RESEARCH ON THE MOLECULAR STRUCTURES AND
NANOSTRUCTURES OF MATERIALS AND THEIR APPLICATIONS
OVERVIEW
The focus of my research is on the provenance and inter-relationships of the thermodynamic properties of engineering materials
with molecular-level mechanisms. My current interests are in computational nanotechnology with emphasis on nanostructured
materials (e.g., starburst dendrimers, carbon nanotubes, aerogels, and nanoporous media). Dendrimers (a tree-like branched polymer)
have found applications in molecular recognition/chemical detection, health, and memory storage areas.. We investigate the use of
dendrimers as sensors. We apply Monte Carlo simulation and Ornstein-Zernike integral equations to obtain quantitative information
on the sensing capacity. In addition, carbon nanotubes can be used as heat conductors as well as surface modifiers for drag reduction.
There are 157 and more structures of zeolites listed by IZA. A few can be exploited for selective gas separation. We are
thus interested in gas adsorption in nanopores.
My research has both a fundamental and a practical side. Since the 1970’s, I have engaged in projects from the National Science
Foundation, the Department of Energy, the Gas Research Institute, the Office of Naval Research, and a number of industrial sources
with combined funding in excess of $3 million. I did experimental work (vapor-liquid equilibrium measurements for Master's thesis),
industrial research and production (natural gas processing, solvent extraction, polymer extrusion and spinning for films and fibers,
sulfonation of detergents, etc.), and theoretical investigations (molecular simulation and liquid-state theories). I have interest in electrolyte
solutions, and supercritical fluids. I shall describe these at some detail in the following.
|
EXAMPLES OF NANOSCALE RESEARCH |
|
|
|
|
|
| Figure 1. Dendrimers as nanosensors. | Figure 2. Water flows past a single carbon nanotube* (CNT). | Figure 3. Heat Conduction through a "Model" CNT Composites. |
TOPICS OF CURRENT INTEREST (Click on the navigation buttons):
Nanosensors--Dendrimers Drag Reduction on Hydrophobic Surfaces Nanocomposites Nanopores
Absorption Refrigeration Acid Gas Treating Supercritical Fluid Solubility
Starburst dendrimers are polymer molecules with Cayley treelike branches (dendron = Greek for tree),
arising from the polyfunctionality of the moieties at the branching points. The size (diameter) of a
dendrimder molecule, depending on the "generation", can range from few nanometers to over ten nanometers,
thus a truly nanoscale material. Figure 1 is a model of a 4th generation polyamidoamine (PAMAM) dendrimer.
Dendrimers have found applications in molecular recognition, as catalysts and catalyst supports, drug delivery
and gene therapy carriers, surface modifiers (tribology, and information storage) and electronic devices and antennae.
Our research focuses on the ability of dendrimers in gas and chemical detection based on their design for molecular
recognition.
Dendrimers can dissolve in solvents for processing and application. We study dendrimer solutions from the point of
view of the liquid-state molecular theory. The PAMAM dendrimer is dissolved in a binary solvent
of an analyte species (to be detected) and a placebo species (neutral to detection).
The major thrusts of this research are: (1) to establish a
solvent-averaged interaction potential Ueff (Fig. 7) between the dendrimers and the solvent molecules,
according to the soft-matter colloids theory; and
(2) to use the simulation and the integral equation techniques to determine if there is "molecular recognition".
We propose, as a general molecular-level indicator for all sensing devices, the Excess Number
Nex of the fluctuation theory, as the figure of merit in molecular reconnoitering.
The analyte molecules (with specific affinity) can be differentiated (detected) by the dendrimers with surface coating of
receptor end groups (exo-receptors), functioning as sensors, from the placebo molecules in the solution.
We use a solvent-explicit off-lattice Monte Carlo simulation (Fig. 4)
and the integral equation theories to study the behavior of the
polyamidoamine dendrimers
in solution. We first establish the colloidal-limit interaction potentials between the dendrimer and the
solvent gas. The analyte
(detected gas) is given stronger interactions with the corona of the dendrimer and this is reflected in the effective
potential Ueff (Fig. 7).
The purpose is to find out if there is "molecular recognition" by the dendrimer’s corona for the analyte gas
(i.e., sensor detection) versus the rest, the neutral placebo gas.
This should be reflected in the excess numbers Nex =
r int dr h(r).
We obtain qualitative and quantitative indication of adsorption of the analyte molecules (i.e., sensor effectiveness)
by the sensitized dendrimers (Fig. 6).
The excess numbers Nex furnishes concrete evidence that there is molecular recognition by the
exo-receptors on the corona for the analyte molecules (showing peaks at distances 10-12
s, where the exo-receptors are located, Fig. 6).
During the course of the research, we have elucidated the dense-core or dense-shell nature of the dendrimers
(Fig. 5).
We also succeed in obtaining accurate structures of the gas-gas and gas-dendrimer pairs by using a self-consistent
theoretical solution of the Ornstein-Zernike integral equations (Fig. 8).
Considering the large size differences between the
gas and dendrimer (7~10 time diameter ratio) as well as the long-range interaction of the dendrimer
Gaussian potential, the theoretical approach is remarkably accurate.
Publication:
Rationale:
Application of nanostructures in sensing technology. Use of functionalized dendrimers for chemical detection.
|
|
|
|
Figure 4. A Monte Carlo simulation box with a 4th generation dendrimer (PAMAM) and two gas species A (purple molecules) and B (green molecules). The end groups are enlarged indicating chemical functionalization (to increase affinity d, i.e., selective detection of the analyte gas B). About 4000 gas molecules are used. |
Figure 5. The dendrimer end group density rEG as a function of the strength of affinity d. d augments the Berthelot combining rule for the interaction energy e between the gas (B) and the dendrimer (D). For small values of d =0, the end groups are closer to the core: i.e. the dense-core behavior (peaks at small r-distances: r measures the core-to-periphery separation). The larger the values d are, the more the end groups are "stretched" towards longer distances. This causes the dendrimer to "swell"-- approaching the dense-shell structure. There is a gradual transition from the dense-core to the dense-shell structure, depending on the strength of the solvent. "Good" solvents relax the dendrimers (thus approaching the dense shell). |
|
|
|
|
|
Figure 6. Excess adsorption number Nex in statistical mechanics is employed to quantify the "loading" or "detection capacity" of the dendrimer surfaces (the coronas) in adsorbing the analyte gas B molecules. As the affinity d (the strength of attraction of the corona towards B) is increased (from, 1 2, to 3), the Nex values (the peaks) are increased (values > 1) near r ~ Rg (~10 to 12 s ) i.e., the radius of gyration of the dendrimer in solution. |
Figure 7. Effective "colloidal" potentials Ueff as obtained from Monte Carlo simulation for the small gas molecules (A & B) surrounding the large dendrimer molecule (D) (about 10 times the diameter s of the gas molecule). Since the B gas molecules are made to attract more to the corona (surface moieties) of the dendrimer (i.e., affinity d, in this case d=2), there is a "deeper valley" for Ueff of B-dendrimer interaction (green squares, the "analyte" gas) than that of the A-dendrimer interaction (the brown squares for the "placebo" gas). The broad depths of the Ueff are at longer distances close to the radius of gyration of the dendrimer. The top curve (orange color) is the singlet density rEG distribution of the end groups of a single dendrimer as the gas B is "tugging" them at the surface. There is clearly inflation of the size of the dendrimer in strong solvents. |
Figure 8. The radial distribution functions g(r) as obtained from Monte Carlo simulation (symbols) and from the theory: the Ornstein-Zernike integral equations (lines). Close agreement is obtained upon using a self-consistent closure (closure that satisfies thermodynamic consistencies as well as the pointwise zero-separation theorem consistencies). |
(Work supported by the National Science Foundation**)
There is increasing evidence, experimental and from molecular simulation, that on hydrophobic (non-wetting) surfaces
fluid flow experiences a non-zero slip velocity (instead of stick) at the boundary. This has important implications for
nano- and microfluidics, as well as for macroscopic flows over ships or submarines. For the past century, the “stick”
boundary condition has been assumed to be appropriate for most macroscopic flows. However, it has received new critical
attention recently as engineering applications are reaching down to the micrometer and nanometer scales, and as
diagnostic techniques enhance our ability to probe the chemistry and physics of the fluid/solid interface at the molecular level.
In particular, alterations in interfacial interactions from attractive to repulsive, as reflected in the dewetting behavior
of the solid, can affect the ability of the fluid to exchange momentum with the surface at the molecular scale, resulting in
a velocity slip at the solid wall.
A number of simulation works using nonequilibrium molecular dyanmics (NEMD)
have shown that the wetting (or non-wetting) properties of the surface can
affect the stick-to-slip flow boundary condition.
Barrat et al. simulated Couette flow and Poiseuille flow on surfaces
with contact angle reaching 140o and found a slip length of 30 molecular diameters (about 10 nm).
In our research we modify
the surfaces with implanted micro/nanoscale polymer chains and/or carbon nanotubes (Figure 9) to
produce a hydrophobic surface. Both
experimental and simulation work are carried out. The NEMD is set up for the Couette flow (Figure 10).
In addition,
phalanges of carbon nanotubes are positioned with water flowing across the bank (Figure 11).
The contact angle, indicative of the wetting of the interface by water on carbon nanotubes,
is estimated to be around 103o to 129o, depending on the diameter of the nanotube.
New superhydrophobic polymer coatings have been discovered that give contact angles approaching 170o.
Rationale: The chemistry of surface influences the boundary layer flow characteristics depending
on the wetting properties of the surface. A velocity slip can be realized on superhydrophobic
surfaces (contact angle > 150o) for reduction of the drag coefficient.
|
|
|
|
|
Figure 9. Banks of carbon nanotubes on substrate (SEM: Choi 2001). |
Figure 10. Nonequilibrium molecular dyanmics (NEMD) simulation* for Couette flow of water molecules confined between two flat plates. The upper plate moves with a velocity U. |
Figure 11. Water molecules flowing past array of carbon nanotubes.* |
(Work supported by the Office of Naval Research)
Carbon nanotubes (CNT) are excellent conductors of heat: at about 3000 W/(m.K) for multiwall carbon nanotubes (MWNT)
and about 6000 W/(m.K) for single wall ones (SWNT).
On the other hand, cooling in nanoscale computer chips is critical for
their functioning correctly. For example, the computer chips industries are entering 90 to 65 nm manufacturing
in order to make it possible
to put more transistors on a single chip or shrink the size of existing chips,
effectively increasing performance. But the developments
have encountered problems, not the least of which is overheating. One thought is for making heat sinks with composites with CNT fillers,
capitalizing on CNT’s high thermal conductance. An unexpected problem, which was actually known since 1941, is the Kapitza resistance
that exists at the interfaces between different solids and between solid and liquid in contact.
Since heat is transferred by phonons in dielectrics,
mismatch of phonon frequencies (acoustic mismatch) causes this additional resistance that could be large even at room temperatures.
Figure 12 shows the interfacial temperature drops for different solid walls in serial contact.
Keblinski (2004) investigated the CNT-in-oil
heat transfer via nonequilibrium molecular dynamics, NEMD (Figure 13).
In the simulation by Barrat et al. (2003) the Kapitza length (Fig.ure 14),
indicative of thermal resistance, can reach 50 molecular diameters (about ~17 nm).
We carry out NEMD to characterize the heat transfer in CNT composites.
Knowledge of the relationships of heat conductance
with phonon density of states, interfacial mismatch, CNT size, arrangement, and
volume fraction can aid in optimum design.
Rationale: Heat dissipation in nanodevices requires high thermal conductance. Carbon nanotubes due to their high thermal conductivity are dispersed in composites to facilitate heat transfer.
Publication:
|
|
|
|
|
Figure 12. Temperature drops across the interfaces of solid slabs in contact due to the Kapitza thermal boundary resistance. X-axis is the direction of heat flow. |
Figure 13. Liquid (octane) surrounding a carbon nanotube in a NEMD simulation of Kapitza resistance. (Keblinski 2004). |
Figure 14. Definition of the Kapitza length, LK, at the interface of Solid 1 and Solid 2. dT is the temperature drop at the interface (due to Kapitza resistance). The dashes are hypothetically extended length (from Solid 1) of the temperature slope, according to Fourier's law: q = - k dT/dz. |
(Work supported by the National Science Foundation** and the Oklahoma Nanonet)
Aerogels and zeolites contain nanometer-sized pores and are used as gas sensors, heat-insulating materials, adsorbents,
catalysts and catalyst supports. Figure 15 shows the ball-and-chain structure of aerogels that is generated from
the gellation-condensation manufacturing process. It is the result of diffusion-limited cluster aggregation (DLCA).
The neutron scattering structure shows unique signatures.
Aerogels are translucent materials with >95% void spaces--highly porous.
Under TEM, it shows cobweb-like features with pore size in the 20 nm range.
Aerogels allow gas permeation. They will deform under nitrogen pressure.
How the gas distributes amongst the ball-and-chain structures of the aerogels,
and what effects they have on their properties are of interest in design and operation of these nanostructures.
The topics of phase behavior of mixtures in pores have received much of attention recently.
The simplest gas that can exhibit liquid-liquid separation is the nonadditive hard sphere mixture.
This is a model gas, however, it embodies the role of repulsive forces and
their contributions to the phase behavior.
We employ Monte Carlo molecular simulation and the Replica Ornstein-Zernike theory to model gas
molecules infusing into a quenched hard sphere matrix (Fig. 16).
This matrix is derived from equlibrated (annealed) hard spheres, which can be altered
to other types of matrices, either from diffusion-limited cluster aggregation (DLCA for aerogels),
or from regular lattices for zeolite channels.
The theory is coupled with a self-consistent closure (ZSEP closure, Figure 20).
This new theory outperforms the
conventional closures (such as the Percus-Yevick or Martynov-Sakisov formulation). It can give accurate
pair correlation functions (Figure 19), especially at contact.
The chemical potentials determined from the theory check well
with grand-canonical Monte Carlo results and a cavity-biased version of same (Figure 17).
The liquid-liquid phase diagram is determined for a nonadditivity
D =0.2. The lower consolution point is then evaluated at
r* =0.33, the bulk value (free of matrix restrictions)
is 0.415 for the same mixture. Thus the confinement inside the pores tends to lower the consolution point
of the liquid mixture.
Works for aerogel matrix and zeolite matrix are under way.
Publications:
Rationale: Molecular structure and thermodynamic properties of gases and liquids enclathrated in nanopores
are important in nanostructures (such as aerogels, carbon nanotubes) and nanoscale devices (such
as in microfluidics and nanofluidics).
|
|
|
|
|
Figure 15. Top: The ball-and-chain structure of aerogels (diffusion limited cluster aggreation, DLCA). Bottom: Gas moelcules invade the interior of the aerogels. |
Figure 16. Gas mixture (nonadditive hard spheres: red and blue spheres) inside an quenched equilibrated hard sphere matrix (grey spheres) (very different from DLCA), the simplest mixture that exhibits phase separation. This is an idealized model of inclusion gas in porous matrices. |
Figure 17. Chemical potentials determined via grand-canonical Monte Carlo simulation (plus cavity-biased method of Mezei) and integral equations (replica Ornstein-Zernike: ZSEP) for the model of Figure 16. Non-additivity = 0.2. |
|
|
|
|
|
Figure 18. Liquid-liquid phase envelope for nonadditive hard sphere mixtures in pores. The lower crtitical consolution density for this case r = 0.33 (cf. the bulk value r = 0.415 for the same mixture with no confinement). |
Figure 19. The pair correlation functions for gas-gas g22 (left panel), and gas-matrix g20 (right panel) pairs as obtained from simulation and from three different theories (Percus-Yevick PY, Martynov-Sakisov MS, and ZSEP self-consistent closures). The ZSEP closure is highly accurate compared to the alternatives. |
Figure 20. Explanation of the ZSEP closure. The bridge function B(r) is expressed as a function of the "renormalized" indirect correlation (g*(r) = h(r)-C(r)+ buref(r)). The parameters a, f, and z are adjustable parameters that are "modulated" to satisfy the thermodynamic conditions plus the pointwise zero-separation theorems and contact value theorems). |
(Work supported by the National Science Foundation**)
As we enter the 21st century, there are urgent needs to find replacements for the CFC refrigerants in order
to mitigate the ozone depletion and global warming effects. This is mandated by the Montreal Protocol,
and signed on by many countries.
The phasing-out schedule for CFCs and HCFCs is imminent. The new refrigerants should be environmentally
friendly. This research is carried out to screen and characterize promising working fluids as replacements.
The first candidates tested here are electrolyte solutions:
the likes of aqueous-based lithium bromide. Several engine cycles (single-effect and double-effect)
are designed to implement the new refrigerants (Figure 22).
The colligative and thermal properties of the salt solutions are determined. The coefficient of
performance (COP) is evaluated and their quantitative dependence on molecular parameters (such as molecular
weight, ion (hydration) size, valence, dielectric coefficients, etc.)
is ascertained. This is made possible by a modern
electrolyte theory, i.e., the mean-spherical approach from statistical mechanics.
We have developed a Windows-based software “Abscycle” to calculate the refrigeration cycle performance
(e.g., the coefficient of performance, COP) for different input conditions by using an electrolyte theory MSAEXP
as the basis for calculating the vapor pressure, enthalpy, and equilibrium properties of the LiBr solution.
An example GUI (Graphical Users Interface) is shown in Figure 23. We consider the aqueous salt refrigerants
promising, achieving a COP higher than 0.7 for single effect and over 1.0 for double effect engine cycles. In late
1980s, the LiBr chillers were actually manufactured in Japan (Hitachi) and in USA (Trane etc.), and are now on
the commercial markets. Figure 21 is a picture of the Trane Absorption Chiller using aqueous lithium bromide as the working fluid.
Publications:
Rationale: To find alternative refrigerants as replacements for the CFC's (Chlorfluorocarbons)
|
|
|
|
Figure 21. Trane Absorption Chiller (height ~ 10ft). |
Figure 22. Absorption Chiller Flow Sheet |
|
|
|
Figure 23. The GUI Interface in the "AbsCycle" software for a Single-effect Lithium Bromide-Water Chiller |
(Work supported by the Gas Research Institute)
Natural gas as produced contains CO2, H2S, etc.
in addition to the hydrocarbons. These gases should be removed before utilization.
This process is called "gas sweetening".
We are interested in the thermodynamic properties of the solutions so that design work
could be carried out for the loading requirements and heat duties.
There are a number amine-treating packages in
existence. We use a statistical mechanical framework for
multisolvent (non-aqueous) electrolyte solutions that satisfies the
Gibbs-Duhem relation. It has been generalized to water, amine, salt, and gas mixtures.
The treating liquids are amine solutions. The process involves three simultaneous equilibria:
(i) vapor-liquid equilibria, (ii) ion dissociation equilibria, and (iii) chemical reaction equilibria.
We use a group contributions (GC) framework with molecular modeling for electrolytes (MSA).
We succeed in predicting key equilibrium and enthalpic properties for solutions of single and blended amines.
Our framework permits accurate predictions for these complex solutions with a minimal number of adjustable molecular
parameters. The method is valid over wide treating conditions. It is applied to MDEA, DEA, MEA, and other solvents.
Furthermore, it can estimate hydrocarbon solubility, effects of phosphoric acid addition, COS, CS2, and trace sulfur compounds
(mercaptans). It can predict VLE, heat of absorption, speciation, heat capacity, bubble point, activity coefficients, and P-V-T
condition. It is useful for obtaining acid gas solubilities, for operation near the pinch point, estimating corrosion, hydrocarbon
emissions, solvent/feed ratio, enthalpy requirements, and gas quality. We present a complete thermodynamic
package for acid gas treating (Figure 27). A typical process is shown in Figure 24.
We have also developed a software “AGAS” to carry out calculations of
the vapor-liquid equilibria in terms of partial
vapor pressures (kPA) of acid gas versus loading of amines. An interface is shown in Figure 29.
The MSA and group contributions based method (ElecGC) is capable of giving details of speciation in
the loaded amine solutions. A typical result is shown in Figure 28, where the DEA speciation in
CO2 absorption
(with species DEA, DEAH+, carbamate, as well as H+, OH-,
HCO-, CO3= ions) is depicted at 40oC.
Publication:
Rationale: To remove the acid gases: carbon dioxide and hydrogen sulfide from natural gas streams by treating with aqueous amines. To provide a complete thermodynamic package for the design calculations of the vapor-liquid equilibria and heat requirements in acid gas treating.
|
|
|
Figure 24. Amine Treating Flow Sheet. |
|
|
|
|
|
Figure 25. Natural Gas Sweetening Plant. |
Figure 26. Bubble: the Vapor-Liquid Interface and Chemical Reactions on a Sieved Tray. (The detailed mechanism of absorption of the CO2 and H2S gas). |
Figure 27. The Computing Package "ElecGC"--Electrolyte Group Contribution Theory Applied to Gas Treating. It funishes robust and accurate calculations for the vapor pressure-loading, heat of absorption, hydrocarbon solubility and speciation values. |
|
|
|
|
|
Figure 28. Speciation During Treating CO2 with DEA (Diethanolamine). The formation of the carbamate ions is in tandem with the bicarbonate ions. |
Figure 29. The GUI (garphics users interface) for the AGAS software developed for calculating the vapor-liquid thermodynamics in acid gas treating. |
Figure 30. The output from the software AGAS, giving the loading and vapor pressures, in this case for MDEA amine solution absorbing CO2. |
(Work supported by the Gas Processors Association and the Gas Research Institute**)
Supercritical solvents have contributed to the environmentally benign manufacturing because of their “cleanliness” (solvents such as CO2 or water), their high solvent power, and operating at lower temperatures than is usually done. Thus these fluids have been used in nicotine removal in tobacco, extraction of pharmaceutical products, coffee decaffeination, and the clean reactions and laundry-cleaning industries. The molecular structures of supercritical solvents around a solute molecule is described using the liquid-state integral equation theories. We found that the augmentation in the long-ranged behavior of the pair correlation function, guv(r), is at the root of the "clustering" phenomenon that gives rise to the solubility enhancement. We further derived the influence through the Kirkwood-Buff factors on the chemical potentials and partial molar volumes. The molecular mechanisms of supercritical solubility enhancement are elucidated.
Publications:
(Work supported by the Department of Energy and Oak Ridge National Laboratory)
