CONCLUSIONS
The present study has demonstrated the feasibility of using added foaming agents to fractionate organic compounds which show little inherent ability to foam (and have relatively low surface activities). The fact that the product is contaminated with an added agent should be relatively unimportant in analytical determinations. However, if charged agents are used in preparative work, removal should be rapid and complete using a short column of ion exchanger. Foam fractionation should receive more attention from analysts. The simplicity of the technique and the fact that separability is usually improved in dilute solutions make foam fractionation attractive for trace analysis. Studies designed to evaluate structural effects on enrichment probably should be carried out using simple
solutes and agents, but there is little question that ultimately the technique will be more useful in dealing with the more complex systems to which it was originally applied-i.e., those involving large polar heat-sensitive molecules of the type encountered in biological samples. LITERATURE CITED
(1) Cassidy, H. G., “Technique of Or-
ganic Chemistry,” Vol. V I Interscience, New York, 1951. (2) Cassidy, H. G., Zbid., Vol. X, 1957. (3) Catoggio, J. A., Olver, J. W., Rogers, L. B., unpublished data, 1961. (4) Gaden E. L. Jr., Schnepf, R. W., J . B i o c h ~idicrobiol. Technol. Eno.
1,1(1959). ( 5 ) Jirgensons, B., Strsumsnis, M. E.,
“A f3,)jm-t Textbook of Colloid Chemistry, Wiley, New York, 1958. (6) Kenrick, F. B., J . Phys. Chem. 16, 513 (1912).
(7) London, M., Cohen, M., Hudson, P. B., Arch. Biochem. Biophys. 46, 141 ( 1953).
(8) London, M., Cohen, M., Hudson, P. B., Biochem. et Biophys. Acta 13, 111 (19541. (9) Perri, J. M., Hazel, F., Ind. Eng. Chem. 38, 549 (1946). (10) Sargent, R., Rieman, W., 111, J . Om. Chem. 21, 594 (1956). (11) ’schne f, R. W., Gaden, E. L., Miroczni!, E. Y., Schonfeld, E., Chem. Eng. Prog. 55, 42 (May, 1959). (12) Schutz, F., Trans. Faraday SOC. 38, 85 and 94 (1942). (13) Thurman, W. C., Brown, A. G., McBain, J. W., J . Am. Chem. Soc. 71, 3129 (1949). (14) Walling, C., Ruff, E. E., Thornton, J. L., Jr., J . Phys. Chem. 56, 989 (1952). \ - - - - , -
RECEIVED for review December 12, 1960. Accepted May 5, 1961. Taken in part
from a thesis submitted by Barry L. Karger in partial fulfillment of the requirements for a B.S. degree, Massachusetts Institute of Technology, May 1960. Work was supported in part by the Atomic Energy Commission under Contract AT(30-1)-905.
Particle Size Distribution Measurements below Five Microns R. R. IRAN1 and E. F. KAELBLE Research Department, Inorganic Chemicab Division, Monsanto Chemical Co., St. louis 66, Mo.
b Centrifugal sedimentation and electron microscopy coupled with electronic counting and sizing are intercompared for the measurement of particle-size distributions of calcium phosphate and silica below 5 microns. Weight-size distributions from the two techniques agree within experimental error for materials whose particle shape does not deviate significantly from sphericity. Similar agreement was observed on flour, clay, and glass particles.
P
from this laboratory on particle-size technology have dealt primarily with the size range above 5 microns (3,4, 7, 8). This paper compares centrifugal sedimentation and electron microscopy for determination of particle-size distributions below 5 microns and in the submicron range. REVIOUS PUBLICATIONS
EXPERIMENTAL
Sedimentation. The apparatus was a commercial unit manufactured by Mine Safety Appliance Co., Pittsburgh, and has been described in detail previously (12). It utilizes the layer sedimentation technique (9), so that the amount of particulate matter collected at the Centrifugal
1168
ANALYTICAL CHEMISTRY
bottom of the sedimentation tube is directly proportional to the per cent by weight of particles with a size equal to or greater than that computed from Stokes’ equation. The speeds of the various centrifuges were checked using a tachometer, and agreed well with the manufacturers’ values. However, the time corrections for a centrifuge start-up and stoppage must be determined every 2 to 3 months. Obviously, the minimum t i e for running any centrifuge is the time required to reach full speed. One of the assumptions in this method is that sediment height is proportional to sediment weight. When complete dispersion of particulate matter is achieved, this is a good assumption because monosized particles are essentially settling at any one time, and void space is independent of size for monodisperse systems. However, when strong aggregates of smaller particles exist, the sediments compact more as higher centrifugal fields are used and bulk density correction factors must be used (8). For example, 2-micron silica particles are strong aggregates of many ultimate particles with a size of 0.01 to 0.03 micron. Therefore, in all determinations, the product of sediment volume times bulk density a t the specific revolutions per minute of centrifugation was taken as proportional to sediment weight. Absolute ethyl alcohol was the sedi-
mentation fluid, and the concentration of particulate matter in the suspension was 0.1% by volume. Spatulation (10) was utilized to disperse the powders, followed by vigorous stirring in a highspeed homogenizer. Electron Microscopic Technique. The electron photomicrographs mere made with an RCA EMU-3E microscope. Magnification mas 4080 diameters, as determined with a carbon replica of a ruled 28,800-line-per-inch diffraction grating. Suspensions of calcium phosphate and silica containing 0.1% solids were sprayed from a compressed air atomizer onto thin nitrocellulose films supported on 200-mesh specimen screens. Absolute ethyl alcohol was the dispersion medium for calcium phosphate, and silica was suspended in 2-ethylhexyl alcohol. Dispersion was accomplished by vigorous stirring in a high-speed homogenizer. The photomicrographs were recorded on 3l/, x 4 inch plates, and positive plates were prepared for counting and sizing purposes. The electron photomicrographs were counted and sized (size corrected for magnification) using a Cintel flying spot particle resolver manufactured by Cinema Television, Ltd., London ( I , 11). In all determinations, a total of a t least 2000 particles were counted and sized to give the number-size dis-
,tribution. The resulting number-size distributions were converted to weightsize distributions as follows:
*'
'
Figure 1. Electron photomicrogroph of tricalcium phosphate partides at a magnification
of 4080
.
:.
.
.
,,%
..
,
,b'.
..**:
. .. . . -.... *.. -.:
..
'
"
.8
4
where y is the per cent by weight greater than the size M ,and f( is the frequency of occurrence by number of the size Mi. I t was assumed that the particle density is independent of size, and the summation terms were evaluated by perfnrming trapezoidal integration on the histograms.
'.'
- ..... .. ?. ':e. '. . . - . .-*, ... . e.- \ .... . .._ .,+,' - - .-. \ . ,
i' . I ' .I
*.
' Y
'8
* .
&..
.
.r'
.$
*.
, + I * . '
RESULTSAND DISCUSSION
Table I is a tabulation of the particle-size distributions obtained on a c:ilcium phosphate and a silica sample. Statistical analysis of the data from the t\vo runs on each material shows that the per cent by number greater than any s1:ecific size can be determined rrith the electron photomicrograph-electronic counting and sizing technique with a precision of *4.4% absolute at the 95% confidence limits. The per cent by weight gr.eater than any specific size as' obtained from the weight-size transforms of the numbersize distributions has a precision of 5~9.7% absolute. These precision limits are consistent with data obtained on a large number of samples, each run several times. The cause for the lowerprecision of \\-eight transforms is obvious from Equation 1. Weight-size distributions are very sensitive to the very coarse region of the number-size distribution (6). Thus, a single 10micron particle has the same contribution to the weight-size distribution as 1000 1-micron particles. Since the precision of counting particles decreases as the number of particles being counted decreases, the coarse end of the numbersize distribution is not very precise, causing relatively larger errors in the weight transforms. However, the weight transforms of number-size distributions are very well resolved in the smaller size end of the distribution. The precision for determining the per cent greater than a specific size from centrifugal sedimentation was 14.9% absolute a t the 95% confidence limits. Obviously, if number-size distributions are calculated from these data, the prwision suffers. As can be seen from Figure 1, the particles of the materials studied do not deviate considerably from sphericity. Hciss and Coull (8) found that the settling velocities of particles having the same degree of sphericity were within a few per cent of the settling velocities of spheres having an equal volume. Therefore, for this material
sedimentation and microscopy should give equivalent particle size distributions vithin experimental error (6). This is consistent with the experimental measurements that show data from the two independent techniques to agree within experimental errors. When the same techniques described in this paper were applied to oddshaped particles--e.g., rods or platesin the size range of 0.05 to 1.0 micron, the microscopic technique consistently gave a coarser particle-size distribution (6). This is expected because the particles tend to sit on the miscroscope grid in the most stable position, which is also that of a relatively large projected area. Also, the settling veloci-
ties of oddahaped particles are not equal to those of spheres of an equal volume. Nevertheless, when shape factors were found to be independent of size, correction factors could he utilized to make the two techniques agree with one another within experimental error. The adoption of one technique in preference to another must he based on several considerations. The precision required in a specific range, the cost of equipment, time per analysis, and caliber of technicians are some of the factors that must be considered. The end use of the data is also important. AE a routine quality control tool, the centrifugal sedimentation technique is more applirable in the size range over
Table I. Particle Size Distribution of Calcium Phosphate and Silica Particles Size,
Per Cent Greater than By weight* Runl Run2 Run2 CALCIUM PHOSPHATE
By weight$
By number.'
Microns
Run 1
0.1 0.2 0.3 0.5 0.7 1 .o 1.5 2.0 3.0 5.0 7.0 10.0
92.4 85.0 77.6 63.9 51.0 35.2 16.5 9.2 3.4 0.28 0.11 0.06
95.5 87.0 80 65 53 37 16.7 10.9 4.8 0.2 0.07 0.04
0.1 0.2 0.3 0.5 0.7 1.0 1.5 2.0 3.0 4.0 5.0
92.4 82.1 73 58 47 31 16 8.5 2.3 0.5 0
90.0 80.1 70 60 44 34 13 6.7 1.1 0.2 0.0
99:99 99.94 99.6 98.8 95.6 86.0 74.0 46 23 11 3
99:9s 99.92 99.6 98.4 94.0 87.0 78.0 50 13 9 2
Runl
Run2
... ...
... ... ...
... 99.0
... 97.0 ... 83.0 50 19 10
2.5
99.5
... ...
99.0 82.0 '55 22 12 3.0
SILICA
e
6 6
99.99 99.97 99.91 99.4 97.2 91 76 58 24 1 0
99.99 99.96 99.85 99.0 97.0 90 68 50 14 3 0
... ... ... 98.4 ... 92 ... 54 20 0.1 0
...
... ...
98.4
... ...
89
51 18 0.1 0
From electron photomicrographs and electronic counting and sizing. Weight transforms of number-size data. From centrifugal sedimentation.
VOL 33. NO. 9, AUGUST 1961
1169
0.5 micron, requiring less expensive equipment and offering more foolproof operation. Yet, in few cases*.g., metal oxides-this technique could not be utilized successfully because of the large bulk density correction factors. When 100% of the particles are below 1 micron, the technique utilizing electron microscopy is much more practical, requiring only a fraction of the time per determination needed for centrifugal sedimentation. Obviously, the particular centrifugal sedimentation technique utilized in this work cannot be used for materials with particles below 0.1 micron because of the excessive time
required per determination and inaccuracies due t o Brownian motion. ACKNOWLEDGMENT
The authors thank J. R. Condray and W. W. Morgenthaler for running some of the experiments. LITERATURE CITED
(1) h e s , D. P., Irani, R. R., Callis, C. F., J.Phys. Chem. 63,531(1959). (2) Heiss, J. F.,Coull, J., J. Chem. Eng. Progr. 48, 133 (1952). (3) Irani, R. R., ANAL. CHEM.32, 1162 119601. (4) Irani, R. R.,Cereal Sci. Today 6, 35 (1961).
(5) Irani, R. R., J. Phys. Chem. 63, 1603 (1959). (6) Irani, R. R., Ames, D. P., ASTM Bull., in press. . (7) Irani, R. R., Callis, C. F., ANAL. CHEM.31,2026(1959). (8) Irani, R.R., Fong, W. S., Cereal Chem. 38,67 (1961). (9) Marshall, C. E., Proc. Roy. SOC. (London),Ser. A 126,427(1930). (10) Michaels, A. I., Weaver, T. L., Nelson, R. C., ASTM Bull. No. 247, 140 (1960). (11) Young, J. Z.,Roberts, F., Nature 167.231 (1951). (12) Whitby, K.'T.,Heating Piping Air Conditioning 61, 33 (1955); 61, 449 (1955). RECEIVEDfor review January 30, 1961. Accepted June 5, 1961.
Determination of Mixed Lead Alkyls in Gasoline by Combined Gas Chromatographic and Spectrophotometric Tech niques W. W. PARKER, G. Z. SMITH, and R. L. HUDSON Ethyl Corp., Baton Rouge, l a . ,An accurate and simple method for the determination of mixed (methylethyl) lead alkyls in gasoline is described. This rapid method utilizes gas chromatography to separate the lead alkyls and a dithizone lead procedure to measure them.
N
o
SIMPLE METHOD has been published for the determination of mixed (methyl-ethyl) lead alkyls in gasoline. The elution characteristics of gasoline mixtures prevent the use of normal chromatographic techniques. However, chromatography may be combined with a spectrophotometric analysis to effect a rapid and accurate determination. The lead alkyls are separated by a chromatographic column, individually collected in iodine scrubbers, and measured by a dithizone spectrophotometric lead analysis procedure (1, 3, 4) . A simple mixture of T M L (tetramethyllead) and T E L (tetraethyllead) may be analyzed in approximately 25 minutes with an accuracy of *2%. This is accomplished by a direct dithizone analysis of the sample for total lead, followed by similar analysis of the eluted TML. The T E L is then calculated by the difference between the total lead and the lead present as TML. While the method is primarily designed for the determination of mixed TML-TEL in gasoline, it can easily be adapted to the analysis of more complex mixtures containing in addition MeJPbEt (trimethylethyllead), Mer
1 170
ANALYTICAL CHEMISTRY
PbEtl (dimethyldiethyllead), and MePbEta (triethylmethyllead). This mixture in gasoline requires five scrubbers and can be analyzed in 11/* hours. Elution times for each of the five components are shown in Figure 1. There may be temperature, gas flow rate, or packing variations that could vary the elution times using apparently similar columns. The only critical time involved is the 3.75-minute time for exchanging the T M L and Me3PbEt scrubbers. This time can be easily determined for any column from its relationship to retention time of toluene: Mid-point time between TML and MesPbEt = toluene retention time X 1.30
The precision of the method was calculated at three lead concentrations. The standard deviation a t 1 pg. is A0.49 pg. At 4 pg. it is +0.43 pg. and a t 50 pg. it is *0.66 pg.
1 1 60
PEAK
EXPERIMENTAL
Apparatus. A Perkin-Elmer Vapor Fractometer, Model 154, was used t o separate the lead alkyls. The column used t o effect the separation was a short 1-foot by '/d-inch column packed with 40% Nujol on 35-80 mesh Chromosorb. The helium flow rate was 100 cc. per minute and a chart speed of '/z inch per minute was used, The short column and a low temperature (70" C.) were used to minimize possible decomposition of the organolead compounds. The absorption cells used as gas scrubbers and the cell container of the Beckman DU spectrophotometer were modified as described by Henderson and Snyder (2). Reagents. Solution A. Dissolve 12.5 grams of ammonium citrate, 5 grams of potassium cyanide, and 30 grams of sodium sulfite in 200 ml. of distilled water. Dilute to 1 liter with concentrated ammonium hydroxide.
Pi
ELUTIW
NO COMPD.TIME. MIN I
TML
2 3
MeSPbEl
4
MezPbEt2 MePbEI,
5
TEL
375 40 11 20 45
50 TIME, MINUTES
Figure 1.
Chromatogram of a mixture of lead alkyls