QUANTUM THEORY OF X-RAY REFLECTION

AND SCATTERING

Part I. Geometric Relations

BY SIR Co V. RAMAN AND D g . N . S. NAGENDRA NATH

(From the Department of Physics, Indian Institute of Science, BŸ
Received Ju]y 16, 1940

t. Introduction
THE simplest example of the optical excitation of the vibrations of a
crystal lattice is that afforded by the absorption of infra-red radiations by
the crystal. In this case, the material vibrations of the lattice and the
electromagnetic waves by which they are excited have the same frequency.
When the incident radiations lie in the visible or ultra-violet region of the
spectrum, they have a much higher frequency than the possible lattice vibra-
tions, but may nevertheless excite the latter, as is shown by the fact that the
light scattered by the crystal includes radiations of diminished frequencyo
The possibility of excitation in this case is due to the lattice vibrations
altering the optical polarisability of the material of the crystal. In both
cases, the excitation is essentially an optical process, as is clear from the
fact that it occurs irrespective of the temperature of the crystal. That it is
also a quantum-mechanical effect is particularly evident in the case of light
scattering; the intensity of the lines appearing with diminished frequency in
the spectrum of the scattered light is proportional to the Planck constant of
action and varies but little with temperature, so long as this is moderate and
the lattice vibrations are of sufficiently high frequency.

That the lattice vibrations of a crystal may be excited by X-rays in much

the same way as in the case of ordinary light has been suggested and support-
ed by experimental evidence in a recent series of papers in these Pro-
ceedings.* It was there shown that the character of the resulting pheno-
mena is different in the cases where the lattice vibrations lŸ respectively
in the acoustic and the optical range of frequencies. This is to be expected,
because in the vibrations of the acoustic class, the average electron density

* Raman and Nilakantan, 1940, 11, 379-408.

83
ATa F
84 C . V . Raman and N. S. N a g e n d r a Nath

in the unit cell of the lattice varies periodically while its distribution within
the cell remains nearly constant, while on the other hand, in the vibrations of
the optical class, the mean electron density in the unit cell remains approxi-
mately, constant while its distribution within the cell varies periodically with
time. The possibility of the vibrations of these two classes being induced
by the incidence of X-rays therefore arises in different ways. In one case,
it is the time-variation of the mean polarisability of the crystal and in the
other case the time-variation of the structure amplitudes (Fourier compo-
nents) of the electron density that is responsible for the excitation of the
vibrations. It is not surprising therefore that the effects produced on the
incident radiation differ even in respect of their geometric characters in the
two cases.
It is proposed in these papers to develop in mathematical forro the ideas
outlined in the papers quoted above. In this first instalment, the purely
geometric aspects of the phenomena will be discussed, the important
question of intensities being only touched upon incidehtally. The fuller
treatment of the latter will be given in Part II.

2. The Acoustic Case

From the remarks made above, it is evident that the cases in which
the frequency of the lattice vibrations lie respectively in the acoustic and
optical ranges must be considered separately. We shall proceed on the
assumption that exchanges of energy and momenta take place between the
incident X-radiation and the waves excited by it in the crystal, the substance
of the latter otherwise taking no part in the phenomena. The conservation
of energy and momentum requires
h v = hv' + hv* (1)

h h h
X = ~ + Z* (2)

~, v' and v* being respectively the frequency of the incident and scattered
X-rays and of the excited acoustic waves ; A, )( and h* are the corresponding
wave-lengths. Equation (1) shows that the frequency of the scattered X-rays
is diminished relatively to the incident X-rays by the frequency of the
sound-waves excited in the process. As, however, v i s far larger than ~*
we may in equation (2) take ~ = ~' without sensible error and proceed to
solve it geomet¡ as shown in Fig. 1.
 QuaJalz~m Theory of X-Ray Re/91 and Sc~~!eri,ag--I 85

0 ',R

Ÿ
/

Fin. 1

With O as centre, we draw a sphere having a radius equal to 1/2~. The

particular radius OI of the sphere which is para!lel to the incident X-rays
being marked out, we draw a second sphere with I a s centre and 1/)~* as
radius whŸ intersects the first sphere along, the circle R R. It is evident
that the generating lines of the cone O R R represent the possible directions
of the scattered X-rays corresponding to the acoustic wave-length u The
semivertical angle of this cone r is given by the relation
2 A* sin r161 ), (3)
When .1" is in¡ r vanishes and the cone of scattered X-rays collapses to
a line along the incident ray. As u diminishes however, the cone openS
out, reaching its maximum semi-vertical angle when the wave-length of the
acoustic waves is the minimum physically possible in the crystal. In the
case of a cubic crystal, this minimum wave-length would certainly be greater
than the grating constant of the crystal, and it is clear from (3) that in any
setting of the crystal, the directions in which the Bragg reflections by the
crystal could be observed would lŸ outside the cone of monochromatic radia-
tions scattered by it. The distribution of energy in the cone of scattered
X-rays is a special problem into which we shall not here enter. It is obvious,
however, that if the ac'oustic spectmm of the crystal has a concentration of
energy at or near a particular wave-length which may be greater than the
minimum possible, the scattered X-rays m a y be expected to show a con-
centration of intensity in the vicinity of a corresponding cone Of rays.

3. The Optical Case

Equations (1) and (2) are equally applicable in the optical case, v* being
now one of the r optical frequencies of the crystal lattice and
86 C . V . R a m a n and N. S~ N a g e n d r a N a t h

;~* is the wave-length of the periodic stratifications of electron density resulting

from the lattice vibrations of that frequency excited in the crystal by the
incident X-rays. u* is, of course, much larger than in the acoustic case,
but even so, it is small compared with the X-ray frequency. Hence, though
the change of frequency in the act of reflection indicated by equation (1) is
an essential part of the phenomenon, we may without sensible error take
;~= ~' in equation (2). We shalt, however, retain the ascent in ~' to indicate
that the vector 1/;V refers to the direction of the reflected X-ray. In the
acoustic case, the rector 1/;~* has an arbitrary direction and an arbitrary
magnitude subject to only a maximum limit, with the result that the vector
l/1' representing the direction of the scattered X-rays has an arbitrary
direction lying within a certain cone. In the optical case, however, the condi-
tions are different, as we shall presently see, with the result that the vector
1/;~' is restricted to specifiable directions and the observed effect is in the
nature of a well-defined reflection and not an irregular scattering.
We shall first consider the particular case in which the optical vibrations
excited in the cells of the crystal lattice are everywhere of identical phase;
in other words, the phase-waves of the lattice vibrations have an infinite
wave-length (ZX-----~). As already remarked, the possibility of optical
excitation of the lattice vibrations depends on the fact that these vibrations
produce time-periodic variations in the Fourier components of the space
distribution of electron density, in other words, produce periodic pulsations
of the structure amplitudes of the crystal. When /X = ~ the wave-length
of these pulsations is the same as that of the crystal spacings, o r a harmonic
thereof, and the wave-fronts also run parallel to the crystaI spacings.
Accordingly, we may write
1 11
;~* - - d ( 4)

where n is an integer (1, 2, 3, 9 9) and lq is a vector normal to any

chosen set of crystal spacings, its magnitude being the reciprocal of that
spacing. Combining (2) and (4), we have

1 1 n
X; = d (5)
Representing this graphically, (Fiff. 2), we derive the formula 2 d sin 0 = nA,
0 being the glancing angle of incidence and of reftection. Equation (5)
thus shows that when the phase-wave length 91 of the lattice vibrations is as-
sumed to be infinite, the quantum reflection of X-rays with diminished frequency
occurs under the same geometric conditions as the classical X-ray reflections.
 Quanlum Theory o{ X-Ray Re#ecUon and Scatterin~~Z 87

FIG. 2

The case when A is finite corresponds to a lattice vibration in which

the phase of the atomic vibrations alters progressively from cell to cell.
The phase of the pulsation of structure-amplitude will therefore also alter
from eell to cell, and in general, both the direction of the wave-fronts along
which this phase is constant, and the spacing between successive such wave-
fronts will differ from the crystal spacings. These quantities are indicated
by the rector 1/;~* and the reciprocal of its magnitude. We may readily
find the relation between ),*, the lattice wave-length A and the crystal spacing
d/n (see Fig. 3).
A A A A

/
,_:/ ........ / /
:~-.~......... _ / - - - ...... _7---- .... ::~---~.--_

Fta. 3

The crystal spacings ate a set of equidistant planes marked d/n in the
figure and the phase waves of the lattice vibrations are another set of planes
marked A crossing them at an angle as in Fig. 3. It is evident that the
planes along which the phase of the pulsations of electron density would be
constant would be the set of planes crossing both diagonally and marked h*
in the figure. The spacings of the three sets of planes ate evidently connected
by the rector relationship between their reciprocals (see note at the end of
the paper).
1 n 1
= 7i q- Z (6)
88 C . V . Raman and N. S. N a g e n d r a Nath

Combining (6) with (2), we have, in the most general case

...> ...> ...> ...>

1 1 n 1
A ~' = 3 + ~ (7)
Equation (7) reduces, as it should, to equation (5) when 91 is infinite.
It states the most general law of ti-Le quantum reflection of X-rays which
may be put into words thus: We combine vectorially the reciprocals of the
crystal spacbzg and of the wave-length of the optical vibrations of the crvstal
lattice. The reciprocal of the resultant gives the spacing of the dynamic strati-
fications of electron density and its direction that of their normal: these
dynamic stratifications gire a geometric reflection of the X-rays with diminished
frequency.
The optical vibrations of a crystal lattice have well-defined frequencies,
as is shown by the sharpness of the lines observed in the spectrum of the
light scattered by the crystal. The fact that the frequency shifts observed
in light scattering have not so far been found to depend appreciably
on the angle of scattering is also significant and shows that the lattice fre-
quencies are approximately independent of the lattice wave-lengths.
Regarding the obse?ved frequencies as due to the characteristic vibrations of
the unit cells in the lattice, it is evident that the frequency as observed with
an actual crystal would be N-fold degenerate, N being the number of lattice
cells in the fragment of crystal used. Actually, the N frequencies should be
regarded as forming an aggregate which is densest at the frequency corres-
ponding to ah infinite lattice wave-length ( 91 r and rapidly thins out as
A diminishes. These considerations enable us to understand the effects to
be expected on the basis of equation (7). A can always be considered as
large compared with d/n. Hence, the effect of its appearance in the right-
hand side of equation (7) on its scalar magnitude is a small quantity of
the first order. It serves, however, to alter the orientation of the wave-
fronts of the dynamic strati¡ with respect to the crystal spacings.
Indeed ti'lis would be its only effect ir the vectors 1/z~ and n/d are mutually
perpendicular, as the resultant of their addition would then be practically
n/d. In the limiting case when A is infinite, equatio n (7) shows that the
quantum reflection appears under the same conditions as the classical
re¡ and coincides with it in position. It is also then of maximum
intensity, as the N-fold aggregate of optical frequencies of the lattice is
densest for infinite values of A. As the crystal is tilted away from the
setting for the ordinary Bragg reflection, equation (7) shows that ZX must
diminish in magnitude, and the intensity of the quantum reflection must
therefore fall off in proportion to the diminished density of the aggregate
 Quantun,z Theory o / A'-Ray Re/qcch'an and Scctttcriny--/ 89

of frequencies for smaller values of A. As we shall see presently, the

q u a n t u m reflection also then separates out ti'oro the Bragg reflection and can
be observed by itself.
4. Geometric La,*,, of Quantum Reftection
The geometric interpretation of equation (7) is facilitated by first con-
sidering a case in which the vector 1/& besides being small compared with
n/d is perpendicular to it and lies in the same plane with it as the i n d d e n t
X-rays. In this case, it is evident that the result of combining 1/A and
nld vectorially is to tilt the reflecting planes in the plane of incidence with
respect to the crystal spacings but without any alteration of these spacings.
If therefore we c o n s i d e r a case in which the crystal-spacings are so inc!ined
to the incident monochromatic X-rays that they cannot give a Bragg
reflection, in other words, when
.+ .-> .-~
1 1 11
a ;V 4= d (8)
they would nevertheless gire a q u a t u m reflection according to the formula

1 n
a a' -- d' (9)
where d' has the same magnitude as the crystal spacing d, but its direction has
been altered so as to satisfy the vectorial relation (9). This relation
indicates that the incident X-rays gire a reflection with diministwd frequency
in the plane of incidence even wken the Bragg reflection is not possible, and tkat
the direction in wkich it appears witk reference to the incident X-rays is un-
affected by the setting of tke crystal, though as remarked above, its intensity
would fall off rapidly as the crystal is turned away from tke Bragg setting.
EquatŸ (9) is represented graphically in Fig. 4 for two cases in which
the crystal has been turned away f r o m the --tlrac"~_-e,setting in one direction or
\ R
q

FIG. 4
90 C . V . R a m a n and N. S. N a g e n d r a N a t h

the other. In both cases, the direction of the reflected ray OR is the same
and is given by the trigonometric formula
2 dsin 89( 0 + q~)= n A (10)
where 0 and ~ ate the glancing angles of incidence and re¡ with respect
to the crystal spacings.
The postulate made above which results in the formule (9) and (10),
namely, that the vectors 1/A and 1/d are perpendicular to each other,
physicatly interpreted, means that the lattice-waves excited by the X-rays
have wave-fronts perpendicular to the crystal spacings giving the quantum
reflection. There is good reason to believe that this must be actually the
case with most crystals. For, the atomic vibrations which ate effective in
altering the structure-amplitude of any particular spacing are those in which
the displacements are normal to the crystal spacing. If the lattice wave-
fronts are parallel to the crystal spacings, the phase of the atomic vibrations
would alter as we pass from one spacing to the next, whereas if they are
perpendicular to the crystal spacings, the phase of the vibration would
change along the crystal spacings but not perpendicular to them. The
forces brought into play between the neighbouring cells of the lattice by
such phase-changes would obviously be much greater in the former
case than in the latter. Accordingly, the N-fold aggregate of frequencies
would be spread out over a much wider range of frequencies when A and d
are parallel than when they are perpendicular. So far as the reflection of
X-rays is concerned, therefore, we would be justified in considering only the
case where A and d ate perpendicular and ignoring the case where they are
parallel. Equation (10) could then be regarded as the generally valid geo-
metric law for the quantum reflection of X-rays with diminished frequency.
It must be recognised, however, that there may be cases where the con-
siderations set forth above cannot reasonably be expected to be valid. It
may, for instance, be possible that the binding forces which hold the crystal
lattice together are of such a nature that though the atomic vibrations are
normal to the crystal planes under consideration, a change in the phase of
such vibrations along the crystal planes may evoke forces and disturb their
frequencies quite as m u c h a s a change of phase normal to them. In such a
case, the postulate that the vectors 1/A and 1/d are perpendicular may cease
even to be approximately true. In such cases, that is, when the effective
lattice-waves have an arbitrary inciination to the crystal spacings, an a!ter-
native principle suggests itself giving the selection rule for the vector l / A ,
namely, that the direction of the rector is such that its scalar magnitude
necessary to satisfy equation (9) is a minimum. Such a principle would not
 Qua~zLum Theor~,o/ X-Ra), RePeclion alzd ScaX[eri~zg--I 91
be unreasonable in view of the remark already made that the frequency-
aggregate of the optical vibrations becomes rapidly less dense for increasing
values of A. It can readily be shown that this p¡ gives the same
resultao equatŸ (10) when 0 and ~91are nearly equai to each other, that is to
say when the setting of the crystal is not far from the Bragg position. More
generally, however, it deviates from that result.

o~

o ~
FIG. 5

Fig. 5 represents equation (7) graphically for two cases in which the crystal
has been turned away from the Bragg setting in one direction or the other,
and A is chosen to have the minimum necessary length required to satisfy
the equation. The direction of the reflected ray is OR1 in one case and OR2
in the other. It will be noticed that the effective resulting spacing is in one
case greater than din, and in the other case less than d/n. At the correct
Bragg setting, therefore, OR! and OR., become coincident. The geometric
law of the quantum reflection is seen from Fig. 5 to be
d sin (0 + r = 17~, cos r (11)
Equations (10) and (11) give results which differ the more widely, the
greater the difference between 0 and r but become identical when 0 = 4'. It
will be noticed that they correspond respectively to the results which have been
found experimentally valid for the cases of sodium nitrate and diamond.
The divergence between the results given by the two formul~e when 0 =r r
and the fact that they have been derived on quite different considerations
suggests that when the crystal is tilted away from the correct Bragg setting,
the definition of the quantum reflection should progressively deteriorate.
Such ah effect is actually observed, but whether it is due solely to the cause
stated or is due partly also to other disturbing effects, e.g., thermal agitation
or the mosaic structure of the crystal, is a matter for further investigation.
92 C . V . R a m a n and N. S. N a g e n d r a N a t h

Summary
When X-rays rail upon a crystal, the charactefistic vibrations of the
crystal lattice may be excited thereby, in much the same way as in the
p h e n o m e n o n of the scattering of light in crystals with diminished frequency,
the excitation being a q u a n t u m mechanical effect. F r o m the equations for
the conservation of energy and m o m e n t u m , the geometrical relations enter-
ing in this effect ate deduced theoretically for the two cases in which the
lattice vibrations fall within (1) the acoustic range of frequency and (2) the
optical range. In the first case, the incident X-rays ate scattered in directions
falling within a c o n e having the incident ray as axis and with a semi-vertical
angle 2 sin-~ aq )t* where A"~*is the m Ÿ acoustical wave-length. In the
second case, we have a quantum-mechanical reflection of the X-rays with
diminished frequency in a direction which generally follows the geometric
formula 2 d sin 89(0 + ~b)= n ~ where 0 and <k are the glancing angles of inci-
dence and reflection on the crystal spacings. For crystals with specially rigid
bindings, the alternative fomuta d sin (0 + ~b) -- n ~ cos ~ is indicated as being
more appropriate. In either case, the intensity of the reflection should fall
off rapidly as 0 and ~b diverge.

Note : - - E q u a t i o n (6) may be very simply dr from the r shown in Fig. 3.

The vector sum o f the two sides of a triangle is r to the third side. Dividing this equation
by the atea of the triangle, we gr a rector relation between the rer162 of the three perpen-
diculars from the vertiees on the sides.