J. Inorg. Nucl. Chem.. 1961. Vol. 21, pp. 221 to 237. Pergamon PreSs Ltd.

Printed in Northern Ireland

ELECTRONIC ABSORPTION SPECTRA OF ACETYL-

ACETONATO COMPLEXES--I

COMPLEXES WITH TRIVALENT TRANSITION METAL IONS

D. W. BARNUM
Shell Development Co,, Emeryville, Calif.

(Receired 14 March 1961 )

A b s t r a c t - - T h e ultra-violet, visible and near infra-red absorption spectra of the following acetyl-
acetonato complexes are reported: Ti(acac)3, V(acac)3, Cr(acac)3, Mn(acac)s, Fe(acac)~, and Co(acac)s.
In this series o f complexes the *r -,- rr* transition energy increases with increasing number o f d,-
electrons and this is attributed to metal-ligand ~--interaction. In order to explain the " n o r m a l " value
of the ligand field splitting parameter A, it is proposed that the metal-ligand 7r-interaction occurs
between d~-orbitals of the metal and both higher, empty, ~-*-orbitals and lower, filled, rr-orbitals o f the
ligands. Other factors affecting the ~r ---,zr* transition energy are also discussed.
Electron transfer bands, which are observed in the 300-400 m/t region, have been assigned to the
d~ , rr~ transition. The d~ ~ d;, transitions are discussed briefly.
In each complex the assignments are discussed in detail. In the case of the ~ ~ ~* transition in
Cr(acac)s the assignment differs from that made by previous workers.

SINCE 1951 a large number of publications have appeared in the literature concerning
the interpretation of the absorption spectra of transition metal complexes. I~ Because
of the theoretical tool provided by the crystal field theory the greatest success has
been with transitions which occur within the d-orbitals. Very little work has been
done with the intense electron transfer bands and the bands characteristic of the
ligand, mainly because of the difficulty in making sufficiently accurate theoretical
calculations. However, even the simplest calculations are often useful in studying
the absorption spectra of conjugated organic molecules and in the present work
naive LCAO-MO calculations were used to help interpret the spectra of a series of
acetylacetonato complexes of trivalent transition metal ions. The method of calcula-
tion and the numerical results will be presented in a forthcoming paper, czl
The present work is concerned mainly with the ~r--,-~-* transition which is
characteristic of the acetylacetonate ion and with the electron transfer transition
which appears in each of the complexes in the series Ti(acac)a, V(aeac)3, Cr(acac)~,
Mn(acac) 3, Fe(acac)3, and Co(acac)a. For reasons which are discussed below the
electron transfer band is assigned to the d,--~ 7r4 transition and hence involves the
transfer of an electron from the metal to the ligand.

EXPERIMENTAL
With the exception of tris(acetylacetonato) titanium(Ill) all complexes used in this work were
prepared by standard methods described in the literature. Analyses are reported in Table 1.
CtlAKRAVARH~:~' prepared Ti(acac)~ by passing dry ammonia into a suspension of TiCI3 in dry
'a' Recently reviewed by T. M. Dt:NN, Modern Coordination Chemistry (Edited by J. Lr.wts and R. G.
WtLKINS) Chap. 4, Interscience, New York (1960).
,2~ D. W. BARNUM,J. Inorg. Nucl. Chem. In press.
,3~ B. N. CHAKRAVARIt, Naturwissenschaften 45, 286 (1958).
221
222 D . W . BARNUM

TABLE I . - - A N A L Y S t S OF METALCHELATES

Complex Metal (%) Carbon (%) Hydrogen (%)

Sc(CsHTO~), 52"6(52'6)* 6.2(6.2)

Ti(CsHTOI), 14.2(13.9) 51 '9(52'2) 6.1(6.1)
V(C~HvO2)3 51-7(51.7) 6.1(6.1)
Cr(CsHTO,), 14.7(14-7) 52-3(51.6) 6.3(6.1)
Mn(CsH,O,)s 51.3(51'1) 6.0(6.0)
Fe(CsHTO~)s 15'8(15'8) 51"1(51-0) 6.0(6.0)
Co(C~H,O~)s 16"0(16.5) 50.7(50"6) 6.0(5.9)
Zn(C, HTO2)2 42.5(45'6) 5.7(5.4)

*Values required by theory are given in parentheses

TABLE 2 . - - A B S O R P T I O N BANDS* IN ACETYLACETONATO COMPLEXES

3ffi**(m/~); log e
Complex
Ethanol Benzene Chloroform Dioxane

Sc(aca0, 296; 4"55

Ti(acac) s (240); 3.9 (242); 4.0

(265); 4.0 (267); 4-1 (265); 4.0
(280); 4.1 (280); 4.1 (280); 4" 1
(290); 4.1 (290); 4.1 (290); 4.1
308; 4.30 310; 4.32 309; 4'29 309; 4.28
389; 2.84 391 ; 2.92 390; 2.88 ~ 390; 2'89
(470); 2.6 (470); 2.8 (470); 2-8 i (470); 2.8
(550); 3.0 (560); 3.1 (550); 3-1 ] (550); 3-1
593; 3-18 602; 3.27 594; 3.22 i 598; 3-24
(640); 3.0 (650); 3.1 (640); 3.1 iI (642); 3-1

V(acac), 228; 3"93

279; 4'26 283; 4"30 280; 4"28
290; 4"26 292; 4"30 290; 4"28
343; 3"85 348; 3"91 345; 3.89
343; 3'85 348; 3'91 345; 3'89
(390) ; 3"1
(460); 2"4 (445); 2-6 (445); 2,5
(550); 1'6 (540); 1.7 (535); 1-7 ....

Cr(acach 213; 4.34

255; 4.05 255; 4.06 255 ; 4.04
270; 4.02 272; 4.03 273 ; 4.03
(295); 3.8 (297); 3.8 (298); 3.8 (290); 3.8
332; 4-18 337; 4.23 336; 4.20 335; 4"21
(373); 2.69 (374); 2.65
380; 2.63 380; 2.69 382; 2.66 379; 2.66
388; 2.63 389; 2-69 388; 2.65 387:2.65
(410); 2.2 (410); 2.2 (410); 2.2 !
(435); 1"7 (435); 1.7
560; 1-82 559; 1'80 560; 1.82 560; 1.83

* Bands which appear as shoulders or inflexions on the side of a more intense band are enclosed in
parentheses.
 Electronic absorption spectra of acetylacetonato complexes--I 223

TABLE 2. (Contd.)

2m~x (m/0; log

Complex i
i Ethanol Benzene Chloroform Dioxane

Mn(acac)3 272:436 _ m

322; 4.00 318; 3.96

(385); 3.1 (400): 3'0
(445); 2.4
m
(550): 2-18 (550); 2"3
m
1125; 1.93

l-e(acac)3 235; 4.22 m

240; 4.26
272; 4.44 m
273; 4.46 273; 4.45
351; 3"55 353; 3'54 353; 3.52 353; 3-56
431; 3.47 434; 3.53 437; 3-51 434; 3.55
725; 05
975; 0.2

Co(acac)~ 228; 4.60 229; 4"61

256; 4.53 -- 257; 4"54 258; 4"54
(295); 4.0 (295); 4"0 (295); 4"0 (295); 4"0
(304) ; 3"9
323; 3.89 324; 3"92 325; 3"90 323; 3"90
(400); 2.5 (400); 2.5 (400); 2.5 (420); 2"5
595; 2-10 596; 2.15 592; 2.10 597; 2"12
- - (800); 0.5
.-- 1100:028

benzene and acetylacetone. The precipitate of ammonium chloride that was formed was filtered off
and the complex obtained by evaporating the benzene. The following however, is believed to be a
more convenient preparation: In a nitrogen atmosphere 10g (0'10mole) of purified* Eastman
"white label" acetylacetone was dissolved in 40 ml of oxygen-free water by dropwise addition of 6 N
ammonium hydroxide and vigorous stirring. About 20 ml of ammonium hydroxide was used. This
was added dropwise and with stirring to a solution of 50 ml of water and 25.7 g of a 20 per cent solu-
tion of titanium chloride (Fischer Scientific Co.). The titanium trichloride solution first turned red,
then blue, and then a blue precipitate of Ti(acac)3 formed. The precipitate was filtered off with suction,
washed three times with oxygen-free water, sucked dry, and dried in vacuum over magnesium
perchlorate. The dry blue powder was purified in about I g amounts by sublimation in vacuum (oil
pump) at 140-150 C. The analysis is reported in Table I. Ebullioscopic determination of the mole-
cular weight in benzene gave 327 (Calc. for Ti(CsH702)3, 345.23). The complex is readily oxidized by
air.
Absorption spectra were measured by standard methods using a Cary Model 11 or Model 14
recording spectrophotometer. Reagent grade benzene, chloroform and ethanol were used as solvents.
1,4-Dioxane was purified by the method of WEISSBERGERand PROSKAUERJs~ All solvents used for
Ti(acac)a and V(acac)a were freed from dissolved air by flushing with dry nitrogen for 10-15 min just
before use. Also, in measuring the spectra of Ti(acac)3 and V(acac)a glass stoppered absorption cells
with light paths of 0.01, 0'05, 0-10, 0'20 and 0-50 cm were used. These allowed the use of more con-
centrated solutions and thereby reduced any error due to traces of oxidizing agents in the solvent or
air leaks. No oxidation products could be detected in spectra measured in this way.
The observed absorption bands in various solvents are recorded in Table 2.
* Purified by the method of STEINBACHand FIIEISER.t4)
t4, j. E. SIXINllACltand H. FREISER,Analvt. Chem. 25, 881 (1953).
c:,, A. WEXSSaERGERand E. PROSKAUER,Organic Solvents, pp. 139-40. Clarendon Press, Oxford (1935L
224 D . W . BARNUM

5 .0

4.5

4.0

f
3 .5
I
I
I
I
/
3.0
I
I
i \j
z.5 I
• I
I

2.0

e ~
I ,
II
1.5

1.0 I I I I I
1 00000 Z0,000 30,000 40,000 50°000
F R E Q U E N C Y , c m -1
Flo. 1.--Absorptlon spectra.
- - . Ti(acac)a in ethanol
..... Mn(acac)s in chloroform
-- -- -- Acetylacetone in ethanol, 0.0246 M in NaOC2H5
......... Approximate absorption spectrum of Ti(acac)2 + in water, pH 3.4
 Electronic a b s o r p t i o n spectra of a c e t y l a c c t o n a t o c o m p l e x e s - - I 225

ASSIGNMENTS

Tris(acetylacetonato titanium(Ill). The spectrum of Ti(acac) 3 (Fig. 1) shows three

absorption maxima at 308, 389 and 593 robe plus several indistinct shoulders on the
593 m/z band. The 308 m/z band is assigned to the =:3 ~ 7rI transition and the 389 m/t
band to the d, - ,-r.1 transition.
The 593 m u band is assigned to the d , ~ d transition in spite of the fact that it
bears no resemblance to the typical pair of bands observed in other titanium(Ill)
complexes. "~8~ Generally, due to Jahn-Teller distortion from regular octahcdral
symmetry, titanium(Ill) complexes have two overlapping d, ~ d bands which are
about 1,000 to 3,000 cm -1 apart and have an intensity of about 10 I. mole t cm
However, in Ti(acac) z there are at least three shoulders at 635, 555 and 480 m~t and
the intensity at 593 m# is 1,520 I. mole -~ cm ~. Very similar absorption bands have
been observed in tris(benzoylacetonato) titanium(Ill) and in a blue complex with
~-picolinic acid. It has been suggested by LII~11R~'q~ that the shoulders are due to
vibrational line structure.
Complexes of trivalent titanium are sometimes compared with vanadyl complexes
which also have only one 3d-electron on the metal ion, and with copper(ll) complexes,
in which the term splitting is just opposite to that in titanium(Ill) and vanadium(IV).
The strong tetragonal distortion from octahedral symmetry in copper and vanadyl
complexes gives rise to thrce absorption bands rather than one. In both cases the
highcst energy d---~ d transition usually is observed in the 380-400 m/t region. ~s,"~
q-herefore, by analogy, one might argue that the 389 m/z band in Ti(acac):~ is a
d, ~ ~L. transition that arises from strong tctragonal distortion. LX~HR and BAII,-
HAtJSEN~H~ have estimated that Jahn-Teller distortion in titanium(Ill) complexes
should result in spectral splittings of approximately 7,000 cm '~. The 593 and 389 m/~
bands are 8,800 cm ~ apart. Nevertheless, it is believed that such an assignment
would be incorrect. In vanadyl and square planar copper complexes the sourcc of
strong tetragonal distortion is obvious but there is no apparent reason for such a
large distortion in Ti(acac)a other than that predicted by the Jahn-Teller theorem.
It has already been mentioned that this distortion has not been found to exceed about
3.000cm t in other titanium(Ill) complexes. LIEItR himself has stated that the
theoretical estimate of the splitting is probably too h i g h . " Still better evidence
against assigning the 389 mlt band to a d~ -* d.. transition is that the visible absorption
of the 1:2 complex (Fig. 1) consists of two overlapping bands which are typical of
Ti a' complexes (the third band is of such low energy that it should lie in the infra-red
region). In addition, however, there is a band at 396 m/t which cannot be a d , d
transition because these have already been accounted tbr. Upon going to the 3:1
complex the 396 mr, band is shifted slightly to 389 ml~, whereas the portion of the
spectrum due to the d - ~ d transitions changes drastically.
Tris(acetvlacetonato) vanadium(Ill). In V(acac).~ there are two overlapping bands
at 279 and 290 m/~ (Fig. 2) which have almost identical intensities. It is not known

~6~ H. HA:,tlMANN, and H. L. SCrtLAFt-:R, Z. Phys. Chem. 197, 116 (1951).

(7, l-{. HARTMANN,H. L. SCrit.ArER and K. H. HANSEN, Z. Anorg. ('hem. 284, 153 (1956); 41) (19571
I~ C. K. JOR(;ENSEN, Acta Chem. Scand. i l , 73 (1957).
tg~ A. LIEHR. Private communication (1958).
t~0~ R. D. FELTHAM, U.S.A.E.C. Report UCRL-3867 (1957).
I111 A. D. LIFHR and C. J. BAI.I.IIAUSEN, Ann. Phys. 3, 304 (1958).
220 D.W. BARNUM

5.0 3.0

4.5 2.5
%

4.0 2.0
!
r l
I
I
I
3.5
1.5

I
I

8,-1 3.0
-- I I
I
I' I 1.0
i ,
i I
, ' I
/ ,
Z.5 I ' I 0.5
-- /" I

J / i '
I
I
2.0 _ / /"
/ I
#!
I
.
I ~0

/ ,
/ ' I
1.5 -0.5
- I

1.0 i i 1 I -1.0
10,000 20,000 30,000 40,000 50,000
FREQUENCY, cm -I

R o . 2 . - - A b s o r p t i o n spectra.
- - - V(acac)s in e t h a n o l (left h a n d scale)
..... Fc(acac)s in e t h a n o l (left h a n d scale)
-- -- -- Sc(acac)s in e t h a n o l (left h a n d scale)
......... Fe(acac) s in c h l o r o f o r m (right h a n d scale)
 Electronic absorption spectra of acetylacetonato complexes- -i 227

which band is the rra ~ rr4 transition, and indeed, they both may be due to this
transition, the splitting being due to configuration interaction or perhaps to a splitting
of the rr4-orbitals because of a peculiarity of their symmetry properties. These
possibilities are discussed below in more detail. Until this question can be resolved
the 290 mkt band was tentatively assigned to the 7r3 -~ rr4 transition.
The 343 m/~ band is assigned to the de--+ Tr4 transition.
No maxima are observed in the visible region, but there is a shoulder at 550 m/~
which is due to the 3Tlg --~ aT% transition and another shoulder at 460 m/z which is
the 3T1~-~ aTIo(P) transition.
Tris(acet),lacetonato) chromium(lII). Previous investigators have assigned the
332 m # band in Cr(acac)a to the 7ra --~ 7r4 transition (Fig. 3), probably on the basis
that it is the most intense ultra-violet absorption band. However, upon comparison
of this spectrum with those of other complexes, especially Co(acac)3, it was concluded
that this band is actually the tIe-~ 7r4 transition and that the 270 m,u band is the
7r3-+ 7r4 transition. ORGEL(12) has pointed out that the spectra of chromium(Ill)
and spin paired cobalt(Ill) complexes are very similar because their ground states
and excited states have similar symmetries. Even though the spectra of Cr(acac) 3
and Co(acac)3 in Fig. 3 do not appear to be similar at first sight they actually are.
The pair of bands at 256 and 228 m/z in Co(acac)3 corresponds to the 270 and 255 m.u
pair in Cr(acac)3. Both complexes have a shoulder at 295 m/t and both have an
electron transfer band in the same region, i.e., at 332 m/t in Cr(acac)3 and at 323 m#
in Co(acac)3. Bands in the visible region are also similar. Now, one would assign
the 256 m/~ band in Co(acac) 3 to the rr3 ~ 7r~ transition, and the corresponding band
in Cr(acac)3 lies at 270 m#. The main difference between the spectra of the cobalt
and chromium acetylacetonates is the relative intensities of the bands. In Cr(acac)3
the d, -~ ~4 transition is unusually intense and the 7r3-~ rq transition is unusually
weak. A final argument for this assignment is that interpolation between the energies
of the electron transfer bands in V(acac)3 and Mn(acac):~ shows that this transition
should fall near 332 m # in Cr(acac)3.
The 4A9¢ ~ 4T2g transition lies at 560 m/~ and the 4A2g ~ 4Tlg transition is there-
fore expected ca. 7,000 cm -1 higher/~,~'~,u) i.e., at about 400 mp. A band is observed
at 380 m # but unlike the "second" band in chromium(Ill) complexes, it is about
seven times more intense than the first band and it shows fine structure. A survey of
the published spectra of chromium(Ill) complexes failed to produce any with a
similar band. It is therefore believed that the 4A2~ ~ 4T~ band is seen as a slight
shoulder at 430 m # and that the 380 m/z band is due to some other transition.
Tris(acetylacetonato) manganese(Ill). In ethanolic solution Mn(acac).~ appears
to be partly dissociated t~) but not in chloroform or benzene solution.
There is a broad band at 272 m/~ in chloroform which is assigned to the rr3 ~ rr~
transition (Fig. 1). The breadth of the band may be duc to a small splitting of the
band as already discussed for the case of V(acac)3.
The d~ -~ rr a transition is clearly distinguishable at 318 m/z.
The shoulder at 550 m/z is assigned to the ~Eo ~ ~T2~ transition, which is the only
spin allowed d ~ d band expected in an octahedral complex. Actually, Mn(acac):~ is
tl~J L. E. ORGEL, J. Chem. Soc. 4756 (1952).
ils~ L. E. ORGEL, J. Chem. Phys. 23, 1004 (1955).
tx4~ R. TSUCrlIDA, Bull. Chem. Soc. Japan, 13, 388 (1938).
II~,l G. [-l. CARTt.EDGE, J. Amer. Chem. Soc. 73, 4416 (1951).
228 D . W . BARNUM

5.0 3.0

4.5 / 2.5
/ %

/ P',
4.0 Z.O
e,,J

3.5 1.5

I
w
I
3.0
I
J 1.0

I I
I
i
2.5
/ / 0.5
/
i
Z.0
-I
e

/',, /i
1.5
I
I
i/t!l
/'V -0.5
/
I I
1.0
J I
I
i I I I -1.0
10,000 ZO,O00 30°000 40,000 5C 000
FREQUENCY, cm-*
FiG. 3.--Absorption spectra.
...... Cr(acac)3 in ethanol (left hand scale)
- - - - - - Co(acac), in ethanol (left hand scale)
. . . . . . . . . Co(acac), in chloroform (right hand scale)
 Electronic absorption spectra of acetylacetonato complcxcs--I 229

probably distorted to lower symmetry c~,tT) but it cannot be detected in this spectrum.
The 1125 m/~ band must therefore be the spin forbidden :'/:',j -~ '~T1~transition, although
it has the unusually high intensity of 85.3 1. mole -t cm-L FURLANI and CIANAI~''
also observed exceptionally intense spin forbidden transitions in some other mangancsc
complexes.
Tris(acetvlacetonato) iron(Ill). The 7r:~-~ ,-r4 transition is located at 272 m/~ and
thc d, --, n4 transition at 351 m # (Fig. 2). The unusually low energy of the electron
transfer band is due to the exceptionally small splitting of the 3d-orbitals by the
ligand tield in a spin-free iron(Ill) complex, and will be discussed in a forthcoming
rcport. ~') The origin of the band at 431 mt.~ is unknown.
In chloroform, two weak spin-forbidden bands were observed at 725 and 975 m/~.
These are the ~;A~,~ .... 4T.,~ and the 6A1,, --~T~,, transitions, respectivcly.
Tris(acetvlacetonato) cobalt(HI). The ultra-violet spectrum of Co(acac):~ has
already been discussed in connection with the spectrum of Cr(acac):~.
In chloroform solution, two weak shoulders were observed at 1100 and 800 m/~
(Fig. 31. These are assigned respectively to the spin forbidden transitions tar, --~ aT~,,
and IAt~, - :~T.,,~.
~r.--,- rr* T R A N S I T I O N S

The enol form of acetylacetone (the hydrogen chelate) has an intense absorption
band which appears at 272-274 m/~ (log e ~-~ 4.0) in a number of different solvents.
in basic solution the enolate ion is formed, the absorption band shifts to 294 mt~
(Fig. I), and the intensity increases slightly. ¢~s) The ~r-energy level diagram of the
enolate ion, which was obtained from a Hiickel L C A O - M O calculation, ~z~ is shown
in Fig. 4, and the 294 m/t band is assigned to the ~-:~--~ rr 4 transition. ¢tg-z~
Upon chelation with a metal ion the rr --~ rr* transition may shift either to longer
or to shorter wavelength with respect to the enolate ion. Attempts have been made
by previous authors to correlate this shift with some property of the metal ion, the
structure of the complex, the nature of the metal.-ligand bonds, or the formation
constants. ~9.'2,'a,z~ In the most recent and complete study HOLM and C O T T O N (2°)
measured the spectra of a large number of complexes and concluded that several
factors, which cannot be easily separated, contribute to the position of this band.
Although this difficulty has not been resolved in the present work, it has been found
that the effect of metal-ligand rr-interaction in the trivalent transition metal complexes
is large enough that it stands out over the other effects.
Valence bond considerations. BASt; and CIIATTERJI(22) observed that in some
acetylacetonato complexes there are two intense ultra-violet absorption bands rather
than one. Thcy tirst assumed that these two bands indicate that the two C --O bonds
(,and consequently the two metal-oxygen bonds) in a chelate ring are different. In a
morc recent paper 12:~) they apparently abandon this view and assume instead that
upon chelation the 274 mtt band of the enol form of acetylacetone splits into two
(1,;I A. FORMAN and L. E. ORtilL. Mol. Phys. 2, 362 (1959).
tl:, C. Ft:RLANI a n d A. ClAN,A, Ann. Chim. (Rome) 48, 286 (1958).
c~K~E. E. BLour, V. W. EAC;~.R and D. C. SILVERMA>,, J. Amer. ('hem. Sot'. 68, 566 (19467.
(~a~ R. L. B~.FORD, A. E. MARrFH. and M. CAt.viy. J. lnorg. Nucl. Chem., 2, 1 I, (1956L
tz0) R. H. HOLM and F. A. C o t t o N , J. Amer. Chem. Sot'. 80, 5658 (1958).
~z~ E. L. Z~ROSK~, U.S.A.E.C. Report BC-63
I~-a) S. BASU and K. K. CHA'lq'ERJI, Naturwissenschaften. 42, 413 (1955~; 43, 124 (1956L
~.~a~S. B~su and K. K. (2HATTI:RJI, Z. Ph.vs. ('hem. 209, 360 (1958).
~z~) A. 1. K~ss and J. CSASZAR, Acta Chim. Acad. Sci. Hung. 13, 49 (1957).
230 D . W . BARNUM

M (acac)3 M (acac) 3
Or-BONDING WITH SYMMETRY
-I0,000 acac NEGLECTED n-BONDING O F M.O.

~5 | / ~ n i | ' ~ m n n ~ • e

-ZO,O00

-30.000 7]'4 n / ~ =mature n ~ al , e

-40,000
dy mm imam e

-50,000
O

dE mmmmmmjmmmmmm aI , e
-60.000
Z

-70,000

-80,000

-90,000

-I00,000
7t, 41J'-'~" "--t'--t' i,ha-L,M, a] , e

~ ~ j l . t ~ - - --~.N,/.t a2 , e
-110,000

FIG. 4.---Molecular orbitals of the acctylacetonate anion and of a trivalent

transition metal complex.
 Electronic a b s o r p t i o n spectra o f acetylacetonato c o m p l e x e s - - I 231

bands. One band shifts to lower wavelength if the metal-oxygen bond is more
covalent than the hydrogen-oxygen bond in the enol. This is illustrated in the series
enolate, enol, and O-methyl ether of acetylacetone shown below. The second band
was assumed to be due to metal-ligand ~--bonding, and it was further assumed that
it shifts to lower energy since ~--bonding results in an increase in the length of the
conjugated system. It has since been shown that this assumption is incorrect. That
is, metal-ligand rr-bonding results in a shift of the ,r -+ rr* transition to higher energy, t2~
Furthermore, the splitting of an absorption band into two new bands implies a
splitting of the excited state by configuration interaction or by the symmetry proper-
ties of the atomic orbitals which are combined in the complex. It does not seem
likely that such phenomena can be explained on the basis of valence bond structures.
Equivalence of metal-oxygen bonds. If the two metal-oxygen bonds in a chelate
ring were not equivalent it can be calculated that the , r - ~ ,r* transition would be
shifted as much as 4,000 cm -1 to higher energy (from 294 m # to 263 m/~) due to loss
of delocalization energy. This is believed to be the reason why the enoi form of
acetylacetone has maximum absorption 2,700 cm -1 above that of the enolate ion.
That is, the hydrogen bond is not symmetrical and therefore the two C:--O and C = C
bonds in the ring are not equivalent. The loss of delocalization energy and the
resulting shift to higher energy can be seen in the following series of compounds:
O O H CH 3
, \ //
C C O O O O
/ ~,,, , / \,\ il
t I
CH~ C CHs C C C C
I ./'/ " ' \
H CH 3 C CH 3 CH z C CHa
E I
H H
Vmn x -- 34,000 cm -x *'max = 36,600 cm -1 *'max -= 39,300 cm -1 (25)
KEUSSLER and ROSSMYc2sl observed that the deuterated enol shows maximum
absorption very close to that of the hydrogen complex (37,000 cm i). and interpreted
this to mean that the hydrogen atom lies in a potential well with only a very small
barrier between the two minima, since resonance in the enol would require that the
proton change its position by tunneling through any potential barrier. However, in
the interpretation presented here one would not expect a large difference between
the rr3 ~ 7r4 transition energies in the hydrogen and dueterium chelates since the loss
in delocalization energy should be about the same in both cases.
In other chelates, however, structure determinations by X-ray diffraction have
shown that, within experimental error, the metal-oxygen bonds are equivalenU z~)
Therefore, except for the hydrogen chelate, non-equivalence of these bonds is not
expected to be a factor which influences the ultra-violet absorption spectrum.
~s~ V. KEUSSLERand G. ROSSMY,Z. Elektrochem. 60, 136 (1956).
12el V. AMIRTHALINGAM, V. M. PADMANABHAN a n d J. SHANKAR, Acta Cryst. 13, 201 (1960); G . J. BULLEN,
Acta Cryst. 12, 703, (1959); R. P. DODGE, U.S.A.E.C. Report UCRL-8225 (1958); H. KOYAMA,
Y. SAITO and J. KUROYA, Inst. Polytech. Osaka City Univ. Ser. C, 4, 43 (1953), Chem. Abstr., 48, 3097 b ;
V. M. PADMANABHAN,Brae. Indian Acad. Sci. 47A, "329 (1958); R. B. ROOF JR., Acta Cryst. 9, 781 (1956) ;
E. A. SHUGAM, Trudy Vsesoyu2. Nauch. -Issledovatel. Inst. Khim. Reaktivov No. 21, 108-11 (1956),
Chem. Abstr. 53, 21025 g; J. V. SILVERTON and J. L. HOARD, Abstracts of Papers, Amer. Chem. Soc.
"~teeting, DiE. of Phys. Chem. Sept. 11-16, (1960).
232 D.W. BARNUM

Effect of positi~:e charge on the metal ion. BELFORD eta/. (l°) made a Hiickel
LCAO-MO calculation of the atomic charges in the ground state and the excited
state of the acetylacetonate ion and then calculated, on the basis of the coulomb
forces, the change in energy of both of these states which is brought about by placing
a point positive charge in the position of the metal ion. They found that an increase
of one unit in positive charge should shift the ~r ~ rr* transition about 3,200 cm -~ to
lower energy, but HOLM and COTTON¢2°~ have shown that such a shift is seldom
observed, even in some especially favourable cases in which the metal-ligand bonding
is apparently "ionic" and the sizes of the metal ions of charge + I and ÷ 2 which
were compared were about equal. The calculation itself appears to be valid, however,
since it correctly predicts the effect of a positively charged quaternary nitrogen atom
on the rr ~ rr* transition in some ~,fl-unsaturated ketones ~2r) and SCHL,~FER and
K/3N1G~zs,291 have used essentially the same method to predict the effect of metal ions
on the ~" ~ 7r* transitions in pyridine and 1,10-phenanthroline complexes.
Actually, metal-ligand bonds are neither 100 per cent ionic nor 100 per cent
covalent ~l,3°,ax) and the charge on the metal ion is less than the formal charge.
Pauling's electronegativity values were used to estimate the amount of covalent
character, and hence the actual charge on the metal on, but this did not lead to any
improvement in the correlation between positive charge and transition energy.
Effect ofmetal-ligand ~r-bonding. In Fig. 4 the calculated energy level diagram of
the free acetylacetonate ion is shown with its six rr-electrons in the ~r1, ~'2 and ~'a-
orbitals. When this ion complexes with a metal the MO calculations show that the
7r-orbitals are all raised slightly in energy and the d e and d7-orbitals fall in between
the 7rz and Tra-orbitals. These calculations and the order of the energy levels are
discussed in a forthcoming paper/2~ In the ground state the appropriate number of
electrons are distributed in the molecular orbitals according to Hund's rules and with
the additional stipulation that if there are between four and seven 3d-electrons, and
if the de and dv-orbitals are close enough together, it may cost less energy to put the
last electrons into the d:,-orbitals rather than pairing them with electrons already in
the d,-orbitals. That is, the ground state is the spin free configuration rather than the
spin paired/z2~ The electronic configurations found by magnetic susceptibility
measurements are given in Table 3.
As seen in Fig. 4 the LCAO-MO calculations show that the effect of metal-
ligand rr-interaction is to shift the ~r3 ~ rra transition to higher energy. Previous
investigators have assumed that this transition will shift to lower energy since an
increase in metal-ligand r-bonding involves an increase in the length of the resonance
path. However, this rule applies only to transitions from the highest occupied
rr-orbital to the lowest unoccupied rr*-orbital, and even then there are exceptions/aTI
Since the d~-orbitals, which are the ones that participate in double bonding with the
ligands, lie in between the ~r3 and the ~'a-orbitals, the d,-orbitals become the highest
occupied "~-orbitals". The effect of inserting a new 7r-orbital in between two others
is to spread them apart, shifting the electronic transition between them to higher
energy.
c27~ D. W. BARNUM. To be published.
c~s) H. L. SCHL~FER, Z. Phys. Chem. 8, 373 (1956).
~ H. L. SCHL~FER, and E. K~SNIG, Z. Phys. Chem. 19, 265 (1959).
isol C. K. JORGENSEN,Acta Chem. Scand., 12, 903 (1958); Disc. Faraday Soc. 110 (1958).
tsu W. Low, Paramagnetic Resonance in Solids, p. 96. Academic Press, New York and London (1960).
~32~ L. E. SUTTON, J. lnorg. Nucl. Chem. 8, 23 (1958); J. Chem. Educ. 37, 498 (1960).
 Electronic absorption spectra of acetylacetonato complexes--I 233

Now, there are two types of ~--interaction. In the first type, which is illustrated in
Fig. 5a, the d~-orbitals interact with higher energy, empty, anti-bonding ~'-orbitals
of the ligand. This leads to increased stability by lowering the energy of d~-electrons
and decreasing the coulomb repulsion between the d,-electrons by allowing them to
spread out over the ligands. Partial donation of d,-electrons to the ligands also
enhances the donation of n-electrons from the ligand to the metal ion. Since thc

TABLE 3.. -[~I.ECrRONIC CONFIGURATIONS IN SOMI ACETYI.A('I-I()NATO COMPI EXES

Complex No. of 3d-electrons Config. Ref. r max., ,-7 . ~'~

( c m ~)

Sc(acac).~ 0 d," 33.800

ri(acac).~ I d/ 32,500
V(acac)a 2 d, -' . • 34,500
Cr(acac)3 3 d,.:~ 33 37,100
Mn(acac)a 4 d, ~d.) 33. 34 36.800
Fc(acac)3 5 d, ~d.2 35 36,800
Co(acac).~ 6 d,'; 33, 36 39. 100

energy of the d,-electrons is decreased the value of A. the energy difference between
the d, and the d.-orbitals, is increased, Therefore, complexes which have this type
of n-bonding have unusally high energy d--~ d absorption bands.
In the second type of ~--interaction, shown in Fig. 5b, the d,-orbitals interact with
lower energy, filled, 7r-orbitals of the ligands. This lowers the energy of the ligand
¢r-orbitals but at the same time it raises the energy of any d,-electrons, and hence
decreases the value of A. if the d~-orbitals are filled, as in Co(acac) 3, there is no
net gain in electronic energy since six electrons have been lowered and six have been
raised. In fact, an increase in energy is expected due to the coulomb repulsion between
tile d,-electrons and the rr-electrons oil the ligands. Furthermore. in this type of
or-interaction, or-electrons are transferred from the ligand to the metal and they are
therefore competing with the ligand or-electrons for the quota of electron density
that the metal ion can accept.
The acetylacetonate ion has an empty ~'4-0rbital, above the d,-orbital, but it also
has a filled ¢rzcorbital which is lower than the d,-orbital. Thus, both types of 7r-
interaction discussed above seem possible. The Hficke[ LCAO-MO calculations
show that the ~r3-0rbitals are depressed and the ¢r4-orbitals remain unaffected by
n-interaction, suggesting that the interaction is of the second type.
The literature contains contradictory remarks about the amount of n-bonding in
acetylacetonato complexes. NAKAMOTOand MARTELLta~) concluded from the infrared
absorption spectrum and a normal coordinate treatment of Cu(acac)~ that the
metal ligand bonding is intermediate between that found in Ni(CO)~ and in Fe(CN)d I ,
two species in which it is believed to be quite strong. This is in agreement with a
(:):~' H. S. JARRErr, J. Chem. Phys. 27, 1298 (1957).
13.1} L. C. JACKSON, Proc. Roy. Soc. 47, 1029 (1935).
~a:,~ I.. C. JACKSON, Proc. Roy. Soc. A 140, 695 (1933).
~:";~ R. O. WHIPeLr, R. W E s r a n d K. EMERSON, J. Chem. Sot'. 3715 (1953L
c:3~ A. PULLMAN a n d B. Pt-LLMAN, Disc. Faraday Sot'. 9, 46 (1950).
' : ~ K. NAK^MO'rO a n d A. MARI~;t.L, J. Chem. Phys. 32, 588 (1960).
234 D . W . BARNUM

statement made by MCGARVEYta°) but in a subsequent paper MAKI and McGARvEY c~°)
conclude from paramagnetic resonance data that in C u ( a c a c ) ~ " . . . the out-of-plane
*r-bonding is ionic". The position of the *r3 ~ *r4 transition in Cu(acac)z supports
this latter conclusion.
As for other acetylacetonato complexes, another argument against the existence
of strong metal-ligand ~--bonding is the position of the acetylacetonate ion in the
spectrochemical series. (41) Those ligands which are known to form strong ,r-bonds
a)
71"*
m l m

<...../
b)

2L
dE
~ ~ Immml ff

dE-ORBITALS M O ' s IN 7r-ORBITALS

BEFORE COMPLEX WITH BEFORE
7r-BONDING ;r-BONDING 7r-BONDING
FIG. 5.--Effect of different types of metal-ligand ,r-bonding on the energy of the d, and 7r-
orbitals. (a) Empty *r-orbitals of higher energy than the d,-orbitals. (b) Full 7r-orbitals of
lower energy than the de-orbitals.

of the first type, such as CN-, NOz-, and 1,10-phenanthroline, have unusually high
positions in this series. Ligands which give rise to rr-interaction of the second type,
such as the halides, lie somewhat low in the series. The acetylacetonate ion, however,
has an intermediate position, very close to that of water, and this indicates that
*r-interaction is not strong.
On the other hand, the rr a --~ ~r4 transition energies indicate strong *r-interaction
in some complexes. In Table 4 it can be seen that the energy varies over a range of
6,500 cm -x (from 308 m/z to 256 m/~). It does not seem likely that some simple
property of the metal ion such as the electronegativity or charge/radius ratio could
cause such a large spread. The direction of the shift suggests that zr-interaction is
tst~ B. R. McGAR~Y, ,1".Phys. Chem. 60, 71 (1956).
(to) A. H. MAgi and B. R. MCGARVEY, J. Chem. Phys. 29, 31 (1958).
(tt) V. SHIMURAand R. TSUCHIDA, Bull. Chem. Soc. Japan, 29, 311 (1956).
 Electronic absorption spectra of acetylacetonato complexes--I 235

increasing as one goes through the series from Ti(acac)~ to Co(acac)3. In fact, as can
be seen in Table 4, the wavelength of the % --~ ~'4 transition depends on the number
ofd,-electrons. The d,-electrons are the ones which satisfy the symmetry requirements
for combining with the ligand ¢r-molecular orbitals.
In order to explain these observations it is suggested that the actual ¢r-interaction
in acetylacetonato complexes is a combination of the two different types described
above. Thus, the d~-orbitals interact with both lower, filled 7r-orbitals and the higher.
empty ~r-orbitals. This has the effect of spreading the % and ~r4-orbitals apart and
shifting the transition between them to higher energy. At the same time, rr-interaction
of the first type tends to lower the energy of the d,-orbitals, but interaction of the
TABLE 4.--OBSERVED E N E R G I E S OF 7r3 ---* rr 4 A N D d e ~ rr.l TRANSI'FIONS

Number of 773 ---~ "JT41

Complex 3d~-electrons (cm-') (cm-')

Sc(acac)a ' 0 33,800 none

Ti(acac)s 1 32,600 25,700
V(acac)3 2 34,500 29,200
35,900?*
Cr(acac)3 3 37.100 30,200
39,200 ?
Mn(acac)3~ 3 36,800 31,400
Fe(acac)3 3 36,800 28,500
Co(acac)a 6 39, 100 31,000
43,900 ?
* Bands followed by a question mark (?) are due to apparent splitting of the
7r3-~Tra transition, as discussed in the text. Solvent is absolute ethanol unless
stated otherwise.
t Chloroform solution.
second type tends to raise their energy. The net result is that the energy of the d,-
orbitals and hence the vaue of A, remains about the same as if no rr-interaction were
present. Both types of ~r-interaction depend strongly on the number of d,-electrons,
the first type because it lowers the energy of these electrons and there are more
electrons available to transfer charge back to the ligands, the second type because the
interelectronic repulsion increases with increasing number of d,-electrons.
In the acetylacetonato complexes of divalent transition metal ions the 7r3 - ~ rr 4
transition energy is very close to that of the free acetylacetonate anion. This indicates
that 7r-interaction is not strong, in agreement with the paramagnetic resonance studies
of MAKI and MCGARVEYt4°) on Cu(acac) 2. In the chelates of Ni(acac) 2, Cu(acac)2,
and Zn(acac)., the d,-orbitals are full, and by analogy with Co(acac).~ one might
expect strong ~r-interaction in these complexes as well. Although certain rationaliza-
tions might be made, this is not yet understood. It is a further reminder, however,
that other factors are of importance besides the number of dcelectrons. Apparently,
in the above series of trivalent transition metal complexes the effect of ,-r-interaction
is large enough that it overshadows these other effects. It would be naive to assume
that these other effects were even nearly constant upon going through the series since
Sc(acac)3 does not fall in order with the rest of the series, and even in the series
Be(acac)2, Mg(acac)z, Ca(acac)2, Sr(acac)2, and Ba(acac) 2, the position of the rr~ ,- 7r4
transition shows no regular trend, t'°)
Configuration hlteraction. The effect of coulomb repulsion between electrons
236 D . W . BARNUM

(configuration interaction) is neglected in the Hiickel LCAO-MO model. This

interaction often causes an absorption band which involves degenerate orbitals in
either the ground state or the excited state to split into two or more bands as, for
example, in benzene34z~ It is important to inquire into the possibility of such splitting
in the absorption spectra of metal chelates.
Even though the rr,-orbitals in a M(acac)3 complex are triply degenerate (Fig. 4),
each of the three pair of electrons is on a separate chelate ring. Since they are so far
apart the repulsion between them, and hence the magnitude of any splitting of the
rr3 ~ ~'4 absorption band is expected to be much smaller than in benzene where the
degenerate orbitals are on the same ring. With increasing metal-ligand ~r-interaction
the repulsion between electrons in different rings should increase. This may be the
reason for the apparent splitting of the ~'3 ~ rr4 transition that is often observed.
For example, in Cu(acac) 2, Al(acac)a, Th(acac) 4, and many other complexes in which
rr-interaction is very weak if it occurs at all, a slight shoulder appears on the long
wavelength side of the rra ~ r: 4 band. However, in the complexes with trivalent
transition metal ions the splitting is even more distinct and appears to increase with
increasing number of d,-electrons. In V(acac)3 there are two distinct humps with
approximately equal intensity in the 7ra ~ rr 4 region. Similarly, in Cr(acac)3 and
Co(acac)z there are two close bands that might be attributed to configuration inter-
action. These bands are listed in Table 4. Even though the splitting is appreciable, it
is still small with respect to that observed in benzene.
There is another possible explanation for the apparent splitting of the ~r3--~ rr 4
band. Since the d~-orbitals belong to the a x and e representations of D z they may
interact with 7r-orbitals on the ligands which belong to a I or e but not to the a 2
representation. The HiJckel LCAO-MO calculations do not predict an interaction
with the ~r4-0rbitals, but a more elaborate treatment may show that the ~r4(e)-orbitals
are raised in energy by 7r-interaction while the rr4(az)-orbitals remain unaffected.
Thus, the ~r4-0rbitals would be split by ~--bonding and, just as observed in the ultra-
violet spectra, the splitting would increase with increasing metal-ligand rr-interaction.

ELECTRON TRANSFER TRANSITIONS

The trivalent transition metal acetylacetonates studied in the present work have
an absorption band between 300 and 400 m/z which is characteristic of neither the
metal ion nor the ligand. The intensities are on the order of 1,000 to 10,000 L mole -1
cm -t. From the results of the MO calculations and the direction of shift upon going
through this series of complexes these bands are assigned to the d, ~ ~r4 transition.
In Fig. 4 it can be seen that there are several possible one electron transitions that
involve both an orbital which is principally on the metal and an orbital which is
principally on the ligand, and which have energies less than 50,000 cm -1. Certain of
these are not possible in specific cases. For example, transitions to the d,-orbitals

7r3 -+ dr I Electron transfer from d, --~ % Electron transfer from

rr~ --~ d , ligand to metal d v ~ rr4 metal to ligand
n---~ d e dy ___~rr 5
n --+ d,I
~'~' F. A. MA'rSEN, Chemical AppBcations o f Spectroscopy, (Edited by W. WERT) Part 2, Chap. 5, p. 667.
Interscienee, New York (1956).
 Electronic absorption spectra of acetylacetonato complexes--I 237

are not possible in Co(acac)3 since they are already tilled. Similarly, transitions from
the d,-orbitals are only possible in Mn(acac)3 and Fe(acac)z because these are the
only complexes which have electrons in the d..-orbitals in the ground state.
The trivalent metal ions of Ti, V, Cr, Mn, Fe, and Co become more difficult to
oxidize on going through this series. With the exception of Fe(acac)3, the observed
electron transfer bands shift to higher energy on going through the series Ti(acac).~.
V(acac) 3, Cr(acac)3, Mn(acac)3, Fe(acac)3, and Co(acac)3 (Table 4). Therefore the
transition must involve "oxidation" of the central metal ion. That is, the electron is
transferred from thc metal to the ligand, and it is assigned to the d, • , 7r~ rather than
to the d, --* rr~ transition on the basis of the calculated encrgy differences between
these orbitals.

Note added hi proof-. FERGUSON has concluded from an analysis of the polariza-
tion of absorption bands in the crystal spectrum of Cu(acac)2 that metal-ligand
~r-bonding is strong [J. Chem. Phys. 34, 1609 (1961)]. He also points out that the
conclusion from paramagnetic resonance studies that the 7r-bonding is "ionic ''~4°~
depended on an erroneous assignment of the d - ~ d transitions. This conclusion is
not necessarily at variance with the U.V. absorption spectrum of Cu(acac) z since
more complete MO calculations ~2) for M(acac),~ complexes have shown that the
7r.a-* rrj transition is split into four transitions by n-bonding. One transition is
independent of metal-ligand 7r-bonding and the other three shift to higher energy.
A similar splitting is expected in the M(acac)2 case. Thus, the 295 m/~ band ot"
Cu(acac)., is probably a ~r3 ---* 7ra transition that is independent of ~r-bonding. A band
is also observed at 242 m p which may be a 7r:~~ 7r~ transition that is shifted to
higher energy by 7r-bonding.
CHARET'rr et al. [J. Phy's. Chem. 65, 735 (1961)] arrived at the opposite conclusion
from this author regarding the equivalence of the two C=-O and the two C,: C
bonds in a chelate ring. Although these bonds are not equivalent in the copper(ll)
tropolone complex they are equivalent, within experimental error, in all acetyl-
acetonato complexes for which X-ray structure determinations have been carried
out/'~) Actually, these workers made an error in the calculation of the absorption
maximum and the value refers to the cnolate anion rather than the enol. With a
smooth potential well the correct value should be 314 mp, which is in fair agreement
with the observed value of 295 mp. It is therefore not necessary to assume that the
pairs of C - O and C ~ C bonds are not equivalent.

Acknowledt,,ements-The author is indebted to Dr. D. P. SII'.'VtNSONand Dr. J. D. SwAI~, for many

helpful discussions and suggestions.