1: %\documentclass[byrevtex,preprint,prb,aps,nobibnotes]{revtex4}%endfloats
2: \documentclass[byrevtex,prb,aps,nobibnotes,twocolumn]{revtex4}%endfloats
3: \usepackage{graphicx}%
4: \usepackage{dcolumn}
5: \usepackage{amsmath}
6: \usepackage{bm}
7: \usepackage{dcolumn}
8: \usepackage{longtable}
9: \voffset=1.5cm
10: %\hoffset=0.8cm
11: \begin{document}
12:
13: \title{\centering\Large\bf Reorganization asymmetry of electron transfer in
14: ferroelectric media and principles of artificial photosynthesis }
15: \author{Dmitry V.\ Matyushov}
16: \email[E-mail:]{dmitrym@asu.edu.}
17: \affiliation{
18: Department of Chemistry and Biochemistry and the Center for the
19: Early Events in Photosynthesis, Arizona State University, PO Box
20: 871604, Tempe, AZ 85287-1604}
21: \date{\today}
22: \begin{abstract}
23: This study considers electronic transitions within donor-acceptor
24: complexes dissolved in media with macroscopic polarization. The
25: change of the polarizability of the donor-acceptor complex in the
26: course of electronic transition couples to the reaction field of the
27: polar environment and the electric field created by the macroscopic
28: polarization. An analytical theory developed to describe this
29: situation predicts a significant asymmetry of the reorganization
30: energy between charge separation and charge recombination
31: transitions. This result is proved by Monte Carlo simulations of a
32: model polarizable diatomic dissolved in a ferroelectric fluid of
33: soft dipolar spheres. The ratio of the reorganization energies for
34: the forward and backward reactions up to a factor of 25 is
35: obtained in the simulations. This result, as well as the effect of
36: the macroscopic electric field, is discussed in application to the
37: design of efficient photosynthetic devices.
38: \end{abstract}
39: \maketitle
40:
41: \section{Introduction}
42: \label{sec:1}
43: Electron transfer (ET) reaction is a basic elementary step in
44: photoinduced charge separation occurring in both natural and
45: artificial photosynthesis. The fundamental mechanism behind the
46: conversion of light energy into the energy of a charge-separated state
47: is illustrated in Figure \ref{fig:1}. Optical excitation $h\nu$ of the
48: donor unit, D$\to$D$^*$, within the donor-acceptor complex, D--A, lifts
49: the energy level of the donor to create conditions for the
50: photoinduced charge separation step, D$^*$--A $\to$ D$^+$--A$^-$.
51: Charge separation is generally an activated transition. The
52: activation barrier, according to the Marcus-Hush
53: theory,\cite{Marcus:93} can be obtained from the equilibrium free
54: energy gap between the final and initial ET states, $\Delta F_0$, and the
55: free energy $\lambda$ required to reorganize the nuclear degrees of freedom
56: (reorganization free energy). When the equilibrium free energy of the
57: acceptor is below the free energy of the donor by the amount of free
58: energy $\lambda$, the charge separation transition is activationless. Such
59: an activationless transition is often fast and efficient, but it
60: requires loosing the energy $\lambda$ from the photon energy $h\nu$.
61: Increasing the energetic efficiency of photosynthesis therefore demands
62: $\lambda$ to be as low as possible. It is often assumed that an important
63: role played by the hydrophobic protein matrix in facilitating ET in
64: natural photosynthesis is to screen highly polar aqueous environment
65: and to reduce $\lambda$.
66:
67:
68:
69: Once the charge-separated state has been created, it needs to be
70: sufficiently long-lived to be used for energy storage. A mechanism
71: often suggested to be of primary importance in natural systems is the
72: use of a sequence of activationless ET steps to move the electron away
73: from the primary donor. Each activationless step $i$ lowers the system
74: energy by $\lambda_i$ thus resulting in the overall low energetic efficiency
75: of the photosynthetic unit. The reaction competing with photoinduced
76: charge separation is the normally highly exothermic return electron
77: transfer to the ground D--A state. The classical reaction path then
78: goes through the relatively high activation barrier in the inverted ET
79: region predicted by the Marcus-Hush theory (point ``C'' in Figure
80: \ref{fig:1}).\cite{Marcus:93} However, the system avoids this
81: activated path by nearly activationless transfer to a vibrationally
82: excited state, D--A(v), of the electronically ground donor-acceptor
83: complex (dashed line in Figure \ref{fig:1}). The vibrational energy
84: is subsequently released to heat through vibrational relaxation ($v\to
85: 0$ in Figure \ref{fig:1}).
86:
87: The design an optimization of a sequence of activationless ET
88: reactions requires a very precise molecular tuning, which is hard to
89: achieve in synthetic systems despite some significant progress
90: achieved in this field in recent years.\cite{Wasielewski:92,Moore:93}
91: It is therefore desirable to search for mechanisms of reducing the
92: rate of return ET in molecular photosynthesis with the goal of
93: achieving higher quantum yield for the charge-separated state.
94: According to the current understanding of radiationless transitions in
95: molecules,\cite{BixonJortner:99} an efficient way to reduce the return
96: rate would be to move the reaction into the normal region of ET.
97:
98: The electronic states D$^*$--A and D--A are distinct and, in
99: principle, the charge-recombination transition D$^+$--A$^- \to$ D--A can
100: be characterized by a pair of parabolas with the curvatures different
101: from those of the charge-separation transition D$^*$--A $\to$
102: D$^+$--A$^-$. In the reactions classification adopted here we label
103: transition D$^*$--A $\to$ D$^+$--A$^-$ as charge separation and
104: transition D$^+$--A$^-$ $\to$ D--A as charge recombination. Backward
105: transitions at each step are not considered as separate steps in the
106: reaction mechanism. When the photoexcitation energy is kept constant,
107: shifting the recombination reaction into the normal region would
108: require increasing the curvature of the charge-recombination parabolas
109: (dash-dotted lines in Figure \ref{fig:1}), i.e.\ lowering the
110: reorganization energy. The inverted-region activated state C then
111: shifts to the normal-region activated state C$'$ in Figure
112: \ref{fig:1}.
113:
114: It is easy to realize the pitfall of the picture shown in Figure
115: \ref{fig:1}. The separation of the minima of two dash-dotted
116: parabolas must be equal, in the Marcus-Hush theory, to twice the
117: reorganization energy, which is clearly violated once the curvature is
118: increased without corresponding shift of the parabolas. Therefore,
119: the goal of bringing the recombination reaction to the normal region
120: can be realized only within models extending beyond the Marcus-Hush
121: picture of equal-curvature parabolas. On needs flexibility, built into
122: the model, that would allow decoupling of the Stokes shift from the
123: curvatures. The use of parabolic free energy surfaces with different
124: curvatures, as was suggested by Kakitani and Mataga,\cite{Kakitani:85}
125: is prohibited by the requirement of energy conservation. When the
126: energy gap between the donor and acceptor energy levels is taken for
127: the reaction coordinate, the free energy surfaces of the initial,
128: $F_1(X)$, and final, $F_2(X)$, ET states are connected by the linear
129: relation established by Warshel\cite{Hwang:87} and
130: Tachiya\cite{Tachiya:89}
131: %
132: \begin{equation}
133: \label{eq:1-1}
134: F_2(X) = F_1(X) + X
135: \end{equation}
136: The parabolic surfaces with different curvatures clearly violate this
137: requirement and, therefore, cannot be used for the modeling of ET
138: reactions. The problem can be resolved within a three-parameter model
139: of ET free energy surfaces\cite{DMjcp:00} which allows different
140: reorganization energies and, at the same time, does not violate eq
141: \ref{eq:1-1}. The free energy surfaces $F_i(X)$ are then necessarily
142: non-parabolic.
143:
144:
145: \begin{figure}[htbp]
146: \centering \includegraphics*[width=6cm]{fig1}
147: \caption{Energetics of photoinduced electron transfer involved in
148: natural and artificial photosynthesis. The charge-separated state
149: D$^+$--A$^-$ is placed below the photo-excited state D$^*$--A by
150: the reorganization energy $\lambda$ to ensure activationless transition.
151: The dashed line indicates the vibrationally excited electronic
152: ground state for which the recombination transition
153: D$^+$--A$^-\to$D--A does not require an activation barrier. The
154: dash-dotted lines mark the free energy surfaces of states
155: D$^+$--A$^-$ and D--A with the reorganization energy four times
156: lower than the value of $\lambda$ used to draw the free energy surfaces
157: shown by the solid lines. C and C$'$ indicate the classical
158: transition states. }
159: \label{fig:1}
160: \end{figure}
161:
162: Once different reorganization energies are allowed within a theory of
163: free energy surfaces, one needs to address the following major
164: question: What would be a mechanism resulting in dramatic changes of
165: reorganization energies between charge separation and charge
166: recombination reactions? A question relevant to studies of natural
167: photosynthesis is the following: Is the role of the protein matrix can
168: be reduced to providing a low dielectric constant or there are some
169: other properties of proteins beneficial for high efficiency of natural
170: photosynthesis? Note that shifting the transition point from C to C$'$
171: requires a very significant change in the reorganization energy (four
172: times decrease in Figure \ref{fig:1}). The possibility of different
173: equilibrium-point curvatures of the two ET free energy surfaces has
174: been actively pursued in the last decades by computer simulations
175: searching for the effects of non-linear solvation on
176: ET.\cite{Hwang:87,Kuharski:88,Marchi:93,Yelle:97,Hartnig:01,DMjcp2:04}
177: However, essentially in all reports, the difference in reorganization
178: energies between the initial and final ET states has never come close
179: to the magnitude that would dramatically change the mechanism of
180: charge recombination.
181:
182: In this paper, we address the problem of reorganization asymmetry by
183: considering ET reactions in polarizable donor-acceptor complexes
184: immersed in polarization anisotropic media. A part of the motivation
185: for this approach comes from studies of bacterial photosynthesis. The
186: protein matrix in natural bacterial reaction centers is highly
187: anisotropic with the electric field amounting $\simeq 10^7$
188: V/cm.\cite{Steffen:94} On the other hand, the primary pair is highly
189: polarizable with the polarizability change upon photoexcitation about
190: 800--1100 \AA$^3$.\cite{Middendorf:93} Since bacteriochlorophyll
191: cofactors are much less polarizable,\cite{Kjellberg:03} primary charge
192: separation occurs with a large negative polarizability change. On a
193: more fundamental level, polarizability change in the course of ET
194: leads to very significant reorganization asymmetry, as follows from
195: both analytical theories\cite{DMjpca:99} and computer
196: experiment.\cite{DMjacs:03,DMjpca:04} The asymmetry arising from the
197: coupling of changing polarizability to the solvent recation field is
198: by far the largest achieved so far in computer simulations thus
199: promising the desired alteration of the energetics of photosynthetic
200: reactions.
201:
202: In Section \ref{sec:2}, we present an analytical model of ET free
203: energy surfaces when both the polarizability and the dipole moment of
204: the donor-acceptor complex change in the course of ET. The
205: donor-acceptor complex is immersed in a polar medium which, in
206: addition to usual polar response, is characterized by a macroscopic
207: electric field. In order to mimic high electric fields present in
208: molecular systems (thin films, proteins, etc.), we use a polar model
209: fluid capable of producing ferroelectric order (Section \ref{sec:3}).
210: The ordered phase is obtained by Monte Carlo (MC) simulations. The
211: possibility of ferroelectric order in bulk polar fluids is still
212: actively debated in the literature,\cite{Wei:92,Zhang:95,Morozov:03}
213: and it has been suggested that boundary conditions employed in the
214: simulation protocol significantly affect the possibility of creation
215: of the macroscopically polar phase.\cite{Wei:93} For our current
216: purpose, this subject is not relevant since we are mostly interested
217: in reactions in condensed phases with macroscopic electric fields
218: strong enough to influence properties of electronic transitions.
219: Whether the macroscopic polar order is stabilized by surface effects
220: or some other reasons is not central to our study. On the other hand,
221: ferroelectrics and ordered polar films may be used as solvents for ET
222: reactions. The present analysis then directly applies to such systems.
223: We return to discussing the role of reorganization assymetry in
224: improving the efficiency of phtosynthesis in Section \ref{sec:4}.
225:
226:
227: \section{Energetics of electron transfer}
228: \label{sec:2}
229: \subsection{Model}
230: \label{sec:2-1}
231: Consider a donor-acceptor complex immersed in a medium with a nonzero
232: macroscopic electric field $\mathbf{F}$. The complex has the dipole moments
233: $\mathbf{m}_{0i}$ and polarizabilities $\bm{\alpha}_{0i}$ in two electronic
234: states, $i=1,2$. These states are coupled to nuclear fluctuations in
235: the solvent and are ``dressed'' with the field of the electronic
236: solvent polarization following adiabatically the changes in the charge
237: distribution within the complex. Because of the electronic
238: polarization, the dipole moments and polarizabilities are different
239: from their gas phase values $\mathbf{m}_{0i}$ and
240: $\bm{\alpha}_{0i}$:\cite{DMjpca:99}
241: %
242: \begin{equation}
243: \label{eq:2-1}
244: \mathbf{m}_i = \left[\mathbf{1}-2\bm{\alpha}_{0i}\cdot\mathbf{a}_e \right]^{-1}\cdot \mathbf{m}_{0i}
245: \end{equation}
246: and
247: %
248: \begin{equation}
249: \label{eq:2-2}
250: \bm{\alpha}_i = \left[\mathbf{1}-2\bm{\alpha}_{0i}\cdot\mathbf{a}_e \right]^{-1}\cdot \bm{\alpha}_{0i}
251: \end{equation}
252: Here, $\mathbf{a}_e$ is the linear response function such that the
253: free energy of solvation of dipole $\mathbf{m}_0$ by the electronic
254: solvent polarization (subscript ``e'') is
255: $\mathbf{m}_0\cdot\mathbf{a}_e\cdot\mathbf{m}_0$. Once the electronic degrees
256: of freedom have been adiabatically eliminated, the electronic energy
257: levels of the charge-transfer complex are affected by the microscopic
258: nuclear field $\mathbf{R}_n$ (nuclear reaction field, subscript ``n'') and the
259: macroscopic field $\mathbf{F}$ of the solvent:\cite{DMjpca:99}
260: %
261: \begin{equation}
262: \label{eq:2-3}
263: E_i[\mathbf{R}_n] = I_i^{\text{np}} - \mathbf{m}_{i}\cdot(\mathbf{R}_n+\mathbf{F}) -
264: \frac{1}{2}(\mathbf{R}_n+\mathbf{F})\cdot\bm{\alpha}_{i}\cdot (\mathbf{R}_n +\mathbf{F})
265: \end{equation}
266: Here $I_i^{\text{np}}$ are the energies of the ET complex which
267: include the gas-phase energies and free energies of solvation by the
268: solvent electronic polarization expressed as a sum of induction and
269: dispersion solvation components.
270:
271: The system Hamiltonian
272: %
273: \begin{equation}
274: \label{eq:2-4}
275: H_i[\mathbf{R}_n] = E_i[\mathbf{R}_n] + H_B[\mathbf{R}_n]
276: \end{equation}
277: is the sum of energies $E_i[\mathbf{R}_n]$ and the solvent bath
278: Hamiltonian $H_B$. The Gaussian (linear response) approximation is
279: adopted for the latter
280: %
281: \begin{equation}
282: \label{eq:2-5}
283: H_B[\mathbf{R}_n] = \frac{1}{4} \mathbf{R}_n\cdot\mathbf{a}_n^{-1} \cdot \mathbf{R}_n
284: \end{equation}
285: where $\mathbf{a}_n$ is the linear response function of the solvent
286: nuclear polarization. In the following, for simplicity, we will assume
287: collinear dipole moments $\mathbf{m}_i$ and the polarizability
288: changing its value only along the direction of the dipole moment. We
289: will also assume that the solute polarizability increases for the $1\to
290: 2$ transition, i.e., $\Delta\alpha=\alpha_2-\alpha_1>0$. These approximations allow us to
291: consider $\mathbf{a}_n$ as a scalar with the only non-zero projection
292: along the solute dipole. We will also define the scalar $F_m$ as the
293: projection of the external field $\mathbf{F}$ on the direction of the
294: solute dipole moment. The consideration of a more general case does
295: not present fundamental difficulties.\cite{DMjpca:99}
296:
297:
298: \subsection{Free energy surfaces}
299: \label{sec:2-2}
300: The classical Hamiltonian $H_i[\mathbf{R}_n]$ in eq \ref{eq:2-4} can
301: be used to build the free energy surfaces of ET along the reaction
302: coordinate associated with the fluctuating donor-acceptor energy gap
303: required to reorganize the nuclear degrees of freedom
304: %
305: \begin{equation}
306: \label{eq:2-6}
307: X =\Delta H[\mathbf{R}_n] = H_2[\mathbf{R}_n] - H_1[\mathbf{R}_n]
308: \end{equation}
309: The free energy surfaces for the initial ($i=1$) and final ($i=2$) ET states are
310: obtained by constrained integration over the nuclear reaction field
311: %
312: \begin{equation}
313: \label{eq:2-7}
314: e^{-\beta F_i(X)} = A \int \delta(X- \Delta H[\mathbf{R}_n]) e^{-\beta H_i[\mathbf{R}_n]} d\mathbf{R}_n ,
315: \end{equation}
316: where $A$ is used to account for the units of the field $\mathbf{R}_n$ and
317: $\beta=1/ k_{\text{B}}T$.
318:
319: \begin{figure}[htbp]
320: \centering
321: \includegraphics*[width=7cm]{fig2}
322: \caption{Free energy surfaces of ET $F_i(X)$ for the initial ($i=1$,
323: ``1'') and final ($i=2$, ``2'') ET states. Shown in the plot are
324: the free energy minima $X_{0i}$, the free energy
325: gap $\Delta F_0=F_{02}- F_{01}$, and the upper boundary $X_0$ for the
326: reaction coordinate $X$. }
327: \label{fig:2}
328: \end{figure}
329:
330:
331: The Gaussian integral in eq \ref{eq:2-7} can be taken exactly with the result
332: %
333: \begin{equation}
334: \label{eq:2-8}
335: F_i(X) = - \kappa_i X - \beta^{-1}\ln\left[\frac{\sinh \chi(X)}{\chi(X)}\right]
336: \end{equation}
337: where
338: %
339: \begin{equation}
340: \label{eq:2-9}
341: \chi(X) = 2\beta\sqrt{\kappa_i^3\lambda_i(X_{0} - X)}
342: \end{equation}
343: The parameters in eqs \ref{eq:2-8} and \ref{eq:2-9} are related to the
344: properties of the polarizable donor-acceptor complex and the polar solvent by the following
345: set of equations. The parameters $\kappa_i$ are
346: %
347: \begin{equation}
348: \label{eq:2-10}
349: \kappa_i = \left( 2a_n f_{ni} \Delta \alpha \right)^{-1}
350: \end{equation}
351: The parameter $f_{ni}$ in eq \ref{eq:2-10} is responsible for the
352: enhancement of the effective solute dipole by the interaction of the
353: solute polarizability with the nuclear reaction field
354: (cf.\ to eqs \ref{eq:2-1} and \ref{eq:2-2})
355: %
356: \begin{equation}
357: \label{eq:2-11}
358: f_{ni} = (1 - 2a_n \alpha_i )^{-1}
359: \end{equation}
360: Further,
361: %
362: \begin{equation}
363: \label{eq:2-12}
364: X_0 = I_2^{\text{np}} - I_1^{\text{np}} - \Delta mF_m - \frac{1}{2}\Delta\alpha F_m^2 + \frac{(\Delta m + \Delta \alpha F_m)^2}{2\Delta\alpha}
365: \end{equation}
366: is the boundary of the fluctuation band of the reaction coordinate $X$
367: (Figure \ref{fig:2}). The requirement $X< X_0$ implicit in the
368: definition of $\chi(X)$ is eq \ref{eq:2-9} does not allow the energy gap
369: fluctuations to exceed $X_0$. Finally, $\lambda_i$ in eq \ref{eq:2-9} is the
370: solvent reorganization energy
371: %
372: \begin{equation}
373: \label{eq:2-13}
374: \lambda_i = a_n f_{ni} \left[\Delta m + \Delta\alpha(R_i + F)\right]^2
375: \end{equation}
376: where $\Delta m = m_2 - m_1$ and reaction field $R_i$ is
377: %
378: \begin{equation}
379: \label{eq:2-14}
380: R_i = 2a_nf_{ni} m_i
381: \end{equation}
382:
383:
384: The function $\chi(X)$ in eq \ref{eq:2-8} does not carry index specifying
385: the ET state because of the relation connecting the parameters of the
386: two ET states:
387: %
388: \begin{equation}
389: \label{eq:2-13-1}
390: \kappa_1^3 \lambda_1 = \kappa_2^3 \lambda_2
391: \end{equation}
392: In addition, the parameters $\kappa_1$ and $\kappa_2$ are related by the
393: additivity constant
394: %
395: \begin{equation}
396: \label{eq:2-15}
397: \kappa_1 - \kappa_2 = 1
398: \end{equation}
399:
400: The free energy surfaces defined in the range $X<X_0$ are
401: non-parabolic, as is depicted in Figures \ref{fig:2} and \ref{fig:3}.
402: Each of them passes through the minimum at $X=X_{0i}$ defined by the
403: condition
404: %
405: \begin{equation}
406: \label{eq:2-16}
407: d F_i(X)/ d X\bigg|_{X=X_{0i}} = 0
408: \end{equation}
409: This equation can be reduced to the algebraic equation for $\chi(X)$
410: which always has a non-zero solution
411: %
412: \begin{equation}
413: \label{eq:2-17}
414: \chi\coth \chi - 1 = (2\beta\kappa_i^2\lambda_i)^{-1} \chi^2
415: \end{equation}
416: The solution is simplified when $\chi_{0i}=\chi(X_{0i})\gg 1$. One then gets
417: %
418: \begin{equation}
419: \label{eq:2-18}
420: X_{0i} = X_0 - \kappa_i \lambda_i
421: \end{equation}
422: and the free energy at the minimum is
423: %
424: \begin{equation}
425: \label{eq:2-19}
426: F_{0i} = - \kappa_i X_0 - \kappa_i^2 \lambda_i
427: \end{equation}
428: In this limit, the second derivatives taken at the position of the
429: minima define the two solvent reorganization energies, as is the case
430: in standard formulations of ET theories
431: %
432: \begin{equation}
433: \label{eq:2-20}
434: \frac{d^2 F_i(X)}{ dX^2}\Bigg|_{X=X_{0i}} = \frac{1}{2\lambda_i}
435: \end{equation}
436: Note that from eq \ref{eq:2-19} the fluctuation boundary $X_0$ is related to
437: the free energy gap $\Delta F_0 = F_{02}- F_{01}$ and two reorganization energies by
438: the relation
439: %
440: \begin{equation}
441: \label{eq:2-20-1}
442: X_0 = \Delta F_0 +\kappa_2^2\lambda_2 - \kappa_1^2 \lambda_1
443: \end{equation}
444:
445:
446: Taken together, eqs \ref{eq:2-8}--\ref{eq:2-20-1} provide an exact
447: model for the free energy surfaces of ET (Figure \ref{fig:2}) based on
448: three thermodynamic parameters, $\lambda_1$, $\lambda_2$, and $\Delta F_0$. The present
449: model thus extends the two-parameter Marcus-Hush theory, based on the
450: assumption $\lambda_1=\lambda_2$,\cite{Marcus:93} to three-parameters space. In
451: view of the connection between $\kappa_1$ and $\kappa_2$ (eq \ref{eq:2-15}), one
452: of them can be considered as the non-parabolicity parameter:
453: %
454: \begin{equation}
455: \label{eq:2-21}
456: \kappa=\kappa_1 =\left(1 - \sqrt[3]{\frac{\lambda_1}{\lambda_2}} \right)^{-1}
457: \end{equation}
458: In terms of the solute polarizability and the solvent nuclear response
459: function, the non-parabolicity parameter becomes
460: %
461: \begin{equation}
462: \label{eq:2-22}
463: \kappa = \frac{1 - 2 a_n \alpha_1}{2a_n\Delta\alpha}
464: \end{equation}
465: Because of the choice $\Delta\alpha>0$ adopted here, $\lambda_1 < \lambda_2$ and $\kappa_i>0$.
466:
467:
468: \begin{figure}[htbp]
469: \centering
470: \includegraphics*[width=7cm]{fig3}
471: \caption{Free energy surfaces $F_i(X)$ at $\lambda_1=1$ eV and $\lambda_2=2$ eV.
472: The free energy gap $\Delta F_0$ is: $-2$ eV and $-7$ eV in (a) and 1
473: eV and 4 eV in (b). The free energy surfaces corresponding to $\Delta
474: F_0=-7$ eV in (a) do not have the classical crossing point $X=0$
475: indicated by the dashed lines in both panels. }
476: \label{fig:3}
477: \end{figure}
478:
479:
480: It is easy to prove that $F_i(X)$ from eq \ref{eq:2-8} obey the energy
481: conservation requirement given by eq \ref{eq:1-1}. Note that the
482: present model exhausts the space of thermodynamic parameters available
483: for the description of ET reactions. Only three thermodynamic
484: parameters, free energy gap between the minima and second cumulants of
485: the fluctuations around the minima, are allowed by the two-state
486: nature of the problem. Any extension of the theory for ET free energy
487: surfaces beyond the present level will require non-equilibrium
488: parameters to be involved.
489:
490: The requirement of thermodynamic stability of the polarization
491: fluctuations limits the range of possible values of the reaction
492: coordinate, $X \leq X_0$. The reaction coordinate boundary $X_0$ is shown
493: by the dash-dotted line in Figure \ref{fig:2}. There is no
494: singularity of $F_i(X)$ at $X=X_0$ and the free energies $F_i(X)$ are
495: infinite at $X>X_0$. For reaction coordinates sufficiently far from
496: $X_0$, the free energy surfaces can be conveniently re-written in the
497: form
498: %
499: \begin{equation}
500: \label{eq:2-23}
501: F_i(X) = F_{0i} + \kappa_i \left[\sqrt{X_{0i} + \kappa_i \lambda_i - X} - \sqrt{\kappa_i\lambda_i} \right]^2
502: \end{equation}
503: When the reorganization energies $\lambda_i$ are close to each other, the
504: non-parabolicity parameter $\kappa$ in eq \ref{eq:2-21} tends to infinity
505: and one obtains the standard Marcus-Hush parabolas by expanding eq
506: \ref{eq:2-23} in $1/ \kappa$. From eq \ref{eq:2-23}, the activation
507: energy of ET is
508: %
509: \begin{equation}
510: \label{eq:2-24}
511: \Delta F_i = F_i(0) - F_{0i}=\kappa_i \left[\sqrt{X_{0i} + \kappa_i \lambda_i} - \sqrt{\kappa_i\lambda_i} \right]^2
512: \end{equation}
513:
514:
515: \subsection{Qualitative results}
516: \label{sec:2-3}
517: The three-parameter model predicts some novel results regarding the
518: dependence of the reaction rates on the free energy gap (energy gap
519: law). In order to illustrate them, we will consider the transition
520: from the activated normal region to the activated inverted region
521: through the activationless transition for the forward reaction $1\to 2$
522: and the backward reaction $2\to1$ while maintaining $\lambda_1 < \lambda_2$ and the
523: definition of the free energy gap as $\Delta F_0 = F_{02}-F_{01}$ (Figure
524: \ref{fig:2}). Our analysis here assumes $\chi_{0i}\gg1$, which holds for
525: most cases of interest. For the forward reaction $1\to 2$, the transition from
526: the normal to the inverted region is marked by the equation $X_{01}=0$,
527: which corresponds to a line in the space of parameters $\lambda_2/ \lambda_1 > 1$
528: and $\Delta F_0 / \lambda_1$ (Figure \ref{fig:4}):
529: \begin{equation}
530: \label{eq:2-25}
531: \Delta F_0 / \lambda_1 = \kappa(\kappa+1) - (\kappa -1)^2(\lambda_2/ \lambda_1)
532: \end{equation}
533: The line shrinks into the point $\Delta F_0/ \lambda_1 = \pm 1$ in the Marcus-Hush
534: limit of equal reorganization energies (marked as ``MH'' in Figure
535: \ref{fig:4}).
536:
537:
538: Lowering the free energy gap in the inverted region for the forward
539: reaction 1$\to 2$ leads to a new region of ET absent in the Marcus-Hush
540: theory. When the boundary of the band of allowed energy gaps $X_0$
541: crosses the transition state $X=0$, the classical crossing point of
542: two free energy surfaces falls outside the range of allowed reaction
543: coordinates. No crossing of free energy surfaces $F_i(X)$ is then
544: possible ($\Delta F_0 = -7$ eV in Figure \ref{fig:3}(a)), and the classical
545: reaction channel is closed, $\Delta F_1 = \infty$. The reaction occurs only
546: through vibrational excitations of the final ET state 2 effectively
547: lowering the free energy gap. The transition to this new region, which
548: may be called ``quantum tunneling region'' ($\Delta F_1=\infty$ in Figure
549: \ref{fig:4}(a)), is marked by the line $X_0=0$ is the space of
550: parameters $\lambda_2/ \lambda_1 $ and $\Delta F_0 / \lambda_1$.
551:
552:
553: \begin{figure}[htbp]
554: \centering
555: \includegraphics*[width=7cm]{fig4}
556: \caption{Normal and inverted region in the space of parameters $\Delta
557: F_0/ \lambda_1$ and $\lambda_2 / \lambda_1$. Shown are the results for the forward
558: reaction $1\to 2$ (a) and for the backward reaction $2\to 1$ (b). The
559: quantum tunneling region with $\Delta F_1 = \infty$ in (a) is separated from
560: the inverted region by the condition $X_0 = 0$, which puts the
561: classical crossing point outside the range of reaction coordinates
562: $X< X_0$ allowed by the condition of thermodynamic stability. The
563: dots labeled as ``MH'' refer to the Marcus-Hush limit in which the
564: transition from the normal to inverted region is given by the
565: conditions: $\Delta F_0/ \lambda = - 1$ for $1\to 2$ and $\Delta F_0/ \lambda = 1$ for $2\to
566: 1$. }
567: \label{fig:4}
568: \end{figure}
569:
570: Due to the asymmetry of the free energy surfaces for $\lambda_2 / \lambda_1 > 1$,
571: the normal region spans different ranges of $\Delta F_0$ for positive and
572: negative free energy gaps. Figure \ref{fig:4} shows a broader
573: normal-range ET for $\Delta F_0 >0$. This observation suggests that the
574: goal of bringing the exothermic recombination reaction to the normal
575: region (Figure \ref{fig:1}) is easier to achieve when the
576: reorganization energy of the charge-separated state D$^+$--A$^-$ is
577: much higher than the reorganization energy of the ground state D--A
578: (see Discussion).
579:
580: \begin{figure}[htbp]
581: \centering
582: \includegraphics*[width=7cm]{fig5}
583: \caption{Energy gap law for the reaction 1$\to$2 ($\lambda_1 < \lambda_2$) at $\lambda_1=0.2$ eV
584: and $\lambda_2=0.25$ eV (dash-dotted line), $\lambda_2= 0.4$ eV (dashed line),
585: and $\lambda_2=2$ eV (solid line). The dotted lines separates the normal
586: (N), inverted (I), and quantum tunneling (QT) regions for $\lambda_2=2$
587: eV. }
588: \label{fig:5}
589: \end{figure}
590:
591:
592: The overall classical energy gap law (no vibrational excitations) for
593: the present model is illustrated in Figure \ref{fig:5}. With
594: increasing the ratio of the reorganization energies $\lambda_2/ \lambda_1$, the
595: bell-shaped dependence of $-\Delta F_1$ on $\Delta F_0$ becomes increasingly
596: shallow in its right wing with positive energy gaps $\Delta F_0>0$,
597: approaching the linear energy gap law $\Delta F_1 \propto \Delta F_0$. On the other
598: hand, the approach of the fluctuation boundary $X_0$ to the transition
599: point $X=0$ squeezes the left wing narrowing the inverted region of
600: ET.
601:
602:
603: \begin{figure}[htbp]
604: \centering
605: \includegraphics*[width=7cm]{fig6}
606: \caption{Inverted-region reaction $1\to2$ with $\lambda_1=0.2$ eV, $\lambda_2=1.0$ eV,
607: and $\Delta F_0=-2$ eV (shown by the vertical arrow). Classical
608: transitions between vibrationally ground-state surfaces are
609: forbidden by the condition $X_0<0$. The dashed surfaces obtained
610: for $n=3$ do not contribute to the Franck-Condon weighted density
611: of states since $X_0 + n\hbar\omega_v<0$ for $n=3$. For $n\geq 6$, $X_0 +
612: n\hbar\omega_v>0$ ($\hbar\omega_v=0.2$ eV) and these vibronic transitions contribute
613: to the overall density of states. The dash-dotted lines indicate
614: the fluctuation boundary $X_{0n}$ at $n=0$ and $n=3$; $X_{0n}=0$ at $n=6$. }
615: \label{fig:6}
616: \end{figure}
617:
618:
619: The quantum vibronic excitations of the donor-acceptor complex can
620: be included in the standard way by considering the Franck-Condon weighted
621: density of states (FCWD) as a Poisson-weighted sum over the vibronic excitations
622: leading to $n$ vibrational quanta in the final ET state.\cite{BixonJortner:99} For
623: instance, for the forward reaction one gets
624: %
625: \begin{equation}
626: \label{eq:2-26}
627: \mathrm{FCWD}(X;1\to 2) = e^{-S}\sum_{n=0}^{\infty} \frac{S^n}{n!} e^{-\beta F_{1n}(X)}
628: \end{equation}
629: Here, $S=\lambda_i/ \hbar\omega_v$ is the Huang-Rhys factor, $\omega_v$ is the characteristic vibrational
630: frequency, and $F_{1n}(X)$ is obtained from eq \ref{eq:2-8} by replacing there
631: $X_0$ with
632: %
633: \begin{equation}
634: \label{eq:2-27}
635: X_{0n}= X_0 + n\hbar\omega_v
636: \end{equation}
637:
638: The Arrhenius factor for the rate constant can be obtained (within a
639: factor) by putting $X=0$ in eq \ref{eq:2-26}. Since the thermodynamic
640: stability of the nuclear fluctuations requires $X<X_{0n}$, only terms
641: corresponding to $X_{0n}>0$ in eq \ref{eq:2-27} will contribute to the
642: sum in eq \ref{eq:2-26}. This fact implies a modification of the
643: standard picture of the FCWD made of a Poisson-weighted sum of
644: vibronic transitions. Vibronic transitions with $n< -X_0/ (\hbar \omega_v)$
645: will not contribute to the FCWD as is illustrated by $n=0$ and $n=3$
646: in Figure \ref{fig:6}. Only starting from vibrational excitation
647: number making $X_{0n}$ positive ($n=6$ in Figure \ref{fig:6}) will a
648: given vibronic transition contribute to the FCWD. This fact will
649: result in a greater asymmetry of emission lines making the red-side
650: wing of an optical band more shallow than the blue-side wing.
651:
652:
653: \section{Electron transfer in ferroelectrics}
654: \label{sec:3}
655: %
656: Equation \ref{eq:2-13} for the solvent reorganization energies of a
657: polarizable donor-acceptor complex is the central result of the formal
658: theory relevant to our discussion of reorganization anisotropy in
659: ferroelectrics. Equation \ref{eq:2-13} predicts that the
660: polarizability change $\Delta \alpha$ couples to the macroscopic and local
661: (reaction) fields to modify the change in the solute dipole moment in
662: the course of electronic transition. Depending on the relative signs
663: of $\Delta m$ and $\Delta\alpha(R_i+F_m)$, the solute polarizability may increase or
664: decrease the solvent reorganization energy compared to the
665: non-polarizable limit. The main question in this regard is whether
666: the term $\Delta\alpha(R_i + F_m)$ can achieve a magnitude comparable to $\Delta
667: m$. Below, we address this question by MC simulations of a model
668: donor-acceptor diatomic in a dipolar ferroelectric solvent. Here, we
669: first provide some relevant estimates.
670:
671: The mean-field Weiss theory of spontaneous dipolar
672: polarization\cite{Zhang:95} relates the macroscopic field in a sample
673: of aligned dipoles $m$ to their number density $\rho$ as
674: %
675: \begin{equation}
676: \label{eq:3-6}
677: F = (4\pi/3)\rho m
678: \end{equation}
679: The term $\Delta\alpha F_m/m$ at $\rho^*=\rho\sigma^3=0.7$ is then $\simeq 3\Delta\alpha/ \sigma^3$ when the
680: solute dipole is aligned with the macroscopic field, $\sigma$ is the
681: solvent diameter. The polarizability change may vary substantially
682: between different donor-acceptor units, but the ground-state
683: polarizability is close to $\sigma_0^3/16$ for many molecular systems
684: ($\sigma_0$ is the effective solute diameter). If $\Delta\alpha$ is of the same order
685: of magnitude as the ground-state polarizability, $3\Delta\alpha/ \sigma^3$ scales as
686: $(\sigma_0/ \sigma )^3$. For example, for primary charge separation in
687: \textit{Rhodobacter sphaeroids}, $\Delta m\simeq 53$ D.\cite{Bixon:95} If we
688: accept $\sigma=2.87$ \AA{} and $m=1.83$ D for water and $\Delta\alpha\simeq 800$
689: \AA,\cite{Middendorf:93} then $\Delta m/m\simeq 29$ and $\Delta\alpha F_m/m \simeq 100$ thus
690: resulting in comparable order-of-magnitude contributions from the
691: change in the permanent charges and from the interaction of the
692: polarizability change with the non-homogeneous electric field. In
693: addition, from the estimate of the local electric field of the protein
694: matrix at the position of the primary pair\cite{Steffen:94} $F\simeq 10^7$
695: V/cm, the induced dipole $\Delta \alpha F\simeq 27$ D is also comparable to $\Delta m \simeq
696: 53$ D.
697:
698:
699: \subsection{Simulations}
700: \label{sec:3-1}
701: A fluid of soft spheres is known to spontaneously form a ferroelectric
702: liquid phase when conducting boundary conditions are employed in the
703: simulation protocol.\cite{Wei:92} This system was used here to model a
704: disordered phase with a molecular-scale macroscopic polarization
705: coupled to the electronic states of the donor-acceptor complex. Two
706: sets of simulations have been carried out. The fluid of soft dipolar
707: spheres was simulated in the first set in order to establish the range
708: of parameters for which ferroelectric phase can be detected. This was
709: followed by the second set of simulations in which polarizable
710: donor-acceptor complex was dissolved in the ferroelectric liquid.
711:
712:
713: MC simulations of the pure solvent employed NVT ensemble of $N=500$
714: particles in a cubic simulation cell at the reduced density of
715: $\rho^*=\rho\sigma^3=0.7$. The molecules are interacting with the potential
716: %
717: \begin{equation}
718: \label{eq:3-7}
719: v(12) = 4 \epsilon \left(\sigma/r_{12}\right)^{12} - \mathbf{m}_1\cdot\mathbf{T}_{12}\cdot\mathbf{m}_2
720: \end{equation}
721: where $\epsilon$ is the repulsion energy, $\sigma$ is the diameter, $\mathbf{m}_j$
722: is the dipole moment with the magnitude $m$,
723: $\mathbf{T}_{12}=\nabla_1\nabla_2r_{12}^{-1}$ is the dipolar tensor, and
724: $r_{12}=|\mathbf{r}_1-\mathbf{r}_2|$. Simulations were done at
725: $\beta\epsilon=1.35$ and varying reduced dipole $(m^*)^2 = \beta m^2/ \sigma^3$. Periodic
726: boundary conditions and reaction-field correction for dipolar
727: interactions with conducting boundary conditions\cite{Allen:96} have
728: been employed. The length of simulations was $10^6$ cycles long far
729: from the transition point and up to $10^7$ cycles long close to
730: the transition to ferroelectric phase.
731:
732: \begin{figure}[htbp]
733: \centering
734: \includegraphics*[width=7cm]{fig7}
735: \caption{First ($S_1$, circles) and second ($S_2$, squares) order
736: parameters of the fluid of dipolar soft spheres vs $\beta m^2/ \sigma^3$
737: (a). Also shown are the dipolar susceptibility $\chi_P$ (b) and
738: constant-volume heat capacity $c_V$ (c). The dotted lines connect
739: the simulation points. The dash-dotted lines in (b) indicate the
740: points of phase transition to ferroelectric fluid ($m_F^*$) and
741: fcc crystal ($m_C^*$). }
742: \label{fig:7}
743: \end{figure}
744:
745: Transition to ferroelectric phase was monitored by calculating
746: the first-order, $S_1$, and second-order, $S_2$, parameters (Figure
747: \ref{fig:7}(a)). The first-order (ferroelectric) parameter quantifies the
748: spontaneous polarization
749: %
750: \begin{equation}
751: \label{eq:3-8}
752: S_1 = M/Nm
753: \end{equation}
754: where $\mathbf{M}$ is the total dipole moment of the solvent. The
755: second-order (nematic) parameter is defined as the largest eigenvalue
756: of the ordering matrix\cite{Allen:96}
757: %
758: \begin{equation}
759: \label{eq:3-9}
760: \mathbf{Q}=(2N)^{-1}\sum_j (3\mathbf{\hat e}_j\mathbf{\hat e}_j - \mathbf{1})
761: \end{equation}
762: where $\mathbf{\hat e}_j=\mathbf{m}_j/m$.
763:
764: Both order parameters change smoothly with $m^*$ and do not allow a
765: reliable identification of the transition point.\cite{Weis:05}
766: Susceptibilities, which are expected to show sharp spikes at the
767: points of phase transition,\cite{Binder:92} are better indicators.
768: Indeed, the dielectric susceptibility
769: %
770: \begin{equation}
771: \label{eq:3-10}
772: \chi_P = (\beta/V) \langle (\delta \mathbf{M})^2\rangle
773: \end{equation}
774: shows the first peak at $(m_F^*)^2=6.9$ and the second peak at
775: $(m_C^*)^2= 7.5$ (Figure \ref{fig:7}(b)), where $V$ in eq
776: \ref{eq:3-10} is the solvent volume. The heat capacity
777: %
778: \begin{equation}
779: \label{eq:3-11}
780: c_V = 3/2 + \beta^2 \langle (\delta E)^2 \rangle/N
781: \end{equation}
782: on the contrary, shows only a little bump at $m^*_F$ and a strong peak
783: at $m^*_C$ (Figure \ref{fig:7}(c)), where $E$ in eq \ref{eq:3-11} is
784: the total energy of the fluid. The order parameters also show weak
785: discontinuities at $m_C^*$. The examination of the density structure
786: factors reveals that the system is in the liquid state below $m^*_C$
787: and crystallizes into an fcc lattice at this value of the reduced
788: dipole. The point $m^*_F$ corresponds to the transition to a
789: ferroelectric liquid. Note that the value of $m^*_F$ obtained here is
790: very close to that predicted by a linear extrapolation of the phase
791: transition line recently reported by Weis\cite{Weis:05} for dipolar
792: hard spheres.
793:
794:
795: \begin{figure}[htbp]
796: \centering
797: \includegraphics*[width=7cm]{fig8}
798: \caption{Charges and polarizabilities of the donor-acceptor complex
799: in three states involved in photoinduced charge separation (CS) and
800: charge recombination (CR). }
801: \label{fig:8}
802: \end{figure}
803:
804: The ferroelectric liquid at $(m^*)^2=7.0$, $\rho^*=0.7$, and $\beta\epsilon=1.35$
805: was used to study the dependence of the solvent reorganization energy
806: on solute polarizability. In these simulations, a donor-acceptor
807: diatomic made of two hard fused spheres of diameters $\sigma_0/ \sigma = 1.5$
808: with the center-to-center separation $d/ \sigma = 0.6$ was inserted in the
809: center of a cubic simulation cell containing $N=500$ soft dipolar
810: particles at $\rho^*=0.7$. Two opposite charges $q_0^*= \beta|q_0|/ \sigma = 10$
811: where placed at the centers of two spheres in the charge-separated
812: state thus producing the dipole moment $m_{02} = d |q_0|$. The induced
813: point dipole moment $\mathbf{p}_{0i} = \mathbf{\hat e}_0 \alpha_{0i} R_i$,
814: caused by the dipolar polarizability $\alpha_{0i}$ and reaction field
815: $R_i$, is placed at the midpoint of the line connecting the two
816: centers of the diatomic. $\mathbf{p}_{0i}$ is aligned along the unit
817: vector $\mathbf{\hat e}_0$ along the same line. The solute induced
818: dipole was equilibrated to each instantaneous configuration of the
819: solvent by iteration algorithm (more details on the simulation
820: protocol are given in ref \onlinecite{DMjpca:04}). Ewald sums with
821: conducting boundary conditions\cite{Allen:96} were used to calculate
822: the solute-solvent electrostatic interactions.
823:
824: Some complications arise from the fact that the director,
825: $\mathbf{\hat d}= \mathbf{M}/Nm$, fluctuates in the simulations. In
826: real liquid crystals, fluctuations of the macroscopic director occur
827: on the time-scale of microseconds to seconds,\cite{Vertogen:88} much
828: slower than orientational motions of molecular solutes. The director,
829: therefore, does not fluctuate on the time-scale of the reaction. In
830: order to account for this hierarchy of relaxation times, the
831: simulation protocol was set up to keep the orientation of the
832: donor-acceptor complex fixed relative to the director. The initial
833: configuration was created by inserting a sphere of diameter $\sigma_0$ into
834: the soft dipolar solvent followed by a short, $5\times10^3$ cycles,
835: equilibration run designed to establish the director $\mathbf{\hat
836: d}$. The two spheres making the diatomic were then pulled apart at a
837: given angle to the director and this angle was adjusted after each
838: cycle over the $N$ solvent molecules. The data were collected from
839: simulation runs of the length $6\times 10^5-1.2\times 10^6$ cycles. In the
840: present simulations, the long axes of the diatomic was always aligned
841: with the director in such a way that the charge-transfer dipole $\Delta
842: \mathbf{m}_0$ is always parallel to the macroscopic field $\mathbf{F}$.
843:
844:
845: \begin{figure}[htbp]
846: \centering
847: \includegraphics*[width=7cm]{fig9}
848: \caption{The linear response parameter $\xi$ from eq \ref{eq:3-15}
849: (a) and the distribution function of the ferroelectric order
850: parameter (b) obtained from MC simulations at $(m^*)^2=7.0$,
851: $\beta\epsilon=1.35$, and $\rho^*=0.7$. The dotted line in (a) connects the
852: simulation points. }
853: \label{fig:9}
854: \end{figure}
855:
856: \subsection{Energetics of photoinduced electron transfer}
857: \label{sec:3-2}
858: We have followed the basic design of photoinduced ET outlined in
859: Figure \ref{fig:1} and detailed in terms of charges and
860: polarizabilities in Figure \ref{fig:8}. It is assumed that the
861: donor-acceptor complex has no appreciable dipole moment and
862: polarizability in its ground state ($q_0=0$, $\alpha_g =0$).
863: Photoexcitation of the donor to state D$^*$--A lifts the
864: polarizability to $\alpha_{01} >0$ but does not substantially change the
865: charge distribution, $q_0=0$. Both the charge distribution and
866: polarizability change upon charge separation resulting in D$^+$--A$^-$
867: state ($q_0>0$, $\alpha_{02} > 0$). Because of the coupling of the
868: polarizability to the reaction field $R_i$ and the macroscopic
869: electric field $F_m$ (eq \ref{eq:2-13}), the reorganization energies
870: will differ for forward and backward electronic transitions for
871: both the charge separation and charge recombination steps of the
872: reaction mechanism. Therefore, we need four reorganization energies,
873: $\lambda_1$ and $\lambda_2$ for charge separation and $\lambda_3$ and $\lambda_4$ for charge
874: recombination (Figure \ref{fig:8}).
875:
876: \begin{table}[htbp]
877: \centering
878: \caption{Energetic parameters (eV) of the charge separation reaction (Figure \ref{fig:8})
879: in which $\alpha_{01}$ is varied and $\alpha_{02}/ \sigma^3 = 0.2$ is kept constant;
880: $\beta q_0/ \sigma=10$ in the charge-separated state.}
881: \begin{tabular}{ccccc}
882: \colrule
883: $\alpha_{01} / \sigma^3$ & $\lambda_1$\footnotemark[1] & $\lambda_2$ & $\Delta X_0$\footnotemark[2]
884: & $\Delta X_0$\footnotemark[3] \\
885: \colrule
886: 0.2 & 0.324 & 0.456 & 0.817 & 0.770\\
887: 0.3 & 0.344 & 0.371 & 0.750 & 0.715 \\
888: 0.4 & 0.373 & 0.295 & 0.706 & 0.665 \\
889: 0.5 & 0.477 & 0.229 & 0.704 & 0.666 \\
890: 0.6 & 0.587 & 0.171 & 0.665 & 0.647 \\
891: 0.7 & 0.761 & 0.123 & 0.612 & 0.640 \\
892: 0.8 & 1.314 & 0.083 & 0.645 & 0.733 \\
893: \colrule
894: \end{tabular}
895: \footnotetext[1]{Calculated from the simulation data assuming $\beta=40$ eV$^{-1}$. }
896: \footnotetext[2]{From the simulation data.}
897: \footnotetext[3]{From eq \ref{eq:3-14}. }
898: \label{tab:1}
899: \end{table}
900:
901:
902:
903:
904: The reorganization energies were calculated from MC configurations as
905: second cumulants of the solute-solvent interaction
906: potential.\cite{DMjpca:04} The polarizability of the photoexcited state
907: $\alpha_{01}$ was varied at constant $\alpha_{02} /\sigma^3 =0.2$ for charge
908: separation and the polarizability of the charge-transfer state
909: $\alpha_{02}$ was varied at $\alpha_g=0$ for charge recombination. Tables
910: \ref{tab:1} and \ref{tab:2} and Figure \ref{fig:10} summarize the
911: forward and backward reorganization energies for charge-separation and
912: charge-recombination reactions along with the Stokes shift
913: %
914: \begin{equation}
915: \label{eq:3-13}
916: \Delta X_0 = X_{01} - X_{02}
917: \end{equation}
918: From eq \ref{eq:2-18}, the Stokes shift becomes
919: %
920: \begin{equation}
921: \label{eq:3-14}
922: \Delta X_0 = \kappa(\lambda_2-\lambda_1) - \lambda_2
923: \end{equation}
924: The Stokes shift in eq \ref{eq:3-13} is equivalent to the difference
925: of absorption and emission maxima due to eq \ref{eq:1-1}. Equation
926: \ref{eq:3-14} also applies to the Stokes shift defined in terms of
927: two first spectral moments
928: %
929: \begin{equation}
930: \label{eq:2}
931: \Delta X_0 = \langle X\rangle_1 - \langle X\rangle_2
932: \end{equation}
933: in the limit $\kappa_i\sqrt{\beta\lambda_i}\gg 1$.
934:
935:
936: \begin{figure}[htbp]
937: \centering
938: \includegraphics*[width=7cm]{fig10}
939: \caption{Reorganization energies and the Stokes shifts for the charge
940: separation (CS) and charge recombination (CR) reactions (Figure
941: \ref{fig:8}) vs polarizability of the photoexcited state $\alpha_{01}$
942: (CS) and the polarizability of the charge-separated state
943: $\alpha_{02}$ (CR); $q_0^*=10$, $\alpha_g=0$. }
944: \label{fig:10}
945: \end{figure}
946:
947: The direct comparison of the results of simulations to the
948: three-parameter model discussed in Section \ref{sec:2} is complicated
949: by the narrow range of solvent dipoles for which ferroelectric phase
950: can be detected. The close proximity of two phase transition points
951: makes the statistics of polarization fluctuations non-Gaussian, in
952: contrast to the Gaussian approximation adopted in eq \ref{eq:2-5}.
953: There are several indications of significant deviations from the
954: Gaussian statistics. The ratio
955: %
956: \begin{equation}
957: \label{eq:3-15}
958: \xi = - \langle v_{ss} \rangle / \beta \langle(\delta v_{ss})^2 \rangle
959: \end{equation}
960: is equal to one ($\xi=1$) when polarization fluctuations are Gaussian
961: (linear response approximation).\cite{DMjcp1:99} Here, $\langle v_{ss}\rangle$ is
962: the average electrostatic interaction energy of a liquid particle with
963: the rest $N-1$ particles in the liquid and $ \langle(\delta v_{ss})^2 \rangle$ is the
964: variance of this electrostatic interaction. As is seen from Figure
965: \ref{fig:9}(a), this parameter is higher than one in the paraelectric
966: phase and drops sharply at the transition to the ferroelectric phase.
967: In addition, the distribution of macroscopic polarization $M$ in
968: the simulation box is non-Gaussian. A shoulder seen in Figure
969: \ref{fig:9}(b) points to the importance of a $\propto M^4$ term in the
970: polarization functional such as that present in the Landau theory of
971: phase transitions.\cite{Vertogen:88}
972:
973: In terms of solvation properties, in the absence of polarizability
974: change of the solute, $\Delta X_0/2$ should be equal to $\lambda=\lambda_1=\lambda_2$. The
975: first lines in Tables \ref{tab:1} and \ref{tab:2} correspond to
976: exactly this situation. The difference between two reorganization
977: energies, as well as deviations of both of them from $\Delta X_0/2$ is
978: another indication of the non-Gaussian statistics of the solvent
979: polarization leading to non-linear solvation. Despite these
980: complications, the direct calculations of the Stokes shift from
981: simulations compare semi-quantitatively to the results of applying eq
982: \ref{eq:3-14} (Tables \ref{tab:1} and \ref{tab:2}). The analytical
983: model developed in Section \ref{sec:2-2}, therefore, captures the
984: basic thermodynamics of ET with markedly different reorganization
985: energies.
986:
987:
988:
989:
990: \begin{table}[htbp]
991: \centering
992: \caption{Energetic parameters (eV) of the charge
993: recombination reaction (Figure \ref{fig:8}) for which $\alpha_{02}$
994: is varied and $\alpha_g=0$ is kept constant; $\beta q_0/ \sigma=10$ in the charge-separated state. }
995: \label{tab:2}
996: \begin{tabular}{ccccc}
997: \colrule
998: $\alpha_{02}/ \sigma^3$ & $\lambda_3$\footnotemark[1] & $\lambda_4$ & $\Delta X_0 $\footnotemark[2] & $\Delta X_0$\footnotemark[3] \\
999: \colrule
1000: 0.0 & 0.369 & 0.279 & 0.649 & 0.642 \\
1001: 0.2 & 0.654 & 0.284 & 0.911 & 0.871 \\
1002: 0.3 & 1.115 & 0.287 & 1.115 & 1.051 \\
1003: 0.4 & 1.466 & 0.290 & 1.466 & 1.352 \\
1004: 0.5 & 2.270 & 0.293 & 2.270 & 1.728 \\
1005: 0.6 & 2.562 & 0.296 & 2.562 & 2.357 \\
1006: 0.7 & 3.240 & 0.299 & 3.240 & 2.928 \\
1007: 0.8 & 7.466 & 0.303 & 4.079 & 3.446 \\
1008: \colrule
1009: \end{tabular}
1010: \footnotetext[1]{Calculated from the simulation data assuming $\beta=40$ eV$^{-1}$.}
1011: \footnotetext[2]{From the simulation data.}
1012: \footnotetext[3]{From eq \ref{eq:3-14}. }
1013: \end{table}
1014:
1015:
1016: \section{Discussion}
1017: \label{sec:4}
1018: %
1019: The basic design of an artificial photosynthetic device, as advances
1020: by Meyer and co-workers,\cite{Huynh:04,Acevedo:05} is shown in Figure
1021: \ref{fig:11}. It anticipates the creation of ordered arrays where donors
1022: and acceptors are connected to catalytic sites where high-energy
1023: reactions can occur (e.g., splitting of water or reduction of carbon
1024: dioxide to carbohydrates). This design requires efficient separation
1025: of the electron and the hole. Unless this separation is achieved by
1026: high mobility of each charge in molecular arrays\cite{Hoertz:05} or in
1027: valence and conduction bands of a semiconductor,\cite{Lewis:05} high
1028: branching ratio between charge separation and charge recombination is
1029: required at each step of productive charge transfer to facilitate the
1030: redox reactions which are normally significantly slower than ET steps.
1031:
1032: \begin{figure}[htbp]
1033: \centering
1034: \includegraphics*[width=7cm]{fig11}
1035: \caption{Reaction center model.\cite{Huynh:04,Acevedo:05} The unit C
1036: is photoexcited to obtain the electron from donor D and transfer it
1037: to acceptor A. The recombination rates $k_r$ at each stage should
1038: be sufficiently low to allow normally slow catalytic reduction
1039: (Ox$_1\to$Red$_1$) and oxidation (Red$_2\to$Ox$_2$) reactions to
1040: occur. }
1041: \label{fig:11}
1042: \end{figure}
1043:
1044: The model presented here offers some novel opportunities in terms of
1045: varying the parameters of ET reactions and activation barriers. The
1046: model emphasizes the coupling of the solute polarizability change,
1047: achieved in the course of electronic transition, to the
1048: electric field at the position of the donor-acceptor complex. The
1049: electric field has two components, the reaction field of the polar
1050: environment $R_i$ and the macroscopic field $F_m$. The coupling of the
1051: overall field $R_i+F_m$ to the polarizability change creates the effective
1052: dipole moment change of the donor-acceptor complex (eq \ref{eq:2-13})
1053: %
1054: \begin{equation}
1055: \label{eq:4-1}
1056: \Delta m_{\text{eff},i} = \Delta m + \Delta \alpha (R_i + F_m)
1057: \end{equation}
1058: The reaction field $R_i$, which depends on the electronic state of the
1059: donor-acceptor complex, is responsible for the distinction between the
1060: forward and backward reorganization energies, $\lambda_1 \neq \lambda_2$ and $\lambda_3\neq
1061: \lambda_4$ as is shown in Figure \ref{fig:10} and listed in Tables
1062: \ref{tab:1} and \ref{tab:2}. Note that the asymmetry in the reorganization
1063: energies up to a factor of 25 obtained in our simulations (Table
1064: \ref{tab:2}) is the largest ever observed in either computer or
1065: laboratory experiment.
1066:
1067:
1068: \begin{figure}[htbp]
1069: \centering
1070: \includegraphics*[width=7cm]{fig12}
1071: \caption{Free energy surfaces of photoinduced ET in the Marcus-Hush
1072: picture (a) and in the present model with different reorganization
1073: energies (b). The free energy surfaces are plotted against the
1074: reaction coordinate corresponding to the energy gap between the
1075: charge-separated and photoexcited states. The parameters are
1076: chosen to show the activationless pathway from the photoexcited
1077: state 1 to the charge-separated state 2 and from 2 to the ground
1078: state 3. The free energy surfaces 2 and 3 coincide in the
1079: Marcus-Hush picture, but are distinct in the present model. In (a)
1080: the dashed line indicates the free energy surface for the
1081: normal-region charge recombination when energetic efficiency is
1082: below 50\%; $h\nu$ indicates the energy of photoexcitation. The
1083: reorganization energies are: $\lambda=1$ eV (a) and $\lambda_1=0.2$ eV,
1084: $\lambda_2=2.0$ eV, $\lambda_3=1.0$ eV, and $\lambda_4=0.5$ eV (b). }
1085: \label{fig:12}
1086: \end{figure}
1087:
1088: A note on the dual nature of the reorganization energy is relevant
1089: here. The reorganization energy for each ET state is calculated on
1090: the configurations of the solvent in equilibrium with the
1091: donor-acceptor complex in that given state. Therefore, the
1092: reorganization energy carries information about a particular
1093: electronic state of the donor-acceptor complex. On the other hand, the
1094: value that is being averaged is the change in the interaction
1095: potential, which is a characteristic of a given transition. As a
1096: result of this duality, the electronic charge-separated state of the
1097: donor-acceptor complex D$^+$--A$^-$ can be characterized by two,
1098: generally unequal, reorganization energies $\lambda_2$ and $\lambda_3$ reflecting
1099: transitions to two different electronic states, photoexcited and
1100: ground.
1101:
1102:
1103: In the present picture, all reorganization energies $\lambda_1$, $\lambda_2$,
1104: $\lambda_3$, and $\lambda_4$ are allowed to be different, producing reach
1105: energetics of photoinduced charge separation and charge recombination.
1106: The difference in scenarios for photoinduced ET in the Marcus-Hush
1107: model and in the present formulation is illustrated in Figure
1108: \ref{fig:12}. The plots show the free energy surfaces vs the reaction
1109: coordinate corresponding to the energy gap $X$ between the
1110: charge-separated and photoexcited states. The energy gap for the
1111: charge recombination reaction is then $-\Delta F_{\text{CS}} - \Delta
1112: F_{\text{CR}} + X$, where $\Delta F_{\text{CS,CR}}$ are the free energy
1113: gaps for charge-separation and charge-recombination reactions. In the
1114: Marcus-Hush model, activationless charge separation and charge
1115: recombination is achieved at the photoexcitation energy $h\nu = 2 \lambda$
1116: (Figure \ref{fig:12}(a)). Therefore, the desire to move the
1117: recombination reaction to the normal region requires energetic
1118: efficiency of the photosynthetic apparatus to be below 50\% (dashed
1119: line in Figure \ref{fig:12}(a)). This situation changes when
1120: different reorganization energies are allowed in the construction of
1121: the free energy surfaces. The plots in Figure \ref{fig:12}(b) are
1122: made with $\lambda_2/\lambda_1=10$ and $\lambda_3/ \lambda_4 = 5$. When activationless
1123: reactions are realized for $1\to 2$ and $2\to3$ transitions, the asymmetry
1124: of the energy gap law leads to a higher energetic efficiency of 70\%.
1125: The disadvantage of this scheme is a shallow shape of the
1126: charge-separated free energy surface making activation energy for
1127: charge recombination rather weakly dependent on changes in the free
1128: energy gap.
1129:
1130:
1131: The role of the reaction field factor in eq \ref{eq:4-1} will diminish
1132: in weakly polar media characteristic of protein electron transfer. The
1133: effect of a strong local electric field $F_m$ then becomes more
1134: important. If the term $\Delta \alpha R_i$ is small compared to $\Delta m$, we can
1135: simplify the effective dipole moment change to the form
1136: %
1137: \begin{equation}
1138: \label{eq:4-2}
1139: \Delta m_{\text{eff}} = \Delta m + \Delta \alpha F_m
1140: \end{equation}
1141: Since $\Delta m_{\text{eff}}$ does not depend on the electronic state, the
1142: forward and backward reorganization energies are the same ($\lambda_1=\lambda_2$,
1143: $\lambda_3=\lambda_4$), according to the Marcus-Hush theory. However, the factor
1144: $\Delta \alpha F_m$ can be responsible for the difference in the reorganization
1145: energies between charge separation and charge recombination reactions.
1146:
1147: \begin{figure}[htbp]
1148: \centering
1149: \includegraphics*[width=7cm]{fig13}
1150: \caption{Free energy surfaces in the Marcus-Hush picture when
1151: the reorganization energy for charge recombination $\lambda_3=\lambda_4=2.0$ eV
1152: is higher than reorganization energy for charge separation $\lambda_1=\lambda_2=0.5$ eV
1153: due to the coupling of the polarizability change to the macroscopic electric field
1154: (eqs \ref{eq:2-13} and \ref{eq:4-2}). }
1155: \label{fig:13}
1156: \end{figure}
1157:
1158: Imagine a situation in which the photoexcited state is significantly
1159: more polarizable than the ground state and the charge-separated state
1160: has about the same polarizability as the photoexcited state,
1161: $\Delta\alpha_{\text{CS}}\simeq 0$. Then, in weakly polar solvents, $\lambda_1\simeq\lambda_2$ and
1162: both reorganization energies are small because of the low polarity of
1163: the medium. For the charge recombination reaction, $\Delta m < 0 $ and
1164: $\Delta\alpha_{\text{CR}} < 0$. The permanent dipole $\Delta m$ and the induced
1165: dipole $\Delta \alpha F_m$ in eq \ref{eq:4-2} add up when $F_m>0$ and subtract
1166: when $F_m<0$. In the former case, the charge-recombination
1167: reorganization energy $\lambda_3\simeq\lambda_4$ is higher than the charge-separation
1168: reorganization energy $\lambda_1\simeq\lambda_2$. Note that this picture does not
1169: contradict to the energy conservation requirement given by eq
1170: \ref{eq:1-1} since the photoexcited state D$^*$--A and the ground
1171: state D--A are two different electronic states characterized by
1172: different polarizabilities.
1173:
1174: The normal-region ET can be realized within the Marcus-Hush picture
1175: when the reorganization energy is greater than the
1176: charge-recombination free energy gap $\lambda_3 > \Delta F_{\text{CR}}$ (Figure
1177: \ref{fig:13}). As in Figure \ref{fig:12}(a), the energetic efficiency
1178: is about 50\%, and it drops when charge recombination is shifted into
1179: the normal region. However, even in this limit of reduced flexibility
1180: of altering the ET parameters, the presence of the macroscopic
1181: electric field in the expression for the reorganization energy (eqs
1182: \ref{eq:2-13} and \ref{eq:4-2}) opens the door to manipulations of
1183: reorganization parameters through proper design of mutual signs of the
1184: polarizability change and the direction of the macroscopic field.
1185:
1186: We can conclude that the combination of a highly polarizable
1187: donor-acceptor complexe with macroscopic electric field creates a
1188: principal possibility for suppressing the recombination reaction in
1189: photosynthetic ET. From a more general perspective, the notion of
1190: non-equal reorganization energies and non-parabolic free energy
1191: surfaces, compliant with the condition of energy conservation (eq
1192: \ref{eq:1-1}), opens the door to new models of ET activation which may
1193: eventually lead to a practical solution of the photosynthesis
1194: problem.
1195:
1196:
1197: \begin{acknowledgments}
1198: This research was supported by the National Science Foundation
1199: (CHE-0304694).
1200: \end{acknowledgments}
1201:
1202: %\bibliographystyle{apsrev}
1203: \bibliographystyle{achemso}
1204: \bibliography{/home/dmitry/p/bib/chem_abbr,/home/dmitry/p/bib/photosynth,/home/dmitry/p/bib/et,/home/dmitry/p/bib/liquids,/home/dmitry/p/bib/solvation,/home/dmitry/p/bib/etnonlin,/home/dmitry/p/bib/dm,/home/dmitry/p/bib/dynamics,/home/dmitry/p/bib/ferro,/home/dmitry/p/bib/lc}
1205:
1206: \end{document}
1207:
1208: \newpage
1209: \begin{center}
1210: \textbf{FIGURE CAPTIONS}
1211: \end{center}
1212: \begin{description}
1213: \item[Figure 1] Energetics of photoinduced electron transfer involved
1214: in natural and artificial photosynthesis. The charge-separated
1215: state D$^+$--A$^-$ is placed below the photo-excited state D$^*$--A
1216: by the reorganization energy $\lambda$ to ensure activationless
1217: transition. The dash-dotted line marks the free energy surface of
1218: the initial state D--A with the reorganization energy four times
1219: lower than the value of $\lambda$ used to draw the free energy surfaces
1220: shown by the solid lines. The dashed line indicates the
1221: vibrationally excited electronic ground state for which the
1222: recombination transition D$^+$--A$^-\to$D--A does not require an
1223: activation barrier. C and C$'$ indicate the classical transition
1224: states.
1225: \item[FIGURE 2] Free energy surfaces of ET $F_i(X)$ for the initial
1226: ($i=1$, ``1'') and final ($i=2$, ``2'') ET states. Shown in the
1227: plot are the free energy minima $X_{0i}$, the free energy gap $\Delta
1228: F_0=F_{02}- F_{01}$, and the upper boundary $X_0$ for the reaction
1229: coordinate $X$.
1230: \item[FIGURE 3] Free energy surfaces $F_i(X)$ at $\lambda_1=1$ eV and
1231: $\lambda_2=2$ eV. The free energy gap $\Delta F_0$ is: $-2$ eV and $-7$ eV
1232: in (a) and 1 eV and 4 eV in (b). The free energy surfaces
1233: corresponding to $\Delta F_0=-7$ eV in (a) do not have the classical
1234: crossing point $X=0$ indicated by the dashed lines in both panels.
1235: \item[FIGURE 4] Normal and inverted region in the space of
1236: parameters $\Delta F_0/ \lambda_1$ and $\lambda_2 / \lambda_1$. Shown are the results for
1237: the forward reaction $1\to 2$ (a) and for the backward reaction $2\to
1238: 1$ (b). The quantum tunneling region with $\Delta F_1 = \infty$ in (a) is
1239: separated from the inverted region by the condition $X_0 = 0$,
1240: which puts the classical crossing point outside the range of
1241: reaction coordinates $X< X_0$ allowed by the condition of
1242: thermodynamic stability. The dots labeled as ``MH'' refer to the
1243: Marcus-Hush limit in which the transition from the normal to
1244: inverted region is given by the conditions: $\Delta F_0/ \lambda = - 1$ for
1245: $1\to 2$ and $\Delta F_0/ \lambda = 1$ for $2\to 1$.
1246: \item[FIGURE 5] Energy gap law for the reaction 1$\to$2 ($\lambda_1 < \lambda_2$)
1247: at $\lambda_1=0.2$ eV and $\lambda_2=0.25$ eV (dash-dotted line), $\lambda_2= 0.4$
1248: eV (dashed line), and $\lambda_2=2$ eV (solid line). The dotted lines
1249: separates the normal (N), inverted (I), and quantum tunneling (QT)
1250: regions for $\lambda_2=2$ eV.
1251: \item[FIGURE 6] Inverted-region reaction $1\to2$ with $\lambda_1=0.2$ eV,
1252: $\lambda_2=1.0$ eV, and $\Delta F_0=-2$ eV (shown by the vertical arrow).
1253: Classical transitions between vibrationally ground-state surfaces
1254: are forbidden by the condition $X_0<0$. The dashed surfaces
1255: obtained for $n=3$ do not contribute to the Franck-Condon weighted
1256: density of states since $X_0 + n\hbar\omega_v<0$ for $n=3$. For $n\geq 6$,
1257: $X_0 + n\hbar\omega_v>0$ ($\hbar\omega_v=0.2$ eV) and these vibronic transitions
1258: contribute to the overall density of states. The dash-dotted lines
1259: indicate the fluctuation boundary $X_{0n}$ at $n=0$ and $n=3$; $X_{0n}=0$ at $n=6$.
1260: \item[FIGURE 7] First ($S_1$, circles) and second ($S_2$, squares)
1261: order parameters of the fluid of dipolar soft spheres vs $\beta m^2/
1262: \sigma^3$ (a). Also shown are the dipolar susceptibility $\chi_P$ (b) and
1263: constant-volume heat capacity $c_V$ (c). The dotted lines connect
1264: the simulation points. The dash-dotted lines in (b) indicate the
1265: points of phase transition to ferroelectric fluid ($m_F^*$) and
1266: fcc crystal ($m_C^*$).
1267: \item[FIGURE 8] Charges and polarizabilities of the donor-acceptor
1268: complex in three states involved in photoinduced charge separation
1269: (CR) and charge recombination (CR).
1270: \item[FIGURE 9] The linear response parameter $\xi$ from eq
1271: \ref{eq:3-15} (a) and the distribution function of the
1272: ferroelectric order parameter (b) obtained from MC simulations at
1273: $(m^*)^2=7.0$, $\beta\epsilon=1.35$, and $\rho^*=0.7$. The dotted line in (a)
1274: connects the simulation points.
1275: \item[FIGURE 10] Reorganization energies and the Stokes shifts for
1276: the charge separation (CS) and charge recombination (CR) reactions
1277: (Figure \ref{fig:8}) vs polarizability of the photoexcited state
1278: $\alpha_{01}$ (CS) and the polarizability of the charge-separated state
1279: $\alpha_{02}$ (CR); $q_0^*=10$, $\alpha_g=0$.
1280: \item[FIGURE 11] Reaction center model.\cite{Huynh:04,Acevedo:05} The unit C
1281: is photoexcited to obtain the electron from donor D and transfer it
1282: to acceptor A. The recombination rates $k_r$ at each stage should
1283: be sufficiently low to allow normally slow catalytic reduction
1284: (Ox$_1\to$Red$_1$) and oxidation (Red$_2\to$Ox$_2$) reactions to
1285: occur.
1286: \item[FIGURE 12] Free energy surfaces of photoinduced ET in the Marcus-Hush
1287: picture (a) and in the present model with different reorganization
1288: energies (b). The free energy surfaces are plotted against the
1289: reaction coordinate corresponding to the energy gap between the
1290: charge-separated and photoexcited states. The parameters are
1291: chosen to show the activationless pathway from the photoexcited
1292: state 1 to the charge-separated state 2 and from 2 to the ground
1293: state 3. The free energy surfaces 2 and 3 coincide in the
1294: Marcus-Hush picture, but are distinct in the present model. In (a)
1295: the dashed line indicates the free energy surface for the
1296: normal-region charge recombination when energetic efficiency is
1297: below 50\%; $h\nu$ indicates the energy of photoexcitation. The
1298: reorganization energies are: $\lambda=1$ eV (a) and $\lambda_1=0.2$ eV,
1299: $\lambda_2=2.0$ eV, $\lambda_3=1.0$ eV, and $\lambda_4=0.5$ eV (b).
1300: \item[FIGURE 13] Free energy surfaces in the Marcus-Hush picture
1301: when the reorganization energy for charge recombination
1302: $\lambda_3=\lambda_4=2.0$ eV is higher than reorganization energy for charge
1303: separation $\lambda_1=\lambda_2=0.5$ eV due to the coupling of the
1304: polarizability change to the macroscopic electric field (eqs
1305: \ref{eq:2-13} and \ref{eq:4-2}).
1306:
1307:
1308: \end{description}
1309:
1310: \newpage
1311:
1312: \includegraphics*[width=12cm]{fig1}
1313: \vfill
1314: \textbf{Figure 1: Matyushov}
1315:
1316: \newpage
1317:
1318: \includegraphics*[width=12cm]{fig2}
1319: \vfill
1320: \textbf{Figure 2: Matyushov}
1321:
1322: \newpage
1323:
1324: \includegraphics*[width=12cm]{fig3}
1325: \vfill
1326: \textbf{Figure 3: Matyushov}
1327:
1328: \newpage
1329:
1330:
1331: \includegraphics*[width=12cm]{fig4}
1332: \vfill
1333: \textbf{Figure 4: Matyushov}
1334:
1335: \newpage
1336:
1337:
1338: \includegraphics*[width=12cm]{fig5}
1339: \vfill
1340: \textbf{Figure 5: Matyushov}
1341:
1342: \newpage
1343:
1344:
1345: \includegraphics*[width=12cm]{fig6}
1346: \vfill
1347: \textbf{Figure 6: Matyushov}
1348:
1349: \newpage
1350:
1351:
1352: \includegraphics*[width=12cm]{fig7}
1353: \vfill
1354: \textbf{Figure 7: Matyushov}
1355:
1356: \newpage
1357:
1358:
1359:
1360: \includegraphics*[width=12cm]{fig8}
1361: \vfill
1362: \textbf{Figure 8: Matyushov}
1363:
1364: \newpage
1365:
1366:
1367:
1368: \includegraphics*[width=12cm]{fig9}
1369: \vfill
1370: \textbf{Figure 9: Matyushov}
1371:
1372: \newpage
1373:
1374:
1375:
1376: \includegraphics*[width=12cm]{fig10}
1377: \vfill
1378: \textbf{Figure 10: Matyushov}
1379:
1380: \newpage
1381:
1382:
1383: \includegraphics*[width=12cm]{fig11}
1384: \vfill
1385: \textbf{Figure 11: Matyushov}
1386:
1387: \newpage
1388:
1389:
1390: \includegraphics*[width=12cm]{fig12}
1391: \vfill
1392: \textbf{Figure 12: Matyushov}
1393:
1394: \newpage
1395:
1396:
1397: \includegraphics*[width=12cm]{fig13}
1398: \vfill
1399: \textbf{Figure 13: Matyushov}
1400:
1401: \end{document}
1402:
1403:
1404: