GRANTW. SMITHAND HOWARD W. JACOBSON
1008 4oo
Vol. 60
trapolating this straight line to the point where M1/a = 0. I By a plying these procedures to depolymerization of colloidarsilica, values of have been obtained for colloidal T
silica formed a t various pH's. They are shown in Fig. 2 and indicate that larger particles are formed a t higher pH. Depolymerization experiments were carried out by diluting one volume of colloidal silica solution with 100 volumes of the solution containing 1 g./l. NaSCOa (pH 10.8). The colloidal silica solutions used in these experiments were prepared by aging a t various pH's for 6 days and contained 2 g. of SiOzper liter.
h
'8 300
v
c
200 7
6 Fig. 2.-Values
8
9
10
PH.
for colloidal silica formed at various pH's. Suito, et aL.,'O have derived the following equations for the dissolution of fine particles, assuming the dissolution rate to be proportional to the surface area of the particles. n'/a M1/a = K(T t ) of
T
-
~ = b v o
where M is the total weight of particles at time t ; 7 , the time necessary for complete dissolution (in the present study T means the time necessary for complete depolymerization); UO, the initial radius of particle; ~t, the number of particles; and b and K are proportionality constants. Therefore, the MVa versus t plot becomes a straight line for monodisperse systems, and T can be obtained by ex(10) E. Suito, N. Hirai and K . Taki, J . Chem. Soc., 1 3 , 713 (1951).
Discussion I n no case has polymerization reached true equilibrium, since in this 'case the system, a t iiifinite time, would consist of water containing a single large particle of amorphous ' silica. After aging the colloidal solutions for only 6 days, surely only a pseudo-equilibrium is established. It might therefore be postulated that as the monomer disappeared to form colloidal particles, these colloidal particles further polymerize, or at least increase in particle weight, and this rate of increase in particle size is also a function of pH, since the largest particles are formed a t highest pH. From the foregoing evidence, it is clear that hydroxyl ions promote the polymerization of silicic acid over a pH range wider than that reported by previous workers. It is very interesting to note that the polymerization takes place more rapidly a t higher pH, where colloidal silica particles are highly charged and depolymerize very rapidly. This might be evidence for the fact that the same mechanism is involved both in polymerization and in depolymerization. Acknowledgment.-The author is indebted to Prof. Y. Uzumasa, Prof. Q. Okamoto and Assist. Prof. T. Okura for their valuable advice and encouragement ,
~
CHARACTERISTICS OF ADSORPTION OF COMPLEX METAL-AMMINES AKD OTHER COMPLEX IONS OF ZINC, COPPER, COBALT, NICKEL AND SILVER ON SILICA GEL1 B Y G R A N T IT.SMITHAND HOWARD w.JACOBSON Contribution from the College of Chemistry and Physics, The Pennsylvania State University, University Park, Pa. Received April 18, 1066
The adsorption isotherms for complex metal ammines, ethylenediamine-metal complexes, and diethylenetriamine-metal complexes on silica gel are shown. The extremely complex nature of the adsorption of metal ammines is demonstrated for the nickel and co per ammines. Metal ammines are adsorbed ill the fol!owing decreasing order, in millimoles adsorbed per gram of silica gel? zinc, copper, cobalt, nickel, silver. For diethylenetriamine complexes, those with higher formation constants were more highly adsorbed. For ethylenediamine complexes studied, six-coordinate systems were more highly adsorbed than four-coordinate. An interpretation of the adsor tion process in terms of hydrogen bonding of ligand to silica surface is presented. Complex copper ammines, copper etEylenediamines and copper diethylenetriamines dissociate during adsorption on silica gel. This is shown by a comparison of the absorption spectra of solutions of the complex metal ions before and after adsorption. The ratio, animonia:copper ion adsorbed is higher than that of the complex species originally in solution. The more stable a given complex ion, the closer the ratio for the adsorbate agrees with that of the 8peCieS initially in solution.
Introduction This study was undertaken to correlate the relative adsorption of complex metal ammines and other nietaI complexes in which ethylenediamine (1) Tliis paper is baaed on part of a thesis submitted by Howard Wayne Jaoobson in partial fulfillment of the requirements for the degree of Doctor of Philosophy at The Pennsylvania State University, August, 1953.
(abbreviated en) and diethylenetriamine (abbreviated dien) are the ligands, with the structure and stability of the complex entities. The unusually strong adsorPtiol1 of ComPlex metal amnlines on silica gel was first reported by Smith and Reyerson.2 adsorbed On the surface (2) G. W. Smith and L. H. Reyerson, J . A m . Chem. Soc., 63, 2584 (1930).
July, 19%
ADSORPTION OF COMPLEX METAL-AMMINES AND OTHER COMPLEX ION^
of silica gel form a quite stable system. Ordinary electrodialysis fails to remove the adsorbed ions to any extent, but changes do take place after considerable time. This indicates that a surface disintegration of the gel with its adsorbed ions has oc~ u r r e d . ~Studies of the adsorption of copper ammine on silica gel have shown that final equilibrium is reached only after very prolonged shaking, and that the composition of the copper ammine ion undergoes continuous change during the process. Experimental The silica gel was a commercial product of 6-12 mesh size, obtained from the Davison Chemical Corporation. It was cleaned by treating with 6 M nitric acid for 12 hours to remove impurities and was then washed with frequent changes of distilled water over a period of one week. It was then dried a t 120' for 24 hours and finally at 300" for four hours. The silica gel used throughout the entire study came from the same lot. Stock solutions of the complex metal ammines were prepared by placing the desired quantity of metal nitrate in a liter volumetric flask and adding concentrated ammonium hydroxide until the precipitate formed just dissolved. At this point an excess of 10 ml. of concentrated ammonium hydroxide was added. The solutions were then diluted to the correct volume. Solutions of the complex metal ammines of lower concentratioiis were prepared by diluting these stock solutions. The cobalt-ammine solution required 25 ml. more of concentrated ammonium hydroxide per liter than the other metal ammines. Stock solutions of the complex metal ions, in which ethylenediamine and diethylenetriamine were the ligands, were prepared in a manner similar to the complex metal ammines. In both cases an excess of 10 ml. of Concentrated reagent was added per liter of final solution. Solutions of the ligands alone-ammonia, ethylenediamine and diethylenetriamine-were prepared for adsorption by diluting the concentrated reagents. The samples were prepared for the adsorption study as follows: 10.00 g. of silica gel was weighed into 125-ml. polyethylene bottles and then 75 ml. of solution was added. Duplicate samples were run in all cases. The bottles were capped and mounted on a shaker in a 25.0' water-bath and rotated a t a rate of 25 revolutions per minute for a period of 72 hours. Two special series were run with the copper- and nickel-ammines. I n the first of these, the volume of the metal ammine solution was 50 ml. and the weight of silica gel was 5 g. In the second, the volume of the metal ammine solution was 50 ml. and the weight of the silica gel was 2 g. At the end of the shaking period the bottles were removed and the solutions were analyzed for the metal ions in the case of the series of metal complexes, and for ammonia, ethylenediamine or diethylenetriamine in studies of ligand adsorption. The following analytical methods were employed. (1) Silver: precipitated as silver chloride and weighed in sintered glass crucibles. ( 2 ) N.ickel: precipitated as nickel dimethylglyoxime-and weighed in sintered glass crucibles. The precipitation was effected by the hydrolysis of urea from homogeneous solution (3) Copper: iodometric titration in sulfuric acid solution. with sodium thiosulfate which had been previously standardized with freshly recrystallized potassium iodate. ( 4 ) Zinc: precipitated as zinc ammonium phosphate and ignited to the pyrophosphate and weighed as such. ( 5 ) Cobalt: electrolytically deposited on platinum electrodes. ( 6 ) Ammonia, ethylenediamine, diethylenetriamine: titrated with sulfuric acid previously standardized with sodium carbonate. The silver complexes involving ethylenediamine and diethylenetriamine were not studied because they did not form stable systems. A considerable amount of insoluble hydroxide and oxide formed in these caseA which could not, be dissolved by the addition of excess ethylenediamine or die tliylene triamine. (3) L. H. Ryerson and R. E. Clark, TKIB JOURNAL, 40, 1065 (1936). (4) I. hl. Kolthoff and V. Stenger, ibid., 38, 475 (1934).
1005
To exhibit the extent to which complex copper-amines, copper-ethylenediamine5 and copper-diethylenetriaminefl dissociate during adsorption, solutions of the complex metal ions were prepared in which the 1igand:metal ion ratio was varied in integral steps from one to the maximum coordination number of the metal ions. Ammonium nitrate was added to stabilize these systems with respect to insoluble hydroxide formation. The minimum amount of ammonium nitrate required to prevent hydroxide formation in the least stable member of a series was used throughout that entire series. Immediately after the 72-hour shaking period with silica gel, the solutions were subjected to quantitative and spectro hotometric analysis. A Beckman Model DU Photoetctric Quartz Spectrophotometer was used to take the absorption spectra before and after adsorption. In using the spectrophotometer, minimum slit width and maximum sensitivity were employed. To obtain the absorption data, the solutions of the complex metal ions were compared t o an ammonium nitrate solution whose concentration was equal to the ammonium nitrate concentration present in the complex metal ion solution in question. The ammonium nitrate was 6.0 M in the copper-ammine series, 3.0 M in the copper-ethylenediamine series and 2.0 M in the copper-diethylenetriamine series. The pH of each solution was measured both before and after adsorption with a Model G Beckman p H Meter.
Results Figures 1-G show the adsorption isotherms of indicated systems on silica gel. They were obtained by plotting the equilibrium concentrations of the solutions against the number of millimoles of complex ion adsorbed per gram of silica gel. Discussion The ordinary hydrated metal ions are not significantly adsorbed on silica gel, but the replacement of the bound water molecules by ammonia, ethylenediamine or diethylenetriamine results in considerable adsorption of the metal ions. Figure 1 shows that ammonia, ethylenediamine and diethylenetriamine all exhibit a high affinity for the surface of silica gel. The surface of silica gel can be considered to be made up primarily of oxygen atoms since the silicon atom is quite small. It seems conceivable to account for the strong adsorption of the complex metal ions on silica gel by considering hydrogen bonds to be formed between the oxygen atoms in the surface of silica gel and the nitrogen-containing ligands which in turn are coordinated to the metal ion. In the case of the aquated metal ions, a similar bonding could be considered to exist between the hydrogens of water and the surface oxygen atoms, but the water-metal ion bond is not of sufficient strength to hold the metal ion to the surface. Water itself is adsorbed to a considerable extent on silica gel. The structure of the complex entity in solution should be one of the important factors governing the amount of metal complex adsorbed. For the complex metal ammines, a maximum of six hydrogen atoms could lie on a flat surface for tetrahedral structure. For a planar ammine, a maximum of eight hydrogens could attach to a flat surface if the entity were on its face and four if on an edge. In an octahedral structure, the maximum is six, while for the linear type it is four. The structures of the nietal ammines involved in this study are generally thought to be as follows: zinc-ammine, tetrahedral ; copper-ammine, planar; cobalt-ammine, octahedral; nickel-tetraamine, tetrahedral; and silver-
GRANTW. SMITHAND HOWARD U. JACOBSON
1010
0
I
2
Fig. 1.-Adsor tion of ligands by silica gel: I, ammonia; 11, ethygnediamine; 111, diethylenetriamine.
I
I
a
3
3
8
I
W A R CONCENTRATION.
MOLAR CONCENTRATION.
J
Vol. 60
I d
Fig. 4.-Adsorption of diethylenetriamine complexes by silica gel: I, copper-ethylenetriamine; 11,nickel-diethylenetriamine; 111, zinc-diethylenetriamine; IV, cobalt-diethylenetriamine.
I
MOLAR CO(ICEWTRATION.
Fig. 2.7Adsorption of metal ammines by silica gel: I, zinc-ammme; 11, co per ammine; 111, cobalt-ammine; IV, nickel-ammine; silier-ammine.
$
01
II
I
2
I 3
I
MOLAR COWCENTRATION.
Fig. 5.-Adsorption
of copper-ammine solutions by silica gel.
Fig. 6.-Adsorption
of nickel-ammine solutions by silica gel.
MOLAR CONCENTRATION. MOLAR CCUCENTRATION.
Fig. 3.-Adsorption of ethylenediamine complexes by silica gel: I, zinc-ethylenediamine; 11, cobalt-ethylenediamine; 111, nickel-ethylenediamine; IV, copper-ethylenediamine.
ammine, linear. The latter has been determined in the form of silver ammine s ~ l f a t e . The ~ other structures are based on the electronic configuration of the metal ions, magnetic measurements, and by comparison to known structures of other deriva( 5 ) A. F. Wells, "Structural Inorganic Chemistry," Clarendon Press, Oxford, 1950, p. 120. .
tives of the metal ions.6 Considering the adsorption of metal ammines on silica gel on the basis of millimoles per gram adsorbent, the following decreasing order is observed : zinc, copper, cobalt, nickel and silver. I n terms of structure, that would be tetrahedral, planar, octahedral, tetrahedral and linear. No reliable correlation between the geometry of the species and adsorbability is evident from the (6) Ref. 5 , pp. 318.
TABLEI TABLE OF FORMATION CONSTANTS Metal ion
Zinc Copper Cobalt Nickel Silver
Ammonia complexes log ka
log ki
2.18 3.99 ,99 2,67 3.23
2.25 3.34 ,51 2,12 3.83
log ks
2.31 2.76 o.93 1 ,61
log k4
log ks
Ref.
log kc
1.96 1.97 o,64 o,06 -o,74 1.o7 o, 63 -o, o9
Ethylenediamine complexes Zinc Cobalt Nickel Copper
5.92 5.93 7.60 10.76
5.15 1.86 4.73 3.03 6.48 5.03 9.37
8 g 10 10 '
Diethylenetriamine complexes Copper Nickel Zinc Cobalt
1011
ADSORPTION OF COMPLEX METAL-AMMINES AND OTHERCOMPLEX IONS
July, 1956
16.0 5.3 10.7 8.3 8.9 5.5 8.1 6.0
11 11 11 11
zinc, cobalt and nickel, have three values shown for the formation constants indicating six-coordinate systems, Only two values for the copper complex indicates a four-coordinate system. I n this series the six-coordinate systems are all adsorbed to a greater extent than the four-coordinate copper ion. For diethylenetriamine complexes there seems to be afair correlation, i e . , the greater the stability (as indicated by higher formation constants), the higher the adsorption. This trend seems to be reversed for the ethylenediamine series, however, and no correlation is evident for the series of ammonia complexes. For a simple adsorption system, one can change the amount of adsorbent or the amount of solution in contact with an adsorbent and still get points falling on one adsorption isotherm. The curves of Figs. 5 and 6 show that for the adsorption of copper and nickel ammines on silica gel, this is not the case. It appears that variations in the ligand: metal ion ratio adsorbed were such that one reaches a different equilibrium state when the amount of
TABLE I1 Empirical composition
Wave length of peak Before After ads. ads.
Before ads.
E
After ads.
Ligand/metal ion
Aratio ads'a B
Before ads.
c u ++ 815 815 14.80 14.80 .. .. 2.60 CU("I)I 755 765 21.75 19.60 3.9 6.3 4.48 Cu(PJH3)z' + 700 715 30.50 26.00 3.5 4.9 5.23 Cu(NH3)3++ 655 675 40.00 33.00 4.1 4.8 5.95 CU(NH~)~++ 600 645 50.00 39.70 5.7 6.1 6.70 Cu( en)l + + 655 665 30.50 28.60 3.0 4.9 4.95 Cu(en)z++ 555 560 60.00 49.25 2.3 4.3 6.51 Cu(dien)l++ 610 610 52.95 64.80 1.0 .. 5.33 Cu( dien)z + + 580 580 81.00 89.75 2.0 .. 7.96 a Column A, calculated from shift in wave length; column B, calculated from decrease in E . sorption was 0.0400 M .
PH
2.60 4.43 5.02 5.45 5.95 4.76 5.46 4.67 6.43
++
data. However, there is considerable doubt that the nickel tetraamine is tetrahedral. If it is actually octahedral, the position of this ammine in the adsorbability series appears reasonable. In this case, the other two positions in the octahedral arrangemerit would be occupied by water molecules, and the relatively lower adsorption of the nickel ammine would be satisfactorily accounted for. The way a complex ion is oriented on the silica surface probably depends not only upon its shape but also upon the inter-atomic Spacings. If the hydrogen-bonding theory applies, the number of hydrogens able to make contact with a "flat" Surface is probably of less importance than the number of hydrogens spaced properly t'o form good hydrogen bonds with the surface oxygen atoms. The stability Of the entity should also be an important factor COntrOlhlg the amount Of metal complex adsorbed. Table I shows the logarithms of the stepwise formation constants of the indicated complex ions. The ethylenediamine complexes of (7) J. Bjerrum, "Metal Ammine Formation in Aqueous Solution," P. Haase and Son, Copenhagen, 1941, p. 289. (8) J. Bjerrum and P. Anderson, Kgl. Danske Vikenskab. Math-Fya >fed& 11, NO. 582 (1931). (9) L. J. Edwards, Doctors Dissertation, University of Michigan, 1950. (10) F. Basoh and R. K. Murmann, J . Am. Chem. Soc., 74, 2373 (1952). (11) J. E. Prue and Scliwarzenbach, Helv. Chim. Acta, 38, 963 (1950).
After ads.
b
Molar concn. after adsorptionb
0.0400 .0378 .0339 ,0282 ,0267 0388 .0345 .0383 ,0211 I
Concentration before ad-
solution or the amount of silica gel were changed. The copper-ammine adsorption curves are somewhat closer together than those of the nickel-ammines. This no doubt is due to the fact that the copper-ammine complex is more stable than the nickel-ammine complex. This is indicated by the formation constants. There must be some dissociation of the complex entities during adsorption in order to reach a different equilibrium state by varying the weight of adsorbent or volume of solution. By comparing,the absorption spectra of the COPper complex solutions after contact with silica gel with those of the original solutions, changes undergone in the complex ions during adsorption are revealed. I n the case of the copper ammines, the adsorption process causes the maximum absorption peak to be shifted to higher wave lengths and the maximum absorption value to decrease. This indicates that the ammonia :copper ion ratio adsorbed is somewhat higher than that of the primary complex species initially in solution. Dissociatioli of the complex metal ion has apparently occurred while in contact with the ,silica gel. Whether the dissociation of the complex ion has taken place at the surface of the silica gel O r in the bulk solution is not revealed, but it is probably a combination of both. Table 11 shows the changes in wave length and extinction coefficientE duril% the adsorption process along with an empirical value of the ratio of
'
JOHN Foss
1012
the ligand :metal ion adsorbed. These ratios were calculated by linear interpolation by considering shifts in the wave lengths and also by considering changes in the molar extinction coefficients during adsorption. These two methods give quite different values in some cases, but when one considers that for most of these systems, the absorption maximum is not too sharply defined, some differences mould be expected. The values calculated using the shift in wave length method were made by observing the change in the maximum absorption peak with reference to the center of the peak before and after adsorption. The first value for the copper-ammine and ethylenediamine complexes is somewhat out of line with the
Vol. 60
other values since the absorption maximum of the aquated copper ion is very poorly defined. The more stable a given complex ion (Table I), the nearer this ratio approaches the value of the primary complex specics initially in solution. The increase in the molar extinction coefficient after adsorption for the copper diethylenetriamine series is unexpected. There is also a widening of the entire absorption maximum. The pH values show that in all cases there is a lowering of pH with adsorption, but the extent of this lowering follows no simple relationship. The p H of the solutions would not be expected to change much since a rather large quantity of ammonium nitrate is present in the systems.
L
1
INTERMEDIATE ORDER HETEROGENEOUS CATALYSIS AND HEATS O F ADSORPTION BY JOHN Foss Department of Chemistry, University of Utah, Salt Lake City, Utah Received February 83,1966
An explicit relationship, based on a simple model, is developed between the heat of adsor tion of a reactant on a surface and the experimental energy of activation for the unimolecular reaction on that surface. 8urnerical values are calculated for the decomposition of ammonia on tungsten.
Introduction I n the literature of heterogeneous catalysis there appears to be no simple discussion concerning the relationship between exp4rimental heats of adsorption and catalysis other than for the limiting cases of first and zero order reacti0ns.l For the former the experimental energy of activation is usually stated as being equal in magnitude to the difference in energy between the activated state and the gaseous reactant. For the latter it is this same quantity, plus the heat of adsorption of the reactant a t full surface coverage. However, actual catalytic studies will in general not take place in one of these limiting regions but rather in an intermediate order region. I n this paper the relationships between experimental heats of adsorption and activation energies are presented in a form simple enough to permit use. The present paper is limited to unimolecular reactions. On the assumption that the rate-controlling step is the reaction of the adsorbed species on the surface we have, from absolute reaction rate theory for a simple unimolecular reaction2 f* kt v = kt - cB - e-(ra + d / k T M - cae-(co + d / k T h fa h where ca
= concn. of adsorbed molecules in no. per square cm.
partition function for the transition state per square cm. f.f = partition function for the adsorbed gas per square cm. CO = height of the highest energy barrier above the ground state of the initial gaseous reactant e = the heat of adsorption of the reactant
f
=
(1) K. J. Laidler, ”Catalysis,” Vol. I, ed. by P. H. Erninett. Reinhold Publ. Gorp., New York, N. T..p. 135. (2) R. Glaastonc, K. J. Laidler and H. Eyring. “The T l i ~ o r yof Rntr Processes,” hlcGraw-Hill Book Co., New York, N. I’.
Now cB = LO where L is the number of sites per square centimeter on the uncovered surface (around and 0 is the fraction of the surface which is covered. This in turn is given by 1
8 =
1
+ fi h!
f a CY
1
M
1
e-JkT
+! ! e-r/lT
(1)
c,
where fa is the surface partition function, Fg is the partition function per cubic centimeter of gaseous reactant, and cg is the gas concentration in molecules per cc. We should note that in using this Langmuir type adsorption isotherm we are assuming a homogeneous surface, and are attributing decreases in heat of adsorption with coverage (such as introduced below in equations 4 and 5 ) to lateral interaction, or changes of the type suggested by B~udart.~ Combining our equations we then have v = -kT L
CP
e-ea/kT
Fg
1
+ %- ec/kT
(la)
F g
The partition function Fg may be approximated by K I T m- 1 where K1 is a constant dependent only on the mass and moment of inertia of the reactant gas. The gas concentration is given ideally by K2 p / T . Making these substitutions and taking the logarithm of (la) gives 3 lnv = l n c ‘ - ; ; l n T + l n p &
(where c’ contains temperature independent t8erms and K = K2/K1). (3) M. Boudert, J . Am. Chsm. Soc., 74, 3850 (1952).
n
R
