in nitric acid was prepared. Table V shows the results of ratio measurements of the various standard solutions which contain from 0 to 2.5 mg. per ml. of the metals. When the ratios of either the CuKa or NiKa to the WLpl intensities are plotted us. the metal content, straight line calibrations are obtained. Table VI shows the results of determining the copper and nickel content of various solutions. The sample solutions were diluted to a suitable concentration range. Silver electrolyte was also analyzed using synthetic standards. The x-ray beam reduces the silver in the solution, thus inaccuracies result.
standards is shown theoretically and experimentally to compensate for instrumental variations and absorption and enhancement effects. Solutions are most advantageously analyzed by this technique since high scatter intensities give statistically accurate counts in a minimum time. Short count times are of importance in solution techniques because of the action of the x-ray beam on the sample. Comparisons of chemical and x-ray determinations of typical copper refinery solutions show deviations of about *l%. ACKNOWLEDGMENT
DISCUSSION
The use of coherently scattered x-ray tube characteristic radiations for internal
The author thanks United States Metals Refining Co., a subsidiary of American Metal Climax, Inc., for
permission to publish the results of this investigation. LITERATURE CITED
(1) Andermann, G., Kemp, J. W., ANAL.
CHEM.30, 1306 (1958).
(2) Compton, A. H., Allison, S. K.,
“X-Rays in Theory and Experiment,”
2nd ed., pp. 116-40, Van Nostrand,
New York, 1935. (3) Hauk, W. W., Silverman, L., ANAL. CHEM.31, 1069 (1959). (4) Jones, R. W., Ashley, R. W., Ibid., 31, 1629 (1959). (5) Liebhafsky, H. A., Pfeiffer, H. G., Window, E. H., Zemany, P. D., “X-Ray Absorption and Emission in Analytical Chemistry,” pp. 168-70, TViley, New York, 1960. RECEIVED for review January 16, 1962. Accepted April 13, 1962. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March 1962.
Measurement of Skeletal Densities of High Surface Area Inorganic Oxides with a Gas Pycnometer JOHANNES TUUL and R. M. DeBAUN American Cyanamid
Co., Sfamford,
Conn.
,A commercial gas pycnometer is used to determine skeletal densities with helium. The results of these determinations are compared with those obtained with liquids. Reasons for the substantial differences between helium and water are suggested. Experiments with different catalyst materials are described, and an explanation of the results is attempted. Possible deviations of the helium skeletal densities from the true values, due to the finite size of the helium atom and adsorption of helium, are also considered.
T
air comparison pycnometer, manufactured by Houston Instrument Corp., Houston, Tex., is an instrument for measurement of volumes and skeletal densities of irregularly shaped, porous, or pgwdered materials. The details of the method are given in “Prelimipary Operating Instructions, Model 200 Air Comparison Pycnometer,” supplied by the manufacturer. The instrument consists essentially of two cylinders with a piston in each. A differential pressure gage indicates when the pressure in the two cylinders is equal. The displacement of air by the sample is obtained from the positions of the pistons. A scale in the form of a counter enables the observer to read the volume directly. The reading accuracy is 0.01 cubic em. HE
814
ANALYTICAL CHEMISTRY
EXPERIMENTAL
With low surface area samples, the instrument performed satisfactorily. However, when i t was used in our laboratory for examining catalyst materials with large surface areas, unrealistic and even negative volume readings were obtained. This indicated that a substantial amount of air was adsorbed on such specimens during measurement. The instrument was then returned to the manufacturer to be modified so that other gases besides air could be used in it. Essentially this amounted to adding another valve and recalibrating the instrument. The main portion of this investigation was carried out on alumina base catalysts. These were obtained in the
Table 1.
form of pellets or extrudates. Some measurements were also carried out with samples of silica and silica-alumina which were obtained in the form of fine powders. The characteristics of the materials with which the bulk of this work was done, are given in Table I. A skeletal density measurement was made as follows. The cup of the pycnometer was cleaned and weighed on an analytical balance. Then the cup was filled with the sample, its edge was thoroughly cleaned, and the cup was weighed again and clamped to the pycnometer. The instrument was evacuated with a Cenco Hyvac mechanical pump. The evacuating was done slowly by gradually opening the valve which connected the pycnometer with the vacuum pump. This pre-
Materials
Original Calcina-
Surface
Notation A AM AMC
s
Composition Physical State Gamma-alumina 1/16-inchextrudates 90% A1203 10% Moos x 3/18 inch pellets 82% Al,O, 15% MoOa 1/16-lnCh extrudates
+ +
Si02
microspheres
Area Compacted tion (S), Bulk TemperaMeter2/ Densitg, ture, O C. Gram Gram/ c
+ 3% COO
240
0.56
593
200
1.oo
593
270
0.59
482
700
0.43
200
caution was found to be extremely important with fine powders which “boil” when air is removed too rapidly. Pellets and extrudates may also have some loose material on the surface so that care is needed. Experiments indicated that even a 5- to 8-minute evacuation was insufficient. Therefore, the pycnometer was filled with helium, re-evacuated, and once more filled with helium to 1 atm. pressure. Repeated volume readings were taken; increases of up to 1% were observed after the first one. The materials under study were subjected to various heat treatments in a thermostated furnace. After treatments the samples were either exposed to room air by letting them stand in open bottles or they were kept in sealed containers. From time to time the weight and skeletal density were determined. The standard deviation in the helium skeletal density measurement was about 0.01 gram per cc. RESULTS
Preliminary skeletal density measurements with helium on alumina extrudates gave results which in some cases differed considerably from water values. This suggested a study of the effect of moisture. Some interesting results were obtained by subjecting samples of various alumina-base catalysts to different heat treatments and exposing them to room air (temperature, 19” to 27’ C., relative humidity 30 to 70%) for periods ranging up to 110 days. The observed skeletal density increased when the sample was calcined a t successively higher temperatures and decreased when the sample was exposed to room air. This decrease in apparent skeletal density and increase in weight were due to water vapor which was taken up by the
Sample No. 35 40 47 3 46
t15; 61 65 67 71 77 78 .-
91 84 85 88
Material A A A B A A
AM AM AIM AM AM AM AMC AMC AM/= AMC AMC AMC
sample. When moisture was excluded from the sample, either by keeping it in a sealed container, or by drying the air to which the sample was exposed, almost no changes occurred in the weight and skeletal density of the sample. The minute changes observed were mainly due to exposure to room air during weighing and transfer to and from the pycnometer. Results are compiled in Table I1 where the changes, due to aging, in six samples of each of the three materials ( A , A M , and A X C ) are recorded. These samples have been chosen so as to represent the characteristic heat treatments and aging procedures used. Do is the skeletal density of a sample after heat treatment, and D is the skeletal density a t the end of the observation period. In the last column, the apparent density of water taken up by the sample, D‘, is recorded. As can be seen, D’ always exceeds the density of liquid water. Average values of D’ have been evaluated. For Al&, it is about 1.6; for Al2O3-Moo3,the value of D’ lies between 1.3 and 1.4; and for A1203-Mo03-Co0, it lies between 1.4 and 1.5. Water seems to combine with gammaalumina to form a surface compound with a density near that of the trihydrate, A1,03 3H20. Calculations based on the results yield an average value of 2.43 grams per cc. for the compound formed in the range of 0 to 7% weight increase. This is in good agreement with the literature value 2.42 grams per cc., for alumina trihydrate (7’). A quantitative consideration, based on the surface area of the material and increases in its weight and volume, suggests that the formed layer is thin, per-
Table II. Helium Heat Treatment Standing Tempera- in Sealed Duration, ture, Container, minutes O c. Days 120 150 160 70 80 350 120 30 320 160 30 280 400 15 350 400 5
40
550 500 593 800 593 500 430 630 500 550 593 593 500 550 593 550 593 593
4 17 16 7
5k 62
1
15 19 14 13 13 3 7
6 7 6 59
haps only a single layer, a t the end of the observation period of 3 to 4 months. For multi-component materials, matters are more complicated. More water is bound per unit mass and unit surface area than on pure gamma-alumina. Thus far, effects due to the active components cannot be separated from those due to the carrier. COMPARISON WITH OTHER METHODS
The first comparative skeletal density measurements with water and helium were carried out on a series of 10 samples of gamma-alumina. The values obtained for most samples with water were about 5Yc higher than those obtained with helium. Further careful observations indicated that the degree of agreement between the two methods of measurement depends on the extent of exposure to room air after calcination. The results of a typical series of comparative measurements are shown in Table 111. There is a satisfactory agreement for the first three samples which have been “aged” by 42 to 55 days of exposure to room air, although the values obtained with water tend to be slightly lom-er. For the three control samples, which have been kept in sealed containers after calcination, about 4ojO higher skeletal densities have been obtained with water than with helium. Similar results are obtained with other materials-e.g., Table IV, AlsOs Moo3 and Table V, Al2O3--MoO3-Co0. Each water skeletal density given in Table IV is an average of several readings. This procedure was adopted because greater difficulties were experienced in evacuating samples of AlzOr Moos in water than with other mate-
Skeletal Densities
Exposed t o Room Air, Days
Weight Gain
%
Do
D
D‘
53 42 42 82 0 0 23 64 34 34 34 9 26b 64 58 58 10 480 10 48c 0
8.5 8.2 7.5 8.0 0 0 4.0 6.5 6.3 6.2 6.3 2.6 7.8 8.0 8.7 4.7 4.8 0
3.29 3.29
3.08 3.05 3.06 3.13 3.28 3.28 3.26 3.17 3.17 3.16
1.77 1.62 1.57 1.68
+
+ +
Skeletal Densities,“ Gram/Cc.
3....3 0 ~
3.36 3.28 3.32 3.46 3.46 3.45 3.46 3.46 3.46 3.56 3.54 3.54 3.51 3.49 3.54
... ...
1.36 1.37 1.39 1.33
3.52
a DOis the initial skeletal density after calcination, D the skeletal density a t the end of the observation period, and D’ the apparent density of water taken up by the sample. b 9 days in regular room air, 26 days in dry air. 8 10 days in regular room air, 48 days in dry air.
VOL. 34, NO. 7, JUNE 1962
815
rials. There was question whether some gas evolved owing to a chemical reaction of water with that substrate. However, mass spectrometric analysis of the released gas revealed only water vapor and air. The same was the case with other substrates under study. The greater difficulty experienced in evacuating the pelleted molybdena catalyst can be ascribed to denser packing and finer pores of this material. This explanation suggests itself if one considers the bulk density of the substrates studied: for the pelleted catalyst this quantity is about 70% higher than for any one of the other materials. It has been shown that the skeletal density of alumina base catalysts decreased when exposed to water vapor. Although the results obtained with water and with helium mere in qualitative agreement in this respect, there The was a quantitative difference. decrease in skeletal density as indicated by helium measurements amounted to 7 to lo%, whereas measurements with water showed decreases of 11 to 17%.
Table 111.
I
2
4 5 5 7 8 PER CENT WEIGHT GAIN
3
9
IO
Weight Heat Gain Due to Treatment Atmospheric Temperature, Exposure, c. %
37 -.
550 ...
8.1
40 44
500 593 550 500 593
s.2 8.4
37a 42 45
Figure 1. Heat of immersion vs. weight per cent adsorbed water
The explanation for this is that water is bound with these substrates thus forming a surface compound. In room air the uptake of water is gradual but on
Water skeletal Density, Gram/Cc. Observer I Observer I1 2.94 2 89 2.88 3 47 3.44 3.36
0 0
0
3.05
3.03 2.93 3 42 3.38 3.44
Helium Skeletal Density, GramiCc. 3.06 3 05 3 05 3.28 3.28 3.28
Comparison Between Helium and Water Skeletal Densities of
A1203-
MOO3
Sam le Kumier
Heat Treatment Weight Gain Due Temperature, to Atmospheric O c. Exposure, % 630 630
13 16 56 18 19 70
630 630 593
Weight Heat Treatment Gain Due to Temperature, Atmospheric ' C. Exposure, % 8.6 8.4 8.5 8.7 0
..
77
78 91 73
77a
816
0
Water Skeletal Density, Gram/&.
3.25 3.18 3.17 3.46 3.46 3.46
3.25 3.21 3.18 3.66 3.65 3.72
Skeletal Densities of A1203-MoOs-CoO
72
79 89
Helium Skeletal Density, Gram/Cc.
4.4 6.4 5.4 0 0 0
500
Table V.
Sam le Kumier
II
Comparative Skeletal Density Measurements on Fresh and Aged Samples of Gamma-Alumina
Sample Kumber
Table IV.
0
593
ANALYTICAL CHEMISTRY
0 0 0
Water Skeletal Density, Gram/Cc.
Helium Skeletal Density, Gram/Cc.
3.10 3.15 3.16 3.14 3.75 3.73 3.84 3.79
3.17 3.20 3.16 3.18 3.54 3.54 3.52 3.55
Ethyl Alcohol Skeletal Density, Gram/Cc. 3.20 3.20 3.19 3.26 3.55 ...
3.64 3.60
immersion in water this reaction occurs rapidly. Considerable amounts of heat evolve when freshly calcined samples are immersed in water. The heat of immersion was measured with a number of samples of two materials, and the results are presented in Figure 1 where the heat of immersion is plotted us. relative rveight gain. These curves indicate that the first water is bound very strongly. When the weight of &03or A1203-Mo03has increased by about 6% owing to the uptake of water, the slope of the heat of immersion curve has changed markedly. This indicates a corresponding change in the differential heat of immersion. The calorimeter measures essentially the difference between the heat of adsorption and the heat of vaporization. Taking this into account, and using the results of this investigation on the uptake of water, one finds that the initial heat of adsorption of n-ater on gamma-alumina and A1203-Mo03is about 21 kcal. per mole of water, when the material has been calcined at 600' C. The quoted heats of adsorption per mole of water are somewhat uncertain since the amounts of water adsorbed in the calorimetric measurements have been estimated indirectly. According to de Boer (f), only two thirds of the water which has been removed by heating is quickly readsorbed on alumina; the remaining one third is taken up in a slow process. The observations of Guderjahn, Paynter, Berghausen, and Good (2) point in the same direction. If this were the case in the present work, the quoted values would be too low. In the absence of accurate data, definite conclusions in this regard have to be postponed. After 3 months' exposure to room air, the materials under study do not take up water readily. One could then assume that on immersion in water, a fresh sample takes up about the same amount of water as during 3 months of exposure to room air. Therefore, when water is used as a displacement medium, an error is introduced into the skeletal density determination due to the reaction of water with the substrate. This error can be evaluated on the basis of the results of the present work, and predictions can be made regarding skeletal densities determined with water. If the foundations of these calculations are sound, the water skeletal densities of fresh samples of &03, AlUZ03-Mo03, and A1203-Mo03-Co0 ought to be 3.65, 3.65, and 3.84 grams per cc., respectively . These calculated values are in good agreement with the experimental water skeletal densities for fresh samples, Table VI. It IS concluded that when the specimen has been exposed to room air for 3 months or more, or saturated with water :n any other manner, both
water and helium give about the same skeletal density. In other cases, a too high skeletal density is obtained with water. For “aged’, samples, water sometimes yields slightly lower skeletal densities than helium. This is probably due to the larger dimensions of the water molecule as compared with the helium atom. Water molecules are excluded from a fraction of the pores whereas helium atoms still have access to them. A few measurements were carried out with ethyl alcohol and carbon tetrachloride as displacement media. The results obtained with ethyl alcohol for seven samples of A1203-Mo03-Co0 are recorded in the last column of Table V. The agreement is good with the values obtained with helium and lends support to the above conclusions. Table VI1 records the results of a series of comparative measurements with helium, water, and carbon tetrachloride. The agreement between helium and carbon tetrachloride is rather good for fresh as well as aged samples. This fact, and the results obtained with water, once more confirm the conclusions drawn. DISCUSSION
The experimental facts of the present work are in agreement with the applicable ones of Steggerda (7), but the interpretations are different in some aspects. Steggerda has taken as the true skeletal density of alumina the one calculated from the dimensions of the unit cell which in turn have been determined by x-ray diffraction. This procedure yields the correct skeletal density in case of a perfect crystal and, thus, the upper limit which is never achieved in practice. Steggerda’s objections to helium are properly based on two facts-viz., the finite size of the helium atom, and the adsorption of helium on solids. Since the ideal gas law treats atoms as point masses, an error is inherent in the determination of the volume of a solid with a gas. A calculation of the introduced steric error, based on the van der Waals’ radius of the helium atom (1.3 A.), yields a correction of 0.013 cc. per gram, to be applied to the measured specific volume of the solid per 100 sq. meters per gram surface area of the sample. From the data of Steele and Halsey (6,6)on room temperature adsorption of helium, Steggerda has calculated a correction of 0.009 cc. per gram for 100 sq. meters per gram surface area. Since the two corrections act in opposite directions, the net correction would be 0.004 cc. per gram in the case considered. For a sample of alumina with 200 sq. meters per gram surface area, this would mean that the skeletal density obtained with helium is 2 t o 3% too low. Recently Joyce published on the air comparison pycnometer (4). He recognized the fact that incorrect volume
Table VI.
Material
Summary of Helium and Water Skeletal Densities
Weight Gain Due t o Atmospheric Exposure, %
Experimental Skeletal Densities, Gram/Cc. ‘Helium Water 3.30 3.6 3.05 3.0 3.64 3.46 3.17 3.18 3.56 3.80 3.14 3.18 2.25 2.21 2.09 2.10
Aged A1203-M~Os Fresh Al~O~-MoO.-CoO Aged A I 2 O 8 ~ M ~ O 8 % ~ 0 Fresh SiOz Aged Si02
Table VII.
Sample Number 2 128 33 34 67 129 75 90
Theoretical Expectation with Water, Gram/Cc. 3.65 3.05 3.65 3.18 3.84 3.18
... .,.
Comparative Skeletal Densities Obtained with Helium, Water, and Carbon Tetrachloride
Material
Treatment O
c.
800 800 593 550 593 550 500 593
readings are obtained with the air comparison pycnometer when substantial amounts of air are adsorbed by the samples, His results are readily understood and explained in the light of the present work. In a subsequent paper (8) we are going to show that the extent of room temperature adsorption of nitrogen and oxygen on inorganic oxides depends on the moisture content of the sample. Starting with a moderately hydrated sample, the adsorption of nitrogen or oxygen can be increased by more than 100% when water is removed by a suitable heat treatment. The samples of Joyce obviously contained various amounts of water. Part of the water was removed when the samples were evacuated before admitting helium to them, hence the slightly higher skeletal densities in the second series of measurements with helium. Still higher skeletal densities were obtained after heating which indicates that more water was removed. This change varied from one sample to another. Thus, in case of sample A , the skeletal density increased only O.60/, due to the heating, suggesting that sample A had been well dehydrated before the measurements were undertaken. The same conclusion is reached by another consideration. With air, the skeletal density was almost 80% too high for sample A indicating a large amount of air adsorption on this sample, and thus a well dehydrated
Weight Gain Due to Atmospheric Exposure,
%
11.6 0
0.3 13.9 14.5 0
16.6 0
Skeletal Densities, Gram/&. Carbon tetraHelium Water chloride 2.97 3.09 2.83 3.46 3.67 3.51 3- - .. 36 3.63 3.44 2.82 2.88 2.87 2.79 2.83 2.77 3.45 3.47 3.63 2.83 2.86 2.84 3.60 3.61 3.84
sample. For sample B , air yielded a skeletal density which was only 6.7% too high. The small amount of air adsorption suggests a poorly dehydrated sample B , and the large (19%) increase in skeletal density due to heating confirms the assumption. The results for samples C and D indicate a state of dehydration intermediate between A and B. ACKNOWLEDGMENT
The authors are indebted to R. A. Herrmann and E. 0. Ernst for performing the heat of immersion measurements, and to W. B. Innes for helpful discussions. LITERATURE CITED
(1) de Boer, J. H., University of Delft,
Holland, personal communication, 1961. (2) Guderjahn, C. A., Paynter, D. A., Berghausen, P. E., Good, R. J., J . Phys. Chem. 63, 2066 (1959). (3) Innes, W. B., ANAL. CHEM.23, 759 (1951). (4) Joyce, R. J., The Analyzer 2, No. 4, 11 (1961). (5) Steele, W. A., Halsey, G. D., Jr., J . Chem. Phys. 22,979 (1954). (6) Steele, W. A., Halsey, G. D., Jr., J . Phys. Chem. 59,57 (1955). (7) Steggerda, J. J., “De vorming van actief aluminiumoxyde,” pp. 50-8, Uitgeverij Excelsior, Delft 1955. (8) Tuul, J., Innes, W. B., ANAL.CHEM. 34, 818 (1962). RECEIVEDfor review January 19, 1962. Accepted April 9, 1962. VOL. 34,
NO. 7, JUNE 1962
817
