Macroscopic versus microscopic Fick’s laws describe time evolution macroscopically Physical Chemistry Lecture 2 Random walks; microscopic theory of diffusion Diffusion can be understood in terms of a microscopic process – the random walk Probability of event Each event is considered equally likely unless other information is known Can only discuss likelihood of an event, p n = number of possible results Only applies to a “large sample” Examples Coin flips (n=2) Throws of dice (n=6) For events that are not equally likely, one must know (by some external means) the probability, p, of each event p Uses probability concepts Implies randomness in a system with many molecules Time development occurs in a “natural” way Random motion of a single molecule is the underlying driving force described by diffusion. Probability of a sequence Random events Do not specifically consider a single molecule (although we discussed them in that manner) Focus on the tendency to eliminate density gradients Can be explained by ad hoc appeal to kinetic theory 1 n Probability of a sequence, P, is product of the probabilities of the elementary events that make up each sequence When the probabilities of the elementary events are the same, this equation simplifies Example: tossing a coin four times or tossing four coins at the same time Since ph = pt = ½, each P p1 p2 p3 p4 pn sequence has a likelihood of happening P = (½)4 16 possible sequences, so the probability of any one sequence is 1/16 (= (½)4) 1 Probability of a configuration P' Pi s e c n e u q e s e t a i r p o r p p a l l a r e v o Often concerned only ︵ with the number of elementary outcomes in nappropriate sequences P a sequence, not order (if Pi P for all sequences) Example of tossing a coin: probability of getting 3 heads when tossing a coin four times ︶ Four different sequences have three heads P ‘ = 4(1/6) = 1/4 One-dimensional random walk Particle hops from site to site, starting at zero Only one step per hop Probability of hopping in either direction is ½ for each step Determine probability that, after m steps, the particle is at position q after m steps Just like evaluating the probability of a sequence Numbers of configurations Counting number of specific configurations can be difficult if there is a large number of configurations Need an expression for number of sequences having a certain property to use in probability calculation Mathematically equivalent to counting the number of ways of putting balls in boxes, 1 ball in each box Example: 3 heads and 1 tails Nconfig = 4!/(3! 1!) = 4 nboxes ! nh !nt ! nh nt ! nh ! nt ! N config (nh , nt ) Mathematics of random walks Probability has two factors P ' (q; m) P N config m 1 N config (n, p) 2 Number of ways of having p positive and n negative steps to end up at q after m steps N config (n, p) m! n! p! m! mq mq ! ! 2 2 These equations give the probability of being at q after m steps in a random walk 2 Calculation of averages in a one-dimensional random walk Use the probability, P ’, to get averages of functions of the distance in m steps f (q ) m 1 P' (q, m) f (q) 2 q m Examples: q 0 q2 m m m m! m q m q f (q ) qm ( )!( )! 2 2 Example random walk Movement of He in a given time Technically only correct for P ' ( x, t ) dx either even or odd q, but we “smooth” the probability over many steps Gaussian function Normalized probability distribution function Same form as the result of solving Fick’s law Equivalence defines the average jump distance, x 0 2 q2 exp( ) 2m 2 m x2 1 exp dx 4Dt 4 Dt Fick’s Law x0 m z t 1.6 cm 1 minute 12.1 cm 1 hour 93.9 cm 1 day 460 cm 1 week 1220 cm In one minute, a molecule samples a reasonable fraction of the environment in that flask. Gaussian function Occurs in many different experiments having random processes x2 1 P ( x, ) exp( 2 ) 2 2 2 Shape is determine by the standard deviation, x qx0 x2 x 1 second Typical flask is of the order of 10 cm in diameter. Small-step-size, large-stepnumber random walk P ' ( q, m) TIME Distance moved in one direction x The average position does not appear to change with number of steps, but the square of the distance traveled does. It can be shown that P’ may be approximated well by a continuous function, when there is a large number of very small steps. T = 298.15 K P = 1 bar Large , wide function Small , narrow function Random noise is gaussian 2 Dt 3 Three-dimensional diffusion Assume diffusion in the three spatial dimensions is uncorrelated P ( x, y, z; t )dxdydz P( x, t ) P( y, t ) P( z , t )dxdydz x2 y2 z2 1 dxdydz exp 3/ 2 4 Dt 4Dt In spherical co-ordinates it simplifies and depends only on r P ( r , , ; t ) d r2 2 1 r sin drdd exp 3/ 2 4Dt 4 Dt Diffusion in liquids Average distance traveled in a liquid between collisions much smaller than in a dilute gas Mean free path for gas Molecular diameter for liquid Collisions happen on a short time scale Typically 1-10 ps between collisions for liquids Typically 10-100 ps between collisions for gases Average distance in 3D diffusion Average over the 3D distribution Assumes isotropic diffusion Sum of three terms r2 x2 y2 z 2 2 Dt 2 Dt 2 Dt 6 Dt r 3D Compare to 1D and 2D diffusion The average distance traveled in a fixed time is larger for movement in higher-dimensional space x r 2D r2 6 Dt 2 Dt 4 Dt Stokes-Einstein equation A moving particle in a fluid feels a drag force resisting the movement of the particle Proportional to the speed, with a coefficient, f A retarding force Results from collisions with molecules of the fluid For a sphere, Stokes showed the relationship of the coefficient, f, to viscosity, , and the sphere’s radius, r Einstein connected the coefficient, f, to the meansquare displacement The combination of the two theoretical results gives the Stokes-Einstein equation Relates size and viscosity to the diffusion coefficient Knowing any two of D, , and r, one can calculate the third Fviscous Fviscous f f f dx dt dx dt 6r x 2 (t ) D T 2kTt f k 6r 4 Stick, slip, and shape Original SEE derived under “stick” conditions Appropriate for large molecules in smallmolecule solvent For small molecules, can use the “slip” condition D Appropriate for small molecules For ellipsoidal molecules, must correct for shape by introducing a factor, (Pe) D kT 6 r kT 4 r Rotational diffusion ( stick ) ( slip ) 1 kT ( Pe) 6 re 1 kT ( Pe) 5.451 2 / 3V 1/ 3 D Rotational motion opposed by frictional forces Treated similarly to translational diffusion Result for a sphere experiencing stick conditions Translational and rotational diffusion coefficients have different units Drot kT 8r 3 Dtrans kT 6r Summary Use of probability theory leads to a prediction for random movement of particles (Brownian motion) Apparent change in a system is the result of random events at the microscopic level No coherent external forces Results of random-walk theory are equivalent to the macroscopic theory (Fick’s laws) which postulate a force related to concentration gradients Connects macroscopic parameter to theory of the microscopic structure and dynamics of matter Stokes-Einstein relationship between diffusion coefficient and averages over the random motion of particles Rotational motion also hindered by viscous drag and treated as diffusive 5