214
Anal. Chem. 1980, 52, 214-216
Suppression of Monoxide Interference in Surface Ionization Mass Spectrometry of Rare-Earth Elements Shuford Schuhmann, John A. Philpotts, and Patricia Fryer Hawaii Institute of Geophysics, University of Hawaii, Honolulu, Hawaii 96822
A long-recognized problem in the determination of rareearth elements by surface ionization mass spectrometry is interference by monoxides of lighter rare earths with the ions of heavier elements. Masuda ( I ) cautioned against errors due to monoxide ions in the determination of Gd, Dy, Yb, and Lu abundances by isotope dilution. Similarly Hooker, O’Nions, and Pankhurst (2)reported BaO+ interference with Eu+ and Sm+, NdO+ interference with Dy’ and Er’, and GdO+ with Yb+. Nakamura et al. ( 3 ) suspected BaO+ interference on 154Sm+. They reduced monoxide production of Nd and Sm in favor of the elemental spectra by drying the solutions on the sample filaments in an H2 atmosphere. Studier, Sloth, and Moore (4) produced oxide-free, elemental uranium spectra from single filament ion sources by deposition of carbon from a benzene atmosphere before, after, or during deposition of the sample. We have fought the problem unsuccessfully for more than a decade. Table I lists isotopes used in our isotope-dilution rare-earth analyses (6) which are isobaric with potentially interfering monoxide ions, the natural compositions of which are also shown. In the actual analysis, of course, the monoxides will not have natural compositions because of the addition of a “spike” enriched in particular isotopes. The prime enrichments for our “spike” are indicated in the table by underlining. Our prime Sm “spike” isotope is 149 which gives a monoxide interfering with lS5H0+,which we do not determine. A few other points concerning the table are worth noting. Only those isotopes measured in our isotope-dilution determinations are included. Thus minor isotopes such as 15’Gd, 154Gd,156Dy, 158Dy,lmDy, etc. are not listed. Similarly, only those oxides that we have encountered, rather than all possible oxides, are included. Those most notably absent are EuO+ and BaO+ (we do occasionally observe BaCP and BaF+) and hence Table I excludes isotopes lighter than 152Sm.Rare-earth oxide interference is also possible, of course, beyond Lu. Monoxide interference affects isotope dilution results in three ways. Uncompensated interference on a non-“spike” mass ion current will yield analytical results higher than the actual abundances, possibly even negative apparent “spike” to normal ratios. For example, inspection of the table shows that uncompensated CeO+, P r o + , NdO+, or SmO+ will disproportionate one or more of the masses 156, 157, 158, and 160 relative to 155 used in our Gd determinations. Similarly NdO+ will disproportionate the Dy+ spectrum with almost no effect on the “spike”-enriched mass 163Dy;Er+ is similarly affected by both NdO+ and SmO’. Such interference is readily recognizable, in general, by discrepant results from the several isotopes of an element. A more insidious type of interference occurs in the case of L a o + on Gd+, for example, because here only “spike” ls5Gd+is appreciably affected so that the Gd values computed from masses 156,157, 158, and 160 may all be in reasonable agreement and yet, unfortunately, all in error. Interference on a “spike”-enriched mass will result in lower than actual abundance estimates, possibly even negative amounts. Thirdly, data may be degraded by monoxide impingement on a correction mass. Thus we measure mass 161 to remove elemental Dy ions from the Gd+ spectrum a t 160, 158, and 156, but if, in fact, the 161 is NdO+, the correction would obviously be invalid. Finally it should be noted that if the oxide spectra themselves are used in isotope dilution analyses, the problem then becomes the opposite one 0003-2700/80/0352-0220$01 .OO/O
of removing interference by ions of the elements. The source of the oxygen to form the monoxides is obscure. We dry our samples onto rhenium filaments as chlorides. H 2 0 vapor can be eliminated as a suspect because its concentration in the source region is rendered negligible by means of a liquid nitrogen cooled cold-finger. C 0 2 is our prime suspect a t the moment. C 0 2 has a vapor pressure of about 2.7 X Torr @ 77.4 K (7). A C 0 2 atmosphere might be maintained by the source cold-finger and/or by the liquid nitrogen cooled trap to the Hg diffusion pump. Possibly reduction of the C 0 2 by lanthanide chloride might yield COCl2, which would be removed quantitatively by the cold trap, or CO which would be removed as soon as formed by the Hg diffusion pump. We have developed a technique that produces monoxidefree spectra of the rare-earth elements by controlled in-leakage of propane or H2 to the source chamber for a steady-state ion gauge pressure of about Torr during analysis. No precautions are taken against oxide formation during drying of the sample solutions onto the source filaments. We use a standard triple rhenium filament source. EXPERIMENTAL Apparatus. The instruments we modified for the purpose are Shields-NBS with 6-in. and 12-in. analyzers. The ‘/4-in.copper tube between the source chamber and “source toggle” valve was teed into twice, successively, and two additional toggle valves were added for on-off control of (1) reducing gas or (2) oxidant gas. To the stem packing side of each of the added toggle valves are connected respective small metering valves. Beyond each of the independent metering valves are, respectively, a propane (LPG) pressure regulator and tank and an O2 pressure regulator on an O2 cylinder. (An O2inlet was provided to facilitate analyzing GdO+ and other oxides in the Masuda mode, if desired.) Procedure. With a normal operating vacuum in the mass spectrometer and filaments at temperatures to ionize the desired element(s), the spectra of the elements are scanned quickly and inspected for abnormal appearing proportions of the ion currents not principally enriched in the spike for the element or inspected for the oxide of the spike of a lighter element suspected of causing interference. While an ion current suspected of subjection to monoxide interference is being monitored, propane flow through the source region is started and adjusted to steady-state pressures Torr. When the monoxide ion currents are of the order of suppressed and the elemental spectra have stopped rapid growth, the mass range of the element is repeatedly scanned and the isotope ratios are measured in the usual manner. RESULTS AND DISCUSSION Typical Rock Sample. Table I1 shows analytical results for a Mariana Trough basalt (~1316).The results presented for Gd and Dy analyses illustrate the overwhelming interference under normal vacuum of lighter monoxides on these two elements. With the reducing effect of propane, however, the oxide interferences are eliminated and there is internal concordance for all the elements from one analytical isotope ratio to another, across the table. Synthetic “Spike Blank”. T o further test the potential for monoxide interferences and their reduction, a “known” mixture of isotopically enriched (“spike”) rare earths was analyzed by the mass spectrometer. (For completeness in the lighter rare-earth series, a nonenriched 141Pr,with potential for monoxide interference with 157Gd,was included.) This sample is called a blank because presumably the “spike” isotope of each element is diluted by zero amount of the
e 1979 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 1, JANUARY 1980
215
__-
Table I. Rare Earth Isotopes Used and Natural Compositions of Potentially Interfering Monoxidesa La0 Sm Gd
152 154
+
155
+
156 157 158 160 161 162
DY +
+
0.09 9 5 3 0.04 0.20
CeO
Pro’
+
88.27 0.03
11.23
99.76 0.04
0.02
167 168 171 172
Yb’
27.1 23.7 8.3 0.02 5.7 5.6 0.00 0.01
164 166
+
+
17.3
163 Er
NdO
173 Lu
SmO
GdO
+
3.1 0.00 0.01 15.1 11.3 7.4 0.03 26.6 0.01 0.05
TbO’
DyO’
0.20
14.8
20.6 15.7 24.7 0.41 21.7
174 175
+
+
0.19 0.25
0.06 0.10 99.76 0.04
2.33
The natural abundance data are taken from the Chart of the Nuclides ( 5 ) . The isotopic composition of oxygen is incorporated. The data are rounded to the second decimal place where applicable. Table 11. Selected Element Abundances in ppm by Weight for Mariana Trough Basalt (M1316) Analytical isotope ratiosa
element Nd
propane: Sm
propane: Gd normal : propane: DY normal: propane: Er pro pane : Yb
propane:
1410.32 ( 7 ) t0.076 1471149 3.42 ( 7 ) k0.033 1 5 6 / 155
propane:
20.097
1541149 3.40 (ti) i0.028 158/155 160/155 -300 6.7 ( 6 ) t242 i0.67 4.48 ( 5 ) 4.48 ( 5 ) i0.035 t0.037 1641163 17.5 (4) i0.51 6.2 ( 4 ) 20.13
Gd normal: propane: DY normal : propane : Yb normal: propane:
1741173 3.09 (11) tO.012
a Numbers of analyses are given in parentheses and the errors are standard deviations for the internal precision.
element of natural isotopic composition. Table I11 shows the mass spectrometric determinations of the blank. The “normal” vacuum runs for Gd and Dy illustrate similar positive interferences from CeO+ and P r o + as observed in the Mariana Trough basalt data of Table 11, and described in the introduction. The high value for Sm (normal) by the 154/149 ratio is thought to represent 13’LaO+ interference. Similarly, referring to Table I, the unrealistic negative Gd value by 158/155 would appear to be principally from 142CeO+(in an actual sample, possibly also from natural ‘**NdO+).Similarly, the high values for Dy (normal) appear to be caused principally by NdO+; and for Yb (normal) by GdO+. Significantly, the values obtained with “propane” influx are near zero and internally concordant from isotope ratio to isotope ratio across the table. Some of the small discrepancies between the check values may result from small errors in the assumed isotopic composition of the enriched spike. Except for Yb, which we re-determined, the spike abundances were
analytical isotope ratios
element Sm normal:
146/145 10.34 (8)
1521149 3.34 ( 7 ) i0.016 1571155 104 145 i 48 i 18 4.52 ( 5 ) 4.49(5) i 0.024 20.038 161/163 162/163 -63.4 ( 5 ) 190 f0.66 i 27 5.2 ( 4 ) 5 . 2 14) ?: 0.20 i 0.lY 1661167 1661167 3.28 ( 1 0 ) 3.26 ( 1 0 ) i0.0058 i 0 ..0 1~. 4 1711173 1721173 3.10 ( 7 ) 3.03 (8) i0.016 +0.013
Table 111. Synthetic Spike Blank in Micrograms
152/149 0.000 ( 7 ) i0.0000 0.000 ( 5 ) io.0000 1561155 0.151 ( 4 ) *0.0011 ’ 0.000 ( 5 ) r0.0002 1611163 1.392 (ti) i0.0036 0.002 ( 7 ) i0.0004 1711173 0.142(4) i0.006 0.000 ( 8 ) *0.0001 ~~
1541149 0.015 ( 7 ) t0.0002 0.000 ( 5 ) iO.OOOO 157/155 1581155- 1601155 -1.211 ( 4 ) -2.572 ( 4 ) 0 . 0 7 0 ( 4 ) i0.0068 ’ i0.0309 ’ i0.0123 ’ 0.006 ( 5 ) 0.001 ( 5 ) 0.001 ( 5 ) i0.0021 +0.0006 i0.0002 162/163 ___. 164/163 0.367 (8) 0.015 ( 8 ) *0.005 i0.0003 -0.001 ( 7 ) 0.000 ( 7 ) i0.0004 i0.0005 172/173 1741173. 0.005 (4) 0.000 (4) iO.0000 iO.0000 0.000 (8) 0.000 ( 8 ) io.0001 io.0001
provided by the supplier, Oak Ridge National Laboratory. In most cases, the “propane” spectra were measured before the “normal”. Strong, stable elemental spectra were produced easily. The “normal” data were obtained after the propane was turned off and the oxides reappeared spontaneously, in vacuo. Unexpectedly, admission of propane not only eliminates monoxide interference, but also increases production of the desired elemental ions, sometimes by a factor of a hundred. Above a certain pressure, signal strength is relatively insensitive to large (2x) changes in pressure; stable, easily-measured spectra are obtainable. Monoxide spectra are not permanently eliminated by propane flow. These may reappear soon after the flow is stopped. LITERATURE CITED (1) A Masuda, Geochem. J., 2, 111 (1968). (2) P.J. Hooker, R. K. O’Nions, and R. J. Pankhurst, Chem. Geol., 16, 189 (1975). (3) N. Nakamura, M. Tatsumoto, P. D. Nunes, D. M. Unruh, A. P. Schwab, and T. R . Wildemann, Proc. Lunar Sci. Conf., 7th, 2, 2309 (1976). (4) M. H. Studier, E. N. Sloth, and L. P. Moore, J . Phys. Chem.. 66, 133 (1962). (5) F. W. Walker, G.J. Kirouac, and F. M. Rourke, “Chart of the Nuclides”, 12th ed., G.E.Co., Schenectady, N.Y. 12345, April 1977.
216
Anal. Chem. 1980, 52, 216-218
(6) C. C. Schnetzler, H. H. Thomas, and J . A. Philpotts, Anal. Chem., 39, 1888 (1967). (7) R. E. Honig and H . 0. Hook, RCA Rev., 21 (3), 363 (1960).
RECEIVED for review, August 9, 1979. Accepted October 15,
1979. This research was performed on equipment on loan from NASA/Goddard Space Flight Center under agreement No. GSFC 77-5(E). The support of research grants NSG 9071 from NASA and EAR 79-06950 from the National Science Foundation is gratefully acknowledged.
Stable Tracer Iron-58 Technique for Iron Utilization Studies J. J. Carni and W. D. James" Research Reactor Facility, University of Missouri, Columbia, Missouri 652 1 1
S. R. Koirtyohann Environmental Trace Substances Research Center and Department of Chemistry, University of Missouri, Columbia, Missouri 652 1 1
E. R. Morris Nutrition Institute, United States Department of Agriculture, Beltsville, Maryland 20705
Iron is the element most commonly deficient in humans. I t has been estimated that more than 5 million adult women in the United States have iron deficiency anemia, even though typical diets contain several times the quantities of iron necessary to maintain balance (1). The deficiency is prevalent among women of child-bearing age. In early studies (2, 3) of iron absorption in infants, absorption efficiencies, especially in babies less than 3 months old, differed markedly from those in older children. More recently it was shown that fortification of infant foods with iron in a form that was not absorbed served no useful purpose and that further study of the absorption of iron by infants was essential for prevention of iron deficiency ( 4 ) . Studies of the bioavailability and absorption of dietary iron are clearly warranted. Currently the common practice in the study of iron absorption is the application of a radiotracer. Oral or intravenous doses of radioiron (55Feor 59Fe)followed by a delay of about two weeks and a count of nuclear radiation emissions from a blood aliquot can provide useful information regarding gastrointestinal absorption of iron and its red blood cell incorporation efficiency. Unfortunately the use of a radiotracer is difficult to justify for infants, pregnant women, and other radiation-sensitive subjects. This is especially true for experimental studies when the individual's medical status does not warrant the absorption evaluation. In 1963 Lowman and Krivit (5)used the stable iron isotope, 58Feas an in-vivo tracer to determine plasma clearance. The tracer 58Fein the plasma samples was quantified by neutron activation analysis and NaI(T1) y spectroscopy. An attempted extension of this study to red blood cell incorporation proved futile because the small quantities of 58Fetracer could not be determined in the presence of the much larger quantities of the isotope in natural iron. Later Dyer and Brill (6) and King et al. (7,8) overcame that problem by using higher resolution, solid state y detectors and successfully determined iron utilization parameters in women. However, both of these studies were primarily concerned with the application of the technique and did not document precision and accuracy characteristics of the method. T h e limited use (6, 9) of the 58Fe stable isotope tracer technique for iron utilization studies reflects the need for thorough evaluation and documentation of the capabilities of the method. Calculations show that 10% utilization of a 5-mg tracer dose of 58Feby an adult human would result in a maximum relative change in %Fe of about 7.5% in the blood. Since this measurement requires the determination of total iron for the purpose of subtracting natural 58Febackground, 0003-2700/80/0352-0222$01 OO/O
in addition to the 58Fe determination, the precision requirements are obviously high. In this investigation, analytical techniques have been developed to measure both with a high degree of confidence (*1-3% relative). These techniques have been applied to multiple analyses of a single whole blood sample to evaluate reproducibility. Experiments have been performed to determine the accuracy with which spike material can be recovered.
EXPERIMENTAL Apparatus. In this work, the neutron activation analysis facilities at the University of Missouri's Research Reactor were used. Samples were encapsulated in cleaned quartz irradiation vials fabricated from TO8 quartz obtained from Heraeus Amersil Co. Samples were irradiated for 50-60 h in a 1-inch reflector position having a thermal neutron flux of n cm-* s-'. Gamma emissions were detected and quantified with an Ortec VIP series Ge(Li) detector (2.2 keV resolution @ 1332 keV) and a Nuclear Data 4410 multichannel pulse height analyzer. A Perkin-Elmer 576 spectrophotometer was used for the colorimetric determination of total iron. Reagents. Spiking solutions enriched in 58Fewere prepared by dissolving Fe203(73.26% 5sFe)in HCl. The enriched iron was obtained from the Isotope Sales Division of Oak Ridge National Laboratory. All comparisonstandards were prepared by a similar dissolution of high purity (99.999%)iron oxide obtained from Spex Industries, Inc., Methuen, N.J. Collection of Blood Samples. Samples were drawn from healthy males directly into heparinized containers and were immediately frozen and stored at -20 "C. One sample was thawed, subdivided into some 100 aliquots and refrozen. The sample was kept well mixed during the aliquoting procedure by frequent shaking; aliquots were considered t o be identical homogeneous portions of the sample. These aliquots were designated and used as a reference standard. A second sample was drawn, thawed, and subdivided in Beltsville, Md., by one of us (E.R.M.) who was not involved in the actual analysis of the samples. Carefully measured quantities of enriched 5sFewere added to these aliquots. These samples along with a portion of the spiking solution were refrozen and transferred to the analytical facilities in Columbia, Mo., for analysis as blind spiked samples. Selection of Total Iron Measurement Technique and Precision Evaluation. Atomic absorption analysis and colorimetry were considered as possible analytical techniques for total iron. Either technique provided adequate sensitivity so we compared their reproducibilities with instruments available. Precision was superior with a modified 1,lO-phenanthroline-iron complex colorimetric technique (10) and a Perkin-Elmer 576 Spectrometer. Two-hundred microliters of each sample was transferred t o a 250-mL Erlenmeyer flask containing a "OB1979
American Chemical Society
