Professional Documents
Culture Documents
Downloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
764 E. WIGNER AND H. B. HUNTINGTON
Any lattice in which the hydrogen atoms would be the ordinary, molecular lattice of solid hydrogen. This
translationally identical (Bravais lattice) would have minimum-though negative-is much higher than that of
metallic properties. In the present paper the energy of a the molecular form, The body-centered modification of
body-centered lattice of hydrogen is calculated as a hydrogen cannot be obtained with the present pressures,
function of the lattice constant. This energy is shown to nor can the other simple metallic lattices, The chances
assume its minimum value for a lattice constant which are better, perhaps, for intermediate, layer-like lattices,
corresponds to a density many times higher than that of
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi
METALLIC MODIFICATION OF HYDROGEN 765
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi
766 E. WIGNER AND H. B. HUNTINGTON
(a) The difference between the energy of the electron recalculation of the interaction of the electron
clouds (2 in II) and the exchange energy (3 in 1/) clouds and the Fermi hole (exchange energy).
0.284/7,-0.017/7.'(E.-E), (in Rydberg units), (1) We shall see that x. can be assumed to be the
wh~re EK is a mean value of the energy of the higher s sum of a function of r alone (s part) and a
terms and has been estimated' to be about 20 Ry /,.'. function of r multiplied by (I" r) (p part). The
The second term in (1) is, therefore, almost negligible. boundary condition for the x. is periodicity in
(b) The correlation energy, calculated in ref. 2 III,
which can be represented for r, > 1 rather closely by
the lattice, which means zero derivative for the
s part, and vanishing of the p part on the surface
-0.584/(r.+5.1). (2)
of the s sphere.
(c) A correction for the Madelung number will be added Eq. (3) has been solved by a perturbation
in the next section.
method. \Ve shall need only those solutions of the
The energy after the corrections (a) and (b) have unperturbed equation
been added is plotted on a magnified scale on the
lowest curve in Fig. 3. One sees that a great part - D.~nl = 7]nl~nl' (4)
of the binding energy is lost.
which are functions of r alone (I = 0, or the s part
(3) of x), and those which are functions of r multi-
plied by (wr), (1=1 or the p part of x). The
It is very probable that the wave functions boundary conditions for these ~ functions are the
have nearly the form e ip , xih , not only in Na, but same as for the corresponding part; of x. We
also in the other alkalies excepting Li.5 In H this have' II
is not the case, however, and there will be further
corrections introduced into the energy curve ~oo= (3/41rr;) I, ~no=cn(sin wnr)/r,
similar to those calculated for Li by F. Seitz.' IV
~.I=(3!cn/p.wn)(p. grad) (sin wnr)/r, (5)
Still the situation for H is rather different from
2
that for Li. In H the wave function of the t}oo = 0, l1nO= 1Jnl =w n ,
lowest energy electron (p = 0) is strongly modu-
lated but the modulation does not increase much 1.4301l' 1.024
for higher p functions. In Li the wave function with Wl=---, (1=----;
for p = 0 was quite flat and the modulation r, (21l'r,) !
increased with increasing p.
1l'
For the calculation of these effects an ap- w.= (n+!h c---
proximate method has been used. The equation r, n- (21l'r,)!
for the wave function of the electron with the
momentum p, multiplied by e ip . x / h , is 6 when the last two expressions hold well for n> 1
only.
II.x. = - D.x. -iw grad x. - 2x./r= <.x., Next we must calculate the matrix elements of
(3)
H .. All terms give nonvanishing elements be-
tween ~nl and ~n'l only, except for iw grad which
Here the <. does not include the kinetic energy of gives elements between ~nl and ~n'I+I. We have
the flat wave functions, p'/2m=p.'/4 Ry. The
<. - <0 will give us the change in the Fermi (II.)oo; 00= -3/r,;
energy, and the wave functions x. the change in (II.)lo; 10=20.2/r.'-5.65/r,;
charge distribution, which are both caused by
the fact that the electrons are not free in the (II.).o; nO~(H.).I; nl~(n+!)'1l"/r.';
lattice. The charge distribution is used in a (H.)lI; 1I=20.2/r.'-3.66/r,; (6)
'Cf. F. Matt and C. Zener, Proc. Camb. Phil. Soc.
30,249 (1934); also J. Bardeen and E. Wigner, Phys. Rev.
(II.)n'o; nl= -(ip.wn/3 l )o.'n;
48,84 (1935); for Li see reference 2 (IV) and]. Millman,
ibid. 47, 286 (1935). (II.)oo; 10 = 1.36/r,;
Cf. Eq. (11), reference 2 (II). The energy in (3) is in
Rydberg units Ry and the length in Bohr units. (II.) 00; nO = 2(6)!/(n+ !)1l'r,.
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi
METALLIC MODIFICATION OF HYDROGEN 767
3 1.85
Ep.= - - -
r, 20.2 - 2.65r, - 6.721"(20.2 -0.6r,)-1
-0.0093. (8)
.FIG. ~. Bottom .Iine, correction to the Fermi energy;
The mean value of E.-EO with the weight 1", mlddle.1me, correctIon forthe Madelung number; top line,
correctlOfl for the mteractIon energy.
from 1'=0 to !L=3.84/r, is the correction to the
Fermi energy; it is plotted in the lowest curve of
Fig. 2. It is negative, just as in Li, because there and an exponentially decreasing charge distri-
is no p level below the s level of the valence bution around each ion:
electron. The energy E. - EO is no longer very
accurately proportional to 1'\ because the effect p= PI+P2, PI=e(1-c)/(4,,-r,a/3). (9)
of the next Brillouin zone is appreciable.
P2 itself contains two parts: first the exponential
(4) charge distribution around the ion, in the s
sphere in which we are, and second the exponen-
Next the interaction of the electron cloud and tial charge distribution around the other ions.
the more accurate exchange energy must be The latter was averaged over all directions so
derived. The greater density near the nucleus that P2 also became spherically symmetric in each
makes it evident that the first will be greater s sphere
than the free electron value assumed in the
~
previous section. This is true also for the ex- f33 Sh(f3r)]
p,= e-~'+4k--- ,
change energy, because the Fermi hole is more
8". f3r
effective in consequence of the uneven charge (9a)
distribution. The two effects work against each k=e-'P"(l +f3r,+f3'r.'/2) (f3r, _1)-1.
other, but in contrast to Li, the former is greater
here and decreases the binding. The two constants c and f3 were adjusted so as to
The calculation of these quantities is a little fit the square of the calculated wave function. (7)
laborious and contains nothing of principal best.
interest. It will be sketched only. The potential corresponding to this charge
The square of the calculated wave function, distribution contains also two parts: <PI, corre-
i.e., the electric density p, was approximated by sponding to the constant part PI and <P2 corre-
an analytical expression 7 containing a constant sponding to P2 which could be easily calculated,
and hence also the interaction of the electron
7 The reason for this is that onlv for this form were we
able to perform the integrations. . clouds.
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi
768 E. WIGNER AND H. B. HUNTINGTON
For the calculation of the exchange energy the these waves-let their form be u(r)-no longer
wave functions of the electrons with momentum overlap. The total interaction energy per electron
p were assumed to be in the form ei(pxllhx(r) is
where x(r) is the wave function of the lowest
state. Under these conditions, the exchange
}f <P(r) 1 u(r) 1 247rr 2 dr (11)
energy is 2 II. 2 III
where <P(r) is the potential of the charge distri-
- e-
2
2
f 1 1
1
p(T) p(T')I(( T - T' )~~dTdT',
17-7' 1
(10)
bution 1u(r) 12. Since no such interactions actually
exist, Hartree's equation gives an incorrect
result by this amount, which will be compensated
where by the exchange and correlation energy.
If Fock's picture is used instead, the interaction
sin (r/d)-(r/d) cos (r/d)]' energy between electrons with parallel spins
I((r) =9[ vanishes in consequence of the enormously
(r/d)'
increased Fermi hole. There remains still the
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissio
METALLIC MODIFICATION OF HYDROGEN 769
(6) CONCLUSION
EfRy
The great density of the hypothetical metallic
-IOE> ! form of H is not surprising. It seems to be the
general rule that the density of the metallic and
-1.07 valence lattice of a substance be greater than
that of the layer lattice, and this in turn has
greater density than the molecular lattice. Table
I illustrates the point.
FIG. 3. Energy of the lattice as a function of lattice This table makes it natural to assume that all
constant. Lowest curve, for flat wave functions; second atoms are bound together by shared electron
curve with all corrections, except zero-point energy of
nuclei; the dotted lines contain the zero-point energy, the forces in metallic and valence lattices, and that
lower for heavy, the upper for ordinary H. this is the case also for the atoms within a layer in
layer lattices. The different layers are bound
quanta vibration of the nuclei. If the elastic together, however, more weakly, perhaps by
constants of the medium were known, the elastic van der \Naals forces. This is well in accord with
spectrum and the zero-point energy could be the general geometric considerations of A. Reis
calculated following Blackman lO or a paper of and K. Weissenberg. 13 It seems sensible, there-
Barnes, Brattain, and Seitz.u Since, however, fore, to speak about valence-like forces between
only the compressibility can be derived from two atoms in a lattice, if the density of the
Fig. 3 (and this only very inaccurately, as it may outermost electrons of both overlaps considerably
be in error by as great a factor as two), it was at this distance.14
performed by the older method of Debye.12 The heat of formation of the ordinary, molecu-
From the curvature of the upper curve in Fig. lar H2 lattice is 52.5 Cal., practically one-half of
3 one obtains for the reciprocal compressibility the heat of dissociation. Although the figures of
of a lattice of very heavy hydrogen, the preceding section may be in error by quite a
few calories, it is evident that the molecular form
!(Cll +2c,,) = 2.9 X 1012 dyne/cm 2 for r, = 1.5 is so much more stable (by about 40 CaL) than
= 1.5 X 1012 dyne/cm2 for r, = 1.6 the metallic, that the chances of obtaining the
= 1.1 X 10 12 dyne/cm 2 for r, = 1. 7 latter are extremely small.
= 0.8 X 10 12 dyne/cm2 for r, = 1.8. In the ordinary lattice, there are two "nearest
di,tances." The distance between two atoms of
By analogy with other metals cll=2c,,=4c44 was the same molecule is 0.7 5A = 1.45ao, while the
assumed. The zero-point energy is then about nearest distance between two H atoms of
10:V1. Blackman, Proc. Roy. Soc. London A148, 365, 384, "A. Reis, Zeits. f. Physik 1, 204 (1920); 2, 57 (1920);
(1935); 149, 117, 126 (1935). K. Weissenberg, Zeits. f. Krist. 62, 12,52 (1925).
II Barnes, Brattain and Seitz, Ph)"s. Rev. 48,582 (1935). l4 This corresponds also to the ideas developed by L.
I'Cf. e.g., E. Schrodinger, Geiger-Scheel Handbuch dey Pauling", J. Am. Chern. Soc. 53,1367 (1931); J. C. Slater,
Physik, Vol. X. Ph)"s. Rev. 38, 1109 (1931).
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi
770 E. WIGNER AND H. B. HUNTINGTON
wnloaded 09 Jun 2013 to 147.226.7.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissi