HAL Id: jpa-00205809
https://hal.archives-ouvertes.fr/jpa-00205809
Submitted on 1 Jan 1964
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Recent crystal structure determinations by neutron diffraction at Oak Ridge
George M. Brown, Henri A. Levy
To cite this version:
George M. Brown, Henri A. Levy. Recent crystal structure determinations by neutron diffraction at Oak Ridge. Journal de Physique, 1964, 25 (5), pp.469-473. �10.1051/jphys:01964002505046900�.
�jpa-00205809�
RECENT CRYSTAL STRUCTURE DETERMINATIONS BY NEUTRON DIFFRACTION AT OAK RIDGE
By GEORGE M. BROWN and HENRI A. LEVY,
Chemistry Division
Oak Ridge National Laboratory (1),
Oak Ridge, Tennessee, U. S. A.
Résumé.
2014Des structures cristallines ont été déterminées à partir des données tridimensionnelles
aux
neutrons pour le difluorure de xénon, le tétraflorure de xénon, le chlorure de baryum dihydraté l’heptafluoniobate de potassium, l’hydrate de chloral, le
sucroseet l’hydroxyde de strontium octahydraté. Les valeurs ont été relevées grace
audiffractomètre à neutrons d’Oak Ridge position-
nant automatiquement les trois angles d’Euler. Les raffinements, par la méthode des moindres
carrés, ont fourni des paramètres structuraux de haute précision.
Abstract.
2014Crystal structures have been determined from three-dimensional neutron data for
xenondifluoride,
xenontetrafluoride, barium chloride dihydrate, potassium heptafluoniobate,
chloral hydrate, sucrose, and strontium hydroxide octahydrate. Data
weretaken
onthe Oak Ridge automatic three-circle neutron diffractometer. Refinements, by the method of least squares, have yielded structural parameters of high precision.
PHYSIQUE 25, 1964,
Introduction.
-In the last two years, approxi- mately, seven different crystal structures have been determined from three-dimensional data by
the neutron diffraction group in the Chemistry
Division of the Oak Ridge National Laboratory.
Data used in determining the structures of XeF2, XeF4, BaC’2.2H20, KNbF, chloral hydrate, and
sucrose were obtained with the new automatic three-circle neutron diffractometer at the Oak
Ridge Research Reactor. The work on these struc- tures has demonstrated very clearly that auto-
matic collection of neutron data is practical. Data
of high quality can be collected in reasonable time.
The precision of the structures derived from the data is very satisfactory. It is the purpose of this paper to review the six determinations made on
data from the automatic instrument.
For Sr(OH)2.8H20, data were obtained [1] on
the crystal orienter of the automatic diffracto- meter, but before it was in automatic operation.
Refinement of the structure of Sr(OH) . BH O has
not yet been completed, and this structure will not
be discussed further here.
The structure d6terminations.
-General infor- mation on the diffractometer and its operation is presented in an accompanying paper by Busing and Levy [2]. Information on procedures followed in data collection and on data processing is given in
another paper by Brown and Levy [3].
Absorption corrections have been applied to the
data for all of the structures discussed here, except
the data in the determination of the XeF2, which
did not require correction. The corrections were
(1) Operated for the U. S. Atomic Energy Commission by Union Carbide Corporation.
computed with the program of Wehe, Busing, and Levy for the IBM 7090 computer [4].
Final refinement was carried out in every case
by the method of least squares, with adjustment
of individual anisotropic thermal parameters for
each atom. The full-matrix Fortran least-squares
program (OR FLS) of Busing, Martin, and Levy [5],
or a modification of it, was always used. The
observations were the values of F 0 2b,,., the observed
square of the structure factor. Observations were
weighted by the reciprocals of their variances
(see reference 3 for calculation of variances). Qua- lity of each refinement is indicated by specifying
the value of the discrepancy factor,
and the value of the standard deviation of an obser- vation of unit weight,
where w is the weight of an observatiou, It is the
number of observations, and p is the number of
parameters being determined. The expected value
of 6i on convergence is unity, if the errors are truly
random and correctly estimated and if the model is correct.
The Fortran function and error program
(OR FFE) of Busing, Martin, and Levy [6] was
used for calculation of bond lengths, angles, mean
vibrational displacements, and so forth, with the
attendant errors. Bond lengths were corrected
for the effects of thermal motion in some instances, according to the method of Busing and Levy, assuming the " riding " model [7].
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01964002505046900
470
TABLE 1
CRYSTALLOGRAPIIIC DATA
Data on unit-cell parameters and symmetry are presented in Table 1 for all the structures reported,
for convenient reference.
XeF2.
-The structure was determined by
H. A. Levy and P. A. Agron. For a preliminary
account, see reference [14]. A more complete des- cription of the work will appear soon [15].
The crystal specimen used was an irregular hexa- gonal platelet, - 1.5 x 1.0 x 0.5 mm, weighing
- 2 mg, enclosed in a thin-walled capillary of
vitreous silica. A total of 334 separate intensity
observations were made, for 91 independent reflec-
tions. Absorption was so slight, less than 1 %,
that no absorption corrections were necessary.
During the 10 days of data collection, the crystal
grew by sublimation at the expense of other
crystals in the same tube. The data were correc-
ted for this growth by f actors derived from repea- ted measurements of the (020) reflection.
The tetragonal body-centered lattice was indi-
cated from x-ray precession photography and trom
the neutron measurements. Inspection of neutron
intensities quickly resulted in placement of Xe at
the origin and placement ot the two fluorine atoi-ns
on the tetrad axis at (0, 0, z) and (0, 0,
-z), with
z
=2/7, in the symmetry of space group I 4/minm.
In the least-squares refinement that followed,
the parameters adjusted were the z coordinate ot
fluorine, the thermal parameters for fluorine and xenon, the neutron scattering factor of xenon, and an overall scale lactor. In the last cycle, the
value of the xenon scattering tactor from the
neutron study of XeF4 was used. The value of
0.55 X cm [18] was used for the scattering
factor of fluorine throughout the refinement.
RF2
----0.090 ; al
=1.67.
The uncorrected Xe-F bond length in the sym- metrical linear triatomic molecule is 1.984 A
6 =
0.002 A). Correction of this length for the
effect of thermal motion, computed on the reaso-
nable assumption that the fluorine atom "rides" on
the heavier xenon atom, yielded the value 2.00 A (s = 0 . 01 A) for the mean separation of Xe and F.
Each fluorine atom has one fluorine neighbor at
3.02 A and four at 3.09 as. Each xenon atom has eight non-bonded fluorine neighbors at 3.41 A,
a somewhat greater separation than the non-
bonded xenon-fluorine distances in the XeF4 struc-
ture.
XeF4.
-This work was carried out by
J. H. Burns, P. A. Agron, and H. A. Levy [15, 16].
The crystal specimen, weighing about 25 mg,
was grown inside a sealed thin-walled tube of vitreous silica, about 1 mm in diameter. To the limit sin 0 Jx
=0.76, 623 independent reflections
were recorded. In the period of data collection,
about one month, a nearly linear growth of the specimen was observed to occur, by transfer from other crystals in the tube. Corrections for the
growth were applied to the data in the same manner as in the XeF2 work. Absorption correc- tions, amounting to about 5 %, were applied.
By the time data collection was completed,
results of two independent x-ray determinations had become available [9, 17]. Least-squares refi-
nement on the neutron data was started with the
parameters of Templeton et ale [9]. The para- meters adjusted were : coordinates of the fluorine atoms (the xenon atoms are in fixed positions),
individual anisotropic thermal parameters, the
neutron scattering factor of xenon, and an overall scale factor. The 24 strongest reflections were
eventually omitted from the data set, because they clearly showed effects of extinction.
The value established for the neutron scattering
factor of xenon is 0.476 X 10-12 cm. The value 0.55 X 10112 cm was used for fluorine [17].
It is a very pleasing result that the structural pa-
rameters from the neutron analysis and those
from the two x-ray analyses are generally in good agreement. For reasons presently unknown, however, the thermal parameters of xenon are si- gnificantly lower in the x-ray case.
The distances between observed atomic posi- tions, Xe-F(1) - 1.932 A (6
=0.002 A) and Xe-F(2) = 1.939 A (6
=0.002 A), were corrected
on the assumption that the fluorine atoms " ride "
the heavier xenon atom, yielding the bond length
1.95 ak (6
=0.01 ai) for each. Angle F(I)-Xe-F(2)
is 90.0~ (6
==0.1~). Clearly the molecule has almost exactly the symmetry D4h ; it is planar by crystal symmetry.
Each fluorine atom has two intramolecular con-
tacts with fluorine atoms at 2 . 74 A and eight inter-
molecular contacts with fluorine atoms at 2.99 to 3.26 A. Non-bonded Xe-F contacts are at 3.25 and 3.22 A.
Results of an x-ray analysis of the crystal struc-
ture of the addition compound XeF2. XeF4 recen- tly carried out at Oak Ridge by J. H. Burns (2),
R. D. Ellison, and H. ~.. Levy [15, 19] are of great
interest in connection with the neutron work on
XeF2 and XeF~. The compound was originally thought to be a polymorph of XeF4, but the x-ray analysis showed that it is indeed a molecular addi- tion compound, in which the two molecules XeF2
and XeF4 retain almost exactly the same mole-
cular parameters exhibited in their individual crystal structures.
BaC12 . 2H20. - The determination of structure
was carried out by V.. M. Padmanabhan (3),
W. R. Busing, and A. H. Levy. For an abstract
of the work, which reports the atomic coordinates,
see reference [20].
A total of 1 242 independent reflections were
collected and corrected for absorption. Coordi-
nates of the Ba, Cl and 0 atoms were already
available from x-ray work [101. Hydrogen coordi-
nates that had been proposed on the basis of a proton-magnetic-resonance study [21] were found
to be inconsistent with the neutron data. Correct
hydrogen positions were obtained from a three-
dimensional Fourier synthesis, using the signs of
structure factors calculated from the heavy atom
coordinates. After preliminary refinement by
least squares, it was recognized that many of the
more intense reflections suffered from extinction
errors. Final refinement was on 961 observations
judged to be free of extinction. RF2
=0.107 ;
a1
=1.65. Standard errors of coordinates vary from 0.00~ to 0.004 A, the largest errors being, of
course, those of the hydrogen coordinates.
For each water molecule, the two hydrogen
( ~) Reactor Chemistry Division.
(1) Guest scientist from Atomic Energy Establishment, Bombay, India.
atoms and two Ba2+ ions make up an approximate
tetrahedral arrangement about the oxygen atom.
Three of the four hydrogen atoms make good hydrogen bonds to chlorine ions ; the other hydro-
gen is loosely shared between two chlorine ions.
Further discussion of the structure is beyond the
scope of this paper. It is worth noting, however,
that the structure suggests a plausible mechanism
for the unusual phenomenon of pressure twinning
that BaCI2. 2H 20 shows [22].
K2NbF7’
-This refinement has just been com- pleted by G. M. Brown, L. A. Walker (4) and
H. A. Levy.
A crystal weighing 18.5 mg was sealed inside a
thin-walled tube of vitreous silica for protection against moisture. A total of 1 754 observations
were made, of which 1 358 were independent. The
data were corrected for absorption effects, though
the corrections needed were very small, only about 3 ~/ .
Least-squares refinement was carried out very
quickly and easily, starting with the atomic coor-
dinates from the two-dimensional x-ray analysis
of Hoard [11]. In the final refinement the 67 reflections of highest intensity were omitted from the set of observations, because they clearly were subject to extinction errors. 0.089 ;
61
=1.11. Standard errors of coordinates are
about 0 . 001 r A for the niobium atom, 0 . 0016- 0 . 0025 A f or the fluorine atoms, and 0.0025 A for
the potassium ions.
The atomic coordinates are quite close to those given by Hoard, but they are now known much
more
"precisely. Hoard remarked that the NbF?
polyhedron is conveniently visualized as derived from an Nb Fs group in the form of a trigonal prism by the addition of a seventh fluorine atom through
the center of one square face, followed by the appro-
priate distortion ". The distortion is considerable,
and the structure of the complex ion is something
of a curiosity. The Nb-F bond lengths all fall in
the range 1.91 to 1.96 ak. The intramolecular F-F distances vary more widely, from 2.36 to
2 . 91 A.
Chloral hydrate, CC13CH( OH) 2. - This is the work of G. M. Brown and H. A. Levy [23, 24], to
be published in detail soon.
Three different crystal specimens, weighing 29, 6, and 2 milligrams, were used, in an attempt to
minimize extinction errors. The crystals were
enclosed in silica tubes to prevent sublimation.
Some 4 100 observations were recorded for selec- tion of data for averaging to obtain reliable values for about 2 100 independent reflections.
(4) Research participaut, Oak Ridge Institute of Nuclear
Studies, summer, 1963.
472
Approximate coordinates for the chlorine
carbon, and oxygen atoms had become available in November of 1960, from a redetermination of the chloral hydrate structure by two dimensional x-ray analysis that was in progress at that time and which later has been published [12]. As soon
as neutron data had been obtained from the two
larger crystals, a three-dimensional Fourier syn- thesis was computed with signs from calculated structure factors including only the heavy atoms, from which reasonable locations for all three H atoms of the molecule were immediately obtained.
After five cycles of least-squares refinement, the
last two of which included adjustment of indivi-
dual anisotropic thermal parameters, the value
of RF2 reached 0.066, and the value of 6i reached 1.28. The largest standard deviation of an atomic coordinate was 0.0011 A for chlorine atoms,
0.0015 A for oxygen, 0.0011 A for carbon, and
0 . 0031 A for hydrogen.
Subsequently, in a further effort to minimize extinction errors, data were collected using the
2 mg crystal for those reflections showing the highest intensities from the 6 mg crystal. Even- tually it appeared that data for the two smaller
crystals were about equally affected by extinction,
and it seemed advisable to omit entirely about
50 reflections of highest intensity. After further
refinement the value of RFI went to 0.071, but
the value of 61 decreased to 1.02. Coordinate
changes were negligible, and coordinate errors re-
mained essentially unchanged.
General features of the refined structure are, of course, the same as those found in the x-ray work, except that the nature of the hydrogen bonding is
revealed unambiguously in the neutron structure.
All structural parameters are much more precisely
determined in the neutron work.
The and -CH(OH)2 groups are in staggered
conformation about the C-C axis of the molecule.
Differences in length among chemically equivalent
but crystallographically non-equivalent bonds do
not appear significant. Bond lengths are as
f ollows : C-Cl,1. 76 A ; C-C, 2 . 55 A ; C-0,1. 39 Å ;
0-H, 0. 98 A ; C-H, 1.10 A (lengths can be repor- ted to higher precision when more reliable cell para- meters are obtained). The C-0 bond lengths is significantly lower than the normal value of ~ . 42 A.
There are quite significant differences among che-
mically equivalent valence angles, resulting pre-
sumably from the packing of the molecules.
Each oxygen atom is involved in two hydrogen
bonds. Two hydrogen bonds 0-H ... 0 link two
molecules about each of the symmetry centers at (0, 0, 0) and (0, 1/2, 1/2). Each molecule is also linked by two other bonds 0-H ... 0 to two neigh-
bor molecules in a helical chain about one of the
screw axes. The easy cleavage parallel to (100)
is consistent with the structure.
Sucrose (Cl,H22011)’
-The sucrose structure
was determined by G. M. Brown and H. A. Levy.
A preliminary report is in press [25].
Considerable effort was expended to minimize
extinction errors. Three different crystal spe- cimens were employed, weighing approxima- tely 80, 10, and 5 mg. The two smaller crystals
were immersed repeatedly in liquid nitrogen before
data were recorded from them. Averaged data
for some 2 800 independent reflections were obtai- ned by careful selection and averaging from about
5 800 individual observations. Absorption correc-
tions were applied.
The rough structure of Beevers et a2. [13] for
sucrose furnished starting coordinates for the 12 carbon and 11 oxygen atoms of the molecule.
The 14 hydrogen atoms attached to carbon atoms
were inserted at calculated positions, and refine- ment was carried out by a combination of least- squares and Fourier methods. Eventually all
atoms were located precisely. jR~ === 0.046 ;
1.15. The discrepancy factor computed
on F instead of F2 is I~F
=0.035. The standard
errors in the coordinates are as follows : 0 . 001 Q to 0.0015 A for the carbon atoms and the ether oxygen atoms ; 0.001.4 to 0.002~ A for the hydro- xyl oxygen atoms ; 0 . 0024 to 0 . 004~ A for hydro-
gen atoms attached to carbon atoms ; and 0.0023
to 0 . 0054 Å f or hydrogen atoms of hydroxyl
groups. From beginning to end of refinement,
the average shift of position of the 23 oxygen and carbon atoms was 0 . 28 A ; minimum shift, 0 . 06 A. ;
maximum shift, 0 . 91. A.
All of the 8 hydroxyl groups per molecule parti- cipate in hydrogen bonding except one. Two of
the hydrogen bonds are intramolecular. The observed cleavage parallel to (100) is neatly explained by the pattern of the hydrogen bonds.
No attempt will be made here to discuss the wealth of structural information on the sucrose
molecule made available by this determination.
Conclusion.
-The order of presentation of struc-
tures in this paper was deliberately chosen as the
order of increasing scale of the problems, as mea-
sured by the volume of data and the number of parameters to be determined, rather than the chro-
nological order in which the determinations were
made. Perhaps this may serve to emphasize the greatly increased potential of neutron diffraction
analysis accompanying automation, which makes possible for the first time the amassing of a large
volume of good data in a reasonable period. If
one accepts at face value the apparent. quality of
the sucrose determination, for example, than this
determination must be considered at least as satis-
factory as any x-ray determination done for a
crystal of even approximately the same complexity.
The results seem to speak for themselves as to
what can be expected f rom neutron diffraction
analysis in the near future.
Discussion
Dr IBERS.
-Vos corrections d’agitation ther- mique ne sont pas des corrections r6elles mais semblent indiquer les effets possibles de l’agitation thermique sur les distances interatomiques appa- rentes. 11 est clair que 1’on peut maintenant d6ter- miner ces distances avec une precision bien trop grande par rapport a notre capacité de les com- prendre, c’est-A-dire de les relier aux valeurs d’6qui-
libre. L’hypoth6se qu’un atome chevauche l’autre est diflicile a justifier dans la plupart des cas.
Dr BROWN. - Je pense que XeF 2 et XeF4 sont
des cas dans lesquels I’hypoth6se du chevau-
chement est assez bien justifiee. Cependant, il me
faut admettre que les corrections sont plut6t appro- ch6es. Je pense que K2NbF7 est un cas similaire.
Dr COPPENS.
-Avez-vous fait des calculs d’eflets d’extinction et quel crit6re avez-vous uti-
lis6 pour d6cider qu’une reflexion est affect6e par 1’extinction ?
Dr BROWN. - Nous n’avons pas fait de calculés d’extinction. Au contraire nous avons essay6 d’611-
miner toute donn6e sujette a 1’effet d’extinction.
Ce sujet sera trait6 dans ma seconde communi-
cation, en collaboration avec le Dr Levy.
BIBLIOGRAPHY [1] ZOCCHI (M.), BUSING (W. R.) and LEVY (H. A.),
unpublished.
[2] BUSING (W. R.) SMITH (H. G.), PETERSON (S. W.) and
LEVY (H. A.), J. Physique, 1964, 25, 495,
[3] BROWN (G. M.) and LEVY (H. A.), J. Physique, 1964, 25, 497.
[4] WEHE (D. J.), BUSING (W. R.) and LEVY (H. A.),
"
OR ABS, A Fortran Program for Calculating Single Crystal Absorption Corrections ", Report
N° TM-229, Oak Ridge National Laboratory, 1962.
[5] BUSING (W. R.), MARTIN (K.) and LEVY (H. A.),
" OR FLS, A Fortran Crystallographic Least- Squares Program ", Report N° TM-305, Oak Ridge National Laboratory, 1962.
[6] BUSING (W. R.), MARTIN (K.) and LEVY (H. A.),
"