1: %2345678901234567890123456789012345678901234567890123456789012345678901234567890

2: %\documentclass[8pt]{article}

3: \documentclass[12pt]{article}

4: \usepackage{graphicx}

5: \usepackage{float}

6: %\setlength{\pagewidth}{8.5in}

7: %\setlength{\pageheight}{11.in}

8: \setlength{\topmargin}{-.25in}

9: \setlength{\headheight}{0in}

10: \setlength{\headsep}{0in}

11: \setlength{\textheight}{9.5in}

12: \setlength{\oddsidemargin}{.25in}

13: \setlength{\evensidemargin}{.25in}

14: \setlength{\textwidth}{6in}

15: \newcommand{\y }{\'{\i}}

16: \newcommand{\be }{\begin{equation}}

17: \newcommand{\ee }{\end{equation}}

18: %\pagestyle{empty}

19: \pagestyle{plain}

20: 

21: \begin{document}

22: 

23: 

24: \begin{center}

25: {\bf DECAY OF METASTABLE NONEQUILIBRIUM PHASES, ENHANCED REACTION RATE, AND 

26: DYNAMIC PHASE TRANSITION IN A MODEL OF CO OXIDATION WITH CO DESORPTION}\\

27: \vspace{15pt}

28: Erik Machado and Gloria M.~Buend\y a\\

29: {\it Physics Department, Universidad Sim\'on Bol\y var,\\

30: Apartado 89000, Caracas 1080, Venezuela}\\

31: \vspace{5pt}

32: Per Arne Rikvold\\

33: {\it Center for Materials Research and Technology,\\

34: School of Computational Science,  and Department of Physics, \\

35: Florida State University, Tallahassee, Florida 32306-4052, USA\\

36: and Department of Fundamental Sciences, 

37: Faculty of Integrated Human Studies, Kyoto University, Kyoto 606, Japan}\\

38: \vspace{5pt}

39: Robert M.~Ziff\\

40: {\it Department of Chemical Engineering and\\

41: Michigan Center for Theoretical Physics,\\

42: University of Michigan, Ann Arbor, MI 48109-2136, USA}

43: \end{center}

44: 

45: \vspace{15pt}

46: \begin{center}

47: ABSTRACT

48: \end{center}

49: %\vspace{5pt}

50: We present a computational study of the dynamic behavior of a

51: Ziff-Gulari-Barshad model of CO oxidation 

52: with CO desorption on a catalytic surface. 

53: Our results provide further evidence 

54: that below a critical desorption rate the model 

55: exhibits a non-equilibrium, first-order phase transition between low and high

56: CO coverage phases. Our kinetic Monte Carlo 

57: simulations indicate that the transition process between these phases

58: follows a decay  mechanism very similar to the one described by the 

59: classic Kolmogorov-Johnson-Mehl-Avrami

60: theory of phase transformation by nucleation and growth. 

61: We measure the lifetimes of the metastable phases on each side of 

62: the transition line and find that they

63: are strongly dependent on the direction of the transformation, 

64: i.e., from low to high coverage or 

65: vice versa. Inspired by this asymmetry, we introduce a square-wave periodic

66: external forcing, whose two parameters can be tuned to enhance the

67: catalytic activity. 

68: At CO desorption rates below the critical value, we find that 

69: this far-from-equilibrium system undergoes a dynamic phase transition

70: between a CO$_2$ productive phase and a nonproductive one. 

71: In the space of the parameters of the periodic external forcing, this 

72: nonequilibrium phase transition defines a line of critical points. The 

73: maximum enhancement rate for the CO$_2$ production rate occurs near 

74: this critical line.

75: 

76: %\newpage

77: \vspace{15pt}

78: \begin{center}

79: INTRODUCTION

80: \end{center}

81: %\vspace{5pt}

82: The study of phase transitions and critical phenomena in nonequilibrium 

83: statistical systems have recently attracted a great deal of attention due its

84: applications in many branches of physics, chemistry, biology, economics, and 

85: even sociology [1]. Within this field, surface reaction models have become

86: an archetype for studying out-of-equilibrium critical phenomena, 

87: and they have been intensely

88: analyzed with the purpose of designing more efficient catalytic processes [2]. 

89: The Ziff, Gulari, Barshad (ZGB) model [3] with desorption 

90: (the ZGB-k model) [4,5,6] describes kinetic aspects of the

91: gas-phase reaction CO+O $\rightarrow$ CO$_2$ on

92: a catalytic surface in terms of two parameters: the relative partial

93: pressure of CO, $y$, that represents the probability that the next

94: molecule arriving to the surface is a CO, and the desorption rate, $k$, 

95: which is related to 

96: the probability that an adsorbed CO molecule is desorbed without being oxidized. 

97: The overall reaction is assumed to occur according to the 

98: Langmuir-Hinshelwood mechanism,

99: \begin{eqnarray}

100: \mathrm{CO(g)}+\mathrm{E} & \rightarrow & \mathrm{CO(a)} \nonumber\\

101: \mathrm{O}_2\mathrm{(g)} + 2\mathrm{E} & \rightarrow & 2\mathrm{O(a)} \\

102: \mathrm{CO(a)} + \mathrm{O(a)} 

103: & \rightarrow & \mathrm{CO}_2\mathrm{(g)} + 2\mathrm{E} 

104: \;,

105: \nonumber

106: \label{eq:LH}

107: \end{eqnarray}

108: where E is an empty site on the surface, and (g) and (a) refer 

109: to the gas and adsorbed phase, respectively.

110: 

111: \vspace{15pt}

112: \begin{center}

113: SIMULATIONS AND RESULTS

114: \end{center}

115: %\vspace{5pt}

116: The ZGB-k model is simulated on a square lattice of linear size $L$ that 

117: represents the catalytic surface. The kinetic 

118: Monte Carlo simulation generates a 

119: sequence of trials: CO or O$_2$ adsorption with probability $1-k$, 

120: and CO desorption with probability $k$. For the adsorption, a CO or O$_2$ 

121: molecule is selected with probability $y$ or $1-y$, respectively.  

122: We calculate the  coverages $\theta_{\rm CO}$ and

123: $\theta_{\rm O}$,  defined as the fraction of surface sites occupied by 

124: CO and O, respectively, and 

125: the rate of production of CO$_{2}$, $R_{{\rm CO}_2}$. In Fig.~\ref{histos4}  

126: we show $P(\theta_{\rm CO})$, the

127: probability distribution for $\theta_{\rm CO}$ vs $y$, where it is clearly 

128: seen that at a particular value of $y$,

129: $y_2(k)$, a low $\theta_{\rm CO}$ and a high $\theta_{\rm O}$ phase coexist.

130: \begin{figure}[H]

131: %\centering\includegraphics[scale=1.00]{histos4.eps}

132: %\centering\includegraphics[scale=0.45]{histos4.eps}

133: \centering\input{histos4.tex}

134: \caption[]{Order-parameter probability distribution, 

135: $P(\theta_{\rm CO})$, for $k=0.02$ and $L=100$.

136: The distribution for the value of $y$ closest to the coexistence value, $y_2$, 

137: is shown with a bold line.}

138: \label{histos4}

139: \end{figure}

140: 

141: A finite-size-scaling analysis of the statistical fluctuations 

142: of the CO coverage gives strong evidence that below the critical value of  

143: $k$ the model exhibits a first-order, nonequilibrium phase transition 

144: between low and high CO coverage phases with the same characteristics as a 

145: first-order equilibrium phase transition [4,5,7].

146: 

147: We also measured the metastable lifetimes associated with the transition from 

148: the low CO coverage phase to the high CO coverage phase and vice versa. 

149: The system-size dependence of the decay times 

150: strongly suggests that the system follows a decay mechanism very similar to 

151: the one described by the

152: classic Kolmogorov-Johnson-Mehl-Avrami theory of phase transformation by 

153: nucleation and growth near

154: a first-order equilibrium phase transition, with well-defined single-droplet 

155: and multidroplet regimes [8,9]. In this system, the desorption parameter and 

156: the distance to the coexistence curve play

157: the roles of the temperature and the supersaturation 

158: or overpotential, respectively. Near the 

159: coexistence curve the decay times are inversely

160: proportional to $1/L²$, and the decay mechanism consists of the nucleation 

161: and growth of a single droplet of the stable phase. Far from the coexistence 

162: curve, the decay times are independent of the system size, and the decay 

163: proceeds by random nucleation of many droplets of the stable phase [5].

164: 

165: \begin{figure}

166: \centering\includegraphics[scale=.65, clip]{decay2.eps}

167: \caption[]{Decay times

168: as functions of $y$ when the system evolves toward the low CO

169: coverage region $\langle \tau_d \rangle$, and when it evolves toward the 

170: high coverage region $\langle \tau_d \rangle$, shown

171: for $k=0.02$ and $L=100$. Here and elsewhere in this paper,

172: time is measured in Monte Carlo steps per site (MCSS).}

173: \label{decay}

174: \end{figure}

175: We found that the lifetimes strongly depend on $k$ and on the direction of 

176: the process; the mean decontamination time $\langle \tau_d \rangle$ 

177: (from high to low CO coverage) 

178: being different from the mean poisoning time $\langle \tau_p \rangle$ (from

179: high to low CO coverage). At comparable distances from 

180: the coexistence curve, $\langle \tau_d \rangle \gg \langle \tau_p \rangle$, 

181: as seen Fig.~\ref{decay}.

182: Since several experiments indicate that it is possible to increase the 

183: efficiency of a catalytic process by subjecting the system to periodic 

184: forcing [10],

185: we decided to exploit the asymmetry between the decay times by subjecting 

186: the system to a periodic

187: variation of the external pressure with periods related to the decay times 

188: in each direction. We therefore select 

189: a square-wave periodic variation of the CO pressure, $y(t)$, that 

190: in a period $T=t_{d}+

191: t_{p}$ takes the values, $y_l$ (located below the transition pressure) 

192: during the time interval $t_{d}$ and $y_h$ (located above the transition 

193: pressure) during the time interval $t_{p}$, as can 

194: be seen in Fig.~\ref{serie_osc3}(a). 

195: \begin{figure}

196: \centering\includegraphics[scale=0.65]{serie_osc3.eps}

197: \caption{(a) Applied periodic pressure of CO, $y(t)$, 

198: that takes the values

199: $y_l=0.5$ and $y_h=0.5346$ during the time intervals $t_d=10$ and

200: $t_p=50$, respectively.

201: (b) Response of the production rate to the

202: applied pressure given in (a) for $L=100$ with $k=0.02$.

203: The dotted line marked $\langle r \rangle $ 

204: indicates the long-time average of the 

205: period-averaged CO$_2$ production rate $r$, 

206: while the dotted line marked $\langle R_{\rm CO_2} \rangle_{\rm max}$ marks 

207: the maximum average CO$_2$ production rate for {\it constant\/} $y=y_2(k)$.

208: %Time is measured in units of Monte Carlo steps per site (MCSS).

209: }

210: \label{serie_osc3}

211: \end{figure}

212: 

213: We found that the times that the driving force spends in the low and high 

214: coverage regions, $t_d$ and $t_p$, respectively, can be tuned for each set 

215: of $y_l$ and $y_h$ to increase the productivity of the system. 

216: In Fig.~\ref{serie_osc3}(b) it is seen that the CO$_2$ production rate 

217: exhibits an oscillatory behavior in response to the periodic pressure shown in 

218: Fig.~\ref{serie_osc3}(a). The period-averaged value

219: of the CO$_2$ production rate $R_{\rm CO_2}$, defined as

220: \begin{equation}

221: r = \frac{1}{T}\oint R_{{\rm CO}_2}(t) \mathrm{d}t,

222: \label{order_parameter}

223: \end{equation}

224: plays the role of the  dynamic order parameter [6]. For the parameters 

225: selected in Fig.~\ref{serie_osc3}, the long-time average of $r$, 

226: $\langle r \rangle$ takes

227: a value that is about 7\% higher than the maximum average CO$_2$ 

228: production rate for constant $y$. It is likely that more careful 

229: tuning of the parameters could further improve the degree 

230: of enhancement.  

231: 

232: We also found that, for sufficiently low values of $k$,  the driven system 

233: undergoes a dynamic phase transition between a dynamic phase of high CO$_2$ 

234: production,  $r > 0$, and a nonproductive one, $r\approx 0$, as can be seen 

235: in Fig.~\ref{r_vs_td3}. The distinction between these phases disappears for 

236: desorption rates above the critical value [6].

237: \begin{figure}

238: \centering\includegraphics[clip, scale=0.65]{r_vs_td3.eps}

239: \caption{Long-time average of the period-averaged rate of CO$_2$ production, 

240: $\langle r\rangle$, shown vs $t_d$ for two values of $t_p$ and $L=100$;

241: (a) with $y_l=0.52$, $y_h=0.535$, 

242: and $k=0.01$, and (b) with $y_l=0.52$, $y_h=0.553$, and $k=0.04$.

243: Only for the lower value of $k$ 

244: does the system clearly present two dynamic phases: 

245: one with $\langle r\rangle\approx 0$ and the other with $\langle r\rangle>0$.}

246: \label{r_vs_td3}

247: \end{figure}

248: 

249: A detailed 

250: finite-size scaling analysis indicates that for small values of $k$, the 

251: measure of the fluctuations of the order parameter, 

252: \begin{equation}

253: X_L = L^2 [\langle r^2 \rangle - \langle r \rangle ^2]

254: \;,

255: \label{X}

256: \end{equation}

257: diverges as 

258: a power law with the system size, $X_{L}^\mathrm{max} \approx L^{\gamma/\nu}$ 

259: with exponent $\gamma/\nu =1.77\pm 0.02$, while moments of

260: the order parameter at the transition point decay as 

261: $\langle r^n \rangle_L \approx L^{-n\beta/\nu}$ with 

262: $\beta/\nu=0.12\pm0.04$. These exponent 

263: ratios, together with general symmetry arguments, give reasonable evidence 

264: that this far-from equilibrium phase transition 

265: belongs to the same universality class as the equilibrium Ising model [6]. 

266: The long-time average production rate $\langle r \rangle$ is shown in a 

267: density plot vs $t_p$ and $t_d$ in Fig.~\ref{fig:image}. The line of 

268: critical points appears as the sharp boundary of the black, low-production 

269: region in the lower right-hand part of the figure. 

270: The region of maximum average 

271: production is seen to lie very close to the critical line on its 

272: high-production side. 

273: We believe this observation offers a clue to understanding

274: the physical reason for the enhancement. Most likely, the long-range critical 

275: correlations associated with the critical cluster that develops for parameter

276: values near the critical line produce a high density of sites where 

277: CO molecules would be situated next to O atoms, in positions conducive to 

278: rapid oxidation and desorption of the produced CO$_2$. 

279: \begin{figure}

280: \centering\includegraphics[clip, scale=0.75]{image.eps}

281: \caption{

282: Long-time average $\langle r \rangle$ of the period-averaged CO$_2$ 

283: production rate $r$, shown as a density plot vs $t_p$ and $t_d$ for

284: $y_l=0.51$, $y_h=0.535$, $k=0.01$, and $L=100$.

285: }

286: \label{fig:image}

287: \end{figure}

288: 

289: \vspace{15pt}

290: \begin{center}

291: CONCLUSION

292: \end{center}

293: %\vspace{5pt}

294: In this paper we have summarized results of large-scale kinetic Monte Carlo 

295: simulations of the ZGB-k model of heterogeneously catalyzed 

296: CO oxidation with CO desorption. The results of the simulations were analyzed 

297: using finite-size scaling methods. 

298: We found that there are strong similarities between the dynamics of 

299: metastable decay in this far-from equilibrium, non-Hamiltonian system and the 

300: well-known behavior of Hamiltonian systems. From a theoretical point of 

301: view, these similarities could lead to significant advancement in 

302: our understanding of the 

303: dynamics of far-from-equilibrium systems. From a practical point of view, 

304: our results can be exploited 

305: to develop novel ways to increase the efficiency of catalytic reactions. 

306: 

307: \clearpage

308: %\vspace{15pt}

309: \begin{center}

310: ACKNOWLEDGMENTS

311: \end{center}

312: \noindent

313: This work was supported in part by U.S.\ National Science 

314: Foundation Grant No.\ DMR-0240078 at Florida State University and 

315: Grant No.\ DMS-0244419 at The University of Michigan. \\

316: 

317: \begin{center}

318: REFERENCES

319: \end{center}

320: 

321: %\begin{thebibliography}{99}

322: 

323: %\bibitem{general1} 

324: \noindent

325: 1. H. ~J.\ Jensen, in {\it Self-Organized Criticality:

326: Emergent Complex Behavior in Physical and Biological

327: Systems}, Cambridge University Press, Cambridge (1998); 

328: J.~Marro and R.~Dickman, in

329: {\it Non-equilibrium Phase Transitions in Lattice Models},

330: Cambridge University Press, Cambridge (1999)\\

331: 

332: %\bibitem{surface}

333: \noindent

334: 2. K.~Christmann, {\it Introduction to Surface

335: Physical Chemistry}, Steinkopff Verlag, Darmstadt (1991); V.~P.\

336: Zhdanov and B.~Kazemo, { Surf.\ Sci.\ Rep.} {\bf 20}, 111 (1994);

337: G.~C.\ Bond {\it Catalysis: Principles and

338: Applications}, Clarendon, Oxford (1987).\\

339: 

340: %\bibitem{ziff86}

341: \noindent

342: 3. R.~M.\ Ziff, E.~Gulari, and Y.~Barshad, Phys.\ Rev.\ Lett.\

343: \textbf{56}, 2553 (1986).\\

344: 

345: %\bibitem{tome93}

346: \noindent

347: 4. T.~Tom\'e and R.~Dickman, Phys.\ Rev.\ E

348: \textbf{47}, 948 (1993).\\

349: 

350: %\bibitem{machado1}

351: \noindent

352: 5. E.~Machado, G.~M.\ Buend\y a, and P.~A.\ Rikvold, Phys.\ Rev.\ E

353: \textbf{71}, 031603 (2005).\\

354: 

355: %\bibitem{machado2}

356: \noindent

357: 6.  E.~Machado, G.~M.\ Buend\y a, P.~A. Rikvold, and R.~M.\ Ziff, 

358: Phys.\ Rev.\ E \textbf{71}, 016120 (2005).\\

359: 

360: %\bibitem{ziff2}

361: \noindent

362: 7. R.~M.\ Ziff and B.~J.\ Brosilow, Phys.\ Rev.\ A

363: \textbf{46}, 4630 (1992).\\

364: 

365: %\bibitem{kJMA}

366: \noindent

367: 8. A.~N.\ Kolmogorov, Bull.\ Acad.\ Sci.\ USSR, Phys.\ Ser.\

368: {\bf 1}, 335 (1937);

369: W.~A.\ Johnson and R.~F.\ Mehl,

370: Trans.\ Am.\ Inst.\ Mining Metall.\ Eng.\ {\bf 135}, 416 (1939);

371: M.~Avrami, J.\ Chem.\ Phys.\ {\bf 7}, 1103 (1939);

372: {\bf 8}, 212 (1940); {\bf 9}, 177 (1941).\\

373: 

374: \noindent

375: 9. P.~A.\ Rikvold, H.~Tomita, S.~Miyashita, and S.~W.\ Sides,

376: Phys.\ Rev.\ E {\bf 49}, 5080 (1994).\\ 

377: 

378: %\bibitem{imbhil95

379: \noindent

380: 10. R.~Imbhil and G.~Ertl,  Chem.\ Rev.\

381: {\bf 95}, 697 (1995);

382: M.~Ehsasi, M.~Matloch, O.~Frank, J.~H.\ Block, K.~Christmann,

383: F.~S.\ Rys, and W.~Hirschwald,  J.\ Chem.\ Phys.\

384: {\bf 91}, 4949 (1989).\\

385: 

386: %\end{thebibliography}

387: 

388: \end{document}

389: