred coordination sites in EDA-core PAMAM -OH terminated dendrimers.

In low generation dendrimers, coordination to amide oxygen and water oxygen seems to be dominant whereas coordination to the tertiary amine nitrogen in EDA-core is expected to be less frequently found.  Time evolution of the distance between the ion and the tertiary amine N atoms indicates that their residence times are very short, suggesting a weaker coordination of Cu(II) to that site than that to the amide O and water O atoms.

The results from DFT calculations in dendrimer fragments, where branch folding effect is low, can provide insights on how the Cu(II) ion coordinates in the outer pockets of large generation dendrimers. These calculations show cooperativity between the branching tertiary amine nitrogen, amide oxygen atoms and water. However as branch mobility and folding are expected to be higher in large generation dendrimers, the latter statement needs to be taken with caution.

MD simulations incorporating the full solvent effect reveal that both “closed” configurations where the dendrimer branches contribute to a “cage” effect, and “open” structures may be found.  In gas phase, such closed configurations are found energetically favorable (by ~ 6 kcal/mol) to the open ones.  Moreover, closed configurations are also found in structures showing two-ion attachments to G0-OH, where a strong participation of the dendrimer branches in complexation is observed.

Analyses of the various complexation structures of Cu(II) in various degrees of hydration indicates that combined effects of shorter and stronger bonds, H-bonding enhanced stability, and chelate formation can explain why Cu(II) hexahydrates tend to lose some of their coordination waters to bind to the dendrimer sites.

Solvent screening effects can be accounted for obtaining a Cu(II)/dendrimer load ratio of at least 2:1 which is expected to be maximum as the ionic separation is approximately equal to the dendrimer size with branches flatly extended.  We did not observe ion pairing between Cu++ and Cl- but the results cannot be extended to larger generation dendrimers where the available area for counterion contact is larger.

Finally, this study expects to set a motivation for fundamental experimental research related to low generation dendrimers as potential generators of useful insights that can help explaining experimentally observed phenomena in dendrimers of larger generations; such valuable insights can be enhanced upon integration of theory and experiment.

 

5. Polymerization reactions in benign solvents

Organic peroxides are widely used as initiators in free radical polymerizations and studies on the decomposition of organic peroxides are of importance in a variety of processes.  Decompositions of diacyl peroxide and peroxydicarbonate polymerization initiators into alkyl or alkoxy radicals carried out in traditional organic solvents have been extensively studied by experimentalists.  However, current technologies intend to substitute traditional organic solvents with supercritical CO2, and it is of interest to determine how this solvent affects the decomposition reactions. The structure of peroxydicarbonates (Scheme 1) is similar to those of diacyl peroxides with general structure of RC(O)OO(O)CR, where the R groups are alkyl for diacyl peroxides and alkoxy for peroxydicarbonates.


Scheme 1. Molecular structure of DEPDC (diethyl peroxydicarbonate), TFAP and AP (diacyl peroxides) initiators. R = C2H5O for DEPDC; R = CF3 for TFAP; and R = CH3 for AP.

 

Peroxydicarbonates are traditionally included in the diacyl peroxide category. As shown in Scheme 2, three decomposition mechanisms were proposed by previous studies of diacyl peroxides and peroxydicarbonates: (1) a single O-O bond cleavage produces a carboxyl radical pair and a subsequent decomposition of the carboxyl radicals; (2) a stepwise pathway including two steps: an initial two-bond cleavage to a carboxyl radical, an alkyl (or alkoxy) radical and CO2, followed by decomposition of the carboxyl radical; (3) a concerted three-bond cleavage to an alkyl (or alkoxy) radical pair and two CO2 molecules.

 

Pathway 1

 

 

Pathway 2

 

Pathway 3

 

 

Scheme 2. Previously proposed thermal decomposition mechanisms for peroxydicarbonates and diacyl peroxide initiators.

 

We used Møller-Plesset perturbation theory and density functional theory calculations to study decomposition mechanisms of polymerization initiators, such as diethyl peroxydicarbonate, trifluoroacetyl peroxide, and acetyl peroxide, which possess a general structure of RC(O)OO(O)CR.  We found that the decomposition of initiators with electron donating R groups follows two favorable stepwise pathways: a two-bond cleavage mechanism in which the O-O single bond and one of R-C bonds of [R-C(O)O-O(O)C-R]  break simultaneously followed by decomposition of the R-C(O)O• radical, and a one-bond cleavage mechanism in which the single O-O bond cleavage produces a carboxyl radical pair and a subsequent decomposition of the carboxyl radicals. It is also found that the initiators with electron withdrawing R groups follow the two-bond cleavage pathway only. Geometrical and energetic analyses indicate that despite the similar structures of the peroxydicarbonates, quite different decomposition energy barriers are determined by the nature of the R groups.  

 

6. Materials for gas storage

We are investigating storage of H2 and CH4 in C-containing self-assembled structures. 

 

7. Biomimetic materials

The goal of this work is to identify biomaterials which could be emulated specifically for ion carrier and oxygen reduction functions.