373

Acta Gryst. (1956). 9, 373

The Crystal Structure of Ferrocene*
BY J. D. DUNITzt, L. E. ORGEL~: ~ D ALV.~A~D~R Rm~§
The Gates and Crellin Laboratories of Chemistry, California Institute of Technology, Pasadena, Califor~a, U.S.A.

(Received 7 December 1955)

X-ray analysis of crystals of ferrocene shows that the molecule has approximate Dsd symmetry,
with Fe-C, 2.045-~0-01 A; C-C, 1-403-}-0.02 A.

Introduction sources of error, a p a r t from those i n h e r e n t in the visual

The 'sandwich' structure of bis-cyclopentadienyl iron method. Absorption errors of up to a b o u t 3 5 % are
(Wilkinson, R o s e n b l u m , W h i t i n g & Woodward, 1952; possible for i n d i v i d u a l reflexions, although the m e a n
Fischer & Pfab, 1952) has already been confirmed b y absorption error is m u c h less t h a n this. E x t i n c t i o n
p r e l i m i n a r y crystal-structure studies (Efland & Pe- errors are likely to be more serious. Since the iron
pinsky, 1952; D u n i t z & Orgel, 1953). W e h a v e now atoms occur in a C-centred arrangement, reflexions
carried out a more detailed analysis, utilizing three- with h+k even show almost n o r m a l decline, a n d are
dimensional data, a n d h a v e obtained a somewhat therefore almost all strong a n d so subject to extinction.
more precise description of the structure. Reflexions with h+k odd are, on the other hand,
m a i n l y weak. The extinction error t h u s occurs in a
r a t h e r s y s t e m a t i c fashion. W e shall r e t u r n to this
Experimental point in a later section.
The cell constants were re-determined from (hk0) a n d
(hO1) Weissenberg photographs on which sodium
chloride powder lines h a d been superimposed. Least- Scale and temperature factors
squares analysis of the observed spacings leads to The t e m p e r a t u r e factors for the iron atoms and for
the values the carbon atoms could be estimated i n d e p e n d e n t l y .
The iron atom does not contribute to reflexions with
a = 10.561, b = 7.597, c = 5.952/~, ~ = 121.02 °
h +/c odd. The t e m p e r a t u r e factor for the carbon atoms
for the cell constants, with estimated s t a n d a r d error could therefore be obtained b y Wilson's (1942)
of a b o u t 0.1%. The space group is P21/a a n d there method, where the averaging is t a k e n only over odd
are two c e n t r o s y m m e t r i c molecules of Fe (C5H5)2 in reflexions. For reflexions with h + k even, a n average
the u n i t cell, so t h a t the iron atoms m u s t lie at sym- over the ~' values, r a t h e r t h a n the F ~ values, yields
m e t r y centres. the t e m p e r a t u r e factor for the iron atom alone. I n
For the i n t e n s i t y estimation, two principal sets of this way, the t e m p e r a t u r e factors are found to be
photographs were t a k e n about the b a n d c crystal exp ( - 5 . 5 sin 90//t 2) for the carbon atoms and
axes, with M o K ~ radiation (2 = 0.7107 A). The exp ( - 3 . 2 sin s 0//t 2) for the iron atom. The scale fac-
crystals were sealed in thin-walled glass tubes to tors d e t e r m i n e d b y the two methods were in good
prevent loss b y s u b l i m a t i o n and the layers, hkO-hb7, agreement.
a n d hOl-h31, were recorded b y the Weissenberg equi-
inclination technique with several films, separated b y R e f i n e m e n t of c o o r d i n a t e s
t h i n sheets of copper, in the camera. A few reflexions The a p p r o x i m a t e structure based on t h e first projec-
of v e r y large spacing were recorded with Cu Kc~ tions down the b and c axes was used as a starting
radiation (2 = 1.542/~). The IF[ values of almost 1200 p o i n t for a series of least-squares refinements. F o u r
reflexions were obtained from visual estimates of the r e f i n e m e n t stages were first carried out on h+k odd
intensities. reflexions. Two more stages were t h e n done on all
Our i n t e n s i t y estimates are subject to two m a i n reflexions. Coefficients of off-diagonal terms were not
computed a n d were t a k e n as zero, with the exception
* Contribution No. 2050 from the Gates and Crellin Labora- of those involving products of the t y p e (SF/Sxi)
tories of Chemistry.
~f Present address: The Royal Institution, Albemarle (~F/Szi). The final coordinates (Table 1, column (A))
Street, London W. 1, England. h a v e a s t a n d a r d error of 0.024 /~, d e t e r m i n e d from
Present address: University Chemical Laboratory, Cam- the i n t e r n a l consistency of the least-squares results.
bridge, England. A t the end of the least-squares r e f i n e m e n t t h e
§ Sr. Asst. Surg. (R), National Institute of Mental Health,
National Institutes of Health, U.S. Public Health Service, q u a n t i t y R = X[IFoI-IFcI[+Z[Fo] was still as high as
Department of Health, Education, and Welfare. 0.175, which led us to compute a three-dimensional
374 THE CRYSTAL STRUCTURE OF FERROCENE

Table 1. Atomic coordinates The m o s t likely cause of these effects would seem
(A)* (B)* to be extinction errors. The v e r y strongest reflexions,
^ those most seriously reduced b y extinction, h a v e t h e
rx/a y/b z/c x/a y/b z]c carbon contribution in phase with the iron contribu-
C1 0.0169 0.2592 0.0306 0 - 0 1 6 9 0.2613 0.0306 tion, a n d are therefore positive. N o w t h e Fourier
Cz 0.0444 0-1777 0.2115 0.0447 0.1767 0-2103 component corresponding to such a reflexion will, in
Ca 0-1614 0.0576 0.0889 0.1612 0.0590 0.0873 general, be positive at t h e atomic positions, a n d t h e
C4 0.2186 0.0788 0.1737 0.2183 0.0787 0.1733
C5 0-1349 0-1978 0 - 2 2 7 8 0.1352 0.1963 0.2267 effect of extinction is therefore predictable. Since Fo
Fe 0 0 0 0 0 0 is measured too small, the electron density n e a r atomic
* (A) As found by final least-squares calculation. positions is diminished. A n d moreover, since only
(B) Corrected on basis of difference synthesis. reflexions with h+k even are affected, t h e diminution
will occur not only a t atomic positions b u t also a t
Fourier series in t h e hope t h a t this m i g h t show system- positions related b y r o t a t i o n a b o u t t h e b axis. This
atic differences from t h e least-squares model. The follows from t h e trigonometric form, cos2~(hx+lz)
signs of the Fourier-series coefficients were t a k e n from × cos 2reky, of these t e r m s in t h e Fourier-series ex-
the final structure-factor calculation of t h e least- pression. The extinction error i n t e g r a t e d over a n y
squares analysis (in fact, all reflexions with h + b even appreciable p a r t of the unit cell m u s t be close to zero,
h a d positive sign); a b o u t 10 w e a k h + k odd reflexions so t h a t the diminution a t the atomic positions will
were omitted because of a m b i g u i t y of sign. The cal- t e n d to be compensated b y a n a c c u m u l a t i o n of
culations were done on I.B.M. machines. The resulting spurious electron density between t h e atoms.
function showed a considerable accumulation of elec- I n our case we h a v e a m e a n s of m a k i n g a rough
tron density between the calculated positions of t h e correction for t h e extinction effects. I n a d j u s t i n g t h e
carbon atoms. This is evident from Fig. l(a), in which coordinates from the gradient of t h e difference syn-
relevant sections a t constant y are superimposed to thesis (Cochran, 1951), t h e shift t e r m is t a k e n as
give a composite representation of the molecule, a n d ~/~r[D(x,y,z)-D(~,y,~)] instead of simply O/~rD(x,y,z),
also from Fig. l(b), which shows a section t h r o u g h for an a t o m a t (x, y, z). F o r C1, of course, which has
the m e a n plane of the cyclopentadienyl ring. x ~ 0, y---¼, z ~ 0, the correction cannot be applied,
Such a n a c c u m u l a t i o n of electron density also a n d this is u n f o r t u n a t e because, owing to t h e r a t h e r
appears in t h e three-dimensional difference synthesis special character of the coordinates, the extinction
in which t h e Fourier coefficients are Fo-Fc; b u t in error is p a r t i c u l a r l y severe for this atom.
addition, smaller although quite definite accumulations
occur at positions related b y r o t a t i o n a r o u n d t h e b axis. Table 2. Bond distanc2z
The difference synthesis has strongly negative troughs
Bond distances are calculated from the coordinates of Table 1,
at every atomic centre; a n d again, although less columns (A) and (B). The standard deviation quoted is that
marked, a t positions related to these b y r o t a t i o n of the individual results from the mean. The standard deviation
a r o u n d b. I t is clear t h a t the low values of t h e electron- of the mean itself is ~] V5
density function a t the atomic centres a n d the high (A) (B) (A) (B)
values between the atoms of the ring are only in p a r t Fe-C 1 1.994/~ 2.010 A Cx-C~ 1-396A 1.403 A
a t t r i b u t a b l e to rotational disorder; the presence of Fe-C 2 2 - 0 5 9 2.050 C2-Ca 1 . 4 0 0 1-386
corresponding features a t s y m m e t r y - r e l a t e d regions Fe-C a 2.078 2-069 Ca-C4 1 . 3 6 5 1-352
which are, in general, not close to a n y of the a t o m s Fe-C 4 2.074 2.070 C4-C5 1 - 4 1 3 1-399
Fe-C s 2.034 2.024 C5-C1 1-471 1-476
m u s t arise from some systematic error in our intensity Mean 2 - 0 4 8 2-045 1.409 1-403
estimations. S.d. 0.031 0.024 0.035 0-040

Fig. 1. (a) Superposition of relevant sections of three-dimensional Fourier series to give composite picture of Fo(CsHs) 2 molecule.
(5) Section of three-dimensional Fourier synthesis through plane of cyc/opentadienyl ring.
 J. D. D U N I T Z , L. E. O R G E L AND ALEXANDER RICH 375

Adjustment of the least-squares coordinates (Table tion of the rings would lead to complete smearing of
1, column (A)), as described above, leads to the co- the electron density around the circle formed by the
ordinates listed in Table 1, column (B), and the bond rotating group.
distances in Table 2, columns (B). The Fe--C distances That free or only slightly hindered rotation of the
become somewhat more regular, the C--C distances cyclopentadienyl rings occurs in isolated molecules is
slightly less so. Another criterion is the deviation of shown by recent electron-diffraction studies of the
individual atoms from the mean plane of the ring. vapour (K. Hedberg, private communication) and by
The least-squares coordinates give, for these devia- dipole-moment studies of mono- and di-substituted
tions, -0.012, -0.010, -0.076, +0.039, +0.058 J~, ferrocene derivatives (Richmond & Freiser, 1955). No
a = 0.047 J~, while the adjusted coordinates lead to appreciable barrier is to be expected, moreover, from
-0.007, +0.013, -0.046, +0.023, +0.017 A, a = molecular-orbital considerations (Dunitz & 0rgel,
0.025 .~, a significant improvement. 1953, 1955; Moffitt, 1954). The Dsa configuration
found in the crystal is, however, slightly stabilized
because of the staggering of the carbon atoms in
D i s c u s s i o n of r e s u l t s opposite cyclopentadienyl rings, separated by 3.32 J~
in ferrocene. Freezing of the molecules in this configur-
The coordinates of Table 1, columns (B), lead to a ation must arise from the interaction of neighbouring
molecule which still departs quite considerably from molecules in the crystal. The hydrogen atoms cannot,
Ds~ symmetry. The standard deviation from the mean of course, be located from our data, but if the C-H
of the Fe-C distances is 0.024 A, of the C-C distances bonds are taken as about 1.05 /~ in length and lying
0.040 A, and from the mean plane, 0.025 A. Further- in or nearly in the plane of the ring, it is found t h a t
more, this mean plane is inclined at 3 ° 30' to the line free rotation of the rings would lead to t t • • • H ap-
from the iron atom to the mean centre of the ring. proaches of about 2.2 A_.
Because of the systematic nature of the extinction
errors, it is difficult to estimate the standard error of We are indebted to Mrs Jane Rich for help with the
the carbon atom positions from the mean slope of the intensity estimates, and to Dr J. S. Rollett for car-
difference synthesis. The least-squares analysis gives rying out the least-squares analysis on a high-speed
0.024 A, and if this figure is even of the right order computer kindly made available to us by the Electro-
of magnitude the deviations from strict Ds~ symmetry Data Corporation, Pasadena, California.
are of rather doubtful significance. The Fe-C distance
is 2.045 J~ (a ~ 0.01 _~) and the C-C distance 1.403 A References
(a N 0-02 A) if strict Ds~ s y m m e t r y is assumed.
COC~RA~, W. (1951). Acta Crgst. 4, 408.
The difference map, even after correcting for ex- DU~TZ, J . D . & 0~OEL, L . E . (1953). Nature, L o ~ .
tinction by subtracting D@, y, ~) from D(x, y, z), still 171, 121.
shows an accumulation of electron density between DU~TZ, J . D . & ORGEL, L . E . (1955). J. Chem. Phys.
the atoms of the ring and a corresponding deficit at 23, 954.
the atoms themselves, as would result from libration EILA.ND, P . F . • PEPINSKY, i~. (1952). J . Amer. Chem.
of the ring about its mean position. Free rotation, Soc. 74, 4971.
either of the molecule as a whole, or of the cyclo- FISC~R, E. O. & PFAB, W. (1952). Z. Natur]orsch. 7 B,
pentadienyl rings with respect to one another, is, 377.
however, quite excluded by the X-ray results. The MOFFI~r, W. (1954). J. Amer. Chem. Soc. 76, 3386.
RIC~O~D, H. H. & FREISER, H. (1955). J. Amer. Chem.
(010) projection (Dunitz & Orgel, 1953), in which the Soc. 77, 2022.
sign of every reflexion is determined by the iron contri- W~soN, G., ROSEI~-BLUM, M., WHITING, M.C. &
bution, is independent of any assumptions concerning WOODW~D, R . B . (1952). J. Amer. Chem. Soc. 74,
the carbon atom positions, and shows moderately good 2125.
resolution of four of the five carbon atoms; free rota- W~LSON, A. J. C. (1942). Nature, Load. 150, 152.

AC 9 25