99
HEAVY-METAL IRON-CYANIDES
X-RAY DIFFRACTIOX STUDIES @N HEAVY-METAL IRON-CYAKIDES’ HARRY B. WEISER, W. 0. MILLIGAN,
AND
J. B. BATES
Department of Chemistry, The Rice Institute, Houston, Texas Received November 17, 1941
In most textbooks of introductory chemistry will be found the statement that the interaction of ferric salts and alkali ferrocyanides yields Fea[Fe(C?;)6]3 or Prussian blue, whereas the interaction of ferrous salts and alkali ferricyanides yields Fea [Fe(CN)6]2 or Turnbull’s blue. Actually, the oxidation-reduction reactions which take place on mixing iron salts and alkali iron-cyanides complicate the problem to such an extent that certain investigators conclude from chemical analysis that all of the blue gels are ferrocyanides, whereas others arrive a t the diametrically opposite conclusion that all of the blue gels are ferricyanides (for a survey see reference 9). In earlier investigations on complex iron-cyanides (lo), it was reported that the ferrocyanide gels of copper, manganese, cobalt, and nickel prepared with varying ratios of heavy-metal sulfate and potassium ferrocyanide gave ident,ical or almost identical x-ray diffraction patterns irrespective of the nature of the divalent cation or of the relative amounts of thc salts. I t was suggested that the very similar crystal structures of the several iron-cyanides resulted from such an arrangement of the large iron-cyanide ions that channels were formed in which the smaller heavy-metal cations may be grouped. .4t about the same time, van Bever (8) suggested a similar structure for various heavy-metal ferricyanides, but he assumed the presence of combined water molecules within the unit cell of the proposed structure. Similarly, Keggin and Miles (3) claimed that several molecules of combined water were present within the unit ccll. Rigamonti (7), on the other hand, did not assume the presence of combined water in the unit cell of the complex cyanides. The state of the water in heavy-metal ferro- and ferri-cyanide gels has been the subject of a recent investigation (ll), which disclosed that the water in most compounds of this type is adsorbed and no definite hydrates exist. For example, smooth dehydration isotherms were obtained for cupric ferro- and fcrri-cyanides and for the so-called Prussian blue and Turnbull’s blue. Further support of the absence of hydrates is furnished by the observation that a given moist gel, and the same gel completely dehydrated in a vacuum, gaye idcntical x-ray and electron diffraction patterns. I n this connection, it was observed further that thp x-ray and electron diffraction patterns of the so-called Prussian blue and Turnbull’s blue were identical within the limits of accuracy of the experiments. This confirms an earlier report of Levi (5) to the effect that the x-ray diffraction patterns of Prussian blue and Turnbull’s blue are identical, In the light of the above survey, a comprehensive x-ray diffraction study of the Presented a t the Eighteenth Colloid Symposium, which was held at Cornel1 Univeristy, Ithaca, New York, June 19-21, 1941.
100
H. B. WEISEH, W. 0. MILLIGAN AND J. B. BATES
gels of the blue complex cyanides and related compounds was undertaken in an attempt to determine their probable structure. EXPERIMENTAL
A . Preparation of samples Samples of various iron-cyanide gels were prepared by the addition of an excess of a solution of the heavy-metal salt to an iron-cyanide solution, as indicated in table 1. The resulting gels were washed with distilled water by means of a centrifuge until the supernatant liquid was free of chloride ion (sulfate ion in the case of scandium), and were air-dried a t room temperature. The charTABLE 1 Preparation of samples
I
SOLUTION8 MIXED
UALT FORMED
Metal d t
Tia[Fe(CN)&. . . . . . . . , , , . Ti[Fe(CN)sI.. . , . . . . . . . . . In[Fe(CN)n], . . . . . . , . , , . . Fe[Fe(CN)e]. . . . . . . . . . . , . Prussian blue.. . . . . , . . , . . Turnbull's blue.. . . , , , , . . "Iron blue ex. HC1". . . , . .414[Fe(CN)s]s... . . . . . . . . S C , [ F ~ ( C N ). ~. .], ~. ... . , . . In,[Fe(CN)Ja . . . . . . . . . , . Zna[Fe(CN)& . . . , . . . . . . Cds[Fe(CN)&. , . . , . . . . . . C U ~ [ F ~ ( C N. .) .~. .] .~, ,. . . ,
0.4 N 0.4 N 0.3 N 0.2 N 0.3 N
Tic14 TiC1, InC1, FeClr FeCla 0 . 2 N FeCln (See text) 0.3 N AICls 0.3 N S C Z ( S O , ) ~ 1.5N InCla 0 . 2 N ZnClz 0 . 2 N CdCll 0 . 2 N CuClz
Iron-oyanide
0.3 N 0.4 N 0.3 N 0.2 N 0.4 N 0.3 N
KsFe(CN)@ K,Fe(CN)6 KsFe(CN)a KaFe(CN)e K,Fe(CN)# KsFe(CN)a
0 . 4 N K4Fe(CN)6 0 . 4 N K4Fe(CiY)6 2.4 N KSe(CN)b 0.3 N KaFe(CN)e 0.3 N KaFe(CN), 0.3 N KaFe(CN)6 0 . 4 N K4Fe(CN)6 0 . 3 N KaFe(CX)6 0.0004 N K4Fe(CN)s 0 . 4 N K,Fe(CN)e
IQUIYALENTs IF METAL ION PER EQUIVAENT Of' IRONCYANIDE ION
2.0 1.7 1.7 1.7 2.0 2.0
2.0 2.0 1.3 2.0 2.0 2.7 2.0 1.7 2.0 2.0
acteristic brown sol of ferric ferrocyanide was first flocculated by the addition of ethyl alcohol. The white gels of aluminum and indium ferro- and ferricyanides were washed and dried as rapidly as possible in order to prevent decomposition with the formation of a trace of Prussian blue or Turnbull's blue. Indium ferrocyanide prepared from indium chloride and hydroferrocyanic acid is unusually stable, as evidenced by its relatively permanent white color. In an attempt to increase the size of the primary particles of Prussian blue br Turnbull's blue, solutions of 12 N hydrochloric acid were saturated a t 100OC. with each of the blue compounds and were cooled slowly. The resulting relatively large crystals gave identical x-ray diffraction patterns which were very sharp. Attention is called to additional lines of weak intensity in this pattern which cannot be indexed on the basis of a face-centered cubic unit of the size
HEAVY-METAL IRON-CYANIDES
101
suggested for the other isomorphous heavy-metal iron-cyanides. Because of the lack of sufficient information concerning this material it has been designated as “iron blue ex. HC1” in table 1and figure 1.
-‘1
+el
I
I
011
91
I
I,,
I
11
I II
I , 1 1 1 1 1 1 1
I,
I
I t
I I I . , I I
FIQ.1. Chart of x-ray diffraction patterns B. X-ray examination X-ray diffraction patterns were obtained for the various iron-cyanide gels listed in table 1, using Fe K, x-radiation (manganese-foil filter). The indium ferro- and ferri-cyanide salts were examined with Cu K, x-radiation (nickel-foil filter), using a sheet of aluminum foil between the sample and the film in order
102
H. B. WEISER, W. 0. MILLIGAN AND J . B. B A T E S
to diminish the effect of fluorescent x-radiation from the indium and iron. The relative intensities of the diffraction lines from the samples yielding rather sharp patterns (cu,[Fe(CN)~]z,Cda[Fe(CN)~lz,and “iron blue ex. HCl”) were measured with a Type B Moll recording microphotometer. The remaining samples listed in table 1 gave more diffuse patterns, and the relative intensities of these diffraction lines were estimated visually. The results obtained are given in chart form in figures 1 and 2. I
C . Density measurements
The density of dehydrated samples of Prussian blue and Turnbull’s blue was determined by Cnlbertson’s method (2), using xylene as the pycnometric liquid. Before making the density measurements, the adsorbed water was removed by subjecting the samples to the vacuum of a Hyvac pump for 72 hr. a t 67°C. It
L A[FEE NIJ
FIG.2. Chart of x-ray diffraction patterns
had been previously found (11) that such treatment removed practically all of the adsorbed water. The following values were obtained: Prussian blue, d;:’ = 1.785; Turnbull’s blue, d i r = 1.790. DISCUSSION
4.T h e crystal structure of heavy-metal iron-cyanides The similarity of the x-ray diffraction patterns of the heavy-metal ironcyanides given in figure 1 suggests that they possess a similar crystalline structure and are isomorphous. As first suggested by Keggin and Miles (3) and confirmed by Rigamonti (7) for ferrocyanides, and by van Bever (8) for ferricyanides, the systematic extinctions in the x-ray data are, in general, eharacteristic of face-centered cubic symmetry. The present authors suggested (10) in 1938 that the almost identical crystal structures of the relatively large ironcyanides resulted from such an arrangement of the relatively large iron-cyanide
103
HEAVY-METAL IROX-CYANIDES
ions that channels were formed in which the smaller heavy-metal cations may be grouped. The object of the present investigation is to show that combined water molecules are not necessary in the structure proposed by van Bever (8), and to extend the results to ferro- and ferri-cyanides containing trivalent and tetravalent cations, especially such cations as would not be expected to undergo oxidation or reduction during the formation of the gels. The following space groups possess face-centered cubic symmetry: TZ,Ti T:, T:, T",03, O', O!, OE,O:, and 0;. It is assumed that the large iron-cyanide ions are arranged in anion-anion contact in the 4(a) positions,2 and therefore the T:, Ti, 0', Oi, Ol, and 0;space groups may be excluded because of the absence of one set of four equivalent points. Van Bever's (8) observation was confirmed 1 = 4n 2) that the intensity ratios of the ( h b 1 = 4n) to the (h k
+ +
+ +
+
TABLE 2 Crystal-structure data FOUR
Atomzc Positions: 4 4 Fe a t : 4hIat: 4 0 to 4 Jf a t : 8 24 C a t : 24 24 S at: 24
positions (b) positions (c) positions (e) positions (e) positions (a)
SPACE Ciaom ot, IFe(CN1sl OROCPS
OR
~k
PER UNIT CELL
( C H M , O ~ ~+ *)
(ooO,Ot+ 4) (ooO,O+) -)
(W0,Ot) +) (000,Of&--+I
+ f&+ + !?it,4fS)
+ (2100 +, ZLW+) + (2200 2200 -1 4,
Parameters: ZI = 7/36 U O ; 2: = 11/36 aa
*
-
cyclical change of coordinates
reflections demand four of the metal cations to be placed in the 4(b) positions. The T z and 7': space groups may be excluded because of the absence of a set of eight equivalent points. Both the 0," and TE space groups lead to the same calculated intensities. On the basis of the 0;or T: space group, the atomic positions are assumed to be as given in table 2. The values of the parameters z1and x1,as listed in the table, are identical with those of van Bever. -4 diagrammatic representation of one-eighth of the unit cell is given in figure 3. The four iron-cyanide groups per unit cell determine the number of molecules of heavy-metal iron-cyanide compounds in the unit cell. For example, there are two molecules of Y13[Fe(C?;)B]2and four molecules of h12[Fe(CN)6]per unit cell. It follows that the number of metal cations to be placed in the unit cell varies from compound to compound. I t has already been assumed that four metal cations are in the 4(b) position, and that varying numbers of metal cations (table 3) remain to be
* The nomenclature in this paper follows the usages in Internattonale Tabellen zur Bestimmuilg uon Krzstallstrukturen, Gebrbder Borntrager, Berlin (1935).
104
E. B. WEISER, W. 0. MILLIGAN A N D J. B. BATES
placed in the unit cell. In the proposed unit cell, geometrical considerations suggest that there is space for these metal ions to be distributed over the S(C) positions (centers of cube octants). (I) Intensity calculations. In order t o test this proposed structure, the cdculated and observed relative intensities of Prussian blue, Turnbull’s blue, Tb[Fe(CN)6]r, Ti[Fe(CN)e], Fe[Fe(CN)s], Sc,[Fe(CN)s]a, Cds[Fe(CN)&,
4
@
24 C I N 2 4 ( 4 ) POSITIONS
0
24 N I N 2 4 ( 4 ) P O S I T I O N S
@
Fa
I N *(a)
POSITIONS
4 U
IN *(hi PQS1TIOMS
0-4 M
I N 8(s) POSITIONS
FIQ.3. Diagram of one-eighth of the unit cell &[Fe(CN)a]p, and Cu,[Fe(CN)6] were compared. The following equation ww used :
where
I = intensity, k = constant,
+(I
+ cos2 28) = polarization
factor, j = multiplicity factor, and F = crystal structure factor.
The observed and calculated values for the relative intensities of the x-ray reflections for these compounds are compared graphically in figures 4 and 5.
HEAVY-METAL
IROX-CYANIDES
105
The agreement is good except for the (200) reflections. Relatively long cxposure times were required to obtain measurable rcflrctions a t large angles, and therefore the lines at small angles were ovrrexposed, with thr result that in some cases the microphotometer tracings showed complete or nearly complete blackening of the film at small angles. The observed (200) reflection appears weaker relative to the other reflections, thus accounting, at least in part, for the apparent discrepancy between the observed and calculated relative intensities at small
IETAL ION1
PER ITNIT
EXAMPLE
CELL
3 4 5.33 6
5
TI
p%]
- - CALCULATED -OBSERVED
I
~
I
HKL FIG.4 . Comparison of calculated and observed intensities
angles. S o attempt was made to obtain photogiaphs of shorter rxposu~’e times in order to diminish the blackening at small angles. Van Hever (8) assumed that some of thr mrtal ions were distributed o v ~ th(. r 32(f) positions beeaure of the intensity ratios of the (311) and (331) reflections, and brrause the 8 ( c ) positions werc brlievcd to he filled with water molecules. In order to test this possibility, intenFity calculations wcrv made on thc asstimption that the cations not in the 4(b) positions arc distributed ovcr t h r 32(f) positions instead of the 8(e) positions 3.- assumctl above. These calculated intrnsitirs are not in good agrremcnt with the o l ~ x r v e dvalues. This is illiistrated for radmium frrricyanide in figure 5 .
1OG
H. B. WEISER, W. 0. h1ILLIGAN AND J. B. BATES
->
------_____
111 200 220 222
/
311 400
331 42 0 422
333 440
531 GOO 620 533 622 444
711 640 642 I
I
I
I
I
I
111 200 220 222
311 400
331 42 0 42 2
333 440
0
c W V
-n -n p1
531
V
600
0
620
533 622 444
711 640 642
J
0 4
z D
-
m o
-------.
107
HE.IVY-JfET.IL 1RON-CT.iSIDES lSlTY I-
O
N
P
~
C
D
I
I
I
I
I
I
O I
I
I
I
I
I
- - - - _ _ -_- - - _ _ _ ___----
4
W V
c
v) v1
z D m
r c m
I
I
I
I
I
I
I
I
I
I
I
I
-------_-__ ___------
7 ;;, c
I
I
> I
I
I
I
I
- - - - -_-------
7-
108
H. B . WEISER, W. 0. MILLIGAN AND J. B. B.4TES
(2) Efecl of radius of rnekzl ioa. If it is assumed that the relatively large iron-cyanide ions are spherical and in anion contactd the radius of the ironcyanide ion is found to have an average value of 3.6 A . Referring to figure 3: it is seen that metal cations in the 4(b) position fit in the space along the cube edge between the iron-cyanide groups. On the basis of the lattice constant (a, = 10$2 to 10.66 d.) and the calculated size of the spherical i;on-cyanide ion (ca. 3.6 A), the available space for such metal ions is ca. 1.6 A. One would expect the iron-cyanide ion not to be spherically symmetrical, but to bulge out along the cube edges. Thus, the actual seace available for the metal cations in the 4(b) position should be less than 1.6 A , ; this was observed experimentally. Divalent cadmium (radius = 0.97 d.)and trivalent lanthanum (1.15 b.)cations are considerably larger than any other cations studied. It was observed (figures 1 and 2) that cadmium was the largest cation the iron-cyanide salt of which gave the isomorphous face-centered cubic structure (a, = 10.66 d.). Lanthanum and neodymium iron-cyanides (figure 2) gave more complicated structures. However, even the x-ray diffraction pattern of cadmium ferrocyanide gel gave somewhat different relative intensities than the other members of the isomorphous series. It is believed, therefore, that the relatively large cadmium ion somewhat distorts the lattice, whereas the larger lanthanum and neodymium ions disrupt it completely, resulting in an entirely different structure. The various heavy-metal iron-cyanides studied in this investigation may be classified according to the number of metal cations per unit cell, as given in table 3. I t appears that heavy-metal ions within a certain size and polarization range can fit into the channels and interstices between the large iron-cyanide ions, in numbers sufficient to satisfy the valence requirements without materially modifying the lattice.
B. Prussian blue and Turnbull’s blue In the preceding discussion it was concluded that iron-cyanides of both type .IrXa(e.g., Scr[Fe(CS)6]r)and type .I& (e.g., zn,[Fe(CS)s]~),wherein oxidation and reduction is extremely unlikely, possess an isomorphous face-centered rubic structure. Accordingly it was not found possible to distinguish between these AI& and .I& types of compounds, in particular between the formillas Fer[Fe(CS)6]3and Fe3(Fe(Cs)6]?ordinarily written for Prussian blue and Turnbull’s blue, respectively. Under these circumstances, physical and chemical evidence has been relied upon to make further deductions regarding the exact composition of the blue iron-cyanides. In table 1 are given possible theoretiral arid calculated molecular \\rights of Prussian blue and Turnbull’s blue. Thc ralrulated molecular weights are baaed on-the density values reported above and on the lattire constant, a. = 10.18 A. From these considerations it is likewise not possible to distinguish between the formulas Fea[Fe(CS)6]3and Fes[Ft.(CS)6]z,because of the clow correspondence of the density values. Oxidaliu,I-reductio,~ pofrntials. Chemical evidence is available which suggests
109
HEAVY-METAL IRON-CYANIDES
that both Prussian blue and Turnbull's blue may be represented by the formula Fe3[Fe(CK)6]z. Thus the oxidation-reduction potential of the reaction Fe3+ e -+ Fez+
+
was found to be -0.772 by Bray and Hershey (1). The oxidation-reduction potential of the reaction [Fe(CN)a13- e [Fe(CN)s14-
+
was found to be -0.36 by Kolthoff and Tomsicek (4). librium constant, K , calculated by the equation,
The value of the equi-
RT nF
AE=-lnK TABLE 4 Theorefical and calculated molecular weights of Prwaian blue and Turnbull's blue MOLECULAR WEIQHTB CALCULATED PROM DENSITY
MOLECULFB PER UNIT CELL
Pruesisn blue
1 1.33 2 3
1
rouMuLA
AMOU"I8 MIXED
170 194
1,
Turnbull's blue
1142 856 571 381
TEEORXTICAL MOLECULAR WEIGHT FROM FOBMULA
Fe,[Fe (CN) I. .............................. Fer[Fe(CN)&. ..............................
XIFe(CN)a
1
1139 854 569 379
859 591
CAEULATED
FeClr
XIFe(CN)s
99.7 49.8
70.3 144.2
1
I
OBSERVED
KaFe(CN)s
33.2 16.6
1
j 1
K