THE STRUCTURE OF THE HEAVY METAL CYANIDES CLAYTON F. CALLIS University of Illinois, Urbana, Illinois
THE STRUCTURE of the heavy metal cyanides, especially that bf the ferro- and ferricyanides, has long been a matter of considerable controversy. The chemical evidence which appears in the literature is conflicting and therefore it is extremely difficult to assign a correct structure to these compounds. The more recent X-ray and diffraction patterns furnish no evidence which can be used as a basis for determining the valency of the. metal or locating the position of the cyanide groups. I t is well to point out.here that it is not yet possible to give a very satisfactory survey of the structures of the metallic cyanides because of the lack of sound experimental data. Further structural investigations of these compounds would he very desirable. The cyanide ion is found in the ionic cyanides of the alkali metals, as well as in the essentially covalent cyanides where the cyanide group is attached by a covalent bond either from the carbon or from the nitrogen atom: -CSY: or :C=N-. There is the further possibility that the cyanide group may be attached at both ends by covalent bonds, -C=N-, and undoubtedly this is the case in a number of metallic cyanides. The only metals knotvn to form ionic cyanides are the alkali metals, sodium to cesium, and thallium (Tl') (1). Apparently no investigation of the structure of the alkaline earth cyanides has been made. At ordinary temperatures, sodium cyanide, potassium cyanide, and rubidium cyanide crystallize with the rock-salt strncture and cesium cyanide and thallium cyanide with the cesium chloride structure. The cyanide ion in these crystals is freely .rotating, thereby acquiring spherical symmetry ( I ). The general structural relationships between the metallic cyanides appear to be similar t o those between halides. The fluorides are the only halide? of metals other than the alkalies which form essentially ionic three-dimensional complexes. The chlorides, bromides and iodidesform more covalent chain, layer, or molrcular lattices. These changes in structure type may br attributed, at least in part, to the decreasing electronegativity of the halogens as we go from fluorine t o iodine. An analogous change in structure type is found in cyanides, going fromtheelectropositive metals, the alkalies to the'metais of later or B subgroups. Unfortunately very little is known of the structures of these cyanides, but it is likely that chain, layer, and molecular types d l be found. Rlasaki (2) found that the addition of silver cyanide, silver chloride, silver thiocyanate, silver oxide, or silver chromate to an excess of sodium cyanide gives a molal ratio of combined cyanide to silver of 1:1, determined
.
by titration of the excess of cyanide with silver nitrate. Braekken (3) studied the X-ray diffraction patterns of silver cyanide recrystallized from sodium carbonate and proposed a chain lattice structure for this covalent cyanide in which there are infinite chains of alternate silver atoms and cyanide groups so that the cyanide group is attached at both ends t o metal atoms. These chains are arranged parallel in hexagonal packing with the silver atoms laced a t the points of a simple rhombohedron (4). The Ag-Ag distance along the chains (5.26 A,) is consistent with this -Ag-C-N-Agbonding, for if we take the C-N distance to be about 1.15 A. (as in K4Mo(CN)~2Hz0) we should have 4.11 A. for the sum of the Ag-C and Ag-N distances. Now the radius of two-covalent silver is given by Pauling as 1.39 A,, being derived from the Ag-0 distance of 2.05 A. in silver oxide using 0.66 A. as the tetrahedral radius of oxygen. With the values 0.77 A. and 0.74 A. for C and N, respectively, the sum of Ag-C and Ag-N is in good agreement with the observed values. Simple layer structures are possible for the cyanides of metals like divalent palladium, platinum, nickel, and copper,which form four planar bonds (1). This suggested arrangement is shown below.
A very interesting example of the manner in which the requirement of attaining the stable coordination number may bring about the polymerization of coordination compounds is afforded by diethyl cyanogold, EtAuCN (6). If this compound were monomeric, gold would have a coordination number three, and would be coordinatively unsaturated. Since, however, the square planar configuration is stable for trivalent gold, and since the cyanide group may coordinate with either. end, the coordination maximum may be attained by a four-fold polymerization (R = ethyl or propyl):
150
-kl
kt
R-$u-C=N
t 1x1
R-Au
I I
R
- t: I
Au-R
I#
1 I
c X=C-Au-R
-
R
131
MARCH, 1948
X-ray and molecular weight determinations confirm this tetrameric structure. This formulation, a symmetrical twelve-atom planar ring, affords additional support for the existence of coordinate linkages in complex molecules and is in keeping with the stereochemical configuration of the cyanide group. These compounds yield, with ethylene diamine, colorless crystalline derivatives (6), which also are nonelectrolytes described as monoethylenediaminotetraethyldicyanodigold and monoethylenediaminotetra-npropyldicyanodigold, respectively. The constitution of these compounds, from analytical and molecular weight determinations, is represented by the structure given below:
-
CN
I
CN
Such a reaction is similar to the hydrolysis of hydrogen cyanide. A great number of these so-called double complex salts conform to the composition requiredfor the heavy metal partner to attain its stable coordination maximum. I n accordance with this principle, the most widely occurring and most important series of double salts are based upon the coordination numbers six and four for the central atom of the complex anion. The distribution of the types [M(CN)6]-(6-z) and [M(CN)n]-(4-x) (where x = valency of the atom M) among the heavy metals is shown in the table (6). Mnl Mn5 Fe: Fe,, [M(CN),I-@*, V5 V : Crl, CrS, [M(CN)J-(4-s)
Coa Co$ Rh4 Rug, OO, fr" Nil, Cu', Zn, Cd, Hg: PdP,PtP
In addition to these metals, there are a few elements which form thiocyanato compounds of similar type, RI although the cyano compounds are unknown. This may be due to the greater polarizability of the thiocyGibson, Bura.way, and.Holt (6) were unable to isolate anate ion, which enables it to form stable complex ions salts of the type with cations of smaller polarizing effect (i. e., larger R ionic radius) than can combine with the cyanide ion. Scandium fits in this way into the hexacido type of complex ion formation, with Ks[Sc(SCN)J .4H20, and divalent cobalt conforms to the tetracido type with R?[Co(SCN)41. I n e n these cyanides, EtzAuCN and PrzAuCN, are ~h~ gold compounds, which contain the heated they lose half their alkyl groups and a definite [Au(CN)4]-, are less stable and are.easily decomposed compound is produced. .The products contain two tri- in solution (?). . ~h~ liberated hydrogen cyanide valent gold atoms and two univalent gold atoms with a does not give enough cyanide ions to move the equilibcoordination number of two and an effective atomic ,it,, number of 82. Gibson suggests the structure below: R-Au + NHrczH4-H?x
A
R R-
At
u-GzN
Au-R
- t:
Au
lAl
I
1
Au c N=C-Au-R
I
R
Double cyanides are among the most stable complexes known. They are mostly unaltered in acid solution, and differ in color in a remarkable manner from the simple con~poundsof the metals they contain. The general formula for the metallo-cyanide ions is One extra electron is required for ,MyCN),I-(m-z,. each cyanide group, and since the coordination number m. is greater than the electrovalency of the metal, these ions carry a negative charge of (m - x). In complexes of this type, the cyanide ion is apparently coordinated to the metal through the carbon atom. 3he carbon and nitrogen atoms are so nearly the same size, analysis that it is difficult to determine from group is, just \,,hat the of the H ~it is known ~ that~ the cobalticyanide ~ ~ ,,,ill hy~ drolyze in the folloving may: IG[Co(Cx)ll-
H+
NE,+ &[ CoCOO(CN)$ ]
[Au(CN),] -
Au(CN)*
+ CN-
to the left and thus prevent precipitation of the simple cyanide. The salts of this complex are colorless and easily soluble in water. The cyanogen groups can be replaced in part by other halogens, e . g., in
pu
(C:)p]
-
A number of crystalline salts containing metallocyanide ions have been studied by the X-ray method.. The discussion of some of these compounds will illustrate the possible arrangements of the two, four, six, and eight covalent bonds. Conductometric and potentiometric titrations of potassium cyanide and silver nitrate led Dorrance, Ellis, and Matheson (8) to the conclusion that silver cyanide dissolves in excess potassiun cyanide t o form a double cyanide in which the ratio is 1: 1 and hence the formula of the complex is KA~(CN)Z.The ions [NC-Ag-CNI - are linear with the same configuration of silver bonds as is found in the [H,N-Ag-NH,] + ion in tetrammino-silver sulfate (1). The stereochemistry of four covalent cuproils copper has, been shown to be tetrahedral in the salt Kz[Cu(CN)4] (I). The [Cu(CN),]' ion has a tetrahedral configuration like the Ni(CO)&molecule with which it is isoelectronic. Glasstone (9) points out that electrometric titration of CuCN and solubility measurements
JOURNAL OF CHEMICAL EDUCATION
152
in alkali cyanide solutions show that the complex ions CU(CN)~-and CU(CN)~-can both exist. The double cyanides KzZn(CN)4, KzCd(CN)a, and K2Hg(CN)4areisomorphous (10). The existence of the complex Zn(CN)a-, Hg(CN)4=, and Cd(CN)&=ions has been confirmed by Britton and Dodd (11) from pH measurements on the metal salts in the presence of potassium cyanide and by Glasstone (12) from measurements of cathode potential and current efficiency. The isomorphous tetrahydrates of BaPt(CN)4,BaPd(CN)4, and BaNi(CN)&contain planar complex ions, in which the metals form coplanar dspa bonds. This square planar configuration has been confirmed by Brasseur and de Rassenfosse (13, 14, 16) from a comwhere R is Ca, prehensive study Sr of or Ba, the and nineM crystals is Ni, Pd RM(CN)4.nHe0 or Pt. Eastes and Burgess (16) obtained a compound of the formula &Ni(CN)4 by treating an anhydrous liquid ammonia solution of potassium cyanenickelate with excess potassium. If the potassium cyanonickelate was in excess, a different compound, KsNi(CN)a, was ohtained. No explanation of the state of valence of the nickel in these compounds was offered. Deasy (17) in a later article postulates electronic formulas for the compounds on the basis of the modern theory. The negative radical of the compound &Ni(CN)&, [Ni(CN)4]-, is isoelectronic with nickel carbonyl, which also contains neutral nickel. Thus an electronic configuration is postulated which is similar to that suggested by Pauling (18) for nickel carhonyl. Three resonating structures can be considered:
has a stable octet of electrons ahout'the carbon atom; this is not shorn in structure C , which has only a sextet. of electrons. In A four of the 3d orbitals of nickel, as well as the 4s orbital and one 4 p orbital would be used in bond formation. A detailed discussion of the nat,ure of the bonds must he preceded by the determination of physical data. Probably, as in the case of nickel carbonyl, the molecule resonates between the double-honded and singlehonded structures. Two resonance structures are suggested for the Ni(CN),-ion:
[
..:
i
.N-::C::Ni+l::C::N:-
.,
!-
[:N:::c:Ta-:c:::x:/
g: ..
..
D
E
. Formula D may make the greater contribution because here the nickel is electropositive. In this structure four 3d orbitals, the 4s, and one 4 p orbital might be used in bond formation, with the unpaired electron in a second 4 p orbital. A similar structure can be postulated for E, which one 3d, the 49, and one 4 p orbital used for bond formation, and the lone electron in a 4 p . orbital. Either of these would be an electronic configuration similar to that of cobalt in the C o ( C N ) s ion, in which it is postulated that the unpaired electron of .. N .cobalt occupies the outer unstable 4d orbital. This outer lone electron is believed to account for the unstability of the complex. The chemical behavior supports the above formulation for the [Ni(CN),l-. .. Magnetic susceptibility measuremkts of nickel cyan-:N::c::,~::c::N.ide complexes have been very helpful in evaluating their suatial confieurations. The cvanides. K9SiICNL ,. and K ~ N ~ ( C N ) ~which . ~ O contai"n , N ~ ( c N ) ~groups ,are diamagnetic and planar. Nickel cyanide hept* hydrate has a normal moment for the Nil ion, but on dehydration the para-magnetism decreases. Si(CN)? containing from two to four molecules of water has a moment of only about one-half that for ionic nickel compounds. This suggests that these hydrates contain approximately equal numbers of planar covalent, Ni(CN)4=, complexes and ionic Ni(OH2)4++or Ni(OH&++ groups. On further dehydration a substance is obtained with an apparent moment of about 0.5 Bohr magneton or less. Bose ($1) concludes t,hat completely dehydrated nickel cyanide mould be diamagnetic. Many salts containing ions of the type [M(CN)6] have been prepared, the best,known being the simple ferro- and ferricyanides. The complex M3(CS)6]=is Structure A seems more satisfactory than the single- produced when solutions containing trivaient metal ions bonded structures since the nickel atom is neutral, are treated with cyanide ions; e. g., when excess of rather than negative; and in its general behavior is potassium cyanide is added to solutions of the trivalent electropositive rather than electronegative. It also metal salts or when the hydroxides are dissolved in
1
~
~
~
MARCH, 1948
potassium cyanide. Intermediate compounds are often formed on treating salt solutions with potassium cyanide; t,heir composition is variable, and sometimes they dissolve only in boiling potassium cyanide. These complex cyanides can also be prepared by oxidation of [M2(CN)e]-- compounds (7). Thus K3[Fe3(CN)a] can be made by oxidizing K4[FeZ(CN)a]with chlorine, bromine, permanganate, or lead peroxide; however, in many instances, the reduction potential of the lower double cyanide is so small that oxidation of some compounds takes place even in the air; e.g., formation of [ M ~ I ~ ( C N )from ~ ] = [Mn2(CN)6]- or of [CO~(CS)~]' from [CO~(CN)~]-. Indeed, if air is absent, this latter radical, [CO~(CN)~]-,decomposes t,he vater in which it is dissolved, and hydrogen is evolved, as in the action of vanadous and chromons rompounds. The salts of the double cyanides were among the first to be isolated, Both the alkali and alkaline earth metal salts of the ferrocyanides are soluble in water, and contain the simple [Fe(CN)J=- ion which exists in the free acid HbWCNh. The alkali salts of the chromium double cyanides are pale yellow, those of manganese red to black, of cobalt very pale and of ferric iron brownish red, ~h~ alkali salts of the double cyanides of chromium, manganese, cobalt, iron, and iridium are isomorphous. No simple ferrocyanide has yet been thoroughly studied ( I ) , but in a number of cobaltammines the octahedral configuration of certain hem-cyanido ions has been demonstrated. Thus the salts [Co(NH,),] [Co(CWel, [Co(N&)61 [Cr(CN)d, and [Co(NHa)6H~O] Fe(CN)6]crystallize with octahedral ions which pack together in much the same way as the cesium and chloride ions in cesium chloride. The cyanides of iron have low or diamagnetic susceptibilities. X,Fe(CN)e is diamagnetic, and &Fe(CN)e has a moment corresponding to one unpaired electron (82). The complexes are octahedral, and the reason for the magnetic moments being as they are is clear from the following consideration of the electron distribution: 3d 44s ' 4p
153
with single covalent bonds from the iron atom to each of the six carbon atoms, is not logical since it places a charge of 4- on the iron atoms. Iron tends to assume a positive charge, as in the ferrous ion. The cyanide group is an electronegative group, and the Fe-C bonds accordingly have some ionic character, which, however, can hardly be great enough to remove the negative charge completely from the iron atom. Now if the assumption is made that the cyanide group in this complex can function as an acceptor of electrons, the bonds can resonate among the following types: A Fe (CN)B Fe-:C:::N: C
F~::C::N:-
The first of these represents an electrostatic bond be: tween the iron atom and the cyanide ion, the second a single covalent bond from iron to carbon, and the third a double covalent bond, with use of another 3d orbital of the iron atom with its pair of electrons. The first and the third of these place a negative charge on the cyanide group, and the second leaves the group neutral. Resonance among these with the second structure contributing only about one-third would make the iron atom in the complex electrically neutral, the negative charge 4- being divided among the six cyanide groups. By using all the Sd, 4s, and 4 p orbitals of the iron atom the valence-bond structure
can be written for the ferrocyanide ion. This structure is, of course, in resonance with the equivalent structures obtained by redistributing the bonds. It is probable that the ionic character of the bonds is great enough to transfer further negative charge to the nitrogen atoms, making the iron atom neutral or even positive. It is convenient, however, to represent the complex ion by the conventional structure, just as, for convenience, the benzene molecule is often represented by a single Kekul6 structure. For other anionic cyanide complexes of the transition elements, such as [Fe(CN)e]=, [Co(CN)6]=, [Mn1 , [Cr(CN)6]-, [No(CN)r]-, [Zn(CN)rl=, [Cu(CN)z]-, similar structures involving partial double bond character of the metal-carbon bond can be written. Pauling (M)is of the opinion that the structural forThe double cyanides of some of the heavy met& are mula usually written for the ferrocyanide ion, highly colored insoluble compounds which are structurally quite different from the simple soluble ferrocyanides. The nature of these compounds will be discussed later in this paper. Molybdenum and tungsten form complex octacyanides of the type &M(CN)c2Hp0. Hoard and Nordsieck (19) have determined the structure of potassium
154
molybdocyanide dihydrate and the configuration of the molybdenum oct,acyanide group by X-ray analysis. The exact nature of the molecule or complex ion of formula M& bad not been established with certainty until X-ray methods were employed. The cube and the square Archimedean antiprism have been regarded as geometrically plausible possibilities. Hoard and Norsieck proved that the eight cyanide groups in the [Mo(CN),- complex ion are arranged a t the apices of a dodecahedron around the metal atom. The C-N bonds are collinear with the Mo-C bonds, their mean lengths being: Mo-C 2.15, and C-N 1.15 A. The orbitals used in these octacovalent complexes are presumably the four 4d, one 5s, and three 571 orbitals, but .this spatial arrangement of bonds had not been predicted and no theoretical treatment of such dodecahedral bonds has been given. Crystals of pot,assium tungstocyanide dihydrate, IGW(CN)8.2Hz0, and potassium molybdocyanide dihydrate are almost surely isomorphous, and the IW(CN)8]4- ion probably has the configuration established for the [MO(CN)~] group (19). When solutions of the complex cyanides [M(CN)s] of such metals as Fe, Co, Mn, Cr, etc., are added to solutions of salts of transitions metals or divalent copper, insoluble compounds are precipitated in many cases. The precipitates formed are mostly flocculent, slimy or gelatinoun, and always contain alkali metals, but the nature of the substances makes it difficult to determine whether these are only adsorbed from solution or in chemical combination. There is no doubt, however, that the adsorption is an important factor. The addition of ferrocyanide ions to ferric salt solutions results in the production of blue substances bearing the generic name Prussian blue. When an excess of ferric ion is employed, the product has the formula Fe4[Fe(CN)6]2and is known as insoluble Prussian blue. From equimolecular quantities of ferric and ferrocyanide ions, a product is formed which contains the metallic ion of the soluble ferrocyanide employed. Thus, with potassium ferrocyanide, the blue product has the formula KFe[Fe(CN)6]and is known as soluble Prussian blue. The term soluble is a misnomer, however, and refers not to the true solubility of the product but to the ease with which it forms colloidal solutions. When dried these complexes are very stable toward acids, and are, as we shall see, quite different in structure from the simple soluble ferrocyanides. The structure of Prussian blue and its related compounds has been a problem of particular interest for a number of years. The chemical evidence, which is due largely to the efforts of Hofmann (84), Eibner and Gerstacker (35), and to Erich Muller and his students (867, is very confusing. Substances termed a, P, and 7 soluble Prussian blue have been reported in the literature. Actually, these compounds differ only in the ease with which they may be dispersed cofloidally, an$ both Wells (37) and Emeleus and Anderson (88) state that all such compounds are actually identical, with the general formula RFe[Fe(Cl\')s] (where R = an alkali
JOURNAL OF CHEMICAL EDUCATION
metal, usually potassium). That is, all the compounds variously described as Turnbull's blue, from a ferrous salt and ferricyauide, Prussian blue, from a ferric salt and ferrocya~de,and %oluble Prussian blue,'' have the composition KFeFe(CN)aH20, though there may be more water associated with the material if it is in the colloidal form. Because of the methods of preparation, Turnbull's blue was originally formulated KFe2[Fea(CN)s] and Prussian blue as KFe3[FeZ(CS)6]. The chemical evidence is not conclusive since the action of NaOH on Prussian blue gives ferric hydroxide and Na4Fe(CN)e, whereas ammonium carbonate yields (NH&Fe(CN)6. Davidson (BS), in summarizing most of the chemical work done on the Prussian blues, stresses the importance of the existence of an equilibrium reaction Feat + [Fe(CN)sl'- eFe" [[Fe(CN)a13between ferric, ferrous, ferrocyanide, and ferricyauide ions. This agrees with the experimental facts that the formation of soluble Prussian blue is a slow reaction, which is accelerated by the ions Fe2-, [Fe(CN)6]3-, and [Fe(CN)6 4-, but retarded by ferric ions. Such data tend to support the ferric ferrocyanide structure of Pnlssian blue. The reduction of ferric ferricyanide, Fe3[Fe3(CN)e],with hydrogen peroxide (which reduces ferricyanide, but not ferric ion) will produce Prussian blue; it can also be formed by reduction of ferric ferricyanide with sulfur dioxide (which reduces ferric iron but not ferricyanide). The precipitation of ferrocyanides of the heavy metals has been studied conductometrically by Kolthoff (50) and by Britton and Dodd (11) who concluded that insoluble double salts were formed for which a definite chemical formula could be mitten. Several structures have been proposed for the Prussian blues (89), but it was not until 1936 that the true nature of the complex anion was revealed. Keggin and Miles @I), who were the first to investigate the X-ray diffraction patterns, show that all the Prussian blues can be regarded as derived from the structure of ferric ferricyauide or Berlin green, Fe[Fe(CN)a]. This has a cubic crystal lattice (Figure I), the unit cell being 5.1 A. on a side (C). In (C) all the iron atoms are in the ferric state. Soluble Prussian blue is also cubic with a side of 10.2 A. The essential skeleton is the same but alternate iron atoms are reduced to the ferrous state ( A ) . Every alternate small cube contains an alkali metal ion, taken up to maintain electrical neutrality. If water molecules are present, they presumably occupy similar positions. Lithium and cesium, with very small and very large ions, respectively, do not form compounds with this structure. In ferrous ferrocyanide (B) all the iron atoms are reduced to the ferrous state. Every small cube is now occupied by an alkali ion as is required by the formula IGFeFe-
+
ICNL ,--.,".
The iron atoms of Prussian blue cannot. be distinguished as ferrous and feyric ions; Emeleus and Anderson (88) state that it is likely that the iron atoms
MARCH, 1948
are equivalent, and that the valency state and charge distribution are evened out by a resonance process. Compounds with the Prussian blue structure include KcuFe(cN)~,C~ZF~(CN)~-which contains no alkali ions-KMnFe(CN)a, KcoFe(cN)~,KNiFe(CN)6, and KF~Ru(CN)G(ruthenium purple) ($7). The compound with the empirical formula KMn(CN)3, prepared from a manganous salt and potassium cyanide, is green in color and very soluble, suggesting that it is really Kz[MnMn(CN)e] with the same type of threedimensional complex as [FeFe(CN)#]in Prussian blue. The cyanide groups are not definitely located by the X-ray analysis. However, Emeleus and Anderson ($8) maintain that their position can be determined with certainty from the following considerations. The fact that the cyanide group can coordmate by means of either its carbon or its nitrogen atom leads to the important concept of supercomplex fosniation. The evidence of alkylation ($8) suggests that the carbon atoms in the ferrocyanides are attached to the central atom of iron. The nitrogen of each cyanide group can then coordinate with another heavy metal atom to fill one coordination position:
Each [Fe(CN)@l4-anion then becomes surrounded by six attached cations, and each cation by six [Fe(CN)J ~ r o.u ~ sThe . union of ferricvanide ions with ferric cations then gives Berlin green; that of ferrocyanide with ferric or ferrous ions gives super-complex anions which can.be written loosely as [Fe3[Fe2(CN)6]]-and [FeZ[FeZ(CN)6]]-,respectively. Davidson (99) has suggested that [ F C ~ [ F C ~ ( C Nmay ) ~ ] be ] termed berlinic acid, and the soluble Prussian blues berlinates. The action of excess ferric iron on a ferrocyanide may be represented as forming a ferric berlinate Fe[Fe(CN)#]I,, rather than a true ferric ferrocyanide Fea[Fe(CN)6]3,as is usually written, while an excess of ferrous iron with a ferricyanide similarly gives Fe[Fe[Fe(CN)6]]zrather than true ferrous ferricyanide. Van Bever (32) has investigated the structures of the ferricyanides of divalent cadmium, manganese, zinc, cobalt, copper, and nickel. All of these compounds that were studied were reported to have the cubic structure (33). Welo and Davidson (34), who measured the magnetic susceptibilities of "soluble" and "insoluble" Prussian blue, show that half of the iron atoms, presumably those bonded to the carbon atoms of the six adjacent cyanide groups, form essentially covalent bonds, whereas the other iron atoms form ionic bonds. The data do not show whether the covalently bond iron is trivalent or bivalent. A number of anomalies in the complex cyanides of metals other than iron can be understood on the basis of the existence of super-complexes. Reihlen and Zimmerman (35), in formulating the structures, made this point clear, but it was later shown that their theory did not have a sound basis.
A
Ferric 0. Ferrous 0. Alkali metal
B
Ferrous 0. Alkali metal
-. Ferric
r i m l u r e
1
An interesting series of salts which probably constitute an intermediate stage between a simple complex cyanide and the supercomplex structure are the Stromholm salts. Stromholm (36) suggested a polynuclear anion:
for the structure of these compounds, %Fe(CN)s.3Hg(CN)z. The reactions of the brown copper ferrocyanide, C U ~ F ~ ( C N ) ~ (n . ~ probably H ~ O 7 or lo), formed from a cupric salt and K,Fe(CN)6, show that is not a simple salt. These reactions
suggest that the brown copper compound is a salt of a [CUF~(CN)~ anion, ] = namely Cu[CuFe(CN)a]. Of the substituted cyanides, probably the best known is sodium nitroprusside, NaFe(CN)sNOl.2Hz0. It serves as a starting point for the preparation
JOURNAL OF CHEMICAL EDUCATION
156
of many other compounds, and is obtained by the oxidation of potassium ferrocyanide with dilute nitric acid and then neutralization of the liquid with sodium number. The requirement of attaining the stable coordination number of the metal atom in the majority of cases known directly influences the nature of the complex formed. In many covalent cyanides a polymerization 'of the inorganic compound results. The key to the constitution of these compounds, which has for many years been in dispute, has been supplied by the X-ray analysis of their structure. LITERATURE CITED (1) WELLS,A. F.,"Struoturd Inorganic Chemistry," Clsrendon Press, Oxford, 1945, p. 446. K., Bull. Chem. Soe. Japan, 4,190 (1929). (2) MASAKI, H.. K d . Nmske Vidaskab. Selskabs. Forh. 11. 131 BRAEKKEN. 1929, MOdd.'~o:48, 169 (1930). WEST,C. D., Z. Kn'st., 88, 173 (1934). GIBSON,C. S., et al., J. Chem. Soe., 1935,1024. EMELEUS, H. J., AND J. S. ANDERSON, "Modern kspects of Inorganic Chemistry," D. Van Nostrand Company, New York, 1939, p. 142. EPHRAIM, F., 'Tnorgmic Chemistry," Nordeman Publishing Company, New York, 1943, p. 305. DORRANCE. R. L.. R. C. ELLIS. AND A. D. MATHESON. Trans. ~ketroehkm.Soc., 72, 473 (1937). S., J . Chem. Soc., 1929, 702. (9) GLASSTONE, E., Gazz. chim.ital., 56,180 (1926). (10) CAROZZI, (11) BRITTON,11. T. S., AND E. N. DODD,J. Chem. Soe., 1932, 1940. S., ibid., 1930, 1237. (12) GLASSTONE, H., AND A. DE RASSENFOSSE, Mem. Acad. Roy. (13) BRASSEUR, Relg., Class6 mi., 16, 107 (1937).
BRASSEUR, H.. A. DE RASSENFOSSE, AND J. PIERARD,Z. K7ist.. 88.210 (19341. ~~, BRASSEUR,H., A. DE RASSENPOSSE, AND J. PIERARD, Compt. rend., 198, 1048 (1934). EASTES, J. W., AND W. M. BURGESS, J. Am. Chem. Soc., 64, 1187 (1942). DEASY,C. L., ibid., 67, 152 (1945). PAULING, L., "The Nature of the Chemical Bond,'' Cornell University Press, Ithaert, 1939, p. 232. HOARD. J. L.. AND H. H. NORDSIECK. J . Am. Chem. Soc.. 61.
. ,
--.~
-
-
~
~
,
~
SELWOOD, P. W., "Magnetochemistry," Interscience Publishers, New York, 1943, p. 177. DOSE,D. M., Nature, 125,708 (1930). SELWOOD, P. W., lot. eil., p. 155. PAULING, L.. "The Nature of the Chemical Bond." Cornell ~nive&it$Preq Ithsea, 1939, p. 255. HOFMANN. K. A.. Annala. 337. 1 (19041: 352.54 (1907). -
~
~ -~--,
J . pmkt. Chem., 79, 81 (1909); 80, 153 (1909); 84, 353 (1911); 89,68 (1913); 90,119 (1914). WELLS,A: F.,"Structural Inorganic Chemistry," Clarendon Press, Oxford, 1945, p. 449. EMELEUS, H. J., A N D J. S. ANDERSON, "Modern Aspects of Inoreanic Chemistrv." D. Van Nostrand COmDanv. New . ~ o r k 1939, ; p. 137.' DAYIDSON, D., J. CHEM.EDUC.,14,238,277 (1937). KOLTHOFP, J. M., Z. anal. Chem., 62,209 (1923). KEGGIN, J. F., AND F. D. MILES,Natu~e,137,577 (1936). VANBEVER,A. K., Ree. trau. ehim., 57, 1259 (1938). RICHARDSON, J., AND N. ELLIOTT,J . Am. Chem. Soe., 62, 3182 (1940). WELO.L. A,. A N D DAVIDSON. D.. J. Phw. Chem... 32.. 1191 (19~8). REIHLEN,H., AND W. ZIMMERMAN, Annalen, 451,75 (1927). STROMHOLM, D., Z. anorg. Chem., 84,208 (1914).
".
