molecular coordinate systems for relaxation processes
TRANSCRIPT
Journal of Molacdar Liquids, 62 (1992) 81-96 EXsevier Science Publie.bem B-V_. Amntedam
81
MOLECULAR COORDINATE SYSTEMS FOR RELAXATION PROCESSES+
JAMES McCONNELL School of Theoretical Physics, Dublin Institute for Advanced Studies, Dublin 4, Ireland,
(Received October 18,199O)
SUMMARY The choice of sui table coordinate systems for the theoretical investigation of nuclear magnetic relaxation processes is considered for both rigid and nonrigid molecules. Expl4ci t results are obtained for certain correlation functions, spectral densities and relaxation times,
INTRODUCTION
In the discussIon of relaxation processes associated with molecular
liquids 4t is often found convenient for calculational purposes to work
in more than one coordinate system. Thus for dielectric relaxation one
may work in both a laboratory system, which we shalt denote by S, and a
mol ecul ar system. The same i s true for nuclear magnetic rel axation
ari sing from di pal ar 5 nteraction and from spi n-rotat I onal interaction.
When the nuclear magnetic relaxation arises from chemical shift or
electric quadrupolar interact4on, a second coordinate system fixed in
the molecule is also needed, Nuclear magnet4c relaxation by seal ar
interact4on can be dl scussed ent4rely in the system S-
Dielectr4c ret axati on has al ready been treated at some
length (ref, 1). It was then found necessary to take due account of the
inertia of the molecules involved in the relaxat-lon process. This 4s
not required for the present accuracy of nuclear magnetic relaxat4on
experimental results - Thus rotational d4ffusi on theory wi 11 be adequate
for the subsequent sections which are devoted entirely to nuclear magnetic
processes_
*Dedicated to Professor H .G -Hertz
0167-7322/92/$05.00 8 1992 - E%avisr Science Publie.bem B.V. AU righta raewed
82
In the next section a mathematical treatment of the rotational
Brownian motion of a rigid molecule is summarized and a mol ecu1 ar
coordinate system S’ is defined_ A second molecul at- system S’/ is
introduced in the following section for a discussion of quadrupol ar
relaxation. In the succeeding section the general resul ts derived
earl ier are applied to the symmetric and asymmetric rotator molecular
model s. The final section serves as an introduction to the influence of
molecul ar nonri gi di ty on nucl ear magnetic rel axati on processes.
THE INERTIAL COORDINATE SYSTEM
We consider the rotational Brownian motion of a rigid molecule in a
heat bath. Many of the early studies of this subject were based on the
Fokker-Planck equation (ref. 21. A more recent approach is based on a
study of -the stochastic rotation operator R(t). This describes the
rotation of the molecule from its orientation at time zero to its
orientation at time t, and so R(o) is the identity operator. Now R(t)
obeys the equation (ref. 1, chap.71
dR(tl dt = -i (4 . w(t)) R(t) ,
(1)
where J,, J,, J, are the infinitesimal generators of rotation. We can
in principle solve (1) for R(t), provided that we know the angular
velocity g(tl of the rotating molecule. For most rel axation problems
the value of the ensemble average <R(t)> is all that is required.
Indeed the matrlx element <R(t)>,, related to the basis constituted by
first or second rank spherical harmonics with polar
angles fixed in the molecule often plays the part of
function.
In order to find w_(t) we take a set of molecular
with or
axes of
by II,
to make
igin at the centre of mass and axes coinciding
inertia. We shall refer to S’as the inertial
I I, 1 l the principal moments of inertia at the origin. In order
progress with tl-.z sathematical investigation we assume that the
inertial tensor and the rotational friction tensor are simultaneously
diagonalfzable, Then corresponding to I, , I a , I, we shall have
rotational friction coefficients 1,. 1%’ J,, say, and a frictional _-ag
c,~,(t), czwdt). c,.wdtL The heat bath produces a thermal couple with
and azimuthal
a Green’s
coordinate axes S’
with the principal
system. We denote
83
components denoted by N,(t), N,(t), N,(t) and these satisfy the
relations (ref. 3)
<Ni(t)> =o
<Ni(t) NJ(S)> 3 2kTaij Cib(t-S)- (2)
Since S/ is a system of rotating axes, the components of angular
velocity obey the Euler-Langevin equations
I, doAt) - (1 - 1,)~~ (t)w, (t.) = -cLw,(t) + N,(t), etc. - a (3) dt
We emptoy (2) to find stationary solutions of (3). This has bec;l done
for spherical , 1 lnear, symmetric rotator and asyrmnetric rotator mot ecu1 ar
model s _ We can then proceed to solve (1) for R(t) (ref. l)-
For nuclear magnetic relaxation processes a knowledge of the
rotational diffusion limit of <R(t)> i s al 1 that i s requi red for our
purposes. It is given by (ref. 4; ref, 1, p. 194)
<R(t)> =exp 1-i 7 Jft ]_ (4) i=I i
The rotational dl ffuslon tensor i s di agonal and its nonvanl shi ng
elements D,, D,, D, are given by
We see from (4) and (5) that
<R(t), = exp [-(D, J: + D, J; + D, J:)t 1. (6)
If the value of R(t) is required, as it is for the study of nuclear .
magnetic relaxation by spin-rotational interaction, it may be expressed
in series form by
R(t) =(E + cF(‘)(t) +E* F(*)(t) + . . . ) < R(t)> ,
where E is the identfty operator, G is a small dimensionless parameter
and the F(j)’ s are stochastic functfons defined in ref, 4-
84
TWO MOLECULAR COORDINATE FRAMES
We mentioned in the Introduction that in order to investigate nuclear
magnetic relaxation by ani sotropic chemical shift or by electric
quadrupol ar interaction two mol ecu1 ar coordinate systems
Since the mathematical prob’Iem is much the same for these
mechani sms, we shall confine our attention to quadrupol ar
A theory of nuclear magnetic re7 axat-i on appl icabl e
are required.
two relaxation
interaction.
to asymmetric
rigid mol ecu1 es has been based on the stochastl c rotational operator
(ref. 5). We start with a 1 aboratory frame S having cartesi an
coordinates x, y, z. a constant external magnetic field in the
z-direction and a molecular nucleus situated at the point (x, y, z) at
time t. We regard all
denote by V the scalar
charges _ Taking second
other charges in the molecule as fixed and we
electric potential at (x, y, z) of these other
derivatives of V we write
F 3 = $vzz, Fkl = :;i tVLX ’ ‘%y’
F = *1 2% (V,x - vyy z 2ivxy)*
(7)
The five quantities F1 are the components of a spherical tensor of
rank 2 (ref -6, section 9.1). Since V is a scalar, we deduce from (7)
that the five quantities Fi/ defined by
Fg’ = $V,~,N , Fz, = ;$ (V/,/k iV,/y/)
F?: x’x’ - Vy/y’ 52iV)(/yf I
constitute a spherica tensor or rank 2 in the inerti al frame S .
(8)
The array
I
Vx’x’ Vx’y’ Vx’z’
Vy’x’ VY’Y’ Vy’r’
V-’ _ x’ Vz’y’ VZ’Z’
is the field gradient tensor-
(9)
We now take a second body coordinate system S” with origin at the
relaxing nuctelrs and with x” , y”, z” coordinate axes in directions that
make (9) diagonal so that the only nonvanishing elements are Vx~~x~~) Vyfimy”.
Vz”z” ) and we order these in such a manner that IVZ@~z” 1~1 Vxl~x~~ 1~1 Vy~~yl~ 1
We deduce from IS) that we have in SO a spherical tensor with components
F;s, = 3 V =,,= 1, , Fy , =O, F:, = rl ‘7”7: , -
2(6)3
where
n= V,“,” - V,“,”
. Vz”z”
the asymmetry parameter _ It is usual to write VZm~t~~ as eq and
therefore express (10) as
FZ = 1 eq, Fll =O, F;’ = n eq 2(61$
(10)
(11)
It is to be remembered that S’ and S” are both coordinate frames
fixed in the molecule which contains the nucleus, the origins of the
frames being, respectively, the molecular mass centre and the relaxing
nucleus, Since the rotation of .a rigid body about any axi s through the
origin of S’ is equivalent to a similar rotation about the parallel axis
through the origin of S” together with a translation and since
translational motion of the ml ecu1 e does not i nf 1 uence the relaxational
probl em under 5 nvestigati on, we may say that a rotation about any axis
through the origin of S Is equivalent to an equal rotation about the
parallel axis through the origin of S” _ If then we go from S’ to S” by
a rotation specified by the Euler angles Q ) s, y (ref -6, appendix B).
we can relate the functions in (8) and (10) through the equation
(121
where Glrn (a. e,~) is a Wigner function.
In order to derive expressions for the spin-lattice relaxation time T,
and the spin-spin relaxation time T, we first define the spectral
density J(W) by
j( w 1 = 3 {z<F, (01 F, (t)> giW%lt.
where Fo comes from (7). An alternative expression for f(w) is provided
by (ref.3, appendix)
’ J”1 j(W) = l%, mj=_r $,,’ <R+(t) >mm, Fp, dt
, (13)
with Fm’ , F;, defined by (8) and the matrix element of CR+(t)>
calculated for the basis constituted by the spherical harmonics Y,‘,
associated with S’_ We may obtain F,/ for insertion into (13) by
inverting (12) and employing (11). so that we obtajn
F; us.&2 D;;(a, B, Y) F; (14)
= ( 4)L eq YZ-, (0, a) + qeq
ZP i D~z(a,~.y) + Dmy_, (a.e.r)I-
To complete the calculation of j( 0) from (13) it is convenient to
introduce the operator T ( W) by the equat-i on
T( w) =I; <R+(t)> e -i tidt
and the operator Q ( W) by
a( w) = -C (w) + T (w)+ *
so that (13) becomes (ref. 6, appendix D)
Then for a spin one nucleus with electric quadrupole moment eQ the
longitudinal relaxation time T, and the transverse relaxation tSme T,
are gtven by (ref.6, ~~-120, 121)
+ = g t+-12 ij(mo) + 4jt20,) I
+ = $ ($+)2 (3j(o) + 5j( wg) + 2j (20,)I ,
(15)
(16)
where oO is the Larmor angular frequency corresponding to the external
magnetjc field Introduced at the beginning of this section,
APPLICATION TO RIGID MOLECULAR MODELS
We shall now employ results derived in the previous section to
calculate the spectral density j(o) for quadrupolar relaxation when the
relaxing nucleus is sStuated in a rjgid asymmetric molecule.
For simplicity we work in rotational diffusion theory so that CR(t)>
is given by (6). Then (ref, 6. appendix D-3)
a(dmn=
where
A+A* o D+D* o F+F*
0 B+I3* 0 E+E* o
D*D* o C+Cf 0 D+D*
0 E+E* o 8+B* o
F+F* o D+D* o A+A*
A= ac - d’
a(ac-2d’ ) , B=J b , C= a
b’- e’ ac - 2d’
D=- d ac-2d2
,E= -e b2 -e2
,F= *’ a(ac-2d2 )
(17)
(18)
a= DI + DZ + ~DJ + i UI, lp 9 (0, + Dt) + DJ + iw
c= ~(4 + D* ) + i o, d = ($I* e = (9)’ (D, - Dz)
and D, , Dz, DZ are the rotational dlffusjon coefficients associated
with the S’ coordinate axes. From (15) and (17) we deduce that
j(hl) p &t (A+A*) (F’ F 1 Jc + --F’F’* 1 + (B+ If) (FL Fl,f + F,’ F’*)
+(C+C*) F,’ F,1*-: (o:D*) ‘(;I F,‘* + F,’ F’* + F,’ F’* + F” F,‘*) (19) +(E+E*) (FIEF,‘” + F; FlF) +-;F+F*) (F’-;‘* + F”F’* )< _
We note that the molecular shapzenters iitz(l9) through
A, B. C, D. E, F and that the quadrupolar relaxation process enters
through the F’m’s. In order to express the latter 1 n terms of Euler
angles we employ ( 14) and
D&m [a -6,~ =e ) 1
-1 (mb +IIIY )~,m( e)
dm/m ( e I E I(2+m) ! (2-m ) ! (Z+m’ ) ! (2-m’) !I4
X z (-1s (cos ~ft)4m-m’-2s (-sin ~a)m’-m+2s
, s (2-m/-s)! (2+11-s)! (2+m’-m)! si
where the sulmnation is over all values Of S for which the arguments of the
factorials are nonnegative. Then on substitution into (19) we obtaSn after
a lengthy cal cul ati on
(A+A*) fz Sin’B + : Sdn*B COS2Y [Sln‘jEi + COS'iB]
+ 5 12 sin’iB + t sfn’s cos 4 Y + 2 cosmJq )
+ (B+p*) ( 3 ii;
sir&B cc6 2r
+ g [2 SlIP6 (sllr ltl + co!?~eI - sin*@ cos 4~1)
+& tc + P) (3 cafe - 1 + n sin% COSa)*
+ $ (D + D*) (3 COPE (20)
- 1 + n slnB COSa)
x ((314 sina CoSZa + 2n 2 2 [cossjg COS(2a+2Y) +sin+~ecos(2a-2Y)lI
+ (E+E*) (- 3 sirP 20 cos 2a + 2 sin 0 sin 20 16 4
x [ sinzlB cos(b-2~) -cosaiB cos(aO+211 1
+ z sin28 [J sfn’6 COS 2a 12
- co!9 & c0s(&~1-s~n4~Bc0s(2a-4Y 11)
+ (F+F*+) ( 3 x sln’e cos4a
+ g [cos’~Bcos(~+4Y1
Thus by introducing the three coordinate
from S and S” to 5’ we have succeeded in
+ a sin28 [coscae cos(4a+2Y)+si1+ jecos(4a-2r)l
+ s-ln’$fl cos(4a-4Yl +$ sin*BcosqPl) 3
_
frames S.S’,S” and transporting results
deriving an analytSc expression for the
spectral denstty, The values of T, and T, are then obtainable by substStuting
(201 into (161.
It lsobvious from (181 and (201 that the calculation of explicit
results for the asymmetric molecule can be a complicated process. This Is due
basically to the noncorunutatlvity of the operators J:. Ji. J: that appear In (6)_
For this reason one often simplifies the mathematics by taktng the molecular model
to be a symaetrlc rotator. If the axis of rotational symmetry is the third S’ axis,
then D, = D, and (61 yields
<R(t)> = exp E- [ 4 J= + (4 - D, ) J: It,
= exp. (-D,J’ t) exp [(q - 0, ) J: t].
au
(21)
Moreover, from (la),
d=e=O
so that
A z a-1 = (20, + 4D, + iw )-I
B = b-’ = (5D, + D, I- iw )-I
c = c-’ = (6D, + i w ) -’
D=E=F=O.
Hence eq. (20) will be considerably reduced,
In the rep:*esentation with basis consisting of second rank
spherical harmonics associated with S’ the representative of Jz is 6
times the unit matrix and the representatives of J: is the diagonal
matrix with successive elements 4, 1, 0, 1, 4. It follows from (21)
that d R(t)> is represented by -the diagonal matrix with nonvanishing
el ernents e -6D,t ( e4(D,-D,)t ,e(D,-D,)t ) 1. e(D,-D,)t, e4(4 -D,)t ) ?
that is,
e-(2D, +4D, )t , e-(5D, +4 )t , e-6D.t, ,-(5D, + D,)t,e-_(2D, +4D,)t _ (22)
Let us employ (22) to calculate the autocorrelation function
of Yzm (e(t). 9(t)). Or< going from the laboratory frame S” to the
molecular frame S’ we obtain for t + o (ref, 3. appendix)
4’)lZ @ ’ )I *1,
that is,
<YCm ( O(O), ql(o)) Yam ( s(t), a (t)),
= (16*)-l I (3 COSa B m _ 1 I~ e-6Dlt +3 sin* 2 e I e-(5D~+Ih It
+3 sin* 9 ’ -(2[h + 4[h)lz) _ e
(23)
This agrees with a result of Hertz (ref - 7, eq.(l71). when it is noted
that he took the axis of rotatior;al syrmnetry of the molecule to be the
first coordinate axis of S’.
INTERNAL ROTATIONS
Irl the previous sections we have related the coordinate
frame S” to the inertial frame S’ by a rotation specified by constant
Euler angl es. The molecule has so far been regarded as rigid but in
fact very few liquid molecules are rigid, While nonrigid molecules have
been studied both theoretically and experimentally for more than thirty
years, the rrtolecu’l;ar models have been rather simple. Thus Frech and
Hertz (ref.81 distinguish bet&en fast and slow motions of parts of a
nonrigid molecule- It is assumed that the correlation function belonging
to a part with fast internal motion is influenced by the stow motion of
other parts, but that the carrel ation function of a slower part i s never
so influenced by the faster motions that it contains fast decaying
exponenti al s. The part with the slowest motion is referred to as “the
molecule”. It has al so been called “the standard part” (ref. 91 and
“the framework” (ref, 101.
An early theoretical investigatjon of internal rotations was proposed
by Stejskal and Gutowsky who examined nuclear magnetic relaxation by the
intramolecular interaction of the protons in the symmetric top CH,
molecule (ref. 11). A somewhat similar approach to the study of
relaxation in a two-proton system was made by Woessner (ref. 10).
Taking a 1 aboratory coordinate system S with origin at one proton and
initial line in the direction of a constant ekternal field tie he denotes
by r: the position of the second proton and by e,+ the polar and
azimuthal angles in S of an axis cf rotation fixed in the framework.
The vector c makes a constant angle A with the axis of rotation and a’
is the azimuthal angle of c about this axis with reference to the
framework system.
Let n,(t), n,(t), n,(t) be the direction cosines of g(t) referred
to S- Uoessner introduces the functions Fa (t), E, (t), F, (t), not to be
confused with F’s introduced in the previous sections, through the
equations
F, (t) = 1 -3n:. F1 (t) = (n, + ini)n, , F1 (t) = (n, +in, IpI (24)
91
It follows that In terms of spherical harmonics associated with the
poiar ar.d azimuthal angles of the proton-proton axis at time t
F,(t) = -(16/5$, Y-It)
F,(t) = -(8mNj12 Y,(t), h (t)* = @/,5+ y _,(t) (25) F,(t) = (32&5)iY,(t), Fz (t)+ = (32x/15$ Yz;_z (t) .
We recall that, if T is the gyromagnetic ratio of the proton, then (ref.
6. eq. (4.22) and (C.30))
j( w) = 9 I? Y”;,,, (0) Yzm (t)> emiti dt -m
and that the longitudinal 1, and transverse relaxatlon time T,
given by (ref. 6, eq, (4.58) and (4.59))
(26)
are
(27)
where o. is the Larmor angular frequency. From (25) and (26) with m=l
j( w) = e J-I_ <F, to)* F,(tB euiwt dt. (28)
Sfnce we are no longer dealing with a rjgid body, we cannot
employ the rotation operator technique of the two previous sections to
calculate the ensemble average in (28). Woessner empl oyed spherical trigonometry to express nk, n,. n, in terms of e , o , a , +’ by
n, = sin n cos 8 cos * cos 0’ + cos a sin 8 cos I# - sin A sin d sin&l n, = Sin A cos 9 Sin $I’ + Sin A COS 8 sfn * cos 0’ + COS A sdn osjn 0
n, = cos A case -S-inA sin 9 CDS 9’ and on substitution Into (24) found that
F,(t)+3 (3 c&A - 1) sin ecos e+ 1 sin ACOS A (cosze - sin’0 + cos e)e i@’ - t
+ 4 Sin A cos A (cos* 8 - sin* & - cos 9)e-‘*
- f Sfn*A (sin 8 tos 8 + sin 8 ) e2”’ (29) - ) sin’s (sin e cos e - sin 9 )e-2i@‘l ei’ -
92
Wri ti ;rg
1 = g1 , ei*f = g,, e -i*‘, g 1 V e2i V= gN , e-ZiQ/= g 5
we express (29) as
F,(t) = f,g, + fz gz + f, 91 + f* 91. + f, g, I (31)
We remember that a is constant, that the tumbling motion of the
framework is described by o and 9, and that the internal rotation is
described by 0’. Thus the fi’S are related to the tumbling motion and
the gi ‘s are related to the internal rotation.
Woessner now assumes that the tuabl ing rnotjon and the internal
(30)
rotati on are mutually 1 ndependent - Thi s goes beyond the assumptjon of
Frech and Hertz reported at the beginning of this section, but Woessner
claims that hl s assumption 3 s physical ly reasonable. Accepting it
al lows us to deduce from (31) that
<F’: (01 F, (tb = & i.j=l
<f~(ol fj(U> q;(o) gj(t)’ . (32)
The value of <go gj(‘)> for a Brownian motion model is found by
assuming a Gaussi an probabi 1 Ity dl stri butIon for the angle V’; that is
to say, we employ a djstribution function
PI a/. t) = $(*t =c-V exp (- P r&It). (331
where xc is the reciprocal of the rotational diffusion coefficjent
for the one-dimensional mot1 on. From (301 and (33) we find that
+$(o) gkc[t) > vanishes for i#k, and hence that
<FI*(o) Fl(t)r = <fp (0) f-i(t)> <go gi(t)'_ (34)
Moreover (33) yields
93
<exp c-i*‘(o) ] exp [i ia’(t)l> = exp[ -It I kc]
c exp [.2i 6; (9) ] exp [T21 O’(t) ]> = exp [ -41 tl /‘cc I - (35)
To complete the calculation of < FT(o)F,(t) > from (32) we note
that “91 *(o) gf(t)> is obtainable from (30) and (35). In order to
calculate < A?(o) fj (t)> we note that thi S describes the tumbling
motion of the molecule, whtch we assume to be rotationally diffusive_
Woessner now supposes that the molecule is spherIcal. Then one deduces
from (6) that
<R(t)> = exp [ -D J*t ] , (t5o)
where D is the co-n value of D L , D.- , DB . In the representation
having Y,, as basis elements, J* is 6 tjmes the identity E and so
<R(t)> = e-‘6DtE . (36)
We see from (29) that fi(t) may be expressed as a linear combination of
the Y 2m(t) ‘s, Replacing 6D by T,: we deduce from (36) and section 7.4
of ref .l that
<f;(o) fjW> = <If3 (0) I’> e-W)/TCiL > where T c, is independent of f . On collecting the results it is found
that
<F:(o) F,(t) = & (3 cos2d -1)’ exp (-It1 /TCI )
+ 1
!¶I sin22 A exp [- Itl h,! +‘$)I
+ 10 sfnC A exp [- It.1 (7;: + 4$ ) ] I
Let us define xc2 and T,_, by
(37)
(38)
We then deduce from (28). (37) and (38) that
(39)
On emptoylng (25) we imnediately find that (39) is in agreement with
eq- (21) of ref. IO.
If a = 0, the proton-proton axis coincides wf th the axis of rotat-ion
and there I s no internal rotation. For vani shi ng a eq- (37) reduces to
<F, *“l(o) F, (t)> = $- exp (- Itl / Tc,)
and so the presence of internal rotations introduces two terms proportional
to exp (-Id 1~;: + ~~~ I) and exp (-4ltl [T$ + T~‘I )- It has
been shown by Hertz that precisely the same occurs for a symmetric top
molecule (ref.7, eq. (26))_ Thus in place of the right hand side of
(23) we shall have terms with time dependences
exp (-6D,t) , exp (- [6D1 + T;’ 1 ) , exp (- [6ch + 4~Cl3 )
exp (- [5D, +D, ] t) , exp (- [ 5& + DI + r c ‘I t) ,
exp (- [~DI + DS + Will t)
exp (- 12D1 + 4D,1 t) _ exp (- [ 2D, + 4D, + Till t),
exp kC2D, + 4D, + 4~;’ 1 t)-
Hence each term on the right hand side of (23) Is split Into three terms
as a result of the internal rotation.
CONCLUSION
An account has been given of the various coordinate systems employed
95
in the investigation at molecular level of djelectric and nuclear
magnet-it ret axati on processes. Apart from a laboratory coordinate system
the most fundamental system 1 s the inerti al molecular frame. This is
extremely useful for a discussion of the rotational Brownian motion of a
rigid body and it Is appltcable not only to spherical molecules but also
to ljnear, ci rcul ar pl ate, symmetric rotator and asynmtetric rotator
mot ecu1 ar model s.
For nuclear magnetjc ret axation by quadrupol ar interaction or
anisotropIc chemical shift 1t is necessary to employ both the inertial
frame and another coordinate system fSxed in the molecule. Et is shown
for these relaxatjon mechanisms that It is possible to provide an
analytical discussSon wfthout restricting the magnjtude of the
approprl ate asymmetry parameter.
To discuss relaxation for nontmigid molecules we have taken the simple
model in which a part of the molecule is free to rotate about an axis
fjxed in the main body of the molecule. The motion of the molecule then
consists of a slow tumbling motion of the major part combined with a
fast rotation of the smal ler part. In order to obtain a mathematical
solution it has been found necessary to suppose that the two motions
occur independently. Expressions for correlatfon functions and spectral
densities are then derived.
REFERENCES
1.
::
4.
2
:: 9.
10. 11.
J. McConnell, Rotational Browni an Motion and Diet ectric Theory, Academic Press, London, 1980. J. McConnell, J. Molec. Li . in press. G.W. Ford, J.T. Lewis and . McConnell. Phys.Rev.A 19(1979) 3 907-919. J. McConnell, Physica A 128 (1984) 611-630. J. McConnell, Physica A 117 (1983) 251-264. J. McConnell, The Theory of Nuclear Magnetic Ret axatl on in Liquids, Cambrldge Univ. Press. 1987. H.G. Hertz, Progress In NMR Spectroscopy 16 (1983) 115-162. T. Frech and H.G. Hertz, 3. Molec. Liq. 30 (1985) 237-282. M.S. Ansari and H.G. Hertz, Z.Phys.Chem. Neue Folge 137 (1983) 187-220. D.E. Woessner. J.Chem.Phys. 36 (1962) l-4. E-0. Stejskal and H.S. Gutowsky, J.Chem.Phys. 28 (1958) 388-396.