T H E PHASE DIAGRAM O F T H E SYSTEM SILVER IODIDE-LEAD IODIDE*
BY FRANK E. E. GERMANX A S D CHARLEY F. METZ
N-hile working on the conductivity of solid crystalline compounds, Tubandt and Eggert,' made a study of the phase diagram of the system silver iodide-lead iodide, and from the results obtained, concluded that these two substances formed a compound having the empirical formula 4AgI,.PbI,. From their curve, Fig. I , it is evident that the point represented by the above formula may be merely a suppressed maximum. Certain authors call
FIG.I
such points compounds when the composition is such as can be expressed by a simple formula. There is, however, no justification for this procedure. It appears that Tubandt and Eggert did not determine time cooling curves. The present investigation is devoted to a complete study of the above system by thermal analysis in an attempt to settle definitely the question of the existence or non-existence of a compound between the two iodides Theoretical Considerations. The two-component system studied in this work seems to be similar to the system represented by Fig. 2 , which may be considered a general case. The exact similarities and differences between the
* Contribution from
the Department of Chemistry of the University of Colorado. ITubandt and Eggert: 2. anorg. allgem. Chem., 110, 196 (1920).
PHASE DIAGR.4M OF SILVER IODIDE-LEAD IODIDE
I945
general case and the system AgI-PbIz are pointed out in the following paragraphs. First, howl-er, the behavior of solutions of different concentrations, in the general case will be discussed. Suppose a mixture having an initial concentration between point E and pure B of Fig. 2 , be heated to the molten condition and allowed to cool slowly. This is represented by dotted line S o . I in Fig. 2 . Khen the temperature has fallen to the point (11) the I-curve is reached and solidification begins. The composition of the crystals which separate will then be given by the point (si) of the s-curl-e which corresponds to the temperature (II). Assuming the separating crystals are constantly maintaining a condition of equilibrium
with the melt, this melt will have solidified to a conglomerate of mixed crystals all possessing the composition (s') of the original mixture, by the time the temperature has fallen t o (s'). In consequence, the cooling curve of this mixture can show but one interval, namely, one reaching from (1J to (s'). This cooling curve is represented graphically in Fig. 3 by curve Ko.1, the portion of the curve between (a) and (b) being the crystallization interval. Suppose now a mixture having an initial concentration represented by dotted line Xo, z be cooled from the molten condition as before. Initial crystallization will begin a t a temperature (12) on the 1-curve, and these crystals will have the composition ( s ~ ) . When the temperature has fallen to that of the horizontal CDE the mixture consists of melt of composition C and crystals, which on account of the previously assumed ideal concentration balance, are uniformly of concentration E. If heat is further removed from the system further lowering of the temperature does not a t once result. The first thing that occurs is transformation of B-rich saturated mixed crystals E plus melt into saturated mixed crystals having composition D. This reaction persists until the melt is entirely exhausted. This crystallization change
1946
FRANK E. E. GERMANN A S D CHARLES F. YETZ
is represented graphically in Fig. 3 , curve 50.2,by that portion of the curve between (b) and (c). That portion of the curve between (a) and (b) represents the initial crystallization beginning a t (12) and ending a t the temperature of the horizontal CDE. When point c is reached further abstraction of heat causes temperature fall along the lower branch of the cooling curve. Now, it can readily be seen that the time of duration of this second halt (bc) depends upon the composition of the mixture. When the original composition is that of point D, the duration of this halt will be a maximum, because a t this concentration only, will the solid phase, a t the close of crystal-
T i me FIG.
3
lization, consist only of rrystals of concentration 1). .It, the bottom of Fig. z is found the length of this halt plotted against composition, and is labeled XRP. This is the only method by which point D can be located by thermal analysis methods. If a third mixture, represented in composition by dotted line S o . 3 is cooled from the molten condition, initial crystallization occurs a t the temperature (13). These crystals have the composition (ss). Khen the temperature has fallen to that of the horizontal CDE, the xhole mixture consists of melt C and U-rich crystals of conccntration E, as did mixture S o . z a t this temperature. On further abstraction of heat, transformation of these B-rich crystals plus the melt into the crystalline variety D occurs. When this reaction has proceeded to completion, no crystals of concentration E are left, and the mixture consist,s of crystals of composition I> and melt C. Thus far, the time-cooling curve is identical to curve KO. z of Fig. 3 , and is also represented by curve S o . 3 of the same figure, down to point (c). Further solidification of the melt C follows along the branch CG of the 1-curve, whereby the composition of the crystals with which it is in equilibrium is given by the branch DF. In accordance with our assumption of complete concentration balance between crystals and melt, the latter will have become
PHASE DIAGRAM OF SILVER IODIDE-LEAD IODIDE
=947
completely solidified by the time the temperature ( s f “ ) is reached, the solid phase consisting of a conglomerate of homogeneous crystals of this composition. This last crystallization is represented by a break in the cooling curve, represented by the (cd) section of curve No. 3, Fig. 3. In case of sluggish crystallization and in the absence of complete equilibrium between melt and crystals, the (bc) and (cd) portion of curve No. 3, Fig. 3 are not well defined, and there results a cooling curve similar to No. 4. Thus the determination of the maximum length of the crystallization interval a t the temperature of the horizontal CDE is rendered practically impossible. The system studied in this work seemed to present these difficulties. A careful study of Fig. 2 will reveal the fact that the behavior of mixtures having concentrations between C and pure D is not unlike the behavior of a system showing a “suppressed maximum.” Thus, in spite of the fact that there has been some controversy over whether a crystal of concentratior D should be called a compound, it is believed there can be no serious objection to this, providing the composition a t that point can be expressed by a simple formula. The above theoretical considerations are offered in some detail because it seems that the system studied in this work belongs to a limiting case of the above where point F coincides with G, and point E coincides with pure B, or practically so in each case. Experimental. The silver iodide was precipitated from a solution of silver nitrate in pure water by means of a solution of potassium iodide. The lead iodide was precipitated from a solution of lead nitrate in pure water by the addition of a solution of potassium iodide. The silver and lead nitrates and the potassium iodide were all Mallinckrodt’s C.P. quality. After precipitation, each precipitate was repeatedly washed by decantation with distilled water, until a portion of the decanted solution would not give a test for nitrates when diphenylamine in concentrated sulfuric acid was used as testing reagent. Both precipitates were then dried on porcelain plates for twenty-four hours a t room temperature, and then placed in an oven a t 110°C for six hours. They were then pulverized and dried again for six hours a t IIOOC.Finally they were placed in colored glass desiccators over anhydrous calcium chloride, and kept there until used. The preparation of both iodides was carried on in the absence of daylight due to the fact that h g I is rather easily decomposed by light, while in the moist condition. In order to obtain good cooling curves, a Hoskins combustion furnace, well insulated to allow slow cooling, was used. This furnace was placed in a vertical position in a straight side iron bucket, with a diameter of twelve inches and a height of fourteen inches. The bucket was filled with fine dry sand. This served as a good heat insulator as well as a heat reservoir. The furnace so insulated against rapid radiation was used in all this work, giving excellent satisfaction. Because of its high thermoelectric power, an iron-constantan thermocouple was used for obtaining the temperatures. One wire of this couple consisted of iron, analyzing 99.7y0 having a diameter of 0.320 mm. The
I 948
FRASK E . E . GERMAhT AND CHARLES F. METZ
other wire was of “constantan” having a composition of 60% copper and 40% nickel, covered with black enamel, and having a diameter of 0.30 mm. The hot junction was protected by a quartz tube closed a t one end, having an outside diameter of 3 mm. and an inside diameter of z mm. The length of the tube was 2 5 cm. A small pyrex glass tube just large enough to slide over the iron wire, and of a length equal to that of the quartz tube, was placed over the iron wire, next to the junction. This was for the purpose of insulating the two wires of the couple. The junction was then slipped into the quartz tube. The cold junction was protected by a pyrex glass tube having approximately the same dimensions as the above. A small pyrex tube was used here also to insulate the two wires of the couple. The cold junction was kept by at oo C by means of a thermos bottle filled with ice. The protecting tube of this junction occupied a position in t,he thermos bottle such that it was not immersed in the ice water. I n other words, the tube protecting the junction was in contact with only the fine cracked ice. The couple was compared with a nickel-chromium couple whose temperature-E.M.F. curve was accurately known. The results when plotted gave practically a straight line. The potentiometer used in this work was a Leeds and Northrup Type K, capable of being read to 0.01 millivolts. The galvanometer used was a Leeds and h-orthrup wall type, having a resistance of 1150 ohms. An Eppley cell, having an E.1ZI.F. of 1.01889 volts a t 26°C was used as a standard. A working E.M.F. of two volts was obtained from one cell of an ordinary three cell, six volt storage battery. The different mixtures were kept in double pyrex test tubes and nitrogen gas was circulated above them for several minutes before heating in order to insure the absence of oxygen. A standardized set of weights and an accurate analytical balance mere used in obtaining weights of the two components of the mixtures. The furnace was previously heated to a temperature of 650’ C for twentyfour hours in each case After the mixtures had attained a temperature of approximately 6oo0C, the current was decreased by means of a resistance so the furnace would cool uniformly and slowly. Readings mere taken every minute and the data tabulated. The rate of cooling in the neighborhood of the freezing points of the different mistures was between one and two degrees per minute which was thought to be slow enough. Mixture No. 2 0 was repeated with the rate of cooling about o.s0C per minute, but the results were no different. Each mixture while in the molten condition, was a heavy viscous liquid, dark red in color. As it cooled down, it passed through an orange color to a yellow, The mixtures containing a large amount of silver iodide were a pale yellow, but this color changed progressively to a deep yellow, as the silver iodide content decreased. During the determination of each of these time-cooling curves, the working battery was kept balanced against the standard cell, and the cold junction of the thermocouple was kept a t 0°C in the manner previously described.
PHASE DIAGRAM OF SILVER IODIDE-LEAD IODIDE
I949
Results. The freezing points of the various mixtures were tabulated and plotted. Table I shows the results obtained. The centigrade temperatures obtained in Table I mere obtained from the calibration curve for the ironconstantan thermocouple. The time-cooling curve for each mixture was plotted and it was found the data obtained gave a very smooth curve. The results given in Table I are plotted in Fig. 4. The ordinates are E.M.F. in millivolts and the abscissae are per cent composition.
The following interpretation has been given of the diagram. Between the concentration 18.5 Mol% PbI2 represented by point B, and 5 7 . 5 Mol% Pb12 represented by point C, we have equilibria between pure .4gl and solid solutions of varying composition. Similarly between 8 j and 1007, PbI2 we have solid solutions in equilibrium with pure PbI2. .Is the temperature falls the solid solution first appearing breaks down. Between the concentrations represented by the points C and D the time-cooling curves showed two halts, the first halt being caused by pure PbI2 crystallizing from the melt, while the second halt was caused by a solid solution of composition C (57. j Moly, PbI2) crystallizing from solution. The length of time of this second halt grew less with increasing concentration of PbI2, and was scarcely noticeable a t a concentration of 85 M 0 l 7 ~PbI2. Point D then represents the limit of miscibility of AgI in PbI2. At concentrations greater than 85 Moly, PbI2, only one halt was obtained on the time-cooling curves of these mixtures.
I950
FRANK E. E. GERMANN AND CHARLES F. METZ
Point A represents pure AgI, and only one halt was obtained on the timecooling curve. Silver iodide crystallizes very sluggishly, also having a tendency to supercool. The cooling curve for this substance is far from ideal. Mixtures having an original composition of j , IO, 15, 15.7 j , and 16.66 Mol% PbIz each showed two intervals of crystallization on their time-cooling curves. The temperature of the second crystallization of these mixtures was identical
TABLE I Freezing Paint
Mixture NO. I 2
3 4
5 6 i
8 9
First
20.20
19.79 '9.43 19.08 18.68 18.32
17.93 1 ; . 7,;
I1
1;
I2
18.08
'3
18.40
14 15
18.70 19.0;
16
19.44
17
19.84 20.29
.80
I9
20,47
24
20.70
25
2I.lj
20
21.60 23.75 25.84 28,13
21
22
23
(Degrees Cent.)
20.92 20.58
IO
I8
E.M.F. Millivolts
in each case. The length of time of the second crystallization in each of these mixtures increased with increasing Pb12 content. However, owing to very sluggish crystallization, the curves were not ideal, making the determination of the length of time of this second crystallization rather inaccurate. In so far as this portion of the diagram has the possibility of a suppressed maximum, the relative lengths of time of the second crystallization in the above mixtures is the only method of determining the composition of the compound. Suppose a compound is formed and has the composition 5AgI.Pb12 corresponding to 16.66 Mol% PbI2, then the mixture having a composition greater than 16.66
PHASE DIAGRAM OF SILVER IODIDE-LEAD IODIDE
195'
Mol% Pb12 and less than 18.5 MolOJ, would show two crystallizations on its time-cooling curve, but at the end of the first crystallization (crystallization of the compound itself) there would still be some melt left having a concentration of B. I n this case the melt would begin solidifying a t the same temperature as the compound, thus increasing the total interval of crystallization. Thus it is impossible to determine the length of the crystallization interval of the compound itself. This particular difficulty is explained by time-cooling curve No. 4 of Fig. 3. I n view of the above we are inclined to believe that if a compound is formed, it probably has the composition jhgI.PbI2, since no other simple formula corresponds to any composition between 16.66 Molyo PbI2 and including 18.5 Molyo Pb12 (point B). The results obtained here prove conclusively that the two iodides do not form a compound having the formula 4AgI.PbI2, as stated by Tubandt and Eggert. In Fig. 4, below the horizontal FB, will be found the curve FG, which indicates the time of crystallization a t the temperature of FB plotted against composition. As was stated above, the cooling curves obtained mere not ideal, making the determination of these intervals rather difficult. However, the curve FG indicates in a general way the relative lengths of time of these halts. Referring again to Fig. I , we note a halt on the cooling curves of mixtures having concentrations up to and including 20 Mol% PbI,, a t a temperature of 144'C. I n the above region of concentrations, the time-cooling curves were determined in the neighborhood of the above temperature, and it was found that this crystallization was not present in mixtures having concentration greater than 15.75 Molyo Pb12. Just what significance this fact might have on the interpretation of the above results is not clear. However, it is known that h g I undergoes a polymorphic transformation a t this temperature, as well as another change a t about xxg0C. This being the case it might lead to the conclusion that possibly the formation of a compound a t concentration of 16.66 Mol% PbI2 would prevent or a t least slow up this change of the AgI so it would not be noticeable on a time-cooling curve. The melting point of silver iodide was found to be 558'C, while that of These results appear to be higher than lead iodide was found to be 412%. any previously reported. From what could be found in the literature, the highest reported value for the melting point of silver iodide was 556OC. The differencebetween this value and the one found in this work is only two degrees, which is within the limits of experimental error. The highest, reported value for the melting point of lead iodide, previous to this time was 402'C. The difference between this value and the one found in this work is ten degrees. It seems hardly possible to attribute this much of a difference to experimental error. The two sources of error most likely to creep in, in work of this kind are first, impurities in the salt, and second, incorrect calibration of the thermocouple. I n view of the fact that an impurity in a substance or compound usually lowers its melting point, it seems possible that this has been one of
1952
F R A S K E . E. G E R Y A S N AND CHARLES F. NETZ
the chief causes of some of the low results that have been reported in the literature. The salts used in this work were washed by decantation until entirely free from nitrates and it is believed they were as free from impurity as it was possible to make them. The thermocouple used in this work was calibrated in two different ways and since the results obtained checked very closely, it is believed this couple was accurate to within one degree over the entire range. In view of the foregoing, it is believed the results obtained are accurate to within one degree. Conclusions. I . The system AgI.Pb12 may possibly form a compound having the empirical formula 5 AgI.Pb12, but not 4AgI.Pb12. 2. The method of thermal analysis ordinarily used in determining the composition of the compound formed in cases of a “suppressed maximum” cannot be used in this case due to a region of formation of solid solutions, and very sluggish crystallization. 3. The system formed a series of solid solutions, the extreme limits of PbI2. composition being 18.5 Molyc to 5 7 . 5 Mol% PbI2, and 8 j to 1007~ 4. Silver iodide is miscible with lead iodide up to a concentration of approximately 15 Mol% AgI. 5 . The melting points of pure lead iodide and pure silver iodide have been redetermined and higher values than any previously reported have been obtained. Since impurities usually tend to lower melting points, it is believed that the new values are more nearly correct than any previously reported. The values suggested for adoption are, silver iodide 558” I’C, and for lead iodide 412’ =t1°C. 6. It appears that Tubandt and Eggert failed to find the region of limited miscibility, and the region of limited formation of solid solutions. From the diagram obtained in this work it is impossible to assume that a compound having the composition 4AgI.Pb12 is formed, which is miscible in an excess of PbI2. 7 . It is believed that the phase diagram for this system is the first of its kind to be reported. Although theoretical treatment has been given to this same t,ype of diagram, this seeins to be a limiting case. Even though the possibility of a “suppressed maximum” may exist, the composition of the compound cannot be determined by thermal analysis alone.
*