896
DONALD GRAHAM
Vol. 59
CHARACTERIZATION OF PHYSICAL ADSORPTION SYSTEMS. 111. THE SEPARATE EFFECTS OF PORE SIZE AND SURFACE ACIDITY UPON THE ADSORBENT CAPACITIES OF ACTIVATED CARBONS BY DONALD GRAHAM Contribution No. 179 from Jackson Laboratory E. I . du Pont de Nemours and Co., Wilmington, Delaware Received February $6, 1066
The effective adsorbent surface of an activated carbon is limited by pore screening and by the effects of impurities and substituent groups in the carbon surface. Of possible impurities and substituents, acid groups are among the most important. The effects of pores and surface acidity are separated and measured for a series of commercial carbon adsorbents using as adsorbates, cationic and anionic dyes sufficiently large in molecular dimensions to average out the effects of heterogeneity. The fraction of the total surface which adsorbs a cationic dye (methylene blue) is in most cases determined by pore screening alone. However, an anionic dye (metanil yellow) is repelled by acidic substituents on the carbon and finds a receptive surface smaller than that for methylene blue by an amount related to the surface acidity.
Introduction Physical adsorption on a non-porous, essentially unsubstituted carbon surface (such as that of Graphon) is non-specific in that the entire surface is receptive to any adsorbate. The strength of the adsorption bond varies only with those factors which determine the van der Waals forces involved. Activated carbon, on the other hand, is porous and varies widely in purity and in the degree of substitution of its surface. Pore screening in adsorption from aqueous solution has been repeatedly recognized and an attempt has been made to measure its effect.l For a particular adsorbate on a number of different activated carbons, adsorption capacity was plotted vs. the area of all pores with diameters larger than a selected value. That limiting diameter which gave the straightest line was considered the smallest which would admit the adsorbate molecules. Adsorbents were treated in groups which could include some in which other factors obscured the effect of pore size. Since any such effect would impose a further limit on adsorption capacity, the indicated limiting pore diameter would tend to err on the high side. The pore size distribution of an activated carbon always limits its capacity for adsorbate molecules too large to enter the smallest pores. Substituent groups or impurities in the carbon surface may also limit adsorption if they repel or fail to attract the molecules of the adsorbate. I n the characterization of an adsorbent, it is therefore desirable to select an adsorbate for which the capacity is measurably related to the surface concentration of the substituent or impurity under study. Acidic groups produced by oxidation during or following activation, constitute an important class of substituents in the surfaces of porous activated carbons. It will be shown that these groups limit the capacity of the adsorbent for an acid (anionic) dye. The present work illustrates a method for measuring separately the effects of pore size and acidic substitution, upon the fraction of the total carbon surface accessible to the first monolayer of cationic and anionic adsorbates. This method has been (1) A. J. Juhola, W. H. M a t s and J. W. Zabor, paper presented April 1, 1951, before the Division of Sugar Chemistry at the 119th meeting of the American Chemioal Society, Boston, Mass.
used in the study of over forty different carbon adsorbents. It has proven applicable to the majority of porous activated carbons, without serious interference from other factors. The method provides for the detection of such interference, when present, as illustrated later in this paper. Some adsorbents, of course, require a different approach. An extreme example is found in the non-porous carbon black prepared by thermal decomposition of acetylene. Its surface substitution is largely hydrocarbon in character and its adsorbent properties are therefore very different from those of carbons activated by oxidation. The adsorbates used in this study were two dyes of comparable moIecular dimensions and opposite ionic character. Methylene blue, a cationic (basic) dye
was selected for use in measuring the effect of pore screening alone. This dye when ionized in aqueous solution carries a positive charge. Any interaction with negative acidic groups in the carbon surface is therefore attractive and the quantity of dye adsorbed in the first monolayer may, in the absence of other factors, be considered a measure of the total physically accessible surface, or the surface limited by pore screening alone. Metanil yellow, an anionic (acid) dye, was used in studying the effects of acidity of the carbon surc>-N=N-~-E-c>
I
SOaH
face. As an anion, in aqueous solution, it tends to be repelled by acidic groups on the carbon. Therefore, although subject to practically the same pore screening, it finds less receptive surface on an acidic carbon than does methylene blue. The molecular dimensions of these dyes are important. The molegules of both are approximately 18 A. long, and 9 A. wide. They are thus closely comparable with respect to pore screening. They are sufficiently large to average out most of the heterogeneity of the carbon surfaces and a t the same time small enough for the anionic metanil yellow to
897
CHARACTERIZATION OF PHYSICAL ADSORPTION SYSTEMS
Sept., 1955
TABLE I INORQANIC IMPURITIES OF CARBON ADSORBENTS Sulfate ash,
Sulfur,
Phosphorus,
Iron,
No.
Starting material
%
Graphon 1 2
Spheron 6 (heat treated) Wood Black ash from paper mill liquor Soft coal Wood f 4 (heat treated) Lignite Wood
...
...
...
...
4.27 3.06
0.02 1.60
0.01 0.02
0.11 0.06
22.70 1.80
1.40 0.40
0.09 0.65
1.32 0.02
15.70 2.47
0.61
0.03 0.007
0.14 0.03
3 4 4A 5 6
%
...
%
%
...
...
...
0.40
TABLEI1 SUMMARY OF ADSORPTIONDATA Methylene blue
Carbon
Graphon 1
2 3 4 4A 5 6
Metanil yellow
Total surface area, m.'/g.
(5L)m, P./&
Accessible area, m.a/g.
Accessible area, % of total surface
83.9 1120 1130 1300 1300 1300 600 580
0.0265 .239 .264 .288 .I90 .300 .136 .118
83.9 757 836 912 602 950 430 374
100 68 74 70 46 73 72 64
be measurably sensitive to the effect of acidic surface substituents.
Materials and Methods A. Adsorbates.-The methylene blue used was tho Eastman Kodak Company Certified Grade. Analyses for nitrogen, chlorine and sulfate ash indicated a purity of about 99% and the dye was used on this basis. The metanil yellow was repeatedly recrystallized from water until a small sample dried first at low temperature and finally a t 140' showed a purity of a t least 99.9% us. a standard Ticla solution. The product was then dried in the same way and bottled. Both dyes were used as 0.100% solutions in distilled, deionized water. The solutions were made up by weight and checked for concentration by optical density measurements in the Cenco Photelometer. The molarity of the methylene blue solution was 0.00267 and the metanil yellow, 0.00266. B. Adsorbe&.-Six different, commercial, active carbons, recommended by their manufacturers for adsorption from aqueous solution, were studied. They are identified by numbers rather than trade names as the data may be applicable only to the specific samples used. Graphon was employed as the reference standard. Inorganic impurities, iron, sulfur, nitrogen and phosphorus were determined by routine analytical methods. The results are given in Table I. Total surface areas were obtained by application of the BET equation* to the nitrogen isotherm a t -195.8' using 16.2 A.2 as the area of the nitrogen molecule. The results are listed with other adsorption data in Table 11. Pore area distributions wcre determined by the Barrett, Joyner, Halenda method.* The results are plotted in Fig. 1. C. Adsorption Equilibrium Measurements.-A temperature of 80" was used for the adsorption measurements to provide hi h diffusion rates. The equilibration time was determinecfby a series of measurements for each carbon extending from 0.5 to 100 hours. Graphon reached essentially complete equilibrium in about one hour but the porous activated carbons required 6 to 20 hours to approach their (2) 8. Brunauer, P. H. Emmett and E. Teller, J . Am. Chem. SOC., 60,309 (1938). (3) (a) E. P. Barrett, L. G. Joyner and P. P. Halenda, i b i d . , 73, 373 (1951). (b) Measurementa of pore area distributions were made at Juniata College by Dr. Raymond T. Davis. Jr., present address U. S. Steel Corporation, Pittsburgh, Pennsylvania.
Acidity Accessible a ~ ~ , t ~ r ~ of&carbon Accessible area, % metanil yellow area, of total m.z/g. aurface mctliylene blue m.'
(:)m, P./&
0.0291 .250 ,218 .305 .I14 .250 .112 .lo2
83.9 721 629 880 329 721 323 294
1.00 0.05 .76 .97 .55 .76 .75 .79
100 64 56 68 25 55 54 51
0
.12 .45 .06 .94 * 80 .70 .60
100-hour values. The timc for all runs was therefore fixed a t 20 hours. This is consistent with earlier results on Ti024 which showed up to 9 days required for equilibration at 25' but only 1 day at 60'.
16 20
30
40
60
80 100 120 160 200
D, pore diameter in A.
Fig. 1.-Pore
area distributions.
The procedure employed was as follows: 50 ml. of: 0.100% dyc solution was placed in a flask, heated to 80 , and a predetermined quantity of carbon added. The flask was closed, placed in a water-bath at 80' ( = k l . O o ) and shaken gently with a reproducible rocking motion for 20 hours. The water-bath accommodated 21 samples and the usual charge comprised three checks for each of seven different points on an isotherm. At completion of the run, the carbon was removed by filtration a t 80" on a 5.5 cm. Whatman No. 5 paper under slight suction. A hot filter was used on a cold 500-ml. flask. (The usual centrifugal method proved poorly suited to some of the carbons used.) Filtration introduced no appreciable error in the metanil yellow data but it was necessary to apply a correction (which varied with concentration) to the methylene blue results. The final e uilib rium concentration was determined by means of a %enco Photelometer with an empirical calibration for each dye. The amount adsorbed per unit weight of carbon was calcu(4) W. W. Ewing and F. W. J. Liu, J . Colloid Sci., 8,204 (1953).
898
DONALD GRAHAM
70
60
-
m
METANIL YELLOW
-I
X
50
METHYLENE BLUE
3Ol 0
1
500
1
1000
I
JC
I
x
2000
1
1
3000
10-3
Fig. 2.-Reciprocal plots of 80" isotherms for the adsorption on graphon of methylene blue and metanil yellow from aqueous solution. lated from the measured change in concentration and the weight of carbon used. D. The Measurement of Surface Acidity.-The concentration of acidic substituent groups in the carbon surface was determined by titration. A quantity of carbon representing 1000 square meters of total surface was slurried in 50 ml. of 0.0100 N NaOH solution a t 80" for 30 minutes. The slurry was then titrated back to the pH of 0.0100 N NaOH solution with 0.100 N NaOH solution. The results are listed in Table I1 in milliequivalents per 1000 square meters of surface (meq./1000 m.2).
Discussion of Results A. Areas Occupied by Adsorbate Molecules.Methylene blue and metanil yellow lying flat on a solid surface in aoclose packed monolayer would cover 160-180 sq. A. per molecule (based on molecular dimensions). However, their multiple bonds to the adsorbent tend to limit mobility and to give a more random packing. The areas occupied by each molecule were therefore determined experimentally by adsorption on Graphon. Determination of the content of a complete adsorbed monolayer is greatly simplified if the system can be treated as "ideal" (uniform sites and no interaction between adsorbed molecules). Graphon is exceptionally uniform and its adsorption of a wide variety of compounds is essentially free from evidence of heterogeneity. Also, effects of lateral interaction between adsorbed molecules are rarely observed in aqueous systems. Under these "ideal" conditions, a plot of (mix), the reciprocal of the amount adsorbed per unit mass of adsorbent, vs. 1/C, the reciprocal of the corresponding equilibrium solution concentration for isothermal adsorption, is a straight line. The
Vol. 59
intercept ( m / z ) a~t 1/15' = 0 is the reciprocal of ( ~ / mthe ) ~content , of the filled monolayer per unit mass of adsorbent. Such reciprocal plots of data representing the 80" adsorption isotherms of methylene blue and metanil yellow on Graphon (Fig. 2) are linear as expected. The intercepts (obtained by the method of least squares) gave values of ( z / m ) , of 0.0265 for methylene blue and 0.0291 for metanil yellow. These results in combination with the BET surface area of the Graphon, 83.9 m.2/g., gave values for the areaso occupied by the individual molecuJes of 197 sq. A. for methylene blue and 179 sq. A. for metanil yellow. B. Pore Screening and Accessible Surface Area.-Use of the "ideal" reciprocal plot to determine the content of a monolayer adsorbed on an activated carbon with a heterogeneous surface is made possible by proper selection of the adsorbate. We have found (in work to be reported separately) that much of the heterogeneity of carbon surfaces is of molecular dimensions and that its effects disappear when the adsorbate molecules are large enough to average out the discontinuities. Both of the dyes used in this investigation fill this requirement. The isotherms for the adsorption of methylene blue and metanil yellow on the different activated carbon samples gave reciprocal plots comparable in linearity to those of Fig. 2. From the intercepts, values of (zlm), were obtained as before. These, combined with the molecular areas, determined on Graphon, gave values for the surface of each activated carbon accessible to each of the two dyes, in square meters per gram (Table 11). The difference between the total surface area of a carbon and that accessible to methylene blue is (unless otherwise noted) considered due t o the action of pore screening. This premise is shown later to be consi.stent with experiment. The still smaller area accessible to metanil yellow involves the additional factor of carbon surface acidity which is also considered later. C. Hindered Adsorption.-The surface areas accessible to methyleneblue (Table 11) are greater than the areas of pores with diameters exceeding 16 8. (Fig. 1) except for Carbon No. 4. Since the accessible surface of Carbon No. 4 for methylene blue was only 46% of the total surfaze vs. 75% for pores of diameters exceeding 16 A. (and since that accessible to metanil yellow was only '25% of the total) it was concluded that the surface carried substituent groups which hindered the adsorption of both dyes. A sample of Carbon No. 4 was therefore heated at 900" for 20 hours in an atmosphere of nitrogen and cooled to room temperature before contact with air. The product, Carbon No. 4 A , was unchanged in total surface area but the surface areas accessible to both dyes were greatly increased. The nature of the chemical change in the carbon surface caused by the heat treatment has not yet been determined. D. The Limiting Pore Diameter for Methylene Blue.-The pore area distributions of Fig. 1 involve the assumption of round, tubular pores. The nominal diameter of an irregular shaped pore (as
c
CHARACTERIZATION OF PHYSICAL ADSORPTION SYSTEMS
Sept., 1955
-3 0.6
899
d
0 0.2 0.4 0.6 0.8 1.0 Acidity of carbon surface in meq./1000 maz. Fig. 4.-The effect of surface acidity upon the carbon surface area receptive to metanil yellow.
pore diameter for the adsorption of methylene blue by that carbon (and presumably also for metanil yellow). CarbonNo. Limiting&
1 2 3 5 6 12.0 14.0 14.6 12.2 13.7 Av. = 13.3 b.
Part of the observed variation between the individual values is of course due to error but part may be real as the pore shape distribution may be expected to vary with the source and method of activation of the carbon. The over-all general agreement between the results from the different carbons confirms the earlier suggestion that, except for Carbon No. 4, pore screening is the principal factor limiting the surface accessible to methylene blue. The average nominal limiting pore diameter of 13.3 A. for methylene blue is in reasonable agreement with the 15.0 A. value obtained from the very different approach of Juhola, el u L , ~particularly since error in the latter value would tend to be on the high side. E. The Effects of Surface Acidity.-For every activated carbon studied, the surface area accessible to the first adsorbed monolayer of metanil yellow is less than that for methylene blue. The observed differences vary widely. The molecular dimensions of the two dyes are closely similar and the observed differences in accessible surface area show no relation to the pore area distributions of the different carbons. The measurement of accessible areas is based on the premise that an unsubstituted, nonporous carbon surface (Graphon) is equally receptive to both. The observed differences are therefore ascribed to interactions of the adsorbate molecules with specific substituents in the carbon surfaces. Acidic substitution of the carbon surfaces could explain the observed effect and a correlation was sought. The effect was measured as the fraction of the surface accessible to methylene blue which would also receive metanil yellow, or the ratio of the two accessible surface areas. A plot of this ratio against surface acidity (Fig. 4) was roughly linear. The data for Carbons 4 and 40. are not seriously out of line, indicating that the general hindrance noted in the single case of Carbon 4 is separate from the effect of acidic surface substitution. These results indicate that acidic groups in the
3 L 00
5
IO
15
20
25
30
Dp PORE DIAMETER IN A'.
Fig. 3.-Extrapolated
pore area distributions.
calculated by the Barrett, Joyner, Halenda method) is more representative of its average diameter than of its smallest. The nominal limiting diameter is therefore somewhat larger than that of the smallest round, tubular pore which can pass methylene blue molecules, At this point it can be stated that the nominal limiting eore diameter for methylene blue lies between 16 A., s h o p in the preceding section to be too large, and 10 A., the diameter of the smallest round pore that would freely pass the dye molecules. A simple average of these two values (13 A,) might provide an acceptable approximation but it is useful to approach the solution in a different way, considering the different carbons individually. The calculation of pore area distribution employs the Kelvin equation, which is of questioned validity for diameters below those of Fig. 1. Therefore, rather than extend the calculation to lower diameters, a simple extrapolation was used, based on two assumptions: (1) only a very small fraction of the total surface area is found in pores with diameters less than 10 8. This is in agreement with all available data. (2) The total surface as measured by nitrogen adsorption includes no pores with diameters less than 4 8. (approximately twice the diameter of the space occupied by a carbon atom in amorphous carbon). A nitrogen molecule would not be expected to enter a pore representing the absence of a single carbon atom. The extrapolation (excluding Carbons 4 and 4 A ) is shown in Fig. 3. The percentage of the total surface area of each carbon which is accessible for the adsorption of methylene blue (from Table 11)is marked on the extrapolated curve for the same carbon in Fig. 3. The corresponding abscissa represents the nominal limiting
900
NORMAN HACICERMAN AND EMERSON H. LEE
Vol. 59
carbon surface tend to reduce the capacity of a carbon adsorbent for metanil yellow, and probably for anionic adsorbates in general, roughly in proportion to the concentration of these groups in the surface.
groups, in the carbon surface, hinder the adsorption of either dye. I n the single case observed, the effect was greatly reduced by heating in nitrogen a t 900". (4) The nominal limiting pore diameter for summary methylene blue (and presumably also for metanil (1) The areas occupied by individual molecules yellow) is approximately 13 8. of methy1en.e blue and metanil yellow in mono(5) Acidic substituents in the carbon surface layers adsorbed on Graphon have been determined reduce the surface area accessible to the anionic and used to measure the surface areas of activated adsorbate, metanil yellow, roughly in proportion to their concentration in the surface. carbons accessible to these molecules. (2) The surface area of an activated carbon Acknowledgment.-The author wishes to thank accessible to Methylene Blue is usually limited Dr. F. C. Chromey of this Laboratory for assistonly by the size of the pores. ance in the statistical treatment of experimental (3) I n some cases, unidentified substituent data.
THE EFFECT OF GASES ON THE CONTACT POTENTIALS OF EVAPORATED METAL FILMS BY NORMAN HACKERMAN AND EMERSON H. LEE Department of Chemistry, University of Twxs, Austin 11,Texas Received February 86, 1966
The effect with time of oxygen, nitrogen, water vapor and air on the contact potential differences between aged bulk platinum and evaporated metal films was studied by the vibrating condenser method. The metals used were aluminum, lead, nickel, chromium and iron. Both reversible and irreversible effects were observed. An explanation is offered based on sorption of the gases which provides either a dipole barrier or an ion barrier to the emission of electrons.
Introduction
If two unlike metals are brought into contact in
cleavage of a metal crystal,1° out-gassing of a pure metal in a vacuum,11v12and evaporation of metal films.13 Use of evaporated films offers some advantages since occluded gases are boiled out prior to and during evaporation. The disadvantage in using metal films is that of correlating the properties of the films with those of bulk metals.'4~~~ Two important factors to be considered are the effect of crystal orientation and film thickness on the work function of the films. I n this work, contact potentials between evaporated films and an aged platinum reference were measured in a vacuum system a t room temperature as a function of time after evaporation. Potentials were measured by the vibrating condenser method.16 The effects of air, oxygen, nitrogen and water vapor on evaporated films of aluminum, chromium, nickel, iron and lead were studied. The stability of the platinum reference was also studied for the experimental conditions used.
air and then separated slightly, a static electrical potential difference exists between them. This potential difference is called' the contact potential, or Volta potential difference. Contact potentials are a function of the electronic work functions of the two metals involved; the difference of the two work functions, in electron volts, is numerically equal to the contact potential difference, in volts. The work function of a metal is changed by adsorbed gas films or chemical films'; therefore contact potentials can be used to study adsorption and reactions of a gas with a metal. This may be done by measuring the potentials between a freshly formed surface and an aged surface. Since the aged metal is relatively stable, any change in contact potentials is ascribable to a change in work function of the fresh metal surface as it adsorbs a gas or reacts with it. (10) F. B. Daniels and M. Y. Colby, Phys. Rev., 52, 1200 (1937). The work functions of pure metals have been (11) J. R. Anderson and A. E. Alexander, Australian J . Ch.. 6 , 109 found to be anisotropic2-4 and to vary with allo- (1953). (12) C. W. Oatley, PTOC.Phys. Soc., 6 1 , 318 (1939). tropic modifications.6J Therefore crystal strucP. A. Anderson, Phys. Rev., 47, 958 (1953); 48, 320 (1936); ture and orientation in metals affect contact po- 57,(13) 122 (1940); 76, 388 (1949); R. Kh. Burstein, M. D. Surova and tential measurements. I. A. Zuidenman, Zhur. Fiz. K h i m . , 24, 214 (1950) [C. A , , 44, 6743 Methods for obtaining a bare metal surface (1950)l; E. W.J. Mitchell and J. W. Mitchell, Proc.Roy.Soc. (London), include abrasion in air1 or in a vacuum s y ~ t e m , ~AalO, , ~ 70 (1952); I. Ogawa, T.Doke and I. Nakada, Oyo Butsure, aa, 101 (1953) [C.A., 47, 10380 (1953)l; T. V. Kalish and R. Kh. Bur(1) R. Suhrman, Z . Elektrochem., 66, 351 (1952). (2) R. Smoluchowski, Phys. Rev., 60, 661 (1941). (3) 8. T. Martin, ibid., 6 6 , 947 (1939). (4) P.A. Anderson, ibid., 59, 1034 (1941). (5) A. Goetz, ibid., 33, 373 (1929). (6) H.B. Wahlin, ibid., 61, 509 (1942). (7) H.II. Uhlig. J . Applied Phys., 28, 1399 (1951). (8) F. Fianda and E. Lange, Z. Elektrochem., 5 6 , 237 (1951). (9) J. Giuer and E. Lange, Nalurwissenschajlen. 40, 506 (1953).
stein, Dolclady Acad. Nauk SSSR., 81, 1093 (1951) [C.A . , 46, 3829 (1952)l; C. F. Ying and H. E. Farmworth, Phys. Rea., 8 6 , 485 (1952); Y.Yashiro, Bull. Nagoya Insl. Technol., 8 , 333 (1951); J. C. P. Mignolet, J. Chem. Phys.. ao. 341 (1952); Disc. Faraday Soc., 185 (1950). (14) J. -4. Allen, Rev. PUTS A p p . Ch., 4, 133 (1954). This is a comprehensive review on evaporated metal films. (15) R. A. Sennett, T. A. McLaughlin and G. D. Scott, Can. J . Phys., 30, 370 (1952). (16) W. A. Zisman, Rev. Sei. Instr., 3, 367 (1932).