Perfect.

Can we expect that all consequences of band theory, such as the Fermi surface and
energy gaps, will no longer exist? Will insulators become conductors?
Experiment and theory13 agree that the consequences of the destruction of the
translational symmetry are much weaker than one expects at first sight. If the impurity belongs to
the same column of the periodic table as the host element in replaces, then the effects are
particularly small, in part because the average number of valance electrons remains constant.
One measure of the effect of alloying is given by the residual resistivity. One atomic
percent of copper dissolved in silver increases the residual electrical resistivity by 0.07 ohm
cm, wich corresponds to a scattering cross sections of perhaps 0.03 of the area of the impurity
atom. Similary, the mobilility of electrons in Si Ge alloys is much higher than one would expect
from the simple geometrical argument that a germanium atom (32 electrons) is very unlike
aslicon atom (14 electrons) and therefore a Si atom in Ge or a Ge atom in si should act as an
efficient scattering center for change carries. The theory shows us that the effective scattering
potential of impurity may be very small.
There is no experimental evidence for an intrinsic reduction in band gap due to the
random aspect of alloying. For example, silicon and germanium form solid solutions over the
entire; the energies of the band edges in the alloys vary continuously with composition (fig.16).
It is believed that the desinty of orbitals near the band edges are somewhat smudged out by
alloying.
We now discuss substitutional solid solutions of one metal A in another B of different
valance. We suppose that atoms A and B occupy equivalent lattice positions at random. The
distinct effects which accur when the accupancies are regular and not random are consideret
under the heading of the order discorder transformation.
Hume Rothery has discussed empirical requirements for solid solutions to occur. One
requirement is size. It is difficult to form solid solutions if the atomic diameters 14 of A and B
differ by more that 15 percent. The sizes are favorable in the Cu (2,55 A0) Zn (2,66 A0) system :
zinc dissolves in copper as a fcc solid solutions up to 38 atomic percent zinc, the sizes are
somewhat unfavorable in the Cu (2,55 A0)Cd (2,97 A0) system: only 1,7 atomic percent.
13

These questions can be discussed using the method of orthogonalized plane waves and effective potentials
described in chapter 10. A low concentrations of impurity atoms cannot have much effect on the fourier components
UG of the potential U (r) responsible for the band gaps and for the behavior of the energy surfaces near the band gap:
An impurity will introduce fourier components of U (r) at wavevectors which are not reciprocal lattice vectors, but
such components are never large if the impurity atoms are located at random. Thus one obtains sharp x ray
diffractioins lines from random alloys. See also the references toward the end of chapter 11.
The atomic diameter is taken as the closest distance of approach in the crystal structure of the element
( tabel 1.5).

Figure 16 (a) calculated variation of the major band edges in Gesi alloys as a functions of the silicon
concentration band edge in Ge. (Compare with fig. 11.15) The calculations are by F. Bassani and D. Brust, phys.
Rev.131, 1524 (1963), who give extensive references to the experimental data. (b) The calculated variations as a
function of pressure for pure Ge. Qualitatively the variation of the bands with pressure is similar to the variations
with alloying, as we see by comparing energy values at equal values of the lattice constant.

Cadmium is solube in copper. The atomic diameters referred to copper are 1.04 for zinc and
1.165 for cadmium.
Although the sizes may be favorable , solid solutions will not from if there is strong
tendency for A and B to form stable compounds of defnited chemical proportions. If A is strongly
electronegative and B strongly electropositive, it is likely that compounds such as AB and A2B
will precipitate out of solution. Although the atomic diameter ratio is favorable (1.02) for As in
Cu, only 6 percent As is soluble. The ratio is also favorable (1.09) for Sb in Mg, yet the solubility
of Sb in Mg is very small.
Saveral aspects of the electronic structure of alloys can be discussed in terms of the
average number15 of conduction electrons per atom, denoted by n. The value of n in the alloy 50
percent Cu50 percent Zn is 1.50; in 50 Cu50 Al, n=2.00. Some of the principal effects of
alloying elements of different valence come from the change in electron concentrations. 16
Hume Rothery first drew attention to the importance of the average electron concentration as
determining structural change in certain alloy systems.
11

This number is often called the electron concentration.

16

J. friedel, Advances in Physics 3, 446 (1954).

Figure 17. Equilibrium diagram of phases in the copperzinc alloy system . the phases is fcc; and
are bcc; is a complex structure; and are both hcp, but which has a c/a ratio near 1.56 and (for pure
Zn) has c/a=1.86. The phase is ordered bcc, by which we mean that most of the Cu atoms occupy sites on
one sc sublattice and most of the Zn atoms occupy sites on a second sc sublattice which interpenetrates the
first sublattice. The phase is disordered bcc: any site is equally likely to be occupied by a Cu or a Zn
atom, almost irrespective of what atoms are in the neighboring sites.

The phase diagram of the copperzinc system 17 is shown in fig.17. The fcc structure of
pure copper (n=1) persists on the addition of zinc (n=2) until the electron concentration reaches
1.38. A bcc structure occurs at a minimum electron concentration of about 1.48. The phase
exists for the approxi mate range of n between 1.58 and 1.66, and the hcp phase occurs near
1.75.
The term electron compound denotes an intermediate phase (such as the phase of Cu
Zn) whose crystal structure is determined by a fairly welldefined electron to atom ratio. The
values of the ratio are called the humeRothery rules; they are 1.50 for the phase, 1.62 for the
phase, and 1.75 for the phase. Representative experimental values are collected in table 3,
based on the usual chemical valence of 1 for Cu and Ag;2 for Zn and Cd;3 for al and Ga;4 for
Si,Ge, and Sn.
17

The phases of interest are usually denoted by metallurgist by greek characters; in the Cu Zn system we have
(fcc), (bcc), (complex cubic cell of 52 atoms), (hcp) and (hcp); and differ considerably in c/a ratio. The
meaning of the character depends on the alloy system.

The HumeRothery rules find a simple expression in terms of the band theory of nearly
free alectrons. The observed limit of the fcc phase occurs close to the electron concentration of
1.36 at which an inscribed Fermi sphere makes contact with the Brillouin zone boundary for the
fcc lattice. The observed electron concentration of the bcc phase is close to the concentration
1.48 at which an inscribed Fermi sphere makes contact with the zone boundary for the bcc
lattice. Contact of the Fermi sphere with the zone boundary for the phase is at the concentration
1.54. Contact for the hcp phase is at the concentration 1.69 for the idea c/a ratio.
Why is there a connection between the electron concentration at which a new phase
appears and the electron concentration at which the Fermi surface makes contact with the
Brillouin zone boundary? It is costly in energy to add further electrons to an alloy once the filled
states reach the zone boundary. Additional electrons can be accommodated only in states above
the energy gap which accurs at the boundary or in the states of high energy near the corners of
the lower zone. It may therefore be energetically favorable for the crystal structure to change to
one which can contain a larger Fermi surface before contact. In this way the sequence fcc, bcc,
,hcp was made plausible by H.Jones
.

Figure 18. Number of free electron orbitals

Figure 19 Bohr magneton numbers of

per unit energy rangr for the first Brillouin

ferromagnetic nickelcopper alloys.

zone of the fcc lattices, as a function of energy.

The transformation from fcc to bcc is illustrated by fig. 18; this shows the number of
orbital per unit energy range as a function of energy, for the fcc and bcc structures. As the
number of electrons
is increased, a point is reached where it is easier to accommodate
additional electrons in the Brillouin zone of the bcc lattice rather than in the Brillouin zone of the
fcc lattice. The figure is drawn for free electrons, and it may perhaps be objected that the actual
desinty of orbitals in pure fcc metals is somewhat different see the Fermi surface of Cu in
figs.10.26, 10.35, and 10.38.
We consider the relationship of the s and d bands in pure nickel at 0 0K as suggested by
fig. 16.7b. There is a certain arbitrariness in the proposed distribution: we can transfer electrons
from the d subband than from the other d subband. Evidence that our particular choice may
correspond to reality is provided by fig.19 which shows the effect on the magneton number of
adding copper to nickel. We add one extra electron with each copper atom because the atomic
number of copper is larger by one than the atomic number of nickel. The desinty of electron
orbitals in the d band is over ten times greater than in the s band, so that the extra electron goes at
least 90 percent into the d band and less than 10 percent into the s band. The ferromagnetic
magneton number is observed to go to zero at about 60 atomic percent copper.
At 60 atomic percent copper we have added about 0.54 electron per atom to the d band
and about 0.06 electron to the s band. But 0.54 electron added to the d band of fig. 16.7b will just
fill both d subbands and will bring the magnetizations to zero, in excellent agreement with
observations. The distribution of electrons for 60 Cu Ni is shown in fig. 20. At low Cu
concentrations the decrease in magneton number from that of pure Ni should be a linear function
of the concentration of copper, in agreement with fig.19.

Figure 21 Desinty of orbitals in copper 3d and 4s bands. Calculations after C.Y.Fong and M.L Cohen
(unpublished); Experimental, after W.E. Spicer, in international colloquium on optical properties and
electronic structure of metals and alloys, F. Abeles, ed, NorthHolland, 1966.

For simplicity the block drawings show the desinty of orbitals as uniform in energy. The
actual desinty may be quite far from uniform: fig.21 gives the results of a calculation for copper.
The d band is characterized by a high desinty of orbitals. The desinty of orbitals at the Fermi
surface determines the electronic heat capacity and the pauli paramagnetic suscepbility; in the
transition metals they have higher values than in monovalent metals.

Figure 22 Magnetization of a free electron gas in neighborhood of a point magnetic moment at the origin r=
0, according to the RKKY theory. The horizontal axis is 2k Fr, where kF is the wavevector on the Fermi
sphere. ( after de genes)

Magnetic Alloys and the Kondo Effect

In dilute solid solutions of a magnetic ion a nonmagnetic metal crystal (such as Mn in
Cu) the exchange coupling between the ion and the conduction electrons has important
consequences. The conduction electron gas is magnetized in the vicinity of the magnetic ion
,with the spatial dependence shown in fig.22. This magnetization causes an indirect exchange
interactions18 between two magnetic ions, because a second ion perceives the magnetization
induced by the first ion. The interaction, know as the RKKY interaction, also plays a role in the
magnetic spin order of the rare earth metals, where the spins of the 4f ion cores are coupled
together by the magnetization induced in the conduction electron gas.
A dramatic consequence of the magnetic ionconduction electron interaction is the
Kondo effect. A minimum in the resistivitytemperature curve of dilute magnetic alloys at low
temperature has been observed in alloys of Cu, Ag, Au, Mg, Zn, with Cr, Mn, Fe, Mo, Re and Os
as impurities, among others.

18

A review of indirect exchange interactions in metals given by C.Kittel, Solid state physic 22, 1 (1968); a review of
the kondo effect is given by J.kondo, Theory of dilute magnetic alloys, solid state physic 23, 184 (1969) and A. J.
Heeger, Localized moments and nonmoments in metals the Kondo effect, Solid state physic 23, 248 (1969)

Figure 23 : a Comparison of experimental and theoretical result for the increase of electrical resstivity at
low temperature in dilute alloys of iron in gold. The resistance minimum lies to the right of the figure ,for the
resistivity increases at high temperature because of scattering of electrons by thermal phonons. The experiment are
due to D.K..C. Mac Donald, W. B. Pearson, and I.M. Templeton , Proc. Roy.soc. (London) A266,161 (1962): the
theory is by J.Kondo,Prog.Theo.Physics 32, 37 (1964).

The occurrence of a resistance minimum is connected with the existence of localized magnetic
moment on the impurity atoms. Where a resistance minimum is found, there is inevitably a local
moment. Kondo showed that the anomalously high scattering probability of magnetic ions at low
temperatures is an esoteric consequence of the dynamic nature of the scattering and of the
sharpness of the Fermi surface at low temperatures. The temperature region in which the Kondo
effect is important is shown in fig.23. No simple physical explanations of the effect has
appeared, but the original paper is quite accessible.