A novel sample loading method and protocol for monitoring sample fractionation for high precision Ca stable isotope ratio measurements using double-spike TIMS

Abstract

The external reproducibility (2σ_SD) of Ca stable isotope ratio measurements (δ^44/40Ca) using double-spike thermal ionization mass spectrometry (TIMS) shows a large range from <0.1‰ to 0.5‰. We demonstrate that using a ⁴³Ca–⁴⁸Ca double spike, which allows simultaneous measurements of δ^44/40Ca and δ^44/42Ca, and analyses at moderate signal strengths (⁴⁰Ca ranging from 6 to 10 V), high precision δ^44/40Ca (external reproducibility better than ±0.08‰, 2σ_SD) as well as δ^44/42Ca can be obtained if samples and standards are analyzed under similar fractionation conditions. To monitor the fractionation conditions, the usage of a parameter β is proposed, which measures the deviation of the ⁴³Ca/⁴⁸Ca ratio of the sample–double spike mixture from that of the pure double spike. A novel, low-cost sample loading technique using a combination of Re and Ta filaments and a Ta₂O₅ activator is presented which results in a steady signal. We report the δ^44/40Ca and δ^44/42Ca values, calculated w.r.t. NIST SRM 915a and reported as deviations in parts per mil (‰), of NIST standards SRM 915a (0.01, −0.02) and SRM 915b (0.73, 0.32), NASS 6 (seawater, 1.80, 0.89), USGS silicate rock standards BHVO-2 (0.86, 0.47) and BCR-2 (0.89, 0.48), Geological Survey of Japan carbonate standards JCp-1 (coral, 0.85, 0.45) and JCt-1 (clam shell, 0.83, 0.46) and ECRM 752-1 (limestone, alternative name BCS-CRM 393) (0.83, 0.46) from the Bureau of Analyzed Samples Ltd. UK. The Ca stable isotopic compositions of JCt-1 and ECRM 752-1 are reported for the first time.

---

1 Introduction

Calcium is the fifth most abundant element on the Earth with six stable isotopes: ⁴⁰Ca, ⁴²Ca, ⁴³Ca, ⁴⁴Ca, ⁴⁶Ca, and ⁴⁸Ca. The natural abundance of ⁴⁰Ca is the highest (96.9%) followed by ⁴⁴Ca (2.1%), ⁴²Ca (0.65%), ⁴⁸Ca (0.19%), ⁴³Ca (0.14%) and ⁴⁶Ca (0.003%). Due to the large mass difference between the isotopes of Ca, significant Ca isotopic fractionation has been observed in various natural reservoirs, which provides insights into understanding diverse processes including early Solar System evolution,^1–3 crust–mantle processes,^4–7 seawater evolution,^8–11 weathering,^12,13 and paleotemperature reconstruction.^14–19 Even in biological systems, large Ca isotopic fractionations have been observed and these variations have implications for understanding trophic levels²⁰ and forested ecosystems²¹ and also have biomedical applications.^22–24

The first successful Ca stable isotopic measurements were performed using a single-collector thermal ionization mass spectrometer (TIMS) where the “double spike” (DS) technique, using a mixture of ⁴²Ca and ⁴⁸Ca, was used for correcting mass fractionation.²⁵ Improvements in mass spectrometry, particularly the transition from single collector to multi-collector systems, has led to improvements in the internal precision of Ca isotope ratio measurements while the advent of the multi-collector inductively coupled plasma mass spectrometer (MC-ICPMS) has allowed rapid measurements of Ca isotope ratios in samples. While MC-ICPMS allows rapid measurements, isobaric interference of ⁴⁰Ar⁺ with ⁴⁰Ca⁺ precludes the measurement of the ⁴⁴Ca/⁴⁰Ca ratio and hence, only δ^44/42Ca is reported for such measurements. Even for δ^44/42Ca measurements, molecular isobaric interference of ¹²C¹⁶O₂⁺ with ⁴⁴Ca⁺ needs to be minimized. Additionally, MC-ICPMS measurements of δ^44/42Ca use sample–standard bracketing to correct for instrumental mass fractionation.²⁶ The use of sample–standard bracketing requires careful monitoring of the yield of Ca (>99%) during ion-exchange chromatography, unlike for TIMS measurements using DS, and minimization of the ‘matrix-effect’ for accurate and precise measurements.

While the internal precision and external reproducibility of δ^44/40Ca measurements using DS-TIMS have improved with multi-collection, the external reproducibility of the measurements shows a large range (0.04–0.5‰),^27,28 the exact reason for which is not well understood. Several factors might affect the accuracy and the external reproducibility of Ca isotopic measurements using TIMS. These factors include the double spike composition, sample to spike ratio, type of acid and filament used for sample loading, reservoir mixing, choice of the fractionation law to correct for instrumental mass fractionation and changes in detector properties over time.^29–36

In this study, we report a method for improving the external reproducibility of δ^44/40Ca measurements using TIMS to <0.1‰ using the ⁴³Ca–⁴⁸Ca double spike, which also allows simultaneous measurements of δ^44/42Ca. We propose a novel sample loading protocol that generates a stable Ca signal for several hours resulting in good internal precision and a method to monitor the extent of fractionation in the sample which allows repeated measurements of the same sample to be performed under similar conditions. Improved reproducibility of operational conditions results in external reproducibility of δ^44/40Ca to be <0.1‰ for samples with different matrices (seawater, carbonates and silicates). Using the above measurement protocol, we report the Ca isotopic composition of multiple carbonate reference materials, seawater, and silicate rock standards. The Ca isotopic composition of the Geological Survey of Japan (GSJ) carbonate standard JCt-1 (clam shell) and the limestone CRM from the Bureau of Analyzed Samples Ltd. UK (ECRM 752-1, alternative name BCS-CRM 393) is reported for the first time in this study.

2 Experimental methods

2.1 Preparation of the ⁴³Ca–⁴⁸Ca double spike

The ⁴³Ca–⁴⁸Ca double spike was prepared using artificially enriched ⁴³Ca and ⁴⁸Ca single spikes procured from the Oak Ridge National Laboratory with batch numbers NX 169191 and RT 219801, respectively. The single spikes were dissolved separately using 3 M HCl in pre-cleaned Teflon containers. The single spike solutions were mixed gravimetrically to achieve a ⁴³Ca/⁴⁸Ca ratio of 0.721 ± 0.02 in the double spike, which is close to the natural abundance ratio. The error is estimated based on the precision of our weighing balance. The isotopic composition of the ⁴³Ca–⁴⁸Ca double spike (Table 3) was estimated using a total evaporation technique on a TIMS (Thermo Fischer Scientific Triton Plus) at the Centre for Earth Sciences, IISc, Bangalore. The double spike with ∼3 μg of Ca was loaded on a Ta filament (evaporation filament) with 5% HNO₃ with an additional 0.5 μl of tantalum oxide (Ta₂O₅) activator on top of the sample. During analysis, the evaporation filament current was slowly increased to 2000 mA at a rate of 200 mA s⁻¹. The ionization filament (zone-refined Re) current was also increased at 200 mA s⁻¹ to 3300 mA. Data acquisition was initiated when the ⁴⁸Ca signal was ∼10 V. Data were acquired for 500 cycles until the ⁴⁸Ca signal reduced below 2 V. The total time of analysis was approximately 8 h.

2.2 Sample preparation and purification of Ca

Silicate samples were dissolved using a combination of Suprapur grade HF from Merck and Emsure brand (ACS, Reag. Ph. Eur) HNO₃ and HCl acids from Merck while carbonate samples were dissolved in 1 M HCl. The human blood sample, which was obtained by our lab in an oxidized form, was dissolved using a combination of 1 : 1 HCl and HNO₃. All samples were processed in a class 100 working area using established laboratory protocols.³⁷ Aliquots of sample solutions, with approximately 10 μg of Ca, were mixed with the ⁴³Ca–⁴⁸Ca double spike in screw cap Teflon vials. Each sample–DS mixture, mixed in optimal proportions as discussed below, was equilibrated at 80 °C on a hot plate for 4 to 6 hours and subsequently, evaporated. The dried mixture was taken up in 2.5 M HCl and processed for the separation of Ca using ion-exchange chromatography. Calcium was separated from the sample matrix using Teflon columns (8 cm long with an inner diameter of 0.4 cm) filled with BioRad AG50W-X8 (200–400 mesh) resin using 2.5 M HCl. Each sample was processed twice using ion-exchange chromatography for better purification. The Ca yield was between 80 and 90% for all the samples. Potential Ca isotopic fractionation due to low yields is also corrected by using the double-spike technique.³⁸ The purified Ca fraction was dried and treated with 50 μl of concentrated, ultra-pure HNO₃ to oxidize any resin that may have leaked out of the column. This HNO₃ treatment results in better signal stability during the isotope ratio measurements.

The relative contributions of Ca from the sample and the DS in the sample–DS mixture, referred to as the sample/spike ratio, is a critical parameter which needs to be optimized, failing which, δ^44/40Ca measurements can be inaccurate.³⁹ The sample/spike ratio was monitored by measuring the ⁴⁰Ca/⁴⁸Ca ratio in the mixture where ⁴⁰Ca is dominantly derived from the sample and ⁴⁸Ca from the DS. The optimal range of ⁴⁰Ca/⁴⁸Ca in the sample–DS mixtures was determined by mixing different amounts of the DS to a fixed amount (10 μg Ca) of the NIST SRM 915b Ca isotopic standard. Our results indicate that accurate δ^44/40Ca can be obtained if the ⁴⁰Ca/⁴⁸Ca in the sample–DS mixture is kept between 30 and 70. At higher values of ⁴⁰Ca/⁴⁸Ca (under-spiked samples), the δ^44/40Ca of the sample deviates from the true value (Fig. 1). The optimal range of ⁴⁰Ca/⁴⁸Ca determined in our study is consistent with earlier studies.⁵ For all the samples and standards analysed in this study, the amount of double spike was added in a manner such that the ⁴⁰Ca/⁴⁸Ca ratio of the sample–DS mixture was between 30 and 70 and mostly between 40 and 60.

Fig. 1 A plot of the measured δ^44/40Ca of the SRM 915b standard versus different values of the ⁴⁰Ca/⁴⁸Ca of SRM 915b and the ⁴³Ca–⁴⁸Ca double spike mixture. Since ⁴⁰Ca is dominantly coming from the samples and ⁴⁸Ca from the double spike, the ⁴⁰Ca/⁴⁸Ca of the mixture essentially reflected the sample/spike ratio. Accurate δ^44/40Ca ratios for the SRM 915b, shown by the horizontal grey bar, were obtained when the ⁴⁰Ca/⁴⁸Ca ratio of the sample–spike mixture was kept between 30 and 70. At high values of ⁴⁰Ca/⁴⁸Ca (under-spiked mixture), the δ^44/40Ca values became inaccurate.

2.3 Sample loading

In order to test the effects of different sample loading methods on δ^44/40Ca measurements, we analysed standards using three different loading techniques, all using a double-filament geometry. Details of these techniques are shown in Table 1. In the first technique (labeled loading #1), 5 μg of Ca was loaded on pre-degassed (kept at 4.5 A for 60 min) zone-refined Re filaments (H. CROSS Co., NJ) with 5% double-distilled HNO₃ (Emsure brand, Merck, ACS, Reag. Ph. Eur). This loading technique was adopted from Farkaš et al.^8,9 In the second technique (labeled loading #2), 5 μg of Ca was loaded on pre-degassed (kept at 3.5 A for 60 min) Ta filaments with 1.3 M H₃PO₄. The Ta filaments were degassed seven days prior to loading the samples to allow for mild oxidation of the degassed Ta filament under atmospheric conditions. In a slightly different version of loading #2, the samples were loaded on Ta filaments using 0.03 M H₃PO₄. No difference in data acquisition or signal stability was observed with change in the molarity of the H₃PO₄ and hence, we used 1.3 M H₃PO₄ for our study (Table 1). In the third type of sample loading (labeled loading #3), we loaded approximately ∼3 μg Ca with 5% double distilled HNO₃ (Emsure brand, Merck, ACS, Reag. Ph. Eur) on pre-degassed Ta filaments. These Ta filaments, after degassing, were not left for atmospheric oxidation. This step was followed by loading 0.5 μl Ta₂O₅ on the top of the loaded sample. For all three types of loading, zone-refined Re filaments were used as the ionization filament. The cost of analyses for loading #2 and #3, involving Ta and Re filaments, is lower than that of loading #1, involving double-Re filaments, as Ta filaments are commercially available at cheaper prices compared to zone-refined Re filaments.

Table 1 Summary of the three different sample loading methods attempted in this study. Loading #3 is our preferred protocol

Parameters	Loading #1	Loading #2	Loading #3
Amount of Ca loaded	5 μg	5 μg	2.5–3 μg
Filament assembly	Double	Double	Double
Evaporation filament	Re (zone refined from H. CROSS Co., NJ)	Ta (from H. CROSS Co., NJ)	Ta (from H. CROSS Co., NJ)
Ionization filament	Re (zone refined from H. CROSS Co., NJ)	Re (zone refined from H. CROSS Co., NJ)	Re (zone refined from H. CROSS Co., NJ)
Loading acid	5% HNO3	1.3 M H3PO4	5% HNO3
Activator	None	None (Ta filament allowed to oxidize in air post degassing)	0.5 μl of tantalum oxide (Ta2O5) added on the top of the sample
Ionization current	2900–3200 mA	3000–3400 mA	3000–3400 mA
Evaporation current	225 mA	1900–2000 mA	1900–2000 mA
Pyrometer reading (°C)	1400–1500	1400–1500	1400–1500

---

In all three loading methods, after loading the sample on the filament, the filament current was increased to 0.8 A, until the sample dried. In loading #3, Ta₂O₅ was added on top of the dried sample and again dried at 0.8 A. The filament current was further increased to 1.8 A and kept for 60 s and then to 2.2 A for 30 s after which the current was slowly reduced to zero. Samples loaded with 5% HNO₃ turned greenish white in colour upon drying while samples loaded with H₃PO₄ looked brownish in colour after drying. Within an hour of loading, the samples were inserted into the TIMS source.

2.4 Calcium isotope ratio measurements

Calcium isotopic measurements were performed using a TIMS at the Centre for Earth Sciences, IISc, Bangalore which was operated in positive ionization mode and is equipped with nine Faraday cups, all with 10¹¹ Ω resistors. For samples loaded with loading #1, the evaporation filament current was increased at 100 mA s⁻¹ and kept at 225 mA while the ionization filament current was increased at 300 mA s⁻¹ and kept between 2900 and 3200 mA. For samples loaded using loading #2 and loading #3, the evaporation filament current was increased at 100 mA s⁻¹ and kept between 1900 and 2000 mA, while the ionization filament current was increased at 300 mA s⁻¹ to a maximum of 3400 mA (Table 1). Data acquisition was performed at a pyrometer reading of 1400–1500 °C although, for double filaments this temperature does not reflect the filament temperature. During analyses, the ⁴⁰Ca signal was kept between 6 and 10 V during which the ⁴²Ca signal is less than 100 mV. Calcium isotope ratio measurements were performed in a dynamic mode. ⁴⁰Ca, ⁴¹K, ⁴²Ca, ⁴³Ca, and ⁴⁴Ca signals were measured in the first sequence in cups L3, L1, C, H2, and H4, respectively, while ⁴⁴Ca, ⁴⁶Ca, and ⁴⁸Ca signals were measured in the second sequence using cups L2, C, and H3, respectively (Table 2). Amplifier rotation was not applied. The ⁴¹K signal was monitored during the analysis for correcting the contribution of ⁴⁰K to ⁴⁰Ca. The intensity of ⁴¹K (natural abundance of 6.73%) was <10 mV for the silicate samples and <1 mV for the carbonate samples thereby ruling out any significant contribution of ⁴⁰K (natural abundance of 0.117%) to ⁴⁰Ca. Hence, external correction for ⁴⁰K on ⁴⁰Ca was not applied. For each analysis, the integration time was set to 16.77 s with 6 s idle time between the two measurement sequences. For focusing and peak centering, ⁴⁰Ca was used in the first sequence while ⁴⁴Ca was used for peak centering in the second sequence. Baseline measurements and lens focusing were done after every 4^th block. Data acquisition was done for a minimum of 10 blocks with 10 cycles in each block. The TIMS operational conditions are summarised in Table 2.

Table 2 TIMS operational conditions during Ca isotopic measurements

Parameters	Parameter values
^40Ca beam intensity	6–10 V
Accelerating voltage	10 kV
Collectors	Faraday cups with 10^11 Ω resistors
Source pressure	<1.5 × 10^−7 mbar
Analyser pressure	<9 × 10^−9 mbar
Sequence 1 (S1)	^40Ca (L3), ^41K (L1), ^42Ca (centre), ^43Ca (H2), ^44Ca (H4)
Sequence 2 (S2)	^44Ca (L2), ^46Ca (centre), ^47Ti (H1), ^48Ca (H3)
Idle time	6 s
Integration time	16.7 s
Blocks	Minimum 10
Cycles per block	10
Gain calibration	Once a day before the start of measurements
Focus	At the beginning of each analysis
Peak centre (control mass)	Every 4 blocks (control mass: ^40Ca-S1; ^44Ca-S2)
Baseline	Every 4 blocks
Total analysis time	∼4 h

---

2.5 Data reduction using an iterative algorithm

The ⁴⁰Ca/⁴⁴Ca, ⁴²Ca/⁴⁴Ca, and ⁴³Ca/⁴⁴Ca ratios measured in the first sequence and the ⁴⁸Ca/⁴⁴Ca ratio measured in the second sequence were used to determine the δ^44/40Ca values of the samples. Instrumental mass fractionation correction using the double-spike technique was done offline using an iterative algorithm adapted from Heuser et al.³⁰ which was originally proposed by Compston and Oversby⁴⁰ for lead isotope ratio measurements. The “Russell value”³⁸ was used as the initial value for the iteration (Table 3). The isotopic composition of the “IUPAC value” was not used as the starting value as it results in large errors in δ^44/40Ca calculations, as suggested by Heuser et al.³⁰

Table 3 Isotopic compositions of the ⁴³Ca and ⁴⁸Ca single spikes and the ⁴³Ca–⁴⁸Ca double spike used in the present study. The isotopic composition of the ⁴³Ca–⁴⁸Ca double spike was obtained using total evaporation using TIMS and the measured ⁴³Ca/⁴⁸Ca of the double spike is consistent with that estimated using gravimetry. See the text for more details

Isotopes of Ca	Natural isotopic abundance (%)	^43Ca spike isotopic composition (%)	^48Ca spike isotopic composition (%)	^43Ca–^48Ca double spike composition	Bulk terrestrial Ca “Russell Values”^a
a Data from Russell et al.^38
^40Ca	96.9	10.13	2.1	^40Ca/^48Ca = 0.1222561	531.54
^42Ca	0.65	0.78	0.024	^42Ca/^48Ca = 0.0072123	3.5194
^43Ca	0.14	83.93	>0.01	^43Ca/^48Ca = 0.7210064	0.73115
^44Ca	2.1	5.06	0.072	^44Ca/^48Ca = 0.0443720	11.273
^46Ca	0.003	>0.001	>0.01
^48Ca	0.19	0.09	97.8

---

3 Results

In this study, NIST carbonate standards SRM 915a and SRM 915b and a seawater standard NASS 6 were analysed for their Ca isotopic compositions using three different loading techniques (Table 1). The results of these analyses are shown in Table 4 and Fig. 2 and are consistent with previous studies.^{5,8,21,28,31,32} Calcium isotopic compositions are expressed as deviations in parts per thousand with respect to the NIST standard SRM 915a using the delta notation, where δ^44/40Ca = [(^44/40Ca_sample/^44/40Ca_{SRM 915a}) − 1] × 1000 and δ^42/44Ca = [(^42/44Ca_sample/^42/44Ca_{SRM 915a}) − 1] × 1000. The ⁴⁴Ca^/40Ca ratio of the SRM 915a standard used for all our measurements is 0.02157692. We have also analysed silicate rock standards from the United States Geological Survey (USGS) which include BCR-2 (Columbia River basalt) and BHVO-2 (Hawaiian basalt), carbonate standards from the Geological Survey of Japan (GSJ) which include JCp-1 (Porites coral),⁴¹ JCt-1 (clam shell),^41,42 a limestone CRM from the Bureau of Analyzed Samples Ltd. UK, ECRM 752-1 (alternative name BCS-CRM 393),⁴³ and a sample of human blood which was obtained by our lab in an oxidized form (Table 5 and Fig. 5). The data presented in Tables 4 and 5 and plotted in Fig. 5 were obtained using loading #3 (Table 1), which is our preferred loading technique. For loading #3, our internal precision (2σ_SEM) is better than ±0.03‰ for δ^44/40Ca and ranges from ±0.04 to 0.06‰ for δ^44/42Ca measurements. The external reproducibility (2σ_SD) for δ^44/40Ca, based on multiple measurements of the standards SRM 915a, SRM 915b and NASS 6 seawater, is better than 0.08‰ (Table 4). Using loading #3, the external reproducibility for δ^44/42Ca ranges from ±0.11 to 0.20‰ (Tables 4 and 5). For loading #1 and #2, the external reproducibility for δ^44/40Ca is worse and ranges from ±0.16 to 0.26‰ (Table 4 and Fig. 3).

Table 4 Ca isotopic compositions of SRM 915a and 915b and the seawater standard NASS-6 analysed using the three different sample procedures. Loading #3 is our preferred protocol

Sample Id	Loading #1 (Re–Re–5% HNO3)	Loading #2 (Re–Ta–4 N H3PO4)		Loading #3 (Re–Ta–Ta2O5)
	δ ^44/40Ca	δ ^44/40Ca	δ ^44/42Ca	δ ^44/40Ca	δ ^44/42Ca
SRM 915a_1				0.06	−0.03
SRM 915a_2				0.02	0.04
SRM 915a_3				0.00	−0.06
SRM 915a_4				−0.02	0.01
SRM 915a_5				0.00	−0.13
SRM 915a_6				0.00	0.13
SRM 915a_7				0.04	−0.08
SRM 915a_8				0.06	0.01
SRM 915a_9				−0.03	−0.03
Average				0.01	−0.02
2SD				0.07	0.11
SRM 915b_1	0.78	0.72	0.40	0.71	0.21
SRM 915b_2	0.62	0.74	0.45	0.73	0.34
SRM 915b_3	0.74	0.77	0.39	0.72	0.31
SRM 915b_4	0.69	0.76	0.49	0.72	0.2
SRM 915b_5	0.70	0.61	0.49	0.74	0.4
SRM 915b_6	0.71	0.59	0.56	0.69	0.34
SRM 915b_7		0.75	0.43	0.78	0.43
SRM 915b_8		0.87	0.30	0.72	0.37
SRM 915b_9		0.78	0.34
SRM915b_10		0.77	0.41
Average	0.71	0.74	0.43	0.73	0.32
2SD	0.11	0.16	0.15	0.05	0.16
NASS 6_1	1.79	1.8	0.98	1.78	0.89
NASS 6_2	1.96	1.82	0.93	1.76	1.02
NASS 6_3	1.70	1.88	0.83	1.85	0.87
NASS 6_4	1.88	2.01	0.81	1.75	0.79
NASS 6_5	1.70	1.72	1.15	1.78	0.88
NASS 6_6		1.93	0.88	1.85	0.84
NASS 6_7		1.66	0.88	1.84	0.92
NASS 6_8		2.04	1.10
NASS 6_9		1.78	0.96
NASS 6_10		1.82	0.94
NASS 6_11		2.12	0.98
NASS 6_12		1.94	0.86
NASS 6_13		1.79	1.02
NASS 6_14		1.77	0.97
Average	1.81	1.86	0.95	1.80	0.89
2SD	0.22	0.26	0.19	0.08	0.14

---

Fig. 2 δ ^44/40Ca of the NIST standards SRM 915a and SRM 915b and seawater NASS 6 analysed with our preferred sample loading technique (#3, Table 1) using a ⁴³Ca–⁴⁸Ca double spike. The external reproducibility (2σ_SD) for δ^44/40 Ca was better than ±0.08‰.

Table 5 The Ca isotopic compositions of GSJ carbonate standards JCp-1 (coral) and JCt-1 (clam shell), ECRM 752-1 (limestone), USGS silicate rock standards BCR 2 (Columbia River Basalt) and BHVO 2 (Hawaiian basalt) and a human blood sample. Also shown for comparison are the recommended values for BCR 2 and BHVO 2 from He et al.²⁸

Sample Id	δ ^44/40Ca	δ ^44/42Ca
JCt-1_1	0.81	0.48
JCt-1_2	0.85	0.46
JCt-1_3	0.81	0.38
JCt-1_4	0.86	0.51
Average	0.83	0.46
2SD	0.05	0.11
JCp-1_1	0.80	0.41
JCp-1_2	0.84	0.46
JCp-1_3	0.88	0.54
JCp-1_4	0.83	0.39
JCp-1_5	0.88	0.45
Average	0.85	0.45
2SD	0.07	0.12
ECRM 752_1	0.77	0.39
ECRM 752_2	0.85	0.56
ECRM 752_3	0.85	0.48
ECRM 752_4	0.84	0.52
ECRM 752_5	0.83	0.36
Average	0.83	0.46
2SD	0.07	0.17
USGS BCR 2_1	0.91	0.6
USGS BCR 2_2	0.93	0.42
USGS BCR 2_3	0.86	0.36
USGS BCR 2_4	0.88	0.56
USGS BCR 2_5	0.89	0.45
Average	0.89	0.48
2SD	0.05	0.20
BCR 2, He et al. (2017)	0.839 ± 0.145	0.423 ± 0.078
USGS BHVO 2_1	0.93	0.47
USGS BHVO 2_2	0.87	0.51
USGS BHVO 2_3	0.85	0.4
USGS BHVO 2_4	0.80	0.39
USGS BHVO 2_5	0.83	0.56
Average	0.86	0.47
2SD	0.10	0.14
USGS BHVO 2, He et al. (2017)	0.759 ± 0.044	0.384 ± 0.055
Human blood	−1.85	−1.0
Human blood	−1.8	−0.85
Average	−1.83	−0.92
2SD	0.07	0.18

---

Fig. 3 δ ^44/40Ca of the NIST standards SRM 915a and SRM 915b and seawater NASS 6 analysed using the three different sample loading techniques. All measurements were done using a ⁴³Ca–⁴⁸Ca double spike. The external reproducibility of these measurements (2σ_SD) is the least when loading 3, our preferred technique, is used. See Tables 1 and 4 and the text for more details.

4 Discussion

The external reproducibility of δ^44/40Ca measurements is the limiting factor for wider application of Ca stable isotopes.⁴⁴ We investigate the factors that could hinder the external reproducibility of δ^44/40Ca measurements and suggest a new sample loading protocol and a method to monitor the fractionation of the samples, which significantly improves the external reproducibility of δ^44/40Ca measurements.

4.1 Choice of the double spike

To correct for isotopic fractionation during ion-exchange chromatography as well as mass spectrometry, Russell et al.³⁸ proposed mixing a ⁴²Ca–⁴⁸Ca DS with the sample prior to ion-exchange chromatography. The authors claimed that the large mass difference between ⁴²Ca and ⁴⁸Ca reduces the instrumental mass fractionation per amu for ⁴⁴Ca/⁴⁰Ca ratio measurements. Other research groups have used different combinations of double spikes like ⁴³Ca–⁴⁸Ca,^5,3042Ca–⁴³Ca^31,44 and ⁴³Ca–⁴⁶Ca⁴⁵ for δ^44/40Ca measurements. As the average mass of ⁴²Ca and ⁴³Ca in the ⁴²Ca–⁴³Ca DS (mean mass 42.5) is close to the average mass of ⁴⁴Ca and ⁴⁰Ca (mean mass 42), it was suggested based on the “mean mass law” that the use of the ⁴²Ca–⁴³Ca DS should minimise the error in ⁴⁴Ca/⁴⁰Ca measurements compared to when using the ⁴³Ca–⁴⁸Ca (mean mass 45.5) or ⁴²Ca–⁴⁸Ca (mean mass 45) double spikes.⁴⁴ Lehn et al.^31,32 argued that the ⁴²Ca–⁴³Ca DS yields better external reproducibility for δ^44/40Ca measurements compared to the ⁴³Ca–⁴⁸Ca DS. For δ^44/40Ca measurements using the ⁴³Ca–⁴⁸Ca DS, Lehn et al.³¹ reported internal precisions ranging from 0.013 to 0.081‰ and an average external reproducibility of ±0.223‰ for standards with different matrices. In contrast, when using the ⁴²Ca–⁴³Ca DS, the external reproducibility for δ^44/40Ca measurements for a silicate standard was 0.033‰ which is only marginally worse than the internal precision of 0.023‰ using the ⁴²Ca–⁴³Ca DS but significantly better than the external reproducibility when using the ⁴³Ca–⁴⁸Ca DS.³¹ Additionally, when the ⁴²Ca–⁴³Ca DS is used, only the ⁴⁰Ca, ⁴²Ca, ⁴³Ca and ⁴⁴Ca signals have to be measured. This measurement can be done in a single sequence thereby reducing the δ^44/40Ca measurement time. However, theoretical studies by Rudge et al.³⁹ have shown that error propagation is approximately half when using the ⁴³Ca–⁴⁸Ca DS compared to the ⁴²Ca–⁴³Ca DS. In addition, using the ⁴³Ca–⁴⁸Ca DS allows simultaneous determination of δ^44/40Ca and δ^44/42Ca values and hence, a direct comparison with δ^44/42Ca values from MC-ICPMS measurements is possible.

In this study, the Ca isotopic compositions of NIST SRM 915a, NIST SRM 915b, and seawater (NASS 6) were measured multiple times using the ⁴³Ca–⁴⁸Ca double spike. The measured external reproducibility for δ^44/40Ca (2σ_SD) is ±0.07‰ (n = 9) for SRM 915a, ± 0.05‰ (n = 8) for SRM 915b and ±0.08‰ (n = 7) for seawater (NASS 6) (Table 4). The external reproducibility achieved in the present study for δ^44/40Ca measurements using the ⁴³Ca–⁴⁸Ca double spike is comparable with those reported in studies using the ⁴²Ca–⁴³Ca double spike.^{16,21,31,32,46} It may be noted that some of the studies using the ⁴²Ca–⁴³Ca double spike analyse the samples at much higher operational voltages compared to those in the present study (6–10 V on ⁴⁰Ca). The operational voltages in the present study were intentionally kept low as high operational voltages are known to degrade the Faraday cups.⁴⁷ The external reproducibilities for δ^44/42Ca measurements in the present study (0.11–0.2‰) are worse compared to those for δ^44/42Ca measurements using MC-ICPMS.²⁶ We believe that the lower precision of δ^44/42Ca measurements is due to the low signal (<100 mV) of ⁴²Ca. At higher intensities of ⁴²Ca, the external reproducibility of δ^44/42Ca using the proposed protocol is expected to be much better. We believe that the ⁴³Ca–⁴⁸Ca DS can yield high precision δ^44/40Ca and δ^44/42Ca values. Instead of the composition of the DS, the sample loading technique and quantitative monitoring of sample fractionation during analysis play a key role in improving the external reproducibility of the δ^44/40Ca measurements.

4.2 Sample purification

A pre-requisite for generating a stable ion beam during isotope ratio measurements is the quantitative removal of the sample matrix by using ion-exchange chromatography. Small amounts of impurities (e.g., Fe, Al, and Mg) present in the purified sample being analyzed can suppress Ca ionization in the TIMS source. The presence of potassium (K) in the samples can result in direct isobaric interference of ⁴⁰K with ⁴⁰Ca. As K has a low ionization potential (3.43 eV) compared to Ca (6.11 eV), Fletcher et al.⁴⁸ proposed that rapid heating of the central filament (for a triple filament assembly) or evaporation filament (for a double filament assembly) can reduce the interference from K by evaporating the K on the filament before reaching the optimum ⁴⁰Ca⁺ ion beam. During analysis, we monitored the ⁴¹K signal for correcting the interference of ⁴⁰K⁺ (⁴⁰K/⁴¹K = 0.001738) with ⁴⁰Ca⁺. For the samples and standards analysed in this study, the ⁴¹K⁺ signal was less than 1 mV during Ca isotope data acquisition and the typical signal of ⁴⁰Ca during these measurements was about 6–10 V. Hence, the contribution of ⁴⁰K to ⁴⁰Ca was negligible. Some studies^8,31 have skipped the ion-chromatographic purification of NIST SRM 915a or SRM 915b carbonate standards because of their pure Ca carbonate matrix. However, the ⁴³Ca spike has a certified value of <100 ppm of K and 5000 ppm of Sr, and hence, the chemical purification of the sample–DS mixture is essential. In the present study, we have chemically purified the samples and standards using cation exchange chromatography after the addition of the double spike.

For samples with high K, for e.g., old crustal rocks, variations in ⁴⁰Ca could also be due to the decay of ⁴⁰K and this radiogenic ⁴⁰Ca component can affect the δ^44/40Ca values in these samples. The contribution of the radiogenic ⁴⁰Ca can be corrected using the K/Ca ratio and the age of the samples to yield accurate δ^44/40Ca values.²⁶ Additionally, an unspiked aliquot of the sample needs to be analyzed to determine the starting isotopic composition which is used for the iterative procedure to calculate the δ^44/40Ca of the rock.

4.3 Sample loading

Sample loading on the filament plays a critical role in TIMS measurements as it affects the stability of the signal and the extent of isotopic fractionation, which in turn could affect the accuracy and internal precision of the measurements. For δ^44/40Ca measurements, several loading techniques have been proposed. Russell et al.³⁸ loaded 5–10 μg of Ca as a chloride or nitrate salt on a pre-oxidized V-shaped zone refined single Ta filament. Subsequently, different filament assemblies involving single, double and triple filaments on different generations of TIMS manufactured by different companies as well as different loading techniques using HNO₃, HCl, and H₃PO₄ acids^21,49,50 have been attempted. The use of activators like Ta₂O₅ has also been proposed.³⁰ In the last few years, Ca stable isotope data (δ^44/40Ca) have mostly been generated using either a GV IsoProbe-T or a Thermo Scientific Triton (Plus) TIMS, with the Triton (Plus) being the more widely used TIMS for Ca isotopic measurements. The IsoProbe-T uses either a single or a triple filament configuration whereas the Triton (Plus) uses either a single or a double filament geometry. In the present study, a double filament geometry has been used.

We have tried three different loading techniques in this study as summarized in Table 1. Loading #1 and #2 involve Re double filament configurations. We have tested two types of Re ribbons (synonymously used as filaments) procured from two different commercial sources. Our initial measurements were performed using Re filaments procured from Thermo Fischer Scientific, while most of our measurements were performed using zone-refined Re ribbons procured from H. CROSS, NJ, USA. We have used the same DS for all our experiments. We have noticed that the extent of fractionation of the samples when using the two different Re ribbons was different (Fig. 4). While the exact reason for this difference is unclear, minor differences in the properties of these filaments procured from the two different manufacturers, e.g., their thickness, could affect the heating characteristics and hence, the fractionation of the sample–double spike mixture.

Fig. 4 Double spike corrected ⁴⁴Ca/⁴⁰Ca ratio on the y-axis plotted against the time of analysis. Measurements were done using two different types of Re filaments. Here note that the systematic increase of δ⁴⁴Ca/⁴⁰Ca = 0.43‰ for measurements with Re filaments procured from Thermo Scientific compared to zone refined Re filaments provided by HCROSS. For SRM 915b and NASS 6 analyses, loading #1 was used. Values of SRM 915a, loaded using loading #3, are shown for comparison.

Fig. 5 A plot of δ^44/42Ca versus δ^44/40Ca for the different standards and samples analysed in this study using our preferred loading method and proposed fractionation monitoring technique. For comparison, the black dotted line represents the theoretically calculated exponential fractionation line. External reproducibility (2σ_SD) for δ^44/40 Ca is less than ±0.08‰ but for δ^44/42Ca measurements, it is worse due to the low intensity of ⁴²Ca. See the text for details.

In our preferred loading protocol (#3, Table 1), we used a combination of Re and Ta filaments using a double filament configuration. Calcium was loaded as a nitrate on Ta filaments, followed by the addition of Ta₂O₅. Zone-refined Re filaments were used as ionization filaments. This sample loading protocol yielded stable Ca signals of 6–10 V which could be sustained for ∼5 hours. While the GV IsoProbe-T TIMS collectors have a dynamic range of 10 V, the Thermo Triton Plus has a higher dynamic range of 50 V, both using the standard 10¹¹ Ω resistors. Most Ca isotope studies using a Triton (Plus) use high operational voltages (e.g., 20 V ^31,32) on ⁴⁰Ca. While high operational voltages yield better counting statistics, the collectors are easily damaged which in turn affects the long-term reproducibility of δ^44/40Ca measurements. Low operational voltages, as used in the present study, improve the life-time of the Faraday cups which are expensive to replace. The use of Ta ribbons for loading the samples also reduces the cost of Ca isotopic analyses as Ta ribbons are cheaper than zone-refined Re ribbons. The Ta–Ta double filament configuration was also attempted but the signal stability was poor with this filament combination and hence, not used for further measurements. While stable Ca beams with controlled fractionation improved the internal precision (2σ_SEM) of our analyses (better than ±0.03‰), the external reproducibility of the data, based on multiple analyses using different filaments, requires reproducible fractionation patterns. We propose a method for ensuring reproducible running conditions which is described in Section 4.4.

4.4 Monitoring isotopic fractionation during analysis

Although, the combination of filaments and sample loading protocols as well as filament heating conditions in the TIMS are kept constant for all the samples, the sample spot size after drying and its exact position on the filament can vary from one filament to another. Due to such differences, the evaporation and ionization pattern of the same sample loaded on different filaments are not reproducible during TIMS measurements. To monitor and compare the fractionation of samples on different filaments, we have introduced a parameter called beta (β) which is defined as:

β = ln[(⁴³Ca/⁴⁸Ca)_DS/(⁴³Ca/⁴⁸Ca)_{sample–DS mix}]/ln(42.958/47.952)

where the mass of ⁴³Ca = 42.958 and that of ⁴⁸Ca = 47.952.

The signal of ⁴³Ca and ⁴⁸Ca in the sample–DS mixture is primarily derived from the DS. So the β value is not affected by the inherent composition of the sample and primarily reflects the extent of fractionation of the sample to DS ratio in the mixture. During the course of analysis of a sample, the β value decreases (Fig. 6). Higher values of β reflect sub-optimal fractionation of the sample–DS mixture whereas very low values of β reflect higher fractionation of the sample–DS mixture at high filament temperatures. For loading #3, using running conditions listed in Table 1, the β values of the samples typically ranged from 0.35 at the beginning of the analyses to 0.1 at the end. By performing repeated analyses of NIST SRM 915a and SRM 915b and seawater (NASS 6), where the ⁴⁰Ca/⁴⁸Ca of the sample–DS mixture is kept between 30 and 70, we determined the range of β values during which the δ^44/40Ca of these standards and the seawater showed accurate and reproducible values (Fig. 6). This β value range is considered as optimal and is between 0.3 and 0.1. All the other samples were analyzed within the same range of β values. At optimal β values, at least 100 cycles (10 blocks) were measured for each sample and standard. The internal precision (2σ_SEM) for δ^44/40Ca of the samples and standards after 10–15 blocks was better than ±0.03‰. Analysis of all samples and standards at optimal β values resulted in an external reproducibility (2σ_SD) that is better than ±0.08‰. When β values are higher than optimal, the δ^44/40Ca values showed lower values (Fig. 6) and the reason for such deviation is not well understood.

Fig. 6 A plot of δ^44/40Ca versus beta values for SRM 915a and NASS 6 showing the range of optimal beta values (yellow vertical shade) where the δ^44/40Ca values (grey horizontal shade) are accurate. Repeated analyses of the samples and standards in the same beta value range yield an external reproducibility better than 0.08‰ for δ^44/40Ca measurements.

5 Conclusions

We propose a novel sample loading protocol and a method to monitor the fractionation of samples during δ^44/40Ca measurements using a TIMS. We demonstrate that high precision Ca isotopic measurements are possible using a ⁴³Ca–⁴⁸Ca double spike, which allows simultaneous measurements of δ^44/40Ca and δ^44/42Ca. After ion-exchange chromatography, 2–3 μg of purified Ca from the sample–DS mixture was loaded on degassed Ta filaments using 5% HNO₃ followed by addition of a Ta₂O₅ activator. A zone-refined Re filament was used as the ionization filament. Using this loading technique, a stable signal with ⁴⁰Ca, intentionally kept between 6 and 10 V, can be sustained for over 5 hours which is essential for the good internal precision of δ^44/40Ca measurements. We also propose the usage of a parameter β to monitor the fractionation of the samples in the TIMS source. When all samples and standards are analyzed under similar extents of fractionation (β value), the external reproducibility of δ^44/40Ca is better than ±0.08‰. We report δ^44/40Ca and δ^44/42Ca of USGS rock standards BHVO-2 and BCR-2 as well as carbonate standards JCp-1, JCt-1 and ECRM 752-1 (alternative name BCS-CRM 393). Our preferred loading method using a Ta–Re double filament assembly is more cost effective compared to the commonly used Re–Re double filament assembly, given the lower cost of Ta ribbons compared to zone refined Re ribbons.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Standards and samples for this study were kindly provided by Shichun Huang, Sambuddha Misra, Q.-z. Yin, Stein Jacobsen, Kaustubh Thirumalai and Ravi Rangarajan. This study was supported by DST grants (SR/FTP/ES-72/2012, SR/S4/ES-650/2012) to RC. We thank three anonymous reviewers and the editor for their comments and suggestions.

References

1. M. Schiller, C. Paton and M. Bizzarro, Geochim. Cosmochim. Acta, 2015, 149, 88–102.
2. M. C. Valdes, M. Moreira, J. Foriel and F. Moynier, Earth Planet. Sci. Lett., 2014, 394, 135–145.
3. S. Huang and S. B. Jacobsen, Geochim. Cosmochim. Acta, 2017, 201, 364–376.
4. X. Zhao, Z. Zhang, S. Huang, Y. Liu, X. Li and H. Zhang, Geochim. Cosmochim. Acta, 2017, 208, 368–380.
5. S. Huang, J. Farkaš and S. B. Jacobsen, Geochim. Cosmochim. Acta, 2011, 75, 4987–4997.
6. J.-T. Kang, H.-L. Zhu, Y.-F. Liu, F. Liu, F. Wu, Y.-T. Hao, X.-C. Zhi, Z.-F. Zhang and F. Huang, Geochim. Cosmochim. Acta, 2016, 174, 335–344.
7. B. D. Marshall and D. J. DePaolo, Geochim. Cosmochim. Acta, 1989, 53, 917–922.
8. J. Farkaš, F. Böhm, K. Wallmann, J. Blenkinsop, A. Eisenhauer, R. Van Geldern, A. Munnecke, S. Voigt and J. Veizer, Geochim. Cosmochim. Acta, 2007, 71, 5117–5134.
9. J. Farkaš, D. Buhl, J. Blenkinsop and J. Veizer, Earth Planet. Sci. Lett., 2007, 253, 96–111.
10. E. T. Tipper, A. Galy and M. J. Bickle, Earth Planet. Sci. Lett., 2006, 247, 267–279.
11. M. S. Fantle and D. J. DePaolo, Earth Planet. Sci. Lett., 2005, 237, 102–117.
12. R. S. Hindshaw, B. C. Reynolds, J. G. Wiederhold, R. Kretzschmar and B. Bourdon, Geochim. Cosmochim. Acta, 2011, 75, 106–118.
13. J.-S. Ryu, A. D. Jacobson, C. Holmden, C. Lundstrom and Z. Zhang, Geochim. Cosmochim. Acta, 2011, 75, 6004–6026.
14. D. Hippler, A. D. Schmitt, N. Gussone, A. Heuser, P. Stille, A. Eisenhauer and T. F. Nägler, Geostand. Newsl., 2003, 27, 13–19.
15. X. Chen, W. Deng, H. Zhu, Z. Zhang, G. Wei and M. T. McCulloch, Chem. Geol., 2016, 435, 71–78.
16. M. Inoue, N. Gussone, Y. Koga, A. Iwase, A. Suzuki, K. Sakai and H. Kawahata, Geochim. Cosmochim. Acta, 2015, 167, 80–92.
17. D. Hippler, A. Eisenhauer and T. F. Nägler, Geochim. Cosmochim. Acta, 2006, 70, 90–100.
18. F. Böhm, N. Gussone, A. Eisenhauer, W.-C. Dullo, S. Reynaud and A. Paytan, Geochim. Cosmochim. Acta, 2006, 70, 4452–4462.
19. T. F. Nägler, A. Eisenhauer, A. Müller, C. Hemleben and J. Kramers, Geochem., Geophys., Geosyst., 2000, 1 DOI:10.1029/2000GC000091.
20. M. Clementz, P. Holden and P. Koch, Int. J. Osteoarchaeol., 2003, 13, 29–36.
21. C. Holmden and N. Bélanger, Geochim. Cosmochim. Acta, 2010, 74, 995–1015.
22. J. Skulan, D. J. DePaolo and T. L. Owens, Geochim. Cosmochim. Acta, 1997, 61, 2505–2510.
23. A. Heuser, A. Eisenhauer, K. E. Scholz-Ahrens and J. Schrezenmeir, Isot. Environ. Health Stud., 2016, 52, 633–648.
24. J. L. Morgan, J. L. Skulan, G. W. Gordon, S. J. Romaniello, S. M. Smith and A. D. Anbar, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 9989–9994.
25. W. A. Russell, D. A. Papanastassiou, T. A. Tombrello and S. Epstein, in Lunar and Planetary Science Conference Proceedings, 1977, vol. 8, pp. 3791–3805.
26. M. S. Fantle and E. T. Tipper, Earth-Sci. Rev., 2014, 129, 148–177.
27. N. Gussone, A.-D. Schmitt, A. Heuser, F. Wombacher, M. Dietzel, E. Tipper and M. Schiller, Calcium Stable Isotope Geochemistry, Springer, Verlag, Berlin, Heidelberg, 2010, ISBN: 978-3-540-68948-5.
28. Y. He, Y. Wang, C. Zhu, S. Huang and S. Li, Geostand. Geoanal. Res., 2016, 41, 283–302.
29. S. F. Boulyga, Mass Spectrom. Rev., 2010, 29, 685–716.
30. A. Heuser, A. Eisenhauer, N. Gussone, B. Bock, B. T. Hansen and T. F. Nägler, Int. J. Mass Spectrom., 2002, 220, 385–397.
31. G. O. Lehn and A. D. Jacobson, J. Anal. At. Spectrom., 2015, 30, 1571–1581.
32. G. O. Lehn, A. D. Jacobson and C. Holmden, Int. J. Mass Spectrom., 2013, 351, 69–75.
33. M. S. Fantle and T. D. Bullen, Chem. Geol., 2009, 258, 50–64.
34. G. Cavazzini, Int. J. Mass Spectrom., 2009, 288, 84–91.
35. E. D. Young, A. Galy and H. Nagahara, Geochim. Cosmochim. Acta, 2002, 66, 1095–1104.
36. K. Habfast, Int. J. Mass Spectrom., 1998, 176, 133–148.
37. A. Banerjee, R. Chakrabarti and S. Mandal, Chem. Geol., 2016, 440, 124–138.
38. W. A. Russell, D. A. Papanastassiou and T. A. Tombrello, Geochim. Cosmochim. Acta, 1978, 42, 1075–1090.
39. J. F. Rudge, B. C. Reynolds and B. Bourdon, Chem. Geol., 2009, 265, 420–431.
40. W. Compston and V. Oversby, J. Geophys. Res., 1969, 74, 4338–4348.
41. T. Okai, A. Suzuki, H. Kawahata, S. Terashima and N. Imai, Geostand. Newsl., 2003, 26, 95–99.
42. M. Inoue, M. Nohara, T. Okai, A. Suzuki and H. Kawahata, Geostand. Geoanal. Res., 2004, 28, 411–416.
43. M. Greaves, S. Barker, C. Daunt and H. Elderfield, Geochem., Geophys., Geosyst., 2005, 6 DOI:10.1029/2004GC000790.
44. K. Gopalan, D. Macdougall and C. Macisaac, Int. J. Mass Spectrom., 2006, 248, 9–16.
45. K. Gopalan, J. D. Macdougall and C. Macisaac, Geostand. Geoanal. Res., 2007, 31, 227–236.
46. A. D. Du Vivier, A. D. Jacobson, G. O. Lehn, D. Selby, M. T. Hurtgen and B. B. Sageman, Earth Planet. Sci. Lett., 2015, 416, 121–131.
47. J. I. Simon, D. J. Depaolo and F. Moynier, Astrophys. J., 2009, 702, 707.
48. I. R. Fletcher, A. L. Maggi, K. J. R. Rosman and N. J. McNaughton, Int. J. Mass Spectrom. Ion Processes, 1997, 163, 1–17.
49. S. Huang, J. Farkaš and S. B. Jacobsen, Earth Planet. Sci. Lett., 2010, 292, 337–344.
50. B. D. Marshall and D. J. DePaolo, Geochim. Cosmochim. Acta, 1982, 46, 2537–2545.

---