Home Search Collections Journals About Contact us My IOPscience

CRYSTAL OPTICS WITH ALLOWANCE FOR SPATIAL DISPERSION; EXCITON THEORY.

II

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1963 Sov. Phys. Usp. 5 675

(http://iopscience.iop.org/0038-5670/5/4/R12)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 35.8.11.2
This content was downloaded on 28/11/2013 at 06:00

Please note that terms and conditions apply.

SOVIET PHYSICS USPEKHI VOLUME 5, NUMBER 4 JANUARY-FEBRUARY 1963

CRYSTAL OPTICS WITH ALLOWANCE FOR SPATIAL DISPERSION; EXCITON THEORY. II

V. M. AGRANOVICH and V. L. GINZBURG

Usp. Fiz. Nauk 77, 663-725 (August, 1962)

CONTENTS
Introduction. 1. Complex Dielectric Tensor еу(о),к) and Normal Waves in the Medium,
a) The Tensor ejj(w, k) and Its Properties, b) Normal Electromagnetic Waves in a Medium.
Transverse and Longitudinal Waves. " F i c t i t i o u s " Longitudinal Waves and "Polarization
Waves", c) Energy and Other Relations for Waves in an Anisotropic Medium. 2. The Tensor
eij(w.k) in Crystals, a) The Concept of the Tensor ец(ш, к) for Crystals, b) The Case of
Weak Spatial Dispersion (а/Л « 1 ) . Cited literature, part I.
3. Crystal Optics with Allowance for Spatial Dispersion 675
a) New Wave near an Absorption Line in Gyrotropic Crystals 676
b) New Waves in Non-gyrotropic Crystals 679
c) Optical Anisotropy of Cubic Crystals. Quadrupole Absorption Lines 681
d) Effect of Mechanical Stresses and External Electric and Magnetic Fields 688
e) The Problem of Boundary Conditions 694
f) Experimental Investigations of Effects of Spatial Dispersion in Crystal Optics 695
4. Quantum Mechanical Calculation of the Tensor ejj( w, k) 698
a) Quantum Mechanical Expression for eij(w, k) 698
b) Mechanical Excitons and the Tensor е^(ш, к) in Molecular Crystals and in the Case of
the Classical Oscillator Model 701
c) Absorption Mechanism and Calculations 705
Concluding Remarks 706
Cited Literature, part II 708

3. CRYSTAL OPTICS WITH ALLOWANCE FOR although pertaining to crystal optics with allowance
SPATIAL DISPERSION for spatial dispersion, has been carried out long ago
and discussed in detail in the corresponding mono-
V>RYSTAL optics with allowance for spatial disper- graphs (see, in particular, E 2 2 ! 2 8 . 3 1 ]). Consequently,
sion deals with the propagation, reflection, and refrac - with respect to gyrotropic crystals, we shall dwell
tion of various normal waves in crystals, using the only on one question of the new waves, a question
tensor ejj(w.k). which as far as we know arose only recently. ^
Formally speaking, the scope is broader here than Second, even if we disregard spatial dispersion (in-
in the case of classical crystal optics, inasmuch as cluding gyrotropy), the analysis of the propagation
many new problems and questions arise (for example, of light in absorbing crystals, particularly in the case
22 28
it becomes necessary to investigate the optical aniso- of low symmetry, is rather cumbersome C > 3. Fur-
tropy of cubic crystals). Actually, however, the situ- thermore, there are special cases. These include the
ation is different, primarily because of the smallness propagation of light along singular optical axes C28>32.32a]>
of the spatial dispersion. Because of the latter, as when we can no longer confine ourselves to the consid-
was already emphasized above, it becomes necessary eration of plane waves of the type (1.13). The role of
to consider only problems in which the spatial disper- spatial dispersion in the case of absorbing crystals,
sion leads to qualitatively new effects or, at any rate, and in the case of complex к in general, was not in-
does not give r i s e to only negligible corrections to the vestigated in any degree of detail.
formulas of classical crystal optics. Third and last, it must be emphasized that only rela-
ш connection with the foregoing, the following dis- tively few investigations have been devoted to an ac-
cussion of several crystal-optical problems is frag- count of spatial dispersion in a nongyrotropic medium
mentary in character and reduces essentially to a dis- [or, more accurately, to the account of second-order
cussion of several phenomena. It is necessary to bear effects proportional to (a/Л.)2] as applied to the optics
in mind here also the following circumstances. First, of crystals, even for a transparent or almost transpar-
the study of the natural optical activity (gyrotropy), ent medium. In other words, although the question of
the effects that are proportional to (a/A.)2 cannot by
*Part I of the article, containing the introduction and Sections any means be regarded as being new (see the Intro-
1 and 2, was published in UFN 76, 643 (1962), Soviet Phys. Us- duction), it has for various reasons remained until
pekhi 5, 323 (1962). recently in limbo. It can thus be assumed that not all

675
676 V. M. A G R A N O V I C H a n d V. L . GINZBURG
the interesting aspects of crystal optics with account investigation of the tensor е ^ ( ш , к ) , and no real sig-
of spatial dispersion have already been noted and dis- nificance need be attached to the corresponding solu-
cussed. Consequently, the relatively narrow framework tions. As regards the roots n t and n 2 , they can be
2
within which we discuss crystal optics with account of determined by replacing n in the right half of (3.3) by
spatial dispersion is due not only to the smallness of the product n 0 1 n 0 2 . It then follows from (3.2) that in a
this dispersion, but also to the present status of r e - wave with refractive index щ we have
search in this field.
(3.4a)
a) New Wave Near the Absorption Line in a Gyro-
tropic Crystal. As already indicated, in gyrotropic while in a wave with refractive index n2
crystals the spatial dispersion manifests itself in
terms of first order of smallness in a/\. Therefore (3.4b)
we can omit in expansion (2.13) the last term in the
right half of the equation. Using relations (1.6) and where
(1.20) and choosing the direction of the wave vector as __ с l—(Л L
the z axis, we obtain the following expressions, which
are satisfied by the components of the transverse in-
(3 4c)
1
duction vector D' (see also I- ^, Section 82): -

Since, generally speaking, n 0 1 * n 0 2 , both waves turn

out to be polarized along ellipses whose principal axes
'i= — ш б ш — D'x. (3.1) coincide with x and y. These ellipses are thus turned
90° relative to each other, and their rotations are op-
(We can, of course, replace here 6j 2 3 by f33 = f 3 3 s 3 or, positely directed. If the vector s is directed along the
if invariant notation is used, by fjjsisj.) The direction
optical axis, when n 0 1 = n 0 2 = n and p = 1, then the el-
of the x and у axes is chosen along the principal
lipses degenerate into circles. Furthermore
axes of the two dimensional tensor e^n, a = x,y and
we denote the principal values of this tensor by l / n ^ and : = По ± (3.5a)
l/n(J2 (we leave out the caret above the n and assume
in most cases, unless otherwise stipulated, that n is and the angle of rotation of the plane of polarization <p
real, i.e., we are dealing with the transparency r e - following traversal of a path I is
gion). Then Eqs. (3.1) assume the following form:
(3.5b)
We note that in the case under consideration each of
(3.2) the two refractive indices corresponds to its own cir-
cular (left- or right-hand) polarization of the wave,
The condition that the determinant of this system van- since when p = 1 we have
ish gives a third-order equation in n 2 D'x = ± iD'y.

The results pertaining to this case in which only the

two roots n t and n 2 of Eq. (3.3) need be taken into
The roots of this equation determine, for a specified
account, are well known and we touched upon them
direction of s, three values of the refractive index
2 here only to emphasize the limits of applicability of
щ, n 2 , and n 3 (we always put n = / n , since the root
these results, and also to compare them with the more
n = — Vli2 merely denotes the reversal of the sign of
complicated situation which may occur near resonances.
s).
In this case the values of n ^ and n(j2 (or of only one of
In the investigation of Eq. (3.3) we shall distinguish
them) are large in a certain region of frequencies, and
between frequency regions away from and close to reso-
all three roots of (3.3) correspond to relatively long
nance. At frequencies sufficiently far away from the
wavelengths, so that all three solutions (as was noted
resonances, the right half of (3.3) is small. Conse-
in tfi ) can be regarded within the framework of the
quently one of the roots, (for example n 3 ) , will be
macroscopic approach. Depending on the frequency,
very large,
Eq. (3.3) has in the absence of absorption either three
real roots, or one real and two complex roots (Fig. 3).*
»
To be specific, we shall assume that 6j 2 3 > 0 and that
Indeed, inasmuch as б ш ~ a ~ 10~3 Л.о, the order of n 3 the wave propagates along the optical axis, i.e., n 0 1
is 103, and the corresponding wavelength in the medium
is \ = Л 0 /п 3 ~ 10~7—10"8 cm. This means that in the *Fig. 3 is taken from M, where it was assumed that (for со = col)
optical frequency band waves with refractive indices (0 2<o?
n 3 for crystals can usually not be considered in the —
<oa — c
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL DISPERSION 677
3
| ы т — w; | /u)j ~ 4 x 10 , which c o r r e s p o n d s when
1 1
WJ я 3 x 10* c m " to | u ) m — uij | ~ 100 c m " . I n a s -
m u c h a s t h e widths of t h e exciton a b s o r p t i o n l i n e s in
33
many, say, m o l e c u l a r c r y s t a l s '- -' a t h e l i u m t e m p e r a -
t u r e s a m o u n t t o s e v e r a l t i m e s 10 r e c i p r o c a l c e n t i -
m e t e r s , t h e r e a r e all g r o u n d s for a s s u m i n g t h a t t h e
" t h r e e - w a v e e f f e c t " in such c r y s t a l s should b e quite
c l e a r l y pronounced at low t e m p e r a t u r e s .
We now stop t o d i s c u s s t h e c h a r a c t e r of t h e s o l u -
t i o n s of t h e field equations when t h e d i s p e r s i o n e q u a -
tion h a s m u l t i p l e r o o t s . The well-known and fairly
frequent c a s e of m u l t i p l e r o o t s c o r r e s p o n d s t o d e g e n -
e r a c y , n a m e l y t o equality of t h e r e f r a c t i v e i n d i c e s
2
n(ш, s ) for w a v e s having different p o l a r i z a t i o n (for
= n 0 2 = n 0 . In this case the values of n on the upper specified ш and s ) . This t a k e s p l a c e for t r a n s v e r s e
2
branch satisfy the inequality n > n§. Therefore, as w a v e s in an i s o t r o p i c m e d i u m , and a l s o in s o m e d i r e c -
follows from (3.2) and (3.3), for the upper branch the t i o n s for w a v e s in an a n i s o t r o p i c m e d i u m . In such
following relations a r e satisfied c a s e s we c a n c h o o s e two l i n e a r l y independent s o l u -
t i o n s of t h e type E/ = Е о д exp [ — i w ( t — ( n / c ) S ' Г ) ] ,
which differ in t h e v e c t o r s E o j , i . e . , in t h e p o l a r i z a -
t i o n . Other c a s e s o c c u r , however, when t h e m u l t i p l e
Thus, the upper b r a n c h of the s o l u t i o n s n2 corresponds
r o o t п(ш, s ) c o r r e s p o n d s to only one solution (we
to the right-hand c i r c u l a r polarization, and the lower
r e f e r to a double r o o t ) . This i s t h e situation in t h e
branch c o r r e s p o n d s to left-hand polarization. If 6 i 2 3 < 0 ,
c a s e when w a v e s p r o p a g a t e along t h e " s i n g u l a r optical
the p o l a r i z a t i o n s of the solutions c o r r e s p o n d i n g to the
a x i s " in a b s o r b i n g c r y s t a l s of t h e t r i c l i n i c and m o n o -
upper and l o w e r b r a n c h e s m u s t be r e v e r s e d . As indi-
clinic s y s t e m s Е28,32,зга] ( i n o r ( j e r for a m u l t i p l e r o o t
cated, the d i s p e r s i o n c u r v e s shown in Fig. 3 have been
to a p p e a r , it i s i m p o r t a n t that t h e p r i n c i p a l a x e s of
obtained in an a p p r o x i m a t i o n in w h i c h the a b s o r p t i o n is
the t e n s o r s е ц ( ш ) and efj(w) do not coincide, s o m e -
disregarded. The quantities n 2 , ( o r njjj and njj 2 ) and
thing t h a t o c c u r s only for t r i c l i n i c and monoclinic
6123 in Eq. (3.3) c a n then b e r e g a r d e d a s r e a l , i.e., t h i s
c r y s t a l s ) . F o r s i n g u l a r axes t o a r i s e , t h e r e i s no
equation has real coefficients. Consequently in the fre-
need for involving s p a t i a l d i s p e r s i o n ( s u c h a x e s a r i s e
q u e n c y r e g i o n w h e r e only o n e of the s o l u t i o n s i s real,
a l r e a d y in c l a s s i c a l c r y s t a l o p t i c s , and in t h e c a s e of
the two o t h e r s a r e c o m p l e x - c o n j u g a t e , and in the case
g y r o t r o p i c c r y s t a l s without account of t h e new w a v e s ) .
of a h a l f - s p a c e t h e y f o r m a standing wave correspond-
In t h e p r e s e n c e of s p a t i a l d i s p e r s i o n t h e m u l t i p l e r o o t s
ing to a z e r o Poynting v e c t o r ( s e e Section lc). Absorp-
a p p e a r even if t h e r e i s no a b s o r p t i o n . This i s i m m e d i -
tion c a n be allowed for by c o n s i d e r i n g n§ a n d 6123 com-
ately c l e a r from F i g . 3 for a g y r o t r o p i c c r y s t a l , and
plex. A s a r e s u l t the p a t t e r n of the d i s p e r s i o n curve
say from F i g s . 4b and 5b for n o n g y r o t r o p i c c r y s t a l s .
changes, particularly in the c a s e w h e n the turning point
Indeed, a s indicated above, t h e p o l a r i z a t i o n of the wave
(see Fig. 3) f a l l s in the r e g i o n of a p p r e c i a b l e absorp-
field i s t h e s a m e for all solutions c o r r e s p o n d i n g to t h e
tion. In t h i s c o n n e c t i o n w e n o t e t h a t m u l t i p l e roots
u p p e r b r a n c h of F i g . 3. Consequently, on a p p r o a c h i n g
(i.e., a turning point) c o r r e s p o n d to v a l u e s of the fre-
the " t u r n i n g p o i n t " ш = w m , n = n m , t h e two solutions
quency w m
satisfying the equation
(8), (10), and (11) of M that for light propagating along the optical
axis we have in this frequency region n20(\,) = 1 + К,Л^/(А^ - \]_),
and the rotation per unit length of ray path i s cp((A<,)// = K2A£/
(А„-А]_)2, where Ao is the wavelength of light in vacuum. A^is the
In this case we have a degenerate root wavelength corresponding to resonance, and K2 = 27r 2 K 2 S, 23 . Thus,
u>Sl2l/c = K2/n\,Kl. According to L67J in cinnabar crystals the ab-
2
sorption band at Aj_ = 4930 A corresponds to K, = 0.56 and
2 \з
K2 = 1.0677 x 10"" cm" 1 , from which we get (u>j_/c)Sm = (2тг/А1)б12,
= 0.8 x 10" J . For the organic crystal of benzyl the band Aj_ = 2400 A
2
and a s m a l l e r root n = n ^ /4. Consequently when corresponds to K t = 0.4 and K2 = 0.35 x 10'' cm' 1 . Therefore
(U)J^/C)S 1 2 3 ~ 10 3 . According to the same data, in the cubic sodium
9 0.2 о
chlorate crystal the band at Aj_ = 900 A corresponds to K, = 1.18
Пл(о)) с* -j г ш ; and w 6 1 2 3 / c ~ 10"° we o b t a i n *
and K2 = О.Обтт x 10~8 cm" 1 , and consequently {o>i/c)S12i » 4 x 10"%
while the band at Aj^ = 1850 A corresponds to K, = 0.08 and
*The quantity cu§ 123 /c can be estimated directly from experi- K2 = -0.077Г x 10" 9 c m ' 1 , s o that ( U > I / C ) 8 1 2 3 = - 0 . 5 x 10". We know
mental data by using t h e frequency dependence of the rotation of that for quartz away from the absorption bands (at A = 589.3 A) we
the plane of polarization and the refractive index of light far from h a v e <p/l = 217 deg/cm, n 0 = 1.54, and according to (3.5b) (<u/c)S 1 2 3
the absorption band under consideration. It follows from Eqs. (7), = 1.3 x 10" 5 .
678 V. M. A G R A N O V I C H a n d V. L . G I N Z B U R G

-ms -иш -от о mi aom OMOS 0.0)

— A-OJ, t-o, fi-ws

--- A-0,1, ff-p'-O

FIG. 5a

•ЦО!
-0.005 -0003 -0,001 0 001 0.003 0.005

FIG. 5b

c o r r e s p o n d i n g to t h i s branch, Ej = E O j i e x p [ — i w ( t —
( r i t / c J S T ) ] and E 2 = E 0 , 2 e x p [ - i w ( t - ( n / c ) s . r ) ]
tend to a s i n g l e solution E ^ = E^Jm e x p [ - i w m ( t -
(n m /c)s.r)].
U n d e r s u c h c o n d i t i o n s , a s i s w e l l known f r o m t h e
t h e o r y of l i n e a r d i f f e r e n t i a l equations with constant
coefficients, the s e c o n d linearly independent solution
has the form

[ w e m a k e a l l o w a n c e f o r t h e f a c t that t h e p o l a r i z a t i o n -tWOS -0,003 -0Ц0) 0 0M1 0103 I

$-W3, [!-Ш~*, A-0,1
i s t h e s a m e a l o n g t h e e n t i r e c o n s i d e r e d b r a n c h of t h e
function n ( u , B ) ; in the m o r e g e n e r a l c a s e the s e c o n d FIG. 5c
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL DISPERSION 679
is quite curious and somewhat unexpected. *
b) New waves in nongyrotropic crystals E7'5^. In
nongyrotropic crystals, the effects of spatial disper-
sion a r e due to the properties of the tensor /Зцйп»
inasmuch as in such crystals [see (2.13)] we have
ey' (со, к) = ву1 (со) + p t y i m (со) fc,fcm. (3.6)
Using this relation, and also (1.57) and (1.58), we ob-
tain
el'у (со, к) = tin e^l (со) r\mi + t) i r p r s l m (со) /c,fcmr)s,-
If, a s i n t h e p r e c e d i n g s e c t i o n , w e a l i g n t h e z axis with
the d i r e c t i o n of the w a v e v e c t o r s , the e x p r e s s i o n for
e 1
e II 1 ii j simplifies:

-m -0.005 -0.003 -Д00! 0 0.001 0003 0.005 e j j i y (со, k ) = TI U еГш (со) r\mj + T\iT$rs33 (со) TISJ- k2
S-IO'3, JS'~1O

FIG. 5d
£ = 1,2,3. (3.7)
s o l u t i o n w i l l b e m o r e c o m p l i c a t e d but a l s o p r o p o r t i o n a l
to t h e c o o r d i n a t e s x , y , and z * ] . T h e f o r e g o i n g p e r - One must not forget here, of course, that in (3.7) the
t a i n s , of c o u r s e , a l s o t o t h e s i n g u l a r o p t i c a l a x e s ( s e e components of the tensors ец(ш) and /Зц2т( ш ) de-
[32,32a])_ j^^ e i a t t e r c a s e , a non-exponential wave pend on the direction of s = k/k by virtue of the choice
c a n a c t u a l l y p r o p a g a t e in t h e c r y s t a l . For multiple of the coordinate system.
r o o t s c o r r e s p o n d i n g to the "turning p o i n t s , " for e x - Inasmuch as the vector D' is transverse, we obtain
a m p l e on Fig. 4b, the situation i s different, s i n c e a l - a system of equations similar to the system (3.1):
l o w a n c e for a b s o r p t i o n e l i m i n a t e s t h e m u l t i p l e r o o t n
( F i g . 5d). One m i g h t think that f o r m o r e c o m p l i c a t e d
c a s e s in c r y s t a l o p t i c s w i t h a l l o w a n c e for s p a t i a l d i s -
p e r s i o n ( a n d p o s s i b l y a l s o in a c o u s t i c s ) o n e a l s o e n -
c o u n t e r s m u l t i p l e r o o t s of t h e d i s p e r s i o n e q u a t i o n
w h i c h e x i s t in t h e p r e s e n c e of a b s o r p t i o n ( w e r e f e r £?„».. И п
to m u l t i p l e r o o t s w i t h i d e n t i c a l p o l a r i z a t i o n and y e t
different from the r o o t s c o r r e s p o n d i n g to the singular (3.8)
a x e s , w h i c h a p p e a r e v e n w i t h o u t a c c o u n t of s p a t i a l d i s -
p e r s i o n ). We s h a l l not d e a l i n w h a t f o l l o w s w i t h t h e If we choose furthermore the x and у axes along the
p r o p a g a t i o n of w a v e s w h e n t h e d i s p e r s i o n e q u a t i o n h a s principal axes of the two-dimensional tensor e ^ o , a,
multiple roots. We t h e r e f o r e m a k e o n l y o n e o t h e r r e - /8 = x, y, then the equation for the determination of n 2
m a r k . It i s u s u a l l y a s s u m e d that i n a h o m o g e n e o u s will assume the form
m e d i u m it i s s u f f i c i e n t t o c o n s i d e r s o l u t i o n s of t h e
type Eo exp [ - i w ( t - ( n / c ) s « r ) ] or E o e x p [ i ( k - r
J L V - L V
— a i t ) ] . F r o m t h i s point of v i e w , i n d e p e n d e n t l y of t h e
I 1 _

q u e s t i o n of w h e t h e r w a v e s c a n b e o b s e r v e d f o r m u l - » 1 " n l

t i p l e r o o t s and i d e n t i c a l p o l a r i z a t i o n , t h e v e r y n e e d
for c o n s i d e r i n g in s u c h c a s e s , g e n e r a l l y s p e a k i n g , a l s o "t~ .4 ^ {PxXZZryyzZ PxyZ ) (3.9)
t h e s o l u t i o n s of t h e t y p e

F.<a> __F4.> / „ \ „ - * ш ' " ( ' - = ? - в г ) *Incidentally, one case of multiple roots for specified polariza-
tion is encountered even in the simplest problem of propagation of
transverse waves in an isotropic medium. The corresponding wave
*In the case when the polarization is constant, i.e., E o is inde- equation is (here E = E x or E = E y , for more details see for exam-
pendent of ы for a given s, we arrive at the indicated second solu- ple W): d2E/dz2 + u>2eE/c2 = 0, and if a2 = e = 0 then
tion by considering the following solution near the turning point: E = E^ 1 ' + Ep^z. Here, however, the roots coalesce for waves propa-
gating in different directions (with n Ф 0) and, in addition, if ac-
) • count of absorption is taken we already have e = e' + i e " = n 2 ^ 0.
We note also that in a magnetoactive plasma in the presence of ab-
In the limit as n 2 -> n, we obtain from this a solution of the type sorption (but even without account of spatial dispersion) we know
of one interesting case of the appearance of multiple roots with
4 - E<^> = E^» = const (sr) E<,!>.
dn identical polarization (see L2J, Sections 11 and 28).
680 V. M. A G R A N O V I C H a n d V. L . GINZBURG
8
We consider first the simplest case of an isotropic x 10 cm, i.e., so long as n 2 « A.0/a. We shall hence-
medium [see (2.33)]. In this case, independently of forth assume this condition satisfied.
the direction of s, we have n 0 1 = n 0 2 = n 0 , j3Xyzz = 0.» Account of spatial dispersion near the absorption
2
Aoczz = /fyyzz = /3. Equation (3.9) then breaks up into line introduces even qualitative changes in the n (w)
two identical equations each of the form curves, as is clear from Figs. 4a and b. Both figures
correspond to the case A = 1 and 6 = 0, but the values
(3.10) 5
chosen for Figs. 4a and 4b are /3' = 10" and /3'
5
= - 1 0 " , respectively. The limiting curve (3.13) with
From this we obtain immediately
A = 1 is shown dashed in both cases (inasmuch as
usually e 0 0 ~ 1, and we are interested in the region
i.2 = ц i O ' + F -
( З Л 1 )
| e01 » 1, we put for the sake of simplicity e 0 0 = 0
where everywhere in Figs. 4a and b ) . We know that in the
2
absence of absorption and when n is real it is obvious
2
that the medium is transparent when n > 0 and total
Bearing in mind the region of frequencies close to reflection of the wave by the medium occurs when n
2

s o m e single absorption line, we obtain the following 2

= - к < 0. When /3' = 0 and /3' > 0, one of these cases
e x p r e s s i o n for € 0 : 2
occurs, since ii is real. But if fi' < 0, then the values
2 2
of n are complex in the region | e 0 1 > 1/2V j /3' | [n
(ffl) = 600 2
= (n + i/c ) even in the absence of absorption]. In the
(со —со,) _J_ case of an exciton absorption line, the sign of /3' is
k
(3.12)
determined by the sign of the "effective m a s s " of the
H e r e e and m a r e t h e c h a r g e and m a s s of t h e e l e c - mechanical exciton. In order to verify this, it is suf-
t r o n , v is the " e f f e c t i v e " frequency of the collisions ficient to consider in place of (3.12) the more general
that cause the attenuation of the waves in the medium, expression for e, in which both temporal and spatial
N e ff/N is the o s c i l l a t o r strength, with N the total dispersion are taken into account:
number of e l e c t r o n s p e r unit volume and N e ff the e (со, k) = (3.16)
fraction of these e l e c t r o n s which "effectively" d e - со|"(к) — со2—i'cov

t e r m i n e s the optical p r o p e r t i e s of the medium in the It i s p r e c i s e l y t h i s e x p r e s s i o n (3.16) w h i c h i s obtained

region of the s p e c t r u m under consideration. under s i m p l e r a s s u m p t i o n s by s t a r t i n g f r o m the m i c r o -
In t h e absence of absorption, when 6 = 0, we get scopic theory ( s e e Sec. 4 ) . By expanding the energy
of t h e " m e c h a n i c a l e x c i t o n " Kwj(k) i n p o w e r s of t h e
(«) = К = 8oo — - (3.13) w a v e v e c t o r w e h a v e i n t h e c a s e of an i s o t r o p i c and
nongyrotropic medium
where nj| is the refractive index neglecting not only
absorption but also spatial dispersion. It is clear from (3.17)
(3.11) that if б = 0 and 2m,,

(3.14) w h e r e m e x c i s t h e " e f f e c t i v e m a s s " of t h e m e c h a n i c a l

exciton. Therefore for small | ш — a>j (0) | we have
we can put
wj (0) (3.18)
Е(со, к) Е0(со) 4ne*Neff/m техс
(3.15)
and in a c c o r d a n c e w i t h (3.6)
When /3' ~ 10~6, condition (3.14) assumes the form
3
njj « 10 or else it reduces when e 0 0 ~ 1 and A ~ 0.1 „ H (°) m
(3.19)
[ see (3.13)] to the inequality | £ | = | ш - wj | /WJ » 10"*.
In the optical (visible) region of the spectrum, where T h u s , t h e s i g n o f /3 a n d /3' [ s e e ( 3 . 1 1 ) ] a c t u a l l y c o i n -
COJ ~ 10 15 s e c " 1 ~ 2 x 104 cm" 1 , this means that expres- c i d e s i n t h e a p p r o x i m a t i o n o f ( 3 . 1 6 ) w i t h t h e s i g n o f
sions (3.15) become valid even at a distance Aw the " e f f e c t i v e m a s s " of the m e c h a n i c a l exciton.
= | w - WJ | » 10" 4 wj ~ 2 c m " 1 from the center of the The p o s s i b i l i t y of o b s e r v i n g the new w a v e n e a r r e s o -
absorption line. If we assume even A ~ 1, which ap- onance d e p e n d s to a c o n s i d e r a b l e d e g r e e on the extent
parently is not realistic, we arrive at the inequality o f a b s o r p t i o n . In t h e a b s e n c e o f a b s o r p t i o n t h e i n f l u -
Aw» 20 cm" 1 . e n c e of the s p a t i a l d i s p e r s i o n i s l a r g e e v e n in the r e -
In this frequency range, obviously, n? я n2,. On the g i o n w h e r e 4 e g | /3' | ~ 1, i . e . , w h e n | | I ~ £ k = 2 A V | / 3 ' | .
other hand, the root n 2 is very large and when e 0 ~ 1, F u r t h e r , if
as is the case far away from the line, | n | | ~ 1/| /3' |
~ 10 6 . In addition Л = A.0/n2 ~ 5 x 10" 8 cm and expres- (3.20)
sion (2.13) is no longer valid. The new root n 2 of the
dispersion equation has therefore real significance only t h e a b s o r p t i o n i s s m a l l f o r f r e q u e n c i e s w i t h | £, \
near the line, in the region where X = \0 /n 2 » a ~ 3 = | и — Wi | /WJ Z Ijj, i.e., it changes little the m a g n i -
 CRYSTAL OPTICS WITH A L L O W A N C E FOR S P A T I A L DISPERSION 681
tude of Re e0, and at the same time | bn e0 | « | Re €( |. on the frequency with account of spatial dispersion does
Assume by way of an example that condition (3.20) not have a Lorentz form, just as the dependence of
is satisfied and that we are considering a frequency n 2 (w) no longer follows the Drude-Sellmeier formula.
for which 4ejj | /3' I = 1. For | < 0 we then have n However, at small values of fi' and 6 ^ 0 , the func-
« Vl + V2~(/3')~l/4 and к = 1 / 6 (6/£k)/3" l / 4 which yields tions n 2 (w) and к(ш) resume their usual form. We
n ~ 50 and к ~ 5 x 10" 4 for fa « 10~3, /3'ra10" 6 , and note in this connection that whereas A is, in accord
б и 10" 7 . Inasmuch as the intensity of an individual with (3.12), directly proportional to the transition os-
wave attenuates as I = I o exp [ - 2oi/cz/c ] = Io exp [ - \a ], cillator strength, ]3 is practically independent of the
this means that the absorption coefficient is fj, transition oscillator strength [see (3.19)], inasmuch
= (2ы/с)к ~ 150 c m " 1 (for Ло ~ 4000 A). For the as the "effective m a s s " of the mechanical exciton is
same values of the parameters, but for /3' = —10~6, roughly speaking inversely proportional to the transi-
we get n = 1/V2e o |is'| = |/S' | " 1 / 4 « 3 0 , к « ( 6 / £ k ) 1 / 2 / tion oscillator strength (we have in mind "mechanical
2 | jS' | 1 / 4 ra 0.15, and ц = 2шк/с ~ 4 x 104 cm" 1 . In this excitons" which can be excited in a dipole transition;
example the intensity decreases by a factor e over a s e e '--•).
path equal approximately to 3 x 10" 5 cm ~ \ 0 , whereas According to (3.19), putting for t h e indicated r e a s o n
X = \ 0 / n ~ 10~6 cm. Thus, when б ~ 10" 7 the attenua- N e f f m e x c / m ~ N, we have
tion in a wavelength in the medium is not so large even a
1
со
—
2

ft _
ba (0) to
• ? /
2

—
in the second case.
Actually, however, for dipole lines the value of б
where w0 = V47re2N/m is the " p l a s m a " frequency. In
for crystals is much greater than 10~7. In particular,
molecular crystals Nra3 x IO 22 — 3 x IO 23 and w2, « 8
in all the investigated molecular crystals, where the x 10 31 — 8 x 10 3 2 sec" 2 , so that when U J » 3 X 10 15 s e c " 1
dipole exciton lines are particularly clearly p r o - we have /3' ra 10~6 —10~ 7 . At such small values of /3',
nounced £33^, the value of б tends as T — 0 to a value absorption assumes a much more important role com-
k, 10~3. This causes the inverse inequality pared with the case /3'га10" 5 considered above, so
S > lk = 24 V\ p' | (3.21) that to observe spatial dispersion effects near dipole
absorption lines it is necessary to produce very spe-
to be satisfied in place of (3.20), and all the deductions cial and favorable conditions. We note that the obser-
of the dispersion theory constructed without account of vation of the effects of spatial dispersion near quadru-
absorption must be reviewed. pole lines is appreciably facilitated by the fact that the
Using (3.11) and (3.12), we obtain* width of these lines turns out to be at sufficiently low
temperatures several orders of magnitude lower than
the width of the dipole absorption lines.
(3.22)
In the analysis of the question of the new waves near
where quadrupole absorption lines ^41-' we must use within the
framework of the phenomenological approach expansions
of the type (2.14)—(2.15). We shall return to this ques-
7»/=[(g2-62+4p'.A2)2+462£2]'/4cos(p, tion in Sec. 3c.
c) Optical Anisotropy of Cubic Crystals. Quadrupole
JV = [(g2 - 62+4p'A2)2+462g2]'/4 sin q>.
Absorption Lines. We have already stated in the Intro-
In relation (3.22) it is necessary to take the arithmetic duction that optical anisotropy of cubic crystals*, which
value of the root, while the angle <p must be deter- was considered theoretically in Сз-5,40,41]^ h a s ь е е п ob-
mined in accordance with the signs of ( £ 2 - 6 2 + 4j3'A2) served experimentally '-6-' (Cu 2 O crystal at low temper-
and 26£; here atures near the quadrupole transition Л. = 6125 A).
The theory of quadrupole transitions in cubic crys-
.'A*) + 2i&l = Qei<v, tals was recently developed, on the basis of the exciton
where concept, in articles t 3 6 " 3 9 ] , where the question of the
influence of stresses and external fields was also con-
sidered. Inasmuch as the exciton wave functions are
The calculated values of n(o>) and K(W) for different generally speaking unknown, it was possible to cast
values of /3', A, and 6 are clearly illustrated in Fig. 5. light in C 36 " 39 ] on only a few aspects of the phenomenon.
It follows from the curves given there that at suffi-
ciently large values of б the anomalous wave c o r r e - *We do not consider here the so-called "latent optical aniso-
sponds to very large damping. On the other hand, the tropy" of cubic crystals, arising because of the possible presence
dependence of the coefficient of absorption of the waves in the crystal of anisotropic centers connected with local lattice
defects (for example, various types of color centers, etc.). If
spatial dispersion is neglected and there are no directional ex-
*We put here eo(&)) ~ -A/(f + iS), which is true only in the ternal influences, cubic crystals with anisotropic centers (impuri-
vicinity of a resonance. ties) remain optically isotropic (for more details see L*SJ).
 682 V. M. A G R A N O V I C H a n d V. L . G I N Z B U R G
In p a r t i c u l a r , t h e p o l a r i z a t i o n of q u a d r u p o l e t r a n s i t i o n s Pi — Pxxxx — Vyyyy — Pzzzzi P2 — Pxxyy — Prazz — Pyi/zzi
a s a f u n c t i o n of t h e d i r e c t i o n of l i g h t p r o p a g a t i o n w a s = =
Рз Pxyxy Pxzxz = PyzyZ'
e s t a b l i s h e d , a l o n g w i t h the c h a r a c t e r of t h e a n g u l a r d e -
p e n d e n c e of the a b s o r p t i o n i n t e n s i t y . This l e a d s to the following s y s t e m of equations for the
U n d e r t h e i n f l u e n c e of e x t e r n a l e l e c t r i c and m a g - c o m p o n e n t s of the v e c t o r D ' [ s e e (1.20)]:
n e t i c f i e l d s and m e c h a n i c a l s t r e s s e s , t h e c r y s t a l s y m -
m e t r y i s g e n e r a l l y s p e a k i n g l o w e r e d and t h e d e g e n e r a t e A-D[ = ( i - D[ (3.23)
exciton l e v e l s may split. The c h a r a c t e r of t h i s s p l i t t i n g
e t c , w h e r e a c c o u n t i s t a k e n of the condition s i D i = 0
v a r i e s w i t h the e x c i t o n s t a t e and d e p e n d s o n t h e s y m m e -
and
t r y of t h e d i s t u r b a n c e . This c i r c u m s t a n c e m a y be e m -
p l o y e d q u i t e e f f e c t i v e l y t o e s t a b l i s h t h e s y m m e t r y of P = P;-Pa-2p;. (3.23a)
q u a d r u p o l e e x c i t e d s t a t e s of a s y s t e m o n t h e b a s i s of
It f o l l o w s f r o m (3.23), n a t u r a l l y , that D ' • s = 0. Let us
e x p e r i m e n t a l data.
consider several particular cases.
O p t i c a l a n i s o t r o p y i n c u b i c c r y s t a l s m a y , of c o u r s e ,
a) W e a s s u m e t h a t t h e v e c t o r s is directed along
m a n i f e s t i t s e l f n o t o n l y i n t h e r e g i o n of q u a d r u p o l e t r a n -
one of the cube e d g e s , for e x a m p l e along the z axis.
s i t i o n s , but a l s o i n t h e r e g i o n of d i p o l e t r a n s i t i o n s . The
Then the equation for the c o m p o n e n t s of the v e c t o r D '
t h e o r y of t h i s q u e s t i o n w a s c o n s i d e r e d p h e n o m e n o l o g i -
simplifies to
c a l l y in ^ and t h e n , w i t h i n t h e f r a m e w o r k of e x c i t o n
40
t h e o r y * , i n '- -'. For the s a m e r e a s o n a s above (the
e x c i t o n w a v e functions a r e unknown) only a few a s -
p e c t s could be explained in this m a n n e r in ^40^ a s w e l l f r o m w h i c h it f o l l o w s that in the c a s e u n d e r c o n s i d e r a -
a s i n M , n a m e l y t h e p o l a r i z a t i o n and t h e n u m b e r of tion fi2 i s independent of the polarization of the v e c t o r
i n d e p e n d e n t w a v e s a s a function of t h e d i r e c t i o n of l i g h t D' and i s d e t e r m i n e d by the equation
p r o p a g a t i o n ; it w a s a l s o p o s s i b l e t o d e t e r m i n e t h e
c o u r s e of t h e d i s p e r s i o n n e a r t h e e x c i t o n a b s o r p t i o n
b a n d s w i t h a c c u r a c y t o t h e unknown o s c i l l a t o r s t r e n g t h s ,
An equation of this type w a s a l r e a d y c o n s i d e r e d e a r l i e r
t h e " e f f e c t i v e m a s s " of t h e e x c i t o n , e t c .
(see Sec. 3b).
We s h a l l s h o w b e l o w that an a c c o u n t of t h e s p a t i a l
b) W e n o w a s s u m e t h a t t h e v e c t o r s is directed
d i s p e r s i o n by expanding the d i e l e c t r i c t e n s o r in p o w e r s
along s o m e principal diagonal of the c u b e : | s t
| = | s21
of the w a v e v e c t o r ( s e e S e c . 2b) m a k e s it p o s s i b l e to
= | s 3 1 = 1/A/3 . We then obtain from (3.23)
obtain in much s i m p l e r fashion all the r e s u l t s indicated
i n C4,36,403_ T h e s a m e method m a k e s it p o s s i b l e to r e -
g a r d t h e i n f l u e n c e of t h e e x t e r n a l p e r t u r b a t i o n s and, i n
particular, to obtain all the r e s u l t s contained in the Consequently, in this c a s e fi 2 i s i n d e p e n d e n t o f t h e p o -
a r t i c l e s C36a~39l. in addition, w e s h a l l o b t a i n b e l o w a l a r i z a t i o n , and to e a c h p o l a r i z a t i o n t h e r e c o r r e s p o n d
s e r i e s of n e w r e s u l t s . two v a l u e s of n2, d e t e r m i n e d f r o m the equation
L e t u s c o n s i d e r f i r s t t h e r e g i o n of f r e q u e n c i e s c l o s e
t o t h e f r e q u e n c y at w h i c h the d i e l e c t r i c c o n s t a n t e o ( u ) )
b e c o m e s i n f i n i t e if s p a t i a l d i s p e r s i o n i s n o t t a k e n i n t o
The seven considered directions (the 3 four-fold axes
account (the dipole l i n e ) . In t h i s f r e q u e n c y r e g i o n it
and the 4 body d i a g o n a l s ) a r e thus optical a x e s of t h e
i s n e c e s s a r y t o u s e a n e x p a n s i o n of t h e t y p e (2.13) t o
crystal.
e s t i m a t e t h e r o l e of s p a t i a l d i s p e r s i o n
c) A s s u m e now that the v e c t o r s is directed along

# (со, k) = (со) t h e d i a g o n a l of t h e f a c e . F o r e x a m p l e , l e t Sj = s 2
= 1/VT, s 3
= 0. Then the equations for Dj, i = 1 , 2 , 3,
[the tensor used in (2.13) is pijlm = (c2/w2)/3ijZmb a s s u m e the following form
As was already indicated in Section 2b, the tensor
fi'ijlm becomes simpler if the x, y, z frame is aligned
with the four-fold axes. In this case, for the crystal
classes Td, O, and Oh, the tensor /3ijjm is determined
by the three numbers T h u s , i n t h e d i r e c t i o n c o n s i d e r e d , t h e v a l u e s o f fi 2 de-
pend e s s e n t i a l l y o n the p o l a r i z a t i o n of the e l e c t r o m a g -
*We note that in the articleM, as well as in all the earlier netic field. If D'3 * 0 a n d D { = D'2 = 0 , then
papers in this direction (see С7'41]), the effective transverse elec-
tromagnetic field is assumed to be equal to the average macroscopic
field. Such an assumption can be particularly critical in the inves-
tigation of quadrupole transitions. The question of the effective
transverse field is discussed in the bookM (see p. 385), but there T h i s e q u a t i o n r e m a i n s u n c h a n g e d f o r a w a v e w i t h D3
is likewise no solution to this problem there. We shall touch upon * 0 and, in the m o r e g e n e r a l c a s e , w h e n s 3 = 0, but
this question again in Sec. 4. st * s2. On t h e o t h e r h a n d , if D3 = 0, and D{ = - D 2
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL DISPERSION 683
2
0, then the equation for n is then obtain for the components of the v e c t o r E, in a c -
c o r d a n c e with (1.16) o r (1.20), the following s y s t e m of
equations [ s e e also (2.34) with a2 = a 4 ] :
i 2 2 2
F o r e a c h p o l a r i z a t i o n t h e r e a r e t w o v a l u e s of t h e r e - n E1 = (e0 + а 2 д ) Ег + ш Л ^ + 2a 3 (Es) я *, + (Es) и ^ (3.24)
f r a c t i v e i n d e x . We n o t e t h a t w h e n D 3 * 0 t h e i n d e x
etc, w h e r e
n f o r c a s e c) c o i n c i d e s w i t h t h e i n d e x f o r c a s e a ) . We
2
note also that the equations obtained for n show —2a 3 . (3.24a)
2
c l e a r l y t h a t t h e r o l e of t h e s p a t i a l d i s p e r s i o n d i f f e r s
with the d i r e c t i o n s and with the p o l a r i z a t i o n . Thus, We c o n s i d e r h e r e , too, s e v e r a l p a r t i c u l a r c a s e s , when
the p h e n o m e n o l o g i c a l a n a l y s i s ' ^ is, a s in o t h e r c a s e s , the v e c t o r s is d i r e c t e d along the s y m m e t r y axes of
n o t o n l y fully a d e q u a t e f o r a c o n s i d e r a t i o n of t h e o p t i c a l the cube:
i s o t r o p y i n c u b i c c r y s t a l s of c l a s s e s T d , O, a n d Oh, a) s p a r a l l e l to the z axis, s 3 = 1, s t = s 2 = 0. In
but also appreciably s i m p l e r than the corresponding t h i s c a s e the s y s t e m (3.24) h a s the following f o r m :
microscopic theory ^ or, m o r e accurately, calcula- 2
n El = (e0 + а2ге ) Ег,
2 2
n E2 = (e0 -f- a2re )
2

t i o n s u s i n g e x c i t o n w a v e f u n c t i o n s . T h i s , of c o u r s e ,
(3-25)
p e r t a i n s a l s o t o c r y s t a l s of o t h e r c l a s s e s .
In t h e d i r e c t v i c i n i t y of a r e s o n a n c e ( p o l e ) , e o ( w )
Thus, in t h i s c a s e t h e r e is a t r a n s v e r s e wave ( E 3 = 0 ) ,
m u s t b e r e g a r d e d a s c o m p l e x and, a s follows f r o m the
for which, independently of the polarization,
foregoing, the absorption coefficient к = Im n, like n (3.26)
na = (n + ЫУ = , ^-
= Re n, depends on the d i r e c t i o n s and the p o l a r i z a t i o n
of the light. Thus, in cubic c r y s t a l s we can have n e a r and a longitudinal wave ( E j = E 2 = 0) with
the poles of the function e o ( w ) not only anisotropy of
2

d i s p e r s i o n , but also of absorption. The poles of e o ( ш ) И = So

ai-Ma',' -

c o r r e s p o n d to lines in the absorption s p e c t r u m which

we shall call dipole absorption lines ( s i n c e they c o r - b) W e a s s u m e n o w t h a t t h e v e c t o r s is d i r e c t e d

r e s p o n d to nonvanishing m a t r i x e l e m e n t s of the c r y s t a l a l o n g a p r i n c i p a l d i a g o n a l of t h e c u b e : s t = s 2 = s 3

d i p o l e - m o m e n t o p e r a t o r , m a d e up of wave functions of = 1//з~. In t h i s c a s e we have in place of (3.25)

the ground and excited s t a t e s of the c r y s t a l ; s e e below
1 2
for d e t a i l s ) . V ° 3 у ' 3 2 3
As is c l e a r from the foregoing, account of spatial dis- (i = l , 2 , 3 ) . (3.27)
p e r s i o n n e a r dipole lines in cubic c r y s t a l s l e a d s to a
s t r o n g change in the v a r i a t i o n of the d i s p e r s i o n c u r v e s , When E = E s (longitudinal w a v e ) n 2 = - 3 e 0 /
and in this r e s p e c t the spatial d i s p e r s i o n effect i s far ( a t + 2 a 2 + 4 a 3 ) , w h e r e a s for t r a n s v e r s e w a v e s
from s m a l l . E • S = 0 i n d e p e n d e n t l y of t h e p o l a r i z a t i o n
We p r o c e e d now to c o n s i d e r the anisotropy of d i s - (3.28)
p e r s i o n and absorption n e a r the quadrupole absorption 1— а/ 3 —
l i n e s , in the vicinity of which е о ( ш ) v a r i e s smoothly, c) Let the v e c t o r s be now d i r e c t e d along the diago-
but at l e a s t one of the components of the t e n s o r nal of the face of the cube. F o r example, let Sj = s 2
a j j ; m ( w ) h a s a r e s o n a n c e ( p o l e ) . To t h i s end we use = 1/V1F, s 3 = 0. In this c a s e we have on the b a s i s of
an expansion of the t e n s o r € j j ( w , k ) [ s e e (2.12) and (3.24)
(2.33)]:
Ег = (e 0 -f a2ra2) Et + ^ n?Ei
6^(0), k) = 6i;-80 (to) + aiilm (to) n2sLsm. (2.33a)
The t e n s o r a j j ; m differs h e r e from the t e n s o r d e s i g - i = \, 2
nated by the s a m e l e t t e r in (2.12) by a factor c 2 / w 2 .
Let u s a s s u m e that the principal absorption of light and
in the c r y s t a l is connected with the quadrupole a b s o r p -
(3.29)
tion l i n e s , by v i r t u e of which the t e n s o r e o (w )6JJ will
be a s s u m e d r e a l and the t e n s o r otijim(u}) complex: It follows from (3.29) that:
Щ1т(ш) = «ijZm(w) + ai'jZm(uj). 1) E 3 = 0, E « s * 0—longitudinal wave:
Inasmuch a s in cubic c r y s t a l s both t e n s o r s a £ j ; m
and aijfrn simplify simultaneously, provided the c o o r - a/2 -)-a2-l-2a3
dinate axes a r e chosen along the four-fold a x e s , the
2) E 3 = 0, E • s = 0 — t r a n s v e r s e wave p o l a r i z e d in
p r e s e n c e of absorption ( а ш т * О) does not c o m p l i -
the plane of the face:
c a t e the a n a l y s i s . We u s e in such a coordinate s y s t e m
the notation at = «xxxx = a y y y y = a z z z z . az = a x x y y e0 (3.30)
l-а,- a/2
= a x x z z = « y y z z . «з = a x y x y = «xzxz = a y z y z - We
684 V. M. A G R A N O V I C H a n d V. L. GINZBURG
Table IV. Characters of irreducible representations
of the Oh group
Irreduc- Symmetry o p e r a t i o n s
ible rep- n
resenta- 0
м *•
0 0 и
tions 4 CO
О
CD CD
0
CD CO
0
00
A-,

E
2
(scaalr) 1

1
1

1 - 1
1

— 1
1 1

1
1

1
1

1
1

— 1
1 1
1
^1^2—У1У1, 2 z i z 2 — 2 2 0 0 2 2 0 0 j

к (pseudovector) —1 0

CO CO
3 1 0 — 1 1
3 —1 — 1 1 0 — 1 1 0
г
+ X^li У& + У2 1

A\
A
E''z (pseusoscalar) 1
1
2
1
1
2
— 1
1

0
—1
1

0 — 1
1

1
—1

—1
—2
- 1

—1
—2
—1

0
-1
1
0
—1
-1
1
x, y, z ( v e c t o r ) 3 —1 1 — 1 0 — 3 1 — 1 1 '0
3 — 1 1 0 — 3 1 1 —1 0

3) E 3 * 0, E • s = 0 —transverse wave polarized does not contain essential screw axes and glide planes,
perpendicular to the plane of the face: the aggregate of "rotational" elements forms a point
group, namely the group of the corresponding crystal
(3.31) class '-43-'. Inasmuch as we are considering in the
I—a,
present article weak spatial dispersion, which c o r r e -
We note that in the case when st * s 2 , s 3 = 0, the sponds to the method of expanding in powers of the
equation for E 3 , as in (3.29), separates and relation wave vector, the tensors ещ{ w ),<*щ т (а)) etc, con-
(3.31) remains in force. tained in an expansion of the type (2.19), are deter-
So far we have not made specific the character of mined by the properties of the exciton states as к — 0.
those excited states in the crystal, the presence of The wave functions of exciton states with к = 0 are
which leads to a resonant behavior of the functions invariant under the elements of the translation sub-
е о (ы) or Q!ij^m(w). Inasmuch as we are interested group [see expression (3) in the Introduction]. There-
here primarily in exciton states, let us stop, before fore the corresponding exciton states can be classified
we proceed to a further exposition of the theory of in accordance with the irreducible representations of
optical anisotropy of cubic crystals, to discuss the the point group of the crystal class*, which character-
classification of these states.* izes the symmetry of the directions in the crystal. This
It is well known that stationary states in a crystal, is precisely the classification of the exciton states
and particularly exciton states (see, for example, which we shall use below.
[42-44]^ c a n jj e c i a s s i f i e d in t e r m s of the irreducible
representations of the crystal space group. Each Let us consider in greater detail crystals of the
space group contains a subgroup of parallel transla- type of Cu2O, which belong to the most symmetrical
tions, including all the possible parallel translations class Oh of the cubic system. The characters of the
of the lattice into itself. The complete space group irreducible representations of the group Oh are indi-
is obtained from this subgroup by adding to it H ele- cated in Table IV (using the notation of №1). The sec-
ments ("rotation" elements), which contain rotations ond column of this table shows how the corresponding
or reflections, with H equal to the number of elements wave functions are transformed under the symmetry
of the group of the corresponding crystal class. Any operations from Oh. Thus, for example, it follows
space group element can be regarded as a product of from Table IV that three wave functions corresponding
one of the elements of the translational subgroup by to the triply degenerate (when к = 0) exciton term and
one of the " r o t a t i o n " elements. If the space group having the symmetry of the irreducible representation
F 2 , are transformed like symmetrized products of un-
like components of two polar vectors (xj, yj, Zj) and
•Transitions whose probability is proportional to k2 can, of (x 2 , У2. z 2 ) .
course, also be classified without resorting to the notion of ex- As in the case of the lowest electron terms of poly-
citons and their wave functions. Actually, these transitions cor- atomic molecules (see t 4 5 ^, Sec. 98), it is customary
respond to radiation of a scalar source [in this case only longitu-
dinal waves can be obtained; see (3.35)—(3.35a) below], and to
radiation from a quadrupole and a magnetic dipole. In other words •The latter is connected with the fact that the point group of
as in the case of dipole radiation, there is no special need to re- the crystal class is isomorphic to the factor group relative to the
sort here to quantum language. The latter, however, will be done translation subgroup. More details on this subject can be found, for
with an aim towards application to exciton lines. example, in L44J.
 CRYSTAL O P T I C S WITH ALLOWANCE FOR S P A T I A L DISPERSION 685
to u s e for c r y s t a l s an e m p i r i c a l r u l e , according to form in accordance with the triply degenerate r e p r e -
which the wave function of the ground s t a t e of the c r y s - sentation F 2 .
tal h a s complete s y m m e t r y with r e s p e c t to the c r y s t a l Each of these quadrupole exciton states can, gener-
s y m m e t r y t r a n s f o r m a t i o n s . In the c a s e under consid- ally speaking, make a contribution to the quadrupole
eration, for example, this m e a n s that the ground state absorption of light by the crystal. However, near a
h a s the s y m m e t r y of the i r r e d u c i b l e r e p r e s e n t a t i o n A t , given resonance one can usually confine oneself to an
which will be a s s u m e d . Then, taking account of the fact examination of one transition, i.e., one excited exciton
that the dipole m o m e n t o p e r a t o r t r a n s f o r m s like a polar state. Here, as will be shown in Sec. 4, for the transi-
v e c t o r , the m a t r i x e l e m e n t of the dipole m o m e n t o p e r - tion from the ground state zero to the excited states of
ator will differ from z e r o only on going from the ground the L-th band (for к = 0) we have
state into such exciton s t a t e s , whose wave functions a r e
transformed when к = 0 in accordance with the i r r e - 0, LQ)
ducible representation F 2 (see Table IV). Nonvanish-
ing matrix elements of the dipole moment operator cor- '. LQ\flm\O)(O\fn\O, LQ)\. (3.33a)
respond to nonvanishing transition oscillator strengths,
Here the operator
meaning that at the transition frequency the dielectric
constant e o (aj) becomes infinite if absorption and spa-
(3.33b)
tial dispersion are not taken into account. Thus, the
results obtained in Item 3b by using for the tensor ejj
p is the number of degenerate excited states (for к
an expansion of the type (2.13) make it possible to take
= 0) in the L-th band, and r a and p a are the coordi-
into account the spatial dispersion connected with the
nate and momentum of the a-th electron in the crystal.
contribution of those exciton bands, whose wave func-
tions are transformed when к = 0 in accordance with The fact that the Tij transform like products of the
the irreducible representation FJ. As regards those components of two polar vectors causes indeed the
exciton bands, whose wave functions are transformed tensor aijftn(<*>) to be determined by the contribution
when к = 0 in accordance with irreducible representa- of the exciton states, whose wave functions are t r a n s -
tions other than F{, these bands manifest themselves formed in accordance with one of the irreducible r e p -
only if spatial dispersion is taken into account. Thus, resentations A t , E, Fj, or F 2 .
for example, as is the case for atoms and molecules In the case of a degenerate exciton term we shall
(see, for example, ^ ) , the contribution to quadru- choose the wave functions such that they transform,
pole absorption and emission of light is obtained only apart from a coefficient, as the corresponding linear
from those exciton bands, whose wave functions t r a n s - combinations of the products of the components of a
form when к = 0 as products of the components of two polar vector. Thus, for example, in the case of a
polar vectors. The aggregate of these products in the doubly degenerate term the wave functions фЦ and ФШ
Ojj group generates the reducible representation V2, will be chosen such that in cubic-symmetry transforma-
which breaks up into a sum of irreducible representa- tions they transform* as /3~П = -/з~ (x t x 2 — z t z 2 ) and
tions* (Table IV and C*2^): III = 2yjy2 — x t x 2 - z t z 2 . It is clear that the scalar
product Г Ф Й Ф Н ^ Г of such functions ФЦ and ф№
V2 = A1 + E + F1 + F2. (3.32) vanishes. (When к = 0 all the functions can be r e -
garded as real, a fact which has been taken into ac-
Here the scalar product of the two polar vectors I count above when writing out the scalar product; when
= x t x 2 + У1У2 + ZjZ2 reduces in accordance with the the coordinate system is rotated through an angle TT/2
representation A t , the two independent linear combi- about the у axis, the function Ф^ reverses sign, and
nations II = x t x 2 — Zjz2 and III = 2y t y 2 — XjX2 — Zjz2 the function Ф^ 1 remains unchanged; thus the invari-
transform in accordance with the doubly degenerate ant quantity—the scalar product—should reverse sign
representation E, the three components of the vector under such a coordinate transformation and conse-
product of two polar vectors IV = y t y 2 — y 2 z t , V = z t z 2 quently vanish.) We can prove analogously the mutual
- zgii, and VI = Х1У2 -^гУ1 transform in accordance orthogonality of the functions Ф ^ , ф ^ , Ф^ 1 , etc. We
with the triply degenerate representation F t , while the shall henceforth use the following identities:
three symmetrical linear combinations VII = y t z 2
+ Уг21> VIII = ZjX2 + x t z 2 , and IX = Xjy2 + y t x 2 t r a n s -

*The representations Aj, A^, E', and F^ (see Table IV) corres- = 4 i x ± 4 vi, з-34)
pond to higher multipoles, which are apparently of no interest. We
note also that we classify here as quadrupole absorption all the т 7 I ~~2
VIII + j - V.
absorption whose probability is proportional to k\ Therefore, in
addition to the true quadrupole absorption (the representations E
and F2), this includes also the magnetic-dipole (representation F,) *The factor \fb follows from the normalization requirement for
and scalar (representation AJ absorptions. the basis functions.
 686 V. M. A G R A N O V I C H a n d V. L. G I N Z B U R G
Let u s stop to d i s c u s s s o m e p a r t i c u l a r c a s e s .
1) We assume that when к = 0 the exciton state L
transforms in accordance with the nondegenerate r e p -
resentation Aj, by which the ground state of the sys- = ~2 ( ii зз) "Ь ~2 ( ^ 22 •" i i
tem is likewise transformed. Then, taking (3.34) into
account, we obtain

we find that
(0, (3.35)
), i.e.
From this we obtain immediately with the aid of (3.33)
[see also (2.31) and (2.32)] Using now (3.33) and also (3.36) and (3.37), we get
a, = a Ф 0, a 2 = 0, a 3 = - -j a,
a! = a=^0, a 2 = 0, a 3 = -^-, a = c^ — a 2 — 2a 3 = 0. (3.35а)
q
a = ax — o 2 — 2a 3 = -к- a, (3.38)
Since in this c a s e we have in accord with (3.24)
For this c a s e , the s y s t e m (3.24) still r e m a i n s c u m b e r -
some:
it is clear that the considered excited state manifests
itself only for longitudinal waves (independently of the ( ) s ) * 1 e t c . (3.38a)
2
direction s, we have for this wave n = — eo/a, where- 2
as for transverse waves n 2 = e 0 ) . Consequently, the expression for n for arbitrary s
2) We now consider the case of the doubly degener- will be derived below by perturbation methods. We now
ate term E. Then note one consequence of (3.26), (3.28), (3.30), and (3.31).
Namely, the presence in the crystal of a quadrupole ex-
?$ (2T22 - fn - f33) ¥ 0 dr = ^ ¥ E n (fu - t33) 4% dr = 0. citon state of the type considered here manifests itself
neither in the dispersion nor in the absorption if the
Therefore in accordance with (3.34) we have light propagates along the edges of the cube, and to the
f 7 1 f contrary, does manifest itself for any polarization, when
(4 s|7 iJ.| J c> = S1jAfi,-, (3.36a)
the light is propagated along the principal diagonals of
where Mjj is the diagonal element and not the trace, the cube. In the case, however, when the vector s is
and directed along the diagonal of the face of the cube, the
exciton state of type E manifests itself only when the
мп (ii) = <Y£ I ru | Y . ) = 4 - f3
electric vector lies in the plane of the face.*
3) Let us discuss the case of the triply degenerate
term F t , whose wave functions transform as the com-
Mn (III) = (T"11 fn | ¥„} = - 4 ^ E 1 1 | 2 - fn - f331 ponents of a pseudovector. In this case, in accord with
the chosen basis, and also by virtue of (3.34), we have
= -±Mt, M22(lll) = ±M2, Л / 3 3 ( Ш ) = - 4
CVFi | Tu | ¥„) = (1 — 6i;) Mu, (3.39)
(3.36b)
Here we used also the relation where, of course, no summation is carried out over i
and j , and
M I1V\ — IWV \f J
I V \ — — {VJ7 I f —f 1ЧМ
23 V / — \ Fi ] 2 3 | 0/ — о \ - ^ l l 2 3 32 I 0/
which holds true by virtue of the fact that the functions
= - M ( I V ) ,
* E and the operator ( Т 1 4 + T 2 2 + T 3 3 ) transform in
3 2

accordance with different irreducible representations Mu (IV) = Мь (IV) = M13 (IV) = M31 (IV) = 0, M13 (V)
of the Oh group. = \ * i**a I * 13 I
T
o/ — 2 \ •Fi 1 31
J
13 1 * 0 / —
J
" 31 к • /» V,
It is easy to verify that
M l t
(V) = M u
(V) = M23 (V) = M 3 2
(V) = 0,

(3.37) лх /•\7T^ /W^* \T 1Ш \ /"Ф^* \ T T W \

J
№ 12 И L) = \ A F J | J i 2 | i d / — y \ i J ? 1 | 12 •* 2 l *0/

Indeed, by carrying out under the integral sign

(3.40)

the operation g (rotation through тг/2 about the z *These deductions, as applied to absorption, coincide with
axis), taking into account the invariance of the inte- those made in L36J. The question of the anisotropy of the disper-
gral, and also the relation sion, i.e., of the dependence of n on s, was not considered in f36].
 C R Y S T A L O P T I C S WITH A L L O W A N C E FOR SPATIAL DISPERSION 687
In addition, it is easy to verify that we get
(3.48)
'1—(
Using now (3.33), and also (3.39), (3.40), and (3.41), we
O n t h e o t h e r h a n d , if t h e v e c t o r s i s d i r e c t e d a l o n g t h e
find that
d i a g o n a l o f t h e f a c e o f t h e c u b e , t h e n f o r t h e w a v e p o -
a
04 = 0, <xa = a a3=— y a =04 — a2 — 2a3 = 0. l a r i z e d i n t h e p l a n e o f t h e f a c e w e h a v e

(3.41a) n2 = e0, (3.49)

Since we have here, in accord with (3.24), a n d f o r t h e w a v e p o l a r i z e d p e r p e n d i c u l a r t o t h e p l a n e

(n2-e0)Ei^akiEi + n2(Es)si{l —a), i = l, 2, 3, of t h e f a c e

we reached the conclusion that the absorption and dis- (3.50)

persion a r e completely isotropic. For transverse 1—a' — ia"
waves we have In the general case of arbitrary s, one has to use
the system (3.24) to determine n 2 . In the case of (3.46)
(3.42)
1— a' — ia" considered here, this system assumes the form
4) Let us consider, finally, the situation that arises !
—e 0
etc. (3.51)
when the exciton term for к = 0 is triply degenerate
and corresponds to the representation F 2 (see E x a c t c a l c u l a t i o n o f n 2
b y m e a n s o f (3.51), w i t h a c -
Table IV). c o u n t o f t h e t e r m s ( a ' ) 2
, ( a " ) 2
, e t c , i s q u i t e c u m b e r -
In this case, in analogy with (3.39), s o m e a n d g e n e r a l l y s p e a k i n g w o u l d c o r r e s p o n d t o a n

(3.43) e x a g g e r a t i o n i n a c c u r a c y , f o r i n o u r a p p r o x i m a t i o n ( f o r

n o n - l o n g i t u d i n a l w a v e s ) it i s m e a n i n g f u l t o d e t e r m i n e
where as a result of (3.34) we get o n l y t h e c o r r e c t i o n s f o r n 2
w h i c h a r e l i n e a r i n a . F o r

t h i s p u r p o s e it i s s u f f i c i e n t t o u s e p e r t u r b a t i o n t h e o r y .

I n d e e d , l e t u s i n t r o d u c e t h e n o t a t i o n

0 )
4 = a6t,. + (1 + a) V ; . L\? = - 2as\bi}

Mlt (VII) = M21 (VII) = M13 (VII) = M31 (VII) = 0,

a n d r e w r i t e t h e s y s t e m (3.51) i n t h e f o r m
M31 (VIII) = \T31\V0) = I ( f3l + f1
/r(0) . r(l)\ и „i? _ _ ™2 Eo

M12 (VIII) = M21 (VIII) = M23 (VIII) = M32 (VIII) = 0, In the zeroth approximation ( L ( 0 ) — p) E = 0 and com-
plete isotropy occurs, the waves can be either strictly
Mlt (IX) = (У1* I transverse (in which case p\ = a) or strictly longi-
tudinal (pj,1 = 1 + 2a).
= M21(IX), Л The first-approximation correction is

(3.44)
In addition, as in the case of (3.41), the following rela-
tion is satisfied where the wave polarization vector is e = E/1 E |.
Thus, for transverse waves we have, accurate to small
M23 (VII) = M3l (VIII) = M12 (IX) = M (F2). (3.45)
quantities of first order of smallness in a,
Hence, and also on the basis of (3.33), we find that
- = a — 2as?ei,
rt л ® о о
Ctj = U , (X 2 = Ct = F U , <X3 = -к- , (X = Ctj — <Z2 — Z ( X 3 = — £CC.
(3.46)
or
T h e r e f o r e , a s f o l l o w s f r o m (3.26), (3.28), (3.30), a n d
= E — ae0( — 2s\e\) y
(3.52)
0
(3.31), t h e q u a d r u p o l e t r a n s i t i o n u n d e r c o n s i d e r a t i o n
For longitudinal waves (e = ± s )
l e a d s t o a n o t i c e a b l e a n i s o t r o p y i n t h e d i s p e r s i o n a n d

a b s o r p t i o n o f light. If t h e v e c t o r s i s d i r e c t e d a l o n g

a n e d g e o f t h e c u b e , t h e n , i n d e p e n d e n t l y o f t h e p o l a r i -

z a t i o n ,
or
(3.47) "" 2 a ( l - s } - s | - s | ) *
1 —a' —ia"
For the vector s, which is directed along the principal It is s o m e t i m e s convenient to consider the optical

diagonal of the cube, independently of the polarization, properties of a m e d i u m in spherical coordinates ( F i g . 6)

688 V. M. A G R A N O V I C H a n d V. L. GINZBURG

Hence

eo f —
2 2
*„ = 3a" VT0 sin 9 sin 2ф, (3.58)
2 2
» = eo + —^- sin 20(3 + cos 2ф), I
3a" V^o sin2 29(3 + cos2 2ф). j
32
d) Influence of mechanical s t r e s s e s and external
electric and magnetic fields. The need for a theoret-
FIG. 6 ical study of the influence of external action on the
form of the tensor ejj(w,k) and on the exciton states
for two mutually perpendicular polarization directions in crystals becomes particularly clear, if one recog-
s
eP and e , where eP corresponds to polarization in nizes that a large number of experimental papers are
the meridional plane containing the z axis and the light devoted to this subject [3M6-*e,*o-S2]> A s a p p i i e ( j t 0
propagation direction. *
s cubic crystals, the subject of the investigation is the
The components of the vectors eP and e are de-
artificial anisotropy of the optical properties of the
termined by the relations
crystal near the exciton absorption lines, as a function
ef = sin<p, e | = — cosq>, eg = O, of the character of the deformation, the direction of the
ef = — cos 6 cos <p, ef = — cos 9 sin ф, ej = sin9. (3.53) magnetic or electric field, etc. As applied to crystals
At the same time, the components of the vectors s of the Cu2O type, many of the aforementioned problems
= s/k are obviously were considered in СЗбЬ-зэ] ^ t h i n the framework of ex-
citon theory.
, s 2 = sin6sinq), s3 =
From the point of view of the phenomenological
Substituting (3.53) in (3.52) for the s- and p-polari- crystal optics with spatial dispersion developed here,
zations, we obtain the influence of the external action can be considered
nl = e0 — <ZB0 (1 — sin2 0 sin2 2<p), in the account of the dependence of the tensors con-
tained in (2.12) and (2.13) on the external fields and
n2p = е„ - ^ ( s i n 2 26 sin2 2q> + cos2 29). (3.54) stresses.
In the presence of external actions, the symmetry
Recognizing now that n = n + i/c, we obtain in first of the crystal is generally speaking lowered, and as a
order of a (we recall that these formulas pertain to result the limitations imposed on the components of the
the level which transforms in accord with representa- tensors € Ц ( « , E<°>H(0), a < m ), т(ш, E<°>, H<0>, a<°>n),
tion F 2 ) : (0 (0)
aijZm(w, E \ H , a$), etc are different from those
/if = e 0 —a < e 0 (l — sin 2 6 sin 2 2q>), prevailing when

(3.55)
-° (sin2 29 sin 2 2q>+ cos2 29), [E < 0 ) and H < 0 ) are the intensities of the external elec-
= 80 -
tric and magnetic fields, аЦ' is the s t r e s s tensor; the
medium is assumed nonmagnetic, by virtue of which we
(sin2 29sin 2 2(p + cos229).
ea

do not distinguish between H ( 0 ) and B ( 0 ) ] .

We can obtain quite analogously, accurate to the An account of this fact enables us to determine,
first power of a, expressions for the complex refrac- similar to what was done in Sec. c), the polarization
tive index n in the previously considered case of a of the normal electromagnetic waves, and also the
doubly degenerate exciton term (E representation). anisotropy of the dispersion and absorption in the r e -
Using (3.38a), we find that for quasi-transverse waves gion of the exciton absorption lines. We are unable to
(i.e., for waves that are transverse when a = 0) we consider this problem in detail for crystals of different
get classes, for lack of space. We therefore confine our-
selves only to a few pertinent problems.
~ll!Llsy y y), (3.56)
+ s + s
Let us consider, for example, the influence of a
constant electric field on the exciton lines in the Cu2O
or crystal. We also assume, for the sake of simplicity,
that the electric field is directed along the four-fold
symmetry axis. If the z axis is directed along the
n'p = e 0 + ^ sin 2 29(3+ cos2 2<jp). (3.57) field, then the operator for the perturbation of the
crystal by the field has in the first order in the field
•The superscript s corresponding to the polarization e s has the form H' = - E ( O > P Z , where P is the crystal dipole
no relation here and below to the vector s = k/k. moment operator. From the form of the operator H'
 CRYSTAL O P T I C S WITH A L L O W A N C E FOR SPATIAL DISPERSION 689
2
Table V. Characters of irreducible represen- =n = and к = K S = 0. For the second wave
tations of the C 4 V group
Irreduc- Symmetry operation
ible rep-
resen- E 2C4 2/C4 where в is the angle between the vector s and the z
tations axis. For weak fields Ле33/ец « 1 and
1 1 1 1 1 2
n p = e u + Де33 sin 0.
A2 1 1 1 —1 —1 Consequently, introducing the notation Де 3 3 = Де 3 3
1 1 —1 1 —1
1 1 —1 —1 1
E 2 —2 0 0 0

2/eu
it follows immediately that in the presence of a field
the crystal symmetry is reduced and characterized by where account is taken of the fact that /cp « np. In
2
the C 4 V group, for which the characters of the i r r e - these formulas there is no factor k , from which it is
ducible representations are indicated in Table V. also clear that one is dealing with dipole transitions.
Comparing the characters of the C 4 v group with the An analogous situation ("flare-up") occurs also
characters of the O n group (Table IV), it is easy to for quadrupole lines, whose wave functions in the ab-
establish in the usual manner that under the influence sence of a field have the symmetry of representations
of the electric field there should arise not a single di- E, Fj, and F 2 . Inasmuch as these terms split under
pole absorption line but a doublet (the representation the influence of the electric field
FJ, which is irreducible in the O^ group, breaks up in
the C 4 V group, so that F[ = Aj + E ) .
In accordance with Table V, one of the components and at the same time the transitions to the exciton
of the doublet should be polarized along the field and states, having the symmetry of the representations
the other perpendicular to the field, so that the crystal Aj and E, according to Table V, are allowed in the
becomes anisotropic (uniaxial) even if the spatial dis- dipole approximation.
persion is neglected, and in this case the nonzero com- As regards the transitions to the exciton states with
ponents of the eij tensor are en = e 2 2 , and e 3 3 . the symmetry of representations B x , B 2 , and A2 of
The quadrupole exciton states also undergo interest- group С 4 У , these transitions remain quadrupole, and
ing transformations.* hi particular, the nondegenerate for an analysis of the corresponding anisotropy of the
exciton state which has the symmetry of representation refractive index n it is sufficient to make use of the
A t of group Oh, has in the presence of an electric field method developed in the preceding section. In the
along the z axis the symmetry of representation At of group C 4 V the quantities transformed in accord with
group C 4 V ; as a result the transitions to this state from the representations A2 are of the type x t y 2 — x 2 y t , in
the ground state of symmetry Aj become allowed in the accord with B t —of the type x t x 2 - y ^ , and in accord
dipole approximation (see Table V). t Independently of with B 2 —of the type x t y 2 + y t x 2 . Therefore, as can be
the spatial dispersion, the intensity of lines of this kind readily verified, in the vicinity of the exciton line A2
should increase with increasing electric field. It is the only nonvanishing components of the tensor а щ т
clear that in the vicinity of these exciton lines, when are the components a 1 1 2 2 = a 2 2 1 1 = a, a 1 2 1 2 = a 2 1 1 2
the crystal behaves like a uniaxial crystal, we have = a 2 1 2 1 = a 1 2 2 1 = - a/2, and in the vicinity of the exciton
line Bj we have а 1 Ш = a 2 2 2 2 = a; ai2i2 = a m i = a 1 2 2 1
8„ (Ф, E(0)) = e u («, E(0)) «„ + Д833 (со, E(0)) 6 i3 6,3. = a
2U2 = ~ a / 2 > while in the vicinity of the B 2 line—the
components a 1 1 2 2 = a 2 2 1 1 = a, a 1 2 1 2 = a 2 1 1 2 = a 2 1 2 1
The equations for the components of the electric
= a 1 2 2 1 = a/2. The knowledge of these nonvanishing
field in the light wave assume in accord with (3.24) the
components of the tensor Q!ij/m makes it possible to
form
make a complete analysis of the anisotropy of the op-
nEi = (Es) tical properties of the crystal in the vicinity of lines
of the type A2, Bj, and B 2 . Inasmuch as the c o r r e -
It follows therefore directly that in any direction s two
sponding calculations are analogous to those in Sec. 2,
types of waves can propagate. For one of these waves
we give here only the results obtained by the perturba-
the electric vector E is perpendicular to the plane
tion theory method.
passing through the z axis and the vector s, with n 2
For transverse waves the following holds true (see
•Many problems connected with the influence of a constant Fig. 6):
electric field on the quadrupole exciton lines in Cu2O were con-
sidered in [ Mb l. n? = e0 + eoa ( [es]2
tThe possibility of "flare-up" of exciton lines in Cu2O in the
presence of an electric field was first pointed out in ["]. for the type A2 transition,
690 V. M. A G R A N O V I C H a n d V. L . G I N Z B U R G
2
n = e0 + 80a (Bi) ( е д — e 2 s 2 ) —
2
e t , (со, k, E<°>, H«») = eOij. (со) + iyin (со)ft,+ aijlm (со) ktkm

for t h e type Bj t r a n s i t i o n , and + AiH (со) E?> + A[n (со) НГ + АЩт (со) H?>km
m m
2 + 43Jm (со) H\ E<$ + A$n (со) E\ km + A$lm (со) Н^Н™
n = e0 + e o a (B 2 ) (e
4?L (со) &РЕ™ + Aijimn (со) ЕГН«!кт + v.. (3.59)
for t h e type B 2 t r a n s i t i o n .
(0)
F r o m t h i s it follows, in p a r t i c u l a r , that Inasmuch as here H is an axial vector while к and
(0)
for the type A 2 t r a n s i t i o n : E a r e polar, the quantities Ащ, A | | | m , А { | | т , and
A a r e
ijlm ordinary tensors, while A<jZ, A $ m , A | ] } m
n\ = 80 + e o a' (At) sin 2 9, x s = у l / e > " (A,) sin 2 6 and A i j ; m n a r e pseudotensors.
The principle of the symmetry of the kinetic coeffi-
cients requires [see (1.10)] that the tensor (3.59) sat-
for t h e type B t t r a n s i t i o n : isfy the relation (as was already pointed out, we do
(0) <0)
not distinguish here between B and H , assuming
nl = e 0 + e o a' (Bj) sin2 0 sin2 2ф, the medium to be nonmagnetic)
l0) <0) <0) <0>
^ ( ^ i ) sin 2 9 sin 2 2<p, eu (со, k, E , H ) = 6 j i (со, - k , E , - H ). (3.59a)
2
= e 0 + -£• eoa" (BJ sin 20 sin 2ф, 2
From this it follows directly that

2 2 = -Aftm, = -Aflm,
4 V ^ a (Bi) sin 20 sin 2ф;

A —A
for t h e t y p e B 2 t r a n s i t i o n :
"ijton — jilmn- (3.60)
*raf= e 0 -r e o a' (B 2 ) sin 2 0 cos 2 2<p,
The condition that there be no absorption calls for the
4, = у ,) sin 2 9 cos 2 2 Ф ,
dielectric tensor to be Hermitian, ец = e«. In this
= e0 + e0 sin2 20 sin 2 2ф, case, the tensors that a r e antisymmetrical in ij, con-
tained in (3.35), a r e pure imaginary. We note also that
*„ = i УГУ (Bt) sin2 20 sin2 2Ф. the pseudotensor Ац1 [and also By/ in (3.61)] produce
magneto-optical effects in crystals (see ^) with
In the foregoing expressions a = a' + ia", and the
argument of a (for example a(A 2 )] indicates the state А'Ф
= e
ijm^mi> В'гц = eijmBml, (3.60a)
corresponding to the value of a. The choice of the
Here A m j and Вщ/ a r e second-rank tensors, which
angles and polarization directions is clear from Fig. 6.
generally speaking a r e asymmetrical.
We can investigate quite analogously the effects p r o -
The symmetry of the crystal greatly reduces the
duced by a constant magnetic field or by mechanical
number of independent components of the tensors con-
stresses, and to determine the resulting new symmetry
tained in (3.59). The components of these tensors
group of the crystal it is necessary to use the following
should be invariant under a change in reference frame
general principle: a crystal under the influence of an
corresponding to any of the crystal symmetry opera-
external action will have only those symmetry elements,
tions. It follows, in particular, that in crystals with an
which a r e common to the Hamiltonian of the crystal in
inversion center the third-rank tensors a r e YijZ
the absence of the action and the part of the Hamiltonian
= Ajjj = 0. Analogously, we reach the conclusion that
dependent on the action (of the magnetic field, stresses,
in such crystals the fourth-rank pseudotensors A j j | m
etc).
and A - j | m vanish, inasmuch as these quantities r e -
So far we have not considered the question of the ex- verse sign upon inversion. A relation analogous to
plicit dependence of the magnitude of the effects on the (3.59) can also be written for the inverse dielectric
magnitude of the applied external actions. In the case tensor ецЧо», к, E ( 0 ) , H < 0 ) ) :
when the external actions a r e sufficiently weak, the
explicit dependence on the intensity of the external a c - ei/ (со, k, E«», H<0)) = e-0\j (со) + *8,-„ (со) + ?„,„,*,&„,
tion can be established by expanding the tensors in
(2.11)-(2.15) in powers of E ( 0 ) , H < 0 ) , and <r{j\ in full + Bin (со) E\m + Bin (») НГ + B«L (a)H\nkm
accord with what is done in ordinary crystal optics + ВЩт (со) H\mES + ВШт (со) Е\°>кт + B%m (со) H\°>H™
(see, for example, '- 26 -'). Inasmuch as the account of
+ ВШт (со) ЕГЕ™ + BiHmn (со) Е\тН%кю +...
spatial dispersion introduces in this case singularities,
we shall discuss some of them, confining ourselves to Here, of course, all the considerations advanced above
the influence of an external electric or magnetic field, regarding the limitations connected with the principle
or of the two simultaneously. of symmetry of kinetic coefficients and crystal sym-
In the presence of weak external electric and mag- metry remain in force. At the same time, the follow-
netic fields we use in lieu of (2.11) the expansion ing very important remark must be made. Expansions
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL DISPERSION 691
of the type (3.59) or (3.61) are valid only if the coeffi- (со, k, E<0), H(0)) | = 0, (kD') = 0, (3.62b)
cients AJ; # (a)) and Bjj #># (w) are sufficiently small.
Otherwise, we can no longer confine ourselves to terms respectively.
with the lowest powers of the field E < 0 ) . This is p r e - If only an external magnetic field is present ( H ( 0 )
cisely the situation that arises, for example, for cubic * 0, E <0> = 0), then we have in crystals with an inver-
crystals in the vicinity of a degenerate level, for which sion center А | Н т = в Ш т = 0, and the spatial disper-
a linear Stark effect takes place.* It is assumed be- sion, as follows from (3.59) and (3.61), manifests itself
low, naturally, that the expansions employed are valid. only in the second order in k, while in crystals without
Under the conditions when the tensors ejj(w, k, an inversion center the expansions (3.59) and (3.61) con-
E ( 0 ) . . . ) or e^Hш, k, E < 0 ) . . . ) cannot be simply ex- tain also terms that are linear in k. This circumstance
panded in powers of the corresponding variable, we leads to the "magnetic field reversal effect": in this
can in all the cases known to us represent these ten- case, generally speaking, eij(w, k, H ( 0 ) ) * eij(w, k,
sors as ratios of polynomials [in the case of expansion — H < 0 ) ), so that the frequencies of the Coulomb prob-
in k, see, for example (2.14)]. For the example given lem, and consequently also the position of the absorp-
above it is obviously necessary to make use of the r e - tion lines in the spectrum, change upon reversal of the
lation magnetic field.
2
For the analysis that follows it is convenient to r e p -
-<4o) 2 - 2a (to»- mfo) - (3.61) resent the tensors A{j} m , and also в Щ т , in the form
off sums:
As was already indicated in Sec. lb, knowledge of
(3.63a)
the dielectric tensor and its inverse enables us to de-
termine the dispersion of the natural frequencies of (3.63b)
the Coulomb problem, the "fictitious" longitudinal
where
waves, and the "polarization waves," which c o r r e -
spond to poles of п 2 (ш) if spatial dispersion is dis- W,— AW.
regarded, and which consequently determine the line
positions in the absorption spectrum. If spatial dis- With r e s p e c t to t h e i r s y m m e t r y p r o p e r t i e s u n d e r rota-
persion is taken into account, the determination of the t i o n s a n d p e r m u t a t i o n of t h e i n d i c e s , t h e p s e u d o t e n s o r s
A
absorption line positions is generally speaking more i i l m a n d B ijlm а г е Р е г ^ е с ^ У analogous to the tensor
a
complicated, but for the case of weak dipole and also ijZm (see Section 2b, Table III, for noncentrosvmmet-
quadrupole absorption lines, when the "mixing" of the ric classes). As regards the pseudotensors А Ш Щ
transverse photon and exciton states of the Coulomb and в | | | щ , they can be represented in the form
problem can be neglected, the line position in the ab-
A§£ = elmnCnij, (3.64a)
sorption spectrum is determined by the value of the
natural frequency of the Coulomb problem, taken for B$£ = elmnDnij. (3.64b)
a wave vector value equal to the light wave vector in Inasmuch as the completely antisymmetrical unit ma-
vacuum Q = 27rs/A.o (for more details see ^ 6 3 ^).t In trix e ; m n is a pseudotensor, C n ij and Dnij are ordi-
the presence of the external actions considered here, nary third-rank tensors, with C n ij = C n ji and Dnjj
the dependence of the frequencies of the "fictitious" = D n jj. Consequently, the symmetry properties of the
longitudinal waves and "polarization wave" frequen- tensors C n jj and D n jj fully coincide with those of the
cies on the wave vector is determined, in accord with tensor which determines the piezoelectric effect (see
(1.35) and (1.36), from the equations [26]
, Chapter VII). In view of the fact that [ 2 6 ] indi-
ei}(со, k, E(0\ H ( 0 ) )s i S j = 0, (3.62a) cates the nonvanishing components of this tensor for
different crystal classes, we shall not dwell on this in
and detail here. Using (3.63) and (3.64) we find that
*In o r d e r t o e x p l a i n t h i s , l e t u s c o n s i d e r a medium for w h i c h = Cni] [Ни>к]„ (3.65a)
6 = 1 - 2a/(<u2 - u>j0) when Е (о > = О, and
= Dnij [H 'k]n <0
(3.65b)
•E(O)
~> ~>"-±- 2<o;o

in the presence of the field. Then in first approximation

Let us consider first the influence of an external mag-
netic field on the dispersion of the frequencies of the
(О2-со20-ц|Е(0)| "fictitious" longitudinal waves. To this end, substi-
Hence tuting in (3.62a) the expression (3.59) with E ( 0 ) = 0,
2au2 I E ( 0 ) I2
2
2a
we conclude that the sought frequencies satisfy the
((o — (Bjo—2e)2 ((02 — <o,o)
2
(0 — cojO — 2a
equation
Obviously, the term with | E ^ | 2 increases without limit in this
expansion as ы -> a>j 0 . y (со) *,•+ + (C [H'0) k])
t F o r the sake of simplicity we disregard here the influence of
other resonances. \°>H% = 0, (3.66)
692 V. M. A G R A N O V I C H a n d V. L . GINZBURG
where the components of the vector С are T a b l e VI. C h a r a c t e r s of i r r e d u c i b l e r e p r e -
s e n t a t i o n s of t h e C ^ g r o u p
Cn = CnijsiSj, и = 1, 2, 3. (3.66a)
\^ Symmetry
Confining ourselves to the vicinity of one of the r e s o - ^ ^ ^ operations
nances, we set Irre- ^^^ E C2 2Ce
ducible ^^
(3.67) representations ^ \

Аи z 1 1 1 1 1 1
where KwjJO) i s t n e exciton energy in the L-th band Л 1 1 1 1 —1 -1
for к = 0. Substituting now (3.67) in (3.66) and con-
вг 1 —1 —1 —1 1 —1
fining ourselves to terms linear in H < 0 ) , we find that £i 1 —1 -1 —1 -1 1
the sought frequency of the "fictitious" longitudinal Ег 2 2 —1 —1 0 0
wave is determined by the relation Ei, x, у 2 -2 1 1 1 0
i .
2(oI,(0)eoi,j1silsil
(3.68) function of t h e d i r e c t i o n of t h e v e c t o r s , i s g e n e r a l l y
speaking not directed along the optical axis. The latter
and thus w(k, H ( 0 ) ) * w ( k , - H ( 0 ) ) . occurs when the wave vector к is either perpendicular
Let us consider in greater detail the influence of the to the optical axis or parallel to it.
reversal of the magnetic field on the frequencies of the To interpret the magnetic field reversal effect, it
"fictitious" longitudinal waves in crystals of the CdS is proposed in E52-' that in the expression for the exci-
type (space group c | v ) - The characters of the group ton energy there enters an additional term ( R / c m e x c )x
Cev are listed in Table VI (in the notation of [ 4 5 J ) . (d • [ H <0) x к ]), corresponding to the interaction be-
It follows from this table that the states of the tween the dipole moment of the exciton and the electric
mechanical excitons at к = 0, which can be excited by field ( Й / с т е Х с Н к x H ( 0 ) ] , which in the opinion of the
light in the dipole approximation, have the symmetry authors of [52] (see also C50>51^) occurs in the p r e s -
of representations A t and E 2 of group Cgy If the ence of an external magnetic field in the coordinate
wave functions of the exciton in the L-th band t r a n s - frame connected with the moving exciton. It follows
form at к = 0 in accordance with representation A1? from (3.68) that in the expression for the exciton en-
then gjj = gA^is^s» a n d ^ t n e v transform in accord ergy ficu(k, H < 0 ) ) a term of the type C [ H ( 0 ) x k ] does
with E 2 , then g^ = g E ^ i i S j j + 6 i 2 6j 2 ). indeed arise but the vector C(s) is not fixed along the
For crystals belonging to class Cgy, the nonvan- optical axis even in a uniaxial crystal. It also follows
ishing components of the tensor Cnij are from (3.68) that the effect of the reversal of the mag-
netic field vanishes when H ( 0 > II k, for in this case
— L-»a» — C113 — C311 — Cs, (3.69)
H ( 0 ) x к = 0 and AJj^sjSjS^m = 0 [see (3.70)].
Using further Table III, we find the nonvanishing com- The same takes place if the vector к is directed
ponents of the pseudotensor AJj| m in crystals of the along the x axis, and the magnetic field along the op-
Cgy class : t tical axis, for in this case AfjJ^SjSjSjH^ = 0 and
лай Л<1)е_ ла)с ла)е C [ H < 0 ) x k ] =0.
We recall that the frequencies of the "fictitious"
Л Л
ш
А , = Л
Ш 2
— — 2212 — — 1211>

(3.70)
longitudinal waves determine the positions of the ab-
It f o l l o w s f r o m ( 3 . 6 6 a ) a n d (3.69) t h a t sorption lines of light propagating only in such direc-
tions, for which the light waves are not transverse.
n ОГ <• с Г 9P с е Г С /o'Xf'V
Ь г — £,U113b1i3, L- 2 — ^ О 2 2 з * 2 6 3 ' ^ 3 — u 8 1 U 4 l T s « For crystals of the CdS type this occurs if the wave
(3.71)
vector к is directed neither along the optical axis
nor perpendicular to it. Otherwise the position of the
Thus, the vector C, being, like the gyration vector, a absorption lines is determined by the polarization
lp) (P) wave frequencies. We therefore consider the influ-
*(CtH k]) = C-[H xk].
tWe assume in (3.70) that the x and у axes lie in mutually
ence of the external magnetic field on the frequency
perpendicular planes of symmetry av and a'v. In order to estab- dispersion of the polarization waves, confining our-
lish the nonvanishing of the component of t he pseudotensor AfjVm. selves to one particular case of type CdS crystals and
we can proceed as follows. We first use the limitations imposed assuming that the vectors H < 0 ) and к as well as the
on the pseudotensor A $ £ by the presence of a sixfold axis. These optical axis form a mutually orthogonal triplet. Let,
limitations cause the nonvanishing components of A $ £ in c l a s s for example, in accordance with this choice, k 2 = k 3
Ce v to correspond to those non-vanishing components of the ten-
= 0 and н{ 0 ) = Нз 0) = 0. As already indicated in Sec. 2,
sor cLijim in the crystal class Сб, which reverse sign upon reflec-
tion in the symmetry planes av and o'v of the group CQV. The
the gyrotropy of crystals belonging to class Cgy leads
latter i s connected with the fact that the components of the pseudo- to second-order effects in powers of ( a / \ ) . We are
tensor A $ £ remain unchanged in this case. interested here in whether the expression for the po-
 CRYSTAL O P T I C S WITH ALLOWANCE FOR S P A T I A L D I S P E R S I O N 693
l a r i z a t i o n - w a v e f r e q u e n c i e s c o n t a i n s t e r m s of t h e i s n e c e s s a r y t o t a k e a c c o u n t h e r e of t h e f a c t t h a t w h e n
o r d e r of H
<0) <0)
and kH . T h e r e f o r e , in investigating a n e x t e r n a l e l e c t r i c field i s a p p l i e d a l o n g t h e h e x a g o n a l
the dipole l i n e s we confine o u r s e l v e s to t h e following a x i s , t h e s y m m e t r y of t h e C d S c r y s t a l d o e s n o t c h a n g e ,
e x p r e s s i o n f o r t h e t e n s o r ejj :
1
s o t h a t t h e n o n z e r o c o m p o n e n t s of t h e p s e u d o t e n s o r
BijZim c o r r e s p o n d t o n o n z e r o c o m p o n e n t s of t h e f o u r t h -
ej/ (со, k, #< ^ + B'mH\m + В%тНГкт. (3.72)
rank pseudotensor В - | | т . On the basis of the remarks
In crystals of class Cgy, the tensor Bmi in (3.60a) is made, we reach the conclusion that in the case under
diagonal, with B u = B 2 2 * B 3 3 . Therefore, and also by consideration the Eqs. (3.62b) have the following form:
virtue of relations (3.65b),
e £ (со) + (Z) 3 a 2
ej/ (со, k, H<°>) = eoh + ewB22Hm + DaijHwk + 2Щ\ЦГтк. m
= 0, (3.76a)
(3.73) £>;=.D;=O,

Inasmuch as, in analogy with (3.69) and (3.70), D3jj •& И + D333H<°>k + B333E«> + B£, (£•)• + B3U23kH<°>E«> = 0, 1
* 0, only if i = j = 1, or if i = j = 2, or if i = j = 3,
while В-]}0 * 0 only if i = j = 1 or i = j = 2, and (3.76b)
furthermore ejjA = 6 ij e ojV E 1 S - ( 3 - 6 2 b ) for the t r a n s -
Let us examine in greater detail the vicinity of the di-
verse components of the vector D' (DJ and D 3 ) have
pole line, polarized along the optical axis. In this case,
the following form [see also (1.37)]:
according to (3.67), for ш « W L ( 0 ) , we have
H«>k) D't = 0, -x „ Ш-0)1(0)
+ D333Hmk)D'a = 0. (3.74)

It follows from (3.74) that the position of the line polar- T h e r e f o r e , a s f o l l o w s f r o m ( 3 . 7 6 ) , t h e p o s i t i o n of t h e

ized perpendicular to the optical axis is determined in a b s o r p t i o n line i s d e t e r m i n e d by t h e following r e l a t i o n :
this case by the equation
« ( k , H<°\ E«>)
) + DtttHmk + 2В™,Нтк = 0, (3.74a)
cH(nEm}. (3.77)
whereas the position of the line polarized along the
optical axis is determined by the equation It follows from (3.77) that for fixed values of k, H ( 0 ) ,
e
and E ( 0 ) / E ( 0 ) , but with variation of E ( 0 ) , the frequency
om (<°) + D333Hmk = 0. (3.74b) of the absorption line is displaced along a parabola,
Thus, the reversal effect can occur both for lines whose position changes, in particular, upon reversal
that are polarized perpendicular to the optical axis of the direction of the magnetic field. An effect of this
(representation E t ) and for lines polarized along the type is indeed observed in CdS crystals'- 50 » 51 -'. Accord-
optical axis (representation Aj) (see Table VI). ing to E50>613, the effect of the absorption frequency
Let us now assume that external electrical and mag- shift is interpreted in the following manner. In the
netic fields are present simultaneously. Let us con- presence of a magnetic field, in a coordinate system
sider, as before, a crystal of the type CdS and assume connected with a moving exciton, an additional electric
that the magnetic field is directed along the у axis, the field arises, (0)
the intensity of which is ( K / c m e x c ) у
vector к along the x axis, and the external electric [к х Н ] . Therefore, the external field acting in the
field along the optical axis. In this case in lieu of exciton on the electron and hole is equal to E < 0 )
(0)
(3.73) we have on the basis of (3.61), accurate to terms + ( K / c m e x c ) [k x H ] , which leads (in media where
linear in H ( o \ the quadratic Stark effect occurs) with a change of
E ( 0 ) to a shift in the exciton energy along a parabola
m
ег/ (со, k, + (D similar to (3.77).
3ij
Were such an interpretation of the effect correct,
m m then the determination from the experimental data of
+ Bijl23kH E . (3.75)
that value of E ( o > which corresponds to the minimum
We note that the tensor Bjjj is analogous in its displacement of the term would enable us to determine
symmetry properties to the tensor Cftj, the tensor the "effective m a s s " of the Coulomb exciton, m e x c ,
B
i i l m i s a n a l ° g ° u s t o the pseudotensor AJj] m , while from the known H < 0 ) and k. Actually, as follows from
the tensors B j | ] m and B j | | m are analogous to the ten- (3.77), it is possible to determine in this manner not
sor aijZm. Using (3.69), (3.70), Table III, and also [ а д the "effective m a s s " of the exciton, but merely the
we arrive at the conclusion that relation between the coefficients contained in (3.77),
and in particular the ratio В 3 3 1 2 3 / в | 3 | 3 , which although
B233 = B323 = 0, B223 Ф 0, B 333 Ф 0,
it does have the dimensionality of the ratio R / m e x c c ,
D(2) -nil) . D(2) (\ D(3) D(3) 73(3) f\
B
•°2228 — ^ 3 3 2 3 — • 2323 ~
U
>
U
-О2221 ~ ^ 3 3 2 1 = - " 2 3 2 1 = > does not, generally speaking, reduce to the latter. We
я25ззз=BSS. = о. в^ Ф о, в^ Ф о. note also that the aforementioned parabola should shift
As regards the fifth-rank pseudotensor В ц ; т п , it upon reversal of the magnetic field not only because
694 V. M. A G R A N O V I C H a n d V. L . GINZBURG
(3.77) contains a term proportional to B33123, but also boundary planes perpendicular to the optical axis. It
because of the presence of the term D 3 3 3 H (0) k, which turned out that the SBC depend essentially on which
leads to a reversal even when E < 0 ) = 0. In this connec- molecules within the unit cell terminate the crystal,
tion it must be borne in mind that no reversal of the and reduces to the vanishing of only one of the t r a n s -
magnetic field was observed in CdS at E ( 0 ) = 0 in C51^, verse components of the polarization vector [ for ex-
in contrast with E52^. Consequently, further research ample, P x ( 0 ) = 0, e t c ] . This makes the directions
in this field is highly desirable. In conclusion, we note x and у no longer optically equivalent. So sharp a
that to investigate the fine effects connected with ex- dependence of the SBC on the structure of the surface
ternal influences on absorption spectra, it is possible layer should either not influence the measured physi-
to use not only crystals but also, for example, gyro- cal quantities, or contradicts the main assumptions
tropic amorphous media, provided the absorption spec- on which the use of the tensor €jj(a), k) is based when
tra of these substances display at low temperatures finite crystals are considered.
sufficiently narrow lines, similar to those observed We note also the article ^83^, in which a conceptu-
in E84-! for molecules in frozen solutions. ally correct method was proposed for considering
e) The problem of boundary conditions. We have electromagnetic waves in finite crystals with account
considered above only the propagation of waves in an of spatial dispersion. In addition, the question of the
unbounded medium. Yet in crystal optics one deals SBC in the vicinity of dipole and quadrupole exciton
always, strictly speaking, with a bounded medium, so absorption bands is considered in '-83-'. It is shown
that the question of boundary conditions arises. If we there that the SBC assumed in ^ are, generally
use the general relation (1.3), it is sufficient to take speaking, incorrect.
into account the ordinary electrodynamic conditions. Let us consider the problem of the SBC from a
On the other hand, the introduction of the tensor point of view that is somewhat different from that in
eij(o), k) corresponds to relation (1.4), which is rigor- [7,82,8з:# W e n o t e f i r s t o f a l l t h a t w h e n a c c o u n t i s taken
ously valid only in an unbounded medium [the p r e s - of spatial d i s p e r s i o n with a c c u r a c y to t e r m s of o r d e r
ence of a boundary leads to the dependence of the ker- k 2 , away from the surface of the c r y s t a l , the p o l a r i z a -
nel of ejj in (1.3) on more than just the difference tion
r - r ' ]. At the same time, it is clear from physical
considerations that the tensor eij(u),k) can actually
be employed for a bounded medium, provided the char- in the vicinity of one selected dipole exciton absorption
acteristic dimensions of the latter are R » a ~ 3 x 10~8 band satisfies the following system of equations (see
cm. Such a conclusion is valid, but at the same time [24,86])
there may arise the need for supplementary boundary
conditions (SBC). Indeed, this usually occurs when
(3.78)
using expansions of the type (2.12) and (2.13), which
are patently connected with an increase in the order where the tensors /3jj, уц1, с * ф т , and Хц do not de-
of the field equations and with the appearance of new pend in first approximation on ш and are determined
waves (in particular, we have in mind longitudinal by the crystal symmetries, hi nongyrotropic crystals,
waves). TijZ = 0, whereas in gyrotropic crystals in first approxi-
mation we can put ayim = 0. In the system (3.78) the
The character of the SBC is determined by the phys- electric field intensity E ( r ) appears as an external
ical properties of the medium and of its surface, and force, in which connection E ( r ) should be contained
consequently is not universal. For example, in the in the boundary conditions for the polarization P ' ( r )
case of a relation of the type (3.13) and specular r e - (these boundary conditions are, strictly speaking, in-
flection of the electrons from a semi-infinite plasma, deed the SBC). The general form of the boundary con-
the SBC have the form E n = 0, where E n is the nor- ditions which the polarization P ' ( r ) should satisfy in
mal component of the field E on the internal surface
the presence of E ( r ) * 0 is obviously the following
of the plasma (we have in mind the problem of a wave
(the subscript zero corresponds to the surface of the
with field E lying in the plane of incidence, impinging
crystal):
from vacuum on the plasma boundary E81^). In crystals
the form of the SBC differs with the spectral region + pi,) P'j (0) + Y«I = KiEi (0), (3.79)
under consideration. Thus, in article И the SBC for
the vicinity of the absorption line corresponding to the and it is very important that the tensor /Зц * 0ц,
excitation of dipole excitons in nongyrotropic crystals Mj A.jj, and .* Inasmuch as the SBC serve
was found to be P = 0, where P is the part of the po-
larization due to the contribution of the considered ex- *Thus, in media with an inversion center, where yy; = 0, the
citon band. In E82^ the method of '-'-' was used to de- tensor Yiji ^ 0. This is connected with the fact that in the pres-
ence of a boundary surface even a nongyrotropic crystal is not
rive the SBC in a gyrotropic crystal with the space invariant against inversion. The problem of boundary conditions
group D|. The case considered was that of propaga- and the meaning of Eq. (3.68) can be understood qualitatively by
tion of light along the optical axis of a crystal with using a one-dimensional chain of coupled oscillators as an example.
 CRYSTAL O P T I C S WITH A L L O W A N C E FOR SPATIAL DISPERSION 695
only to establish the relations between the wave ampli- themselves, and not of the reflection or transmission
tudes, the contribution of the terms containing deriva- coefficients. In this connection we note also that the
tives 9Pj/8x; can be neglected in (3.79) (long waves). dependence щ(ш, s ) , i.e., the function wj(k), can be
Therefore the SBC can be rewritten in the form determined not only from crystal optics. For example,
the function o>/(k) can be found by studying the com-
P[ (0) + Ti}Ej (0) = 0, (3.80)
bination (Raman) scattering of x-rays '-80-'. In this
where the tensor Гц is practically independent of the case the character of the boundary conditions for the
frequency in the spectral region under consideration, excitons generally does not play any role at all.
by virtue of the fact that /?j= */3jj. For specific crys- Finally, we note that when solving the problem of
tals we can determine the tensor Тц only from the reflection and refraction of light on a boundary, as well
microscopic theory. The symmetry properties of this as more general electrodynamic problems, it is some-
tensor are determined by the symmetry of the finite times advantageous to use the reciprocity theorem
crystal. Thus, for example, in the case of a uniaxial (see, for example, ^ , Sec. 69, and M , g e c # 29). In
crystal, bounded by a plane perpendicular to the op- the presence of an external magnetic field this theorem
tical axis (the z axis), the only nonvanishing compo- is not valid in its usual form, but there is a generalized
nents are Гхх = Гуу and r z z . In the case when the reciprocity theorem ^ . When spatial dispersion is
plane bounding the crystal is parallel to the optical taken into account, both the ordinary and the general-
axis, Гхх ^ Гуу * r z z . Relation (3.80) goes over into ized reciprocity theorems remain fully in force ^ .
that obtained in ^ if we put Тц = 0, which generally f) Experimental investigations of effects of spatial
speaking is not correct. In the case of gyrotropic m e - dispersion in crystal optics. Let us dwell on the r e -
dia, the use of (3.80) does not lead to redundant condi- sults of the experimental investigations of spatial dis-
tions only if the transition aijfon —- 0 is made in the persion, first in gyrotropic, and then in nongyrotropic
final expressions for the amplitudes. crystals.
We note that all the SBC assumed in t 5 > 7 ] are of the Gyrotropic crystals. The problem of the nature of
type (3.80). We have considered above dipole transi- the gyrotropy of crystals is far from new, for it was
tions, as manifest in the fact that the tensor Ajj in already pointed out in t 65 > 66 ^ that one is dealing in this
(3.78) was independent of k. For quadrupole transi- case with the effect of spatial dispersion of order a/A.
tions Ajj = MijZm^zkm and the resultant SBC are the Nonetheless, experimental investigations of the optical
conditions (3.80) with F{j replaced by a certain tensor properties of gyrotropic crystals are quite scanty. In
Тц(ш, k) = £2ijfonkzkm, so that when t e r m s of order k2 particular, the data are most skimpy on the frequency
are neglected we have in lieu of (3.80) the conditions dependence of the angle of rotation of the plane of po-
n' л ; -19 4
larization in crystals near individual absorption bands.
/ l = L/, I — 1, £ , O. The only results in this respect are those of '-67-',
The tensor rjj(oj,k) is equivalent within the frame- where several crystals with the symmetry of quartz
w o r k of the p h e n o m e n o n o l o g i c a l t h e o r y to the t e n s o r were investigated (quartz, cinnabar, benzyl). It was
ejj(uj, k ) in the s e n s e that both t h e s e t e n s o r s a r e spe- shown there that in the vicinity of the investigated in-
cified and only their g e n e r a l p r o p e r t i e s and the e x - dividual bands (with the light propagating along the
p r e s s i o n s obtained w h e n they a r e used in the field optical axis), the rotation of the plane of polarization
equations need to be analyzed. On the o t h e r hand, the per unit length of light beam path in the crystal (the
calculation of rjj(w,k), l i k e t h e c a l c u l a t i o n of specific rotation) has the following dependence on the
q j ( w , k ) , i s a problem in the m i c r o s c o p i c theory. frequency of the light*
We note that in the a b s e n c e of n e w w a v e s , the re-
f i n e m e n t of the b o u n d a r y c o n d i t i o n m a y a l s o b e of i n - ^ (3.81)
terest. T h u s , a c c o u n t of the t e r m s of o r d e r a/A in
When the light propagates at an angle to the optical
the m i c r o s c o p i c c a l c u l a t i o n of the r e f l e c t i o n of light
axis, the circularly polarized waves are replaced by
from a cubic lattice is equivalent to introducing a c e r -
elliptically polarized ones with different signs of rota-
tain t r a n s i t i o n l a y e r o n the s u r f a c e of the c r y s t a l (and
tion and with different orientations of the axes of the
the appearance of the transition l a y e r l e a d s in turn to
ellipse. In this case when the waves pass through the
d e v i a t i o n s f r o m t h e F r e s n e l f o r m u l a s ) . C83a3
crystal, a phase difference is produced in them, which
bi c o n c l u s i o n w e w i s h t o e m p h a s i z e t h e f o l l o w i n g .
is determined not only by the gyration effect, but also
T h e s p e c i f i c f o r m of t h e S B C g o v e r n s o n l y t h e r a t i o s
by the ordinary birefringence, which naturally com-
of the a m p l i t u d e s of the d i f f e r e n t w a v e s . The latter
plicates somewhat the investigation of the rotation dis-
feature is essential (many specific results obtained
persion. It is shown in E67^ that in quartz with light
on t h e b a s i s of c e r t a i n p a r t i c u l a r S B C c a n b e found in
[7,82,83]^ kuj. j n e v e r t h e l e s s of s e c o n d a r y importance
s
*The theory of the rotating ability of crystals in the visible
within r a t h e r wide l i m i t s — f r o m the point of v i e w of an
region of the spectrum and in the ultraviolet is developed in
investigation of c r y s t a l s , the principal r o l e i s played [24,6»,69j_ fhjg thgory permits, in particular, an interpretation of
by a determination of the dispersion curves щ(ш, s ) the results of M .
696 V. M. A G R A N O V I C H a n d V. L. GINZBURG
propagating perpendicular to the optical axis the rota- of the intensity of the monochromatic light passing
tion dispersion also obeys a relation of the type (3.81), through a plane parallel plate of the crystal, as a func-
but with a different coefficient K^ * K||. tion of the thickness of the plate. The indication of the
Particular interest is attached to investigations of existence of two waves in a crystal would be their in-
the effect of gyrotropy at low temperatures, the only terference on leaving the crystal; this interference
time when manifestation of the fine effects of spatial should lead to oscillations in the intensity as a function
dispersion, connected in particular with the structure of the thickness. It is assumed here that the oscilla-
of the exciton bands, etc, can be expected. We know of tions connected with the reflection and triple passage
only one investigation of this type E'0^, in which circu- of one of the waves are either insignificant or can be
lar dichroism was investigated at helium and hydrogen suitably accounted for.
temperatures near individual absorption lines in a so- The experimental data obtained in E74^ show that the
dium uranyl acetate crystal. absorption of light in anthracene plates in the region of
For molecules in solution, the magnitude of the c i r - the intrinsic absorption band with maximum at oij
cular dichroism is usually small, of order a/\, where = 25,200 c m " 1 at T = 20°K actually oscillates as the
a is the dimension of the molecule and \ is the wave- thickness is varied. With this, an oscillatory depend-
length of light, so that а/Л. RS 10~2—10~3. In sodium ence on the distance is obtained for In I/I o (I is the
uranyl acetate, however, the circular dichroism, i.e., intensity of the transmitted light and 1ц that of the in-
the ratio of the difference of the coefficients of ab- cident light) in the thickness region 0.05—0.3ц with a
sorption of light circularly polarized in the right-hand distance (difference in plate thickness) between ab-
and left-hand directions, to their sum turned out to be scissas of the maxima amounting to Ad s 0.06 ц. The
close to unity for individual lines.corresponding to oscillations were most clearly pronounced at a light
1
strong absorption of only one of the circularly polar- frequency w = 25,108 cm" . The authors of '-'*-' p r o -
ized waves. Inasmuch as sodium uranyl acetate con- pose that the presence of the oscillations is evidence
sists of nongyrotropic molecules, the observed phe- of the presence of two identically polarized waves with
nomenon is only characteristic of the crystalline state a difference of 6.9 in the refractive indices.
and is the consequence of the difference between the It must be noted, however, that such an interpreta-
structure of the exciton bands, corresponding to the tion of the oscillations contradicts the calculations
right-hand and left-hand circular polarizations of the given in Sec. 3c for the absorption and dispersion
dipole-moment vector (see £63,71]^ Research in this curves. Indeed, according to the data of '-72-' and also
direction is of undoubted interest. '-33-', in anthracene this transition corresponds to a
There are still no experimental investigations of value А ю 0.1 [see (3.12)]; such a value of A at 4;
the effects connected with new waves in gyrotropic = (w-wj)/wj » 4 x 10" 3 yields n 2 = -A/$ = 25, in
crystals (see Sec. 3a). In this connection it should be accordance with ^72-', which gives for б = 10~3,
noted that such research at low temperatures would be I = - 4 x 10~3 (Fig. 5d) for the anomalous wave n + w 35
of great interest, since the presence of three waves in and к + « 9. Consequently, the anomalous wave practi-
a crystal, having the same frequency and different r e - cally cannot " r e a c h " the second surface even when d
fractive indices, absorption coefficients, and polariza- « 0.1 ц, inasmuch as the factor exp ( - 27TKd/\0) ю 10" 6
tion, can lead to the manifestation of a unique oscilla- « 1 when \ 0 = 0.4 ix. The estimates made correspond
5
tory dependence of the specific rotation on the thickness to | /3' | = 10" . Actually, however, the quantity | j3' |
of the crystal. It would be natural here to investigate is apparently (see Sec. 3b) much smaller than this
first of all gyrotropic cubic crystals or the propagation value, so that the role of the absorption increases even
of waves along the optical axis in crystals of lower further.
symmetry (in these cases normal waves are circularly The source of the oscillations observed in ^74-' is not
polarized). clear. It is possible that they may be due to the inter-
Nongyrotropic crystals. Experimental investigations ference of a ray that has passed through the plate once
of optical effects connected with spatial dispersion of and one that has passed through the plate three times.
the dielectric tensor were carried out with the inor- Another possibility is apparently an effect, not con-
ganic crystals CUJJO 1137 ' 46 - 4811 and CdS С 5 0 " ^ , ^ QISO trolled in ^74-', due to the interference of rays reflected
with organic crystals (anthracene, stilbene, and oth- from the front and r e a r sides of the film. If it is a s -
e r s ) C72~743. we have already mentioned (see the In- sumed that the singly and triply transmitted rays inter-
troduction and Sees. 3c and d) the investigations of fere and absorption is neglected, then anthracene should
optical anisotropy in the cubic crystal Cu2O in the have a refractive index n = \ 0 /2Ad я 3.45 (see £74,75:
vicinity of the quadrupole absorption line, and also in- and below). This contradicts the value n k, 5, obtained
vestigations of the external influences on the optical in E72^, but this discrepancy is still not convincing,
properties of Cu2O and GdS. since the relation n = \ 0 /2Ad is incorrect in the p r e s -
A method for proving experimentally the existence ence of noticeable absorption.
of additional (new or anomalous) light waves was pro- Analogous experiments were made in '-76-' with a
posed in [1*1. The method reduces to a measurement cuprous oxide crystal (see in this connection also the
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL DISPERSION 697

t h e o r e t i c a l . p a p e r ^ 7 T ^ ) . The difference between t h e s e r u p o l e a b s o r p t i o n . Thus, for e x a m p l e , for a wave

e x p e r i m e n t s and t h o s e r e p o r t e d in ^ 7 4 ^ l i e s in t h a t the p r o p a g a t i n g along the diagonal of a face, a new wave
o s c i l l a t i o n s w e r e o b s e r v e d in the c u p r o u s oxide in the in the vicinity of the exciton t r a n s i t i o n , c o r r e s p o n d i n g
r e g i o n of the quadrupole exciton a b s o r p t i o n for the to the r e p r e s e n t a t i o n F 2 , c a n a p p e a r only if the v e c t o r
quadrupole line Ло = 6125 A at T = 77°K; t h i s line is E is p e r p e n d i c u l a r to the plane of the face of the cube
a t t r i b u t e d to the " y e l l o w " exciton s e r i e s of Cu 2 O, and [ s e e (3.49) and (3.50); t h i s indeed o c c u r s in Cu 2 O [ 7 6 ] ] .
i s a s c r i b e d a value n = 1. The m e a s u r e m e n t s w e r e We note t h a t in g y r o t r o p i c c r y s t a l s in the region of
c a r r i e d out at T = 93°K, and the d i s t a n c e between the quadrupole absorption, when the wave v e c t o r of the
a b s c i s s a s of the o s c i l l a t i o n m a x i m a was a p p r o x i m a t e l y light i s d i r e c t e d along the optical a x i s , we have g e n e r -
0.2 m m , which is t h r e e o r d e r s of magnitude l a r g e r than ally speaking
the p e r i o d of o s c i l l a t i o n s which would o c c u r in the c a s e k k ш
k
) = ijlm °) l m К - ? (0)
a
of i n t e r f e r e n c e of multiply reflected waves (Ad = Ло / 2 n
RS 10~* m m ) . On the o t h e r hand, if we a s s u m e t h a t the
o b s e r v e d o s c i l l a t i o n s a r e due in t h i s c a s e to the m a n i -
+ Oiilft. И klkm [ » 2 - «? (0) + ? A + gnn-knkn-] !.
festation of the anomalous wave, we m u s t a s s u m e t h a t
the d i s t a n c e between m a x i m a * Ad = \ 0 / 2 ( n t - n 2 ) , It f o l l o w s t h e r e f o r e t h a t a n e q u a t i o n of t h e t y p e (3.82)
w h e r e щ and n 2 a r e the r e f r a c t i v e indices of the o r d i - can have in a definite frequency region not two but
n a r y and a n o m a l o u s w a v e s . T h e r e f o r e , when Ad RJ 2 t h r e e s o l u t i o n s . T h e r e f o r e a s t u d y of t h e d i s p e r s i o n
x 10~ 2 c m and \ 0 « 6 x 10~ 5 c m , we obtain щ - n 2 of w a v e s i n t h e r e g i o n of q u a d r u p o l e a b s o r p t i o n l i n e s
и 10"3. in g y r o t r o p i c c r y s t a l s m a y a l s o t u r n out t o b e quite
hi the vicinity of an isolated quadrupole r e s o n a n c e , promising.
in a c c o r d with (2.14), without account of a b s o r p t i o n , the hi c o n c l u s i o n w e n o t e t h a t a n a t t e m p t t o o b s e r v e
r e f r a c t i v e index s a t i s f i e s the equation o s c i l l a t i o n s i n t h e i n t e n s i t y of t h e t r a n s m i t t e d l i g h t
w a s u n d e r t a k e n a l s o i n '-78-'. H o w e v e r , n o p h e n o m e n a
(3.82) t h a t c o u l d n o t b e e x p l a i n e d o n t h e b a s i s of o r d i n a r y
crystal optics were observed.
where
In o r d e r t o d i s c l o s e t h e r o l e of s p a t i a l d i s p e r s i o n
ft)2 under certain conditions we can use the dispersion r e -
с2
lations. Thus, for an optically isotropic m e d i u m in the
with a b s e n c e of s p a t i a l d i s p e r s i o n a n d of a n e x t e r n a l m a g -
n e t i c field, t h e d i s p e r s i o n r e l a t i o n s (1.12) h a v e t h e f o r m

It follows from (3.82) that

° (1.12a)

On the o t h e r hand, u n d e r the conditions d i s c u s s e d h e r e

When | = 0 we have n2 = 0 and n2 = e0 — v^/^. T h e r e - we have
fore, if e0 is approximately equal to the ratio v^/\x\,

the difference nt — n2 can be very small. Inasmuch as

both e0 and v^/nx can, generally speaking, a s s u m e in- T h e r e f o r e , s u b s t i t u t i n g i n (1.12a) t h e e x p e r i m e n t a l l y

dependently different values for different quadrupole obtained v a l u e s of ej = п 2 - к 2 and е{' = 2пк, we can
lines, and also in different crystals, the small value verify w h e t h e r t h e s e r e l a t i o n s between cjj, ej', and n
n
l ~
n
2
ет
10~
3
obtained in Cu2O is probably not char- and к a r e c o r r e c t o r not [ in the p r e s e n c e of spatial
acteristic. In t h i s sense a study of analogous effects d i s p e r s i o n the f o r m u l a s ej(w) = n 2 — к2 and ej'(w)
in the vicinity of other quadrupole lines is of great in- = 2пк a r e of c o u r s e no l o n g e r v a l i d ; s e e , for e x a m p l e ,
terest, hi expression (3.82), v^ is in fact a certain formula (1.29) for t r a n s v e r s e waves in an i s o t r o p i c
c o m b i n a t i o n of components of a fourth-rank tensor m e d i u m ] . An a t t e m p t to d i s c l o s e the r o l e of spatial
[ s e e (2.24)—(2.15) and (2.33a)]. It i s obvious there- d i s p e r s i o n by t h i s method w a s u n d e r t a k e n in ^ 7 3 ^. F o r
fore that vx * 0 only for w a v e s that experience quad- s e v e r a l c r y s t a l s at low t e m p e r a t u r e s , it w a s observed
that r e l a t i o n s of type (1.12a), which do not take s p a t i a l
•Obviously, we are referring here to the addition of two oscilla- d i s p e r s i o n into account, a r e violated, and that t h e s e
tions Acos(<jt- u>ntd/c) and В cos(wt - <jn 2 d/c). The time aver- violations gradually d i s a p p e a r e d with i n c r e a s i n g t e m -
age of the square of the amplitude of the resultant oscillation is p e r a t u r e . It is obvious t h a t t h e s e investigations should
/=—i:—\-ABcos-7— i—n2) &\ be continued.
2 KQ We note a l s o t h e possibility of o b s e r v i n g a n o m a l o u s
When the thickness of the plate d changes by an amount Ad, the
value of I changes such that Ad = \/2(n, - n2) for the neighboring waves by exciting t h e m with c h a r g e d particles'^ 5 ' 7 9 -',
minima or maxima. This result changes little provided only that the and a l s o the possibility indicated in ^80-' of i n v e s t i g a t -
absorption is sufficiently weak. ing t h e law of d i s p e r s i o n of light w a v e s ( e x c i t o n s ) in
698 V. M. A G R A N O V I C H a n d V. L. GINZBURG
crystals by studying the angular dependence of x-rays in the m e d i u m by the e x t e r n a l s o u r c e s . The,vector
inelastically scattered by a crystal, and accompanying potential of the total field E + E^ will be denoted by
the excitation of excitons. With this, we refer here А„ + б А (E^ is the microscopic field existing in the
primarily to excitons connected with the predominant medium even when E = 0). We denote further by
excitation of electrons. As regards excitons with which Ф п 0 (г, t) the wave functions of the medium in the ab-
the lattice vibrations are essentially connected, these sence of or disregarding the field E, and by Ф п = Ф п 0
can be investigated also by using neutron scattering + бФп the corresponding functions in the presence of a
and a few other methods. field E. The operator of the interaction between the
charges in the external field is, in the linear approxi-
4. CALCULATION OF THE TENSOR eij(o).k) BY mation,
QUANTUM MECHANICAL METHODS
a) Quantum mechanical expression for eij(a),k).
Calculation of the tensor eij(oi,k) for a crystal or for
w h e r e r a is the r a d i u s v e c t o r of t h e a - t h p a r t i c l e
any of its models is a problem in microscopic theory.
with c h a r g e ea and m a s s ma, while £ } denotes
For ionic crystals, such a theory was developed in the
infrared portion of the spectrum long ago (it is de- summation over all p a r t i c l e s of the c r y s t a l and
12 17
scribed in detail in ^ ^; see also ^ ^). Along with it,
several important methodological factors were clari- P « = _ r t _ _ _ i A | 1 ( r e , t).
fied, with a bearing on crystals of arbitrary nature.
In particular, with ionic crystals as an example, the Putting [ s e e (1.40)]
question was considered of the dependence of the fre-
quencies of the normal oscillations (waves) on the
direction of the wave vector к as к — 0. It was shown, we find that
for example, that when full account is taken of the Cou-
lomb interaction, but retardation is neglected, the fre-
quencies of the normal oscillations can be non-analytic where
functions of к (as к - * 0). In this connection it must
be noted that within the scope of a general investigation
of such problems, the consideration of the Heitler- (4.1)
London model of the crystal t 5 3 ' 5 4 ^ introduces nothing
new. It is sufficient to state that in the calculation of Therefore, using perturbation theory, we obtain in
the energy of the excitons in such a model, sums of first approximation (see ^453
dipole-dipole, dipole-quadrupole, and other terms
arise (see, for example, t 4 3 \ the dependence of
« no ^/j
which on к was already considered before (see E тфп
Sec. 30, and also C5S>563). We leave alone the fact that
where
many factors are utterly unnecessary in the micro-
scopic analysis, since the corresponding deductions тоРУпоЛ, ft<omn = Wm-Wn = K(com-oon),
follow directly from the field equations and from the
w
existence of the tensor е^(ш, к ) . By way of an ex- ( m —eigenvalue of t h e energy c o r r e s p o n d i n g to the
ample we can point to relation (1.35) and (1.39), from function Ф т 0 ) . The mean value of the current density
which the dependence of the frequency of the "ficti- in the state Ф п is
t i o u s " longitudinal waves w[\ on the direction of к
as к — 0 is obvious. We have already essentially
dwelt on all these factors in the introduction to this
article.
where
Let us therefore proceed directly to a derivation,
by quantum mechanical means, of a general expression * 161 ¥ n ) = п dx, dx = dtl .. . drN.
for eij(w,k). The corresponding procedure is well
known and consists of finding the density of the current In the linear approximation the mean value of the den-
induced by an external classical electromagnetic field. sity of the current 6j( n ) induced by the field E has the
It is convenient in this case to choose a gauge for the form
potentials, in which the scalar potential of the perturb- tr \
f
ing field is equal to zero, and thus oo/

P 1 ЭЙА

where 6A is the vector potential of the field, induced /\ ш (4.2)

^n
 CRYSTAL O P T I C S WITH ALLOWANCE FOR S P A T I A L D I S P E R S I O N 699
Before we proceed to the Fourier components of the о, к). (4.4b)
current density, let us indicate the selection rules for
T h u s , r e l a t i o n (4.4) i s e q u i v a l e n t t o a r e l a t i o n of t h e
Fmn-
type (2.4). By definition, the tensor €ц relates the
Inasmuch as the wave functions Ф т 0 have the sym-
vector D' with the macroscopic electric field E, which
metry of the irreducible representations of the crystal
is contained in Maxwell's equations (1.1). Therefore,
space group, action on the wave function Ф т о by the in using expressions (4.4) and (4.4b), we should take E
operator T a of the translation of all the electron coor- to mean the macroscopic field, hi this case, conse-
dinates by a whole-number lattice vector a leads to quently, in the calculation of the frequencies and of the
the following result: wave functions of the mechanical excitons, it is neces-
sary to take into account also that part of the interac-
tion, which is related with the difference between the
where the vector q m determines the corresponding i r - field E and the so-called effective field Eg (in the
reducible representation of the translation subgroup simplest case of point-like dipoles, located at the
(see, for example, ^ points of a cubic lattice we have, as is well known,
According to (4.1) Eg = E + 47ГР/3 ). On the other hand, if we take E in
TaM(k)=eik*M(k), (4.3) expressions (4.1) and (4.4) to mean the effective field
Eg, we obtain a relation between D' and Eg. The fre-
b y v i r t u e of w h i c h w e r e a c h t h e c o n c l u s i o n t h a t t h e quencies and wave functions of the mechanical excitons
m a t r i x e l e m e n t s M m n ( k ) , and consequently Fmn(k) must then be calculated in a manner different from that
differ from zero only if q m = q n —к + 2жЪ; the matrix in the preceding case (that is, without account of the
elements M n m ( k ) and F n m ( k ) differ from zero if polarization correction of the type 4тгР/3). It must be
qm = q n + к + 27rb, where b is an arbitrary whole- emphasized that by calculating the characteristics of
number reciprocal lattice vector. Taking the fore- the mechanical excitons with account of the polariza-
going into account and using the expression tion correction, we take into consideration by the same
token, to some extent, also a certain portion of the
long-wave field.* This circumstance, however, does
not lead to a non-analytic dependence of the mechani-
where V is the volume of the crystal, we obtain from cal-exciton frequencies on k.
(4.2)
Above the frequencies of the mechanical excitons,
ь ь k
/Г (M) = 2 < К ) E , (к + 2лЬ, со). (4.4) w m are assumed to be real and it is assumed that the
ь
denominators in (4.4a) do not vanish. In other words,
In t h e d e r i v a t i o n of (4.4) t h e e s s e n t i a l f a c t o r , of c o u r s e , no absorption was considered here (see Sec. 4c). By
a r e t h e foregoing s e l e c t i o n r u l e s for the m a t r i x e l e - virtue of the statements made in Sec. 2a, we shall take
m e n t s of t h e o p e r a t o r M ( k ' ) , b y v i r t u e of w h i c h , f o r into consideration in (4.4) only the term with b = 0. In
example, addition, inasmuch as we are not interested here in the
infrared region of the spectrum, we disregard the con-
6<Fn '0, tribution to the tensor o-jj by the direct interaction be-
tween the ions and the field of the electromagnetic wave
provided only [this means that in expression (4.4a) the mass m a is
assumed to be infinite for ions]. Under these assump-
k' = ± к + 2яЬ. tions, the tensor e( n ) (w,k) for the ground state (n = 0)
As a result we find that the tensor* is of a non-conducting crystal assumes the form

(Mi /45)

where No is the total concentration of the electrons.

We note that the second order pole at ш = 0 in the ex-
> pression for ejj(w,k) is only illusory, since by using
53
the sum rule it can be eliminated- (see '- -'). Since the
where к = к + 27rb, 6 q = 1, q = 0; 6 q = 0, q * 0. region of small w is of no interest to us here, we shall
Knowing the conductivity tensor, we can also deter- not eliminate the pole at w = 0.
mine the dielectric tensor, using the relation
*In this sense the difference between the real and mechanical
*From this relation it follows directly that the components of exciton consists precisely of an account or failure to account for
the tensor ст[" decrease with increasing |b| like Fourier com- the microscopic field E in the equations of motion (see also [12J,
ponents of a smooth function. Sec. 30).
700 V. M. A G R A N O V I C H a n d V. L. GINZBURG
In the ground state of a dielectric, the quasimomen- Accordingly, in place of (4.8a) we have in this case
tum is zero, since the wave function of this state is in-
variant under all crystal symmetry operations. On the B
, 0) ^ -<os(0). (4.8c)
other hand, we note that we are interested here in the
Among the other exciton states, it is convenient to
contribution made to the dielectric tensor of the crystal
separate the quadrupole exciton states, the wave func-
by the presence of exciton states only. These states,
tions of which transform when q = 0 like the products
as is well known, (see, for example, C5T.583) a r e char-
of the components of two polar vectors. Similar to
acterized by a single continuous quantum number—the
what was done in '-34-', it can be shown that for such
quasimomentum q, and also, generally speaking, by a
states, in the first approximation in k, we get
set of discrete quantum numbers s. Therefore
(4.9)
2mc
with [ see also (3)]
The difference between (4.9) and the analogous equa-
(4.6) tion in C343 i s that (4.9) was derived for the wave func-
If the vector q is chosen within the fundamental recip- tions of "mechanical" and not "Coulomb" excitons,
rocal lattice cell, then at small | к | we can readily the wave functions of which, generally speaking, are
verify that non-analytically dependent on the quasimomentum,
something not taken into consideration in ^ 34 ^. Taking
Mlo (k) = M\s., о (k) = Mlks, О (к) 8 q + k ,
the notation (3.33b) into account, we find that*
Ml0m(k) = MJ ; q s (k) = MU.(k)6 k _ 4 . (4>7)

Using (4.7) we find on the basis of (4.5) that

(01 ?-„ 10* r i m 10> A,

I, m=l
3
?{
;
<o-o>8(-k) co+o), (к) (4.8) —5SF 2
In the vicinity of the isolated s - t h absorption l i n e , the
I. m=l
change in the tensor ец(ш,к) with the frequency ш + {O\fim\Os)(Os\fH\O)}klkm.
is determined essentially by one of the resonant terms
in the dispersion formula (4.8). In this case we have If the frequency is ш « O J S ( 0 ) , the main contribution
approximately to the sum (4.8) is made by the s-th term, which is
proportional to [w— w^ 0 *]" 1 . Since the exciton term
Kw s (0) may be degenerate (s = s l t s 2 > . . . ) , we arrive
at the relation (3.43), which we have already used be-
-} , (4.8a) fore.
In connection with the foregoing calculation of
where elj^w) is a smooth function of w in the fre- ejj(ai,k), we make one remark which, incidentally,
quency interval under consideration. For those exci- pertains to the very introduction of this quantity, and
ton states (q, s ) , the wave functions of which t r a n s - could therefore be made even in Sec. 1. The tensor
form when q = 0 like the components of a polar vector, ejj(oi,k), by definition, establishes the connection be-
the matrix elements M J s ; 0 ( 0 ) , i = 1,2,3 (at least tween the harmonic fields D and E:
some of them) differ from zero. It is clear that only
(Tt) = Z)4 (кш) е«кг-ш|) = ei;- (со, к) Ei (r, t)
these exciton states (we previously called them dipole 4

exciton states) contribute to the dielectric tensor = ew((o, k)£ 4 (k, (4.10)
without account of spatial dispersion. Inasmuch as in
this case But if we consider waves in a medium, then by virtue
of the field equation w = w(k), and thus ш and к can-
not be regarded as independent variables. On the other
hand, if a; and к are related, then the tensor eij(w,k)
where D^ 0 ; 0 is the matrix element of the dipole mo- depends only on w and it is not clear what constitutes
ment operator of the unit cell of the crystal and N is the spatial dispersion. The answer frequently given to
the number of unit cells in the fundamental volume of this question is as follows. Equations (1.17) and (1.18)
the crystal, we have
*We note that in Chapt. 5 of M the tensor £у(<и, к) was incor-
rectly expanded in a series in powers of k, for no account was
(4.8b) taken of the dependence of the wave functions of the excited states
where is the volume of the unit cell of the crystal. of the crystal on the quasimomentum.
 CRYSTAL O P T I C S WITH A L L O W A N C E FOR S P A T I A L D I S P E R S I O N 701
are obtained only by starting from equations (1.1) or in the region of the lower excited states this exchange
(1.14) for the free field, equations which are valid if does not play any appreciable role. We introduce func-
there are no extraneous current and charge densities tions cpna which are normalized and antisymmetrized
j 0 and p 0 . On the other hand, in the presence of ex- over all the coordinates
traneous current and charge, the field E ( r , t) is de-
fined by the quantities jg(r, t) and p o ( r , t ) . In partic- tina<tia=E'}4>L (4.12)
ular, j 0 and p can be chosen such that the field has (here f = 0 corresponds to one state of the molecule:
the form in the case when the molecular term is degenerate, the
index f should be regarded as compound: f — (f, r ) ,
with some values of us and к which are not related at r = 1 , 2 , . . . , t, where t is the multiplicity of the de-
all (by way of a simple example we point to the obvi- generacy ).
ous possibility of producing a static electric field with In the Heitler-London approximation the wave func-
arbitrary к by means of a system of external charges). tion of the ground state of the crystal is
The tensor е^(ш,к) then defines D'(k, w) in t e r m s of (4.13)
E(k, ш) and is a function of the independent variables
ш and k. while for the excited state we have
b) Mechanical excitons and the tensor ejj(u),k) in
molecular crystals and in the case of the classical os- ф'=
naAno (4.14)
cillator model. To calculate the tensor eij(co,k) it is
obviously necessary to know the wave functions Ф where
т()
and the eigenvalues of the energy W m for mechanical (4.15)
&=4& П ф°
excitons. With this, in accordance with the statements
made in the Introduction and in Sec. 4a, the excited The energy of the ground state is E o = <Ф0, НФ„). Min
states, called mechanical excitons, correspond to a imizing H = (Ф*, НФ^) over the set {aj^,} subject to
solution of the problem without account of the action the additional condition
of the macroscopic (long-wave) field.
Let us develop here a theory for mechanical exci- nar
tons in the case of molecular crystals in the fixed-
molecule approximation. we find that the quantities a.^a satisfy the system of
We note that the Heitler-London approximation can equations
be used for this purpose only when the interaction be- S М^ г , г па а / ш г р-еЧ Г а = 0, е' = £ - Я 0 - Д „ (4.16)
tween molecules is sufficiently weak. Actually, we en-
counter a whole series of molecular crystals, where
the oscillator strengths are large, and the interaction where Af = Ef - EJj is the excitation energy of the iso-
between the molecules cannot be regarded as weak in lated molecule
many spectral regions (although this interaction does
not lead to violation of the neutrality of individual mole- ^ Ф т ц dr. (4.17)
cules, it does cause an intense mixing of the electron
configurations). The corresponding theory, without From considerations of translational symmetry it is
use of the Heitler-London approximation, was devel- clear that
oped in ^82-l. For the sake of simplicity, however, we nf — i
r
e (4.18)
use here the Heitler-London method, since the results
obtained below can be made more exact in a manner w h e r e N is the total n u m b e r of c e l l s in the fundamen-
similar to that used in ^ 62 ^. tal v o l u m e . Consequently we obtain in place of (4.16)
Let us assume that the elementary crystal cell con-
tains ст molecules. The Hamiltonian of the complete t T (k) af{ (k) - e ' 4 r (k) = 0, (4.19)
Coulomb problem is
w h e r e the p r i m e on the s u m m a t i o n sign denotes that
= the t e r m n = m , a = /3 h a s b e e n left out
" ?j ^*na~r "o" 2j ^nct, mBi (4.11)
£ ' (k) = -W (4.20)
where n and m are whole-number sites of the crystal
lattice; a, /3 = 1, 2,. . . , a; H n Q ! is the Hamiltonian of It follows from (4.19) that the values of the excitation
the molecule (n, a), and V n Q , j i n £ is the operator of energy a r e in the complete Coulomb problem the eigen-
the Coulomb interaction between the molecules na values of the Hermitian m a t r i x r j j ( k ) , while the
and m/3. In the construction of the wave functions of numbers
the ground and excited states of the crystal, we shall , k ) a = l , 2, .... a,
disregard the intermolecular electron exchange, since "a W> r = 1] 2, 3, ...,t
702 V. M. A G R A N O V I C H a n d V. L. GINZBURG

are the corresponding components of the eigenvectors. The coefficients Qjj' ( „ j for a s p e c i f i e d v a l u e of
To change over to the mechanical exciton problem,
let us separate from the interaction matrix f (к) the the wave vector к are determined only by the struc-
interaction due to the long-wave macroscopic field ture of the cell [their form is determined by formula
(we are essentially interested only in the region of (3.31) in the book [ 1 2 ] ] . What is essential is that the
small values of | к |; it is precisely for these values coefficients Qjj' ( *„) are when к = 0 analytic func-
of the wave vector that the Ewald procedure employed
tions of the wave vector, so that the entire non-analy-
leads to the separation of the long-wave field).
To this end, we represent the interaction operator ticity of the sum (4.27) is concentrated in (4.30). The
between molecules ( n a ) and (m/3) in the form of the coefficients Qii' ( ^ ) satisfy the following relations:
sum
V'na, шр = fta, mp + VlJa. ,„3, (4_21) (4.31)
where the index I denotes the operator of dipole- (4.32)
dipole interaction between molecules, while II denotes
the sum of the interaction operators of higher multi- (4.33)
polarity. Analogously, by virtue of (4.17) and (4.20)
we have Omitting in (4.29) the first term in the right half of the
equation, we see that part of the interaction matrix
M%.m = M"&m (4.22) Г ^ а ( к ) has in the mechanical exciton problem the
(4.23) following form

The dipole-dipole interaction operator is = - 2 <4-34>

i> i _ l ii'
* na. nip —r ]— r - [5
I na~ mpl Thus, the complete interaction matrix in the mechani-
r
cal exciton problem, unlike the interaction matrix in
X ] PnaPmp — |i— —r ТГ (Pna, na — 1
L
r
na mpl the problem where the Coulomb interaction is fully
(4.24)
accounted for, is an analytic function of the wave vec-
where P n a
i s t h e d i p o l e - m o m e n t o p e r a t o r of t h e (na) tor, with
molecule. T h e r e f o r e , u s i n g a l s o (4.17), w e find that
Гар(к) = Г%(к) + Г&(к). (4.35)
(4.25)
It is clear that in the problem of the mechanical exci-
where ton the relation (4.18) also holds true, but the ampli-
tudes a ^ ( k ) satisfy, unlike (4.19), the system of
(4.26)
equations
S Ггр'аГ/ (k) 1%* (k) - ? Zf (k) = 0. (4.36)
P£a
For each к the operator Г г ( к ) has a t eigenvalues
r
mp 3<"fo _ Л
* { r f - (4.27) eX /* = 1 , 2 , . . . , a t . The corresponding orthonormal-
f гр ,Гпа— Гщр)|-
'mp
ized eigenvectors will be denoted by {a A*f}. It is
F r o m the f o r m of (4.27) it i s c l e a r that E g ^ d O is
the e l e c t r i c field at the point r = r n a , produced by the clear that in the approximation considered here the
d i p o l e s s i t u a t e d i n t h e l a t t i c e s i t e s o f t h e t y p e /3, w i t h discrete quantum number is s = (f,/j). In crystals
.the m a g n i t u d e o f t h e d i p o l e s v a r y i n g f r o m s i t e t o site containing one molecule per unit cell, the f-th non-
as degenerate molecular term corresponds to only one
exciton band. In this case
P(mP) = Pp''°e~ikrmP. (4.28)

I n a s m u c h a s the o p e r a t i o n of s e p a r a t i o n of the macro-

s c o p i c p a r t of the field f r o m the total field (4.27) is а!=1. (4.37)
d e s c r i b e d in detail in '- 1 2 -', S e c . 3 0 , w e give here only
the final result: In more complicated crystals, containing for example
two molecules per unit cell, the values of e^, ju = 1,2,
\ (к)], =
№ (Д) (4.29) are determined as roots of the equation (the molecu-
lar term is assumed nondegenerate)
where
r'f
j
rf
E jj i s the a m p l i t u d e of the m a c r o s c o p i c part
(PJ, (k) -7) (f '„ (k) -1') -fjfa = 0. (4.38)
i k r
of the field e" :
Using the relations (4.25), (4.29), and (4.30), and also
4Я
taking into account the analyticity of the coefficients
 CRYSTAL O P T I C S WITH A L L O W A N C E FOR SPATIAL D I S P E R S I O N 703
Along with this, (4.40) is replaced by the following
Qjy ( a Л as к — 0, we can readily verify that
relations
pf /Л \ _ pf /П\ «rhoVOQ С P f / b- \ I -^ p f / Ь- \ |
11 CL 11 'K ~\J *>£. |Л ^U

It i s precisely by virtue of t h i s that the molecular

term f corresponds when к = 0 to the two mechani-
cal exciton states, for which
e{ (0) = f^ (0) + fJ, (0), ~a[ (0) = _ «J (0) = - ^ , where P^j is the j-th component of the r-th dipole
moment, located at a type-а site. We assume, for
example, that the Bravais lattice of the crystal is
cubic and that there is one isotropic molecule per
Knowledge of the wave functions of the mechanical ex-
unit cell. In such a molecule, the excited states to
citons makes it possible to represent all the quantities
which transitions are allowed in the dipole approxima-
in the expression for the dielectric tensor directly in
tion are triply degenerate. We choose the correspond-
terms of the characteristics of the individual mole-
ing three wave functions such that the vectors P ,
cules. Using (4.13) and (4.14), and also (4.2), we can
r = 1, 2, 3, are directed along the cube axes. In this
readily show that
case, obviously,

=
But then

= 1- [«o£ ( - k)]* ( - k)*a!U»( - k) pf

(4.40)
= ~[<ol(- k)]* ( - к) а'Л* ( - к) Р).
where v is the number of the electron belonging to the If we neglect the anisotropy of the effective mass of
molecule n a and pv is the radius vector of this elec- mechanical excitons, then accurate to t e r m s ~ k 2
tron relative to the lattice site r n a . As regards the
frequencies o> s (k) contained in (4.8), these frequen-
cies, in accord with (4.16), are determined in the In this approximation, for the model under considera-
Heitler-London approximation by the relation tion, the dielectric tensor
4 k ) ] . (4.41) 8
i,-K k) = 6tye(<B, k)

The molecular crystal model considered above includes

as a limiting case the classical model of a crystal with
point dipoles at its sites.
In order to verify the correctness of the foregoing reduces to a scalar.* A similar expression for the
remark, we must recognize that the crystal has along dielectric tensor, but with wf = const, follows also
with the lattice constant a also a whole set of param- from E15^, where an analogous model was used but
eters ai, equal to the effective dimensions of the atoms spatial dispersion was disregarded. In our model the
and molecules in different states. The values of aj de- refractive index for the transverse waves is obviously
termine the matrix elements for different multipole n = [e(w, шп/с)] 1 ' 2 . The normal waves in any direc-
(dipole, quadrupole, etc) transitions. In place of the tion are either longitudinal or transverse. The fre-
parameters a and aj we can therefore use a and the quencies of the transverse waves, disregarding r e -
values of certain effective multipole moments (more tardation, that is, the frequencies of the "polariza-
accurately, their matrix elements), assuming these tion waves" (see Introduction and Sec. 1), correspond
multipoles to be point like and located at the lattice to the poles of e(a),k), whereas at the frequencies of
sites. In the vicinity of the intense dipole lines we can the longitudinal waves e(oi,k) vanishes.
confine ourselves only to the dipole moments, so that As was already indicated in Sec. l c , to calculate
we arrive at a model of a crystal consisting of point- the refractive indices of light one can also use the
like dipoles. With this, the only interaction present is
the dipole-dipole interaction, so that Г ^ ( к ) = 0 and
by virtue of (4.35) *In the derivation of this formula it must be taken into account
that
704 V. M. A G R A N O V I C H a n d V. L. GINZBURG
tensor* ej^ijf w, k), with the aid of which the t r a n s - scopic t h e o r y ^ these difficulties do not arise. Within
verse induction vector D' is expressed in terms of the framework of this theory the dispersion of the elec-
the transverse part of the electric field intensity Ej_.1 tromagnetic waves in crystals can be developed with
The tensor ejnfcu, k) can be obtained on the basis any degree of accuracy.
of (1.66) and the expression for the tensor ец(ш,к). In conclusion we present the general formula for
However, it is simpler to find an expression for this the refractive index of light in a crystal ^ , which
tensor by using as the system of unperturbed functions could be obtained by substituting in (1.66) an expres-
the functions of the complete Coulomb problem (with- sion for ец(ш,к) and then representing the tensor
out account of retardation), and by using for the per- e
l,ij( w > k) in the form of an expansion in terms of the
turbation the transverse part of the electric field. In poles. Accurate to small quantities proportional to
this case, in complete analogy with the tensor е^(ш,к), ( a / \ ) 2 , we have in crystals with an inversion center
the tensor ex j ij(o), k) is obtained directly as an expan-
sion in terms of the poles. It must be borne in mind
here that both in the derivation presented here for
£ij( w, k), and in the derivation of the expression for 2 2
§/•„ (s) (cos ф! (s, k) —cos ф 8 (s, q)) 1
s
c
l ii ( s e e '"25"') a certain transverse part of the effec- a
a> —0,1 (k) -I
tive field is disregarded (see E12^, Sec. 44). As al-
ready indicated in Sec. 4a, the difference between the ri alF, (s) cos Ф! (s, k) cos ф а (s, k)
effective and the macroscopic field should be taken into (4.42)
consideration in the mechanical exciton problem. The
Ewald method, which we have used earlier, enables us w h e r e wjj i s t h e s q u a r e of t h e p l a s m a f r e q u e n c y ,
to separate only the longitudinal part of this field. As fls(k) = E s ( k ) / f i , a n d E s ( k ) i s t h e e n e r g y of t h e
to the transverse part of the effective field, it is con- Coulomb exciton* in the s - t h band with c o m p l e t e a c -
nected with the retarded interaction in a bounded r e - c o u n t of t h e C o u l o m b i n t e r a c t i o n , <p(s, s ) i s t h e a n g l e
gion. So long as the time necessary to cover this dis- between the vector в and the dipole moment vector
tance with the velocity of light с is small compared Do;os of the transition from the ground state (0) into
with the period of the oscillations under consideration, the state with the exciton (0, s ) ; furthermore <p}(s,k)
the transverse part of the internal field can be neg- with i = 1,2 is the angle between the vector D o > o s and
lected in nonmagnetic crystals. For distances on the the two unit vectors perpendicular to k. In (4.42) spa-
order of 10~8—10~7 cm, the required time is approxi- tial dispersion is taken into account only in the reso-
mately 10~17 sec, whereas the oscillation period in the nant denominators, which is sufficient in the analysis
visible part of the spectrum is of the order of 10~ 15 of spatial-dispersion effects in the vicinity of the di-
sec. Thus, it is not clear a priori whether the contri- pole exciton absorption bands. Formula (4.42) enables
bution of the transverse part of the internal field can us to analyze the dependence of n 2 on the direction of
be neglected in a theory that takes into account spatial light propagation in the crystal and can be used to de-
dispersion the magnitude of which is also connected termine the direction of the transition vector DO;os in
with the small parameter а/Л » 10~3. This entire an experimental study of the function n 2 (w, s ) . Thus,
problem is not fully understood and apparently needs for example, in the vicinity of an isolated exciton tran-
special analysis. We note, however, that in the micro- sition we have

*It follows from (1.58) that the determinant of the matrix тщ nl(a>, s) = e .
((l)
(4.42a)
vanishes, and consequently [see (1.57)1 the determinant of the
matrix ej/y also vanishes. It is easy to see, for example, that in The equations (4.42a) can, of course, be established
a coordinate system where the z axis i s directed along the vector directly on the basis of (1.33) and (4.8a, c) provided
k, all the components of the tensor ejf'ij for which at least one of we neglect the nondiagonal elements of the tensor
of the indices, i or j , corresponds to the z axis, vanish. This
е^(ш) and assume that efj'fw) = e < 0 ) (w)6jj. Indeed,
tensor can therefore be regarded as a tensor in two-dimensional
space, that is, a s е^а/З, <*,/3 = x,у. The tensor e^.i/S' introduced
in this case the approximate expression for the dielec-
in (1.61), is the inverse of the tensor е^'а/з in the same two-dimen- tric tensor has the symmetry of this tensor in a uni-
sional space: £j_,a/3e^Sa' = S a a ' , so that the indices i and j in axial crystal, with
(1.61) can assume only the values x and у (this was unfortunately not
emphasized in Part I). It i s clear that in three-dimensional space e. = e w — - ».(«)•
the tensor e i j ч has a structure similar to that possessed by the
tensor e^.Sj. and is not the reciprocal of the tensor ej^ij simply Hence, bearing in mind (1.31), we find directly that for
because the latter does not exist at all, since || e j ' j j || = 0.
t i t is precisely for this reason that the tensor e L i y i s not suit- *A Coulomb exciton is one corresponding to the exact solution
able for an analysis of the energy losses of a charge moving in a of the Coulomb problem. In accordance with the terminology pre-
medium, since the electric field of the charge acting on the crystal viously exmployed. Coulomb excitons are longitudinal and "ficti-
does not satisfy the condition d i v D ' = 0. tious" longitudinal waves, and also "polarization waves."
 CRYSTAL O P T I C S WITH A L L O W A N C E FOR S P A T I A L D I S P E R S I O N 705
a wave polarized perpendicular to the vector D o ; O s we in connection with the fact that when the lattice vibra-
have n 2 = n 2 = e <0> , whereas for a wave polarized in the tions are taken into account the state of the crystal is
plane formed by the vectors s and D^?JS characterized not only by the state of the electrons,
but also by the state of the nuclear motion.
'-ш,(0))
The need to take into account not only electron mo-
tion but also the motion of the crystal nuclei, generally
A n a l o g o u s l y , w e c a n o b t a i n f r o m (4.42) e x p r e s s i o n s
speaking, complicates appreciably the problem of find-
for n 2 (ш, s ) in such spectral regions where not one but
ing the stationary states of the unperturbed problem
several exciton transitions appear. It should be borne
(that is, the problem in which retardation and the
in mind here that the dependence of the frequencies
macroscopic electric field are disregarded). By far
S2s(q) on q for small | q | can be readily established
not all aspects of the theory of absorption of light in
provided the tensor eij(w.q) is known. In the simplest
crystals have been studied in sufficient detail, and the
case, when it is sufficient to consider one of the m e -
theory needs to be discussed. However, in the present
chanical exciton bands which are not degenerate when
article we are unable to do so. We therefore present
q = 0 in the investigated spectral region, we find by
only a few remarks and assume that the stationary
using (1.39) and (4.8a, c) that*
states of the unperturbed problem are known, at least
2
Q (k) = wij(k) + /a(k)cos <p, 2
(4.43) approximately.* Under this assumption, the calculation
of the current induced by the external electromagnetic
where <p is the angle between the vectors D O ; o s ( k ) field enables us to obtain an expression for the dielec-
and s, while f s ( k ) = (87r/Ah)|D | 2 aj s (k). If not one tric tensor, similar to (4.5). The corresponding expres-
but several mechanical exciton bands must be taken sion must also be averaged over the possible initial
into consideration in the expression for ejj, the angu- states of the unperturbed system, that is, in practice
lar dependence of u(s) becomes more complicated. over the initial phonon distribution function (for a
In some cases it is interesting to know not only the particular model of a molecular crystal see the de-
angular dependence of fi(k) when к = 0, but also the tailed discussion in E 8 ']).
form of the Coulomb exciton band for small | к |. In
The picture is particularly simple in the case of
the case of dipole exciton bands it is necessary to use
weak coupling between the excitons and the photons,
the expansion
when in the initial (zeroth) approximation the t r a n s -
ei/ К q) = e,7 (to, 0) + Р1 (4.44) verse electromagnetic field can be regarded to be the
same as in vacuum. This means that the normal waves
and then determine the frequencies fi^(q) from equa- of the unperturbed problem a r e photons in vacuum, with
tion (1.60), which is obtained by equating to zero the a dispersion lawt w = кдс = 2тгс/Л.о, Coulomb excitons,
determinant of the tensor т)е~1(ш,к)т1Л and phonons.t Under the influence of the perturbing
с) Absorption mechanism and calculations. The electromagnetic field of the photon, and also owing to
damping of electromagnetic waves in crystals is con- the coupling between the Coulomb excitons and the
nected primarily with the possibility of irreversible phonons, transitions occur between the states of the
transfer of energy from the waves under consideration unperturbed problem, as a result of which the energy,
to other degrees of freedom. In non-metals, the ab- say of the photon, can be transformed into lattice vibra-
sorption of electromagnetic waves in the visible portion tion energy, etc.
of the spectrum and in the ultraviolet is due essentially
to the conversion of the electron excitation energy into An analysis of the expressions for the probability of
vibration energy of the crystal lattice nuclei (see, for photon absorption or else for the anti-Hermitian part
example, ^i3^). The possibility of this process causes of the tensor ejj(oj, k) shows that in the region of the
the states of the photon in the medium (the real exci- exciton absorption the damping of the light waves
ton) to become quasistationary, and the tensor ец(ш,к) (photons) is connected principally with processes
to become non-Hermitian even for real w and k. The corresponding to the transformation of the photon with
dielectric tensor in the presence of absorption has been
calculated for several particular crystal models, with *A discussion of the problems related with this can be found
and without account of spatial dispersion, in several in [«.«•»].
papers (see, for example, Е«П ). The procedure for tin order to take into account the influence of other resonances,
we must write here and below Л„/п00 in lieu of A,,, where the re-
determining this tensor is quite analogous to that de-
fractive index n 0 0 is determined in the considered region of fre-
veloped in Sec. 4a, and the main difference arises only quencies by the contribution of the resonances other than the one
iConsidered. In cubic crystals n 0 0 = \J £00 [see (3.12)].
*One must not forget that as к -» 0 the values of &>s(k) and t Phonons, like Coulomb excitons, are among the solutions of
f s (k) are independent of the direction s. the Coulomb problem. We designate as phonons those quasiparticles,
tWe note that this procedure of determining the dependence the consideration of which is essentially connected with an account
O(s), and also the form of t he exciton band at small values of the of the motion of the nuclei (atoms). The arbitrariness in the division
wave vector, holds true regardless of whether the exciton is of of the excitations into phonons and Coulomb excitons is of no im-
the Frenkel type or of the Mott type. portance whatever to what follows.
706 V. M. A G R A N O V I C H a n d V. L. GINZBURG
energy fiw = К2тс/Л0 into a Coulomb exciton with energy propagating in the crystal has the same frequency,
E s ( k ) and one or several phonons. that is,
When we can confine ourselves in the weak exciton-
phonon interaction operator to terms that are linear in
the displacements of the nuclei from the equilibrium
positions, the basic processes are those in which only In the account of the interaction between a real ex-
one phonon with energy Kfij(q) and momentum q par- citon and phonons, the exciton is split into another real
ticipates: exciton and phonons, that is, an apparent combination
scattering of real excitons occurs С14>8та]_ if only one
phonon is emitted, then
%a> = ha>j(k)=:tia>j(k') + bQl(q), %k = ftk' + ftq. (4.47)
where \ 0 * s the wavelength of the monochromatic light
incident on the crystal from the outside, and accurate The dispersion law of the real excitons—the depend-
to a whole-number reciprocal-lattice vector ence oij(k) —is much different for exciton bands cor-
responding to large oscillator strengths from the dis-
ko = k ± q . (4.46)
persion law for Coulomb or mechanical excitons (see,
The plus and minus signs in (4.45) and (4.46) c o r r e - for example, Figs. 1 and 2). This circumstance leads,
spond to the emission or absorption of a phonon. in particular, to a situation wherein relation (4.47) can
Relation (4.45) determines directly that region of also be satisfied in the region w < E s (0)/E, as a r e -
frequencies ш = 27Гс/А.о, which corresponds in our sult of which long-wave absorption appears in the fre-
approximation to the exciton absorption line, but this quency region w < E s (0)/fi. For large oscillator
relation enables us to study the shape of the exciton strengths (in those cases when the point k = 0 c o r r e -
absorption band only approximately. Indeed, let us sponds to a minimum of the exciton band), this long-
assume for the sake of simplicity that the crystal is wave absorption completely determines the shape of
at the temperature of absolute zero, so that the photon the long-wave fall-off of the exciton absorption band
decay can occur only with emission of a phonon [the [63,643.
plus sign in (4.45)]. We assume, in addition, that in Thus, we can find in this manner the absorption
the exciton band under consideration the minimum en- coefficient к in the case of sufficiently weak absorp-
ergy corresponds to a quasimomentum k = 0. In this tion of normal waves (real excitons) in crystals.
case, as can be readily seen, relation (4.45) can be Calculations of ejj(w,k) in the analogous approx-
satisfied only for frequencies ш > E s (0)/K. This imation, that is, with a more accurate account of the
means that the frequency E s (0)/K is in this approxi- interaction between the real excitons and the phonons,
mation the long-wave edge of the exciton absorption particularly in the region of frequencies smaller than
band. But such a conclusion, as applied to not very the limiting frequency of the mechanical or Coulomb
weak dipole absorption lines, turns out to be incorrect, excitons, have not been carried out, as far as we know.
inasmuch as in the vicinity of these lines the real ex- The calculation of £ij(w, k) in this and in the higher
citon differs appreciably both from the photon and from approximations can be made by using the temperature
the Coulomb exciton. The same shortcoming, namely Green's functions [88 ' 89 ^.
the incorrect determination of the course of the absorp-
tion near the long-wave edge, is inherent in the calcu- CONCLUDING REMARKS
lations in which one first determines the tensor
ejj(w,k) using for w m [see (4.5)] the frequencies of In the study of excitons by optical means, the p r e -
vailing practice so far was to measure only the ab-
the mechanical or Coulomb excitons (see, in particu-
lar, [87,25]). sorption and forego the analysis of the absorption line
shape. This procedure is natural so long as one deals
It is quite clear how to approach more correctly the merely with the determination of more or less sharp
question of the absorption of light waves in crystals. excited levels in crystals, the clarification of the cor-
We first neglect the interaction with the phonons and responding series laws, etc. The situation here (par-
consider in this approximation the real excitons (pho- ticularly if the crystal can be assumed to be optically
tons in the medium). This means that we are consid- isotropic) is analogous to a certain degree to that p r e -
ering normal electromagnetic waves in a crystal, neg- vailing in the determination of atomic levels of gases.
lecting absorption* but with full allowance for all the It is quite obvious, however, that in a detailed investi-
remaining electromagnetic interaction. The frequency gation of the energy spectrum of crystals in the region
of the real exciton wj(k) is in this case real. Of of optical frequencies, a more general formulation of
course, in the case when light with frequency ш is the problem is necessary. On the one hand, the ab-
incident on the crystal, the real exciton produced and sorption line shape must be analyzed; on the other
hand, in addition to absorption, it is possible and nec-
*The absorption is thus, for the sake of simplicity, assumed to essary to study the dispersion, that is, to measure the
be solely connected with the energy transfer to the phonons. refractive index. One cannot be restricted here to an
 C R Y S T A L O P T I C S WITH A L L O W A N C E F O R S P A T I A L DISPERSION 707
optically isotropic medium, all the more since even will not only not contradict the macroscopic theory
cubic crystals are optically anisotropic when spatial and will not fail to contain all its consequences as ap-
dispersion is taken into account. In other words, a plied, to this model, but will make it possible to spe-
study of the excitons is inseparably connected both cify several relationships, such as the frequency de-
with classical crystal optics and with crystal optics pendence of €jj(aj,k). It remains still unclear, how-
in which spatial dispersion is included. Some work ever, what depends on the given model and what should
has already been done in this direction, both experi- be obtained for any model (that is, what is entirely in-
mental and theoretical, but much is still left to do. In dependent of the model). It is quite obvious that the
particular, attention should be paid to the question of use of a model or an approximation, and a comparison
new waves in gyrotropic and non-gyrotropic crystals, of the results of the corresponding calculations with
to the dispersion and absorption near the quadrupole experiment is valuable only if one talks of conse-
absorption lines in crystals, to the influence of exter- quences or aspects that are specific for the given
nal electrical and magnetic fields on the optical prop- model, and are not common and independent of the
erties, and also to the influence of stresses and strains. choice of the model. Thus, the use of macroscopic
For a correct analysis of the experimental data and theory generally speaking is not only advantageous
to extract from them definite information on crystal but even necessary for a determination whether a
properties, it is necessary to take into account and to certain model or approximation is of value. In addi-
use the formulas and results of crystal optics. As far tion, under such a formulation of the problem we get
as we know, these results had not yet been expounded rid of the need of carrying out microscopic calcula-
with sufficient detail or from a unified point of view tions for quantities which are essentially derivatives
with account of spatial dispersion. This is precisely of more fundamental quantities [for example, it is
the purpose of the present article, which contains rather advantageous to calculate 6jj(w,k) and not n;(oi,s)].
extensive material. At the same time, it cannot be The foregoing, of course, is not specific to the macro-
stated that all the problems of interest have already scopic and microscopic theory of optical crystal prop-
been investigated within the framework of crystal op- erties, but has a general (and well known) character.
tics with account of spatial dispersion. Suffice it to But, as was already indicated above, in the theory of
say that even in classical crystal optics, which has excitons for various reasons no harmonious combina-
been developing for many decades, new aspects are tion of the microscopic and macroscopic approaches
still being encountered and some cases have not been has yet been reached. The purpose of the present
sufficiently studied (we mention, for example, singu- article will be attained to a considerable degree if
lar optical axes). The number of crystal optics prob- this important and simple aspect is duly recognized.
lems which could be solved with account of spatial dis- As a result of the extensive use of crystal optics with
persion is very large. However, the solution of many account of spatial dispersion and of a correct combi-
such problems will be far from justified from the point nation of crystal optics with microscopic exciton the-
of view of real requirements, which are determined by ory, further development of research in the field of
the experimental capabilities and the value of the par- optical and certain other properties of crystals will
ticular information to the theory of crystals. There- become much easier and will proceed at a much faster
fore, it seems to us that further development of the rate.
theory (we refer now to calculations analogous to those
of Sec. 3) must be primarily closely linked with an
analysis of the experimental data and the experimental Note added in proof: Following the first part of the present
capabilities. Incidentally, even without a direct connec- article (cited below as I), a paper was published in UFN by S. I.
PekarL90], devoted to the same group of problems. A comparison
tion with experiment one can hardly consider an inves- of that paper with ours may give rise to some misunderstanding
tigation of the influence of weak spatial dispersion on on the part of the reader, in connection with the different estimate
the propagation of electromagnetic waves in crystals and different interpretation of several aspects. We therefore find
of different classes, near the optical axes, for crystal it necessary to point out that in the articles of the authorst5'"'24]
plates, etc superfluous. (a bibliography is given in I), many critical remarks regarding the
articles by Pekart7'"'25] were already made both explicitly and
An important point, which we have already empha- implicitly. Inasmuch as we did not encounter in the literature any
sized in the article and mention here once more, is objections to these remarks, we find it unnecessary to repeat the
that the neglect or differentiation of the macroscopic criticism and indicate specifically the sources of the incorrect or
(phenomenological) crystal optics, which uses the inaccurate statements. The gist of the matter has been discussed
tensor ejj(o!,k), and the microscopic theory, is ut- in our article with sufficient detail, and therefore we hope that the
readers will themselves be able to estimate the character of the
terly unfounded. At first glance it may seem that the different works (including Pekar's and ours), particularly if they
microscopic calculations [for example, the calculation take the trouble to read Pekar's articlest7'"], and not only the re-
of ejj(aj,k) or of n ( w , s ) for some particular model] view articlesl14'90], in which opinions and results obtained both by
immediately yield more than the macroscopic theory us and by other authors are to a certain extent included.
(in this case, crystal optics). And indeed, the results Here we are able only to make the following specific remarks
of a correct calculation for a given reasonable model concerning the articleL90J.
708 V. M. A G R A N O V I C H a n d V. L. G I N Z B U R G
1. Pekar's statement that in a magnetoactive plasma with ac- * K . H. H e l l w e g e , Z . P h y s i k 129, 626 (1951).
count of spatial dispersion no new waves occur is in error. In the 5
V . L. G i n z b u r g , J E T P 34, 1593 (1958), Soviet P h y s .
2
absence of spatial dispersion the equation for n in an anisotropic J E T P 7, 1 0 9 6 ( 1 9 5 8 ) ; s e e a l s o P r o c . I n t e r n . Conf. S e m i -
medium (in particular, in a magnetoactive plasma) is quadratic and c o n d u c t o r P h y s i c s , P r a g u e , 1961, p . 394.
h a s two roots п\ and a\, corresponding to the ordinary and extra- 6
E . F . G r o s s a n d A . A . K a p l y a n s k i i , DAN SSSR 132,
ordinary waves. If we disregard the direction along the magnetic
field or the transition to an isotropic plasma in t h e absence of the 98 (1960) a n d 1 3 9 , 7 5 ( 1 9 6 1 ) , S o v i e t P h y s . D o k l a d y 5,
field, when the longitudinal (plasma) wave occurs, no third wave 5 3 0 ( 1 9 6 0 ) , a n d 6, 5 9 2 ( 1 9 6 2 ) .
7
exists at all. On the other hand, in the presence of spatial dis- S . I. P e k a r , J E T P 3 3 , 1022 (1957) a n d 3 4 , 1176
2
persion the equation for n becomes one of the third degree and a ( 1 9 5 8 ) , S o v i e t P h y s . J E T P 6, 7 8 5 (1958) a n d 7, 8 1 3
new wave appears due to the new finite root n, (for more details (1958).
see, for exampleW, Sec. 12). Incidentally, the longitudinal wave 8
G i n z b u r g , R u k h a d z e , a n d Silin, F T T 3, 1835 (1961),
in an isotropic medium i s also a new wave, for if no spatial dis-
S o v i e t P h y s . S o l i d S t a t e 3 , 1337 ( 1 9 6 2 ) ; J o u r n a l of
persion i s taken into account we have not a root but a single dis-
P h y s i c s a n d C h e m i s t r y of S o l i d s , 2 8 9 0 ( 1 9 6 1 ) .
crete oscillation. In general it should be noted that the appearance 9
of new waves—new roots of the dispersion equation —is an obvious Y u . L . K l i m o n t o v i c h a n d V. P . S i l i n , U F N 7 0 , 2 4 7
and well-known consequence of the inclusion of spatial dispersion. (1960), Soviet P h y s . U s p e k h i 3, 84 (1960).
10
2. In an article by one of the authorsM, as in Pekar's first A . A . R u k h a d z e a n d V. P . S i l i n , U F N 7 4 , 2 2 3 ( 1 9 6 1 ) ,
paperM, only dipole lines were considered. In this c a s e account Soviet P h y s . Uspekhi 4, 459 (1961); E l e k t r o m a g n i t n y e
of spatial dispersion in crystal optics i s included in its entirety svoistva plazmy i plazmopodobnykh sred (Electromag-
by expanding the tensor е ч ( ш , к) ot £ ^ ' ( a , k ) in powers of k^, n e t i c P r o p e r t i e s of P l a s m a a n d P l a s m a - l i k e M e d i a ) ,
retaining only the first essential terms of t h e series. However, if
Moscow, Gosatomizdat, 1961.
we consider quadrupole or higher multiple transitions, it may be- 11
come necessary to represent els or вц1 as a ratio of two polyno- V . L . G i n z b u r g , U F N 6 9 , 537 (1959), S o v i e t P h y s .
mials in kj. This was mentioned in ['] and in greater detail in I. U s p e k h i 2, 8 7 4 ( 1 9 6 0 ) .
12
It is important that from our point of view the functions €ц(й), к) M . B o r n a n d K u n H u a n g , D y n a m i c a l T h e o r y of
and e " 1 (<u, k) can be regarded in crystal optics as having no essen- C r y s t a l L a t t i c e s , C l a r e n d o n P r e s s , Oxford, 1954.
tial singularities or branch points. Yet in the articles by PekarL 7 ' M J 13
H . H a k e n , F o r t s c h r . P h y s i k 6, 2 7 1 ( 1 9 5 8 ) ; U F N 6 8 ,
and others the "mechanical e x c i t o n s " chosen were exact solutions 565 (1959).
of the Coulomb problem ("Coulomb excitons") which in some cases 1 4
S . I. P e k a r , J E T P 3 8 , 1786 ( 1 9 6 0 ) , S o v i e t P h y s .
(for " f i c t i t i o u s " longitudinal waves) have singularities at к = 0.
An impression remains therefore that the functions eLi and Ef]1 can J E T P 11, 1286 (1960).
U a
also have essential singularities at к = 0. I . I. H o p f i e l d , P h y s . R e v . 1 1 2 , 1 5 5 5 ( 1 9 5 8 ) .
is U. F a n o , P h y s . R e v . 1 1 8 , 4 5 1 ( 1 9 6 0 ) .
The foregoing will permit the reader to judge the extent to which
16
the remark made at the end of the p a p e r M , concerning the "contra- K . B. T o l p y g o , J E T P 20, 497 (1950).
17
diction" between the deductions of M and the data of [76J concern- К. В. Tolpygo, UFN 74, 269 (1961), Soviet Phys.
ing the new wave near the quadrupole absorption line is unfounded. Uspekhi 4, 485 (1961).
The questions of absorption or of the new wave near the quadrupole 18
M. A. Leontovich, JETP 40, 907 (1961), Soviet
line are discussed in Sees. 3c and 3f of the present article. Phys. JETP 13, 634 (1961).
3. In Sec. 3f of the present article we pointed out that absorp- 19
B. N. Gershman and V. L. Ginzburg, Izv. vuzov
tion near the dipole line in anthracene, investigated in L74J, should
lead to attenuation of the new wave by at least a factor of 10 6 , even (Radiofizika) 5, (1962).
19a
in a film with thickness d = 0.1 ц. We therefore cannot s e e how the V. L. Ginzburg, ibid. 5, (1962).
20
oscillations observed in L741 can be related with the occurrence of a V. L. Ginzburg, ibid. 4, 74 (1961).
21
new wave, unless a radical change takes place in the parameters S . Boguslawski, Ann. Physik 44, 1077 (1914).
22
(we know of no grounds for this). In the a r t i c l e M , however, t h i s G . Szivessy, Hand. d. Phys. 20, 635 (1928).
important item was circumvented by complete silence in the discus- 23
N. G. van Kampen, Math. Rev. 20, 1227 (1959).
sion of the experiments of L"J. u
V. M. Agranovich and A. A. Rukhadze, JETP 35,
Recently S. I. Pekar published still another review a r t i c l e M
982 (1958), Soviet Phys. JETP 8, 685 (1959).
devoted to the same problems. We cannot dwell on it here. One of 25
us hopes to demonstrate in an article in " F i z i k a tverdogo t e l a "
S. I. Pekar, JETP 36, 451 (1959), Soviet Phys.
(Solid State Physics) the incorrectness of many statements made .„ JETP 9, 314 (1959).
26
in ["I. J. Ney, Physical Properties of Crystals, Clarendon
P r e s s , Oxford, 1957.
27
F. I. Fedorov, Opt. i spektr. 6, 85, 377 (1959).
B i b l i o g r a p h y , II 28
G. N. Ramachandran and S. Ramaseshan, Hand. d.
1
L. D. L a n d a u a n d E. M. L i f s h i t z , E l e k t r o d i n a m i k a Phys. 25/1, 1 (1961).
29
s p l o s h n y k h s r e d ( E l e c t r o d y n a m i c s of C o n t i n u o u s M e d i a ) , L. D. Landau and E. M. Lifshitz, Mekhanika
F i z m a t g i z , 1957. sploshnykh sred (Mechanics of Continuous Media)
2
V. L . G i n z b u r g , R a s p r o s t r a n e n i e e l e k t r o m a g n i t n y k h Part II, Sec. 10, Gostekhizdat, 1953.
30
v o l n v p l a z m e ( P r o p a g a t i o n of E l e c t r o m a g n e t i c W a v e s L. N. Ovander, FTT 3, 2394 (1961); 4, 157 (1962);
in P l a s m a ) , F i z m a t g i z , 1960. 4, 294 (1962). Soviet Phys. Solid State 3, 1737 (1962);
3
H . A. L o r e n t z , C o l l e c t e d P a p e r s , Vol. 2, 1936, 4, 112 (1962); 4, 212 (1962).
p . 7 9 ; V o l . 3, 1 9 3 6 , p . 3 1 4 .
 CRYSTAL O P T I C S WITH ALLOWANCE FOR SPATIAL D I S P E R S I O N 709
31 54
A. V. Shubnikov, Principles of Optical Crystallog- S. I. Pekar, JETP 35, 522 (1958), Soviet Phys.
raphy, Consultants Bureau, New York, 1960. JETP 8, 360 (1959).
32 56
S. Pancharatnam, Proc. Indian. Acad. Sci. 48, 227 M. H. Cohen and F. Keffer, Phys. Rev. 99, 1128
(1958). (1955).
32a 56
A. P. Khapalyuk, Kristallografiya, in p r e s s ; Soviet B. A. Nyboer and F. W. De Wettle, Physica 24, 422
Phys. Crystallography, in press. (1958).
33 57
1 . V. Obreimov and A. F. Prikhot'ko, Sbornik F. Seitz, The Modern Theory of Solids, McGraw-
Pamyati S. I. Vavilova (Memorial Volume to S. I. Hill, New York, 1940.
58
Vavilov), Moscow, 1952, p. 197. L. D. Landau and E. M. Lifshitz, Statisticheskaya
34
W. R. Heller and A. Marcus, Phys. Rev. 84, 809 fizika (Statistical Physics), Gostekhizdat, 1951.
59
(1951). V. M. Agranovich, JETP 37, 430 (1959), Soviet
35
E. F. Gross and A. A. Kaplyanskii, FTT 2, 379 Phys. JETP 10, 307 (1960).
60
(1960), Soviet Phys. Solid State 2, 353 (1960). J. I. Frenkel, Phys. Rev. 37, 17, 1276 (1931).
61
36
V. I. Cherepanov and V. S. Galishev, FTT 3, 1085 V. M. Agranovich and A. A. Rukhadze, JETP
(1961), Soviet Phys. Solid State 3, 790 (1961). 35, 1171 (1958), Soviet Phys. JETP 8, 819 (1959).
36a 62
A . G. Zhilich, V. I. Cherepanov, and Yu. A. Karga- V. M. Agranovich, FTT 3, 811 (1961), Soviet Phys.
polov, FTT 3, 1808 (1961), Soviet Phys. Solid State 3, Solid State 3, 592 (1961).
63
1314 (1961). V. M. Agranovich, UFN 71, 141 (1960), Soviet Phys.
36b
V . I. Cherepanov, FTT 3, 1493 (1961), Soviet Phys. Uspekhi 3, 427 (1960).
64
Solid State 3, 1082 (1961). V. M. Agranovich and Yu. V. Konobeev, FTT 3, 360
37
E. F. Gross, A. G. Zhilich, B. N. Zakharchenya, (1961), Soviet Phys. Solid State 3, 260 (1961).
65
and A. V. Varfolomeev, FTT 3, 1445 (1961), Soviet M . Born, Phys. Zs. 16, 251, 437 (1915); Ann. d.
Phys. Solid State 3, 1048 (1961). Phys. 55, 177 (1917).
38 66
A. G. Zhilich, FTT 3, 2041 (1961), Soviet Phys. C. Oseen, Ann. d. Phys. 48, 1 (1915).
Solid State 3, 1483 (1962). 67
S. Chandrasekhar, Proc. Indian. Acad. Sci. A36,
39
V. I. Cherepanov, FTT 3, 2183 (1961), Soviet Phys. 103 (1952); A37, 468, 697 (1953); A39, 243 (1954).
68
Solid State 3, 1583 (1962). V. M. Agranovich, DAN SSSR 97, 797 (1954); Opt.
40
S. I. Pekar and В. Е. Tsekvava, FTT 2, 261 (1960), i spektr. 1, 338 (1956) and 2, 738 (1957).
69
Soviet Phys. Solid State 2, 242 (1960). Yu. A. Tsvirko, JETP 38, 1615 (1960), Soviet Phys.
41
В. Е. Tsekvava, FTT 3, 1164 (1961), Soviet Phys. JETP 11, 1163 (1960).
70
Solid State 3, 847 (1961). B. N. Samoilov, JETP 18, 1030 (1948).
42 71
G. Ya. Lyubarskii, Teoriya grupp i ее primenenie V. M. Agranovich, FTT 2, 1197 (1960), Soviet Phys.
v fizike (Group Theory and Its Applications in Physics), Solid State 2, 1084 (1960).
72
Fizmatgiz, 1958. M. S. Brodin, A. F. Prikhot'ko, Opt. i spektra. 7,
43
A. S. Davydov, Teoriya pogloshcheniya sveta v 132 (1959).
73
molekularnykh kristallakh (Theory of Absorption of M. S. Brodin, A. F. Prikhot'ko, and M. S. Soskin,
Light in Molecular Crystals), Kiev, Izd. AN Ukr SSR, ibid. 6, 28 (1959).
74
1951). M. S. Brodin and S. I. Pekar, JETP 38, 74, 1910
44
H . Winston, J. Chem. Phys. 19, 156 (1951). (1960), Soviet Phys. JETP 11, 55, 1373 (1960).
45 75
L. D. Landau and E. M. Lifshitz, Quantum Mechan- G . B. Dubrovskii, FTT 3, 1305 (1961), Soviet Phys.
ics, Pergamon, 1958. Solid State 3, 943 (1961).
76
46
E . F. Gross, UFN 63, 575 (1957;. 1 . S. Gorban' and V. B. Timofeev, DAN SSSR 140,
47
E. F. Gross and A. A. Kaplyanskii, FTT 2, 1676, 791 (1961), Soviet Phys. Doklady 6, 878 (1962).
77
2968 (1960), Soviet Phys. Solid State 2, 1518 (1960) В . Е. Tsekvava, FTT 3, 1164 (1961), Soviet Phys.
and 2, 2637 (1961). Solid State 3, 847 (1961).
48 78
E. F. Gross, B. P. Zakharchenya, and L. M. S. Nikitine, Proc. Intern. Conf. Semiconductor
Kanskaya, FTT 3, 972 (1961), Soviet Phys. Solid State Physics, Prague, 1961, p. 442.
79
3, 706 (1961). V. M. Agranovich and A. A. Rukhadze, JETP 35,
49
S . A. Moskalenko, FTT 2, 1755 (1960), Soviet Phys. 1171 (1958), Soviet Phys. JETP 8, 819 (1959); V. M.
Solid State 2, 706 (1961). Agranovich, V. E. Pafomov, and A. A. Rukhadze, JETP
50
D . G. Thomas and I. I. Hopfield, Phys. Rev. Lett. 36, 238 (1959), Soviet Phys. JETP 9, 160 (1959).
7 9 a
5, 505 (1960). F . G. Bass, M. I. Kaganov, and V. M. Yakovenko,
51
D . G. Thomas and I. I. Hopfield, Phys. Rev. Lett. JETP (in press).
4, 357 (1960); Phys. Rev. 124, 657 (1961). 80
V. M. Agranovich and V. L. Ginzburg, JETP 40,
52
E. F. Gross, B. P. Zakharchenya, and O. V. Kon- 913 (1961), Soviet Phys. JETP 13, 638 (1961).
81
stantinov, FTT 3, 305 (1961), Soviet Phys. Solid State V. P. Silin and E. P. Fetisov, JETP 41, 159 (1961),
3, 221 (1961). Soviet Phys. JETP 14, 115 (1962).
53
D . Foks and S. Yatsiv, Phys. Rev. 108, 938 (1957).
710 V. M. A G R A N O V I C H a n d V. L. GINZBURG
82 87a
Yu. A. Tsvirko and M. A. Tolmazina, FTT 3, 1393 V. M. Agranovich and A. A. Rukhadze, Fizika
(1961), Soviet Phys. Solid State 3, 1011 (1961). tverdogo tela (Solid State Physics), Collection of
83
V. S. Mashkevich, JETP 38, 906 (1960); 40, 1803 Articles, Vol. 2, 235 (1959).
88
(1960); 42, 135 (1962), Soviet Phys. JETP 11, 653 (1960), V. L. Bonch-Bruevich and S. V. Tyablikov, Metod
13, 1267 (1961); 15, 97 (1962). funktsii Grina v statisticheskoi fizike (Green's Func-
83a
D . V. Sivukhin, JETP 18, 976 (1948); JETP 30, 374 tion Method in Statistical Physics), Fizmatgiz, 1961.
89
(1956), Soviet Phys. JETP 3, 269 (1956). Abrikosov, Gor'kov, and Pitaevskii, Metody kvan-
84
Ё . V. Shpol'skii, UFN 73, 187 (1961). tovoi teorii polya v statisticheskoi fizike (Quantum
85
P. P. Feofilov and A. A. Kaplyanskii, UFN 76, 201 Field Theory Methods in Statistical Physics), Fizmat-
(1962), Soviet Phys. Uspekhi 5, 79 (1962). giz, 1962 (in press).
86 90
V. M. Agranovich and M. I. Kaganov, FTT 4, 1681 S. I. Pekar, UFN 77, 309 (1962), Soviet Phys.
(1962), Soviet Phys. Solid State 4, 1236 (1962). Uspekhi 5, 515 (1962).
87 91
A. S. Davydov and A. F. Lubchenko, JETP 35, 1499 S . I. Pekar, FTT 4, 1301 (1962), Soviet Phys. Solid
(1958), Soviet Phys. JETP 8, 1048 (1959). State 4, 953 (1962).

Corrections to Part 1 of the Review [UFN 76, 643 On p. 333 [first line following Eq. (1.32)]: printed
(1962), Soviet Phys. Uspekhi 5, 323 (1962)]. "indices ej^ and . . . " should be (quantities ej_
In Table II (p. 341) in the column "principal axes On p. 334. The fifth, sixth, and seventh lines follow-
of the t e n s o r " it is shown which axes are fixed from ing Table I are superfluous.
symmetry considerations. This, of course, does not On p. 335. Eq. (1.49). Printed — y-, should read
always pertain to the principal axes of the second-rank 2 a t 47Г
tensor eij(oj). For example, for cubic crystals this в
9t 8тг
tensor has the form ejj = e6jj in any coordinate frame.
On p. 342 (last line of left column). Printed
We hope that this does not lead to misunderstandings,
inasmuch as everything is essentially explained in the ei о ох Л ° °\ Лх ° ° \
text of the article. о еа о , should read I 0 et о | = l 0 e ± 0 J. .
0 0 e3/ \o о E3/ \0 0 e,/
On p. 327 (second line below Fig. 1) it should read
бу(шц) in lieu of e y ( w ) . On p. 337. In place of €ц formula (1.66) should
contain everywhere ejj.
On p. 332 [fourth line following Eq. (1.22a)]: printed
(1.23), should be (1.22a). Translated by Z. Barnea and J. G. Adashko