The Nature of Prussian Blue

Prussian Blue is ferric ferrocyanide, Fe4[Fe(CN)e]3, in which thepotassium of the soluble blue is replaced by ferric ion. Turnbull's Blue is made upof...
4 downloads 0 Views 295KB Size
THE NATURE O F PRUSSIAN BLUE BY DAVID DAVIDSON AND LARS A. WELO

We owe our knowledge of the constitution of Prussian Blue largely to the work of Erich Muller and his students', whose results may be summarized as follows: There is an equilibrium in solution between ferric and ferrocyanide ions on the one hand and ferrous and ferricyanide ions on the other. I11 I1

Soluble Prussian Blue is potassium ferric ferrocyanide, KFe[Fe(CN)c]. It is obtained from equivalents of ferric chloride and potassium ferrocyanide or from equivalents of ferrous chloride and potassium ferricyanide. Insoluble I11 I1

Prussian Blue is ferric ferrocyanide, Fe,[Fe(CX)&, in which the potassium of the soluble blue is replaced by ferric ion. Turnbull's Blue is made up of three molecules of the soluble blue in which two of the potassium atoms have I1 111 I1

I1 111 I1

been replaced by ferrous iron, hence KFe[ Fe[Fe (CK) or KFeFea[Fe (C N)613 . Recently, Reihlen and Zimmerman2 have proposed a constitutional formula for Soluble Prussian Blue in which the two iron atoms are combined in the forin of a complex polynuclear ion, thus:

Cambi' has proposed a similar formula. T o experimental evidence, however, has been advanced in support of such a structure. The present paper deals with the results obtained in the attempt to distinguish between the simple and the more complex structures. Neither the conductivity measurements in solution nor the magnetic measurements in the solid state have produced any evidence of the existence of the postulated polynuclear ion.

Conductivity Measurements As may be seen from the corresponding formulae, Reihlen and Zimmerman's structure would yield only two ions in solution, whereas the simple formula indicates the occurrence of three. Since the insolubility of Soluble Prussian Blue' renders conductivity measurements impracticable, attenI11 111

tion was directed to the closelyrelated ferric ferricyanide, Fe[Fe(CX)6J,which is obtained by treating ferric chloride with potassium ferricyanide. Applying the ideas of Reihlen and Zimmerman to this compound, its structure becomes:

* From the Laboratories of the Rockefeller Institute for Medical Research, New York.

I 192

DAVID DAVIDSON B S D LARS A. WELO

That is to say, while according to the simpler formulation, ferric ferricyanide appears to be a binary electrolyte, according to the second formulation it is manifestly a non-electrolyte. It is actually found, however, that a mixture of ferric alum and potassium ferricyanide solutions has a conductivity only slightly less than that of the component solutions and much greater than that of an equivalent amount of potassium and ammonium sulphates. The specific conductances are given in Table I. TABLE 1 Specific Conductances a t 0.I

0.I 0 .2

0.3

25’

M K3Fe(CN)6 M NH4Fe(S04)2.1zH 2 0 M KBFe(CN)6 0 .2 M XH4Fe(S04)2.12 H20 M &SO4 0 .I M ( S H 4 ) 2 S 0 4

+

+

0.00393 Mhos. 0.00310 ’’ 0.00645



0.00480



Magnetic Measurements Many simple ferric salts, such as FeC13, Fe2(S04),,etc., have been extensively studied. For the purpose of the present paper it is only necessary to state that, in general, solid paramagnetics obey the generalized Curie law. K, (T-8) = C,; that is, I / K &= ( I / C J (T-8) where K,is the susceptibility per gram atom of the paramagnetic ion, T is the absolute temperature on the centigrade scale, 8 is a constant related to the molecular field, and C, is the Curie constant referred to the gram atom. But the constants C, (reciprocal of the slope) and 8 (intercept on the temperature axis) take different values according to the temperature range. The curve showing I/K, plotted against T consists generally of a series of straight lines with well-defined breaks if the range of temperatures is great enough and if the points are close enough together. We have made a close study of the observations for all of the six ferric salts measured by Ishiwaras and Honda and Ishiwara6, and the ferric sulphates measured by ThBodoridirs’ and by Onnes and Oosterhuis8, in order to determine the range of values of C, to be expected in any iron

TABLE I1 Resume of Magnetic Data on Ferric Salts in the Solid State Corresponding number Authors Salt Curie constant. C,. FeC13 FeCI,z(XHKl) 3[FedS04)d XH4Fe(S04)2 N H 4Fe(SO4) Fe(C&Od3 +[FedS04),l SH4Fe(S04)2

+

3.80, 4.08 H20 3.98, 4.29, 5.30 3.85, 4.22, 4.71 1 2 H20 4.43 3.941 4.08, 4.14 3.92, 4.31, 4.81

+

12

+

4.233, 4.245

H2O

4.26

of Weiss magnetons, P. 2 7 . 4 , 28.4 Ishiwara, and 2 8 . 0 , 29.1, 32.4 Hondaand

27.6, 28.9, 30.5 29.6 27.9, 28.4, 28.6 27.9, 29.8, 30.8 28.93, 28.97 29.0

Ishiwara

ThBodorid6s Onnes and Oosterhuis

I193

T H E N A T U R E O F PRCSSIAN B L U E

salt giving the simple ion9 Fe++f. They appear in Table 11. Values calculated from doubtful line segments were rejected. Values of the constant 0 are not shown as we are not concerned about them in this paper. Following common practice we include corresponding values for the Weiss magneton number calculated by the formula P = ?!!! = 14.07?!/, the gas 1123.5 constant R being 83.1j X 106. The ferrocyanides have long been known to be diamagnetic. We have measured the susceptibilities of the ferrocyanides of hydrogen, potassium, sodium, and calcium in the attempt to evaluate the gram atomic susceptibilities of the iron in them. The results appear in Table 111, where K = specific susceptibility. K, = molecular susceptibility. K, = gram atomic susceptibility. Kd = sum of gram atomic susceptibilities of constituents other than iron.

TABLE 111 Magnetic Data on the Ferrocyanides Molecular weight

K

x

xo8

216 -0.328 1 3

K,

x

106

Kd x

108

Mean .(I x 1o8K. X xo6

-70.8

-73.8

+j.o

475 - 0 . 4 7 4

&3

-225.0

-224.0

-1.0

481 -0.465

iI

-223.5

-231.8

+8.3

$4.3

To calculate the values of Kd shown in the table, the following gram atomic susceptibilities given by Pascal10 were adopted: H = - 2.9 X IO-^, K = - 18.j C = - 6 X IO-^, N = X IO-^, X a = - 9.2 X IO+, Ca = - 15.8 X - 5 . j X IO-^, HzO = - 13.0 x 10-6, and the correction 0.8 X IO-& for each of the C N groups. As the table shows, the mean of the four values of the iron atomic susceptibilities is K, = +4.3 x 104. It is to be expected that the individual values of K, should fluctuate considerably because cf experimental errors in K and because of doubt that Pascal’s diamagnetic constants are strictly applicable to these compounds. Errors in the determination of K, for iron in the ferrocyanides are of no importance in connection with the subject of this paper. Here we are concerned with Prussian Blues having one or more ferric ions per molecule, each having a n atomic susceptibility of the order 14000 X IO-^ to 2 1 0 0 0 X IO& within the range of temperature used. The paramagnetic -__ contribution of iron, if any, in the ion Fe(CN)6 is itself entirely negligible.

+

DAVID DAVIDSON AND LARS A. WELO

1194

The magnetic properties of soluble and insoluble Prussian Blue as they have been formulated in the foregoing discussion can now be predicted. SoluI11

I1

ble Prussian Blue, KFe[Fe(Ch’)& should behave as a double potassium ferric salt of hydroferrocyanic acid in which all of the paramagnetism is due

FIQ.I

+++

to the ion Fe. The Curie constant should be, according to Table 11, not less than 3.80 and might be as high as 5.30. In the insoluble Prussian Blue, I11

TI

Fer[Fe(CN)&, only four of the seven iron atoms per molecule play a part in the paramagnetism. We may assume that each of the four contributes equally. We should obtain again, for each ferric ion, a Curie constant within the limits 3.80 and 5.30, The corresponding limits of the Weiss magneton numbers should be 2 7 . 4 to 32.4.

THE NATURE O F PRUSSIAN BLUE

1

I95

Each Prussian Blue was measured a t three temperatures: that of the room; when surrounded by melting ice, giving a final temperature Jf Z . I O C ; and when surrounded by solid COz, giving a temperature of - 7 0 O C . These are sufficient in number and extend over a large enough range to determine whether or not the Curie law is obeyed, and, if so, to enable us to determine the constant C, and if we so desire, 0. The results are shown in Table IV and in Fig. I . The meaning of the symbols T, K, K,, Kd, and K,, has already been stated. K is the dumber of iron atoms per molecule assumed to be responsible for paramagnetism and Kht is the molecular susceptibility_ _K,_ - corretted for diamagnetism and for the small positive residual of iron in Fe ("e by subtracting the quantities Kd. As shown in the figure, the relation between I/K, and T is accurately linear for each of the Prussian Blues throughout the range studied and in accord with the Curie law which may be written in the form, I/K,= (I/Cs)(T-O). The Curie constant-C. can - be '11 - ' 1 2 determined from any two points on the line by the formula, C,= I,KsI - I,,Ks2 T2Ka2 - TiK,, . and, similarly, 8 = K.2 - KaI. TABLE IV Magnetic Data on Prussian Blues hlole-

PrussianBlue

cular N weight

Soluble KFe[Fe(CS)6] 341 -k I.gH20

T

300.5 I

KXIOB K,XIO@ KdXIO@X ~ I X I O ' K ~ X I OI ~

K

14460 -103 275.2 46.73 15940 -103 203.1 65.40 22290 -103 42.45

14563 14563 68.68 16043 16043 62.32 22393 22393 44.70

Insoluble 300.6 49.05 55000 -368 55368 13842 72.35 Fe,[Fe(cX)& 1 1 2 0 4 275.2 53.75 60250 -368 60618 I j I 5 4 66.00 -k 14.5H20 203.1 75.00 84000 -368 84368 21092 47.40 It is convenient to take the values of T and I/K, from the graph. I. Soluble Prussian Blue. When TI = 300, = 68.60. Ka, 300 - zoo c, = 68.60 - 43.90

2.

-

IO0

24.70

=

4.0j.

Giving for the magneton number, P = 14.0jfi = 28.3. For Soluble Prussian Blue, 0 = + 2 2 . Insoluble Prussian Blue When TI = 300, = 72.20. Ka1 T1 = zoo, = 46.70. KC2 300 - 2 0 0 c, = 7 2 . 2 0 - 46.jo - -2150.05 = 3.92.

1196

DAVID DhVIDSON AND LARS A. TVELO

Giving P = 14.0743.92 = 27.8. For Insoluble Prussian Blue, 8 = 14. j . We see, then, that the treatment of the magnetic data dzmanded by the for-

+

111 I1

111

I1

mulations XFe[Fe(CN),J and Fe4[Fe(CN)6]3 for soluble and insoluble Prussian Blue, respectively, leads to the Curie constants 4.0; and 3.92 and the magneton numbers 28.3 and 27.8 which are within the permissible limits for ferric ions, according to Table 11. It is of interest to treat the magnetic data on the basis of the alternative 111

formulations as polynuclear compounds : K[Fe2(CS),] and Fe[Fez(C S )61 3 . In the case of the soluble Blue we must assume that each iron atom in the ion Fez(Ch'js plays the same part and contributes half the total. The Curie constant per gram atom of iron,

and the magneton number, P = I4.074/2.02j = 2 0 . 0 These values of C, and P have never hitherto been observed in any salt of iron. I n the case of the insoluble Blue the measured vnlue of the Curie constant, referred to the gram molecule, is c ~ =t 4 x 3.92 = 15.68 From our measurements on soluble Blue, the contribution of each ion Fez(CS>6 111

is 4.0j and for the three ions Fe2(CN>8 in the formulation Fe [Fez(CN)6]3J

+++

3 X 4.05 =

The Curie constant C, referred to the ion Fe becomes C, = 15.68 - 1 2 . 1 5 = 3.53 and the magneton number, P = 1 4 . 0 7 4 3 7 3 = 26.4 These values of C, and P are, indeed, observed in iron salts, but they are characteristic of the ferrous ion in ferrous salts and not of the ferric ion re12.1

j.

111

quired by the formula Fe[Fez(Ci\')e]3. In both the soluble and the insoluble Blue, therefore, the formulations as polynuclear compounds are inconsistent with the known magnetic properties of iron salts The authors are indebted to Dr. Oskar Baudisch to whom their interest in the subject is due and to Dr. P. G. Colin who very kindly made the conductivity measurements. References J. prakt. Chem., 79,8 I (I 909) ;80,I 53 (1909) ;84,353(191I) ;89,68 ( I 913 1 ;90.I I 9 ( I 9 14!. *Ann., 451,75 (1927). Atti. Accad. Lincei, 79, 324 (1927) 4 The name "Soluble Prussian Blue" simply indicates t h a t the substance readily forms colloidal solutions. 5 Ishiwara: The Science Reports, Tohoku, 3, 303 (1914). 6Honda and Ishiwara: T h e Science Reports, Tohoku, 4,2 1 j (191j). ThBodoridBs: J. Phys. Radium, 3, I (1922). 80nnes and Oosterhuis: Comm. Phys. Lab., Leiden, S o . 139. j7 (1914). We hope soon t o make the details of this study of ferric and other salts the subject of a separate communication. l o Revue g6ndrale des Sciences, July I j (1923). December 22, 1937. 1

'