IOK IKTERCHASGES I N ALUMINUM OXYCHLORIDE HYDROSOLS* BY ARTHUR W. THOMAS A S D THOMAS H. WHITEHEAD
The behavior in solution of many complex inorganic salts (particularly those of cobalt, nickel, chromium, and aluminum) is such that the application of hydrolytic dissociation formulae does not successfully account for this behavior in most instances. Among the several hypotheses formulated to account for the experimental facts, the Werner theory has been successful and consistent in accounting for results obtained and for the prediction of pr-5able behavior of related salts in solution. The original classical postulates of Werner made in 1893 have since been extended by Pfeiffer, Bjerrum, and Stiasny. The explanation of hydrolysis in Werner terms was made by Pfeiffer shortly after Werner published his first papers. Bjerrum,‘ in 1907, found it impossible to interpret the results obtained by heating basic chromic sulfate solutions by hydrolytic dissociation formulae. In addition to this, he found that the molecular weight of these salts had increased in some cases to as high as 7 jo. He therefore postulated that perhaps the hydroxo groups in the complex ions were becoming more firmly bound to form larger complexes, such as,-
HzO_I
Bjerrum called this process "alation." It accounts for decreased activity of hydroxo groups toward neutral salts and sluggishness of attaining equilibrium. Stiasny,’ in order to account for decreased activity of basic chromic salts toward neutral salt solutions and decreased solubility of basic chromic salts when they were heated for long periods of time, further extended the idea of Bjerrum to include the conversion of hydroxo groups to oxygen bridges in the complex ion with formation of acid. This process is called “oxolation,”* Contribution from the Department of Chemistry, Columbia University, No. 637. Bierrum: Z. physik. Chem., 59, 336 (1907);“Studier over Basike Kromiforbindelser,” Kopenhagen (1908),through Stiasny and Grimm: Collegium 691, 505 (1927);Z. physik. Chem., 73, 724 (1910). 2 Stiasny and Grimm: Collegium, 691, 505 (1927).
ABTHUR W. THOMAS AND THOMAS H. WHITEHEAD
28
Stiasny3 also found that the addition of neutral sodium sulfate to basic chromic sulfate solutions decreased the hydrogen ion activity and analysis of the resulting salts showed that the anion of the added sodium sulfate had become a part of the basic chromium salt. He found also that different anions had different degrees of effectiveness in decreasing hydrogen ion activity but in each case the added anion was found in the complex chromium salt. He therefore postulated an equilibrium between basic chromium salt ion and anion, being different for each particular anion and shifted by increasing the concentration of anion with respect to concentration of complex basic ion. This process he called “anion penetration,” for example,HzO Hz0 ++ H 2 0 Cr C1 zC1-+KC2H302 [HzO HzO]
a
1
HzO ++ HzO H20 Cr C2H302zC1Hz0 fKCl [H20
An outstanding extension of these hypotheses is to be found in the vapers by Gustavson4 on the elucidation of the complex process of tanning leather by chrome liauors. The lbrge molecular aggregates obtained by Bjerrum with basic chromic salts led the authors to believe that perhaps particles of colloidal size might be built up in this manner, and, if so, they should act similarly to the basic chromic salts toward heating, aging, and neutral salt solutions. The object of the present investigation was to apply the above mentioned hypotheses to the explanation of the behavior of aluminum oxychloride hydrosols made in the hot (8oo-7o0C)toward ageing and neutral salts and to attempt to give some picture of these hydrosols in terms of these hypotheses.
Materials The C P aluminum chloride hexahydrate used yielded the following results upon analysis: A1 = 99.50 per cent of theoretical; C1 = 99.40per cent of theoretical; Fe = 0.01per cent; Sulfate = Less than 0.002 per cent. Ammonium hydroxide, C P. Sodium hydroxide, C P,-a saturated solution was prepared and let stand for one month. Aliquots were siphoned from this and diluted. The distilled water was always boiled just previous to use. A11 reagents after being prepared were kept in “NonSol” glass bottles. Pyrex ware was used in preparation of reagents. Stiasny and Szego: Collegium, 670,41 (1926). J. Am. Leather Chem. Assocn., 18, j68 (1923); 21, legium, 672, 153 (1926). 3
4
22,
j 3 (1926); 22, 68 (1927); Col-
ION INTERCHANGES I N ALUMINUM OXYCHLOFIDE HYDROSOLS
29
The potassium chloride, bromide, iodide, nitrate, and sulfate were twice recrystallized from boiling distilled water and dried at I IOOC for twelve hours. Potassium oxalate was dried a t IIOOC for three hours. It was tested for chlorides, sulfates, and heavy metals according to Murray5 and found to be satisfactory. Its oxalate content was tested by titration against a standard solution of potassium permanganate which had been standardized the day before against sodium oxalate from the Bureau of Standards. Calculated on the basis of the formula
[email protected]@ the oxalate content was 99.90 per cent of the theoretical. Ammonium acetate was tested according to Murray3 for chlorides, sulfates, and heavy metals. Kone was found, so the salt was simply dried over calcium chloride in a desiccator for two weeks. Preparation of Sols The general method of preparation was to dissolve I O grams of aluminum chloride hexahydrate in two liters of water a t zo°C. Alumina hydrate was then precipitated by addition of ammonium or sodium hydroxide a t 2ooC and the whole immediately heated to the desired temperature. (This must not exceed 80°C because it was found that above 8ooC the alumina hydrate changes from its greyish color to a white color and becomes very much less soluble.) The mixture was kept hot for two hours and then allowed to cool to 2 0 T . The alumina hydrate settled out. The supernatant liquor was siphoned off and the precipitate washed with hot distilled water (7oOC). After washing, the desired amount of distilled water was added and whole mechanically stirred while the minimum amount of hydrochloric acid necessary to peptize was added. (This amount was determined by preliminary preparation of similar sols.) The mixture was again heated (to same temperature as at first) for two hours, put in a “NonSol” bottle and allowed to stand for 24 hours, after which it was centrifuged for one hour a t 1500 r.p.m. (Rotating radius to middle of tube was 42 cm) The supernatant liquor was carefully siphoned off (if any sediment had formed) after standing for 2 4 hours, and again centrifuged at 1500 r.p. m. for one hour. When nothing was thrown out after one hour centrifuging at 1500 r.p.m., the sol was considered peptized. This method gave results which were reproducible and offered a means of making aluminum oxychloride hydrosols of a wide range of alumina hydrate concentration. Four sols were selected varying in concentration of alumina hydrate content, in the temperature a t which they were prepared, and in the precipitating reagent. These preparations were considered colloidal because they conformed to criteria which have been cited as being characteristic of the colloidal state; that is (a) They did not dialyze through nitrocellulose membranes either a t 2o°C or a t 80°C over a period of three days, although HCl in them did. standards and Tests for Reagent Chemicals,” 276
(1920).
30
ARTHUR W. THOMAS AND THOMAS H. WHITEHEAD
(b) They exhibited a Tyndall cone in the carbon arc slit ultra-microscope. (c) They appeared turbid but did not settle out under gravity over a period of six months or in a centrifuge a t 840 times gravity in one hour. (d) The adition of various salt solutions resulted in precipitation of the dispersed phase. Analyses of Sols Aluminum6, Total. Since there were no other interfering elements present except a slight trace of iron, aluminum could be precipitated as the hydrate with ammonium hydroxide, filtered off, and ignited to the oxide. To I O O cc of sol, I O cc of glacial acetic acid were added and the whole left on the steam plate until solution was effected, usually about two hours. This solution was then made faintly alkaline with ammonium hydroxide (to methyl red) and allowed to digest over night a t 2oOC. I t was filtered through Whatman No. 44 filter paper, washed twice with distilled water, and filter paper with precipitate transferred to a previously ignited quartz crucible, heated on an air bath till paper was completely charred, then ignited over a Meker flame for one hour, coolsd in desiccator and weighed to the nearest 0.j milligram. Aluminum, Ionic. No evidence of any aluminum ion was obtained by use of the aluminon test’ for which a sensitivity of IO-^ mol/l aluminum ion is claimed. Chlorine, Total.* One hundred cc of sol and I O cc of 18 h4 nitric acid with not more than I O per cent excess over calculated amount of 0.1M solution of silver nitrate solution were put on steam plate, covered with watch glass, and allowed to stand for I O hours in a dark room. After cooling, the silver chloride was filtered off by means of a porous bottom Gooch crucible and heated in an oven a t I I j°C for three hours. This method had previously been verified in this laboratory by Hamburger in the analysis of ferric oxybromide hydrosols.P Chlorzne,Ionic. Chloride ion was determined directly on the sol by taking IOO cc of sol, adding 2 5 cc of distilled water and 3 cc of 0.1 M potassium chromate. Tenth molar nitric acid was added dropwise till the orange color of bichromate ion was seen, then I gram of powdered calcium carbonate’O was added followed by titration with 0.1 M silver nitrate solution. Ammonia, Total. The standard Kjeldahl method for nitrogen as ammonia was used to determine total ammonia in the sols. Ammonium Ion. After solution of the sol has been effected, ammonium ion was determined by use of the Nessler reagent, as prescribed by the American Public Health Association method.” To do this, 2 cc of 0.5 M potassium oxalate solution were added to 50 cc of sol and it was allowed to stand over night and filtered through Whatman Hillebrand and Lundell: “Applied Inorganic Analysis,” 389 (1929). ‘Hammett and Sottery: J. Am. Chem. Soc., 47, 142(1925). 8 Fales: “Inorganic Quantitative Analysis,” 196 (1924). 9 Hamburner: Diasertstion. “A Studv of Ferric Oxvbromide Hydrosols,” Columbia University (1G26). l o Hillebrand and Lundell: Loc. cit., 590. l1 “Standard Methods of Water Analysis,” 16,17.
ION IXTERCHANGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
31
N o . 44 filter paper. The precipitate was washed carefully with ten portions of 5 cc each of distilled water, allowing the washings to go into the filtrate.
Aliquots of 5 cc each were pipetted out, diluted to I O O cc, z cc Xessler reagent added and the color compared with a standard solution by means of a Duboscq type colorimeter. Determinations were run in triplicate and results checked to 0.5 milli-equivalent. Measurement of Hydrogen Ion Activity Since it has been established’* that the quinhydrone electrode gives reliable results between pH 2 and 7.5, this electrode was chosen for use in this investigation. I t is interesting to note also that Pelling13found the quinhydrone electrode valid for measuring the pH values of aluminum sulfate solutions (o.001 M to 0.4 X I ) . Electrode T’essel This was a 1 2 5 cc glass, wide mouth, bottle fitted with a rubber stopper admitting electrode, salt bridge, and air stirrer. Platznum Electrodes. The electrodes were made of bright platinum wire sealed into a glass tube and the tube filled with redistilled mercury. Two were made and constantly checked against each other. I t was necessary often to leave them over night in 18 hl nitric acid to get consistent and reliable results. Salt Brzdge. The salt bridge was a glass U-tube with stop-cock in the center. The side arms were of equal length and plugged with cotton to prevent diffusion of sol into salt solution, which occurred unless this precaution was taken. The bridge was filled with saturated potassium chloride solution (at z 5 O C ) and the stop-cock kept closed. The stop-cock was well covered with graphite. Azr Stzrrer. This was simply a glass tube tapered off to small opening a t one end and connected through a calcium chloride tube to the source of compressed air by rubber tubing. Calomel Half-cell. Two saturated KC1 calomel cells were made according to Findlay.14 They were checked against each other constantly and against a standard cadmium cell a t intervals. The electrode vessels were immersed in a water thermostat at 2 5 O C * o . I O C and the potentiometric readings were taken to the nearest millivolt which was deemed sufficiently accurate for the purpose of this investigation, Measurement of the Sign of the Electric Charge The apparatus used was the same as that used by Sherman, Thomas, and Caldwellls for measuring the iso-electric point of malt amylase. A diagram with detailed description of parts is given in that paper. lZ LaMer and Parsons: J. Biol. Chem., 57, 613 (1923); Biilmann: Ann. chim., 15, 109 (1921). l 3 J. Chem. Met. Mining SOC.S. Africa, 26, 88 (1925). “Practical Physical Chemistry,” 200 (1928). lb J. Am. Chem. SOC.,46, 1711 (1924).
ARTHUR W. THOMAS AND THOMAS K. WHITEHEAD
32
The bottom of the U-tube was filled with sol, the stop-cocks closed and the side arms washed out with distilled water, drained thoroughly, and filled with 0.01 M KCl solution. The electrode vessels were then inserted, the leveling stop-cock opened and the level of the potassium chloride solution allowed to adjust itself; then the leveling stop-cock was closed. The current, I I O volts DC, was then turned on and both large stop-cocks opened simultaneously. The current was allowed to stay on for 24 hours, after which both large stop-cocks were simultaneously closed, the current cut off and the electrode vessels removed. The contents of each side arm were transferred to clean dry beakers, thoroughly stirred and 20 cc portion pipetted out of each and analysed for aluminum. The colloidal aluminum micelles migrated to the cathode in each case. Summary of the Properties of Sols The data concerning the four aluminum oxychloride hydrosols are given in Table I. All figures are in milli-equivalents per liter.
TABLE I I
2
Sol* Temp. of prepn.
A B C D
3 Total Al
4
5
6
Total C1
Clion
Total NHs
7 ",ion
8
pH
44 I .2 I .22 I .o I 4.51 80' 70 54.90 53.04 ... 48 4.50 70' 44 8.67 8.56 8.15 8 4.68 70" 62 13.40 13.30 ... 0 4.39 * In the cases of sols A, B, and C ammonium hydroxide was used to preci itate the hydrous alumina from the original aluminum chloride solution, while sodium [ydroxide waa employed in the case of the alumina formed for the preparation of sol D. 80"
None of the hydrosols gave any reaction for aluminum ion so that the amounts of aluminum given in column three of Table I were bound up in the complex ionic micelle, which as previously mentioned was cationic in each sol. It may be seen from columns four and five that nearly all of the chlorine present was in the form of chloride ion, the difference being the amount in the complex aluminum ionic micelle. The analytical methods used were equally precise, and the authors think the differences are significant and not the result of different methods of analysis. I n the case of ammonia, however, the slight differences shown in columns six and seven are perhaps due to difference in methods and the authors believe that all of the ammonia was present as ammonium ion. The results in column eight of the table show slight differences in acidity. This is to be expected because the pH depends upon so many factors, e . g . , temperature of preparation, age of the sol, aluminum salt concentration, amount of hydrochloric acid used in peptization of the sol, etc., and since no final state of equilibrium had been reached in any case, no definite relation between the pH and other values was expected.
ION INTERCHANGES I N ALUMINUM OXYCHLORIDE HYDROSOLS TABLE
33
11
Effect of Ageing at zo°C on pH of Sols Time
hour hours 7 days 14 days 30 days 4 j days
Sol A
Sol B
4.54 4 51
4.56 4.59
Sol c
Sol D
Solution AIC13* 4.44 4.42
I
z4
&39 4,4I 4.41
4.39 4.37
4.63 5.39 4 75 5.42 5.49 * .41Cls solution--g grams A1CI3.6HnOper liter solution.
The results listed in Table I1 show that in all cases the hydrogen ion activity of the sols was decreased upon aging at zoo(’. This was found to be the case also for chromium salt solutions which had been heated, but those which had not been heated increased in hydrogen ion activity.I6 The last column in Table I1 shows that the aluminum chloride solution which had not been heated increased in hydrogen ion activity upon ageing. Thomas and Baldwin” found similar results with chromium salt solutions a t ZOT. The decrease in hydrogen ion activity of the sols is therefore significant because all the factors involved would apparently lead to increased hydrogen ion activity; namely, loss of water due to slow evaporation, and slow hydrolysis of the aluminum salt.’* Ah explanation for this phenomenon is given later.
Effect of Heat on pH of the Sols Fifty cubic centimeters of each sol were put in “SonSol” bottleslg and left in an oven at 80°C for four days, when ~ e were y removed, cooled to 25OC, and the pH measured. There was no loss of volume of the sols during this treatment. The results are given in Table 111. TABLE I11 Sol
A B
c
D
Original pH 4 75 4 61 5 39 4 39
After
j
hours
No change T o change ?io change Ko change
After 4 days 4 65 4 15 j
22
4 09
These results are very significant. They show obviously that heating increases the hydrogen ion activity, but the greater significance is that upon cooling back to zs0C, the hydrogen ion activity did not decrease to its former Stiasny and Grimm: Collegium, 694,49 (1928). Am. Leather Chem. Assocn., 13, 192 (1918). 1aTian: Compt. rend., 172, 1179(1921). l9 Sols heated at 80” for four days in ordinary glass reagent hottles became alkaline to phenolphthalein. l6
17J.
ARTHUR W. THOMAS AND THOMAS H. WHITEHEAD
34
value as do salt solutions which undergo hydrolytic dissociation. Hydrolytic dissociation alone then will not account for these results. I t should also be noted that heating for only three hours had no measurable effect. This again indicates that hydrolysis is not all that is involved inasmuch as heat increases the hydrolytic dissociation of salts. This behavior of the sols will be explained later. Effect of Neutral Salts on pH of the Sols All salt solutions were made up just prior to use. Solutions were all a t I normal concentration. In the case of potassium oxalate which was alkaline, sufficient 0.5 molar oxalic acid was added to make the solution slightly acid. This did not, how-
FIQ.I
ever, alter the concentration of oxalate ion available for reaction, as the solution was maintained 0.5 molar with respect to oxalate ion. The pH of the salts solutions at 2 5 O C was: Salt PH
KC1
KBr
KI
5.44
6.71
7.00
KNO, 5.86
KnSOd NHlOOCCHI Oxalate 5.55 7 . IO 5 .a7
Technique. The electrode vessel was rinsed with the sol, then 50 cc added with pipette. Half an hour was allowed for saturation by the quinhydrone; readings on the potentiometer taken every five minutes until constant. Usually they became constant in fifteen minutes. Then the salt solution was delivered into the electrode vessel by burette in I cc portions (except acetate and oxalate) and potentiometer readings taken after each addition until constant. After I O cc had been added in this way, the size of portions was increased to 5 cc and added to a total of 50 cc. I n the case of acetate and oxalate solutions, the portions were one-tenth the size of the other salt solutions. All solutions except oxalate gave constant readings on the potentiometer within five minutes, so that the readings reported for oxalate are those
ION INTERCHANGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
35
taken exactly five minutes after its addition. The drift was probably due to some oxidation-reduction reaction in which oxalate ion is notoriously sensitive. Several facts are evident from the results shown in Table IV and plotted in Fig. I . I n the first place, it will be seen that all salt solutions increased the pH of sol A, but their magnitudes are different, oxalate and acetate having a ten times greater effect than the other salt solutions. This suggests a chemical reaction between the sol and salt solutions. This is further borne out by the fact that water alone had practically no effect on pH; the effect of the salt solutions then was not’ due to dilution SJO of the sol by addition of dilute salt ro socc SOL A AT z s ‘ c . solution. 400 2 3 4 5 6 7 8 9 / 0 1 The question might arise as to L O G O F J E Q U ~ V x 103 whether this effect is due to the action FIG.2 of salt on the hydrogen ion activity shown by Harned:O for solutions of hydrochloric acid. In the case above described, the effect is too great to be accounted for in such a manner. I t is significant that in no case did the pH go above 7 . A maximum was reached with oxalate, acetate, and sulfate after which the pH increased because the salt solutions had a pH of j . 2 2 , 7 . 0 2 , and j . ; j respectively. This maximum is thought to mark the end of the chemical reaction. That there was no salt effect on the quinhydrone which could be measured was shown by the fact that the pH of the salt solutions was independent of the volume measured. The authors believe that the pH changes obtained upon titration of the sol with the salt solutions are real and not due to “salt effects” upon the quinone-hydroquinone ratio. It should be noticed that the order of effectiveness of the anions is : Nitrate
< chloride < sulfate < acetate < oxalate.
In Fig. z the concentration of salt solution is plotted logarithmically and the slopes of the lines represent the effectiveness of each salt in decreasing the hydrogen ion concentration. It is obvious that all three salts are of very nearly the same effectiveness. For this reason, potassium chloride is taken as typical of the halogen salts and used as such in comparisons with other salts. The pH changes upon titration of the four sols and of aluminum chloride solution are shown in Table TI and in Fig. 3. This graph shows first of all that potassium chloride increased the pH of all four sols in a similar manner, but had greater effect upon the sols of low aluminum content (A and C) than upon the sols of high aluminum content 2o
J. Am. Chem. SOC., 44, 2729 (1922).
36
ARTHUR 1%'. THOMAS A S D THOMAS H . WHITEHEAD
(B and D). A similar trend will be noticed with other salts. It is of more than passing interest to point out that sol D was made with sodium hydroxide while B was made with ammonium hydroxide, yet they act alike, indicating that a decrease in hydrogen ion activity is not the result of ammino groups reacting with hydrogen ion. This is further confirmed by the fact that sol A contained only I milli-equivalent of ammonia. while sol C contained 26, yet they acted alike.
6.00
? P 100
4too 0
30
20
50
40
EQUIVS. X 10
FIG.3
TABLE IV 5 0 cc Sol A and
I Sormal Salt Solutions (Figures are pH values at 2 5°C)
KSOI
KC1
K,SO,
NHaAc
KCz0,
0
4.75
4.75
4.75
1 75
0.0
4.75
4.75
I
1.iT
5 "3
5.39
6 4i
0.1
4.99
2
4.78
5.28
5 .63
6 69
0 . 2
3
4.78
5.42
5.80
0
4
4.78
5
4.78
5.51 5.57
5 .87 5.95 6,17
6.73 6 j8 6 81
0 .j
5.75 6 .oo 6.19 6.25 6.32
6.87
I
6.44 ...
6.16
6.54 ...
5.83
6.j8 ...
5.67 5.63
..
6.60 ...
.o
6.60
5.59
Cc added H20only
IO
4.78 4.80
5.71
6.24
6.88
6.28
6.88
cc
'
3
0.4 .o
..
20
4.80
5.76 5.81
25
4.80
s .84
6.34
6.88
..
30
4.80
5.88
6.37
6.85
3.0
35
4.80
5.9'
6.81
..
40
4.81
5.93
6,39 6.41
10
45
4.81
5.93
6.42
6.i9 6.78
50
4.81
5.95
6
6,;s
5
I5
.+I
2
.o
5 .39 6.77 6,73 6.67 ...
... ... ...
ION INTERCHANGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
37
TABLE T Kormal Potassium Halide Solutions added to 50 cc Sol A (Figures given are pH values at 2 5 T ) Cc. added 0
I 2
3 4
5 IO
I5 20
25
30 35 40 45 50
KCl
KBr
KI
4.75 5.39 5.63 5.80 5.87 5.95 6.17 6.24 6.28 6.34 6.37 6.39 6.41 6.42 6.44
4.75 5.28 5.45 5.54 5.61 5.67 5.85 5.96 6 .ox 6.07 6.12 6.15 6.18 6.22 6.22
4.75 5.20
5.33 5.44 5.53 5.59 5.79 5.91 6 .oo 6.05 6.15 6.22 6.27 6.29 6.32
TABLE VI Normal Potassium Chloride added to 50 cc each Sol (Figures are pH units a t 2 5 T ) Sol A
Sol B
Sol c
Sol D
AlC13.6HzO*
0
4.75
4.81
5.10
4.39
I
5.39 5 .s6
4.36 4.36
3 4
5.80 5.87
4.83 4.85 4.87 4.88
4.51
2
5.39 5.63
4.58 4.59 4.62
4.37 4.39 4.41
5
5.95 6.1;
4.89 4.97
5.74 5 .85
4.64
4.43 4.47
6.24 6.28 6.34 6.37 6.39 6.41 6.42 6.44 per liter.
5.01
5.91 5.93 5.95 5.96 5.96 5.96 5.96 5.96
Cc added
IO
'5 20 25
30 35 40
45 50 5g
j .02
5 5.07 5.08 5.10 5 . I2 5.14
5.65 5.7'
4.72 4.78 4.81 4.82 4.85 4.85 4.88 4.89 4.91
4.52 4.54 4.56 4.58 4.60 4.61 4.62 4.62
ARTHUR W. THOMAS AND THOMAS H. WHITEHEAD
38
These results seem to conflict with results reported in the literature by Thomas and Baldwin" and Wilson and Kuan,*l but it should be noted that in their work, large quantities of solid salt were added (3 gram mols per IOO cc) while here the total quantity added was between 0.001 and 0.05 mols per I O O cc. The work of Wilson and Kuan was repeated by the authors, and i t was shown2zthat when dilute solutions are used a single effect is noted, but if solid salt is used, the factor of hydration of the ions of the salt enters and tends to mask the chemical effect. This is confirmed by the work of Harned2: who found that small quantities ( O . O I - O . O ~ M) of sodium chloride increased the dissociation constant of water a t 25'C, but in greater concentration than I molar with respect to the solution, it decreased the dissociation constant.
300
6.00
t
9
~--~oRMAL
100 I
4.00
Also;
SOL'UTIOIN
(,n&5) --
ADDED TO S O cc.EACH SOL AT 25°C.
t
I
0
/O
I
I
20
I
I
30
I
I
40
I
50
EQUIVS. x 103
FIG.4
It will be noted that aluminum chloride solution of the same aluminum content as sol A and C, while showing increased pH values did so to a far less extent than the sols. Fig. 4 shows that potassium sulfate solution also increased the pH of all sols and reference t o Fig. I will show that it does so more effectively than potassium chloride solution of same equivalent concentration. Here again i t should be noted that the sols A and C acted alike and so did sols B and D. The authors hope by this time to have shown that one must look for something else to explain the results other than dilution, salt effect on quinhydrone, and hydration of ions. The results with ammonium acetate and potassium oxalate will be even more convincing. The titration of the sols with ammonium acetate solution is given in Table VI11 and in Fig. 5. The outstanding fact in Fig. 5 is that the effect of acetate ion is ten times greater in increasing the pH of the sols than are nitrate, chloride, and sulfate ions (cf., Fig. I ) . It is also shown that the order of reaction was proportional to the aluminum content of the sols (cf., Figs. 3 and 4). J. Am. Leather Chem. Assocn., 25, I j (1930). Thomas and Whitehead: J. Am. Leather Chem. Assocn., 25, 23 J. Am. Chem. Soc., 47, 930 (1925). 21 2z
127
(1930)
lox
INTERCHAKGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
39
It is again emphasized that no maximum was shown by the curves for B and D ; that is, the curves approached pH 7 on the vertical axis with increasing salt concentration. But the curves for C and A are parallel to the horizontal axis from 40 to 50, indicating that no change in pH occurred with increase in salt concentration after 40 x 102milli-equivalents were addpd.
300
6.00
$, 500
4.00
20
/O
so
30
EQU/VS.X / 0 2
FIG.j
TABLE VI1 Xormal Potassium Sulfate Solution added to 50 cc each Sol a t 25'C (Figures are pH values a t
25OC)
Sol A
Sol B
Sol c
Sol D
0
4.75
5 ' 42
5.49
4.39
4. I4
I
6.47
5.66
6.75
5.73
...
2
6.69
5.75
6.80
5.84
...
3
6.73
5.79
6.81
5.89
4
6.78
5 ' 83
6.82
5.94
... ...
Cc added
AICla.6H10*
5
6.81
5.86
6.83
5.96
4.16
IO
6.87
5.98
6.8;
6.03
4.22
'5 25
6.88 6.88 6.88
6.03 6.07 6 . IO
6.82 6.81 6.80
6.08 6. IO 6 . IO
4.30 4.35 4.41
30
6.85
6.14
6.78
6.12
4.43
35
6.81
6.17
6.76
6.14
4.44
20
40
6.79
6.18
6.75
6.15
4.47
45
6.78
6.20
6.73
6.16
4.5'
50
6.75
6.22
6.71
6.16
4.53
*
IO g
per liter.
40
ARTHUR W. THOMAS AND THOMAS
H.
WHITEHEAD
Table IX and Fig. 6 reveal the effects of titration of the sols with oxalate. It will be recalled that the oxalate solution consisted of potassium oxalate slightly acidified with oxalic acid. It will be noticed a t once that oxalate ion is very active. The pH of the sols was immediately increased to about pH 7, and except for B, the pH then markedly decreased, gradually approaching the pH of the oxalate solution ( 5 . 0 7 ) . This suggests that a reaction took place with A, C, and D; was completed, and the solution became more acid again. In the case of sol B, the second effect was much less readily executed. This sol contained more aluminum micelle than the others.
FIG.6
TABLE VI11 Normal Ammonium Acetate added to 50 cc of each Sol a t 2 j°C (Figures are pH values at zs0C) Sol A
Sol B
Sol c
Sol D
4.75 5.75
4.59 5 . I8
5.39 6.01
4.39 5 . I8
6.00 6.19
5.53 5.66
6.25
5.51
0.3
6.38
5.69
0.4
6.25
5.78
6.47
5.81
0.5
6.32
5.85
6.54
5.89
.o
6.44
6.08
6.67
6.14
Cc added 0.0 0. I 0.2
I
2 .o
6.54
6.29
6.79
6.34
3.0
6.58
6.42
6.85
6.47
4.0
6.60
6.51
6.88
6.58
5.0
6.60
6.56
6.88
6.64
41
ION INTERCHASGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
TABLE IX Kormal Potassium Oxalate Solution added to j o cc each Sol (Figures are pH values at z j"C) Sol A
Sol B
Sol
c
Sol D
0.0
4.75
0. I
4.56 5.69
5.41 6.82
4.37 4.88
6.39 6.62
6.77 6.71
6.41 6. j6
6.78 6.82 6.91
6.59 6.54 6.00
6.64 6.67 6.jo
5.63 5.51
6.03 j.88
5.46 5 ' 42
5.76 j . 76
Cc added
1.0
4.99 5.39 6. 77 6,73 6.67 6.16
2.0
5.83
6.89
3.0
5.67 5.63 5.59
6.85 6.78 6.77
0 . 2
0.3 0.4 0.j
4.0
5.0
Resume of Results between Sols and Salt Solutions Figure I showed the order of the effectiveness of the salt solutions in increasing the pH of sol A to be: Kitrate < Chloride < Sulfate < Acetate < Oxalate. Figures 3, 4, j , and 6, showed all sols acted similarly to sol A, so we can say that the same order would apply to the other sols. Figure 2 showed the similarity of the halogens, so that it may be said that the order of anion effect on pH of positively charged aluminum oxychloride hydrosols is : Kitrate < Halides < Sulfate < Acetate < Oxalate. This order was found for aqueous solutions of chromic salts by GustavsonI4 and by S t i a ~ n yfor ' ~ solutions of chromic salts that had previously been heated. All results showed that sols B and D acted similarly despite the fact that ammonium ion was present in B while absent in D as a result of the initial preparation of the hydrous alumina. Neutral Salt Titration of Aluminum Chloride Solutions of Different Basicities After the titrations of the sols with the salt solutions had been completed it was considered advisable to include in this paper a few similar measurements made upon crystalloidal aluminum chloride solutions. Selecting sol D for comparison, a solution of aluminum chloride hexahydrate ( j gm A1C1,.6H60 per liter) was made containing the same amount of aluminum as sol D. Then a potentiometric titration of the aluminum chloride solution 24
Stiasny and Balanyi: Collegium, 694, ;z
(1928)
ARTHUR
42
w. THOMAS
AND THOMAS
n. WHITEHEAD
with 0.1 N NaOH was carried out a t 2s0C,using the quinhydrone electrode, to determine the exact stoichiometric relation, this was found to agree with the calculated amount, Le., 30.50 cc of 0.1 N Il'aOH for j o cc of aluminum chloride solution. To make the basic solutions, 2 5 cc portions of aluminum chloride solution (containing I O grams A1Cl3.6H20per liter) were mixed with 10.17 cc and 20.34 cc respectively of 0.1 N NaOH mentioned above and the
I
1
1
FIQ.7
final volume made up to 50 cc with previously boiled and cooled distilled water. Thus 50 cc portions of AlCl,, AlOHCl,, and Al(OH)&l, were obtained all containing the same aluminum content. Each of these portions was then titrated with the same I K potassium sulfate solution a t 2 j"C according to the technique previously described,
TABLE X Normal Potassium Sulfate Solution added to 50 cc Portions of Each Aluminum Salt Solution a t 2 5°C. *(Figures are differences in pH units; 25'C) KzSOi
AlCl,
cc
pH
1.0
3.0 4.0
3.77 3.77 3.79 3.81
0.04
5.0
3.83
0.06
10.0
3.91 4.04 4.12 4.17 4.19
2.0
20.0
30.0 40.0
50.0
ApH* 0.00
0.02
0.13 0.2 I
0.25
0.29
.410HC12 pH ApH*
3.88 3.90 3.93 3.95 3.97 4.06 4.19 4.26 4.30 4.39
0.02 0.0 j
0.07
0.09
0.13 0.20
0.24 0.28
Al(0H)LI pH ApH*
4.14 4.16 4.19 4.22 4.25 4.34 4.45 4.52 4.56 4.60
Sol D pH ApH*
0.11
5.73 5.84 5.89 5.94 5.96
0.11
6.03 6.10
0.07
0.18
6.12
0.09
0.22
6.15
0.12
0.26
6.16
0.13
0.02
0.05 0.08
0.11
0.16 0.21
0.23
ION INTERCHANGES I N ALUMINUM OXYCHLORIDE HYDROSOLS
43
The potentiometer readings were made until constant over a fifteen minute period to plus or minus 0.5 millivolt but are recorded only to the nearest millivolt in Table X. The results are plotted in Fig. 7. It is evident that potassium sulfate decreased the hydrogen ion activity of all the aluminum solutions and of the sol. It is also seen that the rate of change of pH with respect to change in concentration of K 8 O 4 is not the same in any two cases but is greatest for sol D and least for AlCl, with the addition of K?S04solution up to and including 5 cc. The addition of from I O to 50 cc of K2S04solution caused a reversal of this order but again there is a progression from sol D through the Al(OH),Cl and the AlOHCl? to the X1C13 state. This suggests then that the rate of change of pH with respect to change in concentration of potassium sulfate is a function of the hydroxo groups contained in the basic complex aluminum ionsince practically every other condition was held constant. (It should be remarked that the sol was made up originally a t 70°C, and cooled to z ~ O C ,while the aluminum salt solutions were made and maintained a t z ~ O C throughout. , This was the factor that was not constant in the comparison between sol and solutions and the numerical difference in change of pH with increased concentration of potassium sulfate, although one of degree only, suggests that a closer analogy would have been obtained if this factor, too, were constant.)
Discussion of Results obtained I n attempting to account for the results reported, the same difficulty is encountered that Werner, Pfeiffer, Bjerrum, Gustavson, and Stiasny encountered in working with other complex basic salts; namely, the difficulty of applying the laws of hydrolytic dissociation and the various hypotheses of salt effects. Hydrolytic equilibria as obtained with simple crystalloidal solutions, such as of potassium cyanide, are not shown by aluminum oxychloride hydrosols. Table I11 showed that the equilibrium was not immediately disturbed by heat, but on prolonged heating it was disturbed. It likewise showed that upon cooling back, the equilibrium was not shifted back to the original stage. Tianlg has suggested that a system like aluminum oxychloride hydrosol may be considered as a heterogeneous system in which the basic salt ion is the dispersed phase. This would explain why upon ageing a t zo°C, the hydrogen ion activity increases because the assumption of a dispersed phase introduces a surface phenomenon which would, of course, decrease the original surface exposed by the ions and cause the equilibrium to be disturbed, for example :
+
+
AIC13 HOH e AlOHCl, HC1 I n this case, AlOHCl? is the supposed dispersed phase, whose formation decreases the surface, and causes a shift in equilibrium to the right; and this results also in hydrogen ion being increased. But if the temperature of such a system should be increased, the degree of dispersity would increase,?; this 15
March: Ann. Physik, 84, 605 (1927).
44
.4RTHUR W. THOMAS AND THOMAS H. WHITEHEAD
would increase the surface and so should shift the equilibrium to the left; L e . , decrease the hydrogen ion activity. Table IV shows that the opposite is true. This was also found to be the case for ferric saltsZ5and by StiasnyZ7 for chromic salts. To explain the effect of ageing and heat, then, one must look to some other source. The Werner-Co-ordination Theory2* offers a consistent explanation for these results. The aluminum chloride hexahydrate used precipitated three mols of silver chloride per mol of aluminum salt with silver nitrate. Therefore the followng structure can be assigned :H:O H2O +++ HzO A1 Hz0 C1-3 [HzO H2Ol Its hydrolysis according to the CrC1,6H20 analogue is :-
[:
H20 A1
[::: rJ+
H2O +++ H2O C1-8 r=i HzO A1 H z J
H2O
CI-2
+ HC1
(I)
To account for lack of mobility in the shift of equilibrium, we can picture the aluminum oxychloride sol as resembling the poly01 basic chromic salts reported by Bjerrum.’ Bjerrum working with chromic sulfate, found that if he heated his salt solutions to boiling that their molecular weights as determined by freezing point method had values around 7 jo. Stiasnyz7working with the same salt, found that heat decreased pH and that on cooling the pH did not return to its original value. Substituting A1 for the Cr of Bjerrum’s picture, we have: r z O HzO A1 H2 0
OH] H20 H20
zC1-
+
[HO HzOT H 2 0 A1 H20 + z C 1 - a HzO H2 0
In the above picture it is postulated that the hydroxo groups become more firmly bound which diminishes their tendency to unit with H+ ion to form aquo groups. Thus the tendency for reversal of the hydrolysis equation by heating is lessened. The authors present the point of view that the colloidal aggregates are formed according to the mechanisms of equations ( I ) and ( 2 ) . That is, heating caused two things to take place: ?‘Thomas and Frieden: J. Am. Chem. SOC.,45, 22 (1923) ?’ Stiasny and Grimm: Collegium, 694, 53 (1928). Thomas: “Inorganic Complex Salts” (1924).
I O 9 ISTERCHASGES IX ALUBIINUM OXYCHLORIDE HYDROSOLS
15
I. Aquo groups gave up hydrogen ions, leaving hydroxo groups in place of aquo groups. 2. The hydroxo groups then “olated,” resulting in the formation of large aggregates, eventually reaching colloidal size. The Werner Theory as extended by Bjerrum and by Stiasny, therefore, offers a consistent explanation for the formation of the sols, for their decrease in acidity upon ageing and increase on heating, for their sluggishness in readjustment of equilibrium, and for effect of heat upon solubility of aluminum hydrates. The explanation of the results obtaincd with salt solutions will now be attempted. It is probable that the more aluminum atoms present in the original aluminum chloride solution, the more olated hydroxo groups there were in the sol produced. (Washing removed the hydrochloric acid of hydrolysis.) It would be expected, then, that the greater the degree of olation, the more sluggish would be the reaction of the complex olated ion. Since in every case this complex ion was positively charged, it suggests that not more than two hydroxo groups were present per atom of aluminum. Kow when a salt such as potassium sulfate was added to a sol, the hydrogen ion activity decreased, suggesting that sulfate ion was replacing hydroxo groups in the Werner complex. The hydroxo uniting with hydrogen ion to form water would decrease the hydrogen ion activity. Stiasny3 analyzed basic chromic salt complexes before and after the addition of sulfate solution, and found that sulfate did go to the chromic complex. This is obviously the case with oxalate ion, because the addition of much oxalate changes the chromium from the cation to the anion.3This offers a simple explanation for the results shown in Figs. I to 4, inclusive. The anion order, then represents the “penetration power” of the several anions toward the complex basic aluminum ion. On this basis, oxalato complexes are least dissociated, and nitrato complexes most dissociated. Gustavson4found this to be true in chrome tanning. The anion penetration order explains why Lamb2$got complete precipitation of chloride ions from green CrC13with silver acetate, but not with silver nitrate. Some chloride was in the complex with chromium. ’Kitrate is below chloride in the penetration order, but acetate is above it and the latter then possesses the property of replacing chloride in the “nucleus” forcing it into the “outer solution.” The agreement among the basic aluminum salt solutions (Fig. 7 ) indicated that their behavior is a function of the hydroxo groups present in the basic aluminum ion and that an aluminum oxychloride sol is averylarge basic aluminum ion carrying a positive charge and in equilibrium with chloride ion. The difference in behavior with respect to small amounts of K2S04,and to larger amounts suggests that hydrolytic dissociation in addition to anion penetration took place a t first and when no further hydrolytic dissociation was possible, only the factor of anion penetration was involved. The curves 99
J. Am. Chem. SOC.,28, 1 7 1 0 (1906)
46
AFTHUR W. THOMAS AND THOMAS H. WHITEHEAD
in the first half of the figure are in contrast to the almost perfect straight lines in the second half of the figure and suggest a net effect of two unequal factors while the straight lines suggest only one factor producing a linear relation. Using potassium chloride as a typical example to show the anion effect we have :OH ff (3 1
HO HO +KOH Since there are many hydroxo groups in a large poly-ol complex ion, the replacement of hydroxo groups will be a function of the concentration of anion added until the maximum penetration for each anion has occurred. This accounts for the maxima in several of salt titration cruves. (Fig. 4, for example.) On the basis of the evidence submitted the authors present Fig. 8 as representing qualitatively their concept of an aluminum oxychloride hydrosol. If in Fig. 8 Cr is substituted for AI it becomes the Bjerrum formulation of a poly-ol chromium complex. As stated, this figure is a mere qualitative picture of a simple 01 compound since the authors do not know how many aluminum atoms are contained in the colloidal ionic micelle. It is possible. for such an ionic micelle to contain “oxo” groups as well as hydroxo groups. The idea of “oxo-lation” originated by Stiasny is shown in equation ( 4 ) ) which illustrates the conversion of one mole of hypothetical octa-aquo-diol-dialumini tetrachloride to the final product of one mol of the dioxo compound plus two mols of HCl.
HO
r
0
l+
With the conversion of hydroxo groups to 01 groups to oxo groups there results increasing resistance to the action of acids. It may be that one of the differences between aluminum oxide hydrosols made by peptization of freshly
10s ISTERCHAPI’GES I N ALUMIST;;\I OXYCHLORIDE HYDROSOLS
47
precipitated hydrous alumina and made by the (!rum3] method of boiling a solution of aluminum acetate is that the latter is more oxolated than the former. The authors hazard the guess that the oxolation of 01 complexes would result in a loss of the reaction with neutral salts described in this paper. I t is seen that oxolation results in a diminished charge on the ionic micelle. The Crum aluminum oxide sols are less stable than those made by peptization by acids of freshly precipitated hydrous alumina.
FIG 6
The ideas submitted by the authors concerning the structure of aluminum oxychloride sols, olation and oxolation were suggested by identical postulates for basic chromium salts made by Bjerrum and by Stiasny and his co-workers. The postulate that the alkalinity resulting upon the addition of neutral salts is due to replacement of hydroxo (or 01) groups is new in this paper. The Werner theory with the extensions by Bjerrum and Stiasny so consistently account for the results obtained that the writers believe its application to other similar colloids will be fertile.
Summary Ageing of aluminum oxychloride sols, prepared as described in this paper, a t zo°C, resulted in decreased hydrogen ion activity. 2. Heating of such sols resulted in increased hydrogen ion activity and subsequent cooling did not immediately decrease it. 3. The addition of neutral salts to the sols produced a decrease in the hydrogen ion activity. 4. An anion order was established, namely:-oxalate > acetate > sulfate > halides > nitrate which is identical to the “anion penetration” order of Stiasny . 5 . Explanations for these results and a suggestion for the constitution of such sols have been offered on the basis of the Werner theory and its extension by Bjerrum and Stiasny. I.
Department of Chemistry, Columbia Uniuersity. N e w Y o r k , A‘. Y .
30Ann.Chirn. Phys., (3) 41, 1 8 j (18jq);lAnn. Chem., 89, 156 (1854)