260

[CONTRIBUTIOX FROM

HARRY GRAY B. AND C. J. BALLHAUSEN

THE

VOI. 86

DEPARTMENT CHEMISTRY, OF COLUMBIA UNIVERSITY, XEW YORK2i, N. Y.]

A Molecular Orbital Theory for Square P a a Metal Complexes lnr

BY HARRY GRAY B. AND C. J. BALLHAUSEN'
RECEIVED JULY 25, 1962
The bonding in square planar metal complexes is described in terms of molewlar orbitals. The relative single electron molecular orbital energies are estimated for the different types of n-electron systems. Both the d-d and charge transfer spectra of the halide and cyanide complexes of NiZf, Pd2+,P t z +and Au3+ are discussed in terms of the derived molecular orbital scheme. The magnetic susceptibilities of diamagnetic d8-metal complexes are considered, and it is concluded that a substantial "ring current" exists in KaNi(CN),.H20.

Introduction The metal ions which form square planar complexes with simple ligands have a d8-electronic configuration. The five degenerate d-orbitals of the uncomplexed metal ion split into four different levels in a square planar complex. Thus three orbital parameters are needed to describe the ligand field d-splittings. Various efforts have been made to evaluate the dorbital splittings in square planar metal complexes, and to assign the observed d-d spectral The most complete calculation has been made for Pt2+complexes by Fenske, Martin and Ruedenberg,B and considerable confidence may be placed in their ordering of the d-orbitals. However, for other complexes the dordering may depend on the metal ion and the type of ligand under consideration. It is the purpose of this paper to develop a molecular orbital theory for square planar metal complexes. Both the spectral and magnetic properties of typical square planar complexes of Ni2+, Pd2+, Pt2+and Au3+ will be considered in order to arrive a t consistent values for the molecular orbital energies. However, in contrast to previous workers, effort will be concentrated on the assignments of the charge transfer bands of representative square planar complexes. Molecular Orbitals for Square Planar Complexes.Figure 1 shows a square planar complex in a coordinate system with the central atom a t the origin, and the four ligands along the x- and y-axes. The orbital transformation scheme in the DQh symmetry is given in Table I.
ORBITAL Representation
a18
a8 2

The single electron molecular orbitals are approximated as

*(m.o.) = Cx(,)(@)(metal)

+ C(L)j@(ligand)

(1)

where the C's are subject to the usual normalization and orthogonality restrictions. The pure u-orbitals are
*[a~,(o)l
*[blg(Q)I

= cM(l)@[fidrpl =

CM*[%dz'-

+ CMW@ + 1)sI + [(n + + + r2] + + -4

CL@[1/2(Ul
Q2

u 3
Q3

Q4)l

(2)

cL@[1/2(Ul

QZ

(3)

The form of the a-orbitals depends on the type of ligand under consideration. Square planar complexes in which the ligands themselves have no n-orbital system (Cl-, Br-, H20, "3) will be case I ; complexes in which the ligands have a n-system (CN-), and thus both a-bonding (nb) and n-antibonding (n*)ligand orbitals must be considered, will be case 2. Case 1.-The pure a-orbitals are
*![aZdT)l
*[@u(T)l

CM@[(n

+
=

@['/Z(Tlh 1)pzl

cL@[1/2(rly

*[bZg(T)] = CM@[fidz,l
*[bzu(~)I

CL@[l/Z(TIh

+ + + + rZv + r4"jl + + T?h T3h Tlh)]

T~~

(4)

(5)
(6) (7)
(8)

XZh

T3h

T4h)l

@['/~(XLV

xzv

TBV

T~Y)]

*r[e8(T)j = ~M@[nd,,l
*rr[e,(?r)~ =
cMo[~~,J

cL@

[.\/z
1

(mr

- my)

cL@

1 [q - inv)] (n?,

(9)

TABLE I TRAXSFORMATION I N SCHEME

Metal orbitals

Dih

SYMMETRY

The mixed u- and n-orbitals are

d2, s

l/Z(Ql

..
P Z

'/2(Tlh 1/2(X1,y2

azu
blZ3
b2g

dZ2dz,

1/2(Ul '/z(Tlh
'/Z(Tl.

bau

..

+ + + + + f + + + ++ - + u2
63
Q4)

Ligand orbitals

T2h

T3h
T3"

X4h)
T 4 T )

* 1 1 [ e d u , ~ ) 1 C M @ [ ( ~ I)P,l =
CL(l)@

TZY

UZ

Q3

84)

Tzh
TTZ"

T3h
T39

T4h)
T4")

[ z
1
(uz

u4)]

CLW@

[ z
1

Case 2.-The

pure a-orbitals are

(1) Permanent address: Institute for Physical Chemistry, Cniversity of Copenhagen, Denmark. (2) (a) J. Chatt, G. A. Gamlen and L. E. Orgel, J . Chem. SOC., (1958); 486 (b) G. Maki, J . Chem. P h y s . , 28, 651 (1958); 29, 162 (1958); 29, 1129 (1958). (3) C. J. Ballhausen and A. D. Liehr, J . A m . Chem. Soc., '71, 538 (1959). (4) M. I. Ban, A d a Chin.Acad. Sci. Hung., 19,459 (1959). (5) S. Kida, J. Fujita, K. Xakamoto and R. Tsuchida, Bull. Chem. SOC. (Japan), 81, 79 (1958). (6) R. F. Fenske, D. S. hlartin and K. Ruedenberg, Inorg. Chem., 1, 441 (1962). (7) J. R. Perumareddi, A. D. Liehr and A. W. Adamson, J . A m . Chein. Sac., 86, 249 (1963). (8) J. Ferguson, J . Chem. Phys., 34, 611 (1961).

Feb. 5 , 1963
2

SQUARE PLANAR METAL COMPLEXES

261
MOLECULAR 0 R B I TA LS LIGAND 0 RBI TA LS

t
I

O R B I TALS

METAL

1%.
A

__

--y
E.

/ /
X

I I

Fig. l.--B molecular orbital coordinate system for a square planar metal complex.

The mixed u- and a-orbitals are

blg (ub), a,g(cb),e,(ub)

t - 4

Fig. 2.-Molecular orbital energy level scheme for square planar metal complexes in which there is no intra-ligand r-orbital system (case 1).

The following general rules were adhered to in estimating the relative energies of the single electron molecular orbitals: (I) The order of the coulomb energies is taken to be u (ligand), ab(ligand),nd(metal), (n 1)s (metal), n*(ligand), (n l)p (metal). (2) The amount of mixing of atomic orbitals in the molecular orbitals is roughly proportional to atomic orbital overlap and inversely proportional to their coulomb energy difference. (3) Other things being approximately equal, a-bonding molecular orbitals are more stable than a-bonding molecular orbitals, and a-antibonding molecular orbitals correspondingly less stable than T antibonding molecular orbitals. (4) Interactions among the ligands themselves are expressed in terms of a ligand-ligand exchange integral p : the sign of this interaction for each molecular orbital is another factor in the final relative ordering. ( 5 ) The relative molecular orbital ordering is considered final only if it is fully consistent with the available experimental results: exact differences in the single electron molecular orbital levels can only be obtained from experiment. The general molecular orbital energy level schemes arrived a t for square planar complexes are given in Fig. 2 (case 1) and Fig. 3 (case 2). Group molecular orbital overlap integrals (for Ni(CN)d2-) and ligand exchange interactions are summarized in Table 11. Almost all square planar complexes with simple ligands are diamagnetic and contain a metal ion with the de-electronic configuration. Thus the ground state is lAIg. The lowest energy excited states will be described separately for case l and case 2. Case 1.-There are three spin-allowed d-d type transitions, corresponding to the one electron transitions b*,(a*) -+ blg(a*) ('AIg + I A z g ) , al,(a*) +blg(a*) .

('Alg + IBlg) and e g ( T * ) + blg(a*) (IAlg+ lEg). All . . . these are parity forbidden as electric dipole transitions. Two allowed charge transfer transitions may be anticipated, corresponding to the one-electron transitions bzu(n) + b ~ , ( u * )(lAlg + ' A d and e,(+) + blg(a*) .
TABLE I1 GROUP OVERLAP ISTEGRALS LIGAND-LIGAND AND INTERACTIONS FOR SQUARE PLANAR METAL COMPLEXES
hlolecular
orbital
*'alp(.)

Metal orbital [?!d,Zl

*[big(u)l

[(n 1 ) ~ l [ ' / z ( ~ i [ n d z 2- yzl fl/z(ui

[L/a(ui

e[uzg(T)]
*Iazu(s)l
*bg(~)l

....

['/Z(Tlh

I(n

+ l)Pzl

['/2(7rlY

[ndzvl

I1/2(T1h

P(br,(n)]

....
[nd,g)

n4h) 11/z(7riv
Tar)

+ + + .a)] + + + u4)l - + .. f s2h + + *4h) 1 + + + . . . -2P(h,h) T"1 4) 0.464 +2@(v,v) - arh + sah 1 0.166 +2,9(h,h) + ray UP

Ligand orbitals
uz uz ua ua
c a

ua)]

Ligandligand G,ja interaction 0.161 ...... .895 ... .. . ,279 , . , .

T8h

82"

TliV

TZV

...

-2P(v,v)

*I[eg(r)l
*~r[e~(s)l

L 1 i4 2

- m")]

0.117
,117

., . ., .

Indyll

[,/i ( m y1

m.11

..., . .

The group overlap integrals are for Ni(CNA)*-. The atomic wave functions are: 3d and 4s radial functions for nickel from ref. 16; 2s carbon radial function from ref. 17; 2p carbon radial function from ref. 18; the 4p nickel radial function is estimated as 9(4p) = Rsp(l.40), where RsP (1.40) is a Slater orbital with exponent 1.40. The Ni-C distance is taken to be 1.90 A., averaged from the data given in ref. 21.

262
METAL ORBITALS MOLECULAR CRB ITALS

HARRY GRAY B. AND C. J. BALLHAUSEN

LIGAND

Yol. 85

0 R B IT A L S

e ( Q *,T

*I

TABLE I11 ORBITALAXD INTERELECTROSIC-REPULSION ENERGIES SOME OF EXCITED STATES O F ISTEREST FOR A4Nnd8 SQUARE PL.4NAR COMPLEX Slater-Condon Term designation Orbital energyn energy"-*
A.
Singlet terms d-d transitions 'AI, (ground state) 'A?, AI 'BiS A I f Ar :Eg Ai A i f A: 'BI, . . . AE(iB8") 'AiU AE('BIu) Ar 'E, A E ( 1 B i d f Az AI JBig Ai f A i gEg Ai f Aa f A:

...

Charge transfer transitions B. E . Triplet terms d-d transitions

+ Aa
-

-3534 4Fn 3F1

...

- 15F4 - 20Fc

Referred to the ground state as zero. For the approximations used t o estimate the Slater-Condon energy for the charge transfer transitions, see ref. 19.

-2QFn -15Fn f 105F4 -12Fn 9Fz

+ 1OOF4
75F4

- 45F4 - 6OF4

b 1q(T , ,a 1q b (cb,e (ub) 1

Fig. 3.-Molecular orbital energy level scheme for square planar metal complexes in which the ligands themselves have a r-orbita: system (case 2:.

(IA?, + lEU). These are transitions from molecular orbitals essentially localized on the ligands to molecular orbitals essentially localized on the metal atom. An examination of the transition moment integrals for these two transitions reveals that, for any reasonable molecular orbitals, the 'Alg ]E, transition will be more intense than 'AIg Case 2.-There are three d-d transitions; 'AIg + 'AZg, lAlg 'Big and 'A1, + lEg, as in case 1. The charge transfer bands of lowest energy are expected to involve transitions from the highest filled metal orbitals to the most stable empty "ligand" molecular orbital, the U ~ ~ ( T * ) The first charge transfer transition is . b?,(T*) U ~ ~ ( T('Alg + lBlU). This transition is or*) bitally forbidden and should have relatively little intensity. The second charge transfer band corresponds to the transition ulg(u*) u2,,(a*) ('AIg + lAzu). The energy of this transition is calculated to be AE('A1, -+ 'BlU) A,, corrected for differences in interelectronicrepulsion energies in the lAzu and 'B1, excited states. The third charge transfer transition is eg(a*) ~ 2 ( T * ) (lAlg + lE,), and is calculated to be A.E(lA1, 7 A2 f A3, again corrected for repulsion difIBl,) ferences in the 'EU and lBIU excited states. A summary of the calculated orbital and interelectronic-repulsion energies of all the excited states of interest for case 2 is given in Table 111. The 'Al, + 'Azu and 'A1, 'EU transitions are allowed, with the 'AI, -+ 'E, transition expected to have considerably greater intensity. Thus a three-band charge transfer spectrum, with an intensity order IAl, + 'E, > 'AIg * 'Azu > 'Al, + 'Blu, should be typical of square planar complexes with case 2 type ligands. Spectral Properties of Square Planar Metal Complexes. A. Halide Complexes (Case 1).-The com-

plexes PdXd2-, PtX42- and AuX4- (X- = halide) are square planar and diamagnetic. The spectra of solid samples of KZPdC14 and KzPdBr4have been measured by Harris, Livingstone and R e e ~ e and by Yamada.12 ,~ Three bands are observed which niay be assigned to d-d transitions. In solution, PdC14?- and PdBrd2- show two charge transfer bands. The interpretation of the spectrum of PtC1d2- is complicated by the presence of one or more "spin-forbidden" bands. However, PtCldz- seems to exhibit the three spin-allowed d-d bands, and one charge transfer band. The spectra of the AuC14- and AuBr4- complexes clearly show the two charge transfer bands.1 Assignments of the spectra of these halide complexes, in terms of the derived molecular orbital level scheme for case I, are given in TabIe IV. The values of the single electron parameters AI, A2 and A3 also are calculated and are given in Table V for comparison purposes. B. Cyanide Complexes (Case 2).-The spectra of the square planar Ni(CN)d2-, Pd(CN)4,- and Pt(CN)42- complexes are given in Table IV. Each complex exhibits the expected three charge transfer bands, but the d-d bands for the most part are obscured by these charge transfer bands. For Ni(CN)4?-, there are two shoulders on the tail of the first charge transfer band. These shoulders probably represent a t least two d-d transitions. The charge transfer spectrum of Au(CN)4- appears a t higher energies than for Pt(CN)d2-. Thus only one of the expected three charge n-. transfer bandsis seen below 50,000 a . ' The significance of this is discussed later. Assignments of the spectra and the calculated values of A,, A2 and A3 for the cyanide complexes are given in Tables IV and V, respectively. Magnetic Properties of Square Planar Metal Complexes.-The magnetic susceptibility of the diamagnetic d8 square planar metal complexes is given by
~

where -

x = '/3(3X.L + XII)
xL .
X A

(22)
(23)

XII = X A

In eq. 23 and 24, X A is the atomic contribution for the molecule in question (the sum of the Pascal constants), Xring represents the expected diamagnetic contribution of the metal-ligand a-molecular orbital system (analogous to the ring diamagnetism in benzene) and

+ + XHF . ( I I ) + XnW

AH F ( ) ' I

(24)

(9) C. hl. Harris, S. E. Livingstone and I. H . Reece, J. Clievi. Soc., 1505

(1959).

(10) A. K. Gangopadhayay and A. Chakravorty, J . Chcm. Phys., 3 6 , 2 2 0 6

(1961).

Feb. 5, 1963

SQUARE

PLANAR METAL COMPLEXES

263

For the d8 complexes under $0 and excited state +i. TABLEV I consideration, the ground state is lAlg, and eq. 25 and SPECTRAL PROPERTIES SQUARE OF PLANAR METALCOMPLEXES
YmX,

Reference to exptl.
c -

26 reduce t o

Complex (sample) PdClet- (solid K2PdClr)

cm. -1

Assignment

work
9,12,20

Case 1
16,700 21,500 23,300 36,000 44,900 16,000 20,000 26,0004 30,100 40,500 17,700 21,000 25,500 30,200 (37,YOO) 46,000 16,700 19,700 24,300 28,200 37,300 31,800 44,200 26,300 39,400 (22,500) (30,500) 32,300 35,200 37,600 41,600 45,400 47,200 35,720 (38,680) 39,180 41,320 30,960 37,850 46,080

(aq. s o h , excess C1-) PdBrcr- (solid KrPdBrc)

(aq. soln., excess Br-)

'Aig + 1Ak 'AI, * 'Big lAip"Eig 12,000 'Aig -+ 'A-

.. . . .. .. .... .... .... ....

30,000 'Alg * 'Eu

'AI, 'A?, 'AI, 'AI, 'AI, 11, 11
+

The magnetic susceptibility of a single crystal of KzNi(CN)4.H20 has been measured by Rogers." The values he obtained were
9

'Atp

PtC1,S- (as. soh.)

PtBrdt- (aq. soln., 1 MBr-)

AuCIc- (aq. soln.) AuBrc- (as. soln.)

10,400 30,400 2.6 15 59 64 250 9,580 5 15 100 120 7000 4,570 29,500 5,010 38,900 2 250 700 4,200 10,600 1,200 7,200 9,000 1,590 26,000 29,500 1,850 51 331 2,400

+ +

'Big

L, E

XI1 = -127.8 X 10-B (c.g.s.) X I = -146 X 10-6 (c.g.s.)

'AtU
lEu

'Aig* 'Aig -+ 'Big

'Aig+ 'E,

'Ai, lAzg

2a

b
*Aig* 'A?, 'AI. * 'Az, 'Atg 'AI, + 'Big 'AI, + ', E lAig + 'A'Aig+ 'As,, 'Aig -C 'Eu 'Aig*lAtu 'Ais + 'E,
'Aig * 'Ai, 'AI, + 'Big 'A * 'BIU 'Ai, * 'Aq, 'AI, 'Eu lAill -+ IBi, 'Aip 'Azu 'Ais + 'E, 'Ai, 'Bin lAig + 'E, 'Aig * 'E, ' A t , + 'Azu 'Ai, * 'Ai, 'Ai, * 'Big
+ + +

This work

10 10

Case 2
Ni(CN)r'- (aq. s o h ) This work

Pd(CN)ez- (aq. soln.)

20

P t ( C N ) 2 - (aq. s o h )

Au(CN)a- (aq. s o h )

22

'Aig

.-c 'BI,

this band is found a t 21,700 crn.-l in the K2PdBrd.2H20 crystal; see ref. 12. bPossibly a "spin-forbidden" charge transfer transition.
TABLE V METAL&ORBITAL ENERGIES FOR SELECTED SQUARE COMPLEXES
PLANAR

a,,-

yt

1'
d m a,, Orbital energy differences (cm.-l) for Fz = 10F4 = 700 cm-1
Complex ion
AI
Ar AI

PdC1,'PdBr42PtC142 PtBraaAuC14AuBrrNi( CN)4aPd(CN)raPt( CN)daAu( CN)da The value of

19,150 6,200 18,450 5,400 23,450 5,900 22,150 6,000 >20,000 .... >20,000 .... 24,950 9,900 >30,000 10,800 >30,000 12,600 33,410 10,620 4 for KsPdBr4.2H10.

1450 5650 (1350)" 4350 3550

....
....

650 50 -4140

....

Subtracting eq. 24 and eq. 23, the unknown XA is eliminated, yielding XI - Xi1 - ( X H . F . ( I - Xti.~.tli)) -Xr,, ) = (29) Using the experimental values for the AE's, XH.F.(L) = 26 X and XH.F.(II) 142 X = are calculated; this gives X r i n g = -98 X loF6. Discussion The metal orbitals involved in u-bonding in square planar complexes are the nd,~, ndxi-p, (n l)s, (n (n l ) P x l)p, and (n l)py. The ndX2-~2,(n lb, and (n l)p, orbitals account for most of the u-bonding, judging from the values of the overlap integrals in Table 11, and the ndz2 makes only a minor contribution. The most important a-molecular orbital is the uzu,consisting of the (n l)p, metal orbital and a combination of the four ligand a,-orbitals. This gives a very stable r-bonding orbital (with only a single node in the molecular plane) which may be called the "ring" 7-orbital. All the other pure n-molecular orbitals have an equal number of electrons in their bonding and antibonding levels. The tendency of nd8-metal ions to form square planar complexes increases in the order NiZT< Pd2f < Pt2+. Two features of the molecular orbital bonding scheme are consistent with this observation. First, the availability of the nd+ya metal orbital for u-bonding undoubtedly increases in going from Ni2+ to Pt2+ (Xc 4 is tetrahedral; PdC1d2- and PtC142- are square 1' planar). Second, the square planar configuration is stabilized by the ring a-bonding, which is expected to increase from Ni2+to Pt2+. The addition of a fifth group above the square plane drastically decreases the ring Xbonding, by tying up the ( n l)p, orbital in u-bonding. The assignments of the electronic spectra of the square planar halide and cyanide complexes of Ni2+, Pd2+, Pt2+ and Au3+ are given in Table IV. The halide complexes will be considered first. The spectra of K2PdC14and K2PdBr4 show three bands which may be assigned to d-d transitions. In the molecular orbital energy level scheme for case 1 ligands, the order of the single electron molecular orbitals is reasonably expected to be eg(7r*), ulg(u*), bzg(a*) and bl,(u*). Thus the bands are assigned 'AIg + lAzg, lAlg + 'B1, and . lAlg + lE, in order of increasing energy. The energy order of the single electron molecular orbitals deduced here is consistent with the results of a "complete" electrostatic calculation of square planar Pt2+ complexes performed by Fenske, et aL6 The spectra of PtC142- and PtBr2- are slightly more complicated; the first band is quite weak and is assigned as the first spin-forbidden transition, 'A1, + 3A2g. The spin-allowed transitions are assigned in the same way as for the PdXh2- complexes. The spectra of single crystals of K2PtC14 using polarized light show that the 25,500 ern.-' band occurs in x,y and the 30,200 cm.-' band occurs in z polarization.12 These polariza-

+ +

+ + + +

where g is the degeneracy of the ground state and , AE0.i is the energy separation between the ground state

(11) M. T.Rogers, J. A m . Chem. SOC.. 69, 1506 (1947). (12) S. Yamada, ibid., 73, 1182 (1951).

2G-L

HARRY GRAY B. AND C. J. BALLHAUSEN

VOl. 85

other hand, the first charge transfer band for Au(CN4tions are expected for the assignments given here, assuming a vibronic intensity giving mechanism. Chatt, occurs a t higher energy than for Pt(CN)42-, expected only if the charge transfer for these cyanide complexes Gamlen and OrgelZapreviously assigned the spectrum is of the metal + ligand type. of P t C L - somewhat differently (17,700 cm.-l, 'AIg -+ The charge transfer band systems observed for the 3A2g;21,000 crn.-', 'Al, + 3Eg; 25,500 cm.-l, 'Alg halide (case 1) and cyanide (case 2) complexes should ' A z a ;30,200 cm.-', 'AIg + lEg). However, the 25,500 serve as a guide in assigning the charge transfer in other cm.-l and the 30,200 cm.-' bands are both quite sensisquare planar complexes as ligand (n)+ metal (case 1) tive to axial perturbations (both bands shift to the red on changing the solvent from 12 d l HC1 to water).13 or metal + ligand [uzu(n*) (case 2 ) . In the halide com] This is evidence against the assignments of Chatt, et d j Z a plexes the two charge transfer bands are separated by 10,000 to 13,000 cm.-l. Thus it can be estimated that and is consistent with the assignments proposed here. The AuC14- and A4uBr4- complexes do not exhibit the bZu(n) and eu(nb) molecular orbitals are separated any bands which can unambiguously be assigned to d-d by a t least 10,000 cm.-' in a typical square planar comtransitions. In these cases the two charge transfer plex. This is a reasonable value since the b2u orbital bands appear a t lower energies (than for PtXd2- comis non-bonding with a ligand interaction of -2/3(v,v), plexes) and mask the weaker d-d bands. and the e, orbital is w-bonding.?Z The only bands which can clearly be identified in the The three charge transfer bands due to metal cyanide complexes are due to charge transfer transitions. ligand [uzU [.*)I transitions and exhibited by the cyanide A rough calculation may be performed to show that the complexes are much more closely spaced. The two charge transfer bands in Ni(CN)i,- appear a t about orbitally allowed metal ligand transitions (ulg(u*) the energy expected for transitions from metal d-orbitals U , ~ ( T * ) , eg(n*) uzu(a*)) are only 2,000-3,000 cm.-' to the ligand uzU(n*) orbital. The ionization potential apart in all the cyanide complexes, since the ulg(u*) of HCN is about 1 e.v.I4 The calculated separation 4 and e g ( x * )metal orbitals are virtually non-bonding and of the single electron nb and x * levels in HCN is 9 e.v.l6 thus very nearly equal in energy in these complexes. Thus these two types of charge transfer show quite Thus the coulomb energy of the x * level of complexed CN- may be estimated a t -5 e.v. The coulomb endifferent properties and should be distinguishable in ergy of the Ni2+ d-orbitals in Ni(CN)4*- may be estiother nda square planar complexes. The ligand (n) to mated a t about - 10 e.v., using the spectroscopic data metal type will have two bands, spaced by about 10,000 compiled by Moore,*4 and assuming that the net charge cm.-', the second more intense than the first. The on the complexed Ni2+ is actually about +1. This metal + ligand [usu(n*)] type will have three closely leaves 5 e.v., or about 40,000 cm.-l, as the difference in spaced bands with intensities bZg(n*) uzU(n*)[small], + energy of the single electron d (Xi2+) and n*(CN-) ul,(u*) + U ~ ~ ( T[intermediate] and eg(n*) + azU(n*) *) [large]. The exact positions of these bands give of orbitals. Since the uZu(n*) level is stabilized by the course the best clues to the relative ordering of the dempty 4p, metal orbital (also by +2p(v,v,)) and since orbital energy levels. In all probability there will be the 3dxy,3dxz,3d,, and 3dz2orbitals are actually weakly antibonding, the appearance of the three charge transcases in which the bZg(n*) level is fairly close to alg(u*) fer bands in Ni(CN)d*- between 32,900 cm.-' and 37,500 and eg(n*) (A, small), and thus only the two orbitally cm.-l is in accord with theory. allowed bands will be observed. In any event the spacing of the two orbitally allowed charge transfer The separation A1 for the cyanide complexes is by far the largest of the three orbital parameters, and illusbands will easily distinguish between ligand -+ metal trates that the d,~-,~ much more strongly antibonding is and metal -+ ligand type transitions for the nds square than any of the other d-orbitals. The separation A3 planar complexes. decreases regularly in going from Xi2+ to Pt2+ (for The "ring" diamagnetism apparently observed for Ni(CN)42- and Pd(CN)J*-, the d,? is more unstable K*Ni(CN)4.HsO may be considered as experimental than d,,, drs; for Pt(CN)4?-, the dZ? more stable than is evidence of electron delocalization through the cyanide system. The electrons in the uZu(nh)orbital, in pard,,, dyJ. This is compatible with the idea that the axial interaction of the water molecules in water soluticular, would be expected to generate considerable tion decreases in going from Ni2+ to Pt2+. The two "ring current." components observed for the intense charge transfer Experimental.-The complex K2Ni(CN)4.H20 was band in Pt(CN)4*- are consistent with its assignment prepared as described in the 1iteratu1-e.~~ The elecas 'Al, + lE,; the 'E, excited state is expected to be tronic spectral measurements were made using a Cary split by second-order spin-orbit effects. 1 spectrophotometer. 4 The assignments given in this paper for the cyanide Acknowledgments.-The authors thank Dr. Andrew complexes differ considerably with assignments given D. Liehr for making available a preprint of his paper by other investigator^.^^^^^ The reason for this diswith Dr. Adamson and Dr. Perumareddi, and for stimuagreement is found in our "band assignment philosophy"-ussign only the bunds which can clearly be iden(16) J. W. Richardson, W. C. Nieuwpoort, R. R. Powell and W. F. Edgell, ibid., 36, 1057 (1962). t$ed as to energy and intensity. In the case of the (17) D. R. Hartree, "The Calculation of Atomic Structures," John Wiley cyanide complexes these bands are almost certainly due and Sons, Inc., New York, N. Y., 1957, pp. 169-171. to charge transfer transitions, as judged by their large (18) C. Zener, Phys. Rev., 36, 51 (1930). intensities. (19) H. B. Gray and C. R. Hare, Inovg. Chem., 1, 363 (1962). (20) C. K. Jp'rrgensen, "Absorption Spectra of Complexes of Heavy A careful comparison of the charge transfer bands for hfetals," Report t o U.S. Army, Frankfurt am Main, October, 1958. the PtC1d2-, AuCld-, Pt(CN)h2- and Au(CN)4- com(21) "Interatomic Distances," The Chemical Society, London, 1958, p. M plexes offers further proof of the assignments given 178. here. Thus the first charge transfer band appears a t (22) A. Kiss, J. Csaszar and L. Lehotai, Acta Chim. Acad. Sci. Hung., 14, 228 (1958). lower energy for AuC14- than for PtCL2-, as expected (23) W. C. Fernelius and J. J. Burhage, Inoug. Synlheses, 2, p. 227. for a ligand + metal type electronic transition, since (24) C. E. Moore, "Atomic Energy Levels," U.S. Natl. Bur. Standards Au3+ is a better electron acceptor than Pt2f. On the Circular 467, 1949 and 1952.
+

(1960).

(13) R . G. Pearson, H. B. Gray and F. Basolo,

J. A m . Chem. SOL., 787 82,

(14) F. H. Field and J . L. Franklin, "Electron Impact Phenomena," Academic Press, Inc., New York, Ti. Y., 1957, p. 273. (15) K . Iguchi, J . Chem. Phys., 23, 1983 (1955).

(25) There is another assignment possible for the two charge transfer bands in the square planar halide complexes. The first hand could contain both the b 2 U ( T ) b l g ( u * ) and a,(nb) * bi,(s*) transitions; this would leave b l g ( c * ) transition responsible for the second intense the allowed eu(ub) band.

--

Feb. 5, 1963

HEXACHLOROMETrZLLATES O F TRIVALENT CHROMIUM,

A N G A S E S E AND

IRON

265

lating discussions. Acknowledgment is made to the donors of the Petroleum Research Fund of the Ameri-

can Chemical Society for partial support of this research.

[CONTRIBUTION FROM

THE

WILLIAM ALBERT NOYES LABORATORYCHEMISTRY, OF UNIVERSITY ILLINOIS, OF URBANA, ILLISOIS]

Hexachlorometallates of Trivalent Chromium, Manganese and Iron BY ~ ' I L L I A M HATFIELD, E. ROBERT F.~Y, E. PFLUGER T. S.PIPER C. C. AND
RECEIVED JULY 16, 1962 Compounds containing hexachlorometallate ions of trivalent chromium, manganese, and iron have been pre111) pared by stabilizing the anion in a crystal lattice with the tris-( 1,2-propanediamine)-cobalt( cation. The existence of the complex anions has been proved by isomorphism with salts of known hexachlorometallate(III) ions and the same cation, and by X-ray data obtained from single crystals of [Co(C3Hl0X&][FeC&]. Magnetic data showed that the ground state of these compounds is the expected high-spin state. Dq values of 1318,1754 and 920 cm.-' for [CrCl~l [MnC16]- and [ F ~ C 3-, Q ] 3-, 3 ~ respectively, were determined from spectral data.

Introduction Complex ions including those not found in appreciable concentrations in solution often can be stabilized in a crystal lattice providing that the energetics of formation of the complex ion from solvated species is not too unfavorable relative to the lattice energy. A good guide to a suitable counter ion is the rule that the lattice energy usually is largest for ions of equal and opposite charge. Furthermore, a counter ion of similar size is best since disparate sizes promote solvation. This approach has been used recently by Gill and Nyholm, who have isolated stable compounds containing such ions as the tetrahalonickellate(I1) ion. Similarly, Mori2 has prepared a number of chlorocuprates by stabilizing the complex anions in a lattice with the hexamminechromium( 111) cation. Although there are several references in the older literature to compounds having stoichiometries consistent with hexachlorometallate salts for trivalent first transition series elements, e.$., T & C T C ~ ~ , ~ Cs3FeC~G.H~O,~ and (h:H4)3CrC16,6 properties of the comthe plex anions are unknown. In fact, evidence reported here casts doubt on the existence of any hexachlorometallate ions in these compounds. mre now wish to report the preparation and some physical properties of compounds which do contain hexachlorometallate ions of trivalent chromium, manganese, and iron. Up to now the position of the chloride ion in the neuphelauxetic series has been based on spectral studies of anhydrous chromium(II1) chloride and complexes of second and third transition series ions.6 Studies on octahedral complexes of first transition series ions coordinated only by chloride should provide additional information concerning the ligand field of the chloride ion. Experimental
Tris-( 1,2-propanediamine)-cobalt(III) Hexachlorochromate(III).-A small amount of the pink compound crystallized from a concentrated hydrochloric acid solution (50 ml.) of chromium(111) chloride hexahydrate (5.4 g., 0.020 mole) and tris-(1,2pr0panediamine)-cobalt(II1) chloride (3.8 g., 0.010 mole) after standing for several days. The crystals were collected on a Biichner funnel, washed well with water, then with alcohol 1' and ether, and dried in an oven a t 1 0 for four days. Anal. Calcd. for CoCgH&T6CrC16: C, 19.80; H, 5.53; N, 15.40. Found: C, 19.87; H, 5.61; N, 15.09. Tris-( 1,2-propanediamine)-cobaIt(111) Hexachloromanganate(III).-This compound was precipitated from a hot solution of manganese(I1) sulfate monohydrate (3.4 g., 0.020 mole) and
(1) N. S.Gill and R. S.Nyholm, J . Chem. Soc., 3997 (1959). (2) M. Mori, Bull. Chem. Soc. Japan. 33, 985 (1960). (3) G. Neuman, A n n . , 244, 329 (1888). (4) P. T. Walden. A m . J. Sci., 131 148, 283 (1894); 2. a m r g . u . allgem. Chcm., 1, 332 (1894). (5) H. I. Schlesinger, J . A m . Chem. Soc., 61, 3520 (1929). ( 6 ) C. E. Schaffer and C. K . JWgensen, J . Inoug. Nuclear Chcm., 8, 143 (1958).

tris-( 1,2-propanediamine)-cobalt( 111) chloride (7.6 g., 0.020 mole) in a minimum amount of hydrochloric acid by the addition of small amounts of sodium chlorate. The dark brown compound was collected on a Biichner funnel, washed with alcohol, acetone, and ether, and dried in an evacuated desiccator over potassium hydroxide. The yield was 8.1 g. (75%). Anal. Calcd. for COC&oN&fnCls: C, 19.70; H, 5.50; N, 15.31; C1, 38.72. Found: C, 20.18; H, 5.61; N, 15.24; C1, 38.98. The compound decomposed slo\vly upon standing presumably evolving chlorine. The magnetic moment of a sample that had been stored for three months was 0.3B.M. higher than that of a freshly prepared sample. Tris-( 1,2-propanediamine)-cobalt(III) Hexachloroferrate(II1). ---Isolution of iron(II1) chloride hexahydrate (2.7 g., 0.010 mole) in 5 mi. of hot water was added to a solution of tris-(1,2propanediamine)-cobalt(II1) chloride (3.8 g., 0.010 mole) in 10 ml. of hot water. The solution was evaporated to about half the original volume. A few milliliters of concentrated hydrochloric acid was added; the yellow crystals formed were collected on a Biichner funnel, washed with alcohol, acetone, and ether, and air dried. The compound was recrystallized from hot water with the addition of hydrochloric acid. The yield was 1.8 g. (29%). Anal. Calcd. for CoCgHsoS+jFeClo: C, 19.66; H, 5.50; N, 16.28; C1, 38.69. Found: C, 19.88; H, 5.14; N, 15.46; C1,

38.69, 38.61. Tris-( 1,2-propanediamine)-cobalt( 111) Hexachloroindate( 111).

-This compound was prepared as was the corresponding iron compound, The yield was 87%. Anal. Calcd. for C O C ~ H ~ O N C, ~ C ~ ~H, 4.96; N, ~ 17.76; : 13.81; C1, 34.96. Found: C, 17.68; H, 5.00; X, 13.08; C1, 35.00. Tris-( 1,2-propanediamine)-rhodium( 111) Hexachlorometallates(II1).-Compounds with the tris-( 1,2-propanediamine)-rhodium(II1) cation were prepared by the methods described above. The analytical data for the new compounds are given in Table I . Hexaamminecobalt(II1) Hexachloroferrate(III).--A solution of ferric chloride hexahydrate (1.4g., 0.005 mole) in 10 ml. of hot water was added to a hot solution of hexaammiriecobalt(II1) chloride (0.43g., 0.0016 mole) in 70 ml. of 3.4 N hydrochloric acid. The resulting solution was heated for a few minutes until an orange product crystallized. The crystals were collected on a Buchner funnel, washed with alcohol and ether and dried in an evacuated desiccator over potassium hydroxide. The yield was 0.5 g. (727,). Anal. Calcd. for CoXcHlaFeCl6: N, 19.54; H, 4.18; C1, 49.55. Found: N, 19.52; H,4.13;C1,48.42. The preparation could not be made a t room temperature because of the insolubility of hexaamminecobalt(II1) chloride in a solution with a chloride ion concentration high enough to facilitate the formation of the hexachloroferrate( 111) ion. Detailed studies of this compound were not undertaken because we experienced difficulties in preparing the corresponding manganese compound. Experiments with Other Cations and Anions.-Proper conditions could not be found for the preparation of hexachloroor ferrate(111) salts of the tris-(2,3-butanediamine)-cobalt(III) tris-( 1,2-cyclohexanediamine)-cobalt(III) cation, An orange product crystallized from a solution of tris-(ethylenediamine)cobalt(II1) chloride and ferric chloride after a few days. The crystals were collected on a Biichner funnel, washed with alcohol and ether, and dried in air. A n d . Calcd. for CoCsH?aSsFeCls.1.5H~O:C, 13.48; H , 5.08; K, 15.74. Found: C, 13.21; H,4.68: N, 15.92.