1: \documentstyle[preprint,aps]{revtex}
2: \tightenlines
3: %\documentstyle[twocolumn,aps]{revtex}
4: \begin{document}
5:
6:
7:
8: \title{Nonlinear Effects In Site Blocking
9: Induced Oscillations}
10: \draft
11: % repeat the \author\address pair as needed
12: \author{Dorjsuren Battogtokh$^{1,2}$ and Ravjir Davaanyam$^2$}
13: \address{$^1$Deparment of Biology, Virginia Polytechnik
14: Institute and State University, Blacksburg, VA 24061\\
15: $^2$Physics and Technology Institute,
16: Mongolian Academy of Sciences,Ulaanbaatar 51, Mongolia}
17: \date{\today}
18: \maketitle
19: \begin{abstract}
20: Higher order nonlinear effects of site blocking,
21: which induces oscillations in the Monte Carlo
22: simulation of CO oxidation, are outlined here. It is shown
23: that the rate equations which include these
24: effects exhibit a supercritical Hopf bifurcation
25: in parameter domains where
26: Monte Carlo simulations lead to oscillations.
27:
28: \end{abstract}
29: % insert suggested PACS numbers in braces on next line
30:
31: \pacs{Pacs: 05.45.-a, 82.20 Wt.}
32:
33: % body of paper here
34: \section{Introduction}
35: Surface chemical reactions play an important role in the processes
36: of heterogeneous catalysis, which are widely used in the chemical
37: industry~\cite{ASM}. Monte Carlo simulation is a powerful tool
38: for a microscopic description of these reactions~\cite{Gil,David}.
39: Kinetic phase transitions, oscillations and chaos, pattern
40: formation,
41: and coupling between catalytic oscillators have been studied
42: by the Monte Carlo method~\cite{ZGB,Kortluke,Zhdanov,Pattern}.
43:
44: The oxidation of carbon monoxide $CO$ is one
45: of the most extensively studied
46: surface catalytic reactions~\cite{Ronald,Imbihl}.
47: In a recent Monte Carlo study, an attenpt of a mean field
48: modeling was made for site
49: blocking induced surface coverage oscillations~\cite{Jansen}.
50: The authors of Ref.~\cite{Jansen} studied
51: a mathematical model based on macroscopic rate equations for the
52: characterization of oscillations seen in the Monte Carlo
53: simulations. However, it was
54: found that the mean field model does not lead to
55: oscillatory dynamics in a parameter region where their
56: Monte Carlo simulations show oscillations.
57: In order to obtain oscillations, they
58: numerically solved a part of their rate equations with
59: data from the Monte Carlo simulation for dynamics
60: of the free site and the inert molecule coverage.
61: Such a combination of
62: deterministic and stochastic time evolutions seems not to be
63: satisfactory; it is more desirable to find a supercritical
64: Hopf bifurcation in the model
65: to ensure the occurrence and
66: stability of oscillations.
67: The aim of this work
68: is to show that inclusion of high order nonlinear terms,
69: which represent interactions inherent in this system,
70: allows us to find oscillations within the rate equations.
71: We find a Hopf bifurcation in the extended system, and
72: we can determine the parameter domain for oscillatory dynamics.
73:
74: \section{Site Blocking Induced Oscillations}
75: The model system under study is the ZGB model~\cite{ZGB},
76: which is extended by the absorption and desorption of an
77: inert molecule:
78:
79: \begin{eqnarray}
80: CO_{gas}+* \stackrel{w_1}{\rightarrow}CO_{ads},\\\nonumber
81: CO_{ads}\stackrel{w_2}{\rightarrow}CO_{gas} + *,\\\nonumber
82: O_2+2*\stackrel{w_4}{\rightarrow}2O_{ads},\\\nonumber
83: CO_{ads}+O_{ads}\stackrel{w_3}{\rightarrow}CO_2+2*,\\\nonumber
84: M_{gas}+*\stackrel{w_5}{\rightarrow}M_{ads},\\\nonumber
85: M_{ads}\stackrel{w_6}{\rightarrow}M_{gas}+*,
86: \end{eqnarray}
87: where $ads$ indicates that the molecule is absorbed on the
88: surface and $gas$ indicates that the molecule is in the gas phase.
89: $*$ indicates a vacant site and $M$ represents the site blocking
90: inert molecule.
91:
92: As was reported by Jansen and Nieminen~\cite{Jansen},
93: one can observe sustained coverage oscillations
94: in the Monte Carlo simulations of Eqn. (1).
95: In Fig. 1 we show results of
96: our Monte Carlo simulation for these coverage oscillations.
97: All variables and time are dimensionless in this paper.
98: As compared with Ref.~\cite{Jansen},
99: we have used a different set of parameters to show
100: that these oscillations can be observed in wide range of parameters.
101: To improve the numerical performance we have implemented
102: {\em lists} in our Monte Carlo simulations. Separate lists for the
103: locations of free sites, $CO$, and inert molecules were
104: updated during the simulations. For example, for an adsorption of
105: a chosen adsorbent, a site was randomly selected from the
106: list of free sites; for a desorption, a site
107: was randomly selected from the corresponding list of $CO$ or $M$.
108:
109: The authors of Ref.~\cite{Jansen} studied rate equations
110: for a macroscopic characterization of the oscillations seen in the
111: Monte Carlo simulation. The rate equations for the coverage of
112: $CO_{ads}$ - $x$, $O_{ads}$ - $y$, and $M_{ads}$ - $z$
113: are~\cite{Jansen}:
114: \begin{eqnarray}
115: \sigma=1-x-y-z,\\\nonumber
116: \dot x=w_1 \sigma-w_2 x-4 w_3 x y,\\\nonumber
117: \dot y=4 w_4 \sigma^2-4 w_{3} x y,\\\nonumber
118: \dot z=w_5 \sigma -w_6 z,
119: \end{eqnarray}
120: where $\sigma$ is the free site coverage.
121: It is known that Eqn. (2)
122: shows oscillations in a narrow parameter domain~\cite{Vigil},
123: however, temporal pattern of these oscillations are very different from
124: the oscillations seen in Monte Carlo simulations as shown in Fig. 1.
125: It was found that
126: Eqn. (2) does not show oscillations in the
127: parameter domain where Monte Carlo simulations
128: lead to oscillations.
129: To obtain oscillations,
130: the authors of
131: Ref.~\cite{Jansen} used data for $\sigma$ and $z$ from the
132: Monte Carlo simulation.
133: We suppose that the reason why
134: Eqn. (2) does not show oscillations in the
135: parameter domains of intereset
136: is that it does not
137: include important higher order nonlinear effects
138: inherent in Eqn. (1), which are essential
139: for oscillations shown in Fig. 1.
140:
141: \section{Higher order Nonlinear effects}
142: Monte Carlo simulations of Eqn. (1)
143: suggest that two nonlinear effects are crucial for
144: oscillations in this system. They destabilize $O$
145: and $CO$ rich surfaces, correspondingly.
146:
147: In Fig. 2 we show snapshots of the surface at two different time
148: moments. Fig. 2a corresponds to the time moment when there
149: are no $CO_{ads}$ molecules left on the surface.
150: Note that in the original ZGB
151: model, for the parameters we have chosen, the system goes to the
152: state fully covered by $O_{ads}$~\cite{ZGB}.
153: However, the presence
154: of the inert molecules shown in Fig. 2a as the brightest spots,
155: prevent such a fully $O_{ads}$ covered state. Moreover, these
156: inert molecules allow for the presence of enough vacant
157: sites; these are shown
158: as black areas in Fig. 2. In these vacant areas,
159: $CO$ molecules can be
160: absorbed; their coverage can grow if they
161: are isolated by the inert molecules from $O_{ads}$ covered
162: areas. By taking into account this nonlinear effect,
163: which includes interaction among $CO_{ads}$, $O_{ads}$, and
164: $M_{ads}$, the reaction term in Eqn. (2) can be modified
165: as $4 w_3 x y(1- \kappa z)$, where $\kappa$ is the blocking
166: coefficient.
167:
168: Fig. 2b shows nucleation of the $O_{ads}$ covered state on the
169: $CO_{ads}$ rich surface. As an adsorption of $O_2$ requires
170: two free sites,
171: nucleation of the $O_{ads}$ island may seem an unlikely
172: process,
173: if one starts from the
174: fully $CO_{ads}$ covered state as an initial condition.
175: However, there is an another nonlinear effect
176: which may lead to explosive growth of an $O_{ads}$ island.
177: Suppose that the desorptions of
178: two $CO_{ads}$ molecules have vacated a pair of adjacent sites.
179: On these sites, $O_2$ can be absorbed, and if this happens,
180: $CO_{ads}$ and $O_{ads}$ will immediately react. As
181: the surface is rich with $CO_{ads}$, the
182: reaction will free four sites. On these four sites,
183: two $O_2$ molecules can be absorbed. This cascade may
184: lead to explosive growth of the $O_{ads}$ island. Therefore,
185: there is an another
186: nonlinear effect which is particularly
187: noticeable during a nucleation of an
188: $O_{ads}$ island on a $CO_{ads}$ rich surface.
189: In the lowest order of nonlinearity,
190: due to the
191: nonlinear increase
192: of available adjacent free sites through the reaction,
193: the growth of $O_{ads}$
194: can be termed as $4\omega w_3 x^2 y$, where $\omega$ is
195: the coefficient for the nonlinear growth of adjacent free sites.
196: We verified that $4\omega w_3 x^2 y^2$ can also serve well for this term.
197:
198: The two nonlinear effects manifest themselves
199: randomly in Monte Carlo simulations. For a macrospcopic
200: description, we consider them deterministic processes.
201: Thus, Eqn. (2) can be modified into the form:
202:
203: \begin{eqnarray}
204: \dot x=w_1 \sigma-w_2 x-4 w_3 x y(1- \kappa z),\\\nonumber
205: \dot y=4 w_4 \sigma^2-4 w_3 x y(1-\omega x),\\\nonumber
206: \dot z=w_5 \sigma -w_6 z.
207: \end{eqnarray}
208:
209: We want Eqn. (3) as simple as possible.
210: We include the reaction term $4 w_3 \kappa x y z$
211: in the first equation only because its effect
212: in the second equation is assumed to be small.
213: This assumption stems from a
214: microscopic process: a pair of $O_{ads}$
215: can not be blocked by a single inert molecule.
216: The reason why we include $4 w_3 \omega x^2 y$
217: in the second equation only is also taken from a
218: microscopic fact: for the adsorption
219: of $CO$, the presence of two free adjacent
220: sites is not essential.
221:
222: We note that the two effects were also described by
223: Jansen and Nieminen~\cite{Jansen}; however,
224: they did not include
225: them explicitly in the rate equations. We also note
226: that Eqn. (3)
227: is much simpler than the rate equations in the pair
228: approximation of Eqn. (1), which do not show oscillations
229: in the parameter range of interest~\cite{Vigil}.
230: Eqn. (3) is an approximate model,
231: and it is desireable to
232: derive it systematically.
233:
234:
235:
236: \section{A Hopf Bifurcation}
237:
238: Because of its nonlinear terms,
239: analytic treatment of Eqn. (3) is difficult.
240: Nevertheless, it can be studied
241: numerically. For example, one may use
242: the software "XPPAUT" by B. Ermountrout~\cite{Brad}.
243:
244: We found that for the occurrence of a Hopf bifurcation
245: it is not necessary that both $\kappa$ and $\omega$
246: were nonzero in Eqn. (3). However, we found that parameter
247: space for oscillations are larger if both
248: $\kappa$ and $\omega$ are nonzero.
249:
250: \subsection{case $\kappa \neq 0$ and $\omega \neq 0$}
251: In the present case, it is difficult to find
252: analytic expressions for the steady states of Eqn. (3).
253: In Fig. 3 we show a numerical integration of Eqn. (4).
254: Here, in contrast to its dynamics shown in
255: Fig. 1, the inert molecule concentration remains high.
256: We suppose that the main reason for this is that
257: $w_3$ is finite in Fig. 3, but $w_3=\infty$ in Fig. 1.
258: Also, a better choice of the rate constants can be made
259: in order to compare dynamics in the rate equations and
260: Monte Carlo simulations~\cite{Gil}.
261:
262: A Hopf bifurcation in this system is shown in Fig. 4.
263: Thus, the nonlinear effects we outlined in the
264: previous section may indeed lead to stable oscillations.
265: A two parameter
266: bifurcation diagram is shown in Fig. 5.
267: We found that with the increase of $w_3$,
268: the parameter region for a Hopf bifurcation
269: widens. We found that such an increase
270: is accompanied by the emergence of saddle nodes.
271:
272: \subsection{case $\kappa \neq 0$ and $\omega = 0$}
273: In the present case, the steady states are given by,
274: \begin{eqnarray}
275: x_0={w_2}^{-1} (w_1 z_0 -4 w_4 {z_0}^2+4 \kappa w_4 {z_0}^3),\\\nonumber
276: y_0=1-x_0-2 z_0,
277: \end{eqnarray}
278: where $z_0$ is a solution to
279: \begin{eqnarray}
280: w_2^2 w_4 z_0+w_3 (w_1+4 w_4 z_0 (\kappa z_0-1)\\\nonumber
281: (w_2(2 z_0-1)+z_0(w_1+4 w_4 z_0(\kappa z_0-1)))=0.
282: \end{eqnarray}
283:
284: Once again we were able to analyze
285: stability of $x_0$, $y_0$ and $z_0$
286: numerically only.
287: Oscillations for the present case are shown in Fig. 6.
288: We note that in the present case, except the Hopf bifurcation,
289: saddle node bifurcations occur in the system.
290:
291:
292: \subsection{case $\kappa = 0$ and $\omega \neq 0$}
293: As in the previous case, the steady states of Eqn. (3)
294: can be expressed through a polynomial. Again, we
295: were able to solve the polynomial numerically only.
296: As a result, a linear stability analysis of these steady states
297: has turned out to be difficult. Numerically detected
298: oscillations for the present case are shown in Fig. 7.
299: Note the exceptions of $w_3$'s value and $\omega \neq 0$;
300: these are the reaction rates used in Ref.~\cite{Jansen}
301: for Monte Carlo simulations.
302: We note that for the parameters used in Fig. 6 - 7, a parameter
303: domain for the Hopf bifurcation
304: was smaller in the present case then it was in the previous case.
305:
306:
307: \section{Discussions}
308: A microscopic study of a reaction system based
309: on Monte Carlo simulations can reveal the molecular
310: mechanisms of catalytic processes. As an example,
311: in this work, such study has allowed us
312: to outline the essential nonlinear processes
313: in site blocking caused oscillations. However,
314: fora more complete study of a reaction system,
315: the role of diffusion should be taken into account
316: ~\cite{Syn}. In that sense,
317: Eqn. (4) can be a useful starting point
318: for a reaction diffusion study of site blocking
319: induced oscillations. The preliminary results of a
320: reaction diffusion system in one dimension,
321: with a diffusion of $x$, $y$, and $z$, show
322: synchronous oscillations, stable long lived islands
323: and diffusion induced chemical turbulence~\cite{Sensei}.
324: More detailed results on this, and its comparison
325: to the Monte Carlo simulations, which include diffusion
326: of adsorbents, will be reported elsewhere~\cite{Bat}.
327:
328: Finally, we suppose that
329: the nonlinear effects and stable oscillations
330: due to site blocking can be detected in experimental
331: studies on $CO$ oxidation.
332:
333: % now the references. delete or change fake bibitem. delete next three
334: % lines and directly read in your .bbl file if you use bibtex.
335: \begin{references}
336:
337:
338: \bibitem{ASM} A. S. Mikhailov and G. Ertl, Chaos {\bf 12}, 107
339: (2002).
340:
341: \bibitem{Gil} D. T. Gillespie, J. Chem. Phys. {\bf 81}, 2340 (1977).
342:
343: \bibitem{David} D. P. Landau and K. Binder, {\em A Guide to
344: Monte Carlo Simulations in Statistical Physics} (Cambridge
345: University Press, Cambridge, 2000).
346:
347: \bibitem{ZGB} R. M. Ziff, E. Gulari, Y. Barshad, Phys. Rev.
348: Lett. {\bf 56}, 2553 (1986).
349:
350: \bibitem{Kortluke} O. Kortluke, V. N. Kuzovkov, W. Niessen,
351: Phys. Rev. Lett. {\bf 72}, 2105 (1998).
352:
353: \bibitem{Zhdanov} V. P. Zhdanov, B. Kasemo, Chaos {\bf 11}, 335
354: (2001).
355:
356: \bibitem{Pattern} R. J. Gelten, et. al., J. Chem. Phys.
357: {\bf 108}, 9921 (1998).
358:
359: \bibitem{Ronald} R. Imbihl and G. Ertl, Chem. Rev. {\bf 95}, 697.
360: (1995).
361:
362: \bibitem{Imbihl} R. Imbihl, Progress in Surf. Sci. {\bf 44}, 185 (1993).
363:
364: \bibitem{Jansen} A. P. J. Jansen and R. M. Nieminen, J. Chem.
365: Phys. {\bf 106}, 2038 (1997).
366:
367: \bibitem{Vigil} R. D. Vigil, and F. T. Willmore, Phys. Rev. E
368: {\bf 54}, 1225 (1996).
369:
370:
371: \bibitem{Brad} B. Ermontrout, {\em Simulating Analyzing, and Animating
372: Dynamical Systems}, (SIAM, 2002); http://www.pitt.edu/=phase.
373:
374: \bibitem{Syn} A. S. Mikhailov, {\em Foundations of Synergetics
375: I. Distributed Active Systems} 2nd. revised edn. (Springer-
376: Verlag, Berlin, 1994).
377:
378: \bibitem{Sensei} Y. Kuramoto, {\em Chemical Oscillations
379: Waves and Turbulence}, (Springer -Verlag, Berlin, 1984).
380:
381: \bibitem{Bat} D. Battogtokh, in preparation.
382:
383: \end{references}
384:
385: % figures follow here
386: %
387: % Here is an example of the general form of a figure:
388: % Fill in the caption in the braces of the \caption{} command. Put the label
389: % that you will use with \ref{} command in the braces of the \label{} command.
390: %
391: \begin{figure}
392: \caption{
393: Coverage oscillations on the square lattice.
394: Open circles: $\bar \theta_{CO_{ads}}$.
395: Long dashed: $\bar \theta_{O_{ads}}$, Solid lines:
396: $\bar \theta_{M_{ads}}$. Rate constants:
397: $w_1=0.44$, $w_2=0.005$, $w_3=\infty$,
398: $w_4=0.56$, $w_5=0.004$, $w_6=0.004$. Time is scaled
399: to the total number of sites
400: $S=N^2$, $N=256$.
401: $\bar \theta=\frac {1}{S} \sum \theta_i,
402: i=CO_{ads}, O_{ads}, M_{ads}$.}
403: \label{1}
404: \end{figure}
405:
406: %
407: \begin{figure}
408: \caption{Snapshots of the surface at two different
409: moments. a.) A phase without $CO_{ads}$ coverage.
410: b.) Emergence
411: of an $O_{ads}$ island on the $CO_{ads}$ dominant surface.
412: Parameters are the same as in Fig. 1.}
413: %\label{}
414: \end{figure}
415:
416: %
417: \begin{figure}
418: \caption{Coverage oscillations in Eqn. (3).
419: Open circles: $x$. Long dashed line: $y$.
420: Solid line: $z$.
421: Parameters are the same as in Fig. 1 except $w_3=5$,
422: $\omega=1$, $\kappa=2$.}
423: %\label{}
424: \end{figure}
425:
426: %
427: \begin{figure}
428: \caption{A Hopf bifurcation diagram. Solid line represents a
429: stable steady state. Circles represent the Hopf
430: bifurcation branch: filled circles show the supercritical
431: Hopf bifurcation, open circles show unstable oscillations.
432: Dashed and thin lines show an unstable
433: steady state. Parameters are the same as in Fig. 3.}
434: %\label{}
435: \end{figure}
436:
437: %
438: \begin{figure}
439: \caption{A two parameter bifurcation diagram.
440: Oscillations occur in the area marked by {\em osc}. Other parameters
441: are the same as in Fig. 3.}
442: %\label{}
443: \end{figure}
444:
445: %
446: \begin{figure}
447: \caption{
448: Oscillations in the case $\omega =0$.
449: Open circles: $x$. Long dashed line: $y$.
450: Solid line: $z$.
451: Parameters are: $w_1=0.2$, $w_2=0.01$, $w_3=5$,
452: $w_4=1$, $w_5=w_6=0.01$ and $\kappa=5.5$}
453: %\label{}
454: \end{figure}
455:
456: %
457: \begin{figure}
458: \caption{
459: Oscillations in the case $\kappa =0$.
460: Open circles: $x$. Long dashed line: $y$.
461: Solid line: $z$.
462: Parameters are: $w_1=1$, $w_2=0.001$, $w_3=5$,
463: $w_4=0.52$, $w_5=w_6=0.0003$, and $\omega=1$.}
464: %\label{}
465: \end{figure}
466:
467:
468:
469:
470: % tables follow here
471: %
472: % Here is an example of the general form of a table:
473: % Fill in the caption in the braces of the \caption{} command. Put the label
474: % that you will use with \ref{} command in the braces of the \label{} command.
475: % Insert the column specifiers (l, r, c, d, etc.) in the empty braces of the
476: % \begin{tabular}{} command.
477: %
478: % \begin{table}
479: % \caption{}
480: % \label{}
481: % \begin{tabular}{}
482: % \end{tabular}
483: % \end{table}
484:
485:
486: \newpage
487: \begin{picture}(3.375,4.5)
488: \special{psfile=Figure1.eps hoffset=-60 voffset=-500
489: vscale=100 hscale=100 angle=0}
490: \end{picture}
491:
492: \vspace{20cm}
493: \centerline{\LARGE {\bf Figure 1}}
494:
495: \newpage
496: \begin{picture}(3.375,4.5)
497: \special{psfile=Figure2.eps hoffset=-60 voffset=-800
498: vscale=100 hscale=100 angle=0}
499: \end{picture}
500:
501: \newpage
502: \begin{picture}(3.375,4.5)
503: \special{psfile=Figure3.eps hoffset=-60 voffset=-500
504: vscale=100 hscale=100 angle=0}
505: \end{picture}
506:
507: \vspace{20cm}
508: \centerline{\LARGE {\bf Figure 3}}
509:
510:
511: \newpage
512:
513: \begin{picture}(3.375,4.5)
514: \special{psfile=Figure4.eps hoffset=-133 voffset=-600
515: vscale=100 hscale=100 angle=0}
516: \end{picture}
517:
518:
519: \vspace{20cm}
520: \centerline{\LARGE {\bf Figure 4}}
521:
522: \newpage
523:
524: \begin{picture}(3.375,4.5)
525: \special{psfile=Figure5.eps hoffset=-133 voffset=-600
526: vscale=100 hscale=100 angle=0}
527: \end{picture}
528:
529: \vspace{20cm}
530: \centerline{\LARGE {\bf Figure 5}}
531:
532: \newpage
533: \begin{picture}(3.375,4.5)
534: \special{psfile=Figure6.eps hoffset=-60 voffset=-500
535: vscale=100 hscale=100 angle=0}
536: \end{picture}
537:
538:
539: \vspace{20cm}
540: \centerline{\LARGE {\bf Figure 6}}
541:
542: \newpage
543: \begin{picture}(3.375,4.5)
544: \special{psfile=Figure7.eps hoffset=-60 voffset=-500
545: vscale=100 hscale=100 angle=0}
546: \end{picture}
547:
548: \vspace{20cm}
549: \centerline{\LARGE {\bf Figure 7}}
550:
551:
552: \end{document}
553:
554:
555:
556:
557:
558:
559:
560:
561:
562:
563: