You are on page 1of 45

Nukes

I study nuclear science I love my classes I got a crazy teacher He wears dark glasses

-Timbuk3

2.1 Review of crystalline defects

One of the first topics discussed in any curriculum of materials science is defects in crystals. If the properties of crystalline imperfections are firmly established, why then are we starting a chapter on nuclear materials with a review of point, line and planar defects? As we shall see in subsequent sections, the nuclear reactor environment generates a large number of defects in materials, many of which are not typically seen in normal applications. The production, movement and interaction of the irradiation induced defects leads to many deleterious materials properties such as embrittlement and swelling. In order to understand the behavior of materials under irradiation conditions a discussion of some of these unusual defects is in order.

The best place to start is with a vacancy, which is simply a lattice site that is missing an atom. The process of removing an atom from a lattice site and allowing neighboring atoms to relax slightly toward the vacancy requires an energy of roughly 1eV. As we have seen in our discussion of band gaps in semiconductor materials, an energy of 1eV implies that some vacancies can be generated in a crystal from thermally activated processes. In fact , much like the probability of an electron hopping to the conduction band, we can write down the equilibrium concentration of vacancies as:

Ef cv = exp v kT

2.1

where Evf is the vacancy formation energy and the concentration is defined as the number of vacancies divided by the total number of lattice sites. Equation 2.1 suggests that the number of vacancies depends very strongly on temperature and for typical metals at the melting point cv is on the order of 10-6. In the subsequent discussion of materials under irradiation one more aspect of a vacancy is worth mentioning and that is: it can move around. Through thermal vibrations an atom adjacent to the vacancy can acquire, occasionally, sufficient energy to exchange places with the vacant site. Indeed, the hopping of vacancies is the mechanism responsible for diffusion in substitutional alloys.

Next up is the interstitial. Interstitials are usually introduced through an illustrative example of some small atom size species dissolved into the lattice of a larger host atom; C in Fe being perhaps the most popular example. Figure 2.1 shows the position of the interstitial C atom in the FCC (austenite) and BCC (ferrite) phases of Fe. Because the Fe atoms surrounding the C form an octahedron the position is referred to as the octahedral site. The name interstitial is derived from the fact that the octahedral site is not a lattice point of the Fe lattice, ie. the C sits in the intersticies.

Fig. 2.1 Interstitial octahedral positions for the FCC structure (left) and a BCC crystal (right). Carbon in the austenite and ferrite phases of steel are examples of an interstitial species in FCC and BCC materials.

The atomic diameter of an Fe atom is approximately 2.48A, whereas the diameter of a C atom is much smaller, roughly 1.5A. Now consider the BCC octahedral site and assume that the Fe atoms can be treated as hard spheres. In the z direction the distance between the surface of an Fe atom at the top of the octahedron and surface of the Fe atom at the bottom is .39A. On the other hand the opening between two Fe atoms along the 110 face diagonal is 1.57A. Overall the C is a tight squeeze in the z direction but is a comfortable fit in the x-y plane. Since the atomic size of C is much smaller than Fe, solubilities can be fairly high. In most instances of interstitial alloys a large size mismatch is found. N in Fe and H in Pd are other examples.

Now consider an Fe atom occupying an interstitial site in Fe, which is a defect known as a self interstitial atom and is frequently abbreviated as SIA. The interstitial atom is equivalent in size to the host (obviously), the defect is accompanied by a large local distortion of the lattice and the energy of forming a SIA is correspondingly high, in fact
Ei f >2eV. Application of Eq. 2.1 to the case of a high energy SIA reveals that the

equilibrium concentration of this defect is very low. So low in fact that SIAs are not even worth mentioning in normal materials applications.

The self interstitial is so large that, in its stable configuration, it doesnt actually occupy a specific site like the octahedral position. Instead the SIA pairs up with a host atom to form what is known as a dumbbell. An example of a dumbbell in the case of a BCC lattice is shown in Fig. 2.2. Here the atom at the body center pairs up with the self interstitial and two atoms align along the 110 crystallographic direction. From the above discussion on the relative space available for a C atom in Fe, it should be clear why the dumbbell would prefer the 110 alignment. Also in Fig. 2.2 is the dumbbell SIA geometry for an FCC metal, where now the low energy alignment is along (100).

Fig 2.2. For self interstitial atoms dumbbell defects are formed. The geometry of a dumbbell is depicted for an FCC (left) and BCC (right) crystal. As with vacancies, SIAs are mobile defects. In fact, and this may seem surprising, the self interstitial is much more mobile than a vacancy. As will be discussed below, the difference in transport properties of the vacancies and interstitials has important consequences on the long term behavior of irradiated materials.

Vacancies and SIAs are both point defects. By contrast a dislocation is a disruption in the crystalline order along a line in the material and thus is referred to as a line defect. The dislocation can be understood by a cut-distort-repair thought experiment. For example, the dislocation pictured on the top in Fig. 2.3 can be constructed by first cutting the crystal along a line (using an extremely sharp nanoscale knife), inserting an extra plane of atoms into the slice and then allowing all the atoms to relax. The result is an edge dislocation. Alternatively, the dislocation depicted on the bottom is formed by again cutting, but this time shearing along the line of the cut by one crystalline repeat distance and then relaxing the structure. This construction creates what is referred to as a screw dislocation.

Fig. 2.3. Illustration of an edge (top) and screw (bottom) dislocation.

Rather than resorting to visualizations, the character of a dislocation can be described in a more quantitative way by the relationship between the line direction, out of the page for the edge dislocation in Fig. 2.3, and what is known as the Burgers vector, b. The determination b is found from a right-hand-finish-start (RHFS) circuit around the dislocation, as depicted in Fig. 2.4. Rather than illustrating the RHFS circuit for a schematic diagram we are demonstrating the construction for a real live material, in this case a NiPt alloy. The image in Fig. 2.4 is taken from a scanning tunneling microscopy

experiment from the surface of NiPt which has been threaded by an edge dislocation. The circuit starts at some atom located down and to the left of the dislocation line and proceeds in some number of lattice sites, 9 here, up followed by 11 to the right (the RH in the RHFS notation). The circuit is completed by 9 atomic spacings down and 11 to the left. The disruption in the perfect crystalline lattice is reflected in the fact that the start of the circuit does not coincide with the finish. The vector connecting the end point with the starting point (the FS part) is b. From the definition of b it is fairly straightforward to understand the difference between an edge and screw dislocation. For an edge, b is perpendicular to the line direction, which is out of the page in Fig. 2.4, and for a screw they are parallel.

Fig. 2.4. Illustration of the RHSF burgers circuit from an scanning tunneling microscopy image of the surface of a PtNi alloy. From www.iap.tuwien.ac.at/.../dislocations.html

Edge and screw dislocations are easy, more or less, to visualize and to draw. However, by discussing only these two examples students are often left with the impression that

screw and edge are the only types of dislocations. This is not true. In general any portion of a dislocation line may have a mixed character, consisting of both edge-like and screwlike components. A general dislocation is much more difficult to draw but a good attempt is shown in Fig. 2.5.

Dislocations cannot simply terminate in the middle of a crystal. They must end at surfaces, grain boundaries or interphase boundaries. Alternatively, dislocations can terminate on themselves creating dislocation loops. Again using our cut and paste procedure we can describe the dislocation pictured on the left in Fig. 2.6 as follows. Cut a circle in the otherwise perfect crystal, displace the material above the circle by a Burgers vector located in the plane of the cut and allow all the atoms to relax. The shear

Fig. 2.4. A dislocation is, in general, of mixed screw and edge character. In this schematic of a curved dislocation the edge portion is on the upper right and the screw segment is at the bottom.

loop formed will have portions where the Burgers vector runs along the line direction, that is screw character, and some portions that are of edge character. A different means of creating a dislocation loop is depicted on the right of Fig. 2.6. Here a circle is cut, but now an extra disk of atoms is placed in the cut. This second loop is known as a Frank loop and from a RHSF analysis at any point along the dislocation line it can be concluded

that the loop is everywhere of edge character. Notice, in constructing the Frank loop we dont have to insert a coin shaped region of atoms, we can instead remove a disk (or insert a disk of vacancies). In this case the Burgers vector has flipped directions, but the loop is still an edge loop.

Shear Extra plane of atoms or vacancies

Fig. 2.6. Dislocations loops in a crystalline material. The shear loop exhibits a Burgers vector lying within the plane of the loop and thus is of mixed screw and edge character. The Frank loop (right) can be viewed as an extra disk of atoms or vacancies. For the Frank loop the Burgers vector is always perpendicular to the line direction.

Dislocations are of primary importance in materials science because they control the process of plastic deformation in metals. However, another aspect of dislocations is central to the discussion of nuclear materials. Dislocations act as sinks for point defects. Imagine a vacancy hopping around on a lattice. Eventually the vacancy may migrate to the core of an edge dislocation (see Fig. 2.3). When the vacancy attaches to the extra half plane of atoms it simply disappears and, in a process known as climb, the dislocation line moves upward in Fig. 2.3. Therefore, if the concentration of vacancies exceeds the equilibrium value predicted by Eq. 2.1, then annihilation of vacancies at dislocations, together with climb, returns the system back to the equilibrium number. Of course, an

edge dislocation can also serve as a SIA sink with the climb motion proceeding in the opposite direction. Furthermore, it should be clear that a dislocation of screw character will not act as an effective sink for vacancies and SIAs. Finally, the sink properties of dislocation lines can also be extended to the case of dislocation loops. Only some portions of the shear loop can act as a sink and the climb will be confined to those regions of near edge-like character. Annihlation of point defects at Frank loops on the other hand will result in the growth of the loop in the case of interstitials and loop shrinkage when vacancies are the defect being eliminated.

The next step up from point and line defects is the planar defects. There are many planar defects grain boundaries, surfaces, twins, etc. but we focus here on what is known as a stacking fault. We know that the FCC crystal structure can be viewed as a periodic stacking of 111 planes given by the sequence . . . ABCABC . . . As shown in Fig. 2.7 a stacking fault is an error in this stacking sequence and two types of stacking faults can be envisioned. Intrinsic faults can be viewed as an extra row of atoms inserted in the sequence thereby changing . . . ABCABC. . . to ABCBABC and the extrinsic stacking fault results from a missing 111 plane changing . . . ABCABC . . . to . . . ABCBC . . .

Notice we have spoken of extra rows of atoms when describing both stacking faults and Frank dislocation loops. It is easy to convinced yourself that a Frank loop is nothing more than a circular stacking fault (at least in FCC systems) and the Burgers vector direction is determined by the intrinsic or extrinsic character of the fault. Alternatively, a stacking fault in FCC crystal is terminated by and edge dislocation line.

Fig. 2.7. Stacking faults in an FCC crystal are disruptions in the ABCABC stacking sequence of 111 planes. The intrinsic stacking fault, top, can be view as an extra layer of vacancies, whereas the extrinsic variety is an extra plane of atoms.

In an FCC crystal there are four types of (111) planes 111, 1 11 , 1 1 1 , and 11 1 each with a different orientation in the unit cell. The 111 planes are also the close packed planes in FCC with the atoms arranged in equilateral triangles within the plane. If we cut an equilateral triangle from each of the four 111 planes and stack them together a tetrahedron is formed. With this geometry in mind we can define one final defect known as a stacking fault tetrahedron. If the equilateral triangles cut from each 111 plane were in fact faulted, then the tetrahedron constructed is what we mean by a stacking fault tetrahedron.

2.2 Nuclear Fission The fuel of choice in a nuclear reactor is the Uranium isotope 235U. In this section we would like to summarize the processes that occur during the fission of 235U and how these processes affect the material used to contain the nuclear reaction.

To start lets take a look at the schematic diagram of the fission reaction depicted in Fig. 2.8. On the left an atom of 235U absorbs a neutron leading to the formation, briefly, of
236

U. As indicated by the squiggly line outlining the atom, the 236U isotope is unstable

and spontaneously fragments into lighter isotopes, that is, we have split the atom. Here we have denoted the fission products as 141Ba and 92Kr, but many possible isotopes are possible, 90Sr and 137Cs are other examples. In addition to the fission products more neutrons are produced.

92

Kr

neutron n
235

n
236

141

Ba

Fig. 2.8. Schematic illustration of one possible nuclear fission reaction. A neutron combines with a 235U atom forming the unstable 236U isotope.

The 236U splits into two other isotopes and, in this case, three high energy neutrons are emitted.

There are four features of the fission process that deserve further scrutiny. First, the production of lighter isotopes during the splitting of 236U means a considerable amount of kinetic energy is generated. This energy, along with radioactivity, in turn generates a great deal of heat. The heat, of course, can be used to produce steam, which drives a turbine and produces electricity. On a per mass basis the amount of heat generated in a nuclear fission reaction is much greater than that obtained in, say, the burning of gasoline in an automobile. That, and the fact that no CO2 is produced during fission, is a major attraction of nuclear power.

Unlocking the tremendous energy of the nucleus however does have its problems, as illustrated by the second important aspect of Fig. 2.8: fission products. The isotopes created after the splitting of 236U are themselves radioactive and many have very long half lives. What to do with this waste is of great political, social and, yes, engineering concern. We will certainly not be able to solve the waste problem within the pages of this text and instead we will focus on the topic of materials under irradiation.

Nuclear waste is not the only sticky political problem associated with nuclear energy. The third important feature of Fig. 2.8 is the neutrons produced. Notice only one neutron is required to initiate the reaction yet more than one is produced at the end, three in the specific reaction depicted in Fig. 2.8 but on average about 2.5 neutrons are emitted. These additional neutrons can be absorbed by more 235U atoms, which produces even more neutrons. If there is nothing in the fuel to absorb, at least partially, the neutrons being produced, then a self sustaining chain reaction results. An uncontrolled chain reaction goes by another, perhaps more familiar, name: an explosion. (We can discuss more about the physics of nuclear bombs, but then Id have to kill you.) A nuclear power generating plant relies on a controlled nuclear reaction, but there is not a huge technological barrier in moving from power plants to bombs.

The fourth aspect of the fission reaction is of central interest to the discussion in this chapter and it has to do with the energy of the neutrons being produced.

2.3 The irradiation cascade

Fig. 2.9 shows schematically how the potential energy of an atom varies with distance from another atom. In constructing this diagram we have somehow accounted for splitting of all the electron orbitals, as described in chapter 1, and have summed up the energy of all the electrons to arrive at a total energy picture. As r decreases from very large separation distances the energy of the system decreases whereas at very small distances the potential arises very quickly. The attractive tail and the repulsive contribution at small distances combine to define an equilibrium separation distance between the two atoms given by the position of the minimum. More important the depth of the potential energy well gives the energy required to take two atoms at equilibrium separation and move them infinitely far apart. In a typical metal the well depth is roughly -0.3eV. Adding up the contribution from all twelve nearest neighbors in an FCC crystal, we find a cohesive energy of approximately -4eV.

Now assume we have chosen a material, whose bonding characteristics are given in Fig. 2.9, to be the wall of our nuclear reactor. Recall that the fission reaction produces many neutrons of which many will escape from the fuel and will be stopped by our choice of containment material. Here is the important fact to keep in mind. The energy of neutrons from the fission reaction is on the order of 2MeV thats 2 million electron volts. Yes the neutron can lose a substantial amount of energy before reaching our containment, but if the atom of the wall material is bound to the lattice to the tune of just 4eV, then, in the interaction of neutron and metal atom, something has to give.

Fig. 2.9. Schematic diagram of the bonding characteristics of an atom in a typical condensed phase. On the potential energy vs. separation distance plot, ro represents the equilibrium interatomic spacing and the potential well depth is proportional to the cohesive energy.

A neutron from the fuel comes screaming into our wall material and collides with some atom in the crystal. The probability that the neutron will strike an atom and how much energy is transferred during the collision are topics beyond the scope of this course, but suffice to say the atom will acquire significantly more energy than is required to move it completely out of its lattice site. The bombarded atom will then move through the lattice striking other atoms along the way. These atoms can collide with other atoms and so forth. The process of atoms striking atoms, striking other atoms, etc. is called a cascade. The initially struck atom is called the primary knock on atom, or PKA for short, and can be viewed as the cue ball in the cascade event.

The PKA, and other displaced atoms, are moving very quickly in the material and the time scale for the formation of a cascade is on the order of ps (1x10-12s). Actually witnessing a cascade forming is therefore pretty much out of the question, but a technique known as molecular dynamics (MD) is very useful in simulating the effects of a PKA on

the defects created and the damage caused by a cascade. MD is not difficult to understand. We can establish a crystal whose interaction energy is known through functions such as that shown in Fig. 2.9. Since the potential energy, call it , is known, the force on an atom from another atom at some distance r is also known through
f = and by summing up the force from every neighbor of an atom the total force is

obtained. With the force we then can model how the atom will move by numerically solving the classical Newton law of motion f=ma, where the mass is known for any atom. (The trick to the MD technique, of course, is determining a potential energy that accurately describes the material being studied and a significant amount of literature exists on developing interatomic potentials.) In any crystal at moderate temperatures or higher each atom vibrates about its equilibrium lattice position with a frequency of about ps-1. Therefore the MD technique requires very small time step (~1x10-15s) to accurately solve f=ma, which severely limits the total time scale of any simulation. Nevertheless MD is of just the right time scale for the study of cascade formation.

Figure 2.10 shows an MD simulation of a 30keV cascade event in Cu at 300K. The top panel corresponds to a time of just 2ps after the PKA started its destructive path through the lattice and the bottom panel shows the resulting damage after 18ps. In the figure not every atom is pictured, instead only the vacant sites and interstitials are highlighted. Notice, early on there are large numbers of point defects. The extent of the cascade is roughly 30nm, meaning thousands of atoms are affected by the initial PKA. Also, notice that the cascade is not quite spherical. The fact that the cascade region extends further in some directions than others is simply a consequence of crystallography. The high energy displaced atoms can move more easily down certain crystal directions, or channels, in the FCC lattice.

Fig. 2.10. A molecular dynamics simulation of the cascade event for a 30keV PKA energy in pure Cu. Only the resulting interstitial and vacancy atoms are shown.

At 18 ps the cascade looks very different, why? The one word answer is: recombination. The number of vacancies and interstitials initially formed in the cascade region is much higher than the equilibrium values. The fastest way for the system to relax back to

equilibrium is an interstitial hopping into a nearby vacant site and in the process annihilating both point defects. As evident from the top and bottom configurations of Fig. 2.10, recombination goes a long way, and is very fast, in decreasing the supersaturation of point defects, but some excess interstitials and vacancies remain. The key question in understanding irradiation damage in materials is: what happens to these excess point defects?

Figure 2.11 shows a compilation of MD simulations of damage cascades in several metals. The notation MD v_MD refers to the number of vacancy-interstitial pairs produced after the PKA atom comes to rest and T denotes the kinetic energy of the PKA. The results suggest that most metals behave in a similar fashion and the production of defects can be described by the power law relationship

MD = AT n

2.2

where A is a constant that varies in the range of approximately 4.5-8 and the exponent n falls within the limit of .71-.83. (The label NRT refers to a continuum model of defect production, see the text by G. Was for details.)

Fig. 2.11. Compilation of MD data of defect formation under irradiation for several metals. The data are fit well to a power law dependence on the PKA energy.

Now that we know a little about irradiation cascades and the significant disruption to a crystal lattice they can cause, we can address the following important question: how do we characterize the total amount of damage incurred by a material irradiated by neutrons? The fluence refers to the total number of neutrons incident upon a unit area of material and has units of n/cm2. However the fluence says nothing about the energy distribution of the neutrons and therefore, given the results of Fig. 2.11, can say little about the total number of defects produced. Thus, the proper description of the total amount of damage is the displacements per atom or dpa. A dpa of 1 means that, on average, an atom has been displaced from its lattice site once. A dpa of unity may seem like a lot and indeed most reactors in service today will not experience a dose of dpa=1, however designs for the next generation of nuclear reactors are calling for materials with the ability withstand doses as high as 200 dpa!

Where do we stand? A nuclear reactor produces heat, steam and electricity in much the same way as a coal burning power plant and thus it is tempting to use the same tried and true pressure vessel steels from the coal industry for nuclear applications. However due to the physics of nuclear fission and, in particular, the high energy neutrons produced, materials for nuclear applications deserve special consideration. In the aftermath of irradiation induced cascades significant numbers of vacancies and interstitials are produced. Since the point defect concentration is way out of whack the system will tend to eliminate the excess. As will be seen in the next few sections, it is the plight of these excess defects that determine the long term microstructure and mechanical behavior of nuclear materials.

2.4 Void formation and swelling

A funny thing happens to an austenitic stainless steel after it receives a high dose of neutrons it expands. Fig. 2.12 plots the percent change in volume vs dpa for several compositions of Fe-Ni-Cr steels at three different temperatures. The data indicates there is an incubation period at low dpa where nothing much happens and the length of this transient regime depends on factors such as temperature, the composition and the dose rate. After the incubation period, however, all steels tend to swell at a rate of about one percent per dpa. Accompanying the swelling is the appearance and growth of a large number of small voids spread throughout the bulk of material. Aside from the changes in materials properties during the swelling process, the dimensional changes themselves present a serious nuclear reactor design concern. Before addressing possible solutions to the swelling problem, we will first try to understand what is happening.

Fig. 2.12. Swelling vs. neutron fluence for several Fe-Ni-Cr austenitic stainless steels. Notice, there is an incubation time after which the swelling occurs at a rate of roughly 1%/dpa. (F. A. Garner, J. Nucl Mater, 122-123, 459 (1984).

In the previous section we argued that the cascade induced excess point defects have to go somewhere. So lets start by writing down an equation which tracks the concentration of vacancies with time, specifically we will attempt to describe the derivative dcv / dt where the subscript v denotes vacancy. The usual process of diffusion suggests that the first term governing the behavior of dcv / dt will be Dv 2 cv where Dv is the diffusion coefficient of the vacancy. Physically this contribution says that any spatial nonuniformities in the vacancy concentration will be smoothed out over time. If we confine the discussion to a time scale that is long compared to the frequency of cascade events, that is, on average the excess point defect production is homogeneous throughout the bulk, then this diffusion contribution can be ignored.

In the description of cv we need a term that reflects the production of new vacancies from cascade events. Under constant irradiation conditions the production rate will be a

constant, denoted by P. All effects due to neutron energy, flux of neutrons (n/cm^2/s), etc can be lumped into P. Competing with the production term are processes that remove vacancies from the material and there are several such sinks. The first is the recombination of interstitials and vacancies. From Fig. 10 it is clear that recombination takes place very rapidly at the very early stages of a cascade, but recombination can still occur at the longer time scales of interest here. Mathematically, recombination can be described by a term given simply as:
4Rc (Di + Dv )ci cv

2.3

where the subscript i denotes interstitial. In words Eq 2.3 states that the probability of a recombination event depends on the probability of finding an interstitial ci and a vacancy in the same spatial region defined by the radius Rc. The atomic volume is denoted by and the units on the sink term is the number of recombination events per atomic site per second.

In our review of defects in crystals it was stressed that edge components of dislocations can annihilate vacancies (and interstitials). To capture the interaction between vacancies and dislocations the following term can be included in the model:
o Dv N d A d Z vd (cv cv )

2.4

where Ad is a geometric factor describing the spatial extent of the dislocation (the superscript d denotes dislocation, the sink type) and Nd is the density of dislocations. The equilibrium concentration of vacancies is defined in eq. 2.4 and its appearance in the above term indicates that the defect sinks only operate, on average, if the vacancy concentration exceeds the equilibrium value. The parameter Z vd , known as the bias factor, is a measure of how effectively dislocations interact with vacancies. More will be said about Z vd shortly.

Finally, any voids that have formed can act as sinks for vacancies. Therefore another term should be added to the rate equation:
o 4RDv N v Z vv (cv cv )

2.5

The Nv and Z vv terms have equivalent meanings as in the dislocation case but the geometric factor is written out explicitly to emphasize the dependence on the void radius R.

Now we are in a position to write down the complete reaction rate model and, because the development for interstitials follows the same line of reasoning as described above for vacancies, we will tack on one additional differential equation. The result reads:
dcv 4Rc (Di + Dv )ci cv Dv N d A d Z vd cv cvo = P dt o 4RDv N v Z vv cv cv

)
2.6

dci 4Rc (Di + Dv )ci cv Di N d A d Z id ci = P dt 4RDi N v Z iv ci

where we have assumed the equilibrium interstitial concentration is negligibly small. In the development of the rate model we have introduced a bias factor for interstitial dislocation interaction. A detailed theoretical treatment of the two bias factors is quite difficult and will not be attempted here, but the results suggest that Z id should exceed Z vd .

The coupled system of differential equations 2.6 cannot be solved analytically. First of all the equations are nonlinear due to the presence of the ci cv product in the recombination terms and second parameters such as R, N d , N v are strictly speaking time dependent. To proceed lets make some simplifying assumption. If we investigate the early stages of the irradiation process, then there will be no voids present and N v = 0 .

Also, the time scale over which the dislocation density is changing is typically small compared to the generation and diffusion of point defects, in which case N d can be taken as a constant. With these assumptions and under steady state conditions ( dcv dt = dci / dt = 0 ) we find:

o cv cv =

Di Z id ci Dv Z vd

2.7

As stated previously the mobility of interstitial is larger than that of a vacancy or Di > DV and together with our previous statement Z id > Z vd eq. 2.7 implies that the concentration of vacancies left over from the irradiation process is greater than the number of interstitials. Because interstitials are more quickly and effectively removed from the system by sinks there exists a supersaturation of vacancies and it is this imbalance that leads to the formation of voids in the material.

The data of Fig. 12 suggests that the formation of voids is not immediate. There exists some initial time period over which N v is approximately zero. This incubation period can be understood in terms of nucleation theory. A complete discussion of the intricacies of nucleation theory would be too lengthy to include here and reviews can be found in standard textbooks on phase transformation, but we can nevertheless provide a brief qualitiative account. A system containing too many vacancies can lower its free energy by condensing them in the form of voids, much like liquid water condensation, ie. fog, is the result of a supersaturation of water vapor in the atmosphere. The driving force to form the voids depends on the volume, the larger the void size the greater is the number of vacancies removed. Counter acting the free energy gain is the penalty in forming a void-metal surface. The surface energy effect depends on the surface area, rather than the volume, and it will be the dominant effect at small void sizes. The volume vs surface area terms imply that a critical nucleus size can be defined, below which the void will tend to shrink and above which the nucleus will continue to grow. At low vacancy supersaturations, ie.low driving forces, the critical radius is very large and the probability

of forming one through random fluctuations in the capture and release of vacancies will be practically zero. At higher supersaturations however the critical radius is much smaller and the probability of creating voids that are stable is increased. The nucleation picture is consistent with the observed delay in the onset of swelling in irradiated material. The supersaturation of vacancies must build up over time until the excess concentration is sufficiently large that nucleation of stable voids becomes favorable.

2.5 Irradiation hardening

Figure 2.13 is a schematic diagram showing the change in mechanical properties of a material as a function of the dose. The first important feature of these stress-strain curves is the fact that the yield strength is increasing with irradiation exposure. Recall the low strain portion of the curve represents the elastic response of the material and is characterized by a linear relation between stress and strain. The point where the curve deviates from linearity, the yield point, represents the onset of plastic behavior. From a microstructural point of view the yield stress is the stress required to move dislocations through the material. Strategies for improving the yield strength of a material involve processing steps that introduce obstacles to dislocation motion. Obstacles can take the form of precipitates, solute atoms, grain boundaries and even other dislocations.

Fig. 2.13. Schematic illustration of the effect of irradiation on the stressstrain behavior of a typical metal. The yield strength increases, the work hardening rate decreases and the ductility decreases.

It would seem then, that the increase in yield strength depicted in Fig. 2.13 must be the result of an increase in the number of dislocation obstacles under irradiation conditions and we already know this is the case. In the previous section we outlined the production of voids from the supersaturation of excess vacancies and we do not have to stop there. Another defect sink that can be included in the rate theory expressions is Frank loops. Fig. xx illustrates a Frank loop visualized as the removal a disk shaped plane of atoms or, more to the point, the loop can be created by the coalescence of excess vacancies on a 111 plane in FCC materials. Just as in the case of voids, the number and size of Frank loops can increase over time. Another source of irradiation induced obstacles to the flow of dislocations is the stacking fault tetrahedron. A simple mathematical model for the increase in yield strength, y , can be written as;

y = A N

2.8

Here as before N is the obstacle density and the square root dependence is valid irrespective of the type of obstacle. The proportionality constant A distinguishes the obstacle and also contains terms like the shear modulus of the material, the dislocation Brugers vector and the square root of the obstacle size. Since the obstacle density is increasing with irradiation dose it is tempting to conclude that the increase in yield strength will increase with time as: y t

2.9

where is the neutron flux. (Note flux times time is the fluence.) Indeed such behavior has been observed experimentally, but only for the early stages of irradiation. At larger

dpa the yield stress tends to saturate to a constant value. In principal we could determine the saturation by numerically solving a set of rate equations, like 2.6, determine the obstacle density over time and use Eq. 2.8 to estimate the yield strength change. Fortunately there is a much simpler approach to the problem, which describes fairly well the saturation effect. Makin and Minter assumed the following. If a displacement cascade takes place within in some volume V surrounding an obstacle, then no new obstacle will form in the region. Saturation will result because as more obstacles are formed the probability of adding even more will diminish. The main assumption of the model can be captured by the following differential equation:

dN B = (1 VN ) dt V

2.10

where B describes the likelihood that a incident neutron will produce a defect/obstacle. Notice, the generation of defects stops when N=1/V. The solution of Eq. 2.10 is given by:

N=

1 1 e Bt V

2.11

and when substituted into eq. 2.11 results in a yield strength vs time relationship given by:

y =

A V

[1 e ]

Bt 1 / 2

2.12

Fig. 2.14 shows experimental data of the yield strength for a variety of 300 series stainless steel as a function of dose. The results of the model Eq. 2.12 prediction are shown by the solid line and clearly the trend in y is reproduced quite well.

Fig. 2.14. The increase in the yield strength as a function of dose for several 300 series steels. The data is fit well to the model described by Eq. 2.12

Fig. 2.15 shows in detail one effect of obstacles on the plastic deformation behavior of a metal. The top left schematic depicts a dislocation moving along its glide plane, from left to right, under the action of an applied shear stress. The dislocation feels a repulsive force due to the presence of two obstacles. In order for the defect to pass through the two obstacles the dislocation line must bow out between the pinning points (middle figure) and the bowing process requires a greater stress than is needed to pass the dislocation in an obstacle free lattice. Hence the yield stress is increased. The remaining diagrams of Fig. xx illustrate the sequence of events after the dislocation moves beyond the obstacles. As shown in the upper right, the dislocation pinches off and two loops are created. The newly formed defects result in two effects. The total dislocation density of the material is increased and the effective size of the obstacles increases, thereby increasing the level of stress needed to pass a second dislocation along the slip plane (see the bottom set of diagrams). The increased stress needed to sustain plastic deformation as the strain is increased is known as work hardening and the effect is manifest by the upward slope of the stress-strain curve after the yield point as seen in Fig. 2.13.

In addition to the yield stress increase, the stress-strain curves of Fig. 2.13 show that the degree of work hardening decreases with increasing dose. This odd behavior can once again be understood by examining the interaction between gliding dislocation of irradiation produced defects. As mentioned above one such defect is itself a dislocation in the form of a Frank loop. To understand how a mobile dislocation reacts with a Frank loop will require an extensive knowledge of dislocation theory, but suffice it to say that there are several mechanisms by which the Frank loop can become unfaulted and the original loop then becomes part of the dislocation network. In other words, unlike the case of typical metals where the obstacles become more effective obstacles as strain increases, in irradiated metals the passage of dislocations actually removes some pinning barriers. The overall effect is a decrease in the work hardening.

One final aspect of the trend illustrated in Fig. 2.13 is worth discussing and that is: the ductility of the material decreases as the dose increases. A measure of ductility is the total strain to failure, which, as noted by the X terminated points in the stress-strain plots, is shifting to smaller strains with increase irradiation exposure. The shift can also be explained in terms of the unfaulting of Frank loops during plastic deformation. As the Frank loops are rearranged the total dislocation of the mobile dislocation network is increased. An increased dislocation density results is a decrease in ductility. There is a particularly annoying aspect of the ducitility shift in the case of austenitic steels. The maximum drop in ductility with increasing dpa takes place at approximately 300C, roughly the same temperature at which core components in a light water reactor are maintained.

Since a metal typically loses ductility under irradiation conditions, it should come as no surprise that the fracture toughness of the material also drops dramatically. The Additional irradiation induced obstacles impeding the flow of dislocations will increase the yield strength, but at the same time will make it more difficult for a propagating crack tip to be blunted and hence arrested. What is not so obvious however is the fact that for ferritic steels the ductile to brittle transition temperature will increase with increased neutron flux.

Recall that the fracture toughness is typically expressed in terms of the total energy absorbed by the material during fracture. The energy is measured in a Charpy impact test, which is nothing more than a heavy pendulum that crashes into a notched specimen. The energy absorbed is manifest in the maximum height of the pendulum arm after fracture takes place. A typical result of the fracture toughness change is illustrated in Fig. 2.16 for a ferrite steel weld containing high concentrations of Cu, Ni and Mn. The open symbols show the Charpy impact energy for an unirradiated specimen. At high temperatures the energy is roughly constant at 118J and this plateau energy is called the upper shelf energy. At around 0oC or slightly higher the impact energy drops precipitously, signaling the ductile to brittle transition. The changes that place under irradiation conditions are summarized by the solid symbols. Here we see that a significant drop in the upper shelf energy, 78J, as well as a 169oC increase in the ductile to brittle transition temperature occurs.

Fig. 2.16. The change in the Charpy impact energy for a ferritic steel in the unirrdaiated vs. irradiated condition. Under irradiation the upper shelf energy decreases and the ductile to brittle transition temperature shifts to higher tmepeatures.

2.6 Phase Equilibria under Irradiation

The phase diagram is an essential roadmap for a materials engineer. For a thermal treatment of any material the temperature-composition phase diagram can answer vital questions about the process, such as: what phases tend to form, what is the composition of each phase and how much of each phase is present in the material? It is important to keep in mind however, that a phase diagram refers to equilibrium conditions. If, for example, the thermal processing step involves holding the material for a fairly short time at a given temperature the phase diagram may not correctly predict the microstructure observed. Under irradiation conditions the same consideration must be considered. It turns out that a constant flux of high energy neutrons and the continual displacement of atoms from their ideal lattice position can not, in general, be considered an equilibrium state.

The effect of irradiation on phase equilibria manifests itself in three important ways. The first is the appearance of equilibrium phases that do not form in the absence of an incident neutron flux. A good example is the Cu-Ni alloy system. The published phase diagram of Cu-Ni reveals a continuous solid solution across the entire composition range; the only two phase region is the solid-liquid lens shaped portion at higher temperatures. But under irradiation conditions it has been found that Cu-Ni also exhibits a miscibility gap at low temperatures (the top of the gap occurs at roughly 400K) and has been verified that the two phase region is an equilibrium feature of the alloy. Why, then, would an equilibrium two phase region reveal itself only in the presence of irradiation? Recall that the basic atomic mechanism of diffusion in a substitutional alloy, like Cu-Ni, is a vacancy exchange. At temperatures around 400K there are very few vacancies in equilibrium (see Eq. 2.1), atomic mobility is quite small and complete equilibrium would require extremely long heat treatment times. We now know that under irradiation there is a much

greater concentration of vacancies than the equilibrium value. As a result the diffusion coefficient of solute atoms increases tremendously and the true equilibrium state can be achieved in reasonable time frames. The enhanced kinetics accompanying irradiation has led to the discovery of equilibrium phases in other alloy systems, for example the Pd8W and Pd8V compounds. In austenitic steels the phase of stoichiometry M6C where M is some metal atom and the M23C6 phase are examples of phases that do not form under usual thermal treatments but can appear under the harsh conditions of a nuclear reactor.

The second important aspect of phase equilibria under irradiation is the fact that nonequilibrium phases can form. To understand why consider again the atomic level changes continually taking place as a high energy neutron flux is incident upon a material. The excess point, line and planar defects created raises the free energy of a phase. However, the increase in energy is not the same for every phase. The energy of formation of say an interstitial may be quite different for a pure metal and some intermetallic phase. The free energy vs composition diagram of Fig. 2.17 illustrates how increasing the free energy of phases by differing amounts can lead to the appearance of an otherwise unstable phase. The solid lines in the figure represent to unirradiated condition. Here there exists a two phase alpha-beta region whose compositions are given by the standard common tangent construction. The phase labeled gamma has a free energy above the common tangent and will therefore never appear in true thermodynamic equilibrium. The dashed lines illustrate schematically the change in free resulting from a neutron flux. Now the free energy of the beta phase has increased to a larger extent than either alpha or gamma such that the new common tangent line predicts the formation of gamma. Notice the beta phase exhibits very limited solubility (the free energy curve is very skinny). It is known that intermetallic phases with limited solubility experience a greater rise in energy under irradiation. Phases that are induced by irradiation include the cubic gamma prime Ni3Si and the tetragonal Cr3P phases in steels.

Fig. 2.17. Illustration of how irradiation can change phase equilibria in a binary alloy system. The solid lines represent the equilibrium free energy under no irradiation and the two stable phases are and . Under irradiation the free energies increase as shown by the dashed lines and now a third phase appears.

The final noteworthy feature of phase formation under irradiation conditions is the modification of phases that can occur. To understand this aspect of the interaction of high energy neutrons and materials, note that a cascade produced in an alloy will tend to mix solute atoms locally. Due to this ballistic mixing process we will need to rethink atomic transport in materials and to illustrate the implications consider the very simple example of a precipitate whose radius R is growing over time in a supersaturated matrix phase. Since growth takes place by attaching solute atoms at the precipitate matrix interface, a solution to the diffusion equation will yield gradient in concentration near the

surface of the precipitate, which in turn yields the flux and hence the growth rate. The result is as follows:

4R 2

dR 3Dcm = R cp dt

2.13

Where D is the diffusion coefficient, cm refers to the composition in the matrix far from the particle surface and cp denotes the concentration of the precipitate, assumed constant throughout. Now imagine what happens when a cascade is formed in the vicinity of the surface. Atoms from the precipitate can be knocked out into the matrix and vice versa. The net result is a shrinkage rate of the particle and this contribution can be written as:

4R 2

dR = 4R 2 dt

2.14

where is the neutron flux (dpa/time) and omega is the atomic volume. Notice the size dependence on the RHS of Eq. 2.14, reflecting the fact that a larger size precipitate is more likely to experience a cascade event. One final ingredient is necessary before establishing a growth rate under irradiation conditions and that is a statement of conservation of solute:

co = 4 / 3R 3 c p + cm

2.15

where co is the average concentration in the system and is the density (number/volume) of precipitates. Eq.2.15 states that any solute atom has to reside in either the precipitate phase or the matrix phase.

Combining Eqs. 2.13, 2.14 and 2.15 leads to:

3Dco dR = + R2 dt 4R

2.16

We need not solve this differential equation to understand the qualitative features of the model. At small sizes the second term on the RHS of Eq. 21.6 will dominate and the growth rate is positive. At larger sizes, however, the first and third term wins and shrinkage will take place. Unlike usual precipitation in alloys, under irradiation there will be a critical radius where growth stops and the higher the neutron flux the smaller the limiting particle size. The above model then is a simple, but illustrative example of what is meant by phase modification under irradiation conditions.

One further example of microstructural evolution in the presence of irradiation will be considered here and as we will see quite dramatic changes are possible. The phase transformation in question is known as spinodal decomposition or more generally phase separation. Consider a quench depicted by the downward pointing arrow in Fig. 2.18. The initial state of the system is a single phase , but after the lowering of the temperature into the miscibility gap the long time equilibrium state of the alloy is a mixture of two phases 1 and 2 whose compositions are given by the boundaries of the gap. For certain average compositions the quenched system is unstable with respect to concentration. What does this mean? Unlike nucleation where a new phase will not form until a sufficiently large cluster forms through fluctuations, the unstable regime represents a state where any composition fluctuation no matter how small will grow in amplitude. The growth in composition will continue until the equilibrium concentrations are approached and after this fairly short time period the domains of 1 and 2 will become larger (coarsen) over time. The phase separation process is depicted in Fig. 2.19 where the dark and light regions represent the two phases.

1 1+2 c1 c c2

Fig. 2.18. A binary phase diagram exhibiting a miscibility gap. A quench, as depicted by the downward pointing arrow, from the single phase results in phase separation between two phases 1 + 2 .

Fig. 2.19. Illustration of coarsening during phase separation. Time is increasing from top, bottom left and bottom right. From T. M. Rogers, K. R. Elder and R. C. Desai, PRB, 37, 9638 (1988)

There is one important aspect of spinodal decomposition that should be clarified before discussing the effects of irradiation. At the very early stages there exists one well defined wavelength of composition fluctuation that grows faster than all others. In any system concentration fluctuations will naturally occur and any spatial variation can be thought as a superposition of many different wavelengths. Spinodal decomposition selects one such

wavelength and that component will dominate the early stage microstructure. Fig. 2.20 is schematic illustration of how this wavelength selection comes about. The top curve represents a long wavelength fluctuation. In order for this component to grow solute diffusion over long distances must occur. On the other hand the very short wavelength composition profile shown in the bottom curve exhibits very high gradients. The gradients are like interfaces and have associated with them a high free energy. The middle curve represents that wavelength which is the sweet spot between the two competing effects. For the fastest growing wavelength, diffusion distances are relatively short and the free energy penalty for gradients in composition is modest.

Fig. 2.20. In the early stages of phase separation there exists a wavelength of composition fluctuation that grows faster than all others.

As weve said before, irradiation introduces a ballistic mixing effect into the transformation. Let us assume that a cascade event affects a local volume in the sample whose radius is R and, moreover, we will assume that the mixing will immediately replace whatever composition profile that exists within R with a completely uniform composition (equal to the average). It should be clear at this point that R will depend on the incident neutron energy. In the discussion to follow we will, for convenience, normalize the actual size of the cascade zone by the fastest growing initial wavelength as

shown in Fig. 2.20. The second important feature of a cascade event is, as before, the flux. How often the concentration is knocked back to co will obviously affect the overall microstructural evolution. Enrigue and Bellon have analyzed mathematically phase separation under the assumptions just described and the results are quite surprising and interesting.

Fig. 2.21 summarizes the behavior. It is plot of then normalized cascade size vs. a normalized neutron flux, denoted by and defines three different scenarios for an alloy quenched to some temperature within the miscibility gap. For small fluxes and small cascade volumes the system will completely phase separate. The kinetics of the process may be slowed by the irradiation damage, but in the long time limit two phases, 1 and 2, of infinite extent will form. At the other extreme for very fast bombardment of very high energy neutrons, the ballistic mixing term is so strong that the system remains at a uniform concentration, that is phase separation reaction is completely supressed. The interesting case occurs at intermediate R and . Here a characteristic concentration wavelength can grow and coarsen over time, but the continual remixing prevents any further growth. When the concentration growth rate is comparable to the neutron flux and when the fastest growing wavelength is close to the cascade size (R~1 in our reduced variables), the system exhibits a state characterized by pattern formation. The system is at an impasse with the thermodynamics promoting further coarsening and the ballistic term favoring a return to a uniform concentration. Within the intermediate pattern forming region the stripe spacing will vary depending on the exact choice of R and .

Fig. 2.21. Schematic diagram showing the possible microstructures that can develop under irradiation according to the Enrique-Bellon model. See text for an explanation of R and . The salient features of the Enrique-Bellon model have been verified by experiments, but more importantly the results illustrate the bizarre microstructural evolution behavior that can occur in materials under irradiation. It is also worth mentioning that other examples of pattern formation in irradiated systems have been identified, such as the non-uniform spatial arrangement of defects and solute composition and the alignment of He bubbles into periodic FCC and BCC arrays.

2.7 Radiation Induced segregation

The region in a crystal in close proximity to a line defect or a planar defect surface, grain boundary, void, interphase boundary represents an energetically favorable, or sometimes unfavorable, position for a solute atom. Thus in equilibrium the solute concentration profile across the defect will exhibit a peak or valley at the defect position.

From the discussion of the previous section it is clear that we cannot always believe equilibrium behavior when discussing materials under irradiation conditions and the situation with solute segregation is no different. Under an incident neutron flux solute segregation can be enhanced or even reversed. Figure 2.22 illustrates what is known as radiation induced segregation (RIS). The data are taken from a scanning electron microscopy study of a 300 series stainless steel irradiated to several dpa and the concentration vs distance from a grain boundary is plotted. At the grain boundary the species Ni, Si and P have increased in concentration relative to their bulk values, whereas Cr has been depleted at the defect. The behavior exhibited in Fig. 2.22 is not observed in un-irradiated steel and thus we are faced with yet another example of microstructure that is radically altered due to irradiation conditions. Since solute segregation can have a major impact on materials properties, most notably embrittlement at grain boundaries, it is important to understand the origin of RIS and at this point it should not be a surprise to learn that RIS has to do with the annihilation of excess point defects.

Fig.2.22. An illustration of radiation induced segregation at a grain boundary in a 300 series stainless steel. Note the increase in Ni, Si and P concentration at the boundary plane and the depletion of Cr.

Although the mathematics underlying RIS can be formidable, the basic idea can be illustrated through the use of flux diagrams. Consider a binary alloy composed of the substitutional species A and B and further assume that in equilibrium there is no preference for A or B at a grain boundary. Far from the boundary there is an excess number of vacancies and interstitials due to the incident flux, but the boundary itself acts as an effective sink such that in the immediate vicinity of the boundary the concentration of point defects are maintained at close to their equilibrium values. Therefore, as shown in Fig. 2.23, the concentration of both defect species must decrease as the boundary is approached or in other words there exists a net flux of vacancies and interstitials toward the boundary. We have stated previously that a vacancy moves through the lattice by trading places with a neighboring atom. For the binary alloy considered here the neighboring atom can be either type A or type B and in general one species will exchange with a vacancy faster than the other. The vacancy mechanism of diffusion in a substitutional alloy means that the net vacancy flux toward the boundary will by countered by a reverse flux of A and B, but JA does not have to equal JB (see the left panel of Fig. 2.23). The imbalance of the reverse flux implies a net enrichment of one species relative to the other. (For those familiar with mass transfer in crystalline materials the name of this process will make sense. It is called the reverse Kirkendall effect.)

A similar mismatch of fluxes will occur as interstitials annihilate at the boundary. Here the diffusion mechanism is not vacancy exchange, but the hopping of interstitials from one available site to another. Therefore, as shown on the right of Fig. 2.23, the net flux of interstitials is composed of the sum of JA and JB, which again do not have to be equivalent. It is the combined effects of vacancy diffusion, interstitial diffusion and recombination that result in the phenomenon of radiation induced segregation.

Fig. 2.23. The fluxes of vacancies, interstitials and species A and B leading to irradiation induced segregation.

2.8 What do we do about radiation damage?

Imagine it is the year 1974 and you have just completed a design for a nuclear reactor. According to your specifications the reactor core that is the control rods, the fuel itself and a moderator will be housed in an austenitic stainless steel, which is known to exhibit high strength and high toughness at high temperatures. After ten or twenty years of operation you decide to evaluate the steel in terms of its mechanical properties and microstructure. At that point you seriously question the wisdom of your choice of materials. The steel has swollen in size, the microstructure is riddled with voids and small dislocation loops, the steel is hardened, and the ductility has dropped. When faced with this not so pretty picture, the obvious question becomes: what can we do to mitigate the effects of irradiation damage in alloys?

Around the mid-1970s it was found that ferritic steels exhibit a much lower swelling rate than austenitic steels and at least three reasons have been proposed to explain the improvement. The ferrite phase is BCC whereas austenite is FCC. We saw at the beginning of this chapter that the more open BCC crystal structure is more accommodating to interstitial species such as C or N. The interstitial atoms carry with them an elastic distortion and this elastic field interacts strongly with the excess interstitials and vacancies created in a cascade event. The interaction implies that C or N

atoms act as traps for the diffusing point defects with the net result that the probability of recombination of self interstitials and vacancies is increased. From the rate equation analysis of section xx, an increased K will result in a lower supersaturation of vacancies and a delayed onset of void formation. Furthermore, the elastic strain of an interstitial atom interacts with network dislocations such that the oversized C atoms tend to segregate towards the side of dislocation which is in a state of tension. The segregation, in turn, lowers the strain field of the dislocation and the net effect is a decrease in the preference of interstitial vs vacancy annihilation at the dislocation. Remember without a bias effect there is no swelling.

A second cause of the decreased swelling rates observed in ferritic steels has to do with dislocation loops. In an FCC metal it is fairly easy to see that the preferred structure of a Frank loop is an extra disk shaped plane of atoms or vacancies lying along the closed packed 111 planes. For the more open BCC crystal the nature of a dislocation loop is not so easy to guess and in fact under irradiation conditions two such loops are found in more or less equal concentration. The first has a Burgers vector given by a<100> and the second type b=a/2<111>. The first Frank loop type is very effective at capturing excess interstitials, but then the excess vacancy concentration is annihilated by the a/2<111> loops. The presence of two loop structures then has the effect of removing the overall bias and decreasing the swelling.

Finally, in steels possessing a two phase martensite-ferrite microstructure there exists an additional mechanism for swelling inhibition. Martensite exists in the form of thin plates throughout the microstructure. The boundary between the plates and the ferrite matrix acts as a planar and unbiased sink for excess irradiation induced defects. The additional high density sink acts to scoop up additional vacancies and interstitials that would otherwise result in void formation and growth.

It is clear that ferrite steels are preferred over austenitic steels for applications where exposure to high neutron doses will occur, but there is a problem with a reliance on ferritic type steels. At high temperatures ferritic steels exhibit diminished mechanical

properties, the creep behavior is of particular concern. In recent years it has become clear that in order to take advantage of the outstanding swelling characteristics of ferritic steels yet maintain mechanical integrity at high temperatures the use of so-called oxide dispersion strengthened ODS steels has shown great promise. Oxide dispersion is just as the name implies, it is the physical addition of small spherical oxide particles during the casting of steels. Oxides, most commonly yittria (Y2O3), in the nanometer size range have been employed in ODS steels. The oxides are very stable stoichiometric compounds that will not dissolve in the steel even at very high temperatures. Moreover, the small particles act as obstacles for the movement of dislocations and pinning centers for grain boundary motion, thereby increasing the high temperature strength and creep properties of the steel.

Over the past decade the phrase nanotechnology has become ubiquitous within the materials science and physics community. Nanowires, quantum dots and other nanoscale structures offer exciting possibilities in terms of electronic and optical devices. Somewhat surprisingly nanoscale structured materials are now finding potential applications for nuclear applications as well. Figure 2.24 shows a transmission electron microscopy image of a material composed of alternating layers of Cu and Nb. The structure was produced by a technique known as sputtering where atoms are removed from a target Cu or Nb and deposited layer by layer onto, in this case, a Si substrate. As is clear from the TEM image the sputtering process is well controlled and well defined layers can be produced. Moreover, the layer thickness can be varied and for the CuNb system layers as small as .8nm were fabricated.

Fig. 2.24 TEM micrograph of a layered Cu-Ng system which is found to be very resistant to irradiation damage. (From, A. Misra, R. G. Hoagland and J. P. Hirth, Acta Mater. 53, 4817 2005)

After exposing the nanolayered CuNb material to irradiation (He ions in this instance) it was found that for layer spacings less than 5nm virtually no evidence of irradiation damage was found. Bubble formation, swelling and excess defect populations were all absent in the microstructure.

From what we now know about the mechanisms of irradiation damage the superior resistance of CuNb multilayers can be understood. Cu is an FCC material whereas Nb is BCC. The Cu-Nb interfaces are incoherent and represent a strong sink for excess point defects. Also, with a 5nm spacing the sink density is ridiculously high. Finally it appears that the interfaces represent an unbiased sink with vacancies and interstitials annihilated rapidly and in equal numbers at the Cu-Nb boundaries.

The specialty CuNb structures are not the only potential use of nanoscale materials in the fight against irradiation damage. Conventional alloys with nanosized grains also offer a high density of unbiased sinks and therefore increased swelling resistance. Although ultra fine grain metals provide a possible improvement in irradiation damage characteristics, the long term stability of the structure and the resistance to grain growth requires further investigation.

You might also like