proximately 300 mg. They can be weighed as quickly and reproducibly as platinum boats and without error due to static charges. The use of the capsules has many advantages in the carbon and hydrogen determination. A sample can be dried in a weighed capsule, sealed completely in a dry atmosphere, and weighed without the need of an additional container, such as a glass pig, to protect the dried sample from moist air. Because the capsules, unlike glass pigs are small and light-weight, they can be weighed on the 1-gram capacity electrobalances used for automated combustions. The capsules are air-tight if properly sealed and are not opened outside the combustion tube after the sample is added; therefore, a sample can be preweighed, easily stored, and analyzed when convenient. Unstable compounds can be encapsulated when freshly prepared; hygroscopic samples can be weighed under optimum conditions and transferred to the combustion tube without change in water composition; and liquid samples can be encapsulated without loss by heat decomposition or vaporization. Indium encapsulation is also faster and simpler for liquids than glass encapsulation. After the capsule is
inside the combustion tube, the combustion system may be purged without loss of volatile sample if the temperature in the capsule area is not high enough to soften the indium and release the sample. Indium capsules collapse in the combustion tube a t approximately 200’ C., a temperature below that required for complete combustion of most organic compounds. Though many compounds vaporize out of the capsule when the indium melts and the seal is broken, partially decomposed sample could be trapped in the melt. The results (Table I) and those obtained with National Bureau of Standards standard samples (anisic acid, benzoic acid, dextrose, sucrose, acetanilide, cystine, 2-chlorobenzoic acid) show that this did not occur. The average percentages were within 3=0.3% of theory. The overall errors were positive, 0.09% above theory for carbon and 0.06% for hydrogen. Indium usually did not spatter or leave a deposit on the wall of the combustion tube even a t maximum burning temperature (750’ C.) except when the encapsulated sample was a solid containing sulfur, iodine, bromine, or chlorine. When these samples were combusted in the capsules, the reaction
a t about 700” C. was sometimes sufficiently vigorous to ignite the melted ifidium to incandescence with a very large increase in the use of oxygen and the formation of a voluminous residue. However, the results were acceptable for the sulfur and halogen compounds analyzed, except for 2-iodobenzoic acid. Consistently good results were obtained for this compound only when it was combusted with cobalt oxide inside the capsule, or when the capsule ignited during the combustion. When the liquid samples containing sulfur or halogens, reported in Table I, were combusted in the capsules neithcr incandescent combustion nor troublesome residue occurred. LITERATURE CITED
(1) Ehrhardt, C. H., Moeller, W. H., Grubb, H. iM.(to Standard Oil Co. of Indiana), U. S. Patent 3,103,277, September 10. 1963. (2) Nerheim; A . G., ANAL. CHEM.36, 1686 (1964). (3) Nerheim, A. G. (to Standard Oil Co. of Indiana), U. S. Patent 3,063,286, Novgmber 13, 1962. Reference to a company or product name does not imply approval or recommendation of the product by the U. S. Department of Agriculture to the exclusion of others that may be suitable.
Computerized Procedure for Editing Data from a Mass Spectrometer Digitizer H. Evert Howard, Union Oil Company of California, Union Research Center, P. 0. Box 76, Brea, Calif. 92621 PRIMARY
ADVAKTAGE
of
mass
A spectrometry is the production of a
wealth of fundamental data. However, many mass spectrometric (MS) methods do not require, nor can they use, all of the raw data produced by the MS. Hence a means is needed for identifying the mass numbers of those MS peaks required for any given analysis. A mass spectrometer digitizer (“LISD” manufactured by Non Linear Systems of Del Mar, Calif.) is used in this laboratory to digitize the mass spectra from a Consolidated Electrodynamics Corporation 21-103 mass spectrometer. This R‘ISD, described by Thomason ( I ) , reads out the height of each RIS peak along with mass identification in terms of ion accelerating voltage. When accelerating voltage is known, mass number may be calculated using the following well known equation:
where R = radius of curvature of the ion beam, B = magnetic induction, and E = ion accelerating voltage. This equation may be simplified to K M/e = E 946
0
ANALYTICAL CHEMISTRY
where K
=
R2B2
- and B is fixed. 2
Therefore, K = E.mass number (3) Thomason (1) reported the derivation of average K factors from known sample runs for different magnet ranges. Using these average factors he obtained mass number reproducibilities of *0.1 mass number up to mass 125 and i=0.25 mass number up to mass 250. Our experience indicates that very diligent control of instrumental conditions (magnet strength, etc.) must be maintained, along with appropriate correction factors, if mass identification is to be kept within these limits using average K factors. Theoretically a K-factor for a given magnet range will not change. However, due to temperature variations in the magnet environment, geometry of the MS ionization chamber and analyzer tube, magnet hysteresis problems, etc., K-factors do change, although not abruptly. Hence this change must be accounted for or mass number identification will be in error. Our approach t o the solution of this problem is described as follows. Preceding the analysis of any group of samples a calibration run on a suitable pure compound (n-butane for gas analysis and n-heptane for gasoline samples) is first obtained. An initial
K-factor is calculated by application of Equation 3 to any readily identified peak in the spectrum. The mass spectrum of the first sample for analysis is next obtained, and the initial Kfactor is used to identify the first mass number ( K / E = mass number) required by the analysis. When more than one MS peak is found for this or subsequent mass numbers, the largest peak is assumed t o be the one desired, on the premise that metastable ions, noise spikes, etc., are always of a loner magnitude than the whole mass ions. After the first required mass number is identified as above, a new K-factor is computed by application of Equation 3 to this mass number. This new Kfactor is then used to identify the next higher mass number required in the analysis, in the same way that the first required mass number was identified. This process is repeated, with intermediate calculation of K-factors for all mass numbers used in the analysis, until all of the mass numbers required for the sample in question have been identified. In this way all “normal” changes in K-factors which occur during the mass analysis of a given sample are sensed and maximum accuracy in the identification of each mass number is maintained. Any “abnormal” changes
in K are tested for by comparing adjacent K values. X change in K by more than 1 part in 300 between adjacent mass values is assumed to represent an abnormal change in K-e.g., discontinuous momentary shift in K-and the immediately preceding K value is used for the following mass value determination-i.e., the abnormal value of K is rejected. The first computed K factor for each sample is used as a starting K-factor for analysis of the next sample, and the overall process is continued until all samples are analyzed. All of the arithmetic operations are carried out by an IBM 1620 computer program in this laboratory.
specific mass numbers starting a t mass 43 and ending a t mass 170. The Kfactors computed during a typical day's production of these analyses vary between two extremes--e.g., 401,000 to 403,000. If either value had been used to determine a mass number where the other K-factor was correct, the mass unit would be incorrectly identified, with a resultant error in the final analysis of the sample. This illustrates the potential errors that can arise by the use of average X-factors.
At the time of this writing we have analyzed hundreds of samples, both gas and gasoline, by means of the described editing routine without a single failure. There are several innovations in the procedure, but the key to its success is the repetitive calculation of new Kfactors each time a mass number is identified. The outstanding advantage of this procedure, as compared to that of Thomason ( I ) , is the certainty of mass number identification (with this burden carried entirely by a computer) while requiring only normal MS operator effort. The gasoline hydrocarbon type method used in this laboratory requires
LITERATURE CITED
(1) Thomason, E. XI., ANAL. CHEM.35, 2165 (1965).
High Sensitivity Oxygen Analysis of Metallic Samples with Fast Neutrons Richard W. Benjamin, Kenneth R. Blake, and Ira L. Morgan, Texas Nuclear Corp., Austin, Texas
for oxygen in metallic T samples has been performed for some time by using fast neutron activaRACE AKALYSIS
tion analysis techniques in conjunction with a pneumatic rapid transfer system ( I , 3, 4). h system has now been designed which is capable of measuring oxygen in concentrations below 10 p.p.m. in cylindrical metallic samples 3//16 t o 0.480 inch in diameter and '/*to 1 inch long. Oxygen activation analysis with fast neutrons may be done accurately and rapidly with the O16(n,p)N16 reaction. Because of the short half life of the 6 " activity, it is advantageous to use some sort of rapid transfer system, such as has been described in some detail by hnders and Rriden ( I ) . h 150-k.e.v. Texas Xuclear Corp. Cockcroft-Walton positive ion accelerator provides an abundant source of 14-m.e.v. neutrons through the H3(d,n)He4reaction. The target used on the accelerator consists of 3 to 5 curies of tritium adsorbed onto a titanium layer which has been vacuum-deposited on a copper backing. The target assembly appears in Figure 1. The deuteron beam impinges on the tritium target (A) which is held against an O-ring by the retaining ring ( B ) ,thus sealing the accelerator vacuum system. The target is cooled by water which flow? directly over the copper backing. This allows higher beam currents and a longer target life. The thin stainless-steel sample tube (C) is fitted flush against the retaining ring, ensuring that the sample is exposed to the highest posijle neutron flux. The sample is centrally and reproducibly constrained in the tube by a polyethylene finger (D), the inside diameter of
PLEXIGLAS TUBING \
w
I
\ L J yA
" B" RETAINING RING
3
-
"A" TRITIUM TARGET
,,,,>
''
8
8
"C" STAINLESS STEEL SAMPLE TUBE
,
I
SILVER SOLDER
PRESSURE FITTING FOR NITROGEN GAS
POLYETHYLENE FINGER. WATER FITTING FOR TARGET COOLING \
b Figure 1.
Target assembly VOL. 38, NO. 7, JUNE 1966
947