Nd - American Chemical Society

Oct 10, 2014 - State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China...
0 downloads 0 Views 946KB Size
Article pubs.acs.org/ac

High-Precision 143Nd/144Nd Ratios from NdO+ Data Corrected with in-Run Measured Oxygen Isotope Ratios Zhu-Yin Chu,*,† Chao-Feng Li,† Ernst Hegner,‡,§ Zhi Chen,†,∥ Yan Yan,†,∥ and Jing-Hui Guo† †

State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China Department of Earth and Environmental Sciences, LMU Munich, Theresienstraße 41, 80333 Munich, Germany § GeoBio-Center, LMU Munich, Luisenstraße 37, 80333 Munich, Germany ∥ College of Earth Sciences, University of Chinese Academy of Sciences, Beijing 100049, China ‡

S Supporting Information *

ABSTRACT: The NdO+ technique has been considerably refined in recent years for high-precision measurement of Nd isotope ratios in low-level samples (1−5 ng Nd). As oxygen isotopic compositions may vary significantly with experimental conditions such as filament material, ionization enhancer and the ambient oxygen in the ion source, great “care” should be taken for using correct oxygen isotopic compositions to do the isobaric oxide corrections for the “conventional” NdO+ method. Our method presented here for NdO+ data reduction and PrO+ interference corrections uses the oxygen isotope composition determined in each cycle of the NdO+ measurements. For that purpose, we measured the small ion signals of 150Nd17O+ and 150Nd18O+ with amplifiers equipped with 1012 Ω feedback resistors, and those of Nd16O+ ion beams with 1011 Ω amplifiers. Using 1012 Ω amplifiers facilitates a precise measurement of the very small 150Nd17O+ and 150Nd18O+ ion signals and calculation of highly accurate and precise 143Nd/144Nd isotope ratios. The 143Nd/144Nd ratios for JNdi-1 standards and several whole-rock reference materials determined with the method on 4 ng of Nd loads are consistent with previously reported values within analytical error, with internal and external precision (2 RSE and 2 RSD) of better than 20 and 30 ppm, respectively.

I

amounts of Pr in Nd cuts separated from geological samples,1−5,17 a correction of a 141Pr18O+ interference with 143 Nd16O+ is also critical for obtaining precise 143Nd/144Nd isotope ratios. An adequate oxygen and PrO+ correction poses a major complication for the NdO+ method, and minimizing this problem is the central topic of this paper. As a 1% variation in the 18O/16O ratio could cause about 11 ppm of absolute variation in the calculated 143Nd/144Nd for a Pr-free Nd sample,4 it is critical to apply a correct oxygen isotope composition to do oxide isobaric correction for the NdO + method. The importance of knowing the correct oxygen isotope composition is further exemplified by the large variation in reported 18O/16O ratios up to 6.85%.1−4,17 The oxygen isotope variation reflects different ambient oxygen reservoirs in different mass spectrometers during sample analysis, which may in part be due to different sample-loading techniques.1−4 To date, the oxygen isotope composition used for data reduction in different laboratories was separately determined using high-purity Pr (by measuring 141Pr17O/141Pr16O and 141Pr18O/141Pr16O to obtain 17O/16O and 18O/16O respectively) or enriched 150 Nd spike solutions (by measuring 150Nd17O/150Nd16O and

n recent years, the high-precision measurement of Nd isotopes as NdO+ species has been considerably refined for very low concentrations of Nd in geological samples.1−5 It has been possible to obtain 143Nd/144Nd ratios with an external precision (2 RSD) better than 20 ppm for only ca. 4 ng of Nd when measured as oxide species and using a tantalum oxide (Ta2O5)−phosphoric acid slurry for improvement of Nd ionization.3 A high-sensitivity NdO+ technique is of particular interest for the application of the 147Sm−143Nd decay system in the following fields of geochronology, geo- and cosmochemistry, and oceanography: (1) highly precise determination of Sm−Nd ages of individual growth zones of garnet;3,6,7 (2) determination of Nd isotope compositions in single grains of minerals, such as zircon8 and titanite;9 (3) ultramafic rocks and chondrites with ultralow concentrations of Nd;10−13 (4) the Nd isotope composition of rare minerals from meteorites and other extraterrestrial samples;14,15 (5) the Nd isotope composition in environmental samples including seawater samples, fossil deepsea corals, and marine authigenic and biogenic materials such as Fe−Mn nodules, and foraminifera.1,16,17 An inherent weakness of the NdO+ method is the requirement of the deduction of Nd isotope ratios using NdO+ measurements, i.e., the correction for Nd18O+ and Nd17O+ interferences with Nd16O+, hereafter termed as oxygen corrections. The NdO+ method also requires correction for isobaric interferences of oxide species of Ce and Pr with NdO+ masses (e.g., 141Pr18O+ with 143Nd16O+). Considering the minor © 2014 American Chemical Society

Received: June 8, 2014 Accepted: October 10, 2014 Published: October 10, 2014 11141

dx.doi.org/10.1021/ac502197u | Anal. Chem. 2014, 86, 11141−11150

Analytical Chemistry

Article

Nd16O+ isotopes with Faraday cups connected to conventional 1011 Ω amplifiers. This analytical setup enabled us to carry out in-run corrections for oxygen and PrO+ interferences. Isotope analyses of reference materials based on this method are presented in the following sections.

150

Nd18O/150Nd16O to obtain 17O/16O and 18O/16O, respectively).1−5,17−19 As the oxygen isotopic composition is relatively “reproducible” for a specific sample loading method, this method of “external” NdO+ data calibration has been practiced successfully in the past. However, it has also been shown that, the ambient oxygen isotope composition may vary for different sample loads using the same sample-loading technique in the same laboratory, and even during mass spectrometric analysis of a sample due to isotope fractionation effects.1,20,21 Therefore, oxygen and PrO+ interference corrections using a constant “in-house” oxygen isotope composition may not be completely accurate and requires frequent checking of the applied oxygen isotopic composition to ensure a correct isobaric oxide correction. Liu et al.22 presented an alternative method for NdO+ analysis and oxygen correction that took advantage of bleeding the sample chamber of the mass spectrometer with 16O-enriched oxygen. Under this condition, it can be assumed that the main oxygen reservoir is represented by the introduced 16O-enriched oxygen. However, a contribution of natural oxygen from oxygenated compounds in the sample load and remnant oxygen in the evacuated sample chamber of the mass spectrometer cannot be precluded and this contribution would be difficult to quantify. Considering these possibilities, it appears as though this method is not well-suited for routine analyses. A major improvement of the currently practiced NdO+ isotope analysis would be the inclusion of oxygen and PrO+ isobaric interference corrections based on the oxygen isotope composition of the sample under ionization, e.g. that specific composition producing the measured NdO+ ion beam. This is feasible if the oxygen isotope composition is determined simultaneously along with the NdO+ signal.20,21 However, a high-precision measurement of the low abundances of 17O- and 18O-related isotopes is technically challenging. Recently introduced Faraday cups connected to amplifiers equipped with 1012 Ω and 1013 Ω resistors23−30 offer a new opportunity to overcome the shortcomings to some extent. The 1012 Ω resistors provide a 10 times higher gain of the amplifier, while the Johnson noise of the resistor increases only by a factor of (10)1/2, resulting in a theoretical 3-fold improvement of the signal-to-noise ratio relative to that of the conventional 1011 Ω resistors.26 This technological improvement enables us now to measure very low isotope abundances at higher precision than previously possible. In this study, accordingly, we determined for the first time the actual in-run 17O/16O and 18O/16O isotope ratios pertaining to a specific NdO+ analysis. For this purpose, we measured the small ion signals of 150Nd17O+ and 150Nd18O+ with Faraday cups connected to 1012 Ω amplifiers, along with the measurement of



EXPERIMENTAL SECTION

Reference Materials and Filament Loading Techniques. The Nd reference material JNdi-1 and modified compositions of it (JNdi-1 doped with 150Nd spike, JNdi-1 doped with Ce and Pr) were used for the test measurements. In addition, we analyzed international whole-rock reference materials, including BCR-2, BHVO-2, AGV-2, W-2a, GSP-2 from USGS and JB-3, JA-2 from GSJ, using the method of Chu et al.4 for Nd separation. Most samples were loaded on degassed W filaments with TaF5 as an ionization enhancer using the methods of Chu et al.,4 except that several samples were loaded on degassed Re filaments with slurried Ta2O5 as an ionization enhancer as reported by Harvey and Baxter.3 Sample loads were ca. 4 ng of Nd for the JNdi-1 solutions and the whole-rock samples. Thermal-Ionization Mass Spectrometry. The isotope measurements were performed on a Thermo Scientific TRITON PLUS thermal-ionization mass spectrometer (TIMS) equipped with nine Faraday cups connected to seven 1011 Ω amplifiers and two 1012 Ω amplifiers. A gain calibration of the amplifiers was performed once a day. We determined the isotope abundances of the reference materials with two mass spectrometric routines: (1) static multi-ion-collection of NdO+ isotopes without simultaneously monitoring 145Nd16O+ and 148Nd16O+ ion signals, and CeO+ interference due to the limited number of cups and, (2) dynamic two-line multi-ion-collection measuring all NdO+ isotopes and monitoring simultaneously all isobaric interferences related to Ce, Pr, and Sm oxides. The cup configurations for these two methods are shown in Tables 1a and 1b. The ionization temperature for the NdO+ measurement for both W + TaF5 and Re + Ta2O5 loading methods was usually 1500 to 1600 °C. Static Multi-Ion-Collection Routine. For static ion collection, the 140Ce16O+ ion signal was measured with the H4 cup at the beginning and end of the sample analysis using 1 block of 10 cycles with 4 s integration time. CeO+ interference with NdO+ and PrO+ was corrected for each cycle by interpolation assuming a linear behavior of the CeO+ signals throughout the measurement. Isotope data were collected in 6 blocks comprising 20 cycles, each with 8 s integration time. Peak center and lens focus were performed every third block using

Table 1a. Cup Configuration for the Static Analysis of Nd Isotopes as NdO+ cup

L4

L3

L2

L1

central

H1

H2

H3

H4

amplifier mass main oxide

1011 157 141 16 Pr O

1011 158 142 Nd16O

1011 159 143 Nd16O

1011 160 144 Nd16O

1011 162 146 Nd16O

1011 166 150 Nd16O

1012 167 150 Nd17O

1012 168 150 Nd18O

1011 170 154 Sm16O

Table 1b. Cup Configuration for the Two-Line Dynamic Multicollection Routine cup amplifier line 1 line 2

L4 11

mass main oxide mass main oxide

10 157 141 16 Pr O 156 140 Ce16O

L3 11

10 158 142 Nd16O 157 141 16 Pr O

L2 11

10 159 143 Nd16O 158 142 Nd16O

L1 11

10 160 144 Nd16O 159 143 Nd16O 11142

central 11

10 162 146 Nd16O 161 145 Nd16O

H1 11

10 164 148 Nd16O 163 147 Sm16O

H2 11

10 166 150 Nd16O 165 149 Sm16O

H3 12

10 168 150 Nd18O 167 150 Nd17O

H4 12

10 170 154 Sm16O 169 153 Eu16O

dx.doi.org/10.1021/ac502197u | Anal. Chem. 2014, 86, 11141−11150

Analytical Chemistry

Article

the 146Nd16O+ ion beam in the center cup. The baseline was measured for 30 s (idle time 10 s) also every third block after deflecting the ion beam. The ion current during sample analysis was kept at 80% to 120% of the value at the beginning of the measurement using 146Nd16O+ as the pilot signal. A sample analysis in static mode took ca. 22 min. The off-line data reduction included the following steps: (1) calculation of the ion beam intensity on mass 164 (148Nd16O+ + 146 Nd18O+, hereafter termed as 164I) that was not included in the measurement using the measured ion beam signal on mass 160 (144Nd16O+ + 143Nd17O+ + 142Nd18O+, hereafter termed as 160 I) and using the relationship 164I/160I = 0.242 45;2 (2) elimination of the 148Nd18O+ contribution to the signal of mass 166 which gives, the net 150Nd16O+ signal. In this calculation, we used the oxygen isotope composition inferred from measurements of high-purity Pr using our mass spectrometer; (3) determination of the 17O/16O and 18O/16O ratios with those of 150 Nd17O/150Nd16O and 150Nd18O/150Nd16O, respectively; (4) the 140Ce16O+ ion beam intensity for each cycle was inferred from the 140Ce16O+ ion signals measured at the beginning and end of the measurement by interpolation. The CeO+ interferences (140Ce17O+ with 141Pr16O+, 142Ce16O+ and 140Ce18O+ with 142Nd16O+, 142Ce17O+ with 143Nd16O+, 142Ce18O+ with 144 Nd16O+) were corrected using the 17O/16O and 18O/16O ratios of step 3 and 142Ce/140Ce = 0.125 65;31 (5) PrO+ interferences (141Pr17O+ with 142Nd16O+, 141Pr18O+ with 143 Nd16O+) were corrected using the 17O/16O and 18O/16O ratios of step 3; (6) oxygen corrections (corrections for interfering NdO+ isotopes, e.g., 142Nd18O+ with 144Nd16O+) were performed with the 17O/16O and 18O/16O ratios of step 3. During this data reduction step, as we did not measure the ion beams on masses 161 (145Nd16O+ + 144Nd17O+ + 143Nd18O+, termed as 161I hereafter) and 164 (148Nd16O+ + 146Nd18O+), 161I and 164I were obtained according to the measured 160I and using the equations 161I/160I = 0.348 96 and 164I/160I = 0.242 45, for the oxygen correction calculations.2,3,12 The calculations of steps 1−6 can be performed iteratively, but it was found not necessary. The following data reduction steps included spike subtraction for spiked samples and correction for thermal isotope fractionation using 146Nd/144Nd = 0.7219 and an exponential mass fractionation law using the method described by Chu et al.32 All of the above calculations were performed cycle-by-cycle. Two-Line Multimass Dynamic Collection Routine. For the two-line multimass dynamic collection routine, each measurement consisted of 4 blocks with 20 cycles, counted for 8 s each line. The idle time after changing the magnetic field was 5 s. Peak center was performed on mass 162 of line 1 and mass 161 of line 2. Lenses were focused using the 146Nd16O+ signal and the baseline was measured at the beginning of the block after ion beam deflection for 30 s (idle time 10 s). The ion current of 146Nd16O+ was kept at 80% to 120% of that measured at the beginning of the analysis. Analysis time of a sample was ca. 40 min. The data reduction procedure was analogue to that of the static collection mode except that the CeO+ interference with NdO+ and PrO+, the 145Nd17O+ interference with 146Nd16O+, and the 148Nd18O+ interference with 150Nd16O+ were corrected with the in-run measured 140Ce16O+, 145Nd16O+, and 148Nd16O+ signal intensities (Table 1b).

ratios for each cycle of NdO+ analysis by simultaneously measuring the 150Nd17O+ and 150Nd18O+ ion beams. Due to the low 17O and 18O abundances of natural oxygen of ca. 0.04% and 0.2%, respectively, the observed 150Nd17O+ and 150Nd18O+ ion beam intensities were very low, in particular, for nonspiked samples. For example, for nonspiked samples, with ion beam intensity for 144Nd16O+ of ca. 2 × 10−11 A, the corresponding 150 Nd16O+ ion beam was only ca. 4.7 × 10−12 A, and the resulting 150Nd17O+ and 150Nd18O+ ion beams only ca. 1.8 × 10−15 A and 9.8 × 10−15 A, respectively. For a precise measurement of the 150Nd17O+ and 150Nd18O+ ion beams, we employed here for the first time 1012 Ω amplifiers. As has been documented previously,23,24,26,33,34 when measuring small ion signals, the JN noise of the resistor is the main source of error of the analytical precision. Consequently, a theoretical 3-fold improvement of the signalto-noise ratio for measuring small ion beams can be achieved if 1012 Ω resistors are used instead of the conventional 1011 Ω resistors. Accordingly, it can be expected that the theoretical analytical precision (1 RSD) on the 17O/16O and 18O/16O isotope ratios can be improved by a factor of ca. 3 by using 1012 Ω resistors instead of the conventional 1011 Ω resistors for measurements of 150Nd17O+ and 150Nd18O+ ions. An example of a calculation of theoretical precision is given in Table S1 (Supporting Information) where, for a 150Nd16O+ ion signal of ca. 0.5 V, measured with a 1011 Ω amplifier, the voltages for 150Nd17O+ and 150Nd18O+ are ca. 2 mV and 10 mV, respectively, using 1012 Ω amplifiers. The resulting theoretical analytical precision (1 RSD) is ca. 2.3% for 17O/16O and 0.5% for 18O/16O when counting the ion signal for 8 s. In the course of this study, the 150Nd16O+ ion signals ranged from ∼3.5 × 10−12 to 1.8 × 10−11 A, with corresponding voltages of 1.3 to 7.0 mV for 150Nd17O+, and 6.8 to 38 mV for 150Nd18O+ measured with 1012 Ω amplifiers (Tables S2a−d, Supporting Information, and Table 2). As a consequence, the theoretical RSDs should be 0.68% to 3.3% for 17O/16O and 0.14% to 0.68% for 18O/16O. Actually, the RSDs for 17O/16O and 18O/16O isotope ratios during a NdO+ measurement ranged from 0.82% to 4.8% and 0.20% to 1.0% in this study, respectively. These actual RSDs are roughly 1.2 to 2.0 times worse than the theoretical values and may be due to an oxide isotope fractionation effect during thermal ionization. Koornneef et al.30 described newly designed 1013 Ω amplifiers recently developed by Thermo Fisher Scientific. If 1013 Ω amplifiers were used, instead of a 1011 Ω amplifiers for the measurement of 150Nd17O+ and 150Nd18O+ ion signals, the theoretical precision for the 17O/16O and 18O/16O isotope ratios would be improved by a factor of ca. 10. Thus, 1013 Ω amplifiers would be even better suited for the here presented cycle-by-cycle in-run oxygen and PrO correction method. Uncertainties Arising from Oxide-Interference Corrections. The reduction of NdO+ ion intensities to Nd+ isotope ratios can be expressed as functions of NdO+ isotope ratios and oxygen isotope ratios from the following equations, if the terms