Feb., 1953
PHYSICAL
ADSORPTION ON SINGLE CRYSTAL ZINC SURFACES
143
PHYSICAL ADSORPTION ON SINGLE CRYSTAL ZINC SURFACES' BY T. N. RHO DIN^ Instilute jor the Study of Metals, University of Chicago, Chicago, Illinois Received June 9. lQ52
The physical adsorption of very small quantities ( 9,) of nitrogen and of argon by the use of a modified quarts vacuuni microbalance has been achieved. The technique permitted the characterization of very small surfaces in terms of surface areas and isosteric heats of adsorption. The adsorbents were single crystal surfaces of copper and of zinc with which considerable effort was made to evaluate surface structure and minimize contamination prior to the adsorption etudies. Surface area values were obtained by the Brunauer-Emmett-Teller method and differential heats of adsorption were calculated by the Joyner-Emmett method. Sigmoidal adsorption isotherms characteristic of type I1 isotherms with no indication of stepwise adsorption within the experimental error were observed. The heats of adsorption were found to vary in a unique manner with the coverage, uniformity and crystallography of the adsorbent and the nature of the adsorbate. Although real distinctions were observed in the heat of adsorption versus surface coverage plots for the various gas-surface combinations a definite maximum in the heat of adsorption was observed in each case around monolayer coverage of the single crystal surfaces.
Introduction With the development of nucleatioii theory and surface energy studies the physical interpretatioii of interface phenomena has assumed new import a n ~ e . ~ - 5The importance of analyzing surface effects in terms of an interface with unique and variant properties rather than merely an inert solid subTo cite a specific strate has been case it has been shown that the physical adsorption of nitrogen is anisotropic on single crystal copper surface^.^ It has also been demonstrated that the study of small single crystal metal surfaces can furnish considerable insight into surface phenomena especially when the preparation of such surfaces is achieved in collaboration with and using the techniques of the physical metallurgists.1° To develop our understanding of the nature of the anisotropy of physical adsorption on single crystal surfaces, adsorption studies on copper have been extended to nitrogen and argon adsorption on single crystal zinc surfaces. It is intended to illustrate the differences in adsorption between a monatomic and a diatomic gas on each of two metal surfaces with the same crystallography but with different atomic spacings and hence different surface electroil configurations. The investigation of simplified gas-solid systems of this type must precede successful efforts to resolve more complicated surface phenomena. By the same token physical adsorption studies are facilitated over chemisorption studies by the relative simplicity of the former although it is the latter type of adsorption which is more critically associated with corrosion and catalytic phenomena.ll Experimental Materials.-Considerable care was taken in the iuetallographic preparation of the surfaces to give a sample of well ( I ) This research was supported in part by Army Air Force Contract AF-33 (038)-8534. (2) Engineering Research Laboratory, Engineering Department, E. I. du Pont de Nemours & Co., Wilmington. Delaware. (3) W. D. Harkins and S. Loeser, J . Chem. P h y s . , 18 (4), 556 (1950). (4) W. D . Harkina, Science, 102, 292 (1945); T. L. Hill, ibid., 17, 520 (1948); ibid., 18, 246 (1950). ' (5) T. N. Rhodin, Jr., J . Applied Plws., 21, (IO), 971 (1950). (6) A. L. McClellan and N. Hackerman, THISJOURNAL, 66, (3) 374 (1951). (7) D . Turnbull, J . Applied Phys.. 2 1 ( l o ) , 1022 (1960). (8) D. Turnbull and R. E. Cech, ibid.. 21 (81,804 (1950). (9) T. N. Rhodin, Jr., J . A m . Chem. Soc., 7 2 , 5102 (1950): 73,.3143 (1951). (10) T. N. Rhodin, Jr.. Faradall Sot. Dzseussions, ( 5 ) 215 (1949). ( 1 1 ) G. D . Halsey, Jr., d i d . , (S), 54 (1960).
characterized surface structure. This in turn facilitates interpretation of the adsorption phenomena. The highest purity commercial zinc was repurified by repeated vacuum casting (2-4 cycles) for 24-hour periods a t nim. and 430'. Spectrographic analysis indicated 0.002% lead, 0.0002% copper and 0.000370 iron residual impurities. Single crystal rods (1 X 15 cm.) were grown using the Bridgman technique by slow movement (0.1 mm./min.) through a vacuum gradient furnace of maximum furnace temperature 430" and 10-6 mm. pressure. The rods were coated with a very thin film of a urea-formaldehyde resin upon removal from the furnace and then cooled to liquid nitrogen temperature and cleaved along the basal plane. For cleavage it was sufficient to form a very small needle point depression on the rod surfaces and induce cleavage by a series of cooling-heating cycles ( -195 to 25') without the application of additional mechanical stress. This technique was remarkably effective in minimizing surface twinning and in the prevention of kink and band formation. The original crystal plates (1 X 1 X 0.1 cm.) were electropolished in a nitric acid-methyl alcohol mixture (1:4) over a period of hours at 0" with a current density of 1 centiampere/cm.l to a final thickness of about 0.03 cm. The final crystal plates showed negligible edge attack, and no twinning lines were observable from microscopic examination. A group of these plates formed a typical adsorbent specimen weighing about 1 g. and possessing an apparent surface area of about 10 crn.l. Test samples were characterized by X-ray and electron diffraction and by electron microscopy to indicate the effectiveness of the preparation from the viewpoint of creating a planar undistorted surface. These methods have been described elsewhere a t some length and will not be repeated here.12*13 It will be indicated later in this paper that the roughness factor of the surfaces indicated by adsorption studies was in good agreement with the supposition that the surfaces actually did possess fine-scale planarity. The rigor with which the vacuum system was outgassed, baked out and gettered is indicated by the fact that the mm. closed-off system decayed from a vacuum of loh7to in 20 minutes and to 10-5 mm. over a period of weeks.14 The gases nitrogen, argon and hydrogen were carefully purified and dried, finishing up with activated charcoal traps a t liquid nitrogen temperatures. Specially activated reduced copper was used to remove all traces of oxygen. Procedures.-Because of the very small actual surface area of the adsorbent it was particularly important to preserve the system and hence the specimen surface free of contamination durin the period of cooling from the reduction temperature to t i e adsorption temperature to ensure the success of precise reversible measurements. The samples were treated to an annealing and firepolishing treatment in situ in addition to the more general precautions previously described in this and other p a p e r ~ . ~ ' ' ~ J 6By J ~ maintaining a (12) A. Raether, O p t i k , 1, 69 (1946). ( l a ) T. N. Rhodia, Jr., J . A m . Chern. SOC.,72, 4343 (1950). (14) Contrary to popular conception even the so-called vacuum of l o - ' mm. of oxygen pressure contains a sufficient number of oxygen inolecules to cover an appreciable portion of the surfare with a monolayer a t room temperature in 15 minutes calculated from the collision frequency of gas molecules with surface. (15) K. F@rland, Tids Kjemt Berovessdn Met., 10, 250 (1950). (1G) J. H. Moore, Australastan Engr., p. 65, Oct. 1950.
T.N. RHODIN
144
hydrogen pressure of 3 atmospheres at 350" for 100 hours the zinc oxide surface was reduced without. significant evaporation of zinc. The water vapor formed in the reduction was condensed out in an adjacent trap a t liquid nitrogen temperatures. The freshly reduced zinc surface was fireylished under these conditions to a mirror bright finish. he manner in which zinc vapor from an independent source was exploited after the reduction to scavenge the residual oxygen present in the reaction volume without contaminating the specimen surfaces with spurious zinc vapor is best illustrated by Fig. 1. It shows the arrangement of the sample, dummy sample, shielding and heating elements around the sample. I t was found that the sample indicated no observable weight pick-up for 15 minutes when a relatively large area of evaporated active zinc surface was produced by condensation of vapor on all the reactor walls. The directionality of the evaporated metal was exploited using suitable shielding to ensure that no measurable amount of the evaporated zinc condensed on the sample surface to cast ambiguity on the known surface structure. This type of application illustrates one manner in which vacuum microgravimetric techniques can be used to prepare and test surface conditions.
VOl. 57
7
2 1 (c l)z V(1 - 2 ) vmc+ where 2 is the relative pressure at which volume V(S.T.P.) of gas is adsorbed, V , is the volume of gas required to form a monolayer, and c is a constant which together with V, can be determined by substituting the experimental values of V and of z in eq. (1). Actually in this study the adsorbed gas is weighed and hence Vm is expressed in micrograms. Using the most likely value for the cross-sectional area of the nitrogen molecule (16.1 the surface areas calculated in this manner were found to check rewonably well with a Harkins-JuraZa analysis of the same data. -3-
I
Q.'
Pressure (crn. 1. Fig. 2.-The adsorption of nitrogen on the (lOT1) single crystal face of zinc a t 78.1, 83.5 and 89.2"K.: 0, first run; 0 , second run; 0 , desorption points. The precision of the weight measurements is indicated by the scatter of the points in runs 1 and 2. The precision of the adsorption measurements in the low pressure region is indicated in Fig. 4.
U Fig. 1.-Adsorption vessel containing metal plates: 1, refrigerant; 2, dewar; 3, quartz tube; 4, thermocouple lead; 5, heater lead; 6, condensed film of zinc; 7, heater for shield; 8, molybdenum shield- 9 dummy plate; 10, sample plate; 11, zinc evaporator; i2, heater lead.
The adsorption apparatus itself was a standard version of the differential adsorption microgravimetric technique developed by the author and described in some detail in other communications.17 Taking readings with care it was possible to observe weight changes to 10-7 g. with a precision of f 0 . 2 X lo-' g. This weight change corresponds to about 0.01 microliter (S.T.P.) of nitrogen adsorbate. I t may be noted that the edge effect of thin crystals contributes less than 0.5% of the total adsorption and was accordingly neglected.
Results All surface area values were obtained by a linear plot of the adsorption data according to the Brunauer-EmmettTeller18-19e20 equation (17) T. N. Rhodin, Jr., J . Am. Chem. SOC.,72, 5691 (1950). (18) P. H. Emmett, S. J. Brunauer and E:. Teller, ibid., 60, 309 (1938). (19) W. C. Mchlillan and E. Teller, THIS JOURNAL, 66, 17 (1051). (20) T. L. Hill, J . Chem. Phys.. 17, 106 (1949)
Y 0
I
.2
I
.4 p/p,.
I .6
I
B
I
1.0
Fig. 3.-The variation of nitrogen adsorption on single crystal zinc surfaces with relative pressure: 0, 78.1"R.; 0 , 83.5OK.; a, 89.2OK. (21) P. H. Emmett, "Advances in Catalysis," Academic Press, New York, N. Y . , 1948, p. 65. (22) W. D. Harkina and C . Jura, J . A m . Chem. Soc., 6 6 , 1366 (1944); 66, 1362 (1944). (23) L. G. Joyner and P. H. Emmett, i b i d . , 70, 2353 (1948); 70, 2359 (1948).
PHYSICAL ADSORPTION ON SINGLE CRYSTAL ZINC SURFACES
Feb., 1953
145
temperatures Taand TI was then substituted in the following integrated form of Eq. (2) TITz Pa H = 2.303 R log P, (3)
1.4
(m)
I.2 -'
The maximum variation in the heat values and the surface coverage values of Figs. 4 and 8 is indicated so as to avoid making the data look yJ 1.0 better than they really are. Adsorptions were 0 measured and compared at pressures from 0.01 E to 150 em. at each of three temperatures in thc .-c0 p .e range 78.1-89.2"K. for each of two samples of 0 a given crystal face. The variation of heat of U adsorption with coverage of the surface could a be determined and plotted in typical heat ( H ) E 6 us. coverage (0) curves. 0 The boundary conditions for application .-c0 of the Clausius-Clapeyron equation and the thermodynamic considerations which limit its = .4 interpretation in terms of isosteric heats of physical adsorption have been considered at great length.24 Some of these treatments have .2 been referred to previously8 and the conclusions drawn that calculation of the heat data on a classical basis may be the most useful approach at present. The absorption isotherms were all measured .I .2 .5 I 2 2o 50 loo 2oo on one group of sample plates and then repeated Pressure ( c m .l)o on another set prepared and treated in the same Fig. 4.-Nitrogen adsorption isotherms from Fig. 2 plotted against manner. Hence comparison between runs 1 logarithm of nitrogen pressure. Isosteric heat of adsorption for a given and 2 indicates not only the precision of the adsorption is proportional to the ratio of the abscissa a t two temperatures microweighings but the degree of reproducibilafter the method of Joyner and Emmett: 0 , first run; a, second run; ity to which the sample surfaces were prepared. 0 , desorption points. The sigmoidal shape characteristic of type I1 isotherms with no evidence of stepwise adThe isosteric heat of adsorption ( H ) , was determined in sorption is illustrated in Figs. 3 and 7 where adsorption is the usual manner by comparing the pressures of gas (P) plotted against the relative pressure. Typical isotherms for in equilibrium with a given weight of gas ( a ) adsorbed on the adsorption runs on the two different set's of single crystal plates at 78.1, 83.5 and 89.2"K. are plotted in Fig. 2 for surface at three different temperatures ( T) nitrogen and Fig. 6 for argon. I t was from the pressure H (2 1 R 6
'I,
I
L
Actually the treatment of Joyner and Emmetta3was used to calculate the values of ( H ) for different amounts of adsorbed gas (a). The isotherms of Figs. 2 and 6 were replotted on a large scale with the logarithm of the pressure ( P ! as abscissa against the amount of adsorption (a) as ordinate. The ratio of the horizontal distances corresponding to the two pressures, P, and PI, for a given adsorption ( A ) at two
1.5
-*
d
1.3
2
Yo
- 1.1
Y
C
.o c .9 n L
0 v)
a .7
.u
c . 0
Y
a
.5
.3 2
4
6
8
1.0
1.2
1.4
1.6
1.8
.I
20
Surface Coveraqe ( V I V m ) .
Fig. crystal 5.-Thezincisosteric of nitrogen adsorption on single surfacesheats calculated from Fig. 4. The surface coverage wa8 calculated from the ratio of the actual adsorption to that required to form a monolayer determined in the classical manner from a typical Brunauer-EmmettTeller Plot. The heabcoverage Plot is a composite curve calculated from the adsorption isotherms pairs, 78.1-83.5, 78.1-89.2 and 83.5-89.2"K. The value of V , in each case was the average one determined for each pak of isotherms. The size of the crosses reflect the experimental uncertainty in the adsorption value, the monolayer value and the corresponding adsorption pressure.
IO,
30
50
70
90
Pr e ssu re (c m. I. ~ i 6,-The ~ . adsorption f, argon on the ( l o i l ) single crystal face of zinc at 78.1, 83.5 and 8 9 . 2 0 ~ ~0:,first run; a, second run; @, desorption points, The precision of the weight measurements is indicated by the scatter of the pointe in runs 1 and 2 except for the low pressure region which is better indicated in the logarithmic plot of adsorption in ~ i 8.~ . (24) J.
F. Duncan. Trans. Fardaay SOC.,46, 879 (1040)
VOl. 57
T. N, I~HODIN
148
TABLE I CHARACTERISTIC D A T A
FOR THE
AIMORPTIONO F NITROGENA N D ARGONON AND
(1) Adsorbate
(2) Adsorption temp., OK.
(3) of no. metal'" (4) Ratio atoms t o no. gas a t o m / u n i t area' Copper Zinc
THE
OCTAHEDRAL CRYSTAL FACES
O F COPPER
ZINC SURFACES (7) (8) Rouglinessc factor Copper Zinc
(6) (6) Isotericb heat ads. unit coverage, cal./mole Copper Zinc
(9) (10) Weight monolayerd X 108 g./cm.l Copper Zinc
Nitrogen 78.1 3.3 3.0 3100 2600 1.20 1.17 3.41 3.32 Nitrogen ' 83.5 3.3 3.1 3250 2670 1.20 1.15 3.39 3.28 Nitrogen 89.2 3.3 3.0 3000 2620 1.20 1.16 3.44 3.18 Argon 78.1 2.9 2.6 3200 2690 1.18 1.14 5.61 5.19 Argon 83.5 2.8 2.6 3110 2740 1.18 1.11 5.65 5.22 Argon 89.2 2.9 2.6 3000 2600 1.16 1.13 5.53 5.15 a This ratio was determined from the number of metal atoms/unit area from crystallographic data for the (lOT1) crystal face of zinc to the number of adsorbate molecules for the same area from adsor tion data using the cross-sectional areas previously listed. * The maximum isosteric differentialheat of adsorption interpoyated from the heat coverage plots. The rou hness factor is defined as the ratio of the experimental values of the wei ht of adsorbate required for monolayer coverage to &at calculated for the same coverage of a geometrically planar surface. can be interpreted as the ratio of the real area to the apparent area of the adsorbent. The weight of adsorbate required for a monolayer coverage of the adsorbent a t a given adsor tion temperature calculated from a Brunauer-Emmett-Teller plot and corrected for the variation of adsorbate density wit! temperature according to Emmett and Joyner. 23
It
values interpolated from these curves that the isosteric differential heats of adsorption were calculated. The precision of the measurements is also indicated clearly in these plots by the scatter of the experimental points. To facilitate application of eq. (3) and to improve the precision of interpolation in the low pressure region the data of Figs. 2 and 6 were replotted against the logarithm of the gas pressure according to the method used by Emmett and Joyner in Figs. 4 and 8 for nitrogen and argon, respectively. The experimental precision of pressure measurement was such as to permit a still larger scale plotting of adsorption value in the low pressure range where eq. (3) is very sensitive to errors in the pressure values. These plots as well as the conventional Brunauer-Emmett-Teller plots of x / v ( 1 2) us. x have been omitted from this paper in the interests of brevity The isosteric differential heats of adsorption as a function oi surface coverage on the single crystal basal plane faces of zinc calculated from Figs. 2 and 4 for nitrogen and Figs. 6 and 8 for argon using eq. (3) are plotted in Figs. 5 and 9, respectively. The surface coverage was calculated from the
-
.
Relo t ive Pressure. Fig. 7.-The variation of argon adsorption on single crystal zinc surfaces with relative pressure: 0 , 78.1'K.; 0, 83.5"K.; (3, 89.2"K.
ratio of the actual adsorption to that required for the formation of a statistical monolayer determined in what has now become the classical BET plot. The heat coverage plots are actually composite curves calculated from the three adsorption pairs: 78.1-83.5, 78.1-89.2 and 83.5-89.2'K. The value of V, used in each pair was the average one in each case. The size of the crosses reflects the experimental uncertainties in the adsorption and pressure values. The variation 'of the monolayer adsorption values and the heat values with tem erature are indicated in Table I in which all available &ta characteristic of the adsorption of nitrogen and argon on the octahedral crystal faces of copper and zinc are summarized. The adsorption data of argon on copper are included because it is SO pertinent to the over-all comparison.
Discussion It has already been demonstrated that the heats of adsorption of nitrogen on homogeneous single crystal copper surfaces vary in a unique manner with the extent of coverage, uniformity and crystallography of the surface^.^ The physical significance of these studies has been evaluated qualitatively within the frame of established theoretical understanding of adsorption phenomena on single crystal metal surfaces. The approximations and assumptions inherent in such treatments continue to present a serious limitation to the development of this type of study. Consideration of this study of zinc surfaces must, therefore, for the present, limit itself to a similar perspective with the interesting additional conclusions that can be made from the significant comparison presented by the four adsorption systems of this study. It is a most striking characteristic of all such adsorption systems that the heat values at low surface coverages are low and that all the heat values pass through a smooth but definite maximum close to, but not necessarily at, monolayer coverage.25 At higher coverage the heat values approach that of the heat of liquefaction of the gas in all cases. This maximum has been interpreted in terms of attractive interactions between the adsorbate molecules that tend to dominate the energetics of adsorption around monolayer c ~ v e r a g e . ~Why this contribution is so sensitive to the crystallography of the substrate especially at monolayer coverage remains somewhat obscure. It is also true that (25) G . Jura and D. Criddle, T n ~ JOURNAL, s 66, (2) 163 (1051).
E’eb., 1053
PHYSICAL ADSOI~PTION ON SINGLE CHYSTAL ZINC SURFACES
147
packed .spheres (1.860 instead of 1.633).%‘j Like the octahedral crystal face of copper each zinc atom has six closest neighbors in 1.2 the basal plane except that the lower layer in the case of copper is much closer. Although the atomic packing and electron configurar 1.0 tion Characteristic of the basal planes of the .-“ two metals has considerable effect on the physical adsorption, it is probably significant P Q .8 that the influence of second layer nearest E neighbors is also more important in determin.I ing the force field characteristic of copper $ .6 surfaces than those of octahedral zinc sur0 a faces. In addition the periodicity of the C force fields of the two surfaces differs in that m .4 zinc has a more “open” surface structure a than copper. This concept may be elaborated as follows. On passing from copper to .2 zinc by the addition of a second s-electron to the fourth shell it is observed that the electron clouds associated with the surface 0 .I .2 .5 I 2 5 IO 20 50 100 ions contract rapidly. The same effect holds Pressure fcrn.). for the analogous transitions from silver t o Fig. 8.-Argon adsorption isotherms from Fig. G plotted against cadmium and from gold to m e r c ~ r y . ~ ’This logarithm of argon pressure to facilitate calculation of isosteric heats is indicated for all three pairs from crystalof adsorntion according to the method of Jovner and Emmett: 0,first lographic data on the distance of closest run; .,Lsecond run; 6 , desorption points.” approach i n the elements and from the the heat maxima are definitely more sensitive to the lattice distortions moduced in solid solutions. crystallography of the substrateg ihan to either its Whereas the electron clouds associated with the nature (compare zinc and copper) or the nature of ions contract rapidly in the transition from the the adsorbate (compare nitrogen to argon). There one metal to the other in each pair, the distance of are, however, some clear-cut distinctions between closest approach in the crystals shows a correspondthe two adsorbents and the two adsorbates evident ing increase. In other words, although the zinc from more careful examination of Table I and Figs. ions are further apart than the copper ions, the elec5 and 9 of this paper. tron cloud can be interpreted to be more contracted Copper and Zinc Adsorbents.-If there exist no for each zinc ion than for each copper ion. The explicit differences between the metal surfaces as a dispersion attractive forces characteristic of van function of preparation, the data may be examined der Waals interaction might be expected to be with profit in terms of the differences implicit in atomic distances and electron density characteristic 300( of the two metal surfaces. It is to be noted, for example, that the maximum molar heats of adsorption 2wc is about the same for both gases but runs 400-500 cal./mole higher in all six isotherms for adsorption 6 ZOO( on the copper than on the zinc surfaces (cols. 5 and 2 6). Although the differences of approximately 0.5 kcal./mole may not seem unusually significant in 5 1500 any one comparison, considerable significance is at? * tributed to this difference because it pioved to be 3 loot distinctly characteristic of all the comparative iso.- 5 0 ( therms measured for zinc and copper octahedral surfaces a t each of the three adsorption temperatures studied. Examination of the data on the 0 ratio of metal atoms to gas molecules per unit areasurface (cols. 3 and 4) also indicates that the accomS u r f a c e Coverape Cv/vm). modation of the surface for gas molecules per unit Fig. Q.-’J’he isosteric heats of argon adsorption on single area surface is greatest for the zinc-argon system, crystal (1011) zinc surfaces calculated from Fig. 8. The intermediate for the nitrogen-zinc and for the ar- values of the adsorption corresponding to a monolayer are the average for the two temperatures of each pair of isogon-copper systems, and lowest for the nitrogen- therms determined from standard Brunauer-Emmett-Teller copper system. These differences may be resolved plots. The heats of adsorption were likewise calculated from in terms of the differences in atomic structure of each isotherm pair; 78.1-83.5, 78.1-89.2 and 83.5-89.2’K. uncertainty of the experimental values are reflected in the adsorbents. Zinc crystallizes in a curious modi- The the size of the crosses as in Fig. 5. fication of the close packed hexagonal structure (26) C. 8. Barrett. “Structure of Metals-Crystallographic Methwhich differs from the normal structure in that it is MoGraw-Hill Book Co., New York, N. Y.,1943, p. 502. extended in the direction of the c axis so that the ods,” (27) W. Hume-Rothery, “Atomic Theory for Students of Rletalaxial ratio is much greater than that for close- lurgy,” Institute of Metals, London, 1946, p. 249. 1.4
L
0 b.
i
0
I
L
L
148
.
1’.N. RHODIN
higher in the latter case since not oidy are the adsorbate molecules somcwhat nearer to the metal ions as well as being associated with more of them per unit area, but the ionic polarizabilities should also be greater for copper than for zinc surfaces. All of these effects are in qualitative ageement with the experimental observation that the heat of adsorption tends to be higher for copper than for zinc octahedral surfaces. How the surface atomic configurations affect the sharpness of the maxima in the heatcoverage curves is probably also indirectly associated with surface electron distribution. Nitrogen and Argon Adsorbates.-Reference has already been drawn to the data of col. 3 and 4 indicative of the accommodation of the surfaces for the two adsorbates in terms of the ratio of metal ions associated with the adsorption of a single gas molecule. Choice of the particular values for theoeffective cross-sectional areas of 16.1 and 14.2 AsZfor nitrogen and argon, respectively, determine the value of what might be called the apparent surface coordination number. These values are based on the fact that the extensive adsorption literature characteristic of all kinds of surfaces, particularly of metals, yields the most self-consistent interpretation in terms of these particular values. It is nevertheless true that the effective cross-sectional areas can depend on the nature of the adsorbent 5s well as the adsorption t e m p e r a t ~ r e . ~The , ~ validity of choosing particular values is the basis of all experimental determinations of absolute surface areas from adsorption data by the B E T method and hence must be defined in terms of each adsorption system under consideration. The area values for nitrogen especially, determined not only for adsorption studies calculated to give absolute surface area values independent of this parameter but physical experiments other than those of adsorption, all tend to confirm the experimental reality of this particular value of the cross-sectional area of the adsorbed molecule.21 The metallurgical evidence of this study indicating that the metal surfaces are planar oii a molecular scale is i n good agreefnent with the unusually low roughness factors indicated in cols. 7 and 8. This furnishes an additional piece of experimental evidence that the molecular areas chosen in the derivation of the roughness values have real significance. The fact that the packing appears to be greater on zinc than on copper is more apparent than real if one assumes that the molecular area of each gas is indifferent to the surface as a first approximation, since there are just ahout lor0more copper ioiis than zinc ioiis per unit area of siirfacc. (That the argoii pacliing deiisity appears t,o be higliw on both surfaces than that of the nitrogen (col. 3 atid 4) is more significant.) It is probably best understood
Vol. 57
in terms of t.he smaller radius (2.1 i.) of argon than that of nitrogen (approximately 4.6 X 3.5 A.). Comparisons with the packing mechanism is further complicated by the dumbbell-like shape of the latter molecule compared to the probable spherical symmetry of the argon atom. Finally any model of surface packing would involve the relative polarizabilities of the two gases in terms of the periodicity,of the electron fields of the interfaces characteristic of each gas-metal pair. Quantitative considerations of the electron-physics of the surface-gas interfaces is unfortunately beyond the province of t,his paper. Other treatments of this problem have been made but they all suffer from insufficiently precise data on ionic polarizabilities,28etc. The remarks on variation of surface packing are also pertinent to the data (cols. 9 and 10) OH the weight of gas corresponding to monolayer coveraoe for each gas-metal pair. The higher values of for copper-argon than for zinc-argon is in direct relation greater to the great>ersurface ion density i n the former case. This trend does not apply to nitrogen adsorption where the monolayer values for nitrogen is essentially the same for both surfaces, probably because of the asymmetry of the nitrogen molecule. The higher values for argon than nitrogen on both metals correspond t o the smaller size and higher atomic weight of argon. Contrary to the sensitive temperature dependence of V , for adsorption on the cubic and rhombohedral crystal faces of copperg the corresponding values for adsorption of both gases on the octahedral faces of both copper and zinc are insensitive to temperature in the range studied. This may be taken t o indicate that the type of packing characteristic of both gases a t liquefaction temperatures on the closely packed octahedral crystal faces of both metals does not irary. In conclusion, consideration of Figs. 5 and 9 indicates a striking similarity between heat-coverage curves of the two gases on zinc. Other data indicate the same resemblance for the heat-coverage curves of the two gases on the close-packed copper surfaces. It is not unlikely that striking differences between the two gases for comparable surfaces may be more typical of less closely packed surfaces such as the cubic or rhombohedral crystal faces. , Acknowledgments.-The effectiveness of this study was determined in no small part by the patient and time-consuming collaboration of the metulliirgical section of the Institut,e with tlie atrt,lror in his efforts t,o determine precisely the st,riictui.al I)i*ol)evtiesof the metal surfaces under study. Tlie s:iml)les Ivere prepared Iiy J. Ccrny. (28) G . Schindtz, “Tecliiiische Oberflacheiikuiido,” J. Spritiger, Berlin, 1936.
.