T ,788
TI.
w. I
m w
r .
i h i s is explained in the following iiiaii~ier:If w e add to the precipitate fresh blood solution, this dissolves the precipitate partially, and with a sufticient quantity, totally. This indicates that before the nearly i n soluble precipitate is fornied there exists in the solution another saltlike product containing less metal ions than tlie precipitate. I f now this first product be a very strong combination, we \vi11 have t o add quailtities of salt proportional to tlie concentr:ition of the blood solution (alivays used in the clunlitit!. ( i f 5 c'c. i het'ore tlie iiearly i~isolul~le 1)recipit:ite is formed. I n other ivords, the necessar!. quantitJ- ( ~ f agglutinating salt will be porportional t o the concentration. Xow the quantity of salt increases more slowly than the concentration, ivhich indicates that there exists a chemical equilibrium Iietn-een the tivo salt-like products, derived froin the blood solution anti the ions of the precipitatiiig salt. 'I'his l)eha\.ior cannot he easily explained from the point of vieiv of the colloidal scliool, just as u i t h the analogous heha\-ior of the aggliiti~iatingquantities. ill1 the facts i n this in\,estigatioii give stroiig arguments i n fa\.or of the physical-chemical tlieor!.. according to ~vhic>li the ohser\.ed phenomena are due t u chemical processes similar t o those encountered in general chemistry, and in which the chemical eciuilibria pl:~!. an iniportant rOle. NOI:Er. P l I Y S I C O - C I I I : ~ I l C . \ L INSTITIT&;. S T O C P I I O I . \ I .
Jurir
12.
[yo'.
ON THE NATURE O F PRECIPITATED COLLOIDS. liY 11.
\V,
FOOTIi.
Keceived J u l y 7 , 1908.
I n this article experimental ei.idence is given tending to show t h a t precipitated ferric and alurniniurii hytlroxides may be regarded as solutions of water in ferric or aluminium oxides or in one of their lower hydroxides or more generally t h a t they are solutions of a liquid i n a solitl. Nuch experimental Jvork has already been (lone on substances of this class. \'an Remmelen in particular has shown by a large amount of very careful work that colloidal precipitates like ferric hydroxide, aluminium hydroxide and silicic acid are not compounds of definite cumposition. He calls them absorption conipoumls. He has recognized clearly t h a t the moist precipitates contain both water mechanicall>. niised kvith the precipitate and absorbed Ivater, tliough he gives 110 means of distinguishing between the two.' He considers that the amount of absorbed water is dependent on the structure of the jelly and that this amount is changed b!. the method of formation, hy time, heating, and li!. thc 1)reserice of ioreign substances. € { e 11i;ilies the statemelit that Ii~tlrcigcls
' %. irtior,q
C.lirv,, 13,
2.4.r;
18,2 1
NATURE O F PRECIPITATED COLLOIDS.
1389
cannot be considered as solid solutions’ but later makes the contradictory statement? t h a t they do behave as such. Hardy3 has suggested t h a t the soluble gels, or heat-reversible gels, like agar are composed of a solution of agar in water and a solid solution of water in agar, and he has shown t h a t the concentration of agar in water solution changes with the temperature. He states, however, t h a t the (‘insoluble gels,” like silicic acid or aluminium hydroxide, are not to be regarded as solid solutions. The experiments described later show t h a t a moist precipitated ferric or aluminium hydroxide free from other substances behaves in many ways like a mixture of two partially miscible liquids such as ether and water. When ether and water are saturated with each other so t h a t two liquid phases are present, the system in general has the following characteristics: I . The composition of each liquid phase is constant a t constant temperature and pressure, independent of the relative amounts of the two phases and independent of the method of preparation. 2 . The composition of each phase changes with a change in temperature. 3. The vapor-pressures of both liquid phases are equal both as to total pressure and the pressure of each constituent. If in the system of ether and water, the water is continually removed, the pressure will retain its original constant value as long as both liquid phases are present. When the water phase disappears, the pressure of the water vapor begins to fall, finally becoming zero when pure ether is left. The point where the vapor pressure begins to fall, indicates the disappearance of the water phase and the liquid left at this stage will be a saturated solution of water in ether. At constant temperature, the composition of the liquid a t this point will be constant and it will change with a change in temperature. Similarly with a moist precipitate of pure ferric hydroxide, if we assume a solid solution, there should be a saturated solution of water in ferric oxide and of ferric oxide in water. As the ferric oxide is practically insoluble in water, its concentration is zero, so the mixture would consist of the solid solution of water in ferric oxide mechanically mixed with water. The vapor pressure of this mixture would thus be equal to the vapor pressure of pure water and when water is removed, the vapor pressure should remain constant as long as free water is present. The vapor pressure should begin t o fall a t the point where the free water disappears and the composition of the residue at this point should give the composition of the saturated solution of water in ferric oxide. At ’ Z. anorg. Chem , 13, 234. Ibzd 49, 1 3 j . I - ’ h y ~ t ,~ 24, l ~
’J
I jj
(~Rgg).
a constant temperature, different samples of ferric hydroxide slioultl show the same composition at tlie point Tvhere the vapor pressure begins t o fall provided they have reached eyuilihrium with the water and tliis composition should change with a change in temperature. To test these conclusions, two samples each of ferric hydroxide and aluminium hydroxide were prepared, using \.er!. different conditions for each of the tn-o samples. I n one case, i c ) o grams of crystallized aluminium chloride or ferric chloride were dissol\wl i l l about six liters of cold water, using just enough hydrochloric acid t o pre\.ent a basic salt from forming:. 'l'his cold dilute solution \vas precipitated ivitli aiiimonia and tllorouglily waslied \ ) y dec:intaticm. 111 t h e c~thercase, the samples were prepared 1)y tlissol\.ing ioo grams oi the same crystallized salts in 011e liter of hydrochloric acid. precipitating ~ v i t hammonia :ind ~vashiiigas before. 'l'lie reaction I)et\veeii the ammonia and ]I!.tlrochloric acid heatetl the solution iiearl!. t o the l~oilingIioint. ( hie sample oi each h!droxide, therefcire, \vas 1)wcipitatetf from a cold dilute solution containing but little aininoiiiuni chloride, and the other \\.as precipitated from a hot concentrated solution \\-it11u large excess of a n moniurn chloride. XI1 the saniples were waslied \.cry tlioroughly, lirst I)!. decantation and then on the filter until iio test for chloride \vas 01) tnined when a little of the precipitate \vas dissolved and tested hvith silver nitrate. l h e s e precipitates aiter being prepared tvere allowed to stand in a moist condition se\.ernl da!.s at the temperature of the experiments. I t was hoped that 11). this metliod tlie hydrates \ ~ o u l dreach a state of eq1iili1)rium with tlie ivater i l i cciiitnct Lvitli tliem. 'ilie coiiil)osition of the moist precipitate \vas deterniiiied I)!. ignition and :I siiinll weighed sample ( I - 2 grams) \vas placed in a weighed porcelain crucible and a l l o u d to dr!. partl!. i n t h e air a t t h e teniperaturc of the esperitneiit. X small pestle of glass or p r w l a i i i rod ivas iveiglied with the crucihle and tliis was used to mix the precipitate frecliientl!. to prevent iine\wi drying. At intervals, the crucihle HXS weighed. t h e n placed on a ruised triangle in a large \veigtiiiig bottle oii the bottviii of which was some pure water. The bottle \vas made tight with melted paraftin and sunk in :I thermostat for from eighteen to fort!.-eight hours. 'I he crucible \vas then remo\,ed and iveighecl ciqickly to determine the change in \\eight. A preliminary experiment s h o u w l t h a t a crucible containing piire \vatellost weight slightly when placed over \\-atc'r ;is clescribetl, altlio~igliit might be expected that tlie Lveight \ v o u l t l remain constaiit. 'I'hct crucible contained i .o5ho grams of ivater :inti lost o.orj7(1 gram i n a1)oLit t n e n t >.-f ou r hours . A11 the moist 1I rec i pi t at e s lie h a\. ed sii n il arl!. , a1wa!.s losing weight sliglitly ~ J \ W Imre Ivatcr, a n d this :ippeurs to 1)e tlic iiorninl behavior \ ~ h i l eniechanical Lvater is present mixed \\.it11 ti!~lrates. :liter tlr!.in:. in the air i o :I c.c,rt:iiii poiiit, h(J\\.('\x'I-. tlie I)recil)itatt.s I~cgaii
r391
NATURE OF PRECIPITATED COLLOIDS.
t o gain weight when placed over water. This was due to the fact t h a t their ,vapor pressure had fallen below t h a t of the water, showing t h a t the water phase had just disappeared, The composition at the point where gain in weight first occurred gave approximately the composition of the saturated hydrate free from mechanical water. It may be of interest t o state t h a t the pcint where the material first began to absorb water could be very closely determined by noting the mechanical condition of the precipitates, which appeared moist up to this point and dry be\.and. The method outlined above was used in every case b u t one-that of aluminium hydroxide at 25'. I n this case, a considerable quantity of the moist hydroxide was slowly dried and samples removed from time to time and tested for loss or gain over water as before. This method was not as convenient and was harder to control. The results obtained are given below. The composition at the point FERRICHYDROXIDE.
T. =
--
I.
,
L
Per ceiit Fe2U8 1st
Change
111 weight.
Series.
33 i I 35 95 37.98 45 2 5
25O.
--.
XI. 7--
Per cent. FelOs. Change i n weight 1st Series.
-0.0043
46 -91
+O.OOZj
23.21 31.26 37.13 40.00 44.94
49 04
+o.mjz
46 a 3 5
-0.0025
-0.0095 -0.0020
-0.0065 -0.0125
---0.0060 -0,004j Q.00Ij
+o.mro
50.18 +0.0035 2nd Series. -0.0040, 23.72 -0 .0060 32.9; 41.60 -0.0030
2nd Series. -0.0040 34.32 42.80 -0.0030 -0.0015 46.28 48 -53
+o.ooro +0.0050
46.09 49.23
-0.0020
50.70
54.74
f0.00go
56.43
+0.0061
+0.0025
T.= 45O. 1st
14.99 30.38 52.68
Series.
1 s t Series.
--O.OIIj -0.0230 -0.OOIO
54 -45
+0.0020
55.62
+O.OOjj
2nd Series. -0.0145 14.90 28.66 -0.0200 52.82 -0.0015 54 .bo
+0.0035
15.74 25.53 41.79 53.81 55.77 57 .I8
--0.0195
-.
0335
-0.oroj -0.0 0 5 0
--0.0033
+0.0040 2nd Series. 16.31 -.OX45 -0.0195 33.78 --o. 0 0 5 0 53.92 -+O.oooO 55.22
56 -30
+o.oor7
59.66
+0.&60
1392
H. W. FOOTE.
where gain in weight first occurred is printed in heavy type. The results under I were obtained with precipitates formed from cold dilute solutions, Under I1 are the results from hot concentrated solutions. Duplicates, called first and second series, were carried out in each case. In calculating the composition, the weight of the oxide was determined by ignition at the end of a series of determinations. The weight of the precipitate was found by subtracting the weight of the empty crucible from the weight of the crucible with the precipitate czj:er it had been over water in the thermostat. The approximate composition of the material a t any point could also be calculated as the composition of the original moist precipitate was known with some degree of accuracy from analysis of similar material. .4LCMINILM
T. =
-
I. ______A_____
Per ceiit. A l ~ u : ,
Change in weight.
HYDROXIDE. 25'. IT. ---A_____?
Per cent. AI2Os.
1st Series.
43.13 45.60 49.97 5 1 -42
1st
41.57 48 -47
-0.Oo;j
-0.0025 -0.0060 4-0 .o226
2nd Series. 32.47 42.26 46.18 48.25 50 -91
Change i n weight.
Series -0.00
10
+0.008,~
2nd Series. -o.oooc)
-0.0098
44.74 5 0 *57
--O.OOjj
52.70
to.04jo
Q.0055
+O.OIlS
--o .0040 +O.Oo35
T. 4j0. 1st
50.9'
52 a 7 0 53.01
Series.
1st Series. Q.OO&l
+O.Ooh
51.39
53 -86
*).ooI 3 +o.ootm
to.ojoo
2nd Series. 51.56 --0.005j 52.43 +0 , 0 0 8 5 53.39 +0.0095
2nd Series.
49.98 .jI
--O. 00,jj
. 116
53.27 53 -77 S 5 . (1 3
A0.
).I
+ +
oooj. 0035
0 ,001.j 0
I
ooj2
The method, it will be seen, can give only approximately the compo-sition of the residue when i t first begins to gain water, but the duplicate results indicate t h a t the results obtained are not largely in error and they appear t o be accurate enough for the purpose. The fact that the losses in weight are irregular in the table, is chiefly because the length of time that the samples remained ox'er water was not alike, the object being to determine whether there was a loss in weight and not how great the loss might be.
NATURE OF PRECIPITATED COLLOIDS.
I393
The summary below gives the average results obtained when the residues first absorbed water, or in other words, when the residues have approximately the composition of the saturated solution. FERRICHYDRATE. Temperature. 25O
45O
I.
Per cent. Fe,Oa.
47.72. 54.53
11.
Per cent. FenOl.
47.79 56.74
ALUMINIUMHYDRATE. Temperature.
I. Per cent. A12O3.
25O
51. I 7
49.52
45O
52.57
53.82
11. Per cent. AlzOs
Different samples, prepared under widely different conditions, have nearly the same composition a t the same temperature, indicating t h a t the moist precipitates have approximately reached equilibrium with their surroundings. The composition changes with the temperature. Certain objections may of course be made t o the statement t h a t ferric hydrate is to be regarded as a solution of water in ferric oxide, If the oxide is treated with water, the latter is not taken up t o form the hydroxide. Again if part of the dissolved water is removed from the hydroxide by evaporation, this water is never completely taken up again. Van Bemmelen’s results have shown this repeatedly and in my own experiments, I noticed t h a t after a hydroxide was slightly dried and then exposed to water vapor, it never absorbed water enough to become saturated. While I have no experimental evidence to offer in explanation of this behavior, it is in general true that those reactions in which a solid directly takes part are often very slow, and the action may be so slow t h a t it practically never takes place. It may also be true that the hydrates are not solutions of water in the oxide but a solution of water in some lower hydrate, a point which has been emphasized by Van Bemmelen in considering the hydrates a s absorption compounds. Another objection is in regard t o the rate a t which the moist precipitate dries in the air. It has been shown by Van Bemmelen t h a t when a moist hydrate is allowed to dry spontaneously, there is no sharp change in the rate of evaporation. Such a change would in general be expected at the point where the free water has disappeared and dissolved water begins t o go off. I believe the reason that no such change in rate has been observed is due t o the conditions of the experiment. I n order t h a t the rate of evaporation from the moist precipitate shall be uniform, the material must not be stirred or disturbed. As a result, the part of the hydrate in contact with the air would lose water rapidly and some of the dissolved water would be removed while the parts out of contact
I394
H . IV. FOOTS.
with the air were still moist. A curve representing the loss b!- evaporation with the time in this case would not show a sharp change in direction. Xr. If. S. Pine has carried out some experiments in this laboratory with zirconium hydroxide, using the same methods t h a t were used with ferric and aluminium hydroxides. Lack of time has prevented him from completing the experiments, but his results as far as obtained agree with the previous ones. .I solution containing ver!. pure zirconium chloride with a slight excess of hydrochloric acid. was divided into two parts. I was diluted to fi\-e liters with water anti precipitated u i t h ammonia. I1 was diluted with m e liter of 11)-drochloric acid and precipitated with concentrated ammonia, the reaction heating the solution nearly to boiling. The conditions are similar t o those adopted in preparing the other hydroxides. I’reliminary determinations had shown the approximate composition of the saturated zirconium h!.droxide. His results were a s follows: T.
_-___--Per
= 2 jo.
1.
cent. ZrO?.
11.
--__
-7
Change i n weight.
Per c e n t ZrO..
1st Series.
1st
30.24
1 3 . 0 0 2 2
32 30
31 .80
+o.oor2
32 -72
32.26
---O.OOIO
+ 0.oooi
--0.0020
+
0 . OOOP
2nd Series.
2nd Series.
32.85
Change in weight.
Series.
2;. 20
---0.002:
30.80
A 0.0010
3rd Series. 2 9 40 32 -54
*
0.0000
+ o.ooL+o
The averages of his results for the composition of the saturated solution of water in zirconia are 32.33 and 32.02 per cent. of zirconia for I and I1 respectively. As before, the composition of the saturated solutions was nearly t h e same in both cases on standing in contact with water, although the original precipitates were prepared under very different conditions. I believe the results given in this article are, with the exception of Hardy’s results, the first quantitative evidence of any case of a solution of a liquid in a solid. I propose t o continue this work in the future. S R E F F I E L D L A B O R A T O R Y . NEW H A V E N . cONN.9 June 18,1908.
