MAY 1, 1933 PHYSICAL RE V I EW VOLUM E 43

Diffraction of Electrons by Metal Surfaces

L. H. GERMER, Bell Telephone Laboratories
(Received March 18, 1933)

Fast electrons scattered from polished metal surfaces and (2) that Debye-Scherrer diffraction patterns are
do not form diffraction patterns. A strong Debye-Scherrer formed only by transmitted electrons. Fast electrons
pattern is produced, however, by electrons scattered from scattered at a small glancing angle from an etched poly-
a surface which has been mechanically roughened in such crystalline surface form a diffraction pattern if the surface
a manner that electrons are able to pass directly through appears mat or roughened, but no pattern is formed if the
projecting irregularities. Small ridges extending from surface shows metallic luster. Here again diffraction
wires, which have been drawn through an imperfect die, patterns appear to be produced only by transmission.
also give rise to a diffraction pattern. These experiments A probable explanation is given for the fact that diffraction
indicate: (1) that there is no considerable layer of amor- rings are not formed by electrons scattered from smooth
phous material (Beilby layer) on a polished metal surface, polycrystalline surfaces.

T T is a well-established fact t h a t electrons of the metal. A strong sharp p a t t e r n is formed

•*- scattered by polished metal surfaces do not when the scratches lie normal to the plane of
form diffraction p a t t e r n s — t h e scattering is incidence, a very much weaker one (relative to
diffuse and featureless as far as the eye can the background scattering) when they lie parallel
detect. This result can be explained on the to this plane.
hypothesis t h a t the atoms in the surface of the T h e p a t t e r n obtained from a scratched nickel
metal, which act as scatterers, have been com- surface with the scratches normal to the plane of
pletely disorganized b y the polishing until they incidence is' reproduced in Fig. 1. A similar
form a layer of amorphous metal as supposed by pattern for tungsten is shown in Fig. 2. In the
Beilby. This explanation of the phenomenon has former there is one weak ring of unknown origin
been proposed by G. P. Thomson 1 and by R. C. and there are eighteen rings due to the face-
French. 2 I t is m y purpose in this paper; (1) to centered cubic structure of nickel. In the latter
present certain experimental evidence opposed are fifteen rings all of which arise from the body-
to this interpretation, (2) to point out a corre- centered tungsten lattice. In each figure the
lation between t h e failure of polished surfaces to position of the primary b e a m has been registered
form diffraction p a t t e r n s and a similar failure in by switching on the current for an instant after
the case of certain heavily etched surfaces, and
(3) to present a new explanation for the absence
of diffraction p a t t e r n s from polished surfaces.
A beam of fast electrons (wave-length 0.05A)
is directed upon the polished surface of a metal
block a t glancing angle about one and a half
degrees. T h e scattered electrons are received FIG. 1. Diffraction pattern formed by electrons scattered
from a scratched nickel block, the scratches of which are
upon a photographic plate, and no detectable normal to the plane of incidence.
diffraction p a t t e r n is produced. Parallel scratches
are then m a d e on the surface b y rubbing the
block upon rather coarse emery paper. Electrons
scattered from the block in this condition form a
Debye-Scherrer diffraction p a t t e r n characteristic
1
G. P. Thomson, Proc. Roy. Soc. A128, 658 (1930). FIG. 2. Similar diffraction pattern from a scratched
2
R. C. French, Nature 129, 169 (1932). tungsten block.
724
 DIFFRACTION OF ELECTRONS 725

removing the specimen block. The photographs know that at least the greater part of the
obtained with the scratches parallel to the plane material of the fins was ordinary crystalline
of incidence are not reproduced. These show a metal.
diffusely scattered background of full speed These two experiments indicate strongly that
electrons upon which a few very weak rings of a polished metal surface is not covered by any
the characteristic patterns can just be detected. appreciable layer of amorphous material.
The interpretation seems obvious: The surface But polished surfaces are not alone in failing
material has been thrown up into ridges by the to produce detectable patterns; even heavily
scratching operation. Parts of these are thin etched surfaces give rise, in some cases, to diffuse
enough to transmit the electrons incident upon scattering only. The scattered electrons which
them, and to give rise to a pattern after the darken the photographic plate have, in these
manner of thin films. As the ridges lie almost cases, the same speed as the electrons in the
entirely parallel to the direction of scratching primary beam. (This fact I have established by
the pattern is intense when this direction is comparing the deflections received by the
normal to the plane of incidence, and extremely primary and by the scattered electrons in trans-
weak when it is parallel to the plane of incidence. verse electric fields.) Yet no detectable pattern
In other words, the diffraction patterns are is produced. There can be no question here of an
produced by electrons which have passed entirely amorphous layer.
through projecting ridges of metal, and when the Whether or not a given etched surface will
ridges are so turned that electrons cannot go produce a pattern can be predicted with con-
directly through them no pattern is formed. siderable certainty from its appearance. If the
Furthermore the ridges are made up of material etched surface is mat or frosted, it will almost
which lay in or very near the polished surface. certainly produce a pattern; if it is smooth and
This material gives rise to the usual ring pattern, shows metallic luster, almost certainly not.
and some of it, therefore, is crystalline. We (These smooth surfaces are sometimes produced
cannot say perhaps that it is all crystalline, but when the etching is very rapid.) The frosted
we can say that it is certainly not all amorphous. surfaces must present to the incident beam many
Another recent experiment also offers evidence sharp crystal edges, while the smooth surfaces
against the amorphous layer theory. In this do not. One is led to conclude that diffraction
experiment electrons were scattered from the patterns are produced only by electrons which
surfaces of drawn wires of nickel and of copper. have passed entirely through crystal edges pro-
The beam of electrons (0.05A) was normal to jecting from the surface. Certainly those surfaces
the wire and just grazed its surface. A wire which do not have such projections are incapable
which, immediately before testing, had been of giving diffraction rings. A similar conclusion
drawn through a good die gave a general scat- was reached independently in the experiments
tering of full speed electrons, but there was no on scratched surfaces. The idea that surfaces
trace of diffraction rings. Other wires were tested covered by minute projections would be favor-
which had been drawn through a broken die and ably formed for the production of diffraction
had projecting from their surfaces minute fins, rings was first put forward by G. P. Thomson. 1
caused by the forcing of metal into the cracks of The conclusion that diffraction patterns of the
the die. These under favorable conditions gave Debye-Scherrer type are in fact always trans-
complete Debye-Scherrer circles, part of each mission patterns is particulary acceptable be-
circle lying in the geometrical shadow of the cause it affords an explanation of the fact that
wire. A pattern of this sort could be produced such patterns never reveal evidence of refraction.
only by transmission of electrons through these It is easy to show that a potential difference of
fins. ten or fifteen volts between the outside and inside
The material of the fins had been cold worked of the metal should have a marked effect upon
to an extreme degree, comparable perhaps with the pattern provided the electrons which form
the working of a surface during polishing. Yet, the pattern have passed in and out of the metal
from the existence of the diffraction pattern, we through the same surface; yet such an effect has
726 L. H. GERMER

not been observed. If, as the present experiments mean plane of the surface by a maximum amount
pretty well prove, the patterns are produced in of as much as one-half of a degree, the average
all cases by transmission then effects due to width of diffraction rings would be of the order
refraction should be too small to be observed. of 2 mm and most rings would overlap neighbor-
But, if we accept this conclusion that all ing rings. If considerable portions of the surface
Debye-Scherrer patterns are formed by trans- were inclined to its mean plane by as much as
mitted electrons, we are at once faced by the half a degree this broadening would be sufficient
problem of accounting for the absence of such to make the observation of diffraction rings
patterns when electrons enter and leave a metal doubtful.
through the same surface. The failure of these I have tested a very highly polished steel
electrons to form patterns may very likely be the surface in the hope of finding diffraction rings
result of refraction and of minute surface irregu- showing the influence of refraction. No rings
larities. For electrons entering and leaving a have appeared. The test has, however, yielded
metal through the same surface, the calculated evidence that at least some well polished metal
effect of a potential difference between the out- surfaces possess sub-microscopic irregularities
side and inside of the metal is a considerable dis- involving variations of glancing angle of several
placement (and change of shape) of all diffraction degrees. The observation is a very simple one
rings. The amount of the displacement turns out —near the forward direction the intensity of
to be a function of the glancing angle upon the scattering varies greatly with the degree of
surface. If the glancing angle varies slightly from polish. An unusually fine surface scatters much
one section of the surface to another, because of more efficiently than does one which is less
surface irregularities, the result is a broadening perfect. I believe that this difference must be
of all the displaced diffraction rings. It seems attributed to shadows cast by irregularities on
quite possible that broadening produced in this the less perfect surface. This shadowing extends
way might be sufficient to cause each ring to at least as far as several degrees above the
overlap the neighboring rings, so that only surface, which means that there is a correspond-
general darkening of the photographic plate ing variation of glancing angle. Such a variation
could be detected. is sufficient to account for the absence of a dif-
I have worked out this idea for the special fraction pattern from the less perfect surface.
case of 50 kilovolt electrons (0.055A) incident And it seems altogether possible that the better
upon nickel, assuming a potential difference of surface would also be found to be very imperfect
15 volts at the surface of the metal. It comes out if we could obtain a still better standard against
that, if parts of the surface were inclined to the which to compare it.
 FIG. 1. Diffraction pattern formed by electrons scattered
from a scratched nickel block, the scratches of which are
normal to the plane of incidence.
 FIG. 2. Similar diffraction pattern from a scratched
tungsten block.