Nonclassicalcrossoversin binary reactionsin one-dimensional systems
PanosArgyrakisa,Raoul Kopelmana,K. Lindenberg
Abstract:We examinethe crossoverbehaviorfor the A + A — 0 andA + B —* 0 reactions,which are highly nonclassical for a randominitial distribution of reactants.We find that both reactionsquickly departfrom the classicalbehaviorto a “depletionzoneregime”, eventuallyreachingthe segregatedasymptoticbehavior.We give the correspondingscaling lawsfor thecrossovertimeanddensity.Finite siLeeffectsarealso discussed. Computersimulationsverify the scalinglaws andprovide the scaling coefficients. Chemical reactionsin low dimensionsthat are circumstances the rate laws are PA — Pa = diffusion controlled do not behaveclassicallybut kPAPB. If PA(t — 0) = PB(t = 0) p(t = 0), then exhibit “anomalous”ratelawsfor the densitiesp(t) the densities of the two speciesare equal, so of the reactingspecies[1 8]. Thus, for the model ~ = — kp2. But, again it is well known that the reaction A + A —~ 0 while the normal rate law actual rate law in dimensions d < 4 is instead is ~ = — kp2, for dimensionsd < 2 it is ~ — = — kp” +4/d) Here the causeof the anomalous — kp~’ + 2fd) [1,4,8]. It is now well understoodthat behavioris the formationof aggregates of like parthis behavioris a consequenceof the spatialdistritides [1 8]. The regionswherethe densityof one bution of A’s: Initially the distributionhasa maxtype of particlesis overwhelminglygreater than imum at zeroseparation,but as the reactionprothat of the othergrow in time. Since the reaction ceedsthe particlesthat arein closeproximity have canessentiallyoccuronly at the interfacesbetween the chanceto react,thus leavinga depletionzone aggregates, andsincethenumberof theseinterfaces and a distribution that peaksat a finite, nonzero decreaseswith time, the reactionslowsdown relanearestneighbor separation.A similar situation tive to the rate that would describea randommixholds for the reactionA + B —~ 0. Undernormal ture of reactants.Again, this behaviorreflects the fact that diffusion is not an effectivemixing mechanismin low dimensions.Initial spatialfluctuations in relative densitiescan thus grow in size as the * Correspondingauthor. Permanentaddress:Departmentof reaction that eliminatesclose oppositepairs proPhysics,University of Thessaloniki,GR-54006Thessaloniki, ceeds.In this paperwe briefly discussthe various Greece. regimes of kinetic behaviorfrom the initial time 0022-2313/94/S07.00© 1994 Elsevier ScienceB.V. All rights reserved 55D10022-23l3(93)E0218-M 414 P. Argvrokis et at. Journal qi Luminescence 58 (1994) 413 416 Fig. 1. <~> p 0 1 versust for latticesof varioussizesL in onedimensionwith a randominitial condition. Theinitial densityof each speciesin numberper lattice site is P0 0.2 in all cases until the asymptoticbehavioris reached.We chardecay 2 hasbeenwell documentedto arisefrom acterizethe variousregimesandestimatethe crossthe “depletion zone” that forms aroundeachsurover times. We presentnumerical simulation reviving reactantafter the initially very close pairs suitsandanalyzethemin termsof our model.Only havequickly reacted[1]. Diffusion in one dimenqualitative results are presentedhere with some sion is known to be an ineffective mechanismtocharacteristicexamples,while a detailed analysis ward refilling the depletionzone. The samehapwill begiven in [9]. pensherein theA + B —* 0 reaction:After therapid The summaryof our results is given in Fig. 1, initial reactionof very nearbyA B pairs (classical which is a plot of <P> ‘ Po 1 versus t for one regime)eachsurviving A and B is surroundedby dimensional lattices of various sizes, and is the a region relativelypoor in the other species[10]. result of MonteCarlo simulationdata.We observe These depletion zones are not efficiently rethe various regimesin this figure as follows: First populatedby diffusion. In Fig. 1 we see that the comesa classicalregimeat t 0. Therateequation slopein this region is about0.55 (over the first 10 here is the classical rate equation: ~5(t)~1 ~ — steps). kp~(0),wherepg(t) is the averageglobal density, After a sufficientlylongtimehaselapsed, it is well and pg(O) = <p(r, 0)> Po. Thus, very near t 0 known [1 8] that the systemevolvestowardsmacthe solution is pg(t) — po/(l + kp0t) and also <p(r, roscopicsegregationand that <p(r, t)> —~ d4 for t)> — Po/(l + kpot). The initial slope at very early d < 4. At very long timesthe largestlattices show times is the same as that of the A + A problem, the Zeldovich behavior.For the largestlattice we which is the classicallyexpectedvalueof 1. find an averageslopeof 0.249over the lastdecade Then comes an intermediateregime where we of the simulation The crossovertime from the have<p(r, t)> d 2 for d < 2, and<p(r, t)> t A + A depletionzonebehaviorto the Zeldovich for d > 2. This behavioris nonclassicalin one dibehavior is discussed below and found to be mension,andis completelyanalogousto the nont, 102. classical regime of the reactionsA + A —+ 0 and Finally, at long times finite volume effects will A + A —* A. In thosecasestheslower-than-classical affectthe resultsof computersimulations.It is well P. Argyrakis et at. Journal of Luminescence 58 (1994) 413 416 415 known that thiscausesan exponentialdecayof the Thedimension-dependent fractiong~is expectedto density, <p(r, t)> ‘~ e zi, where x is a constantinbe independentof the initial density, and mustbe verselyproportionalto the size of the system.This determinedfrom simulations.For L = 100 we find behavioris expectedto take overwhenthe size of that finite size effects begin to set in when each aggregateis of the order of the size of the t 1 —~ 20. Accordingto Eq. (2), the averagesize of the system,or, if the initial densityis too small to lead aggregatesis approximatelyg1L 4(Dtf)” 2 —~ 13, to s gregationbefore finite ize effect set in, when i.e.,g 1 = 0.13. Thus,eachaggregateof eachspecies the averagedistancebetweenparticles is of the is on the averagelarger than 1/10 of the size of the order of the size of the system.This is also shown system(which of coursemustcontain at leasttwo clearly in Fig. 1, wherethe smallerlatticesshowthe aggregates). Sincewe assumethat the fraction g~is “peel off” dueto finite size effects. independent of L, we can use the value obtained The“major” crossovertime t, andaveragecrossabove,g1 0.13, to predictthe crossovertimes tf over density p5 for d ~ 4 occur when the ~ d/4 for the other lattices.We find for L = 1000 that segregationbehaviorsets in. This occursfrom the t1 —~ 210~, and for L = 10000 we obtain depletionzoneregime. Wewrite the crossoverdent1 210~. Both are consistentwith the simulasity as a fraction of the initial density: p. —fdpo, tion resultsof Fig. 1. where fd is a dimension-dependent fraction asWe concludethat the following sequenceof besumedto be of 0(1) andindependent of the initial haviorsis present:First at early times the A + B density.This crossovertime is found to be [9]: reactionbehavesthe sameway as the A + A reaction. Both exhibit classicalbehaviorat very early 1 timesbut quickly crossover to the depletionzone tS 8itDp 21’~f4“ (1) behavior ~ d/2 for d ~ 2. This parallel behavior 0 ci continuesup to a time t 5, whensegregationeffects beginin theA + B reaction.Weestimatethetimet~ Theequationsfor c~5and t5 canbe verified with the in Eq. (1) and, in particular, predict the product numericaldatafrom Fig. 1. Thevalueoffd mustbe t5p~/”to be independentof the initial density Po determinedfrom the simulations(but needonly be Thetransition to the Zeldovich regimeis therefore determinedfor onevalueof po). If our assumption expectedat shortertimesfor higherinitial densities. that fd is independentof Po is correct, then t5p~ We also predict the crossoverdensity p. to be should be independentof the initial density Po• proportionalto Po’ Our simulationsconfirm these Likewise, the ratio of the crossoverdensity to the predictionsquantitatively. Thus, a simulation for initial density,ps/po, shouldbe independent of Po• a single initial density in a particular dimension Our simulationsconfirm thesebehaviors, allows us to predict this crossoverquantitatively If the initial densityis highenoughfor thereto be for any other initial density. We find that in a segregationregime,finite size effects set in when onedimensiont5p0 0(10°)0(101)andp5/pa the aggregatesof like moleculesare of the order of 0.3 0.5. From these results we can also predict the size of the system. It is known [10] that the whetherthe transitionto the Zeldovich regimeoclinear size of the aggregateson the averagegrows cursearlieror later in two dimensionsthan in one with time as ad(Dt) 112. Here a~is a dimensiondimensionfor a givenvalueof Po~We find that t~(2 dependentconstantof proportionality, which apdimensions)/1 5(ldimension)= Kp0, where K is of pears[10] to beapproximately4 in onedimension. 0(101) 0(102). Thus, except for very low initial Finite size effects are thus expectedto set in at densities,the transition will set in later in two a time tf such that the averagelinear size of the dimensions. aggregatesin somefraction g~ 0(1) of the linear The Zeldovich regime in all casesends when size L of the system,ad(Dff)’ 2 g~Li.e., finite lattice size effects set in. Theseeffects begin whentheaveragelinearsize of eachkind of aggregL 2 tf g 2 2 (2) ate(which grows in timeas t~2) is within an order adD of magnitudeof the size of the system,independ416 P. Argyrakis ci al. Journal of Luminescence 58 (1994) 413 416 ently of the size of the system.In otherwords,the References finite size effects for lattices of linear size L start at a time t 1 ‘~ L 2 [cf. Eq. (2)]. Again, we [1] R. Kopelman,J. Stat. Phys. 42 (1986) 185; Science241 have determined dimension-specific parameters (1988) 1620. for this crossoverfrom our simulations.Further[2] V. Kuzovkov andF. Kotomin, Rep.Prog.Phys.51(1988) more,from thefinite sizeresultswe obtaininforma1479. [3] A.A. Ovchinnikov and Y.G. Zeldovich, Chem. Phys.28 tion about the segregation process in infinite (1978)215 lattices. [4] D. Toussaintand F. Wilczek, J. Chem. Phys. 78 (1983) Finally, we stress that