Separation of Fe from whole blood matrix for precise isotopic ratio measurements by MC-ICP-MS: a comparison of different approaches

Abstract

Anion-exchange and precipitation procedures for Fe separation from unspiked human whole blood after microwave digestion and ashing decomposition techniques were thoroughly evaluated in terms of Fe recoveries, decreases in matrix element concentrations and elimination of interfering species for subsequent Fe isotope ratio measurements by multi-collector ICP-MS. During isotope ratio measurements involving ⁵⁴Fe, ⁵⁶Fe and ⁵⁷Fe, on-line mass discrimination correction using Ni isotopes was applied, significantly reducing uncertainties both within and between Fe sample runs. Despite Fe recoveries below 100% for all separation procedures studied, no artificial isotope fractionation was detected. The degree of Fe fractionation in a commercially available, whole blood sample (Trace Elements in Whole Blood, Level 1, Sero AS), expressed as ⁵⁶δ (−2.83 ± 0.06‰) and ⁵⁷δ (−4.23 ± 0.08‰) values relative to IRMM-014 Fe isotopic reference material, agrees well with previously published data. Of the tested separation procedures, precipitation using NH₃ was found to be the most rapid and cost-effective method, yielding high Fe recovery and low levels of concomitant elements.

---

Introduction

Iron was the first minor element discovered to be essential for humans, and its important function of governing oxygen transport in blood makes it an interesting element to investigate.¹ The Fe concentration needs to be maintained in the body at a defined level, but according to estimations, ca. one-fifth of the world population is affected by iron malnutrition.² Since the 1950s, adsorption studies with radioactive Fe tracers have been used to ascertain the bioavailability of iron,³ but ethical considerations⁴ and the development of improved analytical methods have resulted in increased interest in its four stable isotopes, ⁵⁴Fe, ⁵⁶Fe, ⁵⁷Fe and ⁵⁸Fe. Unfortunately, the requirement for large doses of stable isotope spikes increases costs and might lead to unexpected biological implications, which could influence the interpretation of the data.^5,6 Improvements in instrumentation and analytical methodology facilitate the use of smaller amounts of stable isotope spikes to obtain reliable information about the Fe distribution in the human body.

Recently, results have indicated a natural stable isotope fractionation of iron in blood.^7–9 This up to 3‰ deviation of the ⁵⁶Fe/⁵⁴Fe ratio from that observed in a certified Fe isotopic reference material demands high accuracy and precision of the analysis. Of the two dominating types of mass spectrometry (MS) for isotope ratio measurements in heavier elements—thermal ionization-MS (TIMS) and inductively coupled plasma-MS (ICP-MS)—the latter is becoming of increasing importance.¹⁰ The speed of analysis was one reason behind the interest in ICP-MS,⁶ but the initial expectation of a limited requirement for analyte separation and purification of the sample when using this technique has not been completely fulfilled, especially regarding isotope ratio measurements. Accurate and precise measurements of isotope ratios by ICP-MS are dependent on the minimization of bias associated with space charge effects in the plasma, as well as spectral and non-spectral interferences.¹¹ Separation of an analyte from the matrix is the preferred approach, potentially allowing an increase in the sensitivity during the instrumental measurement and a decrease in possible isobaric overlaps. The contribution of spectral interferences can also be minimized by using double focusing sector field instruments with high mass resolution options,⁶ complemented by more laborious mathematical corrections and calibrations than simple internal standardization.¹² The precision of isotope ratio measurements can be enhanced by using an instrument capable of high resolution and multiple ion beam collection. Alternative means of diminishing interferences include dry aerosol sample introduction,^7–9 collision cells and instrument operation under cool plasma conditions.^7–13 However, these approaches require additional components or a comprehensive optimization of the instrument.

A great variety of different schemes for Fe separation from blood matrix has been reported in the relevant literature. The majority of previous studies on Fe isotope composition employed anion exchange procedures for separation of Fe from whole blood.^4,7,8,15 Walczyk et al.⁴ have described a complicated preparation scheme for Fe isotopic analysis by negative ion TIMS using repeated ion-exchange separation followed by three step liquid–liquid extraction of Fe into diethyl ether. Ohno et al.⁹ separated Fe from complex organic compounds in blood using 4-methyl-2-pentanone. Fujimori and co-workers¹⁴ applied the heme–iron precipitation technique for determination of rare earth elements in bovine whole blood. As a rule, these separation techniques require initial mineralization of the whole blood, with microwave digestion being almost exclusively used for this purpose.^4,7,15 However, Bukhave et al.¹⁶ suggested a simplified method for the determination of radioactive Fe in whole blood samples using a dry-ashing procedure and Fe precipitation. Unfortunately, a comprehensive evaluation of Fe recoveries, the efficiency of matrix removal and the amount of potentially interfering elements remaining in the measuring solution has not been reported in these papers. Knowledge of the aforementioned parameters is very important, especially in light of the fact that kinetic fractionation of Fe isotopes, of up to 6‰ in terms of the ⁵⁶Fe/⁵⁴Fe ratio, was observed during sample preparation procedures including separations by anion-exchange^17,18 and precipitation techniques.¹⁹

The purpose of this study was to compare different sample preparation procedures in order to find the optimum method for the precise Fe isotopic analysis of whole blood by multi-collector ICP-MS (MC-ICP-MS). A commercially available, whole blood control material was employed in this work. Although not certified for Fe isotope ratios, the availability of this material will permit the results to be transferred to other laboratories, and provide a basis for comparison.

Experimental

Instrumentation

The single collector, double-focusing, sector field ICP-MS instrument (ICP-SFMS) used in this study was the ELEMENT (ThermoFinnigan, Bremen, Germany). Details on the instrumental operating conditions are given in Table 1.

Table 1 ICP-SFMS instrumental operating conditions and measurement parameters

Parameters
Rf power/W	1400
ICP torch	Fassel torch, 1.5 mm id
Spray chamber	Scott-type (double pass), water cooled
Nebulizer	MicroMist AR40-1-F02
Sampler cone	Nickel. 1.1 mm orifice diameter
Skimmer cone	Nickel. 0.8 mm orifice diameter
Ions lens settings	Adjusted to obtain maximum signal intensity
Gas flow rates/l min^−1:
Coolant	13
Auxiliary	0.85
Nebulizer	0.85–0.92
Scan type	E-Scan
Acquisition mode	Counting
No. of scans	12 for each resolution
Acquisition window (%)	50 in LRM; 120 in MRM
Search window (%)	50 in LRM; 80 in MRM
Integration window (%)	50 in LRM; 60 in MRM
Dwell time per sample/ms	10 in LRM; 20 in MRM
No. of samples per nuclide	30 in LRM, 25 in MRM
Isotopes:
Low resolution mode	^85Rb, ^115In
Medium resolution mode	^23Na, ^26Mg, ^31P, ^32S, ^39K, ^44Ca, ^52Cr, ^56Fe, ^60Ni, ^63Cu, ^64Zn, ^115In

---

The MC-ICP-MS instrument Neptune (ThermoFinnigan) was used for measurements of Fe isotopic composition. Instrumental settings used for the analysis are shown in Table 2. The Neptune is equipped with nine Faraday cups and one ion counter. The stable introduction system used consisted of a tandem quartz spray chamber arrangement (cyclone + standard Scott double pass) together with a low-flow PFA microconcentric nebulizer. Iron isotope ratio measurements were performed using the high resolution setting of the entrance slit. This allows straightforward elimination of many spectral interferences on Fe and Ni isotopes (e.g., ArO⁺, CaO⁺, ArN⁺, CaC⁺, KO⁺, etc.) without applying ‘cold’ plasma conditions, desolvation of sample aerosols or collision/reaction cells.¹³ However, the separation of isobaric overlap from ⁵⁸Ni and ⁵⁴Cr on the corresponding Fe isotopes requires a mass resolution that is unattainable with commercially available instruments. Hence, these interferences may alter the measured Fe isotope ratios unless appropriate chemical separation is applied. Analyses were conducted in static mode, monitoring the following mass-to-charge ratios with Faraday cups: 54(Fe + Cr), 56(Fe), 57(Fe), 58(Fe + Ni), 60(Ni) and 62(Ni). Each analysis consisted of 9 blocks of data, each block comprising 10 scans of 8 s duration. Baseline correction was performed at the beginning of each block.

Table 2 Instrumental operating conditions and measurement parameters for multi-collector ICP-MS

Parameters
Rf power/W	1200
Accelerating voltage/V	−10000
Sampler and skimmer cones	Ni. 1.1 mm and 0.8 mm orifice diameter
Argon gas flow rates/l min^−1:
Coolant	16
Auxiliary	0.60
Nebulizer	∼1.2 (optimized daily)
Magnet setting/u	57.893 (centre cup)
Sample uptake rate /µl min^−1	30-40 (self aspiration)
Isotopes	^54(Fe + Cr), ^56Fe, ^57Fe, ^60Ni and ^62Ni

---

For sample digestion a laboratory muffle furnace (Termolyne 48000, Termolyne Instruments, Dubuque, USA) and a microwave oven (MDS-2000, CEM, Matthews, NC, USA) equipped with 12 low-volume perfluoroalkoxy (PFA) lined vessels (ACV 50) with safety rupture membranes (maximum operating pressure 1380 kPa) were used. The centrifuge used was a Megafuge 1.0, Haraeus Sepatech, Hannover, Germany.

Reagents and samples

All calibration and internal standard solutions used were prepared by diluting single-element standard solutions (SPEX Plasma Standards, Edison, NJ, USA) taking into account inter-element compatibility. Normal Fe isotopic reference material IRMM-014, provided by the Institute of Reference Material and Measurements, Geel, Belgium was also used during measurements of Fe isotope ratios. Analytical-reagent grade nitric acid (Merck, Darmstadt, Germany) was utilized after additional purification by sub-boiling distillation in a quartz still. Supra-pure grade hydrochloric acid (Baker Chemical, Phillipsburg, NJ, USA) and 25% ammonia solution (Merck, Darmstadt, Germany) were used as supplied. NaOH (Merck, Darmstadt, Germany) was of analytical grade. For the preparation of digestion blanks and standards, as well as for dilution of treated blood samples, Milli-Q water (Millipore Milli-Q, Bedford, USA) was additionally purified by sub-boiling distillation in a Teflon still (Savillex Corp., Minnetonka, Minnesota, USA). Trace Elements in Whole Blood, Level 1 Control blood material was purchased in freeze-dried form from Sero AS, Asker, Norway (batch no. OK0336). Reconstitution of the blood was performed in accordance with the procedure recommended by the supplier. The anion-exchange resin used was Dowex 1-8, 50-100 mesh (Serva Feinbiochemica, Heidelberg, Germany).

Sample preparations

The flowchart of the sample preparation procedures tested during this study is shown in Fig. 1. Following is a description of the individual steps.

Fig. 1 Flowchart illustrating the sample treatment operations involved in the preparation of Sero AS whole blood for the measurement of Fe isotope ratios by MC-ICP-MS. Procedure efficiencies were also determined by analysing treated samples using ICP-SFMS.

MW digestion

Microwave-assisted digestion of the whole blood was performed in closed digestion vessels in accordance with a procedure described previously.²⁰ In short, an aliquot of reconstituted blood (1 ml) was digested with 1.5 ml of 6 M HNO₃ for 30 min at 300 W power. After cooling to room temperature, the digest was poured into polypropylene autosampler tubes and diluted to 10 ml with Milli-Q water. Prior to use, these tubes were thoroughly cleaned in a sequence with detergent, water, a mixture of HNO₃ (1.4 M) and HCl (1.1 M) followed by soaking in HNO₃ (0.7 M) for at least 12 h and a final rinse with Milli-Q water.

Ashing

An aliquot of reconstituted blood (1 ml) was transferred into acid-washed Pyrex containers and ashed in a muffle furnace by slowly raising the temperature to 400 °C during 2 h. The containers were held at this temperature for 3 h. 100 ml of Milli-Q water was added to each vessel and water-soluble matrix elements were extracted overnight. The water was discarded and 2 ml of concentrated HNO₃ was added for dissolution of the remaining salts on a hot plate (100 °C for 30 min) situated in a laminar-flow cabinet. Subsequently, 8 ml Milli-Q water was added to the vessels and the solute was separated from carbon particles by filtration through 0.45 µm membrane filters (Schliecher & Schuell, Merck Eurolab, Darmstadt, Germany) into polypropylene autosampler tubes.

Fe separation procedures

Ion-exchange. In this procedure we followed the modified anion exchange separation techniques of Kastenmayer et al.¹⁵ At first, Fe was separated from digested blood without any additional sample handling using ion-exchange columns prepared from 5-ml pipette tips. The columns were filled with anion exchange resin, which was retained at the bottom of the column using plugs of quartz wool. The resin was washed with 80 ml of 3 M HNO₃ and regenerated to the chloride form with 20 ml of 7 M HCl. 5 ml of the digested blood solutions were diluted to 40 ml with 8 M HCl and the solutions were allowed to run freely through the column at a flow rate of approximately 1 ml min⁻¹. To elute concomitant elements (Ca, Mg, Na, etc.), the column was washed with 30 ml of 7 M HCl. Fe was eluted with 30 ml of 1 M HCl.

As a modification of this procedure, 5 ml of the digested blood was first placed into acid-washed Teflon beakers, dried on the hot plate (100 °C) and re-dissolved in 2 ml of 7 M HCl. The solution was passed through a mini-column prepared from 0.2-ml pipette tips at a flow rate of ca. 0.3 ml min⁻¹. The matrix elements were eluted by 2 ml of 7 M HCl and Fe was stripped by 5 ml of 1M HNO₃.

All the elution solutions were collected in 1 ml fractions.

Precipitation. To precipitate heme-Fe two reagents were tested, namely sodium hydroxide (3 M solution) and ammonia (25% v/v solution). The procedure described by Fujimori et al.¹⁴ was followed with minor modifications. To avoid possible contamination introduced during pH adjustment from the electrode or indicator paper, the minimum volume of these reagents necessary to efficiently precipitate Fe from digested blood was determined experimentally and was found to be 2.5 and 1 ml per 10 ml of digested blood for sodium hydroxide and ammonia, respectively. Evaporating the blood digest as described above, followed by re-dissolution in 0.14 M HNO₃ can significantly reduce these volumes. After standing overnight at room temperature, the precipitate was isolated by centrifugation at 4000 rev min⁻¹ for 8 min. The precipitate was rinsed twice with 10 ml of Milli-Q water and then dissolved in 10 ml of 0.14 M HNO₃.

For the reproducibility assessment, all the experiments were performed at least in duplicate.

Mass spectrometry. Quantitative determination of element concentrations was performed by ICP-SFMS using external calibration and In as an internal standard as described elsewhere.²⁰ The list of elements monitored includes Fe, major and minor concomitant elements (Na, S, K, P, Mg, Ca, Zn, Rb, and Cu) as well as Cr and Ni (elements interfering with Fe isotopes). Such measurements were performed for the following solutions (shown as stars with corresponding numbers in Fig. 1):

MW digest of the whole blood (1);

Solutions passed through ion-exchange columns after sample loading and elution of concomitant elements–Fe stripping by 1 M HCl (2), Fe stripping by 1 M HNO₃ (3);

Solutions after re-dissolution of the precipitate for all precipitation experiments (4ab, 5, 6, 8);

Solutions after extraction of water-soluble elements from the ashed blood (7).

This design of experiments allows the efficiency of Fe purification and recovery, as well as pathways of possible Fe losses to be assessed.

Iron isotope ratio measurements were performed by MC-ICP-MS using Ni for on-line mass discrimination correction. Such measurements were performed on IRMM-014 Fe isotopic standard, on MW digested blood solution and on Fe separates, when the separation was assessed to be efficient. All these solutions were diluted to 5.0 ± 0.5 mg Fe l⁻¹ with 0.14 M HNO₃, and spiked with Ni at 5 mg l⁻¹. These concentrations yield an intensity >0.3 V for all isotopes monitored, in spite of the use of the MC-ICP-MS instrument in high-resolution mode. The total measuring time per sample was approximately 35 min.

Results and discussion

Table 3 represents a summary of ICP-SFMS results for the digested blood and different Fe separate fractions. As final sample dilution may vary considerably between different procedures, all concentrations are recalculated to 1 ml of whole blood. For ion-exchange separations, only concentrations in the first 5-ml elution fraction (step 2 in Fig. 1, elution with HCl) or 1-ml fraction (step 3 in Fig. 1, elution of mini-column with HNO₃) are presented.

Table 3 Results of the Fe separation recalculated to 1 ml of Sero AS whole blood. Procedure numbers correspond to those in Fig. 1. Uncertainties are given in parentheses, expressed as one standard deviation of at least three independent sample preparations

Procedure No. Concomitant elements/µg	1	2	3	4a NaOH	4b NH3	5 NH3	6 NH3	7	8 NH3
Na	1200 (50)	3.8 (0.9)	1.3 (0.3)	220 (5)	1.9 (0.8)	2.0 (0.6)	0.7 (0.5)	133, 147	1.6, 2.0
S	1170 (60)	0.25 (0.08)	2.8 (1.3)	2.8 (0.3)	1.4 (0.4)	2.0 (0.3)	<0.5	119, 131	1.2, 1.6
K	1000 (60)	2.6 (0.2)	1.4 (0.8)	3.4 (0.5)	0.6 (0.1)	0.6 (0.3)	0.4 (0.3)	98, 122	0.5, 0.7
P	210 (11)	7.5 (0.1)	3.1 (2.0)	26 (9)	65 (2)	69 (4)	29 (7)	24, 26	12, 16
Mg	18 (1)	0.02 (0.01)	0.05 (0.02)	12 (4)	3.9 (0.2)	6.6 (0.4)	2.8 (1.3)	15, 16	3.5, 4.1
Ca	15 (1)	0.2 (0.1)	0.6 (0.3)	9.4 (0.2)	6.7 (0.5)	9.0 (0.2)	5.6 (2.3)	13, 14	6.1, 6.8
Zn	5.0 (0.3)	1.4 (0.4)	0.5 (0.3)	3.6 (0.1)	0.08 (0.01)	0.17 (0.03)	0.08 (0.05)	4.9, 5.7	0.18, 0.26
Rb	1.38 (0.03)	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.02, 0.03	<0.001
Cu	0.69 (0.05)	0.02 (0.01)	0.13 (0.05)	0.02 (0.01)	0.01 (0.01)	<0.01	<0.01	0.69, 0.81	0.01, 0.03
Total/µg	3620	16	10	277	80	89	39	434	29
Fe/µg	437 (19)	372 (12)	405 (18)	423 (23)	405 (14)	410 (11)	381 (17)	387, 433	368, 402
Fe recovery (%)	100	85	93	97	93	94	87	94	88
Cr/ng	0.60 (0.09)	5.3 (2.7)	2.4 (2.0)	87 (4)	0.3 (0.2)	2.3 (0.1)	1.5 (1.3)	13.2, 25.8	2.9, 3.9
Average error introduced in ^56Fe/^54Fe ratio (‰)	<0.001	0.005	0.002	0.084	<0.001	0.002	0.001	0.018	0.003
Ni/ng	4.6 (1.3)	47 (29)	5.1 (2.1)	85 (14)	4.5 (2.8)	2.9 (0.5)	3.7 (3.2)	7, 12	3.1, 4.7
Average error introduced in ^58Fe/^56Fe ratio (‰)	0.145	1.45	0.16	2.6	0.14	0.09	0.11	0.29	0.12

---

Separation of matrix elements

The efficiency of the removal of major matrix constituents was assessed as the total concentrations of these contaminant elements remaining in the Fe separates. In this sense, ion-exchange separations (procedures 2 and 3 in Fig. 1) provide the most effective means for Fe purification, with combined concentrations of major matrix elements below 0.5% of the initial level in the digested blood, irrespective of the size of the column or reagents used for Fe elution. Approximately 2–3% of the matrix elements remain in the precipitate after single precipitation with ammonia (procedures 4b and 5 in Fig. 1), with no obvious effect from the preceding evaporation step. The efficacy of the ammonia precipitation procedure can be improved further by either a secondary precipitation step (procedure 6 in Fig. 1) or by extraction of water-soluble elements from ashed blood when ashing is used for the decomposition of blood (procedure 8 in Fig. 1). Phosphorus appears to be the most difficult element to separate by this procedure. Precipitation with NaOH (procedure 4a) and extraction of water-soluble elements from ashed blood (procedure 7) were equally inefficient, leaving more than 8% of the matrix elements in the Fe separates.

Fe recovery

Recoveries of the procedures were calculated as the ratio of recovered Fe after different purification steps to native Fe in pooled blood (Table 3). For the ion-exchange procedure using large columns and HCl as eluent, ca. 85% of Fe was recovered in the first 5-ml fraction. The combined Fe concentration in the total 30 ml of eluate was below 95%. For the mini-column, the first 1-ml fraction of HNO₃ elutes 93% of native Fe with an additional 4–5% eluted in the subsequent 4-ml fractions. For both experiments, 0.2–0.3% of Fe was not retained by the resin and 0.3–0.6% was lost during the elution of concomitant elements. It should be noted that an increase in the flow rate above 0.5 ml min⁻¹ during loading of the sample onto the mini-column results in a substantial increase in Fe losses at this stage (up to 20% of native Fe). The unaccounted Fe fraction (if any) was most probably not eluted from the columns with the volumes of eluate used. The separation procedures based on single precipitation exhibit relatively similar recoveries in the 93–97% range, with 0.6–2.0% of Fe remaining in the supernatant. The remaining Fe fraction was probably lost as small particles of iron hydroxides during washing of the precipitate with Milli-Q water. A similar level of recovery was found after extraction of water-soluble elements from the ashed blood, with the remaining part of Fe that resides in the aqueous solution, either in dissolved form or as small organic colloids that pass through the 0.45 µm filter, though less than 90% of native Fe was recovered after secondary precipitation (procedures 6 and 8 in Fig. 1).

Cr and Ni contamination

As Cr and Ni present in Fe separates carry direct risks for inaccurate ratio measurements involving ⁵⁴Fe and ⁵⁸Fe, respectively, a thorough evaluation of possible sample contamination by these elements is of primary importance. Unfortunately, due to the configuration of the collectors in the MC-ICP-MS instrument applied for this study, it was impossible to monitor any minor Cr or Ni isotopes using the central ion counter for on-line mathematical correction of potential isobaric interferences. For Cr, all the separation procedures except precipitation with ammonia introduce significant contamination (Table 3). However, the error introduced to the ⁵⁶Fe/⁵⁴Fe ratio would be less than 0.01‰ for all procedures, other than NaOH precipitation and extraction of water-soluble elements from ashed blood. Hence, isobaric interference from ⁵⁴Cr can be neglected for ion-exchange separations and for precipitation with ammonia.

On the contrary, even at native concentrations in the digested blood, Ni will result in an introduced error on the ⁵⁸Fe/⁵⁶Fe ratio in excess of 0.1‰, and all the separation schemes tested during this study demonstrated no significant improvement in Ni removal from the Fe fractions. Particularly high Ni contamination was found for NaOH precipitation and for ion-exchange separation using large columns. Nickel (as well as Cr) is presented as an impurity in the NaOH for the former procedure, though the anion-exchange resin is the major Ni source for the latter. It should be added that the instrumental blank resulting from the use of cones made from Ni (commonly in the range 20–30 ng l⁻¹) introduces an error >0.05‰ in the ⁵⁸Fe/⁵⁶Fe ratio. As a result, the accuracy of Fe isotope ratios involving ⁵⁸Fe can hardly be better than 0.2‰ unless provision is made to simultaneously monitor additional Ni isotopes of lower abundances and apply a mathematical correction. This would necessitate accurate evaluation of the level of instrumental mass discrimination operative on the Ni isotopes, which was not possible at the Ni concentrations actually present. Instead, the potential to utilize admixed Ni for mass discrimination correction was investigated, and attempts to measure ⁵⁸Fe were abandoned.

Mass discrimination correction

Other than the isobaric interference on ⁵⁸Fe, nickel possesses quite favorable characteristics as a normalizing element. The isotopes ⁶⁰Ni and ⁶²Ni are fairly abundant and free from isobaric interferences. Additionally, the isotopic composition of natural Ni is rather invariant and has therefore been precisely established.²¹ Consequently, processed blood samples were spiked with Ni to a concentration of 5 mg l⁻¹ prior to analysis by MC-ICP-MS. The iron isotopic reference material IRMM-014²² was similarly spiked and used to establish the relationships between measured and true Fe and Ni isotope ratios on the basis of an exponential mass discrimination model, as described by Woodhead.²³ Variations in Fe and Ni isotope ratios were observed to be highly correlated (P ≪ 0.001; data not shown), and applying Ni normalization was found to significantly reduce uncertainties both within and between sample runs, the latter being illustrated in Fig. 2. The average mass discrimination from four measurement sessions, established using IRMM-014, was 3.84 ± 0.03% per mass unit. It should be noted that slight differences in the magnitudes of the correction factors were observed for the Fe solutions after NH₃ precipitation procedures and anion-exchange separation.

Fig. 2 Measured isotope ratios for ⁵⁶Fe/⁵⁴Fe (triangles) and ⁵⁷Fe/⁵⁴Fe (circles) normalized to data for IRMM-014. Data points represented by filled symbols have been corrected using an exponential mass discrimination model²³ with ⁶²Ni/⁶⁰Ni for normalization, those with open symbols have not. Standard uncertainties are smaller than the data points.

Estimate of Fe isotope fractionation during separation procedures

In accordance with the previous MC-ICP-MS study on Fe isotope variations in blood,⁷ we have elected to express the degree of fractionation using δ-notation, defined as follows:²⁴

where the Ni normalized, measured isotope ratios for IRMM-014 are taken as standard values. The results for the Sero AS whole blood sample, following various preparation procedures, are compiled in Table 4. All data are from a single days measurement; day-to-day variations in calculated δ-values were generally better than ±0.08‰. For a total of nine test samples of Sero AS whole blood, independently prepared using methods 3 (n = 3), 4a (n = 2), 5 (n = 1) and 6 (n = 3), average fractionations are δ⁵⁶Fe = −2.81 ± 0.05‰ and δ⁵⁷Fe = −4.18 ± 0.06‰ (uncertainties expressed as one standard deviation). Walczyk and von Blanckenburg⁷ noted significant differences in average Fe isotope fractionation between blood collected from male (δ⁵⁶Fe = −2.75 ± 0.17‰) and female (δ⁵⁶Fe = −2.43 ± 0.20‰) subject groups. Ohno et al.⁹ determined Fe isotope ratios in human red blood cells, observing fractionation of ca. −1.7 ‰ per mass unit, which would correspond to δ⁵⁶Fe ≈ 3.4 ‰. Despite differences in the origin and exact nature of the samples, the present results are clearly consistent with the trends emerging from these other studies.

Table 4 Determined fractionation of Fe in Sero AS whole blood after various sample preparation methods. Unless otherwise stated, uncertainty terms in parentheses are one standard deviation between n independently prepared aliquots. The procedures used are summarized in Fig. 1

Procedure	δ^56Fe, ^0/00	δ^57Fe, ^0/00	n
a Only the first ml fraction of eluent, containing ca. 93% of total Fe, from the anion exchange column was analysed. b First three ml fractions of eluent, containing about 97% of total Fe, from the anion exchange column were combined and analysed. c Duplicate measurements of a single, non-separated, microwave-digested sample. d Uncertainties are one standard deviation of between block means for a single sample (internal precision).
1	−2.81, −2.97	−4.28, −4.45	2^c
3^a	−2.81(0.02)	−4.17(0.03)	3
3^b	−2.89(0.05)	−4.27(0.08)	1^d
4a, 5	−2.81(0.02)	−4.17(0.03)	3
6	−2.82(0.05)	−4.21(0.05)	3
8	−2.73(0.24)	−4.20(0.33)	1^d

---

As is apparent in Table 4 and Fig. 2, all the sample preparation methods tested yielded statistically indistinguishable, Ni-normalized Fe isotope ratios. Johnson et al.¹⁹ noted that rapid precipitation of Fe(III) species, as would be present in the acid digested blood, induces no isotopic fractionation. Such conditions are achieved in our precipitation procedures. The results obtained with anion exchange columns demonstrate that the recovery of 93% of total Fe is also sufficient to avoid detectable fractionation. Even though no isotope fractionation was found between the aforementioned different preparation methods, the possibility exists that a change in the isotope ratio could arise from the presence of unmineralized organic compounds in the blood samples after microwave digestion. Organically-bound Fe might behave differently during the ion chromatography and precipitation than free ions, but this hypothesis can be rejected because the isotopic results from procedure 8 (ashing followed by precipitation) are not significantly different from those furnished by the rest of the separation methods, as seen in Table 4.

Practical considerations

Though ashing requires less expensive equipment compared to MW digestion systems, this method actually provides no improvements in terms of sample throughput or in labour consumption. Moreover, it carries higher risks of contamination and therefore can not compete with MW digestion. As two separation procedures (precipitation with ammonia and ion-exchange using the mini-column following MW digestion of whole blood) show similar analytical performance regarding Fe purification, recoveries and negligible levels of introduced isotopic fractionation, the actual choice would depend on additional (economic) parameters. These include sample throughput, labour consumption, reagent cost, etc, which can be of significant importance for large-scale studies, e.g., on iron absorption from a specific dietary regime. Precipitation with ammonia then would certainly be the method of choice:

—evaporation of digested blood is unnecessary;

—perfect match for high sample throughput provided by MW digestion, practically unlimited number of samples can be prepared concurrently;

—no supervision at precipitation stage necessary;

—low reagent consumption.

Conclusion

Due to the high resolution capabilities of the Neptune MC-ICP-MS instrument, Fe isotope ratios can be measured with high accuracy without complete Fe separation from the blood matrix. Provided that the matrix-induced changes in instrumental mass discrimination are adequately corrected for, even direct isotopic analysis of digested blood after simple dilution is possible, though this approach can hardly be recommended for routine practice. In spite of the fact that all procedures tested in this study result in Fe recoveries below 100%, none of these seems to cause detectable artificial isotope fractionation. The results obtained in this study emphasize that even minor changes of matrix composition may affect the accuracy of isotope ratio determinations due to the mass discrimination effect. Therefore, on-line mass discrimination correction, e.g., using Ni, is mandatory. Iron separation using precipitation with ammonia was found to be the best purification method for highly precise determinations of Fe isotopic composition in whole blood samples. Advantages of the procedure are minimal blank contributions, non-detectable artificial isotope fractionation and high throughput.

Acknowledgements

We would like to thank H. Ramebäck, Geosigma, Kungälv, Sweden and M. Berglund, Institute of Reference Materials and Measurements, Geel, Belgium for providing IRMM-014. The interest in our research from M. D. Axelsson, Division of Applied Geology, Luleå University of Technology led to fruitful discussions. This study was supported by a grant from EU structural fund for Objective 1 Norra Norrland. Analytica AB should be acknowledged for technical and financial support and we are also grateful to Kempestiftelserna for a grant to purchase the Neptune.

References

1. N. N. Greenwood and A. Earnshaw, Chemistry of the Elements, Pergamon, Oxford, 1993, p. 1275.
2. M. B. Reddy, M. V. Chidambaram and G. W. Bates, in Iron Transport in Microbes, Plants, and Animals, ed. G. Winkelmann, D. van der Helm and J. B. Neilands, VCH, 1987, pp. 429–444.
3. K. J. H. Wienk, J. J. M. Marx and A. C. Beynen, Eur. J. Nutr., 1999, 38, 51.
4. T. Walczyk, L. Davidsson, N. Zavaleta and R. F. Hurrell, Fresenius' J. Anal. Chem., 1997, 359, 445.
5. T. Walczyk, Fresenius' J. Anal. Chem., 2001, 370, 444.
6. I. J. Griffin, J. Anal. At. Spectrom., 2002, 17, 1186.
7. T. Walczyk and F. von Blanckenburg, Science, 295, 2065.
8. X. K. Zhu, Y. Guo, R. J. P Williams, R. K. O’Nions, A. Matthews, N. S. Belshaw, G. W. Canters, E. C. de Waal, U. Weser, B. K. Burgess and B. Salvato, Earth Planet. Sci. Lett., 2002, 200, 47.
9. T. Ohno, I. Kouge and T. Hirata, Geochim. Cosmochim. Acta, 2002, 66, A569 Suppl. 1.
10. J. S. Becker, J. Anal. At. Spectrom., 2002, 17, 1172.
11. D. J. Douglas and S. D. Tanner, in Inductively Coupled Plasma Mass Spectrometry, ed. A. Montaser, Wiley-VCH, New York, 1998, pp. 615–679.
12. C. Sariego Muñiz, J. M. Marchante Gayón, J. I. García Alonso and A. Sanz-Medel, J. Anal. At. Spectrom, 2001, 16, 587.
13. F. Vanhaecke, L. Balcaen, G. De Wannemacker and L. Moens, J. Anal. At. Spectrom., 2002, 17, 933.
14. E. Fujimori, T. Hayashi, K. Inagaki and H. Haraguchi, Fresenius' J. Anal. Chem., 1999, 363, 277.
15. P. Kastenmayer, L. Davidsson, P. Galan, F. Cherouvrier, S. Hercberg and R. F. Hurrell, Br. J. Nutr., 1994, 71, 411.
16. K. Bukhave, A. D. Sørensen and M. Hansen, J. Trace Elem. Med. Biol., 2001, 15, 56.
17. A. D. Anbar, J. E. Roe, J. Barling and K. H. Nealson, Science, 2000, 288, 126.
18. P. E. Warwick, S. Severmann and M. R. Palmer, Geochim. Cosmochim. Acta, 2002, 66, A824 Suppl 1.
19. C. M. Johnson, J. L. Skulan, B. L. Beard, H. Sun, K. H. Nealson and P. S. Braterman, Earth Planet. Sci. Lett., 2002, 195, 141.
20. I. Rodushkin, F. Ödman, R. Olofsson and M. D. Axelsson, J. Anal. At. Spectrom., 2000, 15, 937–944.
21. K. J. R. Rosman and P. D. P. Taylor, Pure Appl. Chem., 1998, 70, 217.
22. P. D. P. Taylor, R. Maeck and P. De Bièvre, Int. J. Mass Spectrom. Ion Processes, 1992, 121, 111.
23. J. Woodhead, J. Anal. At. Spectrom., 2002, 17, 1381.
24. B. L. Beard and C. M. Johnson, Geochim. Cosmochim. Acta, 1999, 63, 1653.

---