W. C. WALKER AND A. C. ZETTLEMOYER
182
shown in Table V, obtained for both nitrogen and argon are very close. The agreement in area based TABLE V PHYSICAL ADSORPTION OF GASESON REDUCED AND PARTIALLY OXIDIZEDMOLYBDENUM AT - 195' Reduction
1
GlU
Surface condition before adsorption
VDl,
ml./g.
A
R" 0.248 .253 p.0.4 A .232 P.O. 2 .235 P.O. 0 2 A .211 P.O. 3 A R .201 R ,198 Nz 13 A R .115 co R ,137 .I27 P.O. 0 2 co P.O. .127 A P.O. ,114 14 P.O. .125 0 2 A .112 P.O. co P.O. .124 R = reduced surface; P.O. partially oxidized surface. 0 2
ACTIVE MAGNESIA.
V.
Vol. 57
on Livingston's factors for the cross-sectional area these molecules is also good. Carbon monoxide, however, adsorbs much more strongly on the reduced surfaces than can be accounted for on the basis of either hexagonal close packing or with Livingston's empirical values for molecular areas. Neither nitrogen nor carbon monoxide were chemisorbed a t -195". Nitrogen was not chemisorbed on the reduced sample at 25". It is, of course, impossible to compare the value of the Vm's for oxygen on the reduced surface with those for argon, nitrogen or carbon monoxide because of the chemisorption of oxygen. However, adsorption on the reduced sample which has been previously exposed to oxygen at -195" until the chemisorption of oxygen was complete should furnish a basis for comparison. The values for Vm obtained for the physical adsorption of oxygen on these partially oxidized surfaces were higher than those obtained for argon, as shown in Table IV, and in exact agreement with those obtained for carbon monoxide. No explanation can be offered a t this time for the odd behavior of oxygen and carbon monoxide. Qf
d
ADSORPTION IN PREFERRED POSITIONS'
BY W. C. WALKER AND A. C. ZETTLEMOYER William H. Chandler Chemistry Laboratory, Lehigh University, Bethlehem, Penna. Received March d7, 1968
Physical adsorption of gases on solids is generally considered to take place with a close-packed first layer. Using this concept the area occupied per molecule has been calculated from the density of the normal condensed phase. Areas calculated accordingly from the adsorption of various substances on active magnesias were non-concordant. To explain these discrepancies the implications of adsorption in preferred positions were examined. From. the geometry of the 100 face of a magnesium oxide crystal distances between preferred positions were determined. Values for the area per adsorbed molecule could then be calculated as a function of these distances and the size of adsorbate molecules. These unique values for'the effective molecular areas are aimple multiples of a constant which is a characteristic of the crystal face. This relationship leads to the 'conclusion that, when adsorption takes place in preferred positions, the ratio of the moles of adsorbate A to the moles of adsorbate B required to complete a unimolecular layer on the surface can be expressed in terms of simple whole numbers. Data for the adsorption of a number of gases and for adsorption of iodine from solution on very pure magnesium oxide were found to be in agreement with this postulate.
The measurement of the surface areas of finely divided materials through the use of gas adsorption techniques has become a very useful tool in many fields. There has always been some uncertainty, however, with regard to the absolute values of the areas that are measured. The usual techniques for obtaining surface areas from adsorption data require information from independent sources as to the amount of area present per molecule adsorbed in the first layer. These molecular areas are generally estimated by the method proposed by Emmett and co-workers2 based on close-packing in the first layer similar to either the liquid or solid phase of the adsorbate at that temperature. On the basis of this assumption, the area per molecule has been calculated from the molecular weight of the adsorbate and the density of the condensed phase. Harkins and Jura3 have devised an absolute ( 1 ) Presented before the Division of Colloid Chemistry a t the 112th Meeting of the American Chemical Society in September, 1947. (2) Stephen Brunauer, "The Adsorption of Gases and Vapors," Princeton Univeralty Prena, Princeton, N. J., 1948, p. 287. (3) Wm, D, #arkina end Cleorpe Jura, J . Chsn. Phua,, 11, 481
wa).
calorimetric method for determining surface areas and have used it to determine the magnitude of the area covered by various adsorbed molecules on a number of solid surfaces. Measurements of this kind on two non-porous solids indicated tbat Emmett's area for the nitrogen molecule, 16.2 A.2, was reasonable. When molecular areas are determined by the method of Harkins and Jura, however, the solid must be strictly non-porous. Further support was given the concept of closepacking by Hill4 who showed on the basis of statistical mechanics that the transition from localized to mobile first layers of adsorbed molecules should in general take place at temperatures below the range generally employed for adsorption measurements . Thus, the concept of a mobile, close-packed first layer in physical adsorption has become ingrained in the thinking of workers in this field. Meanwhile, adsorption in preferred or even fixed positions has been accepted for chemisorption where (4) Terrell L. BLlI, (bid., 14, 441 (1946).
I
.
ACTIVE MAGNESIA ADSORPTION IN PREFERRED POSITIONS
Feh., 1953
definite bonds may be formed with units of the solid surface. A few authors, however, have applied adsorption in preferred positions to cases of physical adsorption. Orr6 has calculated the heats of adsorption of argon on crystal faces of potassium chloride and cesium iodide. He concluded that on cesium iodide the most stable configuration consists of a e o n molecules in preferred positions. On potassium chloride, on the other hand, Orr's calculations and measurements showed that a close-packed first layer should have the greater stability. Gaudin6 spoke on adsorption in preferred positions on materials such as galena but presented no experimental evidence for the use of this concept. Others including Beebe' have suggested that the spacings of the solid lattice probably influence the packing of adsorbate molecules. Livingston* has examined a considerable quantity of experimental data which have appeared in the literature for the adsorption of more than one gas on the same surface. He has presented a table of molecular areas based on these data and on an %bsolute area for the nitrogen molecule of 15.4 A.2. This table indicates that the liquid packing concept is too great a simplification and he has introduced the concept of packing factors to account for the discrepancies. I n our work with active magnesiasg the nitrogen areas 'were compared with data for the adsorption of iodine from carbon tetrachloride. The poor comparison obtained through the use of closepacked molecular areas is shown in the first three columns of Table I.
183
obtainable purity. This work was carried out and in the present paper the theory is set forth and the confirming experimental results are reported. Preferred Position Adsorption of Spherical Molecules.-Fortunately, magnesium oxide is a very simple case for such a study. Its crystal is a simple cubic lattice of doubly charged ions and the only face normally found is the 100 face shown as the background of Fig. 1. Here the large circles represent oxide ions while the small ones represent magnesium ions drawn to scale. Let us assume that a molecule being adsorbed on this surface will take some preferred position with respect to the ions in the crystal face. The adsorbed molecule need not be completely immobile, but a t least i t must spend most of its time in a preferred position on the solid surface. These preferred positions are of three types depending on their spacing. The first type is exemplified by positions directly ,over one of the ions. They are arroanged in a square array with an interval of 2.97 A. as shown in the upper left of Fig. 1. The second type is exemplified by positions between adjacent oxide ions. These points are also arganged in a square pattern, but the interval is 2.10 A. as indicated in the lower. part of the diagram. Type I11 positions, shown at the upper right, are conceivable but can accommodate only very small molecules such as hydrogen or helium and will be neglected hereafter.
TABLE I COMPARISON OF NITROGEN AND IODINEAREASOF ACTIVE MAGNESIAS Grade
XP 2624 2652-5 2652 2641 2661'/2 2665
--Close-paokedNnarea 1 2 area (16.2 b.8) (21.2 d.2)
210 151 149 131 79 35 35
134 96 93 83 48.5 19.8 15.9
Mole ratio,
Nn:In
2.05 2.06 2.09 2.06 2.12 1.64 2.88
Preferred positions Nn area Iz area (17.67A . 2 ) (36.33 A.1)
229 164 162 142 85.5 38.2 38.2
223 159 155 138 80.6 46.6 26.5
The iodine areas are only about two-thirds of the nitrogen areas, but in many cases the mole ratio of nitrogen to iodine in the first layer is strikingly close to two. This clue led to an examination of the possibility of adsorption in preferred positions. That approach brought these data into much better agreement as is indicated in the last two columns. Because of this initial success on relatively impure materials it was decided that the postulate of adsorption in preferred positions should be tested carefully using a magnesia of the greatest ( 5 ) W. J. C. Orr, Trans. Faraday Soc., 86, 1247 (1839). (6) A. M. Gaudin, et al., Am. Inst. Minino Met. Engr., Tech. Pnb. No. 2002 and 2005 (1946). (7) Ralph A. Beebe, J . A m . Chem. Soc., 67, 1554 (1945). (8) € H. I. Livingston, J . Colloid Sci., 4, 447 (1949). (9) A. C , ZhttlemoJter snd W,01Walker, I n d , Enq, Chsm,,(11 89,
69 (194T)b
Fig. 1.-Types
of adsorption positions available on the 100 crystal face of magnesium oxide.
I n the positioning of spherical molecules on a square array of preferred adsorption positions the maximum number of molecules which can be placed on the array will be dependent upon the molecular diameter and the interval distance of the array. I n this case all possible values for the number of positions per adsorbed molecule may be expressed 1) mathematically by the expression N 2 ( M 2 where N is a whole number equal to or greater than one, and M is a whole number equal to or greater
+
W. C. WALKERAND A. C. ZETTLEMOYER
184
VOl. 57
than zero. Table I1 shows all possible values up to fifty.
Multiple Proportions. "If two or more gases adsorb in preferred positions on a crystalline surface, then the ratio of their Vm's can be expressed TABLEI1 in terms of simple whole numbers." POSSIBLE VALUESFOR THE NUMBER OF POSITIONS PER Preferred Position Adsorption of Diatomic MoleMOLECULE FOR SPHERICAL MOLECULES ON A SQUARE ARRAY cules.-Another simple type of molecule that can N M O 1 2 3 4 5 6 7 be handled readily in the development of the theory 1 1 2 5 10 17 26 37 50 of adsorption in preferred positions is the diatomic 2 4 8 20 40. molecule. For simplicity let us consider only the 45 3 9 18 case where the bond distance in the adsorbate mole4 16 32 N * ( W + 1) cule is very close to the distance between preferred 5 25 50 N 2 1 positions and the two atoms of the adsorbate moleG 36 111 2 0 tule tend to take two adjacent similar positions. 7 49 For this case there are again several arrangeFrom the numbers in Table I1 and the spacings ments that would be taken depending upon the from Fig. 1, the possible effective molecular areas of radius of the atoms of the adsorbate and upon the simple spherical molecules adsorbed in these interatomic distance. The most important of these preferred positions may be readily calculated and arrangements are depicted in Fig. 2. These arrangements result in two, three, four or six positions are listed in Table 111. being occupied per molecule. The possible effecTABLEI11 tive areas per molecule based on these concepts EFFECTIVE AREASOF SPHERICAL MOLECULES ADSORBEDIN are tabulated in Table IV. PREFERRED POSITIONS O N MAGNESIUM OXIDE '
Positions per molecule
1
2 4 5 8 9 10
Type I positions (2.97. A.) Molecular Effective ' diameter, area,
A.
R. 2
0 -2.97 2.974.20 4.20-5.94 5.94-6.65 6.65-8.40 8.40-8.91 8.91-9.40
8.83 17.67 35.33 44.16 70.66 79.49 88.33
T pe I1 positiona (2.10 A,) &oleaular Effective diameter, area,
A.
A.3
-2.10 2.10-2.97 2.974.20 4.204.70 4.70-5.94 5.94-6.30 6.30-6.65 6.65-8.40 8.40-8.66 8.66-8.91 8.91-9.40
4.42 8.83 17.67 22.08 35.33 39.75 44.16 70.66 75.08 79.49 88.33
0
lG 17 18 20
This table contains all possible values within reason for molecular areas of adsorbates of round cross-section on the 100 crystal face of magnesium oxide when adsorption occurs in preferred positions. This analysis of adsorption in preferred positions leads to a simple law which resembles the Law of
NO CROWD I NG
END CROWDING
2
3
W
SI DE CROWD1N G
4
W END AND S I D E C R O W D l NC
END, S I D E , A N D DIAGONAL C ROWDl NG R
Fig. 2.-Simplest arrangementn of diatomia molecules on a mquare srrby of adsorption podtionr.
TABLE IV EFFECTIVE AREAS OF DIATOMIC MOLECULES ADSORBED PREFERRED POSITIONS ON MAGNESIUM OXIDE Positions per molecule
Type 1 (2.97
Type I1 (2.10 A.)
9 3 4 6
17.67 26.50 35.33 53.00
8.83 13.25 17.67 26.50
The applicability of this concept to a magnesium oxide surface was tested by adsorbing a variety of different gases on a very pure magnesia sample and determining which system of assigning molecular areas most nearly gave the same area for all adsorbates. Special care was used in the preparation of the magnesium oxide sample to eliminate many of the usual impurities such as CaO, S O z , AlnOs, carbonates, chlorides and sulfates. The elimination of impurities is particularly important for this work since there is evidence that some of them have a tendency to gather on the surface of the magnesia crystallites. The sample was prepared from crystals of carbothermic magnesium which was reported to have a purity of at least 99.99%. The metal was treated with dry methanol to produce the methylate. The methylate was then hydrolyzed in boiling distilled water for four hours. (It was found that the hydrolysis was quite slow and that this relatively severe treatment was required to drive it to completion.) The magnesium hydroxide thus obtained was dehydrated in two steps. First the majority of the water was removed by heating in a muffle furnace a t 600" for one hour. The material was then placed in the sample tube of the BET adsorption apparatus and degassed a t 490" for 16 hours to a final pressure of less than mm. After this treatment the magnesia had attained a constant weight and an area of approximately 200 sq. m. per gram. The volumetric adsorption apparatus used and the technique for its operation have been previously described.g
*
e
Feb., 1953
ACTIVEMAGNESIA ADSORPTION IN .PREFERRED POSITIONS
185
areas listed are consideraly below 100 and show an average deviation of 14.3%. Thes: figures indicate that the molecular area of 16.2 A.2for nitrogen or the other molecular areas obtained on the basis of liquid packing are certainly not valid for adsorption on magnesium oxide a t - 195". The fifth column contains similar figures calculated on the basis that nitrogen at -195" shows a solid close-packing. Here the agreement is much better than in the previous case with an average deviation of only 6.2%. Although this hypothesis fits most of the data very well, there TABLE V are two serious objections to it. First of all diffiR I J V M ~ R Y O F EXPERIMEYTAL EVIDENCE FOR PREFERREDculty arises from the lack of agreement with the 1'ORITION A D S O R P T I O N O N M A G N E S I U M O X I D E well established iodine data. The relationship Close-packed between iodine and nitrogen adsorption has been Total arean Preferred position Area per NZ N1 Area per Total established many times with a variety of magnesia Temp., molecule, 16.2 14 0 ~uolecrile, samples so that a satisfactory theory should be Adsorbate 'C. 1.1 .&.2 .&.a .&. 2 A. 1 compatible with it. The second difficulty lies in -195 1'6.20r14.0100.0 100.0 17.67 100.0 Ns A 14.0 87.9 101.7 17.67 101.i -195 the implication that adsorbed nitrogen exists in the 14.4 88.8 102.7 17.67 99.9 A -183 solid state fourteen degrees above the normal 17.67 102,9 14.1 89.5 103 6 On -183 freezing point of the bulk substance. This situa17.0 85.0 98.3 22.08 101.2 -183 Na tion is quite a t odds with the results of several 85.7 99.2 22.08 102.1 coz - 77 17.0 9 0 . 7 102.2 22.08 102.2 CHaOH 0 18.0 workersz~10 mho showed that the freezing points of 85.6 99.0 26.50 105.0 CnHn - 80 19.8 adsorbates are normally much lower than those of 72.5 84.5 35.33 101.5 In (from CC1,) 25 23.2 the corresponding bulk materials. This matter is 35.33 110.2 129.0 a bit nebulous, however, since it is difficult to conCO? (ohemi.) 25 35.33 86.3 100.0 44.16 99.2 46.71 105.0 ceive of the presence of a true normal bulk phase at Average deviation 14.3 6.Zb 2.2 the low coverages encountered here. Area = 100 Vmx2 X / V ~ NI:ZNI. !Av. dev. is 1.6 wit'hIn a similar manner, the last two columns show out iodine values. the result,s obtained when the concept of adsorpI n addition to the measurements of the physical tion in preferred positions is applied to the data. adsorption of gases at low temperatures the chemi- An examination of these area figures shows that sorption of COZ and the adsorption of iodine from there is more general agreement when the data solution were run a t room temperature. Thus, are analyzed according to the concept of adsorption conclusions can be based on a variety of experi- ' in preferred positions than there was in either of the previous ttwo cases. This correlation is based on mental data. the selection of effective molecular area values from Discussion of the Results Tables I11 and IV. Some of these selections require The third, fourth and fifth columns of Table V discussion. show the results obtained when the data were The data indicate that the effective area of the analyzed by use of the concept of close-packing in nitrogen molecule &as increased from 17.67 the first layer. In the third column are tabulated a t - 195" to 22.08 A.2a t - 183'. Kormal thermal the molecular areas calculated according to the expansion of liquid nitrogen over this range would formula proposed by Emmett and Brunauer.2 be expected to produce an increase of only about In the case of iodine where chemisorption may have 5%. The concept of adsorption in preferred positaken place, a preferred position area is also con- tions suggests that in this temperature range the sidered. Since chemisorbed carbon dioxide would effective diameter of the nitrogeno molecule has be very widely considered to be in fixed positions passed the critical value of 4.20 A. When this rather than close-packed, a preferred position area point is passed the arrangement of adsorbed is used for it here. nitrogen molecules on the crystalline surface must I n the fourth column are shown the relative change abruptly. This matter is a fundamentally areas calculated from the experimental V,'s important one that can probably be elucidated by assuming thoat the nitrogen molecule occupies an further experimental work. It would be desirable area of 16.2 A.2and that the other molecules occupy to determine whether this change in the packing of the areas shown in column 3. The area for the adsorbed nitrogen on magnesium oxide is an nitrogen adsorption data a t -195" has been arbi- abrupt one a t some sharply defined transition trarily assigned the value of one hundred units. temperature or a relatively gradual one. It would When the correct molecular area assignments have also be important to known also whether or not this been made, all adsorbates should give relative phenomenon occurs on other surfaces at the same areas of one hundred units. I n other words, the temperature. adsorbent should have a surface area that is inThe adsorption of iodine from carbon tetradependent of the adsorbate used. chloride solution can be considered to take place An examination of the figures in column four in either of two ways. If it is adsorbed as atomic shows very poor agreement between the results (10) W. A. Patriok and W. A. Kemper, THIS.JOURNAL, 42,;369 obtained with the various gases. Almost all of the (1938). The first two columns of Table V show the adsorbate gases that were used and the temperatures a t which they were adsorbed. In each case nitrogen a t - 195" was used as a reference standard and an attempt was made to work with as many gases as could be readily obtained and handled. Measurements were also made with krypton a t -153" and ammonia a t -30", but in these two cases non-linear BET plots were obtained which could not be analyzed to give a precise determination of V , for these gases.
0
186
F. H. HEALRY, J. M. FETSRO AND A. C. Z:~TTJ,EMOYIZR
Vol. A7
iodine, the iodine aioms could be expected to cover theory of adsorption in preferred positions. Part an area of 17.67 A.2 each. On the other hand, of this discrepancy may be attribute’d to experiiodine may be adsorbed as diatomic molecules in mental errors, but it should be pointed out that which case the arrangement may be that depicted even with adsorption measurements of perfect in the lower left of Fig. 2. In either case the accuracy the calculated areas should show some molecular area is 35.33 8.,2in agreement with the scatter from the ideal value of 100 due to edge effects. Perfect agreement with the theory could experimental data. be expected only if the magnesium oxide crystal For the chemisorptios of carbon dioxide the faces were of infinite extent and without impermolecular area of 44.16 A.2 might be selected from fection. Actually, since the cry$allite faces of Table I11 since it gives a relative area of 99.2. This value cannot reasonably be selected on the material used were only about 80 A. on a side, the basis of the known dimensions of the carbon dioxide edges and corners might well produce a measurable distortion of the results. Thus the figures in the molecule, however. A special mechanism has been postulated for the last column of Table V indicate as good an agreechemisorption of carbon dioxide on magnesium ment with the theory as can be expected. oxide which gives a molecular area of 46.71 Conclusions and is also included in Table V where it shows fair The data presented here are not extensive enough agreement with the data. This mechanism is based on the formation of a carbonate ion involving to establish conclusively that adsorption on magone of the oxide ions in the magnesium oxide nesium oxide surfaces does take place in preferred crystal face and with the other two oxygens of the positions. They are explained much more satiscarbonate free to rotate. It is further based on the factorily on this basis, however, than they are by chemisorption being essentially irreversible under any other approach suggested to date. Thus the experimental conditions so that once a molecule there is considerable evidence that adsorption on is chemisorbed it stays in position. Thus, once a magnesium oxide surfaces takes place in preferred molecule has been chemisorbed, all positions where positions. How widespread this phenomenon may be is the adsorption of another molecule would produce interference of rotation are effectively blocked. difficult to determine since the surface geometry of The application of the concepts of probability to m0s.t solids is muoh too complex to be handled by the approach used’here. Also the doubly charged this gives the molecular area mentioned above. The molecular areas tabulated by Livingston8 ion structure of magnesium oxide may make it have also been applied to these data. For the nearly unique in this respect. cases where he has values, the agreement is conAcknowledgment.-The authors wish to acknowsiderably better than that for liquid close-packing ledge the support of the Westvaco Chemical Divibut not as good as for adsorption in preferred sion of the Food Machinery and Chemical Corporapositions. This result might be expected since tion in the early phases of this work and of the these values are not based on any theoretical Lehigh University Institute of Research in the conconcept but are averages of relative experimental cluding phases. They also wish to acknowledge the results reported in the IiteEature. assistance of William E. Kelly, Richard J. MikovIt will be noted that the figures in the last column sky, and the General Aniline and Film Corporation of Table V do not show perfect agreement with the in obtaining some of the adsorption data.
PHYSICAL ADSORPTION ON CADMIUM OXIDE1 BY F. H. HEALEY, J. M. FETSKO AND A. C. ZETTLEMOYER William H . Chandler Chemistry Laboratory, Lehigh University, Bethlehem, Penna. Received March 67, 1062
Cubic crystallites of cadmium oxide were prepared by burning spectroscopically pure cadmium vapor. Adsorption isotherms were obtained with nitrogen, oxygen, carbon monoxide and argon at - 195 and - 183’. A comparison of calculated V , values showed no evidence of the adsorption in preferred position observed by Walker and Zettlemoyer in the case of cubic magnesium oxide. An explanation of the observed difference in the adsorptive properties of ,cadmium and magnesium oxides is offered on the basis of cation polarizabilities.
The state of gases physically adsorbed on solids is usually assumed to be non-localized a t the temperatures ordinarily used.2n3 Due to the lack of constraint on the two-dimensional motion of the adsorbed molecules, the use of molecular areas calculated from liquid densities has often been found satisfactory. However, in the case of a
strongly inhomogeneous surface field, it would not be surprising to find that the motion of the adsorbed molecules was restricted and that an approach was made toward localized adsorption. Evidence pointing toward the latter type of adsorption has been found by Orr4 on CsI and by Walker and Zettlemoyer6 on MgO. Both of these sub-
(1) Presented before the Colloid Division of the American Chemical Society, Houston, Texas, in March, 1950. (2) 8. Brunauer, “The Adsorption of Gases and Vapors,” Prinaeton University Prens, Princeton, N. J., 1048, pp. 448-461. (8) T,L, Hill, J . Chrm. Phua., 14, 441 (l946),
(4) W. J. C. Orr, Trans. Farada?, ~ o c . 36,1247 , (1939). (6) W. C. Walker and A. C. Zettlemoyer, presented before the Division of Colloid Chemistry at the 113th Meeting of the American Cbemiorl looiety.
. r.