330
J. H. SINGLETON A N D G. D. HALSEY, JR.
simplified continuous polymerization hypothesis, such as that used by Gran6r and Sill6nZ7in their interpretation of bismuth hydrolysis, does not appear justified a t this time since, apparently, it can lead to erroneous concIusions regarding degrees of polymerization.28 The hydrolysis products of thorium have been written as complexes with hydroxide ions (e.g., Th(OH)z++ and Thz(OH)z+6)rather than as complexes with oxide ions (e.g., T h o + + and ThzO+6). The techniques used here, of course, do not permit a decision as to which of these formulas is correct. However, recent work by Lundgren and Sill6nZ6 on basic salts of Th(1V) and U(1V) establishes the existence of hydroxide bridges in basic salts of these elements. Using these structure determinations as a guide, one may postulate that the hydrolysis products of thorium are complexes with hydroxide ions and that the polymeric species such as the dimer postulated in this work are held together by hydroxide bridges. In addition, the (27) F. Graner and L. G. S i l l h , Acta Chem. Scand., 1 , 631 (1947). (28) I n preliminary ultracentrifugations of Bi(II1)-perchlorate solutions under conditions where Graner and Sillen calculated a weight average degree of polymerization Nw = 20,a degree of polymerization of less than 6 was found, a8 well aa considerably less polydispersity than indicated by Grandr and Sillen (R. W.Holmberg, K. A. Kraus and J. S. Johnson, unpublished results). Similarly, zirconium perchlorate solutions were found t o have a degree of polymerization of approximately 3 (K. A. Kraus and J. S. Johnson, J . A m . Chem. Soc., 7 6 , 5769 (1953))under conditions where Connick and Reas (ibid., 73, 1171 (1951)), also using a hypothesis of continuous polymerization, suggested a weight average degree of polymerization of ea. 300.
Vol. 58
species are unquestionably hydrated and might be written as Th(OH)2(H20)6+2and ((Hz0)6ThgzTh(HzO)s)+ if a coordination number of 8 is assumed. 4. Hydrolysis in Chloride Solutions.-The early steps in the hydrolysis of ThC14 were studied in KCI-HCI solutions of total concentration 1 M . The data were obtained in an attempt to establish if hydrolysis of Th(1V) is sensitive to chloride complexing according to the equation Th+4
+ C1-
ThCl+a
(12)
for which the concentration quotient
.had earlier been found by Waggener and Stoughton a t p = 1.OO.I6 Thorium was found to hydrolyze to approximately the same extent in chloride and perchlorate solutions. The effect of chloride complexing on n was estimated, and the predicted decrease in n was considerably outside the experimental error. If chloride complexing occurs as postulated by Waggener and Stoughton, one must conclude from the insensitivity of the hydrolytic reaction to chloride ions that basic chloride complexes of the type Th(OH)&I+ and Thz(0H)zC1+6, etc., are formed and that the hydrolysis constants of such chloride complexes are approximately the same as those of the uncomplexed ions.
THE ADSORPTION OF ARGON ON XENON LAYERS1s2 BY J. H. SINGLETON AND G. D. HALSEY, JR. Department of Chemistry, University of Washington, Seattle, Wash. Received November 9, 106S
The adsorption isotherms for argon on pre-adsorbed films of xenon have been measured on graphitized carbon black, silver iodide and anatase. The character of the isotherms change progressively as the amount of pre-adsorbed xenon is increased. The limiting isotherm with inwease in amount of xenon is different for the three solids. This fact indicates that there is a limiting film thickness, beyond which the film of xenon is not stable,
Introduction Adsorption isotherms have been measured on a variety of aubstances in the past. Most of the surfaces have been poorly characterized and chernically complex. Even the ideal case of argon on potassium chloride is not free of difficulties. One may have a number of crystal faces exposed, frozen in defects in crystal structure, and adsorbed water, to mention a few. Such incidental factors make comparisons of isotherms taken on different surfaces difficult, Therefore we have measured a aeries of isotherms on a given solid covered with various thicknesses of xenon, deposited in a reproducible, reversible manner. The films thus formed are as pure as the bulk xenon itself, if we assume that any strongly adsorbed dirt remains on the surface of the solid, with which it presum(1) This research was supported by Contract AF19(604)-247 with the Air Force Cambridge Research Center, (2) Presented a t the la3rd Natlonal Meeting of the American Chemical Society, Los Angeles, Calif,, IMrsrrh 16-20, 1953,
ably has specific interaction. There is then an analogy with the technique of evaporating metal films for chemisorption studies13 with the difference that the xenon films are in equilibrium with the solid, Experimental Vacuum System.-The adsorption measurements were made in a conventional constant volume system. Gas pressures were measured in manometers, 10 mm. in diameter, read with a cathetometer graduated to 0.06 mm. Gases.-Tank helium, from Western Oxygen, Inc., was Nitrogen was produced by heating better than 99.9% pu;e. sodium azide to 350 . Xenon and argon of 99.9% purity were obtained in sealed glass bulbs from Air Reduction Sales co Adsorbents.-Silver iodide was precipitated from hot 0.1 HI by addition of 0.05 j!f AgN03. It was digested for six hours, filtered, washed with hot water and absolute alcohol, and dried in a vacuum desiccator. It was always shielded from light.
.
(3) 0. Beeok. A. E. Smitb m d A . Wheeler, Proc. Roy. Boo. (Landon),
A177, G2 (1940).
33 1
ADSORPTION OF ARGONON XENON LAYERS
April, 1954
sample great care was required to obtain repro ducible results. Slight irregularity in the xenon layers creates heterogeneity and radically alters the isotherm for argon. In the initial experiments the xenon monolayer volume was estimated tielium supply or Vacuum from the “point B” of , , =-Manometer the xenon -isotherm a t - 124.G’ (Table I), The Gas Inlet argon isotherm taken on this nominal one-layer volume of xenon showed a pronounced “point B” a t 0.15 nominal monolayers of argon (isotherm TABLE I labeled 0.851ayer, Fig. 2). vm in ml./g. (areas in parentheses, me2/g.) When 20% more xenon Argon Xenon Nitrogen was pre-adsorbed, the Carbon black 4 . 1 4 ( 1 1 . 9 ) 2.30 (11.6) (12.516 “point B” disappeared 0.47 ( 1 . 3 4 ) 0 . 2 6 ( 1 . 3 2 ) 0 . 2 9 (1.27) Silver iodide (isotherm labeled 1.0, 4 . 1 4 (11.9) 1.88 ( 9 . 5 0 ) (13.8Y4 Anatase Fig. 2). This volume of Outgassing Procedure.-Anatase and carbon black were xenon (1.20 urn), was Silvered Dewar heated to 300’ and evacuated to 10-5 mm. overnight, before use. Silver iodide was outgassed a t room temperature; taken to be one layer and higher temperatures caused loss of area. The weight of each used to compute the sample was chosen to give a total urn of about 3 ml. number of layers preDeposition of Xenon Layers.-Preliminary experiments adsorbed (no similar adSample showed that a slow rate of cooling to liquid nitrogen temperature was necessary to obtain one or more equilibrium justment was made for Vapor Pressure xenon layers. At the final temperature, the vapor pressure urn for argon and the valThermometer of the xenon is less than 10-8 mm. and it will not redistribute ues of e are dotted in at a reasonable rate. To facilitate controlled cooling the multiples of th&volume). sample was enclosed in a small Dewar vessel shown in Fig. 1. isotherms Fig. ].-Dewar vessel for The vertical temperature gradient was minimized by making These two adsorption sample. the entry tubes long and thin. The horizontal gradient was demonstrate the sensicontrolled by evacuating the Dewar until the desired rate tivity of the results t o of cooling was achieved. The conductivity of the sample slight changes in coverage, and explain the necessity was maintained by including a few millimeters of helium for very careful cooling to ensure that all the xenon with xenon. When the sample reached the temperature of liquid nitrogen (2-24 hours) the helium was removed and reaches its equilibrium location. the outer jacket filled with hydrogen. Rough calculations .The isotherms will now be discussed in order. indicated that the thermal gradients were small; however, (The term “step” is used as a traditional way of the ultimate criterion of success in deposition was the reproducibility of the argon isotherm with further decrease in referring to a pronounced point of inflection in the rate of cooling. Results reported here have all been repro- isotherm; it is not to be confused with a first-order duced, in many cases with different sample bulbs of slightly transition, which is a real step.) With no prechanged shape and design. adsorbed xenon the presence of steps indicates Measurement of Temperature and p/po.-In the final design of the Sam le bulb (Fig. 1) an argon vapor pressure that the argon is adsorbed more or less layer by thermometer was focated in the center of the sample. This layer as the pressure increases. With 0.43 layer constant indication of PO was necessary for the carbon black of xenon a sort of dual surface has been prepared, isotherms, because of their complexity. For the other two half xenon and half carbon. The carbon half is samples po was obtained a t the end of the isotherms, by in- covered at p/po < 0.01, which creates a dual surface troducing a large excess of argon; this produced some error in the results for p / p o < 0.9. A check isotherm, using the of xenon and argon. The step corresponding to the refined bulb with enclosed thermometer reproduced the silver second layer of argon on the bare carbon occurred iodide results exactly up to p / p o 0.9. The temperature at p / p o = 0.35. Here a smaller step a t p / p o = for theisothermsreported here was -195.8 =k 0.4’. Measurement of the Isotherm.-Adsorption isotherms of 0.2 reflects the dual nature of the surface, and the argon on the bare surfaces and on a series of pre-adsorbed greater van der Waals attraction of a mixed layer xenon layers were determined two or more times, usually of xenon and argon, than that of a first layer of with a second sample of the adsorbent. The graphs (Figs. argon alone. Further steps are obscured, because 2-7) include only a portion of the actual points taken. The isotherms were reversible. For one or more layers of xenon of the mixed surface. The isotherm with 1.0 layer of xenon shows the the argon was removed completely by pumping for 15 minutes. The isotherm could then be reproduced showing that return of a larger step which is now at p / p , = 0.15 no xenon came off, and that the pre-adsorbed xenon was not because the first layer is all xenon. The step for rearranged by heat from the entering argon. the second layer of argon is very slightly lower than Discussion of Results the third layer step for the bare surface ( p / p o = Isotherms on Graphitized Carbon Black (Fig. 2). 0.6). -Owing to the nearly uniform surface of this It is interesting that the isotherm for the 1.28 layers of xenon crosses and recrosses the one layer (4) W. D. Harkins and G. JUCE,J . Am. Chem. Soc., 66, 1362 (1944). (5) M.H. Polley, W. D. Schaeffer and W. R. Smith, THIB JOURNAL, isotherm because the step in it is a t a lower p/p,, 57, 469 (1953). but smaller. Evidently the fraction of a second (6) P. K. Emmett and S . Rrunauer, J . Am. Chem. Soc., 59, 1553 layer of xenon scattered over the first layer creates (1937).
A sample of standard anatase,4 surface area 13.8 r;n.’/g. was obtained from Professor George Jura, Universrty of California, Berkeley. A sample of graphitized carbon black6 designated P-33(2700’) was obtained from Dr. W. R. Smith, Godfrey L. Cabot, Inc., Cambridge, Mass. It has been shown6 to have the characteristics of a nearly uniform surface. The dead s ace of each sample was determined with helium at -lis, -78 and 0’. The value of om, t,he volume of a monolayer, was dttermined with argon a t -195’ and with xenon a t -124.6 . B.E.T. plots were used for anatase and silver iodide but the plot was unsatisfactory for carbon black, so “point B” was estimated directly from the isot,herm. Areas were then obtained using the figures of Emmett and Brunauer.6 The density of liquid xenon was taken from the “International Critical Tables” (1926). The resulk are tabulated in Table I.
t
w
5
J. H. SINGLETON AND G. D. HALSEY, JR.
332
VOl. 58
I LAYERS XENON PR E ADSORBED -
-
3
-
b 0
6 9 0
P
+
2
o .43 .05 1.0 1.28 2.0 3.0 6.0 9.7 19.4
B'
1
0 0.2
0
Fig. 2.-The
0
Fig. 3.-Argon
0.4
0.6 PlPo. adsorption of argon on layers of xenon on carbon black at
0.8
1.0
- 195".
0.10 PlPo. isotherms on one and two layers of xenon on carbon black a t low partial pressures ( 195"): 0 , one layer; 0, two layers. 0.05
-
April, 1954
ADSORPTION OF ARGON ON XENON LAYERS
,
333
7
4
LAYERS XENON PRE ADSORBED
-
T & 9
3
o 0.5 0.67
0
x -
0 0
$2
1 1.5 2 4
i
1
0
0
0.2
0.4
I
0.8
1.0
P/PQ.
Pig. 4.-The
adsorption of argon on 1:~yersof xenon on silver iodide at - 195’.
some higher energy sites a t the same t’ime as it takes up some of the available space in the second layer. The crossing is repeated near p / p , = 0.G and again faintly a t 0.8. The first step on two layers of xenon is near p / p o = 0.3, very much lower than the corresponding step on the bare adsorbent. The isotherms on 3.6 and 9.7 layers are only slightly different, showing that the van der Waals forces from the solid are almost negligible. The isotherm for 19.7 layers is indistinguishable from that for 9.7, which means that the surface forces are now completely negligible, or perhaps more likely that the extra layers condensed in low surface-area xenon crystals and never spread on the surface. It appears, however, that at least about six layers were successfully de-, posited. Figure 3 shows argon isotherms on one and two layers for partial pressures up to 0.12. All isotherms for more than two layers lie on the curve for two layers, within experimental error. The isotherm for one layer is slightly, but definitely, below. It appears that a really complete, nearly uniform layer of xenon is produced only in the first layer, and that for higher layers a few xenon atoms are out of place, which creates a slight heterogeneity. Isotherms on Silver Iodide and Anatase.-The isotherms on these solids (Figs. 4 and 5 ) are con-
siderably simpler than those for the carbon black. The isotherms show no steps, and the “point B” values are not as well defined. These observations suggest strongly heterogeneous surfaces. I n the absence of a sharp “point B” an independent estimate of the vm for xenon is not possible. The number of layers of xenon was calculated from the In the B.E.T. vm values for xenon at -128.4’. case of silver iodide there is some evidence that this value is correct. The isotherm labeled 1.0 shows a slight “point B” near 0 = 1 at p / p o = 0.25, while the neighboring isotherms do not. Isotherms for more than one layer are Brunauer’s Type 111. There is no change in the isotherm beyond two layers of xenon. In Fig. 6, the adsorption of nitrogen on silver iodide, with and without xenon, is compared with .the argon isot.herms. The isotherms are similar with nitrogen somewhat more strongly adsorbed. It is interesting to note that even for nitrogen, the “point B” values are poorly defined, and the nitrogen is weakly adsorbed. We have been informed? that this is an exception to the generalization that nitrogen is unique in giving good “point B” values and concomitant good B.E.T. plots and surface area measurements. The argon isotherms on anatase are even simpler (7) P. H. Emmett, private communication.
334
J. H. SINGLETON AND G . D. HALGEY, JR.
Vol. 58
4
4
Q .I
3
I
LAYERS XENON PRE -ADSORBED
1
0 0
0.2
Fig. 5.-The
0.6 0.8 P/PQ. adsorption of argon on layers of xenon on anatase at - 195’. 0.4
than those on silver iodide. There is a decline but no disappearance of a vague “point B” as coverage reaches one nominal layer. At 1.5 nominal layers the “point B” is gone, and beyond 1.5 layers there is no appreciable change in the isotherms. Comparison of the Isotherms on Three Solids.The three isotherms with no xenon and the limiting isotherms with many layers of xenon are compared in Fig. 7. Because of uncertainties inherent in
1.0
v,,, values for argon, it is possible that the limiting curves may be as much as 2Q% off in absolute value. However the shapes are different, arid they cannot be made to coincide by a change in the 13 scale.
4 CARBON BLACK
-
3 0:
2
1
0
0.2
0.4
0.6
0.8
PlPo. Fig. 6.-Comparison of the adsorption of nitrogen and argon on silver iodide at - 195’: -, nitrogen; ----, argon.
0 0
0.6 0.8 1.0 Pho. Fig. ’I.-Comparison of argon isotherms on carbon black, silver iodide and anatase at -195’.
0.2
0.4
April, 1954
BEHAVIOH. OF COLLOIDAL SILICATE SOLUTIONS
There are two explanations of the difference in the limiting curves that come to mind: first, that only a few layers can be stabilized on the surface of silver iodide or anatase; that thicker layers are unstable with respect to a bulk crystal and the thinner layer; second, that a structure different on the different solids is established in the xenon and is transmitted out, layer by layer, indefinitely. There are a number of difficulties that make the second alternative unlikely. The results on the graphiti~edcarbon black are in harmony with the generally accepted view of the falling off of van der Waals forces* from the (original) surface with the inverse third power of distance. Because of the non-specific nature of these forces, and the normal nature of the argon isotherms on silver iodide and anatase (bare of xenon) it is clear that a van der Waals field must extend from these surfaces also. I t is unlikely that, in both cases, the hypothetical thickening film of xenon alters in exactly the fight way to match this decaying field. The proportionate change affected by the xenon is greatest for carbon black and least for anatase. This fact is in harmony with the proposed order of maximum layers pre-adsorbed : carbon, over six, silver iodide two or three, and anatase less than two. Finally, on purely thermodynamic grounds, it is impossible that a film of different structure should be induced to grow indefinitely by a foreign body. The stability of the normal crystal favors a separation into a film of finite thickness and a bulk crys(8)
T.H.
Hill, J . Chem. Phya., 17, 590, 668 (1949).
335
tal, if it is impossible to make the transition from surface layer to bulk structure continuously. Conclusion.-The results, especially on carbon black, support the general picture of adsorption as layer formation under the influence of transmitted van der Waals forces.9 The difficulty in reproducing the isotherms, unless extreme care is taken, suggests that the ordinary preparation of adsorbents by chemical means is likely to create defects which cause heterogeneity, even in the absence of impurities. Calculations have been made of the adsorption energy of argon on alkali halide crystals,1° with the assumption that the ions behaved as rare gases. The isotherms on many layers of xenon reported here do not in the least resemble those on alkali halides, with which the calculations agree to a degree. Some doubt is thrown a t once on both the assumption and the calculations themselves. The technique of this paper provides a method of estimating the number of layers of an adsorbent that can be successfully preadsorbed. The impossibility of obtaining xenon films of indefinite thickness on some adsorbents suggests that a surface may be surrounded by a repulsive zone which will prevent the deposition of a solid from the gas phase. The looser structure of a li uid makes the continuous transition from adsor7 3ed film to bulk phase easier, but if the bulk phase is solid, 0 may approach a limiting value, rather than infinity, as p/po approaches unity. (9) G. D. Halsey, Jr., J . Am. Chem. Sac., 73, 2693 (1951). (10) W. J. C . Ow, Trans. Faradag Soc., 35, 1247 (1939); Pror. Roy Soc. (London), A173, 349 (1939).
THE BEHAVIOR OF COLLOIDAL SILICATE SOLUTIONS AS REVEALED BY ADSORPTION INDICATORS1 BY BENJAMIN CARROLL AND ELI FREEMAN Newark Colleges of Rutgers University, Newudc, N . J . Received November l o 3 1859
The adsorption and desorptioh of dyes, as reflected by spectral changes, were used to investigate the effects of pH, aging and ion exchange in colloidal solutions of sodium silicate. Aging effectswere indicated by a loss in the absorption affinity for the dyestuffs. The addition of salts caused the absorption s ectrum to shift hack toward that of the free dye in WatQr, indicating displacement of the dye by the salts, The nature of &e anion appeared to he of minor importance, whereas the cation was found to be the significant factor in t,he binding affinity of salt,s for the silicate sol. The displacement series obtained with the alkaline earth and alkali elements was Ca > Ba > Sr > K > Na > Li. The effect of less than one part per million of divalent cations could be observed by this method. The adsorption isotherm for crystal violet indicated the maximum number of dye molecules which may be bound per formula weight of sodium silicate. The free energy of binding was obtained on the basis that the binding is of a statistical nature.
In a previous publication2 it was shown that by the addition of a small quantity of suitable dye t o a solution of a hydrous oxide, such effects as salt additions and aging upon the colloidal solution could be followed spectrophotometrically in a quantitative manner. h n alternate procedure for this type of investigation is the use of equilibrium (1) This material is taken from the dissertation submitted by Eli Freeman in partial fulfillment of the requirements for the Master of Arts degree of Brooklyn College, New York. Presented in part before the Division of Colloid Chemistry of the American Chemioal Sooiety, Atlantic City, N. J., September 14, 19.52; Abstr. Papers Am. Chein. BOO., 122, 2G (1952). (2) B. Carroll and A. W. Thomas, J . Chem. Phus., 17, 1336 (1949).
dialysis. The latter method has the disadvantages of being time consuming and inexact for hydrous oxides. The inexactness is due in part t o the necessity of using foreign electrolyte to suppress the Donnan effect, thus introducing appreciable changes in the colloidal system. In this work a negative sol, sodium silicate, was investigated spectrophotometrically using the dyes crystal violet and methylene blue. These dyes were selected because it was found that they followed Beer’s law reasonably well, up to a concentration of 2 X 10-5 M . Furthermore, the dyes alone showed inappreciable spectral shifts due to