Part I

ELECTRON DIFFRACTION

Introduction to Part I: Electron Diffraction

BY L. E. SUTTON

In this Volume there are four chapters on topics concerning electron diffraction studies on gases and vapours.

The first, as usual, is one on Results, covering structure determinations published during the period October 1973 to August 1974 and written by Dr. D. W. H. Rankin. This is smaller than those of previous years and so is less elaborately sub-divided. Although fewer results are reported, their total interest remains high. Among the new developments are some determinations of the structures of unstable species and conformational studies of large molecules. However, there is no lack of interest in the more usual measurements on newly synthesized compounds and in the re-determinations of the structures of simple molecules with much improved accuracy. Some of the measurements on large molecules throw light on two long-standing questions viz. whether there are definite radii for sp, sp2 and sp3 hybridized carbon atoms, and if conjugation affects the length of a bond between two sp2 hybridized carbon atoms.

Chapter 2 complements that on Theory and Accuracy by Dr. H. M. Seip in Volume 1, for it is the first of two on Developments in Apparatus which is a key matter in the practice of electron diffraction studies. This year we have a review of developments in Canada, Japan, and the United States of America by Professor Richard L. Hilderbrandt. In Volume 4 we hope to complete the picture by reviewing the parallel developments in Europe (including the U.S.S.R.).

Chapter 3 is an exploration of a borderline subject. What can we learn from measurements on a substance in both the vapour phase and the crystalline form? Dr. Brian Beagley has brought together a large number of results from the very scattered literature so that at last we can see what kinds of specific question there are to answer and how far this can be done. Much of interest emerges, especially for inorganic compounds.

Chapter 4 is concerned with the technique of data reduction. In the early determinations it was commonplace to use information gained from studies on simpler molecules to help in the analysis of more complex ones e.g. C — H bond lengths were usually assumed. This was done in a very simple-minded fashion. Now Professor L. S. Bartell, Dr. D. J. Romenesko and Dr. T. C. Wong bring this procedure up-to-date by showing how it can be done in a much more sophisticated and powerful way.

In the Introduction to the Electron Diffraction Part of Volume 1, which was written in late 1972, I said that it was hoped that a bibliographical volume on vapour-phase structure determinations by diffraction and spectroscopic methods, compiled by Mrs. Olga Kennard's group at Cambridge, would appear in the summer of 1973. Production of this volume has, alas, proved far more difficult than was anticipated; and the current news is that it may appear at the end of this year or, more likely, in 1976.

In conclusion, I have once again the pleasant duty of thanking the contributors to this Section for keeping to deadlines and generally being very helpful.

1 Electron Diffraction Determinations of Gas-phase Molecular Structures

1 Introduction

This chapter reports the results of structure determinations by electron diffraction published during the period October 1973 to August 1974 inclusive. Some 90 relevant papers appeared during this time, compared with ca. 130 included in the previous volume of this report (ref. 1, Part I), which covered rather more than a year. It seems that this is one area of chemistry where the number of publications is not subject to the galloping inflation that afflicts most others. Perhaps this is due partly to the limited number of molecules that are suitable candidates for study by electron diffraction. Indeed it is instructive to consider the types of molecule studied. Figure 1 shows the distribution of compounds investigated in terms of number of atoms per molecule. Most studies fall into one of five groups: (i) studies of molecules that are very reactive, or are only obtainable as gases at high temperatures. Most of these have not had their structure determined before, and are simple molecules, with 3-7 atoms; (ii) redeterminations of known structures of simple molecules, often by the combination of microwave and electron diffraction data, as well as vibrational analyses, giving very precise structural parameters; (iii) routine structural studies of compounds that may have been synthesized only recently; (iv) studies of the effects of neighbouring substituents or nearby multiple bonds on particular bond lengths in organic molecules; (v) conformational studies of large, usually organic, molecules, often combined with calculations of energies for different conformers.

With this wide front on which progress is being made, it is obvious that the stage has not yet been reached where, as one worker in the field put it (seven years ago!) – 'electron diffraction will be finished in three years time – all the volatile molecules will have been done'.

It is also worth noting the range of scattering data used in structure determinations. Figure 2 shows the distribution of the maximum s values used for 96 compounds. Although it was possible in one case to get data out to 60 Å-1, in slightly more than half of the studies the data used did not go even as far as 30Å-1. Thus, although there may be particular reasons for limiting the s range in some cases, there does seem to be room for a general raising of standards, either by improvement of equipment or by making extra effort in collecting data.

The results quoted in the following pages are in the same form as those in the original papers. Distances are therefore rs values, unless stated otherwise, and errors quoted (in brackets) are estimated standard deviations, expressed in terms of the least significant digit given. Any conclusions and comparisons with previous work mentioned in this Chapter are taken from the reports being considered, unless stated otherwise, and references to earlier work will not be given here.

2 Main-group Inorganic Compounds

Groups II and III. — The structures of molecules containing cyclopentadiene groups have long been of interest, and recently determined ones are no less remarkable than earlier ones. Dicyclopentadienyl-magnesium and -chromium are both sandwich compounds with the metal atom centrally placed, unlike in the beryllium compound, which has the metal atom closer to one ring than the other. Both have the eclipsed (D5h) conformation. The magnesium compound has a much longer metal–carbon distance [2.339(4) Å] than the chromium one [2.169(4) Å], and a slightly but significantly smaller ring, the C–C distances being 1.423(2) and 1.431(2) Å. It is also pointed out that the hydrogen atoms in the chromium compound are slightly bent [2.9(11)°] towards the metal, as in ferrocene, whereas in the magnesium compound the evidence suggests that they are bent away from the metal [-1.0(16)°].

The C5H5Be groups in cyclopentadienyl beryllium-bromide and -acetylide have the same dimensions as they have in related compounds. The Be–Br bond length in the bromide [1.943(15)Å] is slightly greater than that in BeBr2 [1.92(2) Å], while the Be–Cl bond in C5H5BeCl is 0.08 Å longer than those in BeCl2. This is interpreted in terms of dative π-bonding in the dihalides, to a larger extent in the chloride than in the bromide. This π-bonding is not possible in the cyclopentadienyls, in which the beryllium 2p orbitals are used in bonding with the C5H5 group. The Be–C (acetylenic) bond length in C5H5BeCCH [1.634(8) Å] is as would be expected from the bond in cyclopentadienyl methyl beryllium, with an allowance for the change from sp3 to sp hybridization of the carbon atom.

In cyclopentadienyldimethylaluminium the ring is neither pentahapto, as in the examples just mentioned, nor monohapto, as in Group IV cyclopentadienes. Four C8 structures fit the data, but CNDO/2 calculations favour the form in which the Al atom is displaced from the C5 axis towards the mid-point of one C–C bond, and with the methyl group carbon atoms in the mirror plane. The ring rotational barrier is calculated to be ca. 5 kcal mol-1. The principal parameters are: Al-ring (perpendicular) 2.10(2); Al–ring axis 0.99(10); C–C 1.422(2); Al–C (methyl) 1.952(3) Å.

Yet another study of beryllium borohydride has been published. This time a mainly linear heavy-atom structure is proposed, with triple boron–hydrogen–beryllium bridges, in agreement with work published in 1946, but at variance with some 1968 results. The Be–B distance is given as 1.790(15) Å, with a maximum possible difference between the two such distances of 0.10 Å. It seems to the authors that there may be two structural forms of this compound in the gas phase, one linear and one triangular: the proportions present in a sample of the vapour depend on its history – how it was vaporized, and the time spent in the vapour phase.

Tris(methylthio)borane has an essentially planar heavy-atom skeleton and a threefold axis of symmetry, with bond lengths of 1.805(2) (B–S) and 1.825(3) Å, (S-C) and angles of 104.5(3)° at sulphur. In contrast, the B–S bond in methylthiodimethylborane is only 1.779(5)Å, although the C–S bond is the same length [1.825(4)Å]. This is accounted for by considering π-bonding between sulphur lone-pair electrons and the vacant boron 2p orbital. The maximum π-bond order in tris(methy1thio)borane is 1/3, whereas in methylthiodimethylborane there can theoretically be a π-bond order of 1. The latter compound also shows interesting effects due to steric crowding. The heavy atoms are nearly coplanar, but the S-methyl group is bent away from the neighbouring C-methyl group [[??]BSC = 107.2(10)°], and the C-methyl groups are also bent away from the S-methyl group, with [??]SBC = 124.0(8) (cis) and 115.3(6)° (trans).

Two carbaboranes with high symmetry have been studied, and show some remarkable differences in bond lengths, which are summarized in Table 1. Thus in 1,5-dicarba-closo-pentaborane(5) (1), which has D3h symmetry, the B-B bonds are very much longer, and the B–H and B–C bonds are much shorter, than in 1,6-dicarba-closo-hexaborane(6) (2), which has D4h symmetry. It is suggested that the bonding in the pentaborane is more closely described by classical models, with little boron-boron overlap, but strongly bonded C–BH–C units.

1,3-Dimethyl-2-chlorodiazaboracyclopentane (3) has an essentially planar ring, with bond lengths of 1.532(9) Å (C–C), 1.455(2) Å (N–C), 1.41 3(3) Å(B–N), and 1.770(4) Å(B–Cl). The B-N bond is therefore longer than that in dimethylaminodichloroborane (1.379 Å), much shorter than that in the boron trichloridetrimethylamine adduct (1.575 Å), and comparable to that in borazine (1.435Å). The B–Cl bond is said to have little double-bond character.

There continues to be interest in the structures of complexes of aluminium, including polymeric gas-phase forms of alkyl aluminium derivatives. Preliminary results have been published for the complexes of aluminium trichloride with ammonia and trimethylamine. These have Al–N bond lengths of 1.996(19) and 1.945(35) Å, respectively, both shorter than the values of 2.06 and 2.10 Å found in the trimethylamine complexes of aluminium–trihydride and -trimethyl. The Al–Cl bond lengths are 2.100(5) and 2.121(4) Å, and ClAlCl angles are 116 and 114°. These values fit well between those for AlC3 (2.06 Å, 120°) and AlCl-4 (2.13Å, 109.5°). The trimethylamine complex has an N-C bond length of 1.516(12) Å, longer than in the other aluminium–trimethylamhe complexes (1.476 and 1.474 Å) and in trimethylamine itself (1.454Å). It seems, perhaps not surprisingly, that as the strength of a complex is increased, the Al–N bond gets shorter, and neighbouring bonds such as Al–CI and N–C get longer.

Dimethylaluminium chloride dimer has bridging chlorine atoms, and D2h symmetry. The Al–C bonds [1.935(4) Å] are shorter than the terminal ones in trimethylaluminium dimer [1.957(3) Å], and the CAlC angle is wider, at 126.9(8)°. This is attributed to higher s character in the Al–C bonds in the chloride. The long Al–Cl bonds in the bridge [2.303(3) Å compared with 2.252(4) in Al2Cl6] are said to arise from the good acceptor properties of Me2AlCl.

Dimethyl-methoxyalumhium trimer has a non-planar Al3O3 ring, for which C3υ symmetry was assumed, and oxygen atoms that are three-co-ordinate, the bonds being co-planar. In this compound the Al–C bonds are of normal length [1.957(3)Å]. The Al–O bond length, 1.851(3) Å, lies mid-way between values accepted as normal for a single covalent bond (1.676 Å) and a dative bond (2.02 Å). This is consistent with the results found for the Al–N bond in dimethylaminodimethylaluminium dimer.

Trimethylgallium vapour at 338 K contains no dimer, unlike trimethylaluminium. The Ga–C bond length (rg) is 1.967(2) Å, and the apparent CGaC angle is 118.6(4)°. Thus, if the skeleton is truly planar, the C···C shrinkage is 0.021(10) Å. The methyl groups were shown to rotate, but it was impossible to decide to what extent this rotation is restricted.

Three studies of Group III compounds that one would not normally consider to be volatile at all have been published. One presents data for the sub-oxides of gallium, indium, and thallium, obtained at temperatures of 1300, 1300, and 900 K, respectively. Data for aluminium sub-oxide at 2400 K were also analysed. Metal-oxygen bond lengths of 1.72(1), 1.82(1), 2.00(1), and 2.15(1) Å are derived for the four metals, but of more interest are the apparent (uncorrected for shrinkage) MOM angles, which are 144(5), 140(5), 144(5), and 133(5)° for Al2O, Ga2O, In2O, and Tl2O, respectively. The bending frequencies of these molecules are unknown, but the authors conclude that if the frequencies are between 100 and 200 cm-1 the bond angles will be between 150 and 180°, and that if the frequencies lie below 100 cm-1, then the molecules are probably linear. From a study of the asymmetry of the In ... In peak for In2O, they conclude that this molecule is linear, with a bending frequency near 75 cm-1.

At 700 K, 80% of the vapour of thallous fluoride is dimer, with some trimer and tetramer. The dimer has a linear FTlTlF structure, with the bond lengths being 2.290(4) Å for Tl–F and 3.678(3) Å, for Tl–Tl.

Thallous sulphate at 1000 K has the same D2d symmetry structure as alkali-metal sulphates and molybdates. The sulphur atom is roughly tetrahedrally coordinated, with two OSO angles of 107.5(40)°, and S–O bonds 1.48(2) Å long (rg). Each of these pairs of oxygen atoms is co-ordinated to a thallium atom, with Tl–O distances of 2.41(2) Å and OTlO angles of 59.0(25)°. The Tl···S distance is thus 2.97(3) Å, close enough for there to be some direct bonding interaction.

Group IV. — The C–Br bond in dibromomethane (which for present purposes is inorganic) is 1.921(2) Å long, and the BrCBr angle in the molecule is 113.2(4)°. The ra bond length compares with a recently determined microwave value (rs) of 1.938 Å. The authors compare the C–Br bond length with those in CH3Br [rs = 1.939(1); re = 1.933(2) Å], CHBr3 [r0 = 1.930(3)Å], and CBr4 [rg = 1.942(3) Å]. Despite the fact that differently defined distances are available in each case, it can be seen that there is no very significant trend as the number of bromine atoms increases, unlike in the case of the fluoromethanes, which have C–F bond lengths in the range 1.32 — 1.38 Å. It is suggested that crowding of the bromine atoms prevents any contraction.

In the chloro- and bromo-germanes contraction of Ge–halogen bonds with increasing halogen substitution does occur. In dichlorogermane, the Ge–CI bond distance is 2.130(3) Å, compared with 2.113(3) Å in GeCl4 and 2.148(3) Å in GeH3C1, while in dibromogermane the Ge-Br distance is 2.277(3) Å, compared with 2.23(3) Å in GeBr4 and 2.298(3) Å in GeH3Br. The ClGeCl and BrGeBr angles of 107.2(5) and 108.4(4)° also suggest that direct Cl···Cl or Br···Br interactions are not important.

Two studies of silicon tetrafluoride enable comparisons of fluorosilanes to be made, as reliable structures are now available for all members of the series. One gives Si–F and F···F distances of 1.555(2) and 2.534(3) Å, while the other quotes values of 1.552(2) and 2.534(3) Å – good agreement. The amplitudes of vibration quoted for the silicon-fluorine bond differ considerably, being 0.029(7) and 0.0434(18)Å. The latter value may well be more reliable, being based on data extending to 50 Å -1. In the former case, anharmonicities were also refined, but the results are not significant. However, the agreed Si–F distance fits well in the series 1.565(5) Å (r0 for SiHF3), 1.577(1) Å (r0 for SiH2F2), and 1.593(3) Å (r0 for SiH3F).

The Si–F bond length in another fluorosilane, trifluorosilylphosphine, is 1.571 (2) Å, and also fits into this series. This molecule has an Si–P bond length of 2.207(3) Å, compared with the value of 2.248(3) Å in trisilylphosphine. There is therefore a shortening of ca. 0.03 Å on replacing an SiH3 group by an SiF3 group: the corresponding shortenings for silyl and trifluorosilyl amines and ethers are ca. 0.06 Å. The molecule adopts a staggered conformation, and there is some evidence that the symmetry of the PSiF3 unit is Cs and not C3υ, with the fluorine atoms being bent towards the hydrogen atoms. ]CH