Characterization of the Solid-Solid Phase Transition of Co2(CO)6(AsPh3)2

Article abstract of DOI:10.1021/ic9807240

The dimeric Co₂(CO)₆(AsPh₃)₂ molecule contains an unsupported Co-Co bond, axial arsines, and equatorial carbonyls. At room temperature, the molecule crystallizes in the R3? space group [a = 15.73(3) ?14.240- (2) ?] packing in chains along c which are hexagonally arranged in the ab plane. When the temperature is lowered, no significant changes are found in the molecular structure until a second-order phase transition [T_c = 206(3) K] occurs by doubling the c axis [at T = 123 K a = 15.233(4) ?, c = 28.440(6) ?, space group R3?]. The major molecular conformational changes related to the phase transition are the rotations of Co(CO)₃ and AsPh₃ fragments about the 3-fold axes and those of the phenyls about their As-C bond. X-ray studies at 16 different temperatures have been performed in order to describe the details of the small continuous structural changes that occur in the low-temperature phase.

Full text of DOI:10.1021/ic9807240

Inorg. Chem. 1998, 37, 6263-6268
6263
Characterization of the Solid-Solid Phase Transition of Co2(CO)6(AsPh3)2
Piero Macchi,*,† Luigi Garlaschelli,‡ Secondo Martinengo,‡ and Angelo Sironi*,†
Dipartimento di Chimica Strutturale e Stereochimica Inorganica, Universita` di Milano e centro CNR
CSMTBO, Via G. Venezian 21, 20133 Milano, Italy, and Dipartimento di Chimica Inorganica
Metallorganica e Analitica, Universita` di Milano, Via G. Venezian 21, 20133 Milano, Italy
ReceiVed June 24, 1998
The dimeric Co2(CO)6(AsPh3)2 molecule contains an unsupported Co-Co bond, axial arsines, and equatorial
carbonyls. At room temperature, the molecule crystallizes in the R3 space group [a ) 15.473(3) Å, c ) 14.240-
(2) Å] packing in chains along c which are hexagonally arranged in the ab plane. When the temperature is
lowered, no significant changes are found in the molecular structure until a second-order phase transition [Tc )
206(3) K] occurs by doubling the c axis [at T ) 123 K a ) 15.233(4) Å, c ) 28.440(6) Å, space group R3]. The
major molecular conformational changes related to the phase transition are the rotations of Co(CO)3 and AsPh3
fragments about the 3-fold axes and those of the phenyls about their As-C bond. X-ray studies at 16 different
temperatures have been performed in order to describe the details of the small continuous structural changes that
occur in the low-temperature phase.
with an authentic sample prepared according to Manning.4 The
previously unreported IR spectrum in tetrahydrofuran solution shows
Introduction
As a part of a continuing investigation of the chemistry and
stereochemistry of metal (particularly Co and Rh) carbonyl
clusters,1 we studied the substitution reactions of Co6(CO)16 with
various ligands. In most cases, degradation of the cluster
occurred to give Co4(CO)12-xLx species2 and, with ER3 (E )
P, As), dimeric species as well. In the case of AsPh3, Co2-
(CO)6(AsPh3)2 (1) was obtained as the major product. Because
the structure of this compound was assigned only on the basis
of IR data, we decided to undertake an X-ray analysis. Further,
because of the good quality of the crystals, the high symmetry
of this molecule, and the presence of an unsupported Co-Co
bond, we thought that this compound was also a good candidate
for an accurate low-temperature study of the experimental
electron-density distribution,3 in order to investigate its interest-
ing bonding features. While performing the low-temperature
X-ray experiment, we observed at T ) 206(3) K a solid-solid
reversible phase transition, which is here reported. In the high-
temperature (HT) phase, the molecular fragment is centrosym-
metric; in the low-temperature (LT) phase, the intramolecular
inversion center is lost, the “hexagonal” c axis is doubled, the
space group remains unchanged.
two bands at 1999(w) and 1980(s) cm-1 in the stretching region of the
terminal CO’s. Because this spectrum is compatible with both D3h
and D3d local symmetry of the unbridged Co2(CO)6 moiety,5 the
staggered D3d conformation of the two sets of radial carbonyl groups
was proposed as the more probable on the basis of reduction of steric
interactions.4 Our structural results fully confirm this hypothesis.
Crystals suitable for the X-ray analysis were obtained by the slow-
diffusion technique: layering cyclohexane over a saturated CH2Cl2
solution and leaving the solvents to diffuse; after a few days, several
violet crystals (with hexagonal prismatic shape) appeared at the interface
between the two phases and were then separated.
Structural Characterization. A crystal of 1 (0.2 × 0.15 × 0.15
mm) was isolated and used for the X-ray analyses. All experiments
were carried on a SMART-CCD diffractometer equipped with a nitrogen
gas stream LT device (Bruker LT3). In this apparatus, a thermocouple
measures the stream temperature inside the nozzle, but the actual
temperature on the sample can only be estimated on the basis of the
nozzle-to-sample distance. A careful test to estimate the actual
temperature on the crystal during variation of the gas flow intensity
has been performed using the cell parameters of a test crystal (LRB/
081)6 which was previously collected on a diffractometer equipped with
the Samson cryostat. Here, the crystal and the head are in thermal
equilibrium at the same temperature which was controlled by a silicon-
diode sensor whose accuracy in determining the actual temperature
reduces the uncertainty below 1°.
Experimental Section
Synthesis. 1 was formed in the reaction of Co6(CO)16 with AsPh3
in dichloromethane at room temperature and identified by comparison
Some general conditions were fixed during the experiments: graphite-
monochromatized Mo KR radiation (λ ) 0.710 73 Å); generator at 45
kV and 40 mA; 20 s/frame; detector at 2.94(1) cm from the sample;
(sinθ/λ)max ) 0.85 Å-1; ω-scan method, ∆ω ) 0.3°.
A preliminary experiment showed the reversibility of the phase
transition. Then, data at 16 different temperatures (alternatively below
and above Tc) were collected. Three kinds of experiments were
performed: (a) at T ) 299, 224, 218, 208, 199, 184, 170, and 123 K,
1000 frames in two different φ settings at negative θ angles were
collected (100 additional frames at positive θ were used for correctly
estimating the cell parameters); complete sets of unique data (up to
† Dipartimento di Chimica Strutturale e centro CNR CSMTBO.
‡ Dipartimento di Chimica Inorganica.
(1) For instance see: (a) Fumagalli, A.; Martinengo, S.; Bernasconi, G.;
Ciani, G.; Proserpio, D. M.; Sironi, A. J. Am. Chem. Soc. 1997, 119,
1450. (b) Fumagalli, A.; Martinengo, S.; Tasselli, M.; Ciani, G.;
Macchi, P.; Sironi, A. Inorg. Chem. 1998, 37, 2826 and references
therein.
(2) (a) Heaton, B. T.; Longhetti, L.; Mingos, D. M. P.; Briant, C. E.;
Minshall, P. C.; Theobald, B. R. C.; Garlaschelli, L.; Sartorelli, U. J.
Organomet. Chem. 1981, 213, 333. (b) Darensbourg, D. J.; Incorvia,
M. J. Inorg. Chem. 1981, 20, 1911. (c) Bojczuk, M.; Heaton, B. T.;
Johnson, S.; Ghilardi, C. A.; Orlandini, A. J. Organomet. Chem. 1988,
341, 473.
(4) Manning, A. R. J. Chem. Soc. A 1968, 1135.
(5) Vohler, O. Chem. Ber. 1958, 91, 1235.
(3) Macchi, P.; Proserpio, D. M.; Sironi, A., J. Am. Chem. Soc., in press.
(6) Destro, R.; Soave, R. Acta Crystallogr. 1995, C51, 1383.
10.1021/ic9807240 CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/13/1998

6264 Inorganic Chemistry, Vol. 37, No. 24, 1998
Macchi et al.
Table 1. Cell Parameters, Agreement Indexes (upon refinement), and Total Number of Reflections Collected at Each Temperature
cHT or (cLT)/2
VHT or VLT/2
reflections
collected
a
b
T (K)
a (Å)
(Å)
(Å3)
R1
unique (Rint )
299
248
233
224
218
208
15.473(3)
15.409(4)
15.392(3)
15.391(3)
15.385(4)
15.375(4)
15.359(3)
15.353(3)
15.347(2)
15.334(4)
15.333(2)
15.329(2)
15.318(2)
15.282(2)
15.267(3)
15.233(4)
14.240(2)
14.229(3)
14.228(2)
14.230(2)
14.228(2)
14.228(2)
14.227(5)
14.235(3)
14.226(2)
14.225(4)
14.234(9)
14.225(4)
14.220(4)
14.223(4)
14.220(2)
14.220(6)
2952.5(1.5)
2925.5(1.5)
2919.2(1.5)
2919.2(1.5)
2916.6(1.2)
2912.8(1.2)
2907.0(1.4)
2906.1(1.4)
2901.6(1.4)
2896.7(1.4)
2898.0(1.6)
2894.7(1.2)
2884.6(1.1)
2876.7(1.1)
2870.4(1.2)
2857.6(1.9)
0.0303
0.0342
0.0355
0.0324
0.0290
0.0295
0.0364
0.0460
0.0581
0.0384
0.0346
0.0364
0.0365
0.0428
0.0422
0.0241
10 467
2 128
1 229
6 532
8 558
12 940
4 651
15 444
2 210
9 888
19 532
2 349
15 547
5 210
1 984
14 519
3147 (0.0275)
1902 (0.0289)
1060 (0.0198)
2887 (0.0291)
3061 (0.0304)
3040 (0.0266)
3445 (0.0288)
5917 (0.0386)
1918 (0.0228)
3619 (0.0231)
5952 (0.0373)
2021 (0.0181)
6019 (0.0308)
3788 (0.0223)
1649 (0.0153)
6054 (0.0281)
202c
199c
194c
186c
184c
182c
170c
155c
143c
123c
b
2
2
2
2
2
a R1 ) Σ||Fo - Fc||/Σ|Fo|; reflections with I/σ(I) > 2.0 have been used for refinements. Rint ) Σ|Fo - Fmean |/ΣFo ; Rσ ) Σσ(Fo )/ΣFo . c For
these temperatures c/2 and V/2 are reported.
sinθ/λ ) 0.7 Å-1) were available. (b) At T ) 248, 202, 186, and 155
K, 300 frames (100 at positive θ) were taken, providing ca. 80% of
unique data. (c) At T ) 233, 194, 182, and 143 K, 150 frames (50 at
positive θ) were collected, providing ca. 50% of unique data.
Absorption correction was applied to all datasets using SADABS;7
no crystal decay was observed. A reproducible dependence of cell
axes on ω regions was found at each temperature; thus, to have an
unbiased picture of the cell dimensions dependence on temperature,
cell parameters were obtained upon least-squares fit of integration
positioned reflections contained in 150 frames (at positive and negative
θ angles) collected in the very same ω-regions at each temperature. A
scaling based on five repeated room temperature experiments was
applied to standard uncertainties (the uncertainties upon integration are
unrealistically small because of the large observations/parameters ratio,
and a scale factor of 3.0 was always applied to produce more realistic
estimated standard deviation).
Positions of Co and As atoms (both in HT and LT phases) were
determined with Patterson methods. All other atoms were located via
Fourier difference maps; hydrogens were kept fixed at constant distances
from bonded C atoms. A mean C-H distance was finally modeled to
1.08 Å on the basis of an accurate refinement (using Stewart scattering
Figure 1. Co2(CO)6(AsPh3)2 molecular structure at (a) T ) 123 K
and (b) room temperature. Ellipsoids of the non-hydrogen atoms are
drawn at 50% probability level.
factors for hydrogens) of the extensive LT data.
molecules pack in chains along c, and the chains pack
hexagonally in the ab plane; however, because neighboring
chains are shifted (by c/6 or c/3), the overall packing is trigonal.
Through examination of the shape of the “enclosure shell”
(not shown) of a Co2(CO)6(AsPh3)2 molecule, the packing motif
of the crystal is easily decoded into three fundamental compo-
nents: (a) S6 “sextuple embraces”,9 which strictly couple the
AsPh3 groups of two contiguous molecules along c (see Figure
2); (b) trigonal-helical π stackings, where each phenyl of a
AsPh3 group “interacts” with a phenyl of a surrounding chain
(see Figure 3); and (c) trigonal gears, arising from phenyl bumps
into the (six) holes generated by the staggered conformation of
the Co2(CO)6 moiety, which is somewhat assisted by
C-H‚‚‚O interactions (see Figure 3 and Table 2). All of these
features have a gearing character which strongly correlates the
movements of the surrounding groups, and this correlation is
probably at the heart of the phase transition which is described
in the following.
Anisotropic thermal parameters were assigned to all non-H atoms.
Refinements at each temperature were all stable; however, for the type
c datasets (only 50% of data) rigid body constraints for phenyl rings
were applied. For the data collected very close to Tc, refinements were
carried out in both lattices. Results of refinements for each data
collection are summarized in Table 1.
Results and Discussion
Co2(CO)6(AsPh3)2 crystallizes in the R3 space group lying
about a 3 symmetry element (b in Wyckoff notation). The
molecule contains two trigonal-bipyramidal cobalt atoms linked
by an unsupported Co-Co bond along the axial direction (see
Figure 1). The two (enantiomeric, i.e., of inverted helicity)
AsPh3 ligands are bonded in the remaining axial sites whereas
the terminal CO’s are equatorially bonded and bent away
markedly from the AsPh3 ligands, i.e., toward the opposite Co
atom (C-Co-Co ≈ 86°). The observed geometrical features
of the AsPh3 fragment fit well into the structural correlation
analysis of Ph3P-X systems:8 the 3-fold symmetry with Co-
As-C-C torsion close to 40° is in the center of the most
densely populated zone of the conformational space. The
Structural and Conformational Aspects of the Phase
Transition. When the temperature is lowered, no significant
changes are found in the molecular structure (apart from the
expected slight contraction of the interatomic distances)10 until
(7) Sheldrick, G. M. unpublished work.
(8) Bye, E.; Schweizer, W. B.; Dunitz, J. D. J. Am. Chem. Soc. 1982,
104, 5893.
(9) Dance, I. In The Crystal as Supramolecular Entity; Desiraju, G. R.,
Ed.; Wiley Chichester, 1995.

Phase Transition of Co2(CO)6(AsPh3)2
Inorganic Chemistry, Vol. 37, No. 24, 1998 6265
Below Tc, a number of reflections with half-integer l indexes
in the HT lattice gain significant intensities. This means that
the c axis is doubled. In the HT phase, there are two kinds of
inversion centers (intra- and intermolecular, respectively; see
Figure 2); however, upon refinement of the LT phase structure,
it appears that the intramolecular inversion center is lost (thus
allowing for the lowering of the molecular symmetry from S6
to C3; see Figure 2).11 Although the lattice has changed its
metric, the space group is unchanged, but the molecule now
lies about the c Wyckoff position. As is usual in solid-state
phase transitions, the HT phase is more symmetric than the LT
phase.
The major conformational changes of the Co2(CO)6(AsPh3)2
molecule, occurring when the temperature is lowered below Tc,
are the rotation of the two Co(CO)3 and the two AsPh3 fragments
about their 3-fold axes (see the O1-Co1-Co2-O2, C11-As1-
As2-C21, and C-As-Co-O torsion angles in Table 2) and
the rotation of the phenyls about their As-C bond (which
affords different conformations for the two arsines; see Co-
As-C-C angles in Table 2). There is also a shift of the whole
molecule along c which differentiates the two intrachain
stackings of the phenyl groups (that were equivalent in the HT
phase). The sextuple embrace is strongly sensitive (only) to
phenyl rotation, as can be derived from the following data: at
room temperature, φ ) 41.7°, and the As‚‚‚As distance (d) is
7.00 Å, but at 123 K, the two unequivalent embraces have φ )
36.8°/d ) 7.23 Å and φ ) 45.5°/d ) 6.76 Å, respectively.
However, because phenyl rotations occur in opposite directions
for the two arsines, the normalized c axis length (i.e., c in the
HT phase and c/2 in the LT phase) does not change substantially.
On the contrary, the packing in the ab plane is sensitive to all
of the above-mentioned changes which can all contribute to a
better intermeshing of the chain, but above all, the rotation of
the two Co(CO)3 and the two AsPh3 fragments is the key
movement which, as it opens three infracarbonyl holes (at the
expenses of the other three), allows a denser packing of the
chains, hence, the strong shortening of a (and b).
Figure 2. Molecular packing along the c axis. The HT phase has an
additional intramolecular inversion center (inversion centers are here
represented by empty squares for the HT structure and by filled circles
for the LT one) which is lost upon phase transition (thus producing
the doubling of c). The main rotational movements occurring above Tc
are also schematically drawn: the opposite rotation of the two “sextuple
embraces” about c (which involves, to a lesser extent, the carbonyls)
and the rotation of phenyl rings about their main librational axis (again
preserving the intermolecular inversion center and breaking the
intramolecular one) which causes alternated stretching and compressing
of the embraces. (For the rotations drawn here, the stretching occurs at
the central one.)
Determination of Tc and Classification of the Phase
Transition. The phase transition has been determined to occur
at Tc ) 206(3) K on the basis of the following experimental
observations:
(a) When the cell volume is plotted against T (V of the HT
phase and V/2 of the LT phase), a significant change in its slope
is observed (see Figure 4) and the two curves, fitted by linear
regression, cross at Tc ) 207 K. The very same behavior is
shown by the a axes (Table 1) while c has only a small decrease
with temperature, and no significant change in its slope (with
respect to c/2 in the LT phase) can be detected. Because the
molar volume does not show discontinuities in the region of
the transition, at variance from its first derivative, the transition
can be classified as second order (according to Ehrenfest’s
criteria).12
(b) As stated above, the loss of the intramolecular inversion
center is due mainly to the motion of phenyl and carbonyl
groups. The C11-As1-As2-C21 and O1-Co1-Co2-O2
torsion angles are constrained by symmetry to 60° in the HT
phase. In the LT phase, instead, torsion angles different from
Figure 3. Packing in the ab plane. The trigonal-helical π stacking,
where each phenyl of a AsPh3 group “interacts” with a phenyl of a
surrounding chain, has been outlined by three bold lines. In the HT
phase (bottom), the O-Co-Co-O torsions are all 60°; thus, each
molecule has six identical holes which host symmetrically the sur-
rounding phenyls. On the other hand in the LT phase (top), the two
carbonyl torsions are no longer identical (see Table 2); thus, there are
two kinds of holes: phenyl 1 bumps into the larger one while phenyl
2 is somewhat “excluded” from the smaller hole. (Note, however, that
dO1-H13 > dO2-H23.)
(11) Note that two different origin choices can be taken in the R3space
group. In the HT phase, one is centered on the intramolecular inversion
center and the other on the intermolecular center. Thus, the doubling
of the c axes can, in principle, speak for the presence in the unit cell
of two centrosymmetric molecules not related by any crystallographic
symmetry or two noncentrosymmetric molecules related by an
intermolecular inversion center.
Tc is reached. As for the cell parameters, a (≡b) strongly
decreases; but c remains substantially unchanged (see Table 1).
(10) Atomic distances were corrected for thermal libration according to:
Schomaker, V.; Trueblood, K. N. Acta Crystallogr. 1968, B24, 63.
(12) Ehrenfest, P. Proc. Natl. Acad. Sci. U.S.A. 1933, 36, 153.

6266 Inorganic Chemistry, Vol. 37, No. 24, 1998
Macchi et al.
Table 2. Behavior of Some Relevant Intra- and Intermolecular Bonding Parameters on changing the Temperaturea
torsion angles
intermolecular distances
T (K)
O1-Co1-Co2-O2
C11-As1-As2-C21
C-As-Co-O
Co-As-C-C
O-H5′ (Å)
O-H3 (Å)
299
248
224
218
208
202
199
186
184
170
155
123
60.0
60.0
60.0
60.0
60.0
60.0
60.0
60.0
34.2(1)
33.2(2)
32.9(1)
32.9(1)
33.0(1)
41.7(2)
41.7(3)
41.5(2)
41.5(2)
41.5(2)
2.66
2.63
2.63
2.62
2.61
2.61 2.61
2.60 2.60
2.59 2.60
2.60 2.60
2.60 2.60
2.61 2.60
2.60 2.58
2.48
2.47
2.47
2.47
2.47
2.46 2.46
2.46 2.46
2.46 2.47
2.46 2.47
2.45 2.48
2.45 2.49
2.44 2.49
60.0
60.0
60.0(4)
60.3(2)
62.0(1)
62.2(2)
63.4(1)
64.2(1)
65.6(1)
61.0(4)
61.2(2)
63.9(2)
64.6(2)
66.9(1)
68.3(2)
70.8(1)
33.4(2) 32.3(3)
33.2(2) 32.3(2)
33.0(2) 31.7(2)
33.8(1) 31.4(2)
33.8(1) 30.8(1)
34.4(2) 30.4(2)
34.7(1) 29.5(1)
40.8(5) 41.9(5)
41.1(4) 41.8(4)
39.8(4) 43.0(4)
39.6(3) 43.3(2)
38.5(2) 44.0(2)
37.9(3) 44.5(3)
36.8(1) 45.5(1)
a ”Long” data collections only. All intermolecular distances reported have uncertainty below 0.01 Å.
Figure 5. C-As-As-C and O-Co-Co-O torsion angles are fixed
to 60° by symmetry in the HT phase (where the molecular symmetry
is 3) while they are unconstrained in the LT phase. Here, R (i.e., the
angular shift of C-As-As-C from 60°) and ø (that of O-Co-Co-
O) are plotted. Through quadratic fitting, Tc ) 206 and 203 K can be
estimated, respectively.
and ø change continuously and drop to zero at the critical
temperature, the transition is indeed second-order within Lan-
dau’s criteria (the two phases at Tc are identical). Through the
use of a quadratic regression, Tc ) 206 and 204 K are estimated
for R and ø, respectively.
(c) The absolute value of the Co-As-C-C angles must be
identical for the two arsines in the HT phase (≈42°, nearly
independent from T; see Table 2). In the LT phase, instead,
the two independent Ph rings have different angles; their
difference, |φ|, can be treated as was done before for R and ø
and, upon regression, Tc ) 206 K was obtained.
According to Landau and to the mean-field theory,14 η ÷
(Tc - T)â, with â ) 0.5, in phase transitions of minerals,
deviations are found only for temperatures very close to Tc.14c
Here, log/log plots for the selected quantities (R, ø and φ) give,
by linear regression treatment, an averaged â ) 0.73(3), which
means that there is not a linear dependence between η and R,
ø or φ.
(d) Reflections with odd l indexes in the LT phase are
monitored at different temperatures (peak integrations of the
HT phases have been performed on the LT lattice as well; thus,
the HT forbidden intensities are also available). At T ) 199
K, 190 of these still have significant intensity15 (upon merging
Figure 4. Properties of the crystals plotted as a function of the
temperature. On the bottom, the cell volumes of the two phases are
reported (note the discontinuous change in the slope that occurs at the
transition temperature, which can be estimated by the temperature at
which the two linear fittings cross, Tc ) 207 K). The central plot
represents the changes of librations of phenyls about their main
librational axis (As-C); the regular and linear dependence on T suggests
that an order-disorder transition can be safely excluded. On the top,
Ueq are grouped for heavy (Co, As) and light (C, O) atoms and fitted
(linearly and quadratically, respectively) on T. Note the abrupt change
that occurs for Ueq of light atoms at Tc, which might be related to the
fact that the heavy atoms are less sensitive to the intermolecular
environment which is much more perturbed than the intramolecular
one by the phase transition.
(13) Actually, in a displacive phase transition (see the discussion below
for the classification of this transition) η ) ∑juj where uj is the shift
of the jth atom from the symmetric HT position.
(14) (a) Landau, L. D.; Lifshitz, E. M. Statistical Physics; Addison-Wesley:
Reading, MA, 1958. (b) Sirotin, Y. I.; Shakolskaya, M. P. Funda-
mentals of Crystal Physics; Mir Publishers: Moscow, 1982. (c) Dove,
M. T. Am. Mineral. 1997, 82, 213.
60° are allowed (at T ) 123 K, they are 70.8° and 65.6°,
respectively). The shifts of these two torsions from their HT
value (R and ø in Figure 5) are related to the order parameter
(η)13 within the Landau theory14 of second-order phase transi-
tions, being null by symmetry in the HT phase. Because R

Phase Transition of Co2(CO)6(AsPh3)2
Inorganic Chemistry, Vol. 37, No. 24, 1998 6267
and correcting the datasets16), but at T ) 208 K, only 10
reflections have I/σ(I) > 2.0.17 Odd l-index reflections are on
average less intense than even l-index ones because the reduction
in symmetry is due mainly to phenyls and carbonyls (Co and
As still being located on the 3-fold axes and very close to their
original position in the HT phase); however, this effect decreases
with sinθ/λ because high-order reflections are more sensitive
to small shifts from the lost symmetry, which also affects Co
and As, whose contribution to diffraction is overwhelming at
high angles.
Additional observations support the phase-transition clas-
sification. A differential scanning-calorimetry (DSC) trace did
not show an exothermal-endothermal peak in the 173-298 K
region but only the sudden change in slope of the curve (centered
about 210 K, 25 K wide; scan speed18 40° min-1) expected for
second-order transformations.19 The LT group is a translational
subgroup of the HT one (HT and LT space groups are identical
while the crystal lattice changes, according to cLT ) 2cHT).
It is also worth noting that, because the point group symmetry
(3) in this phase transformation is preserved, there is only one
diffraction domain also in the LT phase, at variance from what
happens in transformations without conservation of the crystal
point group [for which the loss of one symmetry operation
produces n kinds of domains (n is the multiplicity of the
symmetry operator lost)].
At room temperature, the phenyl ring libration about the
As-C axis is quite large (ca. 70 deg2; see Figure 4), and the
thermal motion analysis of the AsPh3 as a rigid group shows
that a libration about the c axis is also present (ca. 20 deg2).
According to the theory of displacive phase transitions,14c it is
a vibrational (soft) mode, antisymmetric with respect to the
symmetry operator lost in the phase transition, that is responsible
for the observed atomic displacements. Indeed, the major
conformational changes observed in the phase transition can
be considered as movements along some20 of the normal modes
implied by the above librations. A comparison with the phase
transitions observed for poly(phenyls)19c,21 (order-disorder in
p-terphenyl, displacive in biphenyl) might be helpful. In
general, an order-disorder phase transition implies a multiwell
potential surface describing the motion of the HT phase
disordered atoms; for p-terphenyl, a harmonic double-well model
was supported by the reasonable temperature dependence of the
librations around the molecular axis for the (two) disordered
central rings against the abnormally large, temperature-
independent libration (260 deg2 for p-terphenyl at room tem-
perature) “observed” for a single ordered ring.21d,f,22 Here, a
double-well model (and the implied order-disorder phase
transition) can be safely excluded on the basis of the “correct”
behavior of the phenyl ring librations (which are close to that
of an ordered phenyl; see Figure 4) and also on the basis of the
continuous change in R, ø and φ. In this context, we notice
that this libration lacks a sudden change at Tc, while the Ueq
-
(T) functions, in particular those of the lightest atoms (C and
O), cannot be differentiated at Tc (see Figure 4).
Conclusions
The phase transition reported here can be classified as second-
order according to Ehrenfest (because the first derivatives of
chemical potential are continuous at Tc) and Landau (because
the two phases are identical at Tc), antiferrodistorsiVe (because
there is a loss of translational symmetry), antiferroic23 (because
there is conservation of the point group symmetry of the crystal),
a type I Christy’s transition24 (because it shows a group-
subgroup relationship between phase symmetries, corresponding
to a unique, irreducible representation of the higher symmetry),
and displaciVe (because it is caused by only small shifts of atoms
from their HT symmetric positions, and an order-disorder
mechanism has been proved not to occur).
Solid-solid phase transitions in molecular crystals are often
considered “unwanted” phenomena,25 disturbing otherwise
simple structural characterizations. Indeed, many of these have
been observed by crystallographic methods; however, an ac-
curate characterization (with attempts to determine the transition
temperature and the structural changes that occur in the
proximity of the transformation) is rare.26 Here, we have
reported on the phenomenology of a second-order phase
transition which possibly relates HT phase-active internal
motions to the three observed (antisymmetric) deformation paths
of the LT phase. From this point of view, we may state that a
good reason for studying the structural details of phase
transitions is that of adding to the conventional analysis of the
anisotropic displacement parameters some information on the
correlations between the displacements of the individual atoms
which would be otherwise lost.
(15) Note that at T ) 123 K, ca. 2000 unique odd l reflections have I/σ(I)
> 2.0.
(16) It has been demonstrated that a correction for the nonfiltered λ/2
intensities has an insignificant effect on the results of structure
refinements and accurate ED analysis. See: (a) Kirschbaum, K.;
Martin, A.; Pinkerton, A. A. J. Appl. Crystallogr. 1997, 30, 514. (b)
Macchi, P.; Proserpio, D. M.; Sironi, A.; Soave, R.; Destro, R. J. Appl.
Crystallogr. 1998, 583. However, λ/2 contamination has its worst effect
in determining the correct cell parameters and space groups (because
systematic absences may be violated); thus, a careful correction has
been applied to all datasets here.
Acknowledgment. The authors are thankful to Dr. M. Di
Pasquantonio (Gruppo di Supporto Analitico Perkin-Elmer) for
(17) It is worth noting that some of these forbidden intensities (usually for
large θ) are still present even for temperature well above the estimated
Tc; besides the difficulties in integrating these very weak and spread
reflections (I/σmax ) 3.5), this phenomenon cannot be neglected. The
hypothesis of a hysteresis of the LT phase can be safely excluded on
the basis of the second-order character of the transition and because
they have been also observed at room temperature for a new crystal
(before lowering the temperature). It seems to be more reasonable
that these intensities are superstructure reflections due to a diffuse
scattering originated by long-range correlated motions, which violate
the local symmetry.
(18) The nature of a second-order phase transition in a molecular crystal
is such that only a high scan speed can enhance the occurrence of the
phase transformation in the DSC trace; of course, the usage of such
a high rate reduces the instrumental accuracy in the determination of
the critical temperature.
(21) (a) Badour, J. L.; Delugeard, Y.; Cailleau, H. Acta Crystallogr. 1976,
B32, 150. (b) Charbonneau, G. P.; Badour, J. L. Acta Crystallogr.
1976, B32, 1420. (c) Charbonneau, G. P.; Delugeard, Y. Acta
Crystallogr. 1977, B33, 1586. (d) Badour, J. L.; Cailleau, H.; Yelon,
W. B. Acta Crystallogr. 1977, B33, 1772. (e) Badour, J. L.; Delugeard,
Y.; Rivet, P. Acta Crystallogr. 1978, B34, 625. (f) Baudour, J. L.;
Sanquer, M. Acta Crystallogr. 1983, B39, 75. (g) Badour, J. L. Acta
Crystallogr. 1991, B47, 935.
(22) Brock, C. P.; Schweizer, W. B.; Dunitz, J. D. J. Am. Chem. Soc. 1985,
107, 6964.
(23) Salje, E. H. K. Phase Transition in Ferroelastic and Co-Elastic
Crystals; Cambridge University Press: Cambridge, 1993.
(24) (a) Christy, G. A. Acta Crystallogr. 1993, B49, 987. (b) Christy, G.
A. Acta Crystallogr. 1995, B51, 753.
(25) (a) Dunitz, J. D. Acta Crystallogr. 1995, B51, 619. (b) Dunitz, J. D.;
Bernstein, J. Acc. Chem. Res. 1995, 28, 193.
(19) (a) Pope, H. I.; Judd, M. D. Differential Thermal Analysis; Heyden:
London, 1990. (b) Threlfall, F. L. Analyst 1995, 120, 2435-2460. (c)
Cailleau, H.; Baudour, J. L.; Meinnel, J.; Dworkin, A.; Moussa, F.;
Zeyen, C. M. E. Faraday Discuss. 1980, 69, 7.
(26) For a recent study, see: (a) Destro, R. Chem. Phys. Lett. 1997, 275,
463 and other work in progress. (b) Dunitz, J. D. Pure Appl. Chem.
1991, 63, 177. (c) Jagarlapudi, A. R.; Sarma, P.; Dunitz, J. D. Acta
Crystallogr. 1990, B46, 784. (d) Yang, Q. C.; Richardson, F. R.;
Dunitz, J. D. Acta Crystallogr. 1989, B45, 312.
(20) No phase is conveyed by thermal factors.

6268 Inorganic Chemistry, Vol. 37, No. 24, 1998
Macchi et al.
for the room temperature and the T ) 123 K structures, and DSC trace
obtained at 40°/min (12 pages). Ordering information is given on any
current masthead page.
the DSC analysis and to Prof. R. Destro for providing a suitable
crystal for calibration of the LT device. Prof. N. Masciocchi,
Prof. P. Sozzani, Dr. D. M. Proserpio, and Dr. E. Corradi are
also thanked.
Supporting Information Available: Tables listing detailed crystal-
lographic data, atomic positional parameters, bond lengths and angles
IC9807240