Solubility of Ferrous Iron in Aqueous Ammoniacal SoIutions Donald J. Kiockel and A. Norman Hixson2 School of Chemical Engineering, University of Pennsylvania, Philadelphia, Pa. 19104
The solubility of iron was measured in the ferrous state of oxidation in aqueous solutions containing various amounts of ammonia and either chloride, sulfate, or carbon dioxide. Although there i s no detectable iron NH40H HCI or H&04 or HzC03per solubility in these solutions at low concentrations [grams Fe(0H)z liter of solultion], iron became soluble in increasingly larger amounts at concentrations of 300, 400, and 500 g/l. This increase in solubility appeared to be a linear function of the solute concentration. The solubility of iron (as Fe) was relatively unaffected b y changing acid anions; however, the presence of such an anion was necessary for solubility. Employing ion-exchange resins showed the soluble species to be a cationic iron-ammonia complex. Material balance information indicated that the hexammine complex i s predominant at the point of maximum iron solubility.
+
N i c k e l , cobalt, copper, and other transition metals form soluble complexes in ammoniacal solutions. Commercial processes have been developed for the recovery of these metals from ores with an extraction step employing ammonia-ammonium salt solutioins as the leaching agents. The metals dissolve in aqueous solutions as complex ammines and are separated from the g,mgue constituents. I n experimental work on such a process, surprisingly substantial amounts of iron in the ferrous state of oxidation were soluble in high-strength ammoniacal solutions. I n dilute solutions the iron wa's insoluble as expected. However, as the amounts of ammonia and ammonium salt were increased, the solubility of iron increased markedly. This work was undertaken to determine accurately the extent of iron solubility in such solution and to investigate the nature of the soluble iron species. The systems selected were aqueous solutions of ammonia with t h e anion being chloride, sulfate, or carbonate. Representation of Phase Equilibria
The method of representing the phase equilibria was adopted from Dean (1952a,b) who studied similar systems containing the manganous ion. This involved the arbitrary assumption that the systems contained the following compounds: Fe(OH)2, b[H,OH, H20, and HCl or HzS04 or H2C03 (depending on the anion present). Such fourcomponent systems can be represented graphically b y a tetrahedron with each pure component at a n apex. However, by applying the restriction t h a t "the total solute concentration is a constant ') the solubility data can be shown on two-dimensional ternary diagrams. As an example, in the chloride system fQr any one diagram, the sum of the weights of Fe(OH)Z, HCl, and "40H per liter of solution is maintained constant. The complete representation of all possible total solute concentrations for a system can be accoinplished b y using an equilateral triangular prism with the pure solutes [Fe(OH)2,HCl, "*OH] 1 Present address, Mobile Research & Development Corp., Paulsboro, N.J. 08066. * To whom correspondence should be addressed.
+
at the prism apices and the total solute concentration varying along the vertical axis. Neither the assumption of the system components nor t h e concentration restriction affects the species in solution in a n y way. However, this restriction has the disadvantage t h a t it is true only for the portion of the phase diagram where there is complete solubility. The formation of a solid phase violates the restriction of constant total solute concentration. Thus, equilibrium mixtures of solid and solution are not accurately represented b y these ternary diagrams; however, these regions have been included to illustrate the phases present. Experimental Procedure
Basically, two methods were employed t o determine the limits of solubility in the three systems. The first involved visual observation of the appearance or disappearance of a solid phase when solutions with the same total solute concentration were mixed. The second method required chemical analysis of a saturated solution for iron, acid anion, and ammonia (Fischer, 1961; Kolthoff and Sandell, 1945). With the chloride or sulfate solutions, either method could be used. However, only the analytical method was applicable to the carbonate system. Because of the ready availability of pure NH4Cl, (h"4)2SO4, FeC12, and FeS04, the same technique was employed for both systems. Using the chloride system as the example, we prepared stock solutions of known total solute content and ammonia to chloride ion ratio by dissolving NH4C1 in water with subsequent addition of either HC1 or ",OH. Similarly, solutions of known total solute content were prepared by dissolving FeClz with a subsequent HCl addition if a different iron to chloride ion ratio were desired. The total concentrations selected were 10, 300, 400, and 500 g/l. of solution. B y use of solutions of equal total concentrations, t h e phase equilibria were determined b y titrating solutions of various ammonia to chloride ion ratios with a ferrous chloride solution until a solid phase formed or dissolved. Conversely, solutions of various iron t o chloride ion ratios were titrated with ammonium hydroxide until a solid phase appeared. Experiment showed that the volume resulting from mixing these solutions was equal t o the sum of Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1, 1972
141
5
200 300 400 500 TOTAL SOLUTE CONCENTRATION - GM/L Figure 1. Solubility of iron
n
Fe(OH)2
A
HCI
Figure 3. Fe(OH)2-HCI-NH40H-H20 NH40H CONCENTRATION
/H( Figure 2. Diagram of apparatus
the volumes used. Consequently, the total solute concentration remained constant in the absence of a solid phase. The accuracy of the technique for visual observation of the phase equilibria was verified when solutions in equilibrium with a minute amount of solid phase were analyzed for iron, chloride ion, and ammonia. Since the amount of solid phase was small, the total solute concentration remained essentially constant. The two techniques showed excellent agreement. Because the solubility of FeSOa was exceeded a t 400 and 500 g/l., a suspension of a known total amount of the salt in water was titrated with solutions of various ammonia to sulfate ion ratios until the solid phase dissolved. The resulting solution volume was checked t o verify the concentration. A problem caused by the insolubility of ammonium chloride in acid solutions a t these same high concentrations was handled similarly. The determination of equilibria in the Fe(OH)2-H2C03TU"40H-H20 system required a different technique. FeC03 is virtually insoluble in water and unavailable in reagent grade. High-strength H2C03solutions cannot be prepared a t atmospheric pressure because of the vapor pressure of COZ. Ammonium bicarbonate and carbonate were used in preparing the specified ammonia-carbonate ion solutions. The carbonate is a mixture of carbonate and carbamate which necessitated a special analysis to determine the NH3/C02 ratio. The iron solubility was determined b y analyzing solutions of various NH3/COz ratios saturated with iron originally in 142 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1, 1972
Fe(OHl2
A
HCI
Figure 4. Fe(OH)2-HCI-NH40H-H20
the form of Fe(OH)2which was prepared as an aqueous slurry by precipitation from an FeClz solution with NaOH and washing to remove the chloride ion. The slurry was settled, and supernatant water decanted as much as possible. A series of additions of constant H2C03-NH40H ratio solutions was made to the slurry. After each addition, the system was agitated for 30 min, and an aliquot of the saturated solution analyzed for iron, ammonia, and carbonate ion. Thus, the iron solubility was measured with increasing total solute concentration for a particular T\"40H-HZCO3 ratio. Eleven different ratios were measured. The linear plots of these data (Figure 1 is an example) were used to select values a t 300, 400, and 500 g/l. for oomparison with the other systems. The values a t 300 and 400 g/l. were interpolated. Because of the difficulty in obtaining a dense Fe(OH)2 slurry (low water content), the measured total solute concentration exceeded 500 g/l. only once. Most of the solubilities at this concentration represent a n extrapolation from about 460-70 g/l. Description of Apparatus
The mixing processes described were carried out under a nitrogen atmosphere since iron in the ferrous state of oxidation readily oxidizes to the ferric state in these basic solutions. The apparatus employed is shown in Figure 2. Fluted cylinder B with a volume of 180 ml was used as the mixing vessel. Burette C was employed to add small amounts
",OH .t
Table 1. Maximum Iron Content
CONCENTRATION
Fe (OH) 2-HCI-SH40H-H20 Concn,
s/l.
300 400 500 300 400 500
Fe(OH12 Figure
Fe(OH12
A
HCI
300 400 500
Maximum iron content Fe(OH)e, Concn Fe, Wt % g/l.
95% Confidence limit Fe(OHl2, Concn Fe, Wt % g/l.
2.2 1.7 3.5
27.6 1.2 14.8 17.5 44.0 0.7 59.0 1.2 19.0 Fe (OH)2-H,S04-r\"40H-H20 16.2 0.9 30.2 18.2 1.0 43.4 21.3 66.1 1.3 Fe(OH)z-HzCOa- S H I O H - H ~ O 14.7 27.4 2.2 17.6 43.7 1.4 20.7 64.2 1.9
1.7 2.4 4.0 4.0 3.4 58
5. Fe(OH)2-HCI-NH40H-H20
A
H2SO4
Figure 6. Fe(OH)2-H2S04-NH40H-H20
of the titrating solution. Agitation was accomplished b y a magnetic stirrer H. Coiist'ant temperature was niaintained by water bath -4 and heater G. Temperatures m-ere measured by thermometers D aiid E. Flow of the nitrogen inert atmosphere was controlled by valve F. Results
The phase equilibria of the chloride, sulfate, and carbonate systems are plotted in Figures 3-5, 6-8, and 9-11, respectively. The coinplet'e set of experimental data can be found in a dissertation by Klocke (1970). Copies may be obtained from Irniversity Microfilms, Ann Arbor, Michigan. In these figures, Area I is a single liquid-phase region where iron is soluble in a basic aqueous solution containing ammonia and the approlriate ammonium salt. Area I1 is also a single liquid-phase region with the solutio11 containing a n ammonium and ferrous salt and the free acid of the particular system. The region marked I11 is a t'wo-phase region; the solid phase is ferrous hydroxide. Points in this area cannot be represented accuratel:; by these diagrams since the presence of a solid phase violates the restriction of constant total solute concentration. The additional regions shown will be identified for each system. Fe(OH)z-HCl-NH4OH--H20, T h e phase equilibria of this system (Figures 3-5) were determined at total solute concentrations of 10, ,300, 400, and 500 g/l. A and l3 are the equivalent represent'ation of the ferrous chloride and am-
A
H2S04
Figure 7. Fe(OH)2-H2S04-NH40H-H20
monium chloride points, respectively. A line connecting the two point,s divides the diagrani into acidic aiid basic sections. Area IV is a two-phase region with ammonium chloride being the solid phase a t coiicent'rations of 400 and 500 g/l. Area V , which appears only a t 500 g/L, contains a solution and two solid phases, Fe(OH)2 and SH4CI. Also a t this concentration, the maximum concentration of HC1 is 92% owing to the vapor pressure of HC1. The main interest centers on AAreaI \)-here iron is soluble in strongly basic ammoniacal solutions containing the chloride anion. Although t'liere is no detectable iron solubility in these solutions a t a concentration of 10 g/l., iron becomes soluble in increasingly large amount+ a t concentrations of 300, 400, and 500 g/l. as indicated by the increasing size of Area I. The maximum solubility of iron slio~vnon the diagram is 19.0%, the percent of Fe(0H)Z in a soluble mixture containing the equivalent of 500 g/l. of Fe(OH)?, NH40H, and HC1. This amounts t'o a solubility of 59 grams Fe per liter. Table I s h o m the maximum iron solubilities for the various total solute concentratioiis of the three systems studied. T o attain concentrations higher than 500 g/l. is difficult since the limit of ammonia solubility a t atmospheric pressure is approached combined with the increasing size of the ammonium chloride two-phase region. Fe(OH)2-H2S04-NH40H-H20.The phase equilibria of the sulfate syst'em also determined at total solute coiiceiiInd. Eng. Chem. Process Des. Develop., Vol. 11, No. 1 , 1972
143
A
H2S04
Figure 8. Fe(OH)2-H&04-NH40H-HZO NH40H
NH40H CONCENTRATION
CONCENTRATIQN 300 G R A ~ L I T E R
500 GRAM/LITER
A
FeCOHl2
Figure
H2C03
9. Fe(OH)2-H&08-NH40H-H20
trations of 10, 300, 400, and 500 g/l. are shown in Figures 6-8. Again A and B are the equivalent representation of the ferrous sulfate and ammonium sulfate points. Area I V shows a ferrous sulfate insolubility occurring at 400 and 500 g/L, whereas V is the ferrous ammonium sulfate insolubility present a t 500 g/l. Iron is not soluble in aqueous ammoniacal solutions containing the sulfate ion a t a total solute concentration of 10 g/l. However, similar to the chloride system, in solutions with high solute contents there is a large solubility of iron which in this case reaches 67 g/l. (as Fe). Fe(OH)rH~C03-NH40H-H~0. The phase equilibria of this system a t total solute concentrations of 300, 400, and 500 g/l. are presented in Figures 9-11. A, C, and D represent the equivalent ferrous hydroxide, ammonium bicarbonate, and ferrous carbonate points, respectively. Point B, the limit of solubility of an ammonium hydroxide-ammonium bicarbonate solution, varies with the total solute concentration. The region marked I1 could not be prepared owing to the limit of solubility of HzC03. ilrea I V was not investigated because of the insolubility of ferrous carbonate and ammonium bicarbonate. Consequently, region I is the only single liquid-phase portion of the diagram. Again, iron was soluble in this system with a top value of 59 grams F e per liter. 144 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1, 1972
FcCOH12
D
H2C03
Figure 1 1 . Fe(OH)2-H2C03-NHrOH-Hz0
Species in Solution
The solutions of interest are found in Area I of the previous figures where a soluble iron apecies exists in aqueous solutions of ammonium hydroxide and an ammonium salt of the particular system. T o eliminate the possibility that the iron being measured as soluble in these solutions was actually in a colloidal form, a diffusion experiment was performed with a semipermeable membrane. The iron-bearing species was placed on one side, and an iron-free ammonium hydroxide-ammonium salt solution on the other. A migration of the iron through the membrane was detected almost immediately. Since a colloid could not exhibit this behavior, the iron was in a true solution. At t h e maximum iron concentration, the p H values were 9.3-9.7. At such concentrations of hydroxyl ion, the maximum solubility of iron as Fe(OH), would be of the order of 10+ g/l., a n insignificant contribution to the measured solubility. The soluble iron species would appear to be a complex ion. The nature of the charge on the complex ion was determined by selective ion-exchange resins. Successive 3.0-ml volumes of a solution containing iron in the complex form were passed through a bed containing either a cation-exchange resin, Amberlite IR 120 originally in the ammonium form, or an anion resin, Amberlite IRA 400 in the chloride form. The resin beds were initially filled with an ammonium
p'o
"I
,d
9
0.21
{
0.0 0
1
ANIONIC RESIN
1
-VOID VOLUME
d
i ,y.p 10
,
,
,
,
,
,
,
,
20 30 40 50 60 VOLUME REMOVED - M L
1
,
1 70
Figure 12. Iron removal by ion exchange
FcCOHl2
HCI
Figure 13. Fe(OH)2-HCI-NH40H-H20
hydroxide-ammonium s>altsolution to keep the soluble iron in solution. Figure 12 is a plot of the effluent concentration vs. the volume of iron solution removed from the column as effluent for the two resin types. No iron was detected in the effluent from the cation exchanger until a volume equal to twice the column void had been removed. With the anion exchanger, iron appeared before an amount equal t o one void volume had been displaced. As seen from tlhe curves, the ratio of effluent concentration t o original concentration C/Co approached unity much more quickly with the anion exchange resin. With resin beds of approximately equal weight, the cation exchange material removed 87y0 of the iron from 30 ml of solution whereas the anion column removed only 6% from a similar volume. The iron appears t o be in the cation, and since i t cannot be a simple ferrous ion, a complex cation, probably with ammonia, must exist. Several spectrometric techniques were employed to determine experimentally the nature of the cationic species. None of the results was meaningful, undoubtedly because of the high solute concentrations. If a n assumption is made as t o the chemical formulation of the complex species, the analytical results can be used t o determine whether it is possible for such a species t o exist. 'The following equation could represent the ionization of such a complex: Fe(NH3),(H20),Z S Fe(iXH3)z(H20),2+
+ 2*-
the sum of z and y, the ammonia and water coordination numbers, must equal six, the maximum coordination number. 2 is the acid anion. If all of the soluble iron is in this complex form, then b y selecting a value for 2 , a series of material balances indicate the regions of the phase diagrams where it is possible for such a species to exist. Figure 13 is a plot of these calculated regions for the chloride system; the solubility envelope of the chloride system at a total solute concentration of 500 g/l. is also shown. There are two limiting factors in this calculation procedure: There must be sufficient acid anion present to be equivalent t o the complex ion (line A), and the amount of ammonium hydroxide present must be great enough to form the complex molecule and to combine with any excess acid anion (line B for a complex with six ammonia ligands and line C for a complex with two ammonia ligands). Comparison of these regions with the experimentally determined phase equilibria indicates that a t high ammonium hydroxide percentages (greater than 700j0), the solubility envelope approaches line A asymptotically, indicating little or no ammonium chloride formation in this area. .4t ammonium hydroxide percentages near 50.0, the envelope enters t h e region where all of the iron cannot exist in a six ammonia ligand complex. Probably, a mixture of complexes with various ammonia coordination numbers exists in these regions, most probably a t the point of maximum solubility, complexes with a n ammonia coordination number of six are predominant since this point is near the intersection of lines A and B. Possibly, the structure proposed is a n idealized one, and dimeric or trimeric complexes could also be involved. This could result from the loss of protons by complexed water molecules with the resulting hydroxyl ions serving as bridges. Discussion
Iron in the ferrous state of oxidation has a surprisingly large solubility in high-strength aqueous ammoniacal solutions containing a n acid anion. The almost complete insolubility expected occurs only a t low solute contents. The solubility
Table II. Component Ratios at M a x i m u m Iron Content
Fe (OH)?- HCl- NH40H-HzO -
CI __
NH3 -
Fe,
CI,
m/l.
m/l.
NH3, m/l.
300 400 500 Theor.
049 077 108
249 342 394
470 588 739
5 4 3 2
300 400 500 Theor.
0.55 0.81 1.19
0.95 1.32 1.33
4.52 5.66 7.51
1.7 1.6 1.1 1.0
4.8 4.3 5.7 6.0
300 400 500 Theor.
0.50 0.79 1.15
1.03 1.32 1.59
5.49 7.06 8.52
2.1 1.7 1.4 1.0
5.3 5.4 5.4 6.0
Concn, g/l.
"3
Fe
9 6 7 6 6 8 6 0 Fe(OH)z-H$04-iYH40H-H20
8.3 7.0 6.3 6.0 Fe(OH)2-HzC03-NH40H-H20
11.0 8.9 7.4 6.0
CI
Fe
1 4 6 0
1 1 1 3
Ind. Eng. Chem. Process Des. Develop., Vol. 1 1, No. 1, 1972
9 7 9 0
145
a t total solute concentrations above 150 g/l. increases in an apparently linear fashion with the total solute concentration. Changing acid anions from the chloride to sulfate or carbonat’e made little difference, but the presence of such an anion is necessary since t8hereis no solubility of iron in a n acid anion-free aqueous ammoniacal solution. The similarities of the three systems can be seen in Table I1 where the concentrations are shown in moles per liter and mole ratios. Apparent’ly, the limitations on the iron solubility are : There must be sufficient acid anion present to combine with a divalent cat’ionic complex; the SH40H concentration must be greater than 50 wt yo of the total solute concentration; and most importantly, the total solut’e concentration should be greater than about 150 g/l. (In the carbonate system a n Fe(OH), solubility of 2% was measured a t 138 g/l. total solute.) The requirement, not only of excess ammonia, but of high-strength solutions (4.5-8.5 moles of ?U”,OH per liter) indicat,es t’he nature of t’he equilibrium involved in the format,ion of the soluble iron species. These concent’rations appear necessary t o form the ammonia ligands. If the system is diluted, the ammonia is replaced by water, aiid the iron becomes insoluble. As the ammonia strength increases, the NH3/Fe mole ratio of all three systems decreases indicating a higher fraction of the ammonia involved in the soluble complex (Table 11). I n the sulfate system a t 500 g/l., the NH,/Fe, SOa2-/Fe, and XH3/S04z-mole ratios are 6.3, 1.1,and 5.7, respectively, which compares with the theoretical values 6.0, 1.0, and 6.0 for an iron hexammine sulfate. Possibly, this ratio would be reached a t higher solute contents. Dean (1952a) , working with the systems lln(OH)2-NH4OH-H,O-HCl, or HsS04 or H2C0,, found a similar high solubility in high-strengt,h ammoniacal solutions. H e concluded that’ t’he manganese in these solutions existed as an anionic complex based on electrical conductivity and electrolytic transport measurement’s. Electrical conductivity measurements (Daniels et al. , 1962) on the iron system reported here showed that a change
did occur as t’he ammonia t o acid anion ratio was varied in solutions of constant iron content as reported by Dean. However, the plot of the change appeared parabolic rather than an intersection of two lines and might well result from the change in T\;H4C1concentration. The results wit’h the ionexchange resins were felt to be a more reliable indication of the nature of the charge on the soluble species. The extent of the iron solubility in ammoniacal solutions suggests a possible applicat’ion in the winning of iron from low-grade ores. A partial reduction of the iron oxides to t,he ferrous state could be followed by dissolution in a highstrength ammonia-ammonium salt solution. Filtering off this solution separates the iron from the gangue materials such as silica and alumina. The iron could then be recovered from solution by dilution, evaporation of ammonia, or b y air oxidation which precipitates bhe iron as hydrated ferric oxide. The economics of such a system would depend on the recovery of the ammonia aiid the ammonium salt, particularly because of the high-strength solutions involved. The ammonia-ammonium carbonate system would appear to be attractive because of the ease of recovery of ammonia and carbon dioxide. literature Cited
Daniels, F., Nathews, J. H., Wi!!ams, J. W.,Bender, P., Rlurphy, G. W., Alberty, It. A,, Experimental Physical Chemistry,” 6th ed. XcGraw-Hill, Iiew York, X.Y. 1962, p 161. Dean, It. S., Ji‘zAing Eng., 4 ( l ) ,55-60 (1952a). Dean, 11. S., U.S. Patent 2,608,463 (August 26, 1952b). Fischer, It. B., ‘‘Quantitative Chemical Analysis,” 2nd ed., Saunders, Philadelphia, Pa., 1961, pp 279-81, 355-6. Klocke, D. J., PhD thesis, University of Pennsylvania, Philadelphia, Pa., 1970. Kolthoff, I. M., Sandell, E. B., “Textbook of Quantitative Analysis,” RIacmillan, New York, N.Y., 1945, pp 562-4. RECEIVED for review April 26, 1971 ACCEPTEDAugust 19, 1971
Kinetics of Hydrogenolyses of n-Butane and Isobutane on Supported Ruthenium J. Christopher Kemplingl and Robert B. Anderson2 Department of Chemical Engineering & Institute for Materials Research, ,VfcNaster Cniversity, Hamilton, Ont., Canada
R u t h e n i u m is one of the most active e1ement.s for hydrogenolysis of paraffins-e.g., for Group VI11 elements on silica in the hydrogenolysis of ethane (Sinfelt, 1969). For 5y0 ruthenium on silica, a n ethane order of 0.8 and a hydrogen order of -1.3 have been reported (Sinfelt and Yates, 1967). These exponents were explained by a mechanism in which the ethane adsorbs reversibly to form a n unsaturated surface species, and the overall rate is limited by the rupture of the 1 Present address, Department of Chemistry, University of Edinburgh, Edinburgh, Scotland. 2 To Fhom correspondence should be addressed.
146 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1 , 1972
carbon-carbon bond (Cimino et al., 1954). The hydrogenolyses of ethane and propane on ruthenium on alumina have been studied in a continuous stirred-tank reactor similar to that used in this study (Tajbl, 1969). The ethane hydrogenolysis was +1 order in ethane and -2 order in hydrogen; the piopane hydrogenolysis was + l order in propane and -3/2 order in hydrogen. Propane reacted about 75 times faster than ethane. Examinations of the hydrogenolysis reactions of larger hydrocarbons have been principally concerned with the distribution of products. Matsumoto et al. (1970) have contrasted