Multiple Bonds Between Metal Atoms / 01-Introduction and Survey

Multiple Bonds Between Metal Atoms

2

Chapter 1

One of the aspects of this overall development of multicenter transition metal chemistry obviously constitutes an innovation with respect to the entire science of chemistry, namely, the recognition that there exist chemical bonds of an order higher than triple. The existence of quadruple bonds was first recognized in 1964, and since then more than a thousand compounds containing them have been prepared and characterized with unprecedented thoroughness by virtually every known physical and theoretical method, as well as by a wide-ranging investigation of their chemistry.

It is especially to be noted that compounds containing quadruple bonds are in most cases not at all exotic, unstable, or difficult to obtain. On the contrary, many of them can be (and are) easily prepared by undergraduate chemistry students and they ‘live out in the air with us’. Perhaps the most astonishing thing about this chemistry is that it was discovered so late.

1.1.2 Prior to about 1963

It is well to begin with the following observation. Werner, of course, recognized the existence of polynuclear complexes and, indeed, he wrote quite a number of papers on that subject.2 However, the compounds he dealt with were regarded (and correctly so) as simply the result of conjoining two or more mononuclear complexes through shared ligand atoms. The properties of these complexes were accounted for entirely in terms of the various individual metal atoms and the local sets of metal-ligand bonds. No direct M–M interactions of any type were considered and the concept of a metal-metal bond remained wholly outside the scope of Wernerian chemistry, even in polynuclear complexes.

Before Werner’s time, however, there were a few compounds in the literature that could not be accommodated correctly by the coordination theory. The earliest was chromous acetate, to which we shall return later (p. 10). In the period 1857-61, the Swedish chemist Christian Wilhelm Blomstrand3 and co-workers investigated the dichloride and dibromide of molybdenum and found them to have some surprising properties. For example, only one third of the halide ions could be precipitated with Ag+, thus indicating that the smallest possible molecular formula is Mo3X6. Werner himself in the several editions of his Neuere Anschauungen auf dem Gebiete der Anorganischen Chemie proposed the following formulation:

XX

Towards the middle and end of Werner’s life, further discoveries inconsistent with his theory were made. From 1905 to 1910 Blondel and others4 reported dinuclear PtIII compounds, which we now know to contain Pt–Pt bonded [Pt2(SO4)4]2- ions. In 1907, ‘TaCl2υ2H2O’ (which, as shown below, was later correctly formulated as Ta6Cl14υ7H2O) was reported.5 During the 1920s Lindner6 and others attempted to account for the composition of these and other compounds by imaginative (but chimerical) polynuclear structures in which metal-metal bonds were not included.

It was only with the advent of X-ray crystallography and its evolution into a tool capable of handling reasonably large structures that the existence of non-Wernerian transition metal chemistry could be recognized with certainty and the character of the compounds exemplifying it disclosed in detail. The first such experimental results were provided by C. Brosset,7 who showed that the lower chlorides of molybdenum contain octahedral groups of metal atoms with Mo–Mo distances even shorter (~2.6 Å) than those in metallic molybdenum (2.725 Å). Brosset’s publications did not, apparently, stimulate any further research activity.

It was also Brosset8 who showed that K3W2Cl9 contained a binuclear anion, [W2Cl9]3-, with the tungsten atoms so close together that “[t]hey are, apparently, within these pairs, in

Introduction and Survey 3

Cotton, Murillo and Walton

some way bound together.” This promising insight was not pursued.

In 1950, an X-ray diffraction experiment, albeit of an unconventional type carried out on aqueous solutions, showed that Ta6Cl14ʷ7H2O and its bromide analog, as well as the corresponding niobium compounds, also contain octahedral groups of metal atoms9 with rather short M–M distances (~ 2.8 Å). As before, these remarkable observations did not lead to any further exploration of such chemistry.

It was not until 1963, in fact, that attention was effectively focused on non-Wernerian coordination compounds. It was observed at about the same time in two different laboratories10,11 that ‘ReCl4−’ actually contains triangular Re3 groups in which the Re–Re distances (2.47 Å) are very much shorter than those (2.75 Å) in metallic rhenium. In one report10 not only was the molecular structure described very precisely, the electronic structure was discussed in detail, leading to the explicit conclusion10 that the rhenium atoms are united by a set of three Re–Re double bonds. This work was important because it was the basis for:

1.the first explicit recognition that direct metal–metal bonds can be very strong and can play a crucial role in transition metal chemistry, and

2.the first formal recognition that there is an entire class of such compounds to which the name metal atom cluster compounds was then applied.12,13

In [Re3Cl12]3− it was first shown that metal–metal bonds may be multiple, since the MO analysis10(a),12 of this cluster clearly shows that there are six doubly occupied bonding MOs covering the three Re–Re edges of the triangle, thus giving the MO equivalent of double bonds.

It should be noted that during the period of time just considered there were developments in the field of metal carbonyl chemistry that also led to the consideration of direct metal–metal bonds as stereoelectronic elements of molecular structure. In 1938 the first evidence for the structure of a polynuclear metal carbonyl compound, Fe2(CO)9, was obtained by X-ray crystallography. To account for the diamagnetism of the compound, it was considered necessary to postulate a pairing of two electron spins, each of which formally originated from a different metal atom. For many years it was taken as obvious that there exists an Fe–Fe bond. The structural integrity does not require such an assumption because there are three bridging carbonyl groups. Today there are convincing (though not entirely conclusive) theoretical arguments in favor of spin coupling via the carbonyl bridges without direct Fe–Fe bonding. It was not until 1957, with the determination of the Mn2(CO)10 structure,14 that unequivocal evidence for metal–metal bond formation in metal carbonyls was obtained.

1.2How It All Began

1.2.1 Rhenium chemistry from 1963 to 1965

By mid-1963, further studies of the chemistry of the trinuclear cluster anion [Re3Cl12]3- had led to the recognition that the trinuclear Re3 cluster with Re–Re double bonds was the essential stereoelectronic feature of much of the chemistry of rhenium(III), particularly that which used the so-called trihalides as the starting materials. Both the chloride and bromide of ReIII had been shown to contain these Re3 clusters.15

However, it was precisely the use of these ReIII halides as starting materials that posed a practical problem, since their preparation is tedious and time consuming. The idea of obtaining the trinuclear complexes by reduction in aqueous solution of the readily available [ReO4]− ion to give, for example, [Re3Cl12]3− was very attractive. The devising of such an aqueous route into trinuclear ReIII chemistry was regarded at MIT as perhaps the one remaining task to be carried out before leaving the field of ReIII chemistry. During the autumn of 1963, Dr. Neil Curtis (later Professor of Chemistry at Victoria University in Wellington, New

Multiple Bonds Between Metal Atoms

4

Chapter 1

Zealand) was a visiting research associate at MIT, and he set about trying this, with the added objective of obtaining mixed clusters, such as [Re2OsCl12]2-, by using a mixture of [ReO4]− and an osmium compound.

Neither of the original goals has ever been attained because, after a few exploratory experiments, a far more interesting result was obtained by Curtis. He found that by using concentrated aqueous hydrochloric acid as the reaction medium and hypophosphorous acid as the reducing agent (with or without the presence of any osmium compound), the product was an intense blue solution from which materials such as a beautiful royal-blue solid of composition CsReCl4 could be isolated. Since this substance had the same empirical formula as the red Cs3Re3Cl12 we were keenly interested in learning its true nature.

By a coincidence, of a sort that seems to occur rather often in research, there was another visiting research associate in the group at the same time, namely, Dr Brian Johnson (today Professor of Chemistry, Cambridge University), who had been checking a rather puzzling report from the USSR16 to the effect that reduction of [ReO4]- in hydrochloric acid by hydrogen gas under pressure gave [ReCl6]3-. This was obviously relevant to Curtis’s work, since it suggested that aqueous reduction of [ReO4]- might give (previously unknown) mononuclear ReIII chloro complexes. An even more remarkable feature of this curious report was that the precipitated ‘MI3ReCl6’ compounds displayed a variety of colors, depending on the counterion, MI. Johnson showed quickly that the claim of [ReCl6]3- salts was erroneous17 and that the compounds were in fact the rather uninteresting, very familiar, MI2ReCl6 salts. The variety of colors displayed is not easy to explain with certainty, but probably arose from incorporation of impurities. The reaction conditions cause serious corrosion of the steel bomb in which the reaction is conducted.

However, it had also been claimed16 that there was a dark-blue/green product, to which the formula K2ReCl4, was assigned. Johnson found that there was indeed such a product and, in view of its apparent similarity to Curtis’s new blue ‘CsReCl4,’ we immediately wondered if the Soviet chemists had simply got their formula wrong and that they really had ‘KReCl4.’ It did not take long to show that this was precisely the case and that the substance had the empirical formula KReCl4υH2O. Since it formed better-looking crystals than did the cesium compound (which, incidentally, is actually CsReCl4υ1/2H2O18 before drying), and these had a small triclinic unit cell, we considered KReCl4υH2O to be the preferred subject for an X-ray crystallographic study. Mr C. B. Harris (now Professor of Chemistry, University of California, Berkeley), who was just beginning his doctoral research and had never previously done a crystal structure, began a study of these crystals.

The Soviet chemical literature was also examined more carefully to see if there were any further reports of interest on the chemistry of lower-valent rhenium. It was found that between 1952 and 1958 V. G. Tronev and co-workers had published three papers16,19,20 that described an assortment of low-oxidation state rhenium halide complexes in which the metal oxidation state was proposed to be +2. Much of the impetus for their investigations was a search for analogies between the chemistry of rhenium and platinum, an approach which no doubt prejudiced them in favor of the ReII oxidation state. The existence of most of the compounds described in their 195219 and 195416 reports has never been substantiated, for example, products such as ‘Re(C5H5N)4Cl2,’ ‘Re(C5H5N)2Cl2,’ and ‘Re(thiourea)4Cl2.’ Two compounds—namely, the ‘K2ReCl4’ already mentioned and blue-green ‘(NH4)2ReCl4,’ which was also obtained by the action of hydrogen under pressure upon solutions of NH4ReO4 in concentrated hydrochloric acid at 300 ˚C—were further discussed in 1958 when Kotel’nikova and Tronev20 published a more substantial contribution, entitled ‘Study of the Complex Compounds of Divalent Rhenium.’ Additional details were reported for the various materials emanating from a work-up of the blue solutions produced by these hydrogen reductions of perrhenate (KReO4) in concen-

Introduction and Survey 5

Cotton, Murillo and Walton

trated hydrochloric acid. In addition to the rhenium(IV) salts such as K2ReCl6, a remarkable variety of low-oxidation state products of spurious and largely unsubstantiated formulas (e.g., H2ReCl4, KHReCl4, ReCl2υ4H2O, ReCl2υ2H2O, H2ReCl4υ2H2O, KHReCl4υ2H2O, and NH4HReCl4υ2H2O) were mentioned. Other than rhenium and chlorine microanalyses and an occasional oxidation state determination by the old method of I. and W. Noddack21 (see below), no further characterizations were described that supported these formulations.

With respect to the oxidation state determinations, which Kotel’nikova and Tronev reported as supporting the oxidation state +2 for rhenium, two points are pertinent. First, this method (which involves treatment with basic chromate, with intent to oxidize all rhenium to ReVII, while reducing an equivalent amount of chromium to Cr2O3, which is filtered off and weighed) has often been found unreliable. Second, however, when this procedure was repeated at MIT on one of our own compounds,22 it gave an oxidation number of +2.9±0.2. Presumably, the Soviet chemists, for whatever reason, obtained results that they thought required an oxidation number of +2 and, accordingly, adjusted the number of cations, usually by postulating the otherwise unsupported H+, to make this consistent with the analytical data they had.

Before we leave our discussion of these rather confused and largely erroneous early results, consideration of two additional points is appropriate. First, Kotel’nikova and Tronev20 observed the formation of a gray-green material, formulated as (C5H5NH)HReCl4, upon the addition of pyridine to an acetone solution of ‘H2ReCl4υ2H2O’ that had been acidified with concentrated hydrochloric acid. Second, a variety of products, obtained when ‘H2ReCl4υ2H2O’ was dissolved in glacial acetic acid, were described20 once again as derivatives of rhenium(II), namely ReCl2υ4CH3COOH, ReCl2υ2CH3COOHυH2O, ReCl2υCH3COOHυH2O, ReCl2υCH3COOH, and ReCl2υCH3COOHυC5H5N. The isolation of both (C5H5NH)HReCl4 and ReCl2υCH3COOH is of significance since, while both were incorrectly formulated,20 they are now known to have been genuine products that contain quadruple rhenium–rhenium bonds.

Except for one more brief report in 1962, describing23 the formation of crystalline (C5H5NH)HReCl4, by hydrogen reduction of a hydrochloric acid solution of the rhenium(IV) complex ReCl4(C5H5N)2 in an autoclave, the work of Tronev et al. was not further examined, by the authors themselves or anyone else, until 1963. We return now to that story.

While Harris was carrying out his crystallographic study of ‘KReCl4υH2O,’ proceeding rather slowly and deliberately (since he was learning X-ray crystallography as he went), a new issue of the Zhurnal Strukturnoi Khimii was received at MIT, and we noted that it contained an article24 dealing with, ‘(pyH)HReCl4.’ Since we did not read Russian, it was not immediately clear what was being reported, though tables and figures within the article implied that it was reporting a structure determination. Fortunately, S. J. Lippard, a graduate student in the group (now Professor of Chemistry at MIT), had completed a crash course in Russian the previous summer at Harvard University and he was able to enlighten us. The paper reported that (in Lippard’s translation, which is substantively identical to but in exact wording slightly different from the commercial translation that appeared nearly a year later):

Eight chlorine atoms constitute a square prism with two rhenium atoms lying within the prism, whereby each rhenium atom is surrounded by four neighboring chlorine atoms situated at the apices of a strongly flattened tetragonal pyramid. The apices of two such pyramids approach each other generating the prism. In such a structure, each rhenium atom has for its neighbors one rhenium atom, at a distance of 2.22 Å and four chlorine atoms at a distance of 2.43 Å. As a result, the dimeric ion [Re2Cl8]4- is generated.

Multiple Bonds Between Metal Atoms

6

Chapter 1

With regard to the structural situation of the H atoms present in the formula, the following statements were made:

The isolated [Re2Cl8]4- grouping is bonded ionically to the pyridinium ion [C5H5NH]+ carrying a positive charge, and its free hydrogen ions. . . . The detached free hydrogen ion is identified as situated on a fourfold position, which is electrostatically stable. It may be surmised that four hydrogen atoms are situated between ClII atoms on centers of symmetry . . . and serve to bond the [Re2Cl4]4- groups even further to each other.

In addition to the completely unprecedented Re-to-Re distance of only 2.22 Å and a puzzling discussion of the structural role of the ‘hydrogen ions’ (also sometimes called ‘hydrogen atoms’), there had been, according to the experimental section of the paper, severe difficulty with crystal twinning. For all these reasons, we felt that this work was probably in error, possibly because the twinning problem had not, in fact, been successfully handled. Harris therefore hurried to complete his work on ‘KReCl4υH2O.’

To our considerable surprise, he found an anion essentially identical in structure to that described by the Soviet workers. There were some slight quantitative discrepancies, which we later resolved by carrying out a better refinement of the Soviet structure. The structure of the [Re2Cl8]2- ion, exactly as found and reported by C. B. Harris25 in K2Re2Cl8υ2H2O, is shown in Fig. 1.1.

While Harris was completing his structural work, several others in the laboratory had also prepared a number of new compounds containing the [Re2Cl8]2- ion, using both our method (H3PO2 reduction) and the Tronev method (high-pressure H2 reduction), and shown that:

1.the same products were obtained by both methods, although the former was far more practical, and

2.that the charge on the Re2Cl8 unit was indeed 2- and not 4-, as believed by the Soviet workers.

Fig. 1.1. The structure of the [Re2Cl8]2- ion as originally reported in ref. 25. A cartesian coordinate system has been added.

To round out this section, it is pertinent to note several other publications during the period in question, even though they had no bearing on the recognition of the existence of the Re–Re quadruple bond. There were two other very short Soviet papers (neither of which became known to us until much later, anyway) in which a few additional, misformulated, compounds were reported. One26(a) described compounds said to have the compositions ReCl2υCH3CO2HυL, with L = H2O, C5H5N, or (NH2)2CS, while the other26(b) reported substances said to have the formulas (ReCl2υCH3CO2HυH2O)2, Re2Cl3υ3CH3CO2HυH2O, (ReClυ2CH3CO2H)2, ReCl2υCH3CO2HυH2O, ReCl2υCH3CO2Hυ2thiourea, and ReCl2υCH3CO2Hυpyridine. As to possible structures, little was said, none of which was correct.

Introduction and Survey 7

Cotton, Murillo and Walton

Finally, in late 1963 there was a paper27 reporting that reactions of rhenium(III) chloride with neat carboxylic acids give diamagnetic, orange products with molecular formulas [ReCl(O2CR)2]2. It was proposed, by analogy with the known structure of CuII acetate, that the compounds were molecular, with bridging carboxylato groups and terminal chloride ligands.

1.2.2 The recognition of the quadruple bond

In only one of the Soviet papers discussed in the preceding section was anything said about the bonding in the putative ReII compounds, namely in the structural paper,24 where the following statement was made:

It should be noted that the Re–Re distance ~2.22 Å is less than the Re–Re distance in the metal . . . . The decrease in the Re–Re distance in this structure, compared with the Re–Re distance in the metal, indicates that the valence electrons of rhenium also take part in the formation of the Re–Re bond. This may explain the diamagnetism of this compound.

Although it appears that at least by 1977,28 the Russian school fully endorsed the concept of the quadruple bond, they appeared to have remained quite ambivalent for some time about the related problems of composition (i.e. the oxidation state of the rhenium and the question of whether hydrogen is present) and bonding, and the discussions in their papers are sometimes confusing, even as late as 1970. Thus, there is a paper29 entitled ‘Crystal Structure of Re2Cl4- [CH3COO(H)]2υ(H2O), with a Dimeric Complex Ion,’ in which it was stated that, “In the two (_ and `) modifications of (pyH)HReIIBr4 the authors found triple (1μ + 2/) Re–Re bonds.” The correct formulas and oxidation numbers for at least some of their compounds still appeared to elude them. In the formula used in the title, the appearance of ‘(H)’ is certainly an arresting feature, but what it is meant to imply was left entirely to the reader’s imagination, unless it was an attempt to evade ‘the question of whether acetic acid is found as a neutral molecule or as an acetate ion.’ The authors described that question as one which “remains unclear.”

Taha and Wilkinson27 did come to grips with the question of bonding in their [ReCl(OCOR)2]2 compounds (for which they did have the correct formulas). They drew a structure with no Re–Re bond and explicitly stated that “it is not necessary to invoke metal–metal bonding to account for the diamagnetism.”

The explanation for the remarkable structure of the [Re2Cl8]2- ion was put forward by one of the editors of this book in 1964.30 Prior to this the chemistry of the [Re2Cl8]2- ion had been extensively clarified.22 We had shown that the ion could be prepared much more conveniently from [ReO4]- using an open beaker with H3PO2 as the reducing agent, that the analogous bromide could be made, that it reacted with carboxylic acids to give the Taha and Wilkinson27 compounds, and that this reaction is reversible.22

The existence of a bond between the rhenium atoms was proposed and explained in September 1964, as follows:30

The fact that [Re2Cl8]2- has an eclipsed, rather than a staggered, structure (that is, not the structure to be expected on considering only the effects of repulsions between chlorine atoms) is satisfactorily explained when the Re–Re multiple bonding is examined in detail. To a first approximation, each rhenium atom uses a set of s, px, py, dx2−y2 hybrid orbitals to form its four Re–Cl bonds. The remain-

Multiple Bonds Between Metal Atoms

8

Chapter 1

ing valence shell orbitals of each rhenium may then be used for metal-to-metal bonding as follows. (i) On each rhenium dz2−pz hybrids overlap to form a very strong μ bond. (ii) The dxz, dyz, pair on each rhenium can be used to form two fairly strong /-bonds. Neither the μ nor the / bonds impose any restriction on rotation about the Re–Re axis. These three bonding orbitals will be filled by six of the eight Re d electrons. (iii) There remains now, on each rhenium atom, a dxy orbital containing one electron. In the eclipsed configuration these overlap to a fair extent (about one third as much as one of the / overlaps) to give a β bond, with the two electrons becoming paired. This bonding scheme is in accord with the measured diamagnetism of the [Re2Cl8]2- ion. If, however, the molecule were to have a staggered configuration, the β bonding would be entirely lost (dxy-dxy, overlap would be zero). . . . Since the Cl–Cl repulsion energy tending to favor the staggered configuration can be estimated to be only a few kilocalories per mole, the β-bond energy is decisive and stabilizes the eclipsed configuration. This would appear to be the first quadruple bond to be discovered.

In a full paper31 that followed shortly, this proposal was elaborated in detail and supported with numerical estimates of d-orbital overlap. It was proposed that Re–Re quadruple bonds also occur in the Re2(O2CR)4X2 molecules. Finally, the correlation of metal–metal distances with bond orders ranging from <1 to 4 was explicitly discussed, and the concept of an entire gamut of M–M bond orders in an entire field of non-Wernerian compounds was introduced. This broad, synthetic view (and preview) of the field, which is in the nature of a Kuhnian paradigm shift, was presented in more detail very soon after in a review article.32

The quadruple-bond chemistry of rhenium was opened up quickly in several papers,33,34 and before the end of 1966 the first metal–metal triple bond had also been reported35 in the dirhenium compound Re2Cl5(CH3SCH2CH2SCH3)2, which is obtained from the [Re2Cl8]2- ion.

Today the concept of quadruple bonds is no longer novel, with about 1500 compounds known to contain them, and the physical and theoretical characterization of them is very comprehensive, as this book will show. However, prior to 1964 quadruple bonds were totally unknown, and the idea even seemed to alarm some organic chemists, who took some time to accept the fact that d-orbitals can do things that s- and p-orbitals cannot. The newness of the concept of a quadruple bond is well illustrated by Linus Pauling’s comment36 (l960) that no one had ever presented evidence “justifying the assignment to any molecule of a structure involving a quadruple bond between a pair of atoms.” Actually, the notion of quadruple bonds had been broached earlier, when Langmuir had proposed37 to G. N. Lewis that the structure for nitrogen and carbon monoxide might involve “a quadruple bond such that two atomic kernels lie together inside a single octet,” but this possibility (not surprisingly) was quickly eliminated as a realistic description of the bonding in any homonuclear or heteronuclear diatomic molecule formed by nonmetals.

1.2.3 Initial work on other elements

Molybdenum and technetium

The reaction of molybdenum carbonyl with carboxylic acids was apparently examined for the first time in 1959, when the reaction with benzoic acid was reported38 to yield a compound of empirical formula Mo(C6H5CO2)2. It was suggested that this substance might be either mononuclear or an infinite polymer, but, in either case “a novel type of oxygen chelate complex . . . where, in addition, the arene nucleus is bound to the metal atom by a sandwichtype bond, as in the arene metal carbonyls.”

Introduction and Survey 9

Cotton, Murillo and Walton

When, in 1960, it was shown39 that several aliphatic acids also react with Mo(CO)6 to form “(RCOO)2Mo” compounds, the arene–metal structure for the benzoate was pronounced “unlikely.” For all of these compounds an infinite polymer structure with tetrahedrally coordinated metal atoms and no metal–metal bonding was suggested. When this same work was reported more fully in 1964, it was suggested40 that dinuclear molecules (which were pictured as shown below) are present and that, since they “are diamagnetic, this is consistent with tetrahedral coordination by oxygen, . . . with both bridging and chelating carboxylate groups.” Again, no metal–metal bonding was even mentioned as a possibility. Clearly, at this time the true nature of these substances was entirely unrecognized.

R

C

C

R

It was not until late 1964 that such recognition occurred. By then the existence of quadruple bonds in [Re2X8]2- and Re2(O2CR)4X2 compounds had been proposed, as outlined above in Sections 1.2.1 and 1.2.2, and an X-ray investigation of a recently reported41 technetium compound, (NH4)3Tc2Cl8υ2H2O, had been completed. The formula of this compound prompted those who reported it to observe that “the stoichiometry of the [Tc2Cl8]3- ion is unusual and it seems to have no analogs.” One of us was immediately struck by its similarity to [Re2Cl8]2- and, within a few months, had shown42a that the [Tc2Cl8]3- ion had a structure very similar to that of the [Re2Cl8]2- ion, especially in that the conformation was eclipsed. The Tc–Tc distance was even shorter (2.13(1) Å) than the Re–Re quadruple bond distance (2.24 Å), which seemed consistent with the fact that Tc atoms are inherently a little smaller than Re atoms. The correct explanation for the presence of an additional electron in the [Tc2Cl8]3- ion was not at that time evident and the issue was not addressed.

Just as the findings on the [Tc2Cl8]3- ion were being prepared for publication, it was learned by letter from Prof. Ronald Mason (then of Sheffield University) that he had determined the crystal structure of ‘(CH3COO)2Mo’ and found the molecular unit to be as shown in Fig. 1.2. The Mo–Mo distance is nearly the same as the Tc–Tc distance and, since MoII is isoelectronic with ReIII, it seemed clear that Mo2(O2CCH3)4 contains a quadruple bond and that it is a group 6 analog to the Re2(O2CR)4X2 type of group 7 compound. We invited Mason to publish his molybdenum acetate structure back-to-back with our [Tc2Cl8]3- structure, and he agreed. The two manuscripts were submitted together on 30 November 1964, and appeared together in early 1965.42 In our communication on [Tc2Cl8]3- we observed that on the basis of these new results on two compounds formed by metals in the second transition series:

It appears that the formation of extremely short, presumably quadruple, bonds between d4-ions of the secondand third-row transition elements may be quite general.

Subsequent events have shown that this statement erred only in being too cautious.

The chemistry of quadruply bonded Mo24+ derivatives did not undergo further development until late 1967, when a young Yugoslavian chemist, Jurij V. Brencˇic (Professor of Inorganic Chemistry, University of Ljubljana), joined the MIT group and took up the problem of finding the right conditions for the reaction

Mo2(O2CCH3)4 + 8HCl Α [Mo2Cl8]4- + 4H+ + 4CH3CO2H

10Multiple Bonds Between Metal Atoms Chapter 1

Mo

Mo

O

O

C

R O O

C

R

Fig. 1.2. The structure of the dinuclear molybdenum(II) acetate molecule, as first reported by Lawton and Mason in 1965.

which is analogous to our earlier reaction for the smooth interconversion of [Re2Cl8]2- and Re2(O2CCH3)4Cl2. It turned out that unless conditions were carefully controlled, a variety of products were obtained, many of which were insoluble and, for that and other reasons, difficult to characterize.43 Brencˇic sorted out this confusion, and by July of 1968 we were able to submit a report of the preparation and X-ray verification of the first of several compounds containing the [Mo2Cl8]4- ion.44

It was with this discovery that a decade of virtually exponential growth of the field of M–M multiple bonds commenced. The compounds containing the [Mo2Cl8]4- ion are entirely stable thermally and toward the atmosphere (like those of [Re2Cl8]2-); they have provided the starting points for a host of chemical, physical, and theoretical investigations.

In 1979 it was recognized45 that several compounds containing Mo–Mo quadruple bonds had been made as early as 196246 and 196447 but were not at all understood at that time. It was found that MoIII chloride and bromide reacted with liquid ammonia, methylamine, and dimethylamine to produce what were believed to be solvolysis products with suggested stoichiometries such as MoX2(NH2)υ3NH3, MoBr(NHMe)2υ2/3NH2Me, and MoBr2(NMe2)υNHMe2. It is now45 clear that these are Mo2X4L4 type molecules; for example, ‘MoBr2(NMe2)υNHMe2’ is actually Mo2Br4(NHMe2)4 and may be smoothly converted to Mo2Br4(PPrn3)4, which is also obtained by action of PPrn3 on [Mo2Br8]4-.

Thus the prehistory of Mo–Mo quadruple bonds resembles that of rhenium in that several key compounds had been made prior to 1964, but no one had the remotest idea what they really were until after the true nature of the [Re2Cl8]2- ion was made clear.30,31

Chromium

The prehistory of the Cr–Cr quadruple bonds is fairly extensive. Astonishing as it may seem, the story begins with work published as early as 1844. In that year Eugène Peligot (Fig. 1.3) reported for the first time48,49 that from bright blue aqueous solutions of chromium(II) ions, he could isolate, upon addition of sodium or potassium acetate, “little red transparent crystals. . . .

which decompose upon exposure for a few moments to air.” The method of preparation, the properties, and the analytical data leave no doubt at all that the compound Peligot prepared is Cr2(O2CCH3)4(H2O)2. Because of uncertainties prevalent at the time as to the molecular versus atomic weight of hydrogen, the empirical formula given was CrC4H4O5; upon multiplying the number of H atoms by two, this formula becomes precisely correct. Moreover, Peligot showed that thermal decomposition gave an oxide weighing 41.8% of the original weight of the salt: