1: \setcounter{page}{6} \pagenumbering{arabic}

2: \addtocontents{toc}{\protect\vspace*{0.5in}}

3: \chapter[\protect\uppercase{Introduction}]{Introduction}\label{ch.ch1}

4: \addtocontents{toc}{\protect\vspace{0.25in}}

5: 

6: \section[Molecular Dynamics Encoded In Spectra]{\underline{Molecular Dynamics Encoded In Spectra}}

7: \addtocontents{toc}{\protect\vspace*{7pt}}

8: 

9: In its earlier years, spectroscopy was an important tool in determining the structure of molecules.  Today, the equilibrium structure, as well as the related spectroscopic constants, continues to be actively pursued.  However, the accumulated theory, techniques and data have also enabled researchers to use spectroscopy as a probe of dynamical processes, such as collisions, energy transfer, and chemical reactions.  

10: 

11: The question is how to decode dynamical information from the experimental or calculated spectra.  Spectra recorded in either the frequency or time domain should reflect the same physical behavior.  These domains are formally connected by a Fourier transform \cite{Fourier}.  Electromagnetic radiation affects molecules on at least four levels in the order of increasing energy: the nuclear spin, rotational, vibrational and electronic {\it Degrees Of Freedom} (DOF).   The interactions among even these four levels within a molecule make it nontrivial to analyze the total dynamics.  For chemists, the vibrational DOF are of special interest since they are closely related to chemical reaction processes \cite{VDOF1}.   The focus of this thesis is the vibrational dynamics of small polyatomic molecules (3-4 atoms).   On the one hand, they are much simpler than most organic or biological molecule, where explicit treatment of all DOF is usually difficult if not impossible.  On the other hand, these small molecules already exhibit a rich range of dynamical behavior, especially among the highly excited states, that is yet to be fully understood \cite{NesbittField}.  

12: 

13: \section[Modes of Vibration]{\underline{Modes of Vibration}}

14: \addtocontents{toc}{\protect\vspace*{7pt}}

15: 

16: We start with the frequency-domain spectra.  Each resolved level in the spectrum corresponds to one quantum eigenstate of the molecular Hamiltonian.  The decoding of dynamics involves assignment of these levels with {\it quantum numbers} corresponding to their modes of vibration.  The word ``mode" has been rather liberally used to designate patterns in vibrations.  The conventional meaning is that there are certain coordinates in which the vibration appears particularly simple.  Especially, it refers to the case when the vibration can be separated into {\it independent periodic oscillations} in these coordinates.  Such a separation is crucial for comprehending molecular vibrations that may be too complex for direct visualization.  Therefore, mode designation is more than mere labeling because it reflects the pattern of underlying dynamics.

17: 

18: A well-known case is the {\it normal modes}.  When the amplitudes of vibration remain small, the force between any two atoms can be approximated as harmonic, i.e. proportional to the displacement from equilibrium configuration.  Diagonalization of the resulting force constant matrix yields $3N-6$ (or $3N-5$ for linear molecules) normal mode coordinates, $N$ being the number of atoms in the molecule \cite{Wilson}.  The classical motion of the atoms (determined by Newtonian mechanics) is decomposable into oscillations along these normal mode coordinates with characteristic frequencies.  In the quantum system, the nodes of the eigenstates are distributed along the normal coordinates (each intersecting the coordinate perpendicularly).  A set of normal mode quantum numbers can be assigned from counting the nodes.

19:  

20: At higher energy, as the vibrational amplitudes increase, inter-atomic forces deviate from the harmonic approximation.  This requires the inclusion of additional terms, such as anharmonic coefficients and resonance couplings, in describing the vibrational dynamics.  The former distorts the normal modes, while the latter not only distort but also mix them.  Eventually, the mixing would become so extensive that the normal mode picture ceases to be a valid representation of the dynamics.  

21: 

22: It is believed that at {\it sufficiently} high energy, the vibrational dynamics enters the ``bag of atoms" regime \cite{NesbittField}, which is characterized by a general lack of periodicity.  In practice, however, this regime may not be easily attained.  In at least some and perhaps most molecules, even when the normal modes break down dynamics could acquire new types of periodicity corresponding to new modes of vibration.  For example, in H$_2$O \cite{LocalWater} and O$_3$ \cite{KellmanOzone}, the stretching states exhibit a transition from normal to {\it local mode} behavior as the energy of excitation is increased.  In Fig.~\ref{fig1.1}, the normal symmetric and antisymmetric modes have two A-B bonds stretch in phase and out of phase by $\pi$, respectively.  In the local modes, only one of the A-B bonds is vibrating.   The more excited states can be assigned in terms of local mode quantum numbers, but not the normal mode ones.    Here the local mode is not just another way to describe the molecule, but a special one that directly reflects the underlying dynamics.  More exotic modes, such as the ``precessional" mode \cite{XiaoPre} and more complicated collective motions \cite{TaylorCHBrClF}, are also known to exist.  

23: 

24: \newpage \begin{figure}[hbtp]  

25: %==\vspace{3.6in}

26: \begin{center} \includegraphics[width=4.13in]{fig01.eps}\end{center}

27: \cpn{Normal and local stretching modes in an ABA molecule} {Normal and local stretching modes in an ABA molecule.  In the normal symmetric (s) and antisymmetric (a) modes, the two A-B bonds vibrate in concert.  In contrast, the local modes 1 ($\ell_1$) and 2 ($\ell_2$) have most vibrational amplitude in only one bond. \label{fig1.1}}

28: \end{figure}   

29: \newpage 

30: 

31: These vibrational modes strongly influence the molecules' chemical behavior.  When a molecule is between collisions and free of other external interactions, its intrinsic reaction rate, if one exists, is determined by the flow of vibrational energy within the molecule, a process termed {\it Intramolecular Vibrational Relaxation} (IVR) \cite{NesbittField,ZewailIVR}.  The complete characterization of IVR requires a ``map" of all possible paths of energy flow.  Yet, in actual modeling of reactions, such a detailed description is often simplified into a statistical treatment.  A most popular assumption is the {\it Rice-Ramsperger-Kassel-Marcus} (RRKM) model, which assumes unrestricted energy flow among all vibrational DOF \cite{RRKMReview}.  This model is assumed increasingly valid at high internal energy and coupling, especially when the corresponding classical dynamics is chaotic \cite{Berne}.  

32: 

33: In contrast, energy in a stable vibrational mode remains trapped in a small fraction of all the possible energy distributions.   IVR and reaction probability would be very different from the RRKM limit when the molecule is in such a state.  Then, one has to be cautious about using a statistical model.   These stable modes may be of great use in coherently controlling chemical reactions \cite{LaserControl}, because excitation energy in these modes may remain long enough for optical manipulation.

34: 

35: Since the famous Fermi-Ulam-Pasta simulation in 1953 \cite{FermiUlamPasta}, molecular systems have served as an application for the mathematical theory of nonlinear dynamics, and a motivation for its continued development.  The past 50 years have seen an increasing interest in applying classical nonlinear mechanics in studying dynamics of microscopic systems.  

36: 

37: \section[Goal and Structure of Thesis]{\underline{Goal and Structure of Thesis}}

38: \addtocontents{toc}{\protect\vspace*{7pt}}

39: 

40: The identification and assignment of vibrational modes are crucial for analyzing the dynamics of highly excited molecules.  For over a decade, the Kellman group has been using the mathematical tools of bifurcation \cite{Kuznetsov} and catastrophe theory \cite{catastrophe} to study classical and quantum behavior in the vibration of small molecules.  The group developed the method of critical points analysis in order to uncover the nonlinear, anharmonic modes at high excitation via analytical detection (as opposed to numerical search).  Its earliest formulation focused on the integrable two-oscillator-single-resonance systems,  according to the critical point(s) in the reduced classical phase space \cite{Xiao-Kellman}.  This was later extended to coupled 3-oscillator systems, which are nonintegrable \cite{Zi-MinH2O1}.  These two methods are reviewed in references \cite{KellmanReview,KellmanAnnRev}.  This thesis is a further generalization that takes into account Hamiltonians with multiple polyad numbers.  The result is a DOF-independent method of generalized critical points analysis based on a more rigorous foundation.  The vibrational dynamics, especially the pure bending subsystem of acetylene (C$_2$H$_2$) is studied using the generalized critical points analysis.

41: 

42: Chapter 2 reviews the relevant background information: the effective quantum Hamiltonian and polyads, basic tools in classical mechanics, and topics on the quantum-classical correspondence.

43: 

44: Chapter 3 first describes existing procedures of critical points analysis for treating integrable and nonintegrable systems with one polyad number.  Then general considerations are carried out with regard to the behavior of critical points of an arbitrary effective Hamiltonian with polyad number(s).

45: 

46: Chapter 4 applies this analysis to the C$_2$H$_2$ pure bending system.  The result reveals the existence of new families of critical points, born in bifurcations at increasing energy.  The result is compared to those from the separate consideration of single resonances, giving a qualitative description of the nature of the new critical points.  The stable families of critical points correspond to stable quantum modes of vibration.   

47: 

48: Chapter 5 extends the analysis to the stretch-bend acetylene system including all the resonances.  The fate of the C-H normal stretch mode overtones is considered under increasing excitation. The preliminary result suggests that a chain of bifurcations first create a local C-H stretch mode, which then bifurcates into more complex stretch-bend modes.

49: 

50: \newpage Chapter 6 summarizes contributions made in this thesis, and discusses possible future directions of research.

51: 

52: \vspace*{0.25in} 

53: 

54: \begin{picture}(370,5) \thicklines \put(0,0){\line(1,0){370}} \end{picture}

55: 

56: \begin{center}{GLOSSARY}\end{center}\vspace{-6pt}

57: 

58: \begin{picture}(370,5) \thicklines \put(0,0){\line(1,0){370}} \end{picture}

59: 

60: \begin{align}

61: &\mbox{DOF} & \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Degrees Of Freedom} \nonumber\\ 

62: &\mbox{IVR} & \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Intramolecular Vibrational Redistribution} \nonumber\\ 

63: &\mbox{PES} &  \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Potential Energy Surface} \nonumber\\ 

64: &\mbox{PO} &  \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Periodic Orbit(s)} \nonumber\\ 

65: &\mbox{PPS} &  \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Polyad Phase Sphere(s)} \nonumber\\ 

66: &\mbox{RRKM} &  \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Rice-Ramsperger-Kassel-Marcus model} \nonumber\\ 

67: &\mbox{SOS} & \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Surface(s) Of Section} \nonumber\\ 

68: &\mbox{ZOS} & \cdots \cdots \cdots \cdots \cdots &\mbox{\,\,\,\,\,\,\,\,Zero-Order State(s)} \nonumber

69: \end{align}

70: 

71: \begin{picture}(370,5) \thicklines \put(0,0){\line(1,0){370}} \end{picture}