You are on page 1of 16

JOURNAL OF MAGNETIC RESONANCE

2, 286-301 (1970)

Isotropic NMR Shifts in Transition Metal Complexes: The Calculation of the Fermi Contact and Pseudocontact Terms?
ROBERT

J. KURLAND

Department of Chemistry, State University of New York at Buffalo, Buflalo, New York 14214

BRUCE

R. MCGARVEY

Department of Chemistry, Polytechnic Institute of Brooklyn, Brooklyn, New York 11201

Received July 31, 1969; revised January 16, 1970; accepted January 19, 1970 The Fermi contact and pseudocontact contribution to isotropic NMR shifts in paramagnetic complexes are considered for the following cases, for which formulae customarily used are not strictly applicable: (1) there is an appreciable contribution to the magnetic moment of the complex from unquenched orbital angular momentum; (2) there is appreciable orbital contribution (induced by spinorbit effects) from spin density at the ligand nucleus; (3) there is appreciable mixing of the ground electronic state and thermally populated excited states by the applied magnetic field. An approximate density matrix treatment is developed to handle these situations. The following examples are discussed as illustrations: octahedral (f&l and (t2$ complexes and complexes in which zero-field splitting occurs. I. INTRODUCTION

Isotropic NMR shifts in paramagnetic transition metal complexes have been used extensively to determine s-electron spin densities at various magnetic nuclei in the complex (I). The expressioncommonly used to obtain the Fermi contact term from the isotropic NMR shift is of the form (2) AH sPs(s+l) A -=[II h&W 3kT Ho where S is the rotationally averagedg value for the complex, /? is the absolute value of the Bohr magneton, yN is the nuclear magnetogyric ratio, AH is the isotropic shift at an applied magnetic field Ho. The constant A, is proportional to the density of
t This research was supported by the National Science Foundation (Grant Numbers GP-10463 and GP-7133). 1 AH is defined by AH = (H,-Ho), where Ho o is the resonance field measured for the same nuclear species in a diamagnetic reference compound, at the same resonance frequency as that employed for the paramagnetic compound. Thus AH is positive for upfield shifts (negative A., positive yN).
286

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

287

unpaired electron spin occupying s orbitals centered at the nucleus. In Eq. [I] A, is given in units of Hz. McConnell and Robertson (3) have pointed out that an anisotropic magnetic moment will produce an isotropic pseudocontact NMR shift, in addition to that given by the Fermi contact term. They and Jesson (4) have given the expression
(3cos2&- 1) AH p S(S+ 1) -=aI) R3 3kT HO for this shift. In this equation R is the angle between the principal symmetry axis of the complex and the vector between the metal ion center and the nucleus whose NMR is being observed; R is the distance between metal ion and nucleus; and F(g) is an algebraic function of the g-tensor values, the form of which dependson the relative magnitudes of electronic relaxation times and rotational correlation times. In the derivation of this equation it was assumed that R was large enough that the paramagnetic metal ion may be treated as a point dipole. Several other assumptions were made in the derivation of Eq. [l] and [2] that drastically restrict their application. First, it was assumed that the complex had only one thermally populated energy level (in the absence of the magnetic field), and the states of this ground level could be assigned a spin quantum number, S. Second, it was implicitly assumed that for S 3 1, splittings of the ground level in the absence of the field (zero field splittings) could be ignored; and, third, that the orbital contribution to the isotropic NMR shift could be taken into account indirectly through the use of g-tensor components in Eq. [l] and [2]. This last assumption requires that there be at most only a first order orbital contribution to the magnetic moment of the complex, brought about by spin-orbit mixing of the ground level with nonpopulated excited states. Many complexes for which the paramagnetic shifts have been analyzed by means of Eq. [I] and [2] do not satisfy one or the other of these assumptions. For example, complexes with a T, or T, ground term may, by virtue of spin-orbit coupling effects, have an appreciable orbital contribution to the magnetic moment and several thermally accessible excited levels (e.g., octahedral Co(II), tetrahedral Ni(II), octahedral low-spin Fe(III)). Moreover, a number of complexes are known to have relatively large zero field splittings: e.g., distorted tetrahedral Co(II), octahedral Ni(II), tetragonal Fe(II1) and Mn(I1) complexes. Jesson (4) attempted to include the contribution of thermally populated excited states in his calculations on octahedral Co(I1) complexes, but he neglected the significant contribution from the Zeeman mixing of these states. Golding (5) has derived equations for the contact shift in octahedral complexes taking into account thermally populated excited states and the Zeeman mixing of these states; however, his assumption that the spin density is the same for all the thermally populated states does not necessarily apply to many complexes that have been studied, as we shall demonstrate below. The point dipole assumption may not be applicable, in all cases,for the calculation of the pseudocontact shift. McGarvey and Pearlman (6) have suggested that this may be the reason that the 14N NMR pseudocontact shifts found in ferric cyano complexes are much larger than those predicted by Eq. [2].

288

KURLAND

AND MCGARVEY

If meaningful estimates of the distribution of unpaired electrons in paramagnetic complexes are to be obtained from isotropic NMR shifts, a theory is required which is not limited by the restrictive assumptions discussed above. Moreover, it may be particularly important in analyzing isotropic NMR shifts of nuclear speciesother than protons, to consider the orbital contribution from unpaired electrons in orbitals centered at the nucleus (7,8). This orbital contribution cannot be taken into account simply by use of the g tensor of the complex (8). We have developed a method of calculating both the contact shift and the pseudocontact shift which does not require the restrictive assumptions used previously. To illustrate the application of this method we shall discuss the following examples: S = 1, 3/2 and 5/2 states having zero field splittings; and octahedral (tz,) and (t2J5 complexes. Other cases will be discussed in future reports.
II. DERIVATION OF GENERAL EQUATIONS

In order that the paramagnetic NMR shift be observable, the electron spin relaxation rate, l/T,,, must be much greater than the nuclear-electron hyperfine interaction. (We do not consider here cases involving chemical exchange). The nuclear-electron hyperfine interaction can be expressed as z,v = (y,Pn)
A, *I=

(LV/~~) (A, +A,

+ A3

PI

-% = f geB 7 6 (pi) Si
A, = ge p J$ [3(si.rJ A, = 2/l C rsy3 li I ri- r: Si] r,: 5

where I is the nuclear spin operator, SN is the nuclear-electron hyperfine Hamiltonian expressed in units of Hz, and A,, A, and A, are operators representing, respectively, the Fermi contact, the dipolar, and the nuclear-electron orbital interactions in units of oersteds. The operators Si and li are, respectively, the spin momentum and orbital angular momentum for the ith electron; ri is the vector between the locus of the ith electron and the nucleus; and ge is the free electron g factor (= 2.00223). For systems in which G+ is much smaller than the nuclear Larmor frequency, the ~ nuclear spin is quantized along the magnetic field Ho and we can write for I
I = I,, h,

where Ih is the component of I along H, and h, is the unit vector along H,. With very short T,,, AH is obtained from an average over thermally populated states AH = (AN).ho. PI In general, the thermal equilibrium values can be obtained from a density matrix expression (9) CO = tr bW (PI II61 where p is the density matrix operator, A is the operator corresponding to the observable and tr indicates the trace. For a system in thermal equilibrium the density matrix operator is given by p = exp [-S/kT], iI71

CALCULATION

OF CONTACT

TERMS

FOR COMF LEXES

289

where .%is the time independent Hamiltonian for the system, k is the Boltzmann constant, and T the temperature. The form of 2 depends on the relative values of T,,, A,,, (rotational correlation time) and h/(E,, -EL) for a given state. In this paper we are considering systems(complexes with large zero field splittings or slightly distorted Tl or T2 ground states) in which thermal equilibrium is established so rapidly between states that we can treat the complex as fixed in space. This situation has been referred to as the solid state case by McConnell and Robertson (3). In this case 2 can be written as 2 = sfro+Afz PI where &?z is the Zeeman Hamiltonian for the interaction of the complex with the applied magnetic field and & is the Hamiltonian in the absenceof the field. $Pz may e be written as
iVz = -p.H,, = /?(L+g,S).H,,

PI

where II is a magnetic moment operator for the complex, L is the total angular momentum operator and S is the total spin operator. In general, X0 and %?z will not commute and the matrix elements of &z will be much less than those of X0. We label the eigenvaluesof 2Pc,as ar and the eigenstatesfor the Kramers multiplet corresponding to sr. as IIn). Then prn,rtrn,the density matrix element required for the evaluation of Eq. [6], may be put in terms of E,-and the matrix elements of %z. This is done by expanding exp [ - X /kT] by a procedure outlined by Schwinger and Karplus (10) and summarized in the Appendix. To first order in (I~j~? rn) z[I the expansion gives for those matrix elementsin which I = I
$fh,~-~ = e -&r/k:T [b,,- (r n~& m)/kT], and for those matrix elements in which lY # I ,
P m,pm = Qrr~~~+clrw~

cw
[lObI [lOc] Iill al

Qrr,
AH is then
AH=q- c p m,r&- f+b. l-n,l- m {

e-r/kT=
Er-+

e-+/kT .

h$-n) W+bh$W

AH = (kTq)-

~rme-LrlkT(rnl~.Holr~) ,

Cllbl
rn m In obtaining Eq. [I 1b] use was made of the fact that q = C P~,,,~,, = Ce-"rlkT.

[UC]

pzlA,.

hop-n> = 0

for each multiplet state. Equation [I lb] is applicable to solids in which Ho has a fixed orientation with respect to the complex. For complexes in solution we must average Eq. [I lb] over all orientations to obtain

290 AH/H, = (3kT$ c

KURLAND

AND MCGARVEY

i=x,y,2 1.

r~me-"rlkT<rnlPijrm)

(~ml&ilrn)

where ANi and pi are the components of AN and p, respectively, along a set of axes, i = x, y, z, which are fixed with respect to the complex. The commonly used expressions for contact and pseudocontact shifts, Eq. [l] and [2], come from the first term in Eq. [l lb] and Eq. [I 21. The second term arises from the fact that Ir,) are not eigenfunctions of the Zeeman operator, and the interaction with the magnetic field, therefore, results in a second order mixing of the IT states with the ground state. When (+ - +) is of the order of kT, this Zeeman m) mixing term makes a large contribution to the paramagnetic shift and cannot be ignored. The casein which z,,~is lessthan both T,, and h/(E,, -EL), referred to by McConnell and Robertson (3) as the liquid solution case, is more difficult to treat, primarily due to difficulties encountered in properly averaging the Zeeman mixing term, and will be considered in a later work. In systems in which (+ - +) is much greater than kT the result for AH is obtained by first computing the average of the matrix elements of or. and A, before averaging over the Kramer multiplet. As an example, consider the s case of the ground state being a Kramer doublet with two eigenfunctions we can s represent as I+) and I- ). In this case we would obtain for AH AH/H, = p &jkT WI where

P = 3 [(+l~=l+>+i(+I~Lyl-)+(+(~L,I-)l = -6 (szz+~yy+iLx) = --is B. B

AN= 3 [(+IAN,I+>+i<+IAN,(-)+(+lANxl->l.
III. THE PSEUDOCONTACT METAL ION CONTRIBUTION

[W [151

If the isotropic shift is to be calculated via Eq. [12], then, in general, one needs to know the explicit form of the wavefunctions representing the states Ir,). However, even if these wavefunctions are not known explicitly, an expression for the metal ion contribution to the pseudocontact shift can be derived in terms of the susceptibility. The term (A,+A,) in Eq. [5] which gives rise to the pseudocontact shift can be broken down into two contributions: one, which we denote by (A,+ A,)=, from unpaired electrons occupying orbitals centered at the ligand nucleus; the other, (A,+A,), from unpaired electrons occupying orbitals centered at the metal ion nucleus. This decomposition of (A,+A,) is equivalent to neglect of the overlap distribution formed from the ligand and metal ion orbitals. The metal ion term, can be written as an expansion about the metal ion nucleus as origin. G%,+AL~~~ Marshall (8) has shown that the leading term in such an expansion, corresponding
2 This averaging procedure would be appropriate for a polycrystalline solid. However, many complexes will have values of tl sufficiently large that the so-called solid case will apply even to solution samples as, for example, in the case of Fe porphyrin complexes, (17).

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

291

to the point dipole model, has the following form:

(A,+A,)*ho

= R-3 [(P.h,)-3((IL.o)(a.h,))].

WI

In Eq. [16], R is the distance from the metal ion nucleus to the ligand nucleus, IS is the unit vector along the line joining the metal and ligand nuclei and ~1is defined in Eq. [9]. Marshall (8) calculated higher order correction terms in this expansion for

some specific cases and showed these terms were small. Substituting Eq. [16] into Eq. [12] yields for AH, the metal ion contribution
the pseudocontact shift:

to

AH/H,

= $$

[(l-3

cos*!A)(X,,-Xj+sin*fl

cos 2 II/ (x,,-xXX)]

II171
tensor given

where xXx, xy,,, xzz are the principal axis components of the susceptibility by the Van Vleck equation (II)
~ii=(kTq)-' i

rn,rm

e -r/kT(l-nI$rn)

(I-w+$-n)

[I*1
In Eq. [17] 2 = (l/3) (x~~+x~~+x~~), R is the angle between cs and the z axis of com-

plex and $ is the angle between the projection of (T in the xy plane and the x axis.3 For a complex with only one thermally populated multiplet of spin number S

and the pseudocontact shift is given by: AH/H, = ;;(TSR:I! {

bzz2 + (9x,*+ syv2)1 (I-

3 cos*m

+ t (gyy2gxx2) sin2Q cos 2$}. -

I201

For the case of axial symmetry (gZZ= g,,, g,, = gYY = gl), Eq. [20] reduces to the corresponding equation given by McConnell and Robertson (3). For rapid rotation in this case we would average (p. Ho) and (AN. h,) separately, obtaining for the pseudocontact shift the equation :

(S+lE AH&~/H _ B
0

~~-. 9kTR3

[szz - t (9xX+ gyy)] Cl- 3 COS2Q + 3 (gyy- g,,) sin2R cos 2$

Pll

which reduces to the McConnell and Robertson equation (3) for axial symmetry. We emphasize that if the molecular orbitals for the unpaired electrons contain a p orbital centered at the nucleus in question, then, in addition to the contribution of the metal ion d orbitals (which contribution is approximated by Eq. [17]), we must also evaluate the matrix elements giving (A,+AJL. These matrix elements are proportional to (r,r3), the average value over the ligand orbital of rIr3. Even though the
3 Note added in Proof: W. D. Horrocks, Jr., has pointed out to us that he and others (22) have previously proposed for the metal-ion contribution, in the axially symmetric case,a formula in terms of the susceptibility components.

292

KURLAND

AND MCGARVEY

spin density at the ligand nucleus may be much less than that at the metal ion, (r; 3, may be much larger than Rm3, so that the pseudocontact contributions from metal ion and ligand centered spin may be of comparable magnitude. For proton NMR shifts, where the unpaired spin occupies an s orbital, (A,+AJL will be essentially zero, and Eq. [ 171should approximate the total pseudocontact shift. In Section VII we will discuss the calculation of (A, + AL)L. h, for a specific case,octahedral (t2Js complexes.
IV. THE FERMI CONTACT SHIFT AND SPIN DENSITIES

The Fermi contact contribution to the isotropic shift, AHF, may be expressed in terms of the spin density at the nucleus by means of Eq. [12] and [4a], to give

where we have defined a vector spin density matrix at the nucleus by (a) i-n,T = m
<rnlC j 6
(rj)

Sj)r m).

In the general case where Im) is not an eigenfunction of S, it can be expressed as a linear combination of eigenfunctions of S where S,]crM) = MlctM) and a is an index denoting the remaining quantum numbers necessary to specify the eigenfunction. The matrix elements of (oN) will be proportional to those of S using the basis states ]a) such that

In Eq. [25] and [26], the constants Kz,, represent empirically or theoretically derived proportionality constants which relate the orbital spin density contribution to the nuclear spin density. The basis state IaM) may explicitly include unpaired spin in the ligand centered s orbitals, in which case the contribution to (~~~)r~,r,~ be obtained can from the explicit form of the molecular orbital representing ]c&). On the other hand, there may be cases in which the state ]ctM) does not, in a zeroth order approximation, explicitly include ligand-centered s orbitals; but the unpaired spin associated with ]clM) may, via a spin polarization mechanism, produce a net unpaired spin at the ligand nucleus. The situation is analogous to that for organic free radicals, where the s orbital spin density at the nucleus can be related, via empirical or theoretical proportionality constants, to n-orbital spin densities in orbitals centered on neighboring nuclei. If the constants K,,, were all equal to the same constant K then Eq. [26] would become C~>rn, = ~(r+3~rh) l-m and Eq. [22] would reduce to that proposed by Golding (5) for contact shifts. However,

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

293

as we shall demonstrate in Section VI, this assumption is not necessarily valid in complexes having T, or Tz ground states. A simplified form of Eq. [22] can be obtained for those complexes in which all thermally accessible states belong to a single spin state S. In this case Eq. [27] is valid and, further, we can equate the matrix elements of S to those of p by the equation (ri+S,Ir n2) = - (Tnl~ilT m)/giiB. IPI If we define A, as A~ = (4/3) 9, P YN K, PI Eq. [22] can then be rewritten using the Van Vleck (II) susceptibility equation, Eq. [18], as I where xxx, xyy3 are the principal axis components of the susceptibility tensor. For xzz the simple case of no zero field splitting
y,, = bll fii2B2s(s + I>

[301

3kT

and Eq. [30] reduces to Eq. [l]. We shall find Eq. [30] useful when we discuss the case of zero field splittings in Section V.
V. SPIN-ONLY STATES WITH

ZERO FIELD

SPLITTINGS

We consider here zero field splittings associated with an axially symmetric distortion in complexes having orbitally nondegenerateground states characterized by a spin quantum number, S z 1; for example, octahedral complexes of Ni2+ (S = l), Cr3+, Mo3+ (S = 3/2); tetrahedral complexes of Co2+ (S = 3/2); weak field tetragonal complexes of Mnzf, Fe 3+ (S = 5/2). In this case the Hamiltonians % and & e z are given by the equations
sFo = D [Ss,--(l/3) S(S+ l)]

Pll
r321

sz = Pg II&Hz+ P91[UL + S,H,l

where S, is the operator for the spin component along z, the principal symmetry axis of the complex, and D is the zero field splitting parameter. Thus, the zero field levels may be labeled by ]M(, where A4 is the eigenvalueof S, and the energy for the Kramer s doublets in the absence of a magnetic field is given by the equation elMI = [M2-(l/3) S(S+ l)]D. PI We will restrict the discussion here to proton shifts and thereby take Eq. [17] as valid for the pseudocontact shift. For the contact shift Eq. [30] will be valid. We first give expressions for xZZand xxx = xY,,obtained by using Eq. [18]. These equations will be given in terms of the parameter X = D/kT. s= 1.
X 2gIj 2P2e-

x2 = kT(l + 2eeX)

I341 PI

2gL2p2(1 emX) Lx = kTX(1 +2eYX)

294 S = 312.

KURLAND

AND MCGARVEY

g$ (1+9e-2X) x22 = 4kT(l+e-2X)


x xx =F dP [ 1+;x(l-e-2x

WI

s = 512.
Y = izz

,I/

(1 -l-eeZX).

PI

s$ (1 +9e-2X+25e-6X)/(1+e-2X+e-6X) m
9+$-kie- X-2$e-f3X

(l+e-2X+e-6X).

PI

In the absence of zero field splitting, i.e., for the limit X = 0, the susceptibilities should be given by gii2p2 S(S+ I) ___-Xii =- P3kT and it will be noted that Eqs. [34]-[39] give these proper limiting values. At room temperature kT N 200 cm- while values of D are usually less than , 5 cm- Thus, we can expand the expressions given in Eqs. [34]-[39] to first order in . the parameter X = D/kT to obtain the following equations: s= 1. AHJH, AH/H, S = 312. AH,/H, AH/H, s = 512. AH,/H, AH/H, 3% P As 12kThlW 35p2(gi -g;> = 36kT = -sd 315 g (3 cos2i2- 1) R3 1 -2wq x I p(9~+5d) wg; -s:> x 4k ThJW 5P2(gi-gf) = 12kT = 5g PA, I -4(9,, -sJ 159 (3 cos2Q- 1) R3 x = 29 BA,

3WyivlW 2p (g,, -g I ) = 229kT 2

WI

PI WI
t-441

* 1 WI

Two conclusions can be drawn from an examination of Eqs. [40]-[45]. First, since (g,,-gJ is small it would take rather large values of X to produce significant deviations in the contact shift from that predicted by Eq. [l] and thus Eq. [l] will be sufficient to account for the contact shift even for complexes having a sizeable zero field splitting. Second, the pseudocontact term has two terms that could be separated by a variable temperature study. The first term is the contribution from the anisotropy of the g tensor as first derived by McConnell and Robertson (3) with a T-l dependence; the second term arises from the zero field splitting and has a Td2 dependence. In

CALCULATION

OF CONTACT

TERMS

FOR COMPLEXES

295

practice, it may not be feasible to separatethe T- term (which will include the contact term) from the T- term by temperature variation since the experimental range of temperatures may be too small and experimental errors too large; but evidenceof its presence could be obtained by comparing the extrapolated NMR shift at T = co (using a fitted T- plot) to that of analogous diamagnetic compounds. A comparison of the two terms in the pseudocontact shift shows that the zero field term makes a large contribution to the total pseudocontact shift. In octahedral Ni + a good approximation to D can be obtained (12) from the equation

E461
where J. is the spin-orbit coupling constant. In the case of Ni2+ (2 - 200 cm- and L/RT N 1 at room temperature) the zero field contribution is 25 % that of the contribution from g tensor anisotropy. For ions with S = 312and 512,D is generally much greater in magnitude than that predicted by Eq. [46] (12) and in these caseswe might expect the zero field term to make the major contribution to the pseudocontact shift. In Table 1 we give the values of D required for (I) a 1 ppm and (II) a 10 ppm pseudocontact shift when R = 5A, kT = 200 cm- (t - 15 (3 co? Q- 1) = 1, and (Z),
TABLE VALUES OF 1 10 PPM PSEUDOCONTACT SHE-W

REQUIRED FOI< (I) 1 PPM, (II)

s
1

D (cm- ) (I) I ppm shift (II) 10 ppm shift


25.8 5.14 0.924
-b

312

512

61.6 9.65

Values calculated by means of Eqs. [17], [34]-[39], text, and with values of R = 5 A, kT = 200 cm-l, (3 co? Q-1) = 1, Q,, = 9,. = 2. b 10 ppm shift not possible for given values of R, T and R.

g,, = gl. The magnitude of D required for a given shift falls off rapidly as the spin quantum number, S, increases.The results in Table 1 indicate, in accord with the experimental results (I), that the pseudocontact shifts for octahedral nickel come plexes (S = 1) would be too small to detect, since the magnitude of D is probably less than ca 5 cm- 1 for these complexes. Shifts for Cr3+ complexes(S = 3/2) should be small since the largest values of D reported (12) have been - 1 cm- Mo3+ . (S = 3/2) complexes are known (13, 14) to have much larger values of D and, therefore, the pseudocontact shift could be appreciable (15). D values for tetrahedral Co complexes(S = 3/2) are unknown but could be large enough to give significant + pseudocontact shifts. In the high spin hemin, methemoglobin and porphyrin complexes of Fe3+, D is of the order of 5-10 cm-* (16) and the pseudocontactterm could be important in these complexes. However, the observed linear variation of the proton shifts with l/T (17) suggeststhat the pseudocontact term may be unimportant compared to the large Fermi contact shift in these complexes.

296
VI. CONTACT

KURLAND SHIFT FOR

AND MCGARVEY (f& IN AN OCTAHEDRAL FIELD

We consider next the contact shift for an undistorted octahedral (t,,) complex. This example will illustrate what part thermally populated excited multiplets, with incompletely quenched orbital angular momentum, play in the contact shift. In the absence of spin-orbit coupling, the (tz,) configuration gives rise to a sixfold degenerate ground level. This degeneracy is partially lifted by the spin-orbit interaction to give a four-fold degenerate ground level and an excited doublet level, at 3/2 2 above the ground quartet (where 3 the spin-orbit coupling constant). is If only the ground quartet were considered, Eq. [l] would predict no contact shift, since S N 0 for the quartet. Taking into account the thermally populated doublet, Golding (5) derived the equation below for the contact shift: AH F lH 0 = -d!-..
(y,Pn) 6kT

[!g

+ (1 -

5)

r- 3v]li2+r Y] 3/z

[47]

where V = i/kT. The derivation of Eq. [47] assumed (implicitly) that the spin density for each basis state was the same (i.e., all constants K,,, = K). However, we will show below that an equation similar to Eq. [47] can be derived from Eqs. [22] and [26], if certain reasonable assumptions, less restrictive than the above, are made about the constants K,,,. The states of the ground quartet, ]3/2, M), and those of the excited doublet, ] l/2, M), can be expressed as: 13/2, 3/2 > = IT.&), + >
/3/Z 13/L 13/T /l/2, l/2) --l/2) -3/T) l/2) = U/h = (l/h = IT2(-1), Ct'~~T,(o), [d4T,(O), > + > SIG(l), > +IT2(-1), >I + >)I

= (1/43)[JT,(O),

+ > -JZJT,(l),
- > +L:$,(-11,

- )I
+ >I

11/Z, -l/2)

= (l/h

[-IT,(O),

t.481

where the states IT,(O) ), ]T,(l) ), ]T2( - 1) ) are represented by cZs molecular orbitals having the same transformation properties, respectively, as the metal ion orbitals :

id,,,5 (4, - id,,),


l/d (d,, + id,,); ? + and - denote, conventionally, spin functions corresponding to eigenvalues S, = + l/2, S, = - l/2, respectively. In Eq. [48] we have labeled the states to emphasize the isomorphism between tlg molecular orbitals and corresponding metal-ion p orbitals (18). Since the complex is octahedrally symmetric, we need consider only those ligand nuclei located along the z axis of the complex. Although symmetry considerations preclude the direct admixture of ligand centered s orbitals in the t29molecular orbitals, s-orbital spin density at a ligand nucleus can be introduced by a number of other mechanisms, some of which would be analogous to those considered for aromatic free radicals.

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

297

One such mechanism involves spin polarization of electrons occupying ligand s orbitals by unpaired spin occupying metal ion orbitals. For a ligand on the z axis, unpaired spin in metal ion d,, orbitals (corresponding to the ]T2(0) ) basis state) will not produce the same spin polarization at the ligand nucleus as that from metal ion d,., or d,, orbitals (corresponding to IT,(l) ) or IT2( - 1) ) states). Unpaired electrons occupying ligand-centeredp, andp, orbitals will also produce spin ,, polarization at the ligand nucleus. The pX andp, orbitals will mix with metal ion n and d,, orbitals, respectively, but not with metal ion dxy orbitals. Accordingly, the px and p,, orbital spin density will be proportional to that for the metal ion d,, and d,, orbitals. With the above consideration in mind, we can set the proportionality constants K,,, given in Eq. [25], as follows:
K, =K T~(I),T~(I) = KT>(-I).T~(-I) K, = K TZ(O). TAO) [@I

with other constants set equal to 0. Note that K, takes into account polarization by unpaired spin in metal ion dxy orbitals and K,, polarization by unpaired spin in metal ion d,, and d,, orbitals and in ligand px and p,, orbitals. Even in the absence of the ligand source, K, and Kl would not be equal, however. Using Eqs. [26], [48], and [49], we can exhibit the required spin density matrix elements, (c$)~~, J:Mj (i = X, y, z) in the tabular form shown below:
J 312 312 312 312 312 312 112 l/2 l/2 l/2 M 312 112 l/2 -l/2 -l/2 - 312 112 112 -l/2 -l/2 J l/2 l/2 l/2 112 l/2 l/2 l/2 112 l/2 112 M l/2 112

C&lM,JW - KlIJ6
0 - ~,,3& ~,,3Ji 0 G/J6 0 - Ko/6 -Ko/6 0

<a,",,M,J fMT

(aEh,JfMs
0 (Kc, + K,)PJi 0 0 Wo + K,)Pd

X,/J6
0

-l/2
l/2 -l/2 -l/2 112 -l/2 l/2

iKo/3JZ iKo/3J2
0

X,/J6
0 iKo/6 - iKo/6 0 Wo -2oKl)W
0

-l/2 -(K, - 2fCl),6 The matrix elements of the dipole moment operator, p, can be evaluated by use of the tZs-p isomorphism. We tabulate these matrix elements below, setting ge as 2:
J M 312 112 l/2 - l/2 - 312 112 l/2 - 112 --l/2 J w 112 112 112 112 112 112 l/2 l/2 M 112 112 -l/2

wJM,J M

312 312 312 312 312 112 l/2 l/2 112

-iJ3/2p
0

-~lJ2
- ij@i, iP iP 0

-l/2
-112 112

-l/2
112 -l/2

298

KURLAND

AND MCGARVEY

These matrix elements substituted into Eq. [22] give the expression below for the contact shift:

[2 + e-(32)7. c501 Equation [50] is formally the same as that derived by Golding (Eq. [47]), if we set the Fermi constant, A,, equal to $g,&(+Ko++K1). For likely values of the parameter V, the major contribution to the contact shift arises from the (16/3 V) [l - e- (3/2)] term resulting from the Zeeman mixing of the excited doublet with the ground state quartet. In order to appreciate the magnitude of this contribution, note that a value of A = 150 cm-l gives a shift calculated from Eq. [50] equal to that calculated from Eq. [l] for S = 1.47 and S = l/2, values of the Fermi constant, A,, and kT = 200 cm- being the same in both cases (note also , that 1 g 154 cm-l for Ti3 ). We emphasize that cross terms I& have been neglected in the derivation of Eq. [50]. While these terms may be relatively small, by analogy with the situation for organic free radicals (19), they are not necessarily zero. Moreover, we have neglected nucleus spin density introduced into the 13/2, 3/2 ), /3/2, -3/2 ) states by first order spin-orbit mixing with excited, nonpopulated states of e, symmetry.
VII. THE LIGAND CONTRIBUTION TO THE PSEUDOCONTACT (f2$ COMPLEXES SHIFT FOR

In order to illustrate how the ligand contribution to the pseudocontact shift may be calculated, we turn next to the case of octahedral (t2J5 complexes. The states for the hole electron are given by Eq. [48]. Note that for the (t2$ configuration the doublet (J = l/2) is the ground level, the quartet (J = 3/2) is the excited level, and il is negative in sign. The molecular orbitals representing the T2 basis states include ligand centered orbitals, of which ligand p orbitals will be of particular interest. We consider ligand nuclei lying on the z axis of the complex. Denoting the p-orbital ligand contribution to the molecular orbital representing the 1JM) state as $&L), we obtain (in terms of p orbitals centered at the ligand) the expressions in Eq. [51] P
*l/2,

112~~)

= f -$[P. T ipylT
=

$312,

k l/Z(=)

j&k

f &Jr WI

+3/2,

3/2CL)

j&P, f i&l+.

In Eq. [51] f gives the admixture of the ligand orbital in the t,, molecular orbitals (representing 1T,( + 1) ) states) and the superscripts + and - denote the spin functions, corresponding to eigenvalues S, = + l/2, S, = -l/2, respectively; the ligand px and p,, orbitals are directed, respectively, parallel to the x and y axes of the complex.

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

299

The evaluation of the ligand pseudocontact contribution requires the matrix elements of the operators A, and A, (defined in Eq. [4b] and [4c]) between the states given in Eq. [51], which matrix element can be written as: <A,+&.h,.r~~~ = AJM,wPW~)~~ 2. The nonzero matrix elements, A,,,, J,M, are tabulated below: J 1/L l/L l/L 1/L 1/k l/2. M l/2 - l/2 l/2 112 -l/2 - l/2 J l/L l/L 312, 312, 312, 312, M l/2 -l/2 312 112 -l/2 - 312 A JM, w(z) - 16/15 16/15 0 S&l5 8,/i/l 5 0 A,, , J w 6~) 0 0 -i&/15 0 -iJ$l5 0 -&,I5 0 $15

[=I
A JM, J w(x)

Using Eq. [12] the ligand contribution by the equation L\HL/H, = _ 16 2f~~~i)i,

to the pseudocontact shift is found to be given [ 1 + , (1 -c-(3/2) -: )] ,/[I +&?-(312)v] [53]

where, in Eq. [53], (u-~), is the average value of Y3 for the ligand p orbital, V = IJl//<T, and AHL is the ligand contribution to the pseudocontact shift.4 The rather surprising result predicted by Eq. [53], that a pseudocontact shift occurs even for octahedrally symmetric complexes, can be justified by the following arguments: First, the symmetry at the ligand nucleus site is not octahedral but tetragonal; second, the extensive mixing of spin and orbital states arising from spin orbit coupling is responsible, even in the presence of an isotropic g tensor, for the pseudocontact shift. Accordingly, the 13C and 14N shifts observed for Fe(CN)i3 in solution are not necessarily pure contact shifts. Values for the observed 14N NMR shifts (6) yield, when substituted into Eq. [53], a value of 6.10 x 1O+23 cmm3 for f (rm3),, if it is also assumed that V = 2.0. If to be 21.1 x 1024cn-3 for a nitrogen p orbital (20), we obtain we take (rd3), J = 0.029, a not unreasonable value. We may conclude in this case that the ligand 2 pseudocontact contribution is not negligible. However, for the much larger 13C shifts (21), both the magnitude of the observed shift and the sign, which is opposite to that predicted by Eq. [53], indicate that the Fermi contact mechanism is probably dominant in the 13C shifts for Fe(CN)i3 and Mn(CN)i4. We do not suggest that the above results should be interpreted to show that a large pseudocontact shift will always result from occupation of ligand p orbitals. A similar treatment of the octahedral (t2J3 configuration (for which the ground state manifold is not split in zeroth order by spin-orbit coupling) shows no contribution from ligand y-orbitals. Presumably the ligand contribution will be most important for T2 or T, terms, where there is extensive mixing of pure spin and orbital states via a spin-orbit coupling mechanism. 4 For strictly octahedral symmetry, the metal ion contribution to the pseudocontactshift would vanish. ?I

300

KURLAND AND MCGARVEY VIII. SUMMARY

Several important conclusions come from the preceding analysis. The equations presently employed to calculate contact and pseudocontact shifts (Eq. [l] and [2]) are valid only for complexes in which the zero field splitting is small and in which there are no low lying excited orbital states. For large zero field interactions, the contact equation is still approximately correct for the largest zero field splittings that one would normally expect to encounter; however, the pseudocontact equation must be altered to include a zero field term, which can be sizeable for systems with S > 3/2. For complexes with Tl or T, ground states, the equations are greatly different from Eq. [l] and [2]. The main difference comes from the presenceof a Zeeman mixing of excited states into the ground state which produces a large shift even if the excited state is populated to only a few percent. This Zeeman mixing term leads to temperature dependences much different than the T- predicted by the usual equations. Further, it has been shown that ligand nuclei with p orbitals that can combine with the metal d orbitals (to form molecular orbitals) can experience,in some cases,large pseudocontact shifts even in high symmetry and for small populations of unpaired spin in the p orbital.
IX. ACKNOWLEDGMENT One of us (R.J.K.) gratefully acknowledges stimulating discussions on this subject with

Dr. D. G.

Davis.
APPENDIX:
Derivation

of the Density Matrix Formulae

Consider the operator exp (A + B), where the operators A and B do not commute, and the matrix elementsof B are much less than those of A. Let F(n) = exp [L(A + B)] and note that F (A = 0) = 1, the identity operator. Define G(A) = (exp [-LA]) F (I) note that G(0) = 1. The differential equation Eq. [Al] which is obtained for G(I), dG/di = exp [-AA] B F (2) [AlI can be converted to the integral equation, Eq. [A2]: G(A) = 1 + / [exp(-A A)]BF(A . )dA
0

L-421

From Eq. [A21 one infers that the zeroth order approximation to G(I) is G,(A) = 1 and, thus, the zeroth order approximation to F(A) is F,(l) = exp (AA) Go(l) = exp (AA). Similarly, the first order expression for G(n) and F(l) can be written as G,(i) = 1 + / exp (- A B F,(n A) )dh = 1 + / exp (- M )Bexp (A A)d/Z
0 0

and
F,(l) = exp(lA) G,(n).

[A31

The matrix elements, (ml exp (A + B)lr m), may be evaluated to first order in the matrix elements (l-n]BIT m) by putting /? = 1 in Eq. [A31 and by taking the matrix elementsof F, (1. = 1) for a set of basis functions which are eigenfunctions of A (i.e., for ]m) such that Aim) = ArIm)). For our application we let A = -Xo/kT and

CALCULATION

OF CONTACT

TERMS FOR COMPLEXES

301

B = -2JkT and use as basis functions Im), eigenfunctions of A?,, (i.e., such that The nz). Xo,llY~) = &,jm)), to evaluate P~,,,~,~= (ml exp [-(X0 + Xz)/kT]IT (where Ed # Q.) are given in Eq. [lo]. first order expressions for pm, rm and P~~,~,,,,
REFERENCES 1. D. R. EATON AND W. D. PHILLIPS, Advances in Magnetic Resonance (J. S. Waugh, Ed.), Vol. 1, p. 119. Academic Press, New York (1965). 2. H. M. MCCONNELL AND D. B. CHESNUT, J. Chem. Whys. 28, 107 (1958). 3. H. M. MCCONNELL AND R. E. ROBI~RTSON,ibid., 27, 1361 (1958). 4. J. P. JESSON,ibid., 47, 579, 582 (1967). 5. R. M. GOLDING, Mol. Phys. 8, 561 (1964). 6. B. R. MCGARVEY AND J. PEARLMAN, J. Msg. Res. 1, 178 (1969). 7. K. W. H. STEVENS,Proc. Roy. Sot. A219, 542 (1953). 8. W. MARSHALL, Paramagnetic Resonance (W. Low, Ed.), Vol. 1, p. 347. Academic Press, New York (1963). 9. C. P. SLICHTER, Principles of Magnetic Resonance, p. 127. Harper and Row, New York (1963). 10. R. KARPLUS AND J. SCHWINGER, Phys. Rev. 73, 1020 (1948). II. J. H. VAN VLECK, Theory of Electric and Magnetic Susceptibilities. Oxford Univ. Press, London/New York (1933). 12. B. R. MCGARVEY, Transition Metal Chemistry (R. L. Carlin, Ed.), Vol. 3, p. 89. Marcel Dekker, New York (1966). 13. J. OWEN AND I. M. WARD, see K. D. Bowers and J. Owen, Rept. Prog. Phys. 18, 304 (1955). 14. H. S. JARRETT, J. Chem. Phys. 27, 1298 (1957). 15. D. R. EATON, J. Amer. Chem. Sot. 87, 3097 (1965). 16. P. L. RICHARDS, W. S. CAUGHEY, H. EBERSPAECHER, FEBER AND M. MALLEY, J. Chem. Phys. G. 47, 1187 (1967). 17. R. J. KURLAND, R. G. LITTLE, D. G. DAVIS AND C. Ho (to be published). 18. J. S. GRIFFITH, The Theory of Transition Metal Ions, Chap. 9. Cambridge Univ. Press, New York (1962). 19. J. R. BOLTON, Radical Ions (E. T. Kaiser and L. Kevan, Eds.), p. 4. Wiley (Interscience), New York (1968). 20. A. H. MAKI AND B. R. MCGARVEY, J. Gem. Phys. 29, 35 (1958). 21. D. G. DAVIS AND R. J. KURLAND, ibid., 46, 388 (1967). 22. W. D. HORROCKS, JR., R. H. FISCHER, J. R. HUTCHISON AND G. N. LAMAR, J. Amer. Chem. Sot. 88, 2436 (1966).

You might also like