Table I. Comparison of Photon Activation Analysis Results with Certified Values of Standard Reference Materials Certified carbon Photon activation Sample content, ppm results, ppm SRM 131 a low carbon 44 45,48,45,43, silicon steel 40, 46 520 480,480, 500 SRM 170a B.O.H.steel SRM 125a 320 330, 330, 320” SRM 170a used as standard. Table 11. Carbon in Pure Metals Results Sample Carbon content, ppm Gold 0.44,0.39,0.49 Silver (Lot 1) 1.2,1.2,1.8 Silver (Lot 2) 0.22,0.33,0.23,0.33 Zinc 0.3,0.2,0.2 Cadmium 0.32,0.43 Molybdenum 1.8, 1 . 5
is not relevant in the present case; added carbon would not necessarily behave in the same way as that already in the metal. The National Bureau of Standards has certified the carbon content of a variety of metals and alloys. The certifications are based on the results of several laboratories using different methods and the carbon contents of the Standard Reference Materials are well known. Table I shows the results obtained by photon activation analysis compared with the certified values. The precisions and accuracies are in line with most results reported for photon activation analysis. The agreement verifies the validity of the general techniques of sample encapsulation, flux monitoring, carbon dioxide extraction, and counting. Table I1 shows results obtained on the high purity metal samples. The agreement of replicate analyses tends to confirm the reliability of the method at levels of 1 ppm and less. The sensitivity of the method may be estimated by a pro-
cedure described by Currie (7). The background of the coincidence detector system is about 0.2 cpm. An activity of half-life of 20 minutes would require an initial count rate of about 0.8 cpm to be detected with 95 confidence in the presence of this background. One microgram of carbon after an irradiation time of two half-lives of llC has a counting rate of 40 cpm. It takes about one half-life for processing the sample for counting, reducing the count rate to 20 cpm per microgram. This is about 25 times mimimum detectable activity implying a detection limit of about 4 X 10-8 gram. There are a number of factors affecting the accuracy. The possibility of incomplete combustion of the sample or incomplete extraction of the carbon dioxide into the gas stream exists in all methods of carbon combustion analysis. The judicious use of fluxes and ignition accelerators, without regard to their carbon content, may alleviate this when using photon activation. Chemical interference does not appear to be a problem. Decay measurements were made on all samples and no contamination from other radioisotopes was observed. Problems were encountered in the determination of low levels of carbon in materials of chip form. If an adequate post-irradiation etching could not be accomplished, results were usually high and unreproducible by several micrograms. The reaction lBO(y,an)l1Cinterferes. At 35 MeV the ratio of yeild from carbon to yield from oxygen, on an atom basis, is approximately lo3. The oxygen reaction has a threshold at about 26 MeV. Conducting the irradiation at this electron energy eliminates the interference, but results in a loss of sensitivity by about a factor of ten. ACKNOWLEDGMENT
The authors thank the NBS LINAC operators for the very fine services performed. Received for review March 23, 1970. Accepted April 28, 1970. (7) L. A. Currie, ANAL.CHEM., 40,586 (1968).
AIDS FOR ANALYTICAL CHEMISTS New Sample Introduction Technique for Mass Spectrometer F. G. Padrta and J. J. Donohue Universal Oil Products Company, Des Plaines, 111.
DIFFICULTIES were encountered during attempts to quantitatively analyze for wide boiling range mixtures at sample inlet temperatures of 350 “C. The only sampling technique which allowed the introduction of such a wide boiling range mixture without sample disproportionation was via a pipet through a gallium-covered frit. However, there was no control on the amount of sample introduced and the gallium promoted thermal decomposition of certain samples. This paper describes the design of an all-glass ‘‘breakoffl’ sample introduction technique which could be attached to any mass spectrometer. Results are also included which show that the technique is capable of admitting standard amounts of a C6-CZ9 blend, or 1 pl, without sample disproportionation. 950
ANALYTICAL CHEMISTRY, VOL. 42, NO. 8, JULY 1970
EXPERIMENTAL A Consolidated Electrodynamics Corp. Model 21-103C mass spectrometer was modified with a CEC 350 “C glass oven and a Varian 40 literisecond getter-ion pump. The glass frit normally located at the top of the expansion chamber was sealed off. The new breakoff sampling system was connected by a short section of glass tubing extending from it to the solids-inlet valve. Direct sealing to the solidsinlet valve could not be made because of the restricted space within the oven. Sample Inlet System. Sample introduction into the expansion chamber is made through a vacuum lock arrangement by the following procedure. First, a disposable pipet is filled with sample and sealed into a capillary ampoule,
Precision Ground To Fit D r y - O r i f i c e
\ /
/
I
7
,
/
Hole For Sample A r n p o u A
l s p r i n g Sleeve
Figure 1. Ampoule capillary holder Seeied, . -dnE
Co., Bromall, Pa. Volume reproducibility for these pipets is i l % . The pipet is filled with sample and loaded into the open end of a measured length of capillary tube (55 mm long, 1-mm diameter) made from larger disposable pipets available from the same supplier. Immediately after inserting the filled pipet into the capillary, the open end is sealed to make a n ampoule, using the flame from a small torch such as those available from Tescom Corp., Minneapolis, Minn. The thin wall of the capillary permits sealing quickly. Volatile samples can be sealed by first immersing the capillary momentarily in liquid nitrogen. The small volume of air that will be trapped in the sealed capillary does not interfere with sample analysis. If oxygen-sensitive samples are to be analyzed the ampoules can be purged with a n inert gas prior to inserting the sample pipet. Empty, sealed 1-pl pipets are used to reduce the volume of trapped air when solid samples are placed directly into the capillary. A schematic drawing of a sealed capillary containing a 1-pl pipet is shown in Figure 2. Dry-Orifice Inlet. A 60-cc reservoir and dry-orifice are part of the mass spectrometer inlet system, as shown in Figure 3. The dry-orifice and the capillary holder are ground together before sealing the reservoir to the expansion chamber. The short section of the inlet from the dry-orifice to just outside the oven is wrapped with heater wire. This section is kept approximately 25 “C above the oven temperature to ensure vaporization of the sample into the reservoir. The holder is 40 mm in total length. Diameter of the large end is 5 mm, 3 mm at the small (spring) end. The cavity within the holder is 1.1-mm diameter by 28 mm deep; a rather snug fit for the capillary to ensure its not being lost from the holder. The distance from the point of capillary breakoff to the back cavity of the holder as it seats in the dry-orifice is critically important because breakoff must occur just as the holder seats. To ensure proper breaking, it is 5 mm less than the length of the sample ampoule. Once determined, this distance will not change unless a n entire new inlet is made. Several holders can be ground to fit the same dry-orifice in case of’ breakage. In five years of using this technique, however, a holder has never been broken. This is primarily due to the flexible spring sleeve between the handle and capillary holder shown in Figure 3, and detailed in Figure 1.
y s e a l s d Breokoff TIP
~ C o p i l l a t y Tube
Los-or
I - ~ Sample I Pipet
Figure 2. Sealed ampoule with sample pipet
one end of which is drawn out to a fine tip. The ampoule is placed into a specially-designed holder, described later, the end of which has been ground to make a vacuum-tight seal with the dry-orifice of the heated inlet under reduced pressure. The tip of the sample ampoule protrudes from the special holder, and the fine tip breaks off against the wall of the heated inlet just as the holder makes its seat with the dry-orifice. Broken pieces of glass are prevented from entering the 2-liter expansion chamber by a glass frit a t the end of the chamber. Pipet and capillary (minus the tip) are withdrawn after pumpout of the sample. Use of new sample ampoules and pipets for each run prevents contamination from previous samples. Capillary Holder. The three main functions of the holder are that it retains the sample ampoule, guides the protruding tip safely through the dry-orifice for breakoff, and seals the vaporized sample in the reservoir. The holder is attached by means of a spring sleeve to a handle for insertion through the vacuum lock t o the heated dry-orifice. This spring sleeve removes any strain from the glass holder, thereby eliminating breakage and allowing it freedom to seat properly. The holder is reduced in diameter at a convenient point near the spring end to minimize heat loss to the handle. The design of the holder is shown in Figure 1. Sample Capillary and Pipet. The 0.5- and 1-pl disposable pipets (Microcaps) are available from Drummond Scientific
Handle, 2-iller Eapansion
Figure 3. Dry-orifice and sample inlet system
MW 78 134 226 386
Compound
Benzene t-Butylbenzene ri-Hexadecane Di-n-C11H23 Benzene
Table I. Accuracy of “Breakoff” System MS analysis of synthetic mixtures Synthetic mixtures Absolute Normalized A B A B A B 9.8 ... 9.9 ... 9.7 ... 37.6 47.6 5.0
40.2 39.9 19.9
38.5 48.5 5.1
39.7 38.6 19.2
37.7 47.6 5.0
40.7 39.6 19.7
100.0
100.0
102.0
97.5
100.0
100.0
ANALYTICAL CHEMISTRY, VOL. 42, NO. 8, JULY 1970
951
ACCURACY OF “BREAKOFF” SYSTEM
Table 11. Example of Precision for 10 Injections of n-Decane Run No.
Relative intensity @ m/e 142
1
224.7 Div/pl 222.6 222.6 222.6 223.5 226.2 222.6 222.0 220.5 219.9
2 3 4 5
6 7 8 9 10
Estimated std dev
952
= d= 1 . 8
ANALYTICAL CHEMISTRY, VOL. 42, NO. 8, JULY 1970
To demonstrate the accuracy of this technique, two wideboiling (80-450 ’C) samples of known composition were analyzed. Comparison of the normalized mass spectrometer results with the known values is shown in Table I. Such good agreement demonstrates that no sample disproportionation has taken place. Accuracy of the results on an absolute basis is &2.5 %, but it also must be remembered that reproducibility of the pipets is only i l %. An example of the precision of introduction is shown in Table 11. The estimated standard deviation for 10 injections of n-decane based on the peak height of the parent ion is f1.8%. RECEIVED for review September 5, 1969. Accepted April 9, 1970.
