applications. Even with a 10-inch linear rule, components of highest molecular weight can be read directly to 0.570. This corresponds to a n absolute displacement of about 0.5 mm., the smallest interval between graduations on the 1OL scale. Estimation to 0.25% is, therefore, possible. For most components, estimation to 0.1% is feasible. Such accuracy is possible because
all possible ranges of composition need not be covered. Where all possible compositions must be allowed, the largest scale must be set equal to the full length of the rule. If the components exhibit a wide range of molecular weights, the smaller scales (representing components of high molecular weight) might be badly compressed. This difficulty can be overcome in the circular
rule by choosing some intermediate scale as standard and handling by parts compositions that ,cannot be incorporated on the longer scales. It may be necessary to “overrun” the zero setting in some summations, but this presents no difficulty. Issued as N.R.C. No. 5091.
Concentration of Impurities from Organic Compounds by Progessive Freezing Joseph S. Matthews and Norman
D. Coggeshall, Gulf
i t is necessary to determine 0 trace amounts of impurities in organic compounds. Mass spectrometry FTEN
and ultraviolet or infrared spectroscopy have limits of sensitivity and are not always applicable. The usual procedure, when concentrations are lorn, is to concentrate the impurities by distillation or extraction prior to instrumental analysis. These methods have certain limitations. Crystallization by fractional freezing, fractional melting, or zone melting techniques has come to the fore as a means of obtaining ultrahigh purity. Starting with Schm-ab and Kichers (9, 10) various workers have shown that such methods may be used to purify organic compounds (1, 8, 4-6, 8 ) . Goodman (3) suggested t h a t they might also be used to concentrate impurities that occur in high dilution. I n 1957 Schildknecht and Mannl (7) concentrated small quantities of biological material from aqueous solution by zone melting the frozen solution. Progressive freezing can conveniently be used t o concentrate impurities from organic liquids.
Research & Development Co., Pittsburgh, Pa.
frozen is lowered into the freezing bath, TI hile the unfrozen part is stirred by a motor-driven glass stirrer. The tube is lonered until 1 ml. or less of liquid remains unfrozen a t the top of the tube. If all the liquid solidifies, a eniall amount is melted. The liquid is lvithdran n with a long dropper pipet, transferred to a vial, weighed, and analyzed. After removal of the fraction, the tube contents are melted and the cycle is repeated if desired.
wire frame which holds the sample tube and allows it to be stirred while being lowered into the cooling bath. The stirrer motor remains in a fixed position. The top of the sample tube is closed by a rubber stopper fitted with a Teflon sleeve t o serve as a bearing for the stirrer shaft. The cooling bath consists of a copper pipe, 2 inches in outside diameter by 14 inches long, insulated on the outside with glass fiber pipe insulation, and cooled by circulating acetone cooled by a dry ice-acetone bath. The temperature is regulated by adjusting the coolant f l o ~rate. The coolant is maintained a t a constant level by an overflow through which the circulating acetone returns to a reservoir.
Several parameters n-ere studied: effect of stirring, lowering speed, temperature of cooling, and size of fraction removed, and effect of impurity concentration on efficiency of separation of impurities.
EXPERIMENTAL
RESULTS AND DISCUSSION
The tube containing the liquid to be
Stirring was necessary for increasing
Figure 1. Freezing apparatus
LOWERING APPARATUS
The apparatus consists of a lowering mechanism and a cooler. The lowering mechanism is a small synchronous motor with a speed of 1800 r.p.m., geared down through a worm drive which passes the torque on to a ball and disk integrator that serves as a variable speed reducer. The ball cage of the integrator is attached to a calibrated micrometer screw, so that the ball can be moved across the face of the disk to vary the output speed. The torque from the speed reducer is transferred through a worm drive to the output shaft, which can be regulated between 0.25 and 2.87 revolutions per hour. A spirally grooved drum attached to the output shaft carries a very thin metal cable for lowering the sample tube into the cooling bath. The range of speeds at which the cable is unwound can be controlled by the size of the cable drum. With a drum of 1-inch diameter, the speed of the cable can be regulated between 4 and 16 em. per hour. Attached to the end of the cable is a
1 124
ANALYTICAL CHEMISTRY
STIRRER
MOTOR
N
DRY ICE COOLING
ACETONE
PbMP
the efficiency of separation; without stirring impurity concentration was only slightly enhanced. Slower lowering speeds were more efficient than faster. Variations of 10" in the cooling temperature had little effect on the efficiency of concentration. Removing relatively small samples (1/40 to 1/60 of charge) obtained higher impurity concentrations in the samples removed. This takes advantage of the impurity distribution in a liquid R-hich is progressively frozen. This process, however, is best suited for low initial concentrations of impurities, as efficiency decreases as impurity concentration increases. The primary purpose of these studies is the concentration of impurities for identification. I n the identification of impurities in a sample of reagent grade benzene, benzene containing 0.25 weight yo impurities mas placed in a tube 28 mm. in inside diameter and stirred while being lon-ered a t 5 cm. per hour into the freezing bath a t -25" C. As a preliminary concentration step, a large (20%) fraction Tyas removed from the first cycle of the benzene charge. Analysis showed that it contained 65% ?f the total impurities in the charge. Thls fraction was then recycled in a tube 22 mm. in inside diameter and about 1 gram was removed and analyzed. This small fraction contained 44% of the impurities present at the start of the cycle and the concentration of impurities had been raised to 5.14 weight %, equal to a 20-fold increase over the original concentration of impurities, Analysis of the initial benzene charge revealed no detectable quantities of C7 paraffins or C8 and Cg aromatics (Table I). The small sample removed from the second cycle, hon-ever, reveals that all these were present in the original charge, although a t undetectable concentrations. As the material balance for the over-all experiment wis
Table
1.
Concentration of Impurities in Reagent G r a d e Benzene
Weight, g. Impurities, n-t. yG Impurity composition, wt. % Toluene Minimum Ce cyclics and/or mono-olefins Minimum C, crclics and /or mono-olefins" Minimum Ce paraffins Minimum C, paraffins Ca aromatics Cg aromatics
First Cycle Initial 1st charge fraction 73.3315 14 3033 0.25 0.81
Second Cycle Charge 13 6777 0 81
1st
fraction 0 9487
Calcd. Initial Conrn.. Wt. %' ~
~
~~~~
5 14
0.05
0.22
1.25
0 06
0.06
0 27
1.46
0.07
0.03 0.11
0.13 0.19
0.66
0.03 0.02
0.00 0.00 0.00
0.00 0.00 0.00
98.8%, i t is possible to calculate the concentration for these groups in the original charge stock. It is assumed that in the concentration procedure all impurities are affected equally-i.e., the relative composition of the mixed impurities remains the same and no impurity is concentrated preferentially. On this basis, the calculated concentrations for Cs and Cg aromatics in the initial charge are about 0.0005 and 0.001 weight %, respectively. This explains \Thy these groups were not detected in the initial charge by mass spectrometry. The C7 paraffins are calculated to be a t an initial concentration of 0.06 weight yo, but they are not detected in the initial charge, because the parent mass spectral peak for the C7 paraffins ( m / e = 100) is masked b y the mercury peak (m/e = 100) when the C7 paraffin concentrations are low. (Mercury is present, because of its use in the sample introduction system of the spectrometer.) However, progressive freezing raised the concentration of the C7 paraffins to a level rrhere the peak was distinct and easily measured. The parent peak for the C6 paraffins \\-as contributed to by the undetected C7 paraffins; this explains lvhy the calculated value
... . .. ... ... ...
0.45 1.2'3
001 0.02
0.06 0.0005
0.001
for the Csparaffins is much lower than what was found. The total of the impurity concentrations calculated to be initially present in the charge (excluding the very low values for Ce and CS aromatics) is 0.24%. while that actually found in the initial charge is 0.25%. ACKNOWLEDGMENT
The authors thank the Geophysical Development Division, Gulf Research Bt Development Co., for the design and construction of the lon-ering mechanism. LITERATURE CITED
(1) Dickinson, J. I]., Eaborn, C., Chem. & Znd. (London) 1956, 959. (2) Glasgow, A. R., Ross, G., J . Research A7at2. Bur. Standards 57, 137 (1956). (3) Goodman, C. H. L., Research 7, 177
(1954). (4) Handley, R., Herington, E. F. G., Chem. & Znd. (London) 1956, 304. (5) Herington, E. F. G., Handley, R., Cook, A. J., Ibid., 1956, 292. (6) Rock, H., Natur~cissenschaften 43, 81 (1956). (7) Sc'hildknecht, H., AIannl, A., Angew. Chem. 69, 634 (1957). (8) Schumacher, E. E., J . Metals 5, 1428 (1953). (9) Schwab, F. W., Kchere, E., J . Research Natl. BUT.Standards 25. 747 (1940). (10) Zbid., 32, 253 (1944).
System for Sampling the Gases in Processed Transistors C. Gordon Peattie, G. R. Kornfuehrer, R. 6000 Lemmon Ave., Dallas 9,Tex. of a study of the processes occurring within encapsulated and canned transistors during aging tests, the problem arose of analyzing the gas in the transistor can. Because this gas ll-as believed to be present in small amounts and was of unknolyn composition, mass spectrometry was chosen. the problem One Of sampling the gas comprising the atmosphere within a transistor can, a system was devised t o permit crushing of the transistor can in an evacuated system ASPART
E. Trueb, and J.
R. Moreland, Texas Instruments Inc.,
and subsequent transfer of this gas to a mass spectrometer. The cEc h ~ 21-401 ~ d ~spec-~ trometer was used in the analytical, rather than the isotope, mode. T o avoid exposing the sample gas to selective absorption b y the stopcock grease used in the inlet system of the spectrometer, the gas was admitted directly intotheionsource, Inamassspectrometer having a grease-free inlet system, this would have been unnecessary. The sample cylinder was a stainless steel tube 6 inches long and 0.460
inch in inner diameter with a mall thickness of 0.020 inch, closed a t one end with a helium-shielded arc weld. to a The Other end was flange having a 1.25inch face* Each cy1inder was hydrogen-fired and of the The glass Portion before use. inlet system also had a stainless steel flange sealed to it. With a Teflon 0ring and a n interthreaded coupling, a n easily demountable, vacuum-tight joint mas obtained. Three transistors, fastened to a strip VOL. 31, NO. 6, JUNE 1959
1125