Learning Unit 4: Diffusion in substitutional solid solutions Outline Introduction 1. Diffusion in an ideal solution, Fick’s first law of diffusion 2. Diffusion in an ideal solution, Fick’s second law of diffusion 3. The Kirkendall effect 4. Porosity 5. The Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions 6. Solutions to the Fick’s laws of diffusion 7. Solutions to the Fick’s first law of diffusion 8. Solutions to the Fick’s second law of diffusion – Grube method 9. Solutions to the Fick’s second law of diffusion- Matano- Boltzmann method Outline 10. Determination of the intrinsic diffusivities 11. Self-diffusion in pure metals 12. Temperature dependence of the diffusion coefficient 13. Chemical diffusion at low solute concentration 14. The study of chemical diffusion using radioactive tracers* 15. Diffusion along grain boundaries and free surfaces 16. Fick’s first law in terms of mobility and an effective force* 17. Diffusion in non-isomorphic systems 18. Application of diffusion concept to the homogenization of castings Objectives • On completion of this learning unit, you should: – Understand how diffusion occurs, and what the driving force behind it is – Understand Fick’s laws and what factors will determine the rate at which diffusion occurs for some simple cases – Understand the effects of temperature and microstructure on diffusion – Be familiar with why diffusion is important to a range of applications Introduction: diffusion in the context of physical metallurgy • Diffusion in solid state as the means to bring in physical contact atoms involved in phase transformations during the processing/service of metals and alloys • Diffusion in solid state as the phenomenon that control the kinetics of phase transformations occurring during the processing/service of metals and alloys Processing : casting , (cold, warm or hot) working, metal bonding - welding, Properties machining, powder processing, heatheattreatments, surface engineering, etc. Processing involves phase transformations Performance Phase transformation kinetics • Diffusive processes, Kinetics of diffusion •Johnson Melh Avrami Kologoromov (JMA/K) kinetics •Thermal activation Structures Processing • Competitive nucleation kinetics • Growth morphologies • TTT and CCT diagrams Introduction: Driving force of diffusion – Down-hill and up-hill diffusion • Gradient of free energy • Gradient of chemical potential • Gradient of activity • Gradient of concentration • Gradient of temperature • Gradient of electric field • Local state of stress Diffusion always occurs so as to decrease in Gibbs free energy (chemical potential). It could occur down the concentration gradient for down-hill diffusion (absence of strong interaction between A and B atoms: ideal solid solution) or up the concentration gradient for up-hill diffusion (gradient of temperature, presence of strong interaction between A and B atoms – ordering or clustering systems – spinodal decomposition: negative or positive deviation, etc.) Introduction: Diffusion mechanisms (atomic steps) Substitutional mechanisms : vacancy, direct interchange, cooperative interchange a. Vacancy b. Interstitial c. Interstitial- vacancy d. Direct Interchange e. Zener- Ring rotation (cooperative interchange) ! Break of bonds with neighbours and lattice strain → Energy barrier The diffusive jump actually requires additional energy and cooperation from the close-packed neighbours to accommodate the jumping atom. Diffusion is thermally activated. It requires an adjacent vacant site and a high energy of the solute atom and neighbouring atoms. All these factors are strongly temperature dependant. For an active diffusion to occur, the temperature should be high enough to overcome energy barriers to atomic motion. This energy comes from the thermal energy of atomic vibrations (Eav ~ kBT) Theoretical computation of energy required of each mechanism, distortion energy and activation energy rather too complicated compared to experimental determination. Introduction: Diffusion mechanisms (atomic steps) Interstitial diffusion is generally faster than vacancy diffusion because bonding of interstitials to the surrounding atoms is normally weaker and there are many more interstitial sites than vacancy sites to jump to. Interstitial diffusion involves small impurity or solute atoms (e.g. C, H, O, N) to fit into the interstices of the solvent metal. Introduction: • Impact of thermal vibration on the formation of vacancies in metals and alloys Introduction: • Simulation of vacancy diffusion mechanism in substitutional solid solutions Introduction: • Self-diffusion in pure metals Label some atoms After some time C C A D A D B B No driving force exists, but atoms still diffuse. However, the overall movement of atoms is zero, as the flux of atoms is the same in every direction. Introduction: • Simulation of downhill diffusion The direction of flow of atoms is opposite the vacancy flow direction. Introduction: • Interdiffusion (impurity or solute diffusion) in metals and alloys Different atomic species generally have a different probability of jumping into vacant lattice sites. This gives rise to a number of interesting and important phenomena, which are considered in further detail. These effects are illustrated by means of a diffusion couple between initially pure A and pure B. Initially After some time Adapted from Figs. 5.1 and 5.2, Callister 7e. Introduction: Modelling (mathematics) of diffusion Flux of diffusion The flux of diffusion, J, is used to quantify how fast diffusion occurs in non-steady conditions , i.e. when the concentration gradient and the local concentration profile are changing with time. The flux of diffusion is defined as the number of atoms crossing a unit cross-section area per unit time (atoms/m2-second). J = natom / A t [atom m-2 s-1] where natom is the number of atoms crossing the area A during the time t. Introduction: Modelling (mathematics) of diffusion Rate of diffusion The rate of diffusion, ∂n/∂t, is used to quantify how fast diffusion occurs in non-steady conditions, i.e. when the concentration gradient and the local concentration profile are changing with time. Its measures the local rate of change of composition with time. The rate of diffusion is defined as the number of atoms crossing a cross-section area per unit time (atoms/second). ∂n/∂t = natom/t [atom s-1] where natom is the number of atoms diffusing through the area A during time t. Introduction: Modelling (mathematics) of diffusion Steady-state diffusion (Fick’s first law) Flux of diffusion J - the flux of diffusion does not change with time. The J equation does not explicitly consider the change of concentration with time, the concentration remains constant at each point, or the significant parameters do not change with time. Non-steady-state diffusion (Fick’s second law) Rate of diffusion ∂n/∂t - the flux of diffusion do change with time. Approach Solutions to Fick's laws are stated where applicable, but the mathematical derivations (most of which require a knowledge of partial differential equations) are not given. Validity The phenomenological description based on the Fick’s laws is valid for any atomic mechanism of diffusion. Understanding of the atomic mechanisms is important, however, for predicting the dependence of the atomic mobility (and, then, diffusion coefficient) on the type of interatomic bonding, temperature, and microstructure. Introduction: Modelling (mathematics) of diffusion Factors that influence diffusion: Crystal structure If a crystal structure is distorted either by elastic strains or by extensive plastic deformation, the rate of diffusion is usually increased. Another effect of crystal structure is the variation of the diffusion coefficient with crystal direction in a single crystal of the solvent metal .Such anisotropy is nearly or completely absent in cubic metals, but bismuth (rhombohedral lattice) shows a ratio of about one thousand in its self-diffusion constants measured parallel and perpendicular to the c-axis) Diffusing species, impurities The presence of small amounts of additional metals usually has a relatively small effect on the diffusion of solute atoms in a solvent metal (solvent-impurity and solute -impurity interactions). Frequency factor D0, and activation energy Q are different for every solute, solvent pair. Diffusion mechanism - Diffusion by interstitial mechanism is usually faster than by substitutional or vacancy mechanism Concentration Temperature - Diffusion rate increases very rapidly with increasing temperature Microstructure (Grain size and short-circuit paths of diffusion ). Diffusion is faster in polycrystalline materials compared to single crystals because of the accelerated diffusion along grain boundaries. 1. Diffusion in an ideal solution, Fick’s first law of diffusion From experimental, diffusion in FCC, BCC, HCP materials occur by vacancy mechanism Random jumps, same probability of jumps – jump rate not function of concentration. A as solute, B as solvent, Solid solution A-B with variation of [A] along the length. [A]right > [A]left However the [A] is constant throughout the cross-section. 1. Diffusion in an ideal solution, Fick’s first law of diffusion Difference of [A] between 2 adjacent transverse atomic planes, in [atom m-4] Sign “- “ means downhill diffusion, either of A or B atoms The key point is made that the flux of diffusion is proportional to the concentration gradient, driving force of diffusion. Coefficient of diffusion or diffusivity, D, in [m2s-1] Fick’s 1st law of diffusion Flux of diffusion of A atoms in [atoms m-2 s-1] 1. Diffusion in an ideal solution, Fick’s first law of diffusion Unequal jump probability and vacancy drift It is shown that if the jump probability of solute A and solvent B are different, the flux across a given lattice plane will also be different - i.e. A atoms move into the B-rich side at a different rate to that of B atoms into the A-rich side of the couple. DA ≠ DB → JA ≠ JB This imbalance in atomic movement must be offset by an equal movement, or drift of vacancies in the opposite direction. Vacancy creation and annihilation As a result of such a drift, the vacancy concentrations on either side of the couple will depart from equilibrium - there will be an excess of vacancies on one side and a depletion on the other. In order to maintain the concentrations at or near equilibrium, vacancies are destroyed or created at various sinks and sources (such as dislocations, grain boundaries, etc.) 1. Diffusion in an ideal solution, Fick’s first law of diffusion 1. Diffusion in an ideal solution, Fick’s first law of diffusion 1. Diffusion in an ideal solution, Fick’s first law of diffusion 1. Diffusion in an ideal solution, Fick’s first law of diffusion An important result of vacancy creation at an edge dislocation is the extension of the half-plane. Similarly, vacancies sinking into a dislocation cause it to contract. Whole lattice planes can therefore be created or destroyed by vacancy movement, and this gives rise to yet another important effect, that of lattice drift. 1. Diffusion in an ideal solution, Fick’s first law of diffusion Diffusivity D in substitutional solid solutions is a function of T°and composition n, D = f (T°, n) This is largely due to the variation in vacancy concentration, Nv with composition and T°. 1. Diffusion in an ideal solution, Fick’s first law of diffusion In general, the metal with the higher melting T° will contain fewer defects. It is shown that if the two are mixed to form a solid solution, the vacancy concentration for the resulting alloy may take an intermediate value (image of the interdiffusion coefficient D, a weighted average of the diffusion coefficients DA and DB with respect to alloy composition), which will depend on the composition. Since the diffusivity of each species is proportional to Xv, it follows that DA, DB and the interdiffusion coefficient, D, will be composition-dependent. 2. Diffusion in an ideal solution, Fick’s second law of diffusion In many real situations the concentration gradient and the local concentration profile are changing with time. The changes of the local concentration profile can be described in this case by a differential equation, Fick’s second law. Fick’s 2nd law of diffusion 2. Diffusion in an ideal solution, Fick’s second law of diffusion Solution of the Fick’s 2nd law of diffusion is the local concentration profile nA (x,t): Fick’s second law relates the rate of diffusion ∂nA/∂t (local rate of change of composition with time) to the curvature ∂2nA/∂x2 of the local concentration profile nA (x,t) ∂2nA/∂x2 > 0 , ∂nA/∂t > 0 ∂2nA/∂x2 <0 , ∂nA/∂t <0 The Fick’s 2nd law of diffusion is basically saying : if the local concentration profile is a maximum, it will reduce with time, wheareas if the profile is a minimum, it will increase with time Composition nA decays with distance x - An initial step concentration will evolve into a sigmoidal distribution after a diffusion annealing - The longer the diffusion annealing, or the higher the temperature, the further the profile will penetrate. 3. The Kirkendall effect •In binary solid solutions, each of the 2 atomic forms can move with a different velocity. •Gradient of concentration •Fine wires in refractory material in the plan of weld serve as markers. •Assumption: metals are pure at the start •Heating to T°close to melting point of A and B •Holding for a relative long time at the latter T°, e g for 2 days. •Subsequent cooling at the ambient T° •Flow of A atoms from right to left •Flow of B atoms from left to right •Interesting result: the wires “move” during the process. 3. The Kirkendall effect Movement of wires to the right for distance x. The only way to explain the movement of wires is to point out that A atoms diffuse faster than B atoms. Kirkendall effect = confirmation vacancy diffusion mechanism. Displacement of Kirkendall markers Ξ lattice drift of 4. Porosity The element with the low melting point diffuses faster. eg Cu and Ni, Cu diffusion is faster The right side lost more atoms that it gain, then shrinkage at the right and expansion at the left. Undersaturation at the right and supersaturation at the left, tensile stress at the right and compressive stress at the left. Grain boundaries and exterior surfaces are probable positions for both creation and elimination of vacancies. 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions The Kirkendall effect shows that in a diffusion couple composed of two metals, the atoms of the components move at different rates, and the flux of atoms through a cross-section defined by markers is not the same for both atom forms. It is, thus, more logical to think in terms of diffusivities, DA and DB corresponding to the movement of the A and B atoms respectively. These quantities may be defined by the following relationships. DA and DB are function of composition and then of position along diffusion couple. These quantities are valuable because they measure the speed with which the individual atomic forms move during diffusion Darken’s equations make it possible to determine the intrinsic diffusivities experimentally. Assumptions: •volume contraction and volume expansion only in the length direction •Cross-section area = constant •Diffusion zone <<<<<<< diffusion couple •Porosity does not occur. 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions Velocity of Kirkendall markers v - Darken’s first equation 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions Velocity of Kirkendall markers Ξ lattice drift velocity 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions Interdiffusion coefficient D - Darken’s second equation v and ∂nA/∂x could be determined experimentally, but a second equation is compulsory so that to solve the 2 unknowns DA and DB Volume mmnn = 1. dx Rate at which the number of A atoms changes inside mmnn = Difference in the number of A atoms jumping into and out mmnn per second Or difference in flux of A atoms across the surface at x and x+dx 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions Lattice drift velocity v will affect the net flux of diffusion, JA, Net flux of diffusion leads to the → Net flux = flux due to concentration gradient + flux due to lattice drift derivation of Darken’s equations, whereby the net flux of diffusion of A solute atoms is given by: Similarly, the net rate of diffusion of A solute atoms is given by: 5. Darken’s equations as essentially Fick’s laws of diffusion in substitutional solid solutions Interdiffusion coefficient Darken’s equations 6. Solutions to the Fick’s laws of diffusion Fick's two laws of diffusion are differential equations and must be solved to obtain an algebraic equation describing the variation of concentration nA as a function of distance x and (usually) of time t. The ease of solution depends on such factors as the geometry of the specimen in question, the initial conditions of concentration in the specimen, and the "boundary conditions'" [for example, whether diffusion penetrates the entire (finite) specimen or affects only a portion of an (infinite) specimen]. The solutions can be listed in three categories: (1) Those that can be obtained easily by elementary manipulations of the diffusion equations. A majority of the diffusion problems encountered by an engineer can be simplified to permit at least an order-ofmagnitude solution in this manner. (2) A variety of technically important diffusion problems have been solved by more advanced mathematical techniques, and the solutions conveniently collected in reference volumes. Solutions of Fick's second law for cylinders and spheres are examples. (3) Diffusion problems of many types, such as those involving concentration-dependent diffusion coefficients, are solved most conveniently using a digital computer (or perhaps an analog computer). Examples can be found in technical journals, but usually each special problem must be solved individually. 7. Solutions to the Fick’s first law of diffusion A majority of the diffusion problems encountered by an engineer can be simplified to permit at least an order-ofmagnitude solution in this manner. Diffusion of hydrogen through metals is an especially simple example of steady-state diffusion . Oxidation of metals is another example of steady-state diffusion. The treatment of oxidation in the following section illustrates a typical procedure. 7. Solutions to the Fick’s first law of diffusion • It is assumed that diffusion of only the metal atoms need be considered. Experimental data on the high-temperature oxidation of metals frequently show a parabolic relation between the time of oxidation and the amount of oxide formed. This behavior can be explained in terms of a diffusion process. Furthermore, the difference between the metal concentration NA2 at the oxide surface and the concentration NA1 at the metal surface has a constant value ∆NA. Therefore, at any thickness x of the oxide layer the flux of diffusion of metal atoms (rate of transfer of metal atoms per unit area) is: J = - D (∆NA/x) = dm/dt 7. Solutions to the Fick’s first law of diffusion However, dm and the increase in thickness, dx, are proportional: that is, dx =K dm. Therefore : dx = - D’ (∆NA/x) dt where D’ =K D and is a constant when D is assumed to be constant. This equation can be solved as follows: → x2 = 2 D’ ∆NA t x2 = K’ t K' is positive, since the minus sign cancels the minus sign of the concentration difference ∆NA. The thickness of the oxide layer increases as the square root of the time of oxidation. 8. Solutions to the Fick’s second law of diffusion – Grube method 2nd order partial differential equation Grube method assumption: Diffusivity D very slightly variable, in other words D is constant. However, the Grube method may be applied to diffusion in alloy systems where the diffusivity varies moderately with composition if the two halves of a couple are made of alloys differing slightly in composition. A solution, NA = f (x, t) of this equation must satisfy both the 2nd order partial differential equation and also the initial and boundary conditions (for space and time) in question. Examples of solutions to Fick's 2nd law: Infinite solid, : Diffusion across a couple, Thin-film solution, thin or finite source: Deposit of a thin layer of metal A at the end of a long bar of metal B, or in sandwich in between two long bars of metal B Semi-infinite solid, constant concentration - infinite source : Addition of A atoms from a gas phase (diffusion across a semi-couple), e.g. carburisation or case hardening Removal of A atoms via a gas phase (diffusion across a semi-couple), e.g. decarburisation Homogenisation (solute equilibration) Elimination of micro segregation in castings, dissolution of intermetallic compounds 8. Solutions to the Fick’s second law of diffusion – Grube method Note that derivations for these solutions are not provided. The solutions given are as follows: 8. Solutions to the Fick’s second law of diffusion – Grube method 8. Solutions to the Fick’s second law of diffusion – Grube method 8. Solutions to the Fick’s second law of diffusion – Grube method Error function or probability integral function erf (y) erf (y) becomes negative when y [or x in x/2(Dt)1/2 ] is negative 8. Solutions to the Fick’s second law of diffusion – Grube method More accurate values of erf (y) can be obtained from books of standard mathematical functions 8. Solutions to the Fick’s second law of diffusion – Grube method Diffusion distance xd Since erf (0.5) ≅ 0.5 the depth at which NA is midway between NA1 and NA2 is given by [x/2 (Dt)1/2] ≅ 0.5 , that is x ≅ (Dt)1/2 Diffusion distance = xd ≅ (Dt)1/2 Clearly there is no such thing as a “fixed “diffusion distance” The diffusion distance xd is just an estimation of the scale of diffusion, i.e. an estimation of the diffusion effects. The diffusion distance xd is yd = xd /2 (Dt)1/2≅ 0.5 approximately the distance over which the solute will have fallen to half its initial value. yd 8. Solutions to the Fick’s second law of diffusion – Grube method 8. Solutions to the Fick’s second law of diffusion – Grube method Theoretical penetration curve Distance versus composition curve obtained when the solution of Fick's 2nd equation NA is plotted as a function of the variable y = x/2 (Dt)1/2 Thus, if a diffusion couple has been maintained at some fixed temperature for a given period of time (t) so that diffusion can occur, then a single determination of the composition at an arbitrary distance (x) from the weld permits the determination of the diffusivity D 8. Solutions to the Fick’s second law of diffusion – Grube method Thus, suppose that a diffusion couple is formed by the welding together of two alloys of the elements A and B having the compositions NA1, 40 % A (the alloy to the right of the weld), and NA2, 50 % A (left of the weld). Let the couple be heated quickly to some temperature T1 and held there 40 hr (144 000 sec), and let it be assumed that after cooling to room temperature, chemical analysis shows that at a distance 2 x 10-3 m to the right of the weld the composition NA is 42.5 % A We have on substituting the assumed data in Eq. 12.20 8. Solutions to the Fick’s second law of diffusion – Grube method Relationship between distance and time corresponding to the same composition NA during isothermal diffusion Since the composition NA remains the same, this problem requires that the argument of the probability integral has the same value as in the last example. With the assumption of a constant diffusivity (D), that is: In the numerical computation just considered, a composition NA = 42.5 % A was assumed at a distance of 2 x 10-3 m from the weld when the diffusion time was 40 hr. Compute the length of time needed to obtain the same composition, 42.5 % A, at twice the distance from the weld interface. 9. Solutions to the Fick’s second law of diffusion – Matano- Boltzmann method Assuming that : Interdiffusion coefficient D is function of composition D = f (NA), Diffusion process does not change the composition at the ends of the diffusion couple as boundaries conditions, Number of atoms per unit volume (nA + nB) is constant and, No porosity forms at the interface. Boltzmann proposed the following solution to the Fick’s 2nd law of diffusion Analytical evaluation of the derivative and integral in Equation 12.24 are much more difficult. As a consequence, the Matano-Boltzmann method of determining the interdiffusion coefficient D uses graphical derivation and integration to find out the different factors in Equation 12.24 9. Solutions to the Fick’s second law of diffusion – The Matano- Boltzmann method A ‘pure’ diffusion couple is created and annealed at a constant temperature for a given length of time. After removal from the furnace, a concentration profile is generated. The first step, after the diffusion anneal and chemical analysis of the specimen, is to plot a curve of concentration versus distance along the bar measured from a suitable point of reference, say from one end of the couple. From this, the Matano interface is defined as being the plane across which an equal number of atoms have crossed in both directions. The second step is to determine that cross-section of the bar through which there have been equal total fluxes of the two atomic forms (A and B). This cross-section is known as the Matano interface and lies at the position where areas M and N in Fig. 12.13 are equal. The position of the Matano interface is determined by graphical integration, but, in general, it has also been experimentally determined that, in the absence of porosity, the Matano interface lies at the position of the original weld. 9. Solutions to the Fick’s second law of diffusion – The Matano- Boltzmann method Once the Matano interface has been located, it serves as the origin of the x coordinate. Distances to the right of the interface are considered positive, while those to the left of it are negative. With the coordinate system thus defined, the Boltzmann solution of Fick's equation is The integral term, is the area between the penetration curve and Matano interface, whilst ∂x/∂NA is the reciprocal of the concentration gradient (slope of the penetration curve ) at NA. With the aid of concentration profile: The derivative ∂x/∂NA is evaluated by drawing a tangent to the concentration profile at NA and measuring its slope. The integral, limits of which are NA1 and NA, is evaluated by a graphical method such as the Simpson's rule. 9. Solutions to the Fick’s second law of diffusion – The Matano- Boltzmann method Table 12.3 represents this concentration-distance data corresponding to no actual alloy system, but it is representative, in a broad general way, of diffusion data obtained in actual experiments. The diffusion couple is assumed to be formed from pure A and pure B. Figure 12.14 shows the penetration curve obtained when the data of Table 12.3 are plotted. 9. Solutions to the Fick’s second law of diffusion – The Matano- Boltzmann method Let us reconsider the Boltzmann solution of Fick's 2nd law: Suppose that we desire to know the interdiffusion coefficient at a particular concentration, which we shall arbitrarily take as NA = 0.375. (point C in Fig. 12.14). Diffusion time is assumed to be 50 hr (180,000 s) ∂x/∂NA = reciprocal of the slope of line E tangent to the penetration curve at point C = 1 /(610 m-1) = (1/610) m Integral (limits of which are NA1 = 0 and NA = 0.375) = cross-hatched area (F) = 4.66 x 10-4 m 9. Solutions to the Fick’s second law of diffusion – The Matano- Boltzmann method Computations similar to the above may be made to determine the interdifussion coefficient D(NA) at any concentration not too close to the terminal compositions (NA = 0 and NA = 1). Since the desired area approaches zero as the composition approaches one of the terminal compositions, the accuracy of the determination falls off as NA becomes very close to either 0 or 1. Large values of D are obtained as one approaches the concentration NA = 1. There is a minimum in this curve in the middle of the concentration range. A diffusivity-concentration curve of this form has been reported for the diffusion of Zr and U. However, the curves of Fig. 12.3 are more typical of the diffusion data reported to date. 10. Experimental determination of the intrinsic diffusivities There has been, to date, very little success in the development of a theory capable of predicting the numerical values of the intrinsic diffusivities starting from a consideration of atomic processes. While it is generally agreed that diffusion in metallic substitutional solid solutions is the result of the movement of vacancies, the factors that control jump rates into vacancies of the two atomic forms are complex and not completely understood. Thus, in our previous derivation of Fick's first law, several simplifying assumptions were made that do not hold for real metallic substitutional solid solutions. First, it was assumed that the solution was ideal, but, as we have seen, most metallic solutions are not ideal, and in a nonideal solution the diffusion rates are influenced by a tendency for like atoms either to group together, or to avoid each other (negative and positive deviation). Second, it was assumed that the rate of jumping was independent of composition, that is, whether the jump was made by an A or a B atom. The assumption that the rate of jumping is independent of the composition is certainly not true, as may be judged by the fact that measured diffusivities vary widely with composition. Experimental determination of intrinsic diffusivities is the viable alternative. 10. Experimental determination of the intrinsic diffusivities The determination of the intrinsic diffusivities will now be illustrated with the use of the assumed data of Table 12.3. First we must derive an expression for the marker velocity v in terms of the marker displacement and the time of diffusion t. Experimentally, it has been determined that the markers move in such a manner that the ratio of their displacement squared to the time of diffusion is a constant. Thus 10. Experimental determination of the intrinsic diffusivities The flux of A atoms through the marker interface from right to left is approximately 1.2 times that of the flux of B atoms moving from. left to right. 10. Experimental determination of the intrinsic diffusivities 11. Self-diffusion in pure metals In self-diffusion studies, one investigates the diffusion of a solute consisting of a radioactive isotope in a solvent that is a nonradioactive isotope of the same metal. In such a system, both atomic forms are identical except for the small mass difference between the isotopes. The principal effect of this mass difference is to cause the solute isotope to vibrate about its rest point in the lattice with a frequency slightly different from that of the solvent isotope, giving the two isotopes a slightly different jump rate. This difference is easily calculated because the vibration frequency is proportional to the reciprocal of the square root of the mass, and since the jump rate into vacancies is proportional to the vibration frequency, we have where 1/T and 1/T* are the jump-rate frequencies of normal and radioactive isotopes respectively (τ and τ* are the mean times of stay of the respective atoms at lattice positions, and m and m* are the masses of the two isotopes [m* radioactive]). 11. Self-diffusion in pure metals Except for the mass difference, solute and solvent are chemically identical and the solid solution is truly ideal. Considerations of the effect of the departure of a solution from ideality may thus be neglected. Furthermore, the mass correction is usually small so that, to a good approximation, we may assume that the intrinsic diffusivity of the radioactive isotope is the same as that of the nonradioactive isotope. When the intrinsic diffusivities are equal, the interdiffusion coefficient equals the intrinsic diffusivities, as may be seen by considering the Darken equation: where D is the interdiffusion coefficient, D = DA = DB, the intrinsic diffusivity of either the radioactive or nonradioactive isotopes, and by the definition of atom fractions (NA + NB) = 1. Because the intrinsic coefficients do not depend on composition, it is also true that the interdiffusion coefficient does not depend on composition. Therefore experimental determinations of self-diffusivities may be made by using the simpler Grube method. 11. Self-diffusion in pure metals Because self-diffusion in pure metals occurs as in an ideal solution and with diffusivity that is independent of concentration, experimentally determined self-diffusion coefficients of pure metals are usually of high accuracy. Measured diffusivities are capable of theoretical interpretation because the diffusion process occurs in a relatively simple system. The assumptions made in our derivation of Fick's first law are those actually observed in self-diffusion experiments, and Eq. 12.7 is correct for self-diffusion in a simple cubic system: i.e., 11. Self-diffusion in pure metals Frequency of jumps of an atom in a pure metal crystal The average thermal energy of an atom (kBT = 0.026 eV for room temperature) is usually much smaller that the activation energy ∆Hm (~ 1 eV/atom) and a large fluctuation in energy is needed for a jump ra = Frequency of jumps of an atom = Number of successful jumps made per second by an atom ra = Jump rate of atoms into vacancies = Number of jumps made per second by an atom ra = 1/mean time of stay of an atom at a lattice site = 1/τ ra = (Attempted jumps) x (Probability of a successful jump) Attempted jumps = attempt frequency = frequency or number of times per second that an atom moves toward a vacancy = lattice vibration frequency (temperature-independent) = ν Most of the time the site adjacent to an atom will not be vacant and the jump will not be possible. Probability of a successful jump = (Probability that an adjacent site is vacant) x (Probability that any attempt at jumping will be successful) Probability that an adjacent site is vacant = (Lattice coordination number ) x (Probability that any one site is vacant) Lattice coordination number = number of nearest neighbours = Z Probability that any one site is vacant = molar fraction of vacancies in the metal = Nv = exp (-∆Hf/RT) ∆Hf = enthalpy change or work to form a mole of vacancies Probability that an adjacent site is vacant = Z exp (-∆Hf/RT) 11. Self-diffusion in pure metals Probability that any attempt at jumping will be successful = probability that an atom will have sufficient energy to make a jump = exp (-∆Hm/RT) ∆Hm = enthalpy change or energy barrier that must be overcame to move a mole of atoms into a mole of vacancies Probability that any attempt at jumping will be successful = exp (-∆Hm/RT) Probability of a successful jump = Z exp (-∆Hf/RT) exp (-∆Hm/RT) ra = ν Z exp (-∆Hf/RT) exp (-∆Hm/RT) ra = Z ν exp [-(∆Hf+∆Hm)/RT] ra = 1/τ 1/τ = Z ν exp [-(∆Hf +∆Hm)/RT] 12.31 Equation 12.31 neglects entropy changes associated with the formation and movement of vacancies and should be more correctly written 11. Self-diffusion in pure metals Coefficient of self-diffusion D where ∆Gm and ∆Gf are the free-energy changes associated with the movement and formation of vacancies respectively. where ∆Sm is the entropy change per mole resulting from the strain of the lattice during the jumps, and ∆Sf the increase in entropy of the lattice due to the introduction of a mole of vacancies. In the BCC lattice, α is 1/8, while Z is 8, and, similarly, in the FCC lattice, α is 1/12, while Z is 12 so that for both forms of cubic crystals This expression will be discussed further when we consider the temperature dependence of the diffusion coefficient. For the present, it suffices to point out that considerable success has been obtained in the theoretical interpretation of the various factors in this equation. 11. Self-diffusion in pure metals Self-diffusion in pure metals is essentially viewed as a vacancy diffusion, involving ∆Hf >0. Atoms are only able to move when they are adjacent to one or more vacancies. Temperature plays a significant role in the rate of diffusion, as it alters the equilibrium concentration of vacancies, and probability of a successful jump into a neighbouring site. Self-diffusion in pure metals could be viewed as interstitial diffusion by considering ∆Hf = 0. Indeed, interstitial diffusion is nearly always self-interstitial diffusion. In other words, since the formation of a vacant site is not needed for interstitial atoms, interstitial atoms are nearly always free to move. How does the diffusion of vacancies in a crystal relates to that of atoms? Since vacancies are always free to jump, their jump frequency is much greater than that for atoms. It is shown that the diffusivities of vacancies, Dv and atoms DA are related by the expression: Dv = DA /Nv where Nv is the molar fraction of vacancies in the metal 12. Temperature dependence of the diffusion coefficient It has already been seen that the diffusion coefficient is a function of composition. It is also a function of temperature. The nature of this temperature dependence is shown clearly in the equation for the self-diffusion coefficient stated in the previous section, Temperature dependence of the diffusion coefficient D follows the Arrhenius dependence. In this form, the equation can be applied directly to the study of experimental data. 12. Temperature dependence of the diffusion coefficient For interstitial diffusion Q = QID = ∆Hm = enthalpy of interstitial atom migration For vacancy diffusion Q = QSD =∆Hf + ∆Hm = enthalpy of formation of vacancies + enthalpy of atom migration Since the formation of a vacant site is not needed for interstitial atoms, Qinterstitial << Qsubstitional Hence Dinterstitial >> Dsubstitional. Comparison of diffusion distances in vacancy and interstitial diffusion points out that xinterstitial >> xsubstitional It is observed that for materials of a given crystal structure and bond type, the values of QSD are roughly proportional to the absolute melting temperature. 12. Temperature dependence of the diffusion coefficient The slope of the experimentally determined straight line determines the activation energy Q since m = - Q/2.3 R or Q = -2.3 R m. At the same time, the intercept of the line with the ordinate designated by b yields the frequency factor D0, since b = log D0 or D0 = 10b. 12. Temperature dependence of the diffusion coefficient The above method of determining experimental activation energies and frequency factors can be illustrated with the use of some representative data given in Table 12.4. 12. Temperature dependence of the diffusion coefficient The experimentally determined equation for the self-diffusion coefficient is, accordingly, 12. Temperature dependence of the diffusion coefficient The preceding discussion has been concerned only with the temperature dependence of self-diffusion coefficients. However, experimentally determined values of chemical interdiffusion coefficients D, and of their component intrinsic diffusivities DA and DB, also show the same form of temperature dependence. Speaking in general, all diffusion coefficients tend to follow an empirical activation law, so that we have for self-diffusion 13. Chemical diffusion at low solute concentration 14. The study of chemical diffusion using radioactive tracers* 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces Experimental measurements have shown that the surface and grain-boundary forms of diffusion also obey activation, or Arhennius-type laws. Diffusion is, more rapid along grain boundaries than in the interiors of crystals - free-surface diffusion rates are larger than either of the other two. Because the regularity of the crystal lattice of a metal is disturbed at a grain boundary, near a free surface, and adjacent to a dislocation line, diffusion by the vacancy mechanism is greatly enhanced in these regions. Grain boundaries : open structures containing voids, dislocations and possessing already a substantial level of distortions Both the number of vacancies and their mobility may be larger as a result of the local disruption of crystalline regularity. 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces The frequency factor in the grain-boundaries is less than in the lattice. D0b < D0l Energy barrier of diffusion in grain boundaries to be lower because lower strain of the lattice is needed for diffusion to take place, the grain boundary possessing already a substantial level of distortions whose free energy provide for the activation energy of diffusion. For this reason the activation energy Q, is less for grain-boundary diffusion than for volume or lattice diffusion. Qb < Ql Q is still lower for surface diffusion and for “pipe” diffusion along grain boundaries. This difference in Q values explains why short-circuit diffusion plays a significant role only at diffusion temperatures below about three-quarters of the absolute melting point, Tf °K. 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces Because of the very rapid movements of atoms on free surfaces, surface diffusion plays an important role in a large number of metallurgical phenomena. However, grain-boundary diffusion is of more immediate concern because, in the average metallic specimen, the grain-boundary area is many times larger than the surface area. Furthermore, grain boundaries form a network that passes through the entire specimen. It is this latter property that often causes large errors to appear in the measurement of lattice diffusion, coefficients. When the diffusivity of a metal is measured with polycrystalline samples, the results are usually representative of the combined effect of volume and grain-boundary diffusion. What is obtained, therefore, is an apparent diffusivity, Dapp, which may not correspond to either Dl or Db. However, under certain conditions, Db may be small, so that Dapp = Dl . On the other hand, if the conditions are right, Db may be so large that Dapp diverges considerably from Dl. Let us investigate these conditions. 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces Diffusion in a polycrystalline specimen cannot be described as a simple summation of diffusion through the crystals and along the boundaries. Diffusion in the boundaries tends to progress more rapidly than diffusion through the crystals, but this effect is counteracted because as the concentration of solute atoms builds up -in the boundaries, a steady loss of atoms occurs from the boundaries into the metal on either side of the boundary. The nature of this process can be visualized with the aid of Fig. 12.21, which represents a diffusion couple composed of two pure metals A and B. Short arrows represent the nature of the movement of A atoms into the B matrix. Parallel arrows perpendicular to the weld interface represents the volume component of diffusion. Arrows parallel to the grain boundaries indicate the movement of atoms along boundaries, and Arrows perpendicular to the boundaries represent the diffusion from the boundaries into the crystals. The smaller the grain size d, the greater the total grain-boundary area available for boundary diffusion and, therefore, the greater the importance of boundaries in the diffusion process. 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces The importance of grain-boundarydiffusion phenomena in diffusion measurements is also a function of temperature. (Fig. 12.22 12.23) Single crystal specimens Polycrystalline specimens (grain size d = 35 µm before diffusion anneal) Lattice or diffusion when compared with boundary diffusion, is more sensitive to temperature change. Lattice diffusion and Total or apparent diffusivity Dapp grain-boundary diffusion different temperature dependence. have 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces Only a small fraction of the cross section Single crystal specimens Polycrystalline specimens (grain size d = 35 µm before diffusion anneal) of a typical metal is employed by a given short-circuit mechanism. For grain- boundary diffusion the fraction is about 10-5 . Therefore, the ratio Db/Dl must approach 105 before the amount of material transported along the grain boundaries is comparable to that diffusing in the volume of the grains. At high temperatures the ratio is less than this value, but with decrease in temperature Db decreases less than Dl because of the difference in their Q values. The apparent diffusion coefficient, Dapp begins to deviate from Dl at a sufficiently low temperature because of the increasing contribution of grain-boundary diffusion. At still lower temperatures, where both volume diffusion and grain-boundary diffusion may be ineffective, significant atomic transport can occur along the "pipe'" of disturbed lattice surrounding a dislocation line. 15. Short-circuit paths of diffusion, diffusion along grain boundaries and free surfaces • Thus as the temperature is raised, the rate of diffusion through the lattice increases more rapidly than the rate of diffusion along the grain boundaries. • Conversely, as the temperature is lowered, the rate of diffusion along the grain boundaries decreases less rapidly. • The net effect is that at very high temperatures diffusion through the lattice tends to overpower the grain boundary component, but at low temperatures diffusion at the grain-boundary component becomes more and more important in determining the total or apparent diffusivity Dapp. • • Diffusivities determined with polycrystalline specimens are more liable to be representative of lattice diffusion if they are measured at high temperatures. • The reliability of the data can be increased by controlling the grain size of the specimens. The larger the grain size, the smaller the grain-boundary contribution to the diffusivity. • Thus for accurate measurements of lattice diffusivities, using polycrystalline specimens, high temperatures and large-grained specimens should be used. 16. Fick’s first law in terms of mobility and an effective force 16. Fick’s first law in terms of mobility and an effective force 16. Fick’s first law in terms of mobility and an effective force 17. Diffusion in non-isomorphic systems 17. Diffusion in non-isomorphic systems 17. Diffusion in non-isomorphic systems 17. Diffusion in non-isomorphic systems 17. Diffusion in non-isomorphic systems 18. Application of diffusion concept to the homogenization of castings It is often of interest to be able to calculate the time taken for an inhomogeneous alloy to reach complete homogeneity, as for example in the elimination of segregation in castings. 18. Application of diffusion concept to the homogenization of castings 18. Application of diffusion concept to the homogenization of castings 18. Application of diffusion concept to the homogenization of castings Summary • Make sure you understand language and concepts: – – – – – – – – – – – – – – – Diffusion Driving force of diffusion and concentration gradient Energy barrier and activation energy Self-diffusion Interdiffusion Diffusion coefficient Vacancy diffusion Interstitial diffusion Flux of diffusion Rate of diffusion Fick’s first and second laws Steady-state diffusion Nonsteady-state diffusion Solutions to Fick’s laws of diffusion Temperature dependence of diffusion coefficient D