Direct separation of Zn from dilute aqueous solutions for isotope composition determination using multi-collector ICP-MS

Direct separation of Zn from dilute aqueous solutions for isotope composition determination using multi-collector ICP-MS

Chemical Geology 259 (2009) 120–130 Contents lists available at ScienceDirect Chemical Geology j o u r n a l h o m e p a g e : w w w. e l s ev i e r...

775KB Sizes 0 Downloads 4 Views

Chemical Geology 259 (2009) 120–130

Contents lists available at ScienceDirect

Chemical Geology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c h e m g e o

Direct separation of Zn from dilute aqueous solutions for isotope composition determination using multi-collector ICP-MS Jiu-Bin Chen ⁎, Pascale Louvat, Jérôme Gaillardet, Jean-Louis Birck Equipe Géochimie et Cosmochimie, Institut de Physique du Globe de Paris (IPGP), Université Paris-Diderot, UMR CNRS 7154, 4 place Jussieu, 75252 Paris, France

a r t i c l e

i n f o

Article history: Received 18 June 2008 Received in revised form 26 September 2008 Accepted 24 October 2008 Editor: B. Bourdon Keywords: Zn isotope composition Ion-exchange chromatography Dilute natural water Mass fractionation Mass spectrometry Seine River

a b s t r a c t We report here a two-step Zn chemical separation protocol, which allows dilute aqueous samples with low Zn concentrations to be directly loaded onto the chromatography columns, without any preliminary evaporation. This method is particularly suited for river and rain water samples with Zn concentrations between 0.1 and 120 μg/l. We present the different steps of the protocol on Chelex-100 and AG 1-X4 resins (Bio-Rad) with varying eluants under different matrix conditions. The method results in acceptable procedural blanks and quantitative yields of Zn. The mass bias during isotope measurements by the Neptune Thermo® MC-ICP-MS was corrected using empirical external normalization (EEN) with Cu as the internal dopant. We additionally evaluate the sensitivity of the measurement method to instrumental and analytical conditions as tuning, solution pH, Zn concentration, Zn/Cu ratio, temperature, and trace matrix addition. Both the stable introduction system (SIS, double Scott cyclonic spray chamber) and the Apex HF (ESI) were tested for Zn isotope measurement and the Apex was preferred for its higher sensitivity and better stability. External δ66Zn reproducibility for measurements of the standard solution (calibrated over two years) and natural water samples was 0.04‰ (2σ). © 2008 Elsevier B.V. All rights reserved.

1. Introduction Isotope studies have been crucial in the advancement of many fields in Earth and Environmental Sciences. The recent improvements in multiplecollector magnetic-sector inductively coupled plasma mass spectrometry (MC-ICP-MS) instrumentation and the introduction of clean chromatography procedures allow more accurate and precise measurements of the small natural isotope variations of many non-traditional elements (Marechal et al., 1999; Zhu et al., 2000; Wombacher et al., 2003; Poitrasson and Freydier, 2005; Ingri et al., 2005). Among them Zn (with five stable isotopes) has attracted a great deal of attention, because of its biological role and potential toxicity at elevated concentrations in different environmental reservoirs (soils, atmosphere, water, etc.). Preliminary Zn isotopic studies have demonstrated the potential for applications in geochemistry and biogeochemistry (Marechal et al., 2000; Pichat et al., 2003; Luck et al., 2005; Moynier et al., 2006; Viers et al., 2007; Weiss et al., 2007; Borrok et al., 2008; Chen et al., 2008; Cloquet et al., 2008; John et al., 2008; Kavner et al., 2008; Petit et al., 2008; Sonke et al., 2008; Toutain et al., 2008). A total isotopic variation of 3‰ (in δ66Zn units, see Section 4.1) has been determined in natural biological and terrestrial geological samples, with 2σ analytical precisions as good as 0.04‰ (Marechal et al., 1999; Stenberg et al., 2004; Cloquet, 2005; John et al.,

⁎ Corresponding author. Tel.: +33 1 44 27 48 16; fax: +33 1 44 27 49 16. E-mail address: [email protected] (J.-B. Chen). 0009-2541/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chemgeo.2008.10.040

2007; Cloquet et al., 2008). Biological incorporation, abiotic metal adsorption, chemical diffusion and column elution can introduce important fractionations of Zn isotopic ratios (Marechal and Albarede, 2002; Stenberg et al., 2004; Pokrovsky et al., 2005; Weiss et al., 2005; Viers et al., 2007; Balistrieri et al., 2008). However, the majority of previous work has focused either on developing reliable MC-ICP-MS measurements or on determining natural variations of Zn isotopes occurring in solid samples. Little has been reported for the Zn isotope geochemistry in aqueous environment largely because of the inherent analytical challenges. Bermin et al. (2006) analyzed Zn isotopes in seawater using a coprecipitation and ion-exchange column chemistry method. Malinovsky et al. (2005) tested the application of diffusive gradients in thin films (DGT) for isotopic measurement of soluble Zn during fractionation experiments. More recently, Borrok et al. (2007) improved the Marechal et al. (1999) anion-exchange chromatography separation protocol to isolate Zn from concentrated mine drainage waters for isotopic analysis. Isotopic composition of Zn and the mechanisms controlling its variability in very dilute aqueous solutions (i.e. river waters), however, remain poorly documented. Understanding the isotopic distribution of Zn in freshwater systems is critical for evaluating the behavior, transport, geochemical and biogeochemical cycles of Zn in nature. The main difficulty in measuring Zn isotopes in freshwater samples is that it is often present at very low concentrations (b 10 μg/l). This exasperates problems with sample contamination and incomplete chemical separation. Large volumes of water are thus needed (up to 1 l) to get enough Zn for an isotope analysis (at least 4 replicates). Because

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

of the sensitivity of instrumental mass bias to even small matrix effects, obtaining well-purified zinc solutions is the obvious prerequisite for accurate Zn isotope determination. Traditional separation protocols require either (or both) the evaporation of large amounts of natural waters or the addition of large volumes of highly-purified acids to the samples, before proceeding to the column chemistry. Evaporating big volumes of diluted waters introduces difficulty in dissolving the solid pellet in preparation for column chromatography (so does not fit easily traditional microchemistry lab-ware) and the addition of a large quantity of concentrated acid often increases the concentrations of Zn in procedural blanks. The ion-exchange purification on AG-MP1 resin proposed by Borrok et al. (2007) after Marechal et al. (1999) is thus difficult to apply to waters with very low Zn-concentration. Although the MC-ICP-MS has advantages over conventional TIMS (Albarede et al., 2004), one disadvantage is that the instrumental mass bias is large, implying that the measured raw isotopic data must be corrected. Among different analytical mass-bias correction strategies, the “empirical external normalization” (EEN) approach proposed by Marechal et al. (1999) has been proven successful by many previous studies for Zn isotope measurement (Archer and Vance, 2004; Vance et al., 2006). This method is based on the “external” normalization of Zn relative to a Cu standard (Cu-SRM976, NIST) added into each solution. The definition of the mass fractionation relationship between Zn and Cu isotopes is a crucial and delicate step. Mass bias effects on Zn and/or Cu related to both instrumental and external operating conditions may broaden the Zn–Cu standard data cloud or modify the fractionation regression lines. Such effects have been observed on several MC-ICP-MS spectrometers as Isoprobe, Nu and VG Axiom (Mason et al., 2004b; Petit et al., 2008). Isotope mass discrimination occurs during sample introduction (including desolvation), ionisation in the plasma, and ion transport inside the mass spectrometer (Stewart and Olesik, 1999; Vanhaecke et al., 1993; Fraser and Beauchemin, 2000; Albarede et al., 2004; Andrén et al., 2004). These effects are instrument-dependent (Andrén et al., 2004; Mason et al., 2004b) because internal instrumental parameters and external conditions act differently on different machines (Stewart and Olesik, 1999; Vanhaecke et al., 1993; Fraser and Beauchemin, 2000). Many authors have tried to define the factors affecting the mass bias behaviors of the Zn–Cu pair. Some demonstrated that the space-charge effect that favors the heavier ions transmission can lead to different Zn and Cu mass fractionation behaviors (Marechal et al., 1999; Malinovsky et al., 2003). Others suggested that Zn–Cu fractionation generally takes place in the introduction system and between the sampler and skimmer cones, or immediately after the skimmer (Andrén et al., 2004; Mason et al., 2004b). Furthermore, both matrix addition and imperfect chemical separation would lead to additional non-spectral mass discrimination effects (Chang et al., 2003; Archer and Vance, 2004; Wilkinson et al., 2005). We report here a new two-step purification method for the direct separation of Zn from dilute aqueous samples. The validation of the purification protocol is confirmed by multiple tests on varying eluants and concentrations, and by assessing the extraction yield, effect of organic matter, and reproducibility of the measured isotopic ratios. Using the modified empirical Cu-external normalization approach (EEN), Zn stable isotope compositions were accurately determined using our optimal measurement strategy in 50 natural and anthropogenic water samples of the whole Seine basin with Zn concentrations varying from 0.1 to 1700 μg/l (Chen et al., 2008). 2. Reagents and materials Chelex-100 and AG1-X4 resins (both 200–400 mesh, Bio-rad®) are used in this study for Zn purification. Chelex-100 is a commonly used cationic exchanger and has been designed to extract transition elements from seawater without pre-treatment (Pai, 1988; Liu and Ingle, 1989; Jiann and Presley, 2002; Bermin et al., 2006). Its binding capacity according to pH is comparable to that of EDTA. Both resins were carefully

121

washed and saturated in water to discard the finer floating grains before loading them onto the column. All acids (HCl, HNO3, and CH3COOH) used in this study were purified by sub-boiling distillation. The water was 18.2 MΩ Milli-Q water (Millipore). Ammonium acetate powder was dissolved in water to obtain a 1 M solution. This solution and a 1 M NH3 solution were both purified using Chelex-100 resin. HNO3 solutions were carefully adjusted to pH 2.56 (2.75 mM) and 1.90 (12.59 mM) for Zn elution. The H2O blank was determined directly, but the reagent blanks were determined by ICP-MS after evaporation of 10 ml initial solution. Blank concentrations were 0.001, 0.003, 0.05, 0.045, 0.075, 0.024 and 0.019 μg Zn/l for H2O, 1 M HNO3, 0.5 M HNO3, HNO3 solution at pH = 2.56, HNO3 solution at pH = 1.90, 2 M HCl and 1 M ammonium acetate solution, respectively. Zn procedural blanks are additionally very sensitive to the lab-environment and handling of initially very pure reagents as water and HNO3 may increase the blank. Cu-SRM976 metal and Zn JMC 3-0749L solution were used as reference materials. AAS (Alfa Aesar, Germany) Zn (1000 mg/l) and Cu (1000 mg/l) solutions are used as in-house isotopic standard solutions and were calibrated against the Zn-JMC and Cu-SRM976 international standards. All vials (mainly Teflon-made) were conditioned with ultrapure HNO3 and were rinsed with Milli-Q water just before their use. Except for fractions of separation tests whose Zn contents were measured by Neptune ICP-MS, bulk elemental concentrations were measured on ICP-MS X-series II (Thermo®) at IPGP. 3. Zn separation protocol For a precise Zn isotope measurement (i.e. 2σ ~ 0.04‰) of a natural water sample (at least 4 replicates), about 1 μg of Zn is necessary. This means that about 1 l (average total dissolved solids of 400 mg/l) of sample should be processed considering a typical Zn concentration of 1 μg/l in river waters (Gaillardet et al., 2005). A thorough separation of Zn from its aqueous matrix and in particular from elements such as Ca, Mg, Al, V, Ti, Ni, Cr, Ba, La and Ce is necessary, because these elements can generate spectral interferences with Zn (or Cu) isotopes in argide form (Mg, Al), (hydro)-oxide form (Ca, Cr) or double-charged species (Ba, La) (Mason et al., 2004a). As Cu standard was added to the samples for external mass bias correction, Cu must also be quantitatively removed from the sample Zn elution fractions. Zn is separated from the water matrix by two ion-exchange chromatography steps: a pre-concentration step using Chelex-100 resin and a purification step on AG 1-X4 resin. The column used for the pre-concentration is a polypropylene mini-column (volume 6 ml, reservoir diameter 1.7 cm, height 9.5 cm, Eichrom) superposed by a Poly-prep reservoir (volume 25 ml, reservoir diameter 2.5 cm), loaded with 0.5 ml of Chelex-100 resin. The column is first cleaned with 30 ml 2 M HNO3 and 30 ml water and then conditioned by 10 ml 1 M ammonium acetate (pH = 5.5). The sample solution is buffered with 1 M ammonium acetate at a final concentration of 0.05 M and then adjusted to pH = 5.5 using 1 M NH3 and 15.8 M acetic acid. This buffered sample solution (up to 1 l) is introduced directly into the pre-concentration column. Introduction flow is controlled by gravity and is on average 1.6 ml/min. The column is then cleaned with 20 ml 1 M ammonium acetate (pH = 5.5), 10 ml HNO3 at pH = 2.56 (2.75 mM) and 4 ml HNO3 at pH = 1.90 (12.59 mM). Zn is finally eluted in 10 ml HNO3 at pH = 1.90. The elution rate is about 1 ml/min. A faster elution would scatter the elution peaks and larger volume would be needed for quantitative Zn recovery. The final Zn elution fraction still contains substantial amounts of Ni, Al, La and Ce (measured by ICP-MS) and thus requires a second purification. This fraction is evaporated at 60 °C in a closed Teflon system (Evapoclean, Analab®) and is then re-dissolved in 1 ml 2 N HCl. The second separation step is made on a Bio-Rad column (volume 4 ml, reservoir diameter 2.5 cm, height 6 cm) loaded with 0.1 ml AG 1X4 resin (200–400 mesh, Bio-Rad). The column is first rinsed with 4 ml

122

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

All Zn isotopic results in this study are expressed as δxZn (in units of ‰):

2 M HNO3 and 4 ml water, then conditioned with 0.2 ml 2 M HCl. The previous Zn fraction in 2 M HCl is loaded onto the column. After rinsing with 1 ml 2 M HCl, 0.05 ml water is added to eliminate remaining HCl from the resin. Zinc is then eluted by 0.5 ml 0.5 M HNO3 and is ready for isotopic measurement. The elution scheme for this two-step Zn separation protocol is summarized in Table 1. Finally, the concentration factor of Zn from natural waters on the Chelex-100 resin is between 50 and 100, and Zn content in the last fraction of the two-step chemical purification can be 1000 times higher than that of the initial natural solutions. Yield for the entire two-step column is 100 ± 1% (measured by Neptune ICP-MS) and numerous separations of some identical samples (replicates) have validated the process, see Section 5.4.

where xZn is 66Zn,67Zn or 68Zn, JMC represents the JMC 3-0749L international Zn isotopic standard (from Lyon, France). δ68/66Zn is also determined to check the measurement validity (mass dependant fractionation law). Error bars given in this paper are all 2σ external standard deviations, whereas, the internal precision refers to 2 standard errors (2σm) from the 100 cycles of one measurement (Deines et al., 2003).

4. Mass spectrometry

4.2. Mass bias correction

4.1. MC-ICP-MS measurements

In order to accurately measure Zn isotopic ratios in natural samples, raw data must be corrected for instrumental mass bias (Marechal et al., 1999; Mason et al., 2004b). The direct and “modified” sample-standard bracketing (SSB) (Mason et al., 2004b) methods were not used in this study, because of their prerequisite of a perfect instrumental stability and of the large volume of standard solution needed (1 standard for each sample). Instead, mass discrimination was corrected by external normalization through the addition of an internal Cu standard (SRM976) to both standard and sample Zn solutions. Because Cu and Zn do not fractionate identically during MC-ICP-MS measurements, the mass discrimination factor for Cu cannot be directly applied to correct for Zn mass discrimination of the measured sample. The mathematical form that best relates instrumental mass bias in MC-ICP-MS can be expressed using the “generalised power law” (Marechal et al., 1999; Albarede et al., 2004; Mason et al., 2004b):

Zn (Cu) isotopic analyses were performed on a Neptune (Thermo Finnigan®, Germany) MC-ICP-MS at Institut de Physique du Globe de Paris (IPGP). Two sample introduction systems were used: a) the stable introduction system (SIS) made of a double-pass Scott spray chamber and a cyclonic spray chamber with a self-aspiring microconcentric PFA nebulizer, and b) the desolvation inlet system Apex HF (with a Teflon spray chamber, ESI) with the same nebulizer. Except during mass bias tests, Zn isotopic ratio measurements for all water samples were performed using the Apex HF introduction system since it showed obvious advantages compared to the cyclonic chamber SIS (see Sections 5.2 and 5.3). The Zn from standards and sample fractions obtained from the two-column separation were measured in 0.05 M HNO3 at a Zn concentration of 200 μg/l and a Zn/Cu elemental ratio of 2.0 (to give similar ion beam intensities for 63Cu and 64Zn). After positioning the Faraday cups to measure 64Zn, 66Zn 67Zn, 68Zn, 70 Zn 63Cu and 65Cu isotopes, we optimized Ar fluxes, torch position and optical parameters for maximum intensity, stability and optimum peak shape. The low-resolution entrance slit was used throughout in this study. During an analysis session, the Zn–Cu standard solution was measured every three to four sample solutions. Cup gain calibration was done at the beginning of each measurement session. One measurement consisted of 100 integrations of 4.2 s in 5 blocks of 20 cycles. The on-peak baseline was performed before each block and the peak centering before each measurement. Ultra-pure 0.5 M HNO3 is aspired for rinsing after each measurement for about 2 min. Hence, one complete analysis typically takes 13 min. Cycle values outside the 3σ range around the average ratio were automatically rejected. Each sample measurement was repeated 3 times during one measurement session and at least 3 other times in other sessions, giving more than six analysis replicates for most samples. Table 1 Elution sequence of the two-column separation of Zn from dilute water Eluant

ml

0.5 ml Chelex-100 Biorad resin (200–400 mesh) 2 M HNO3 H2O 1 M ammonium acetate at pH = 5.5 Sample (pH = 5.5) buffered with 0.05 M ammonium acetate 1 M ammonium acetate at pH = 5.5 HNO3 at pH = 2.56 (2.75 mM) HNO3 at pH = 1.90 (12.59 mM) HNO3 at pH = 1.90 (12.59 mM)

Eluted

30 Cleaning 30 Rinsing 10 Buffering up to 1 l Loading 20 Matrix 10 Matrix 4 Matrix 10 Zn (Ni)

0.1 ml AG 1-X4 Biorad resin (200–400 mesh) 2 M HNO3 H2O 2 M HCl 2 M HCl sample solution 2 M HCl H2O 0.5 M HNO3

4 4 0.2 1 1 0.05 0.5

Cleaning Rinsing Conditioning Loading Matrix HCl eliminating Zn

δx Zn =



x

Zn=64 Zn

 sample

=



x

Zn=64 Zn

 r = R M2n −M1n g

 JMC

 −1 ×1000

ð1Þ

ð2Þ

where r represents the measured isotope ratio, R the true isotope ratio, g the mass fractionation coefficient, and M1 and M2 the atomic masses of the two isotopes in ratio r. The arbitrary number n describes different mathematical laws. From repeated isotope measurements of standard and sample solutions during an analytical session, two linear relationships can be empirically plotted in a ln(rZn) versus ln(rCu) diagram, following the equation lnðrZn Þ = a× lnðrCu Þ + b

ð3Þ

where the slope a can be related to the previous generalised power law: a=

  n n ln gZn × M2Zn −M1Zn   n −M n ln gCu × M2Cu 1Cu

ð4Þ

the intercept b can be expressed as: b = lnðRZn Þ−a× lnðRCu Þ:

ð5Þ

Regression lines for sample and standard measurements have slopes so close that they can be considered as identical. Therefore, the vertical difference between the two lines (bsample − bstandard) is given by:     ðRZn Þsample ðRCu Þsample + a×ln : Δ = lnðrZn Þsample −lnðrZn Þstandard = ln ðRZn Þstandard ðRCu Þstandard ð6Þ Because the same Cu internal standard solution was added to both the standard and sample solutions, the second term of the Eq. (6) can be eliminated. Using our Cu-spiked sample measurements, we calculate the vertical offsets (Δ) between sample data points and the regression line defined by the repeated Zn standard solution

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

measurements from the ln(rZn) versus ln(rCu) plot. The final δ-notation data (in ‰) of the sample is directly derived from this offset (Δ):   δx=y Zn = eðΔÞ −1 ×1000:

ð7Þ

In the previous mathematical derivation, the exponential law would lead to the same final expression of δx/yZn (Eq. (6)) as for the generalised power low; the same empirical linear relationship between ln(rZn) and ln(rCu) would have been used and RCu would also have disappeared in the derivation. As a consequence, the CuSRM976 standard solution is not absolutely mandatory in this EEN mass bias correction method and any Cu standard would achieve the same result as long as the Cu solution is added to both the Zn standard and sample solutions. The previously described EEN correction method based on one single fractionation line works well for pure synthetic solutions such as a Zn atomic adsorption standard (AAS). However, for purified natural water samples, we observed sporadic “jumps” of the Zn–Cu standard fractionation line every few samples even during one measurement session, and not one but several regression lines of repeated standard measurements (every 3 to 4 samples) must in fact be considered for the mass bias correction. This phenomenon can be seen in Fig. 1: During the measurement session of Dec. 23rd 2006 with the Apex HF, uncorrected isotope data of Zn-JMC and Cu-NIST solution plotted in ln–ln space yielded four distinct straight lines with slightly different slopes (1.2195, 1.2279, 1.2753 and 1.2915) and intercepts (0.4006, 0.4071, 0.4429 and 0.4553). Each line was defined by successive measurements of the standard solution under identical operating conditions. Four data point sets representing the raw results of the same unknown sample (roof runoff RF5) measured during the same session are also plotted on Fig. 1. The vertical offset for the unknowns was calculated for each data point set using the corresponding regression line defined by immediately bracketing standard measurements. The average Zn isotope value of this sample was calculated by averaging the results from all four of the comparisons to the standard lines and the calculated 2σ standard deviation for these four measurements was 0.02‰. Based on the above described measurement characteristics of our instrument, the best measurement protocol was defined as performing at least 3 replicates of the same sample during one measurement session in addition to at least two other independent measurements during two other sessions. The classical EEN correction method has thus been slightly modified to include a “loose” standard-samplesstandard bracketing in that the Zn–Cu standard is measured every 3 to 4 samples in our study. The origin of these “jumps” of the empirical Zn–Cu fractionation line (leading to the definition of several lines during one measurement session) remains unknown, despite the advanced study of the influence upon mass bias of instrumental and chemical parameters (see following Sections 5.2 and 5.3). We have nevertheless tested that our EEN method was more precise than traditional standard-samplestandard bracketing, as shown by previous studies (Mason et al., 2004b; Petit et al., 2008). However, a Zn double-spike method may be a safest way if it gives similar precision (e.g., Bermin et al., 2006), but we did not evaluate this.

123

systems. Almost all measurements (100 cycles of 4.2 s) during this study display an internal standard error below 0.01‰ (2σm). Integrated results of both systems (Apex and SIS) give almost the same long-term 2σ external uncertainty (0.04‰). These 96 measurements lead to average values of 0.44574 ± 0.04‰ (2σ) and 0.56505 ± 0.04‰ (2σ) for the 65/63Cu and 66/64Zn AAS standards, respectively. 5.2. Influence of instrumental parameters on mass bias The accuracy of the isotope ratios corrected by the EEN method critically depends on the precision by which the empirical regression line is defined. Some instrumental mass bias drift (e.g. cluster of points, without relationship) may deteriorate the ln–ln linear relationship and thus preclude its definition (Andrén et al., 2004). Two experiments were carried out on the Neptune to address the influence of instrumental operating conditions on the mass discrimination (Fig. 3). Firstly, four isotopic measurements were performed with a standard solution containing 400 μg/l Zn and 200 μg/l Cu using the SIS. The first and the second measurements (5 blocks of 20 cycles) were performed respectively before and after tuning (higher sensibility, best stability and best peak form adjustment) in which nearly all operating parameters were adjusted. The third was measured after calibration of the Faraday cup amplification gains, and the fourth after a change (from 1 to 3 rpm) of the peristaltic pump speed (outlet of the spray chamber), other operating conditions being kept constant. The internal isotopic ratio variation during one measurement is similar to general internal variations determined for standard solutions during three years of study (b0.0001 for 65Cu/63Cu and b0.0002 for 66Zn/64Zn). However, the total “external” isotope variations both for 65Cu/63Cu (0.0011) and 66Zn/64Zn (0.0016) of the four measurements are much higher (nearly 10 times) than the internal variations, being 16% and 18% of total long-term variations of standard measurements during three years (0.0068 and 0.0088), respectively. Zn and Cu isotope ratios measured before tuning are very different from the 3 others, which are very similar (Fig. 3A). The effect of Faraday cup gain calibration and of peristaltic pump rotation speed is thus very limited on our instrument. For the second experiment, after operating parameters were adjusted for optimum intensity, stability and peak shape, we tested the effect of nebulization gas flow by bracketing Cu-SRM976 and Zn-JMC solution measurements at 1.009 l/min (optimized flux) and at 0.999, 0.989, 0.979, 0.969 and 0.959 l/min. The five measurements at 1.009 l/ min give a total 66Zn/64Zn standard deviation of 0.00018 (Fig. 3B). The small changes of gas-flow rate (up to 0.05 l/min) cause a drastic variation of measured Zn isotope ratios and a modification of the empirical

5. Results and discussion 5.1. In-house Zn and Cu standard calibrations Because of the exhaustion of the Zn-JMC and Cu-SRM976 standards, Zn and Cu AAS standard solutions (1000 mg/l, Alfa Aesar, Germany) calibrated against these international isotopic standards were measured regularly as in-house isotopic standards. Fig. 2 shows the results of 96 calibration measurements from 23 separate sessions during two years both with Apex HF and SIS (cyclonic spray chamber) as introduction

Fig. 1. Empirical external normalization (EEN) method for mass discrimination correction. During one measurement session (23 Dec. 2006) with the Apex HF introduction system, raw standard data define four parallel mass fractionation lines (numbering 1–4). δ66Zn values of the sample RF5 measured simultaneously 4 times were calculated from the vertical offset (Δ) between sample data points and the corresponding regression line. The four data sets give a 2σ precision of 0.02‰.

124

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

acidity for both standard and sample solutions is thus required for optimal Zn isotope measurement. Moreover, we used 0.05 M HNO3 throughout in this study as it gives optimum Zn and Cu intensities compared to more concentrated HNO3 solutions (0.1, 0.25 or 0.5 M).

Fig. 2. Ninety-six measurements of in-house Zn-AAS standard, calibrated against ZnJMC (from Lyon, France) and Cu-SRM976 international standards (23 measurement sessions over 2 years). Zn/Cu concentration ratios are respectively 400/200 μg/l and 200/100 μg/l for the SIS (Scott + cyclonic spray chamber) and Apex HF introduction systems, both giving intensities of about 10 V for 64Zn and 63Cu. Results for both the SIS and Apex HF introduction systems show almost the same external long-term standard deviations (2σ = 0.04‰).

fractionation line. This may be a combined effect attributable to 1) changes in aerosol particle size distribution, 2) the volume of analyte ions actually contained in the plasma, 3) the volume of introduced matrix, and 4) ion transport preference in the instrument (Andrén et al., 2004; Malinovsky, 2004). Despite these measurable variations, the Zn and Cu isotopic ratios of the 10 measurements correlate well (r2 = 0.9976) in a ln–ln plot (Fig. 3B). From these experiments, we conclude that the adjustment of the instrumental parameters can considerably extend the mass bias range. These mass bias drifts generally define one empirical mass fractionation line in Zn–Cu isotope plot. However, a detailed inspection of Fig. 3 shows that slight differences are found for slopes and intercepts of fractionation lines defined by total measurements from those of measurements under optimal conditions. These slope and intercept deviations for the second experiment introduced calculated Cu-normalized δ66Zn maximal variations up to 0.20‰. Though changes of instrument parameters that extend the mass bias range lead to better linearity definition of Zn–Cu isotope fractionation lines, we propose that Zn isotope measurement using EEN Cu-normalized method should be carried out at optimized instrument conditions (best tuning) to improve the external reproducibility.

5.3.2. Effect of concentration Mass bias drift along the regression line was observed for isotope measurements on AAS standard solutions with Zn concentrations varying from 140 to 600 μg/l (Zn/Cu = 2). As shown by Rehkämper and Mezger (2000), no “jump” was found for the regression line in the ln– ln plot under both introduction systems (Apex and SIS). Thus, Zn concentration in sample solutions does not need to match exactly that of the standard solution using the EEN correction method (Wombacher et al., 2003). However, according to an experiment performed with Zn-AAS concentration varying from 1 to 500 μg/l, at lower concentration, blank effect on measured Zn isotope ratios is noticeable (data not shown). This is why we chose to work mainly with Zn concentrations higher than 100 μg/l for the SIS and higher than 50 μg/l for the Apex, respectively. 5.3.3. Matrix effect Several studies have demonstrated the possibility of matrix addition for getting greater mass bias shift and therefore better definition of linear regression compared to the instrumental setting changes (AlAmmar et al., 1999; Woodhead, 2002; Archer and Vance, 2004; Peel et al., 2008). However, other unexpected matrix effects (i.e. regression line movements) may be sometimes introduced by imperfect Zn

5.3. Matrix and temperature effects on mass discrimination For a given instrument operated at optimal instrumental conditions, external factors can yield artificial mass bias drift (Malinovsky et al., 2003; Malinovsky, 2004). Previous studies have shown that solution composition and changes in the sample introduction system can introduce variations in mass bias (Mason et al., 2004b). The presence of a matrix-element spike (e.g. Sr, Pb, U) or even different Cu–Zn ratios in the measuring solution generate extreme variation in mass bias on Isoprobe and Nu instruments (Archer and Vance, 2004; Peel et al., 2008; Petit et al., 2008). Several external condition tests have also been performed on the MC-ICP-MS Neptune at IPGP during this study. 5.3.1. Effect of solution acidity The experiment was carried out on five different nitric acid solutions (0.01 M, 0.05 M, 0.1 M, 0.25 M and 0.5 M) containing the same AAS Zn and Cu concentrations (200 and 100 μg/l, respectively). Fig. 4 shows the raw data measured with the SIS (A) and Apex HF (B) introduction systems. For both systems, the effect of acid concentration on mass bias drift is obvious. However, whereas all data plot on a single regression line for Apex HF (Fig. 4B), they plot on different “jumped” parallel lines for SIS (Fig. 4A). A second set of measurements for these five solutions has confirmed these conclusions. The same

Fig. 3. Influence of instrumental parameters on mass bias (SIS spray chamber introduction). A: Tuning, Faraday cup gain calibration and peristaltic pump (spray chamber evacuation) speed change. Regression line defined from the whole set of data is different from that for the measurement with optimum instrumental parameters (M3). Before tuning, the mass bias is particularly different. B: Nebulizer gas flow rate change. A change of 0.05 l/min for sample gas flow can cause a mass bias spread by a factor of 10. However, all raw data define a different fractionation line from that of measurements at fixed gas flow (1.009 l/min). The induced Cu-normalized δ66Zn shifts are of 0.03 and 0.20‰ for the experiments A and B, respectively.

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

125

separation due to the chemical complexity of these samples. However, repeated ICP-MS measurements of trace element in our final Znfractions did not provide us with any insight into what the matrix effect might be, and no spectral interferences were detected during MC-ICPMS measurements.

Fig. 4. Influence of the solution acidity on the instrumental mass bias with the SIS cyclonic spray chamber (A) and the Apex HF (B) introduction systems. The effects of the standard solution acidity on mass bias are found for both systems. Whereas all data plot on a single regression line for the Apex HF, they plot on different parallel lines for the SIS spray chamber. Our isotope measurements of Zn in samples are carried out in 0.05 N HNO3 as this acidity gives the higher sensitivity compared to more acidic solutions (0.1, 0.25 or 0.5 N).

chemical separation. We have tested the effect of remaining matrix elements by adding a small amount (5%) of column-purified (to remove Zn and Cu) river water to an AAS Zn–Cu standard solution. This doped solution was measured by bracketing with the pure standard solution using the Apex introduction system (Fig. 5). This natural matrix addition (essentially major elements as Na, K, Ca, Mg and organic materials) produces a mass bias dispersion range 2 times larger than that for all measurements of the pure standard during the same session. Similar matrix addition effects have been observed previously and been suggested to be potentially used to induce a greater mass bias for precisely defining the necessary linear relationships in ln/ln space (Archer and Vance, 2004; Peel et al., 2008). However the fractionation line for the matrix-doped solution shows a very different slope (1.1116) compared to that of pure standard measurements (0.8853). Moreover, the measurement of a matrix-contaminated solution can “crosscontaminate” the next pure standard (or sample) measurement. This is illustrated in Fig. 5 by a progressively larger scatter of the data points after each matrix-doped standard measurement. This may be attributable to space-charge effects produced inside the plasma (before or after the sample cone) or to clogging of the cone or a similar memory effect in the inlet system. Even though the blank signal after the measurement was similar to that found initially, the acid-wash procedure does not always immediately eliminate the cross-contamination effect. We think that this matrix effect is partly responsible for the “jumps” of the regression lines found in this study (many regression lines have to be defined during one measurement session, Fig. 1). Our two-column purification protocol that has proven efficient to separate Zn from freshwater may miss some specific matrix (i.e. organic matter) during

5.3.4. Effect of variable Zn/Cu ratios The EEN method dependence on the analyte-spike (Zn/Cu) ratio was recently reported on Isoprobe and Nu instruments (Rehkämper and Mezger, 2000; Thirlwall, 2001; Archer and Vance, 2004; Petit et al., 2008). A similar experiment was also performed on our Neptune instrument. AAS standard solutions with different Zn/Cu ratios (from 0.5 to 8) were measured with both SIS and Apex HF introduction systems. All Zn and Cu concentrations were adjusted between 100 and 1000 μg/l. Fig. 6A shows the raw data obtained using the cyclonic chamber. As reported in the study on Isoprobe and Nu (Archer and Vance, 2004; Petit et al., 2008), all raw results of variable Zn/Cu solutions define different parallel mass fractionation lines. These departures from the Zn/Cu = 2 line induce a negative deviation of Cunormalized δ66Zn value up to −0.43‰ for high Zn solution (Zn/Cu = 8) and a positive variation of 0.07‰ for high Cu solution (Zn/Cu = 0.5). By contrast, data obtained for the same AAS standard solutions using the Apex HF display a single mass fractionation line (Fig. 6B), showing no systematic regression line “jump” in mass bias behaviors of Zn and Cu at different Zn/Cu ratios. Combined with the earlier studies using Isoprobe and Nu MC-ICPMS instruments (Archer and Vance, 2004; Petit et al., 2008), our experiments confirm the effect of analyte to internal standard ratio upon mass bias and imply that extraction and introduction systems act differently in regard of this effect. Generally, Zn isotopic measurements using external Cu standard addition should be done at a constant Zn/Cu ratio both for standard and sample solutions. Though more works should be done to probe the ultimate cause for this mass bias effect on the Neptune, this unexplained effect is readily resolved in our laboratory for Zn isotope measurement by measuring sample and standard solutions at Zn/Cu ratio of 2 regardless of the sample introduction system chosen. This ratio gives similar signal intensities for 63Cu and 64Zn. Though the stable SIS introduction system (cyclonic chamber) gives nearly the same external reproducibility as Apex HF (Fig. 2), our study demonstrates the potential advantages of the Apex HF in correcting the

Fig. 5. Influence of the matrix addition on the instrumental mass bias. A small quantity of riverwater purified by Chelex-100 resin (Zn and Cu are retained) is added to the AAS Zn–Cu standard. This solution is measured 3 times, bracketed with 3 measurements of pure standard solution. Measurements of matrix-added standard show a total data point scatter 2 times larger than that for the pure standard. Three measurements of the pure standard define a regression line with different slope from that of matrix-added standard. The continuous increase in mass bias range from the first measurement (1) to the third (3) of the pure standard probably results in a “cross-contamination” caused by the measurement of the matrix-added standard right before.

126

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

20 cycles) of AAS Zn/Cu standard were performed continuously (Fig. 7). We observed a direct correlation of 65Cu/63Cu (and 66Zn/64Zn, not shown in Fig. 7) with temperature. The 8 °C temperature change leads to a variation of 1.4‰ for 65Cu/63Cu. However, there is a time offset of approximately 30 min between the maximum temperature and the maximum 65Cu/63Cu ratio (Fig. 7A). Temperature also affects the mass fractionation lines in the ln–ln diagram (Fig. 7B), in that measurements during the temperature increase (M 8 and 9) define a regression line with a slope (1.0663) smaller than that for measurements performed at 21 °C (1.1706; M1–7, M12–14). However, data measured during the temperature decrease (M10 and 11) define a line with a steeper slope (1.3143, Fig. 7B). These changes of regression line slope can induce a deviation of −0.15‰ for the calculated δ66Zn. The temperature change influence upon extraction conditions and the related space charge effects are likely to favor the transmission of heavier ions, and this may be the principal cause for the observed Cu and Zn isotope variation with temperature (Vanhaecke et al., 1993; Andrén et al., 2004; Archer and Vance, 2004). Although a 8 °C change is enormous, our feeling is that a temperature variation of 1 °C or 2 °C could also have an influence. Isotopic measurements should therefore be performed under stable temperature conditions. 5.4. Efficiency and reproducibility of the Zn separation protocol Chelex-100 resin has been extensively used for pre-concentrating metals from aqueous solution (Liu and Ingle, 1989; Jiann and Presley,

Fig. 6. Influence of the Zn/Cu ratio on the instrumental mass bias with the SIS spray chamber (A) and the Apex HF (B) introduction systems. With the SIS system (A), increasing Zn/Cu ratio leads to drastic mass bias changes, data points for each Zn/Cu ratio plot on separate parallel fractionation lines, and a systematic movement is found for these lines with the increase of Zn/Cu ratio. Zn/Cu ratio changes from 0.5 to 8 can induce a total δ66Zn variation up to 0.50‰ with the SIS chamber. However, when using the Apex desolvation system (B), the mass bias changes are globally smaller and all data points for 6 solutions fall on the same mass fractionation line. This effect is readily resolved in this study using fixed Zn/Cu ratio of 2.

mass discrimination. As reported by Olesik and Hobbs (1994), the major source of signal fluctuation is the variation in the droplet-to-droplet vaporization point within the plasma. Desolvation in the Apex HF reduces the amount of the carrier solvents (H2O, HNO3) injected into the plasma, which probably diminishes much of the turbulence effects occurring in the plasma during ionisation, vaporization, atomisation and excitation (Stewart and Olesik, 1999). The Apex HF also favors ionisation, and weakens light ion diffusion derived from the charge repulsion. This set of effects leads to the higher sensitivity of the Apex HF (3 or 4 times higher than the SIS) and great improvement in the signal stability. Moreover, the Apex HF nebulizer has a slightly lower uptake rate than the SIS (30 μl/min against 50 μl/min). The advantages of Apex HF can also be confirmed by our acidity experiment and analyte/spike ratio experiment in that the isotope ratio measurements are globally less sensitive to these parameters with the Apex HF. 5.3.5. Effect of temperature The temperature dependency of instrumental mass bias is another prime concern (Vanhaecke et al., 1992; Vanhaecke et al., 1993). We were able to test this effect during a problem with our air conditioning installation. The temperature experiment was made with the SIS introduction system, but the same effect is observed with the Apex HF. After a stabilizing time of 2 h, we turned off the air conditioning (raising the room temperature from 21 °C to about 29 °C in 15 min) and then turned it on again. During this procedure, 14 measurements (8 blocks of

Fig. 7. Influence of room temperature change on the instrumental mass bias. Fourteen measurements (M1–14) of the AAS Zn–Cu standard were carried out with the SIS spray chamber during a T°C change of 8 °C (from 21 to 29 °C). A: The Cu (and Zn, not shown) isotope ratios show similar variation to the temperature. But Cu isotope ratio and temperature maxima are shifted from approximately 30 min. B: The regression line in the ln–ln plot changes also during this temperature change. The measurements (M 8 and 9, triangles) during T°C increase define a regression line with a different slop (1.0663) from those of measurements at 21 °C (1.1706, M1–7 and M12–14, open diamonds) and measurements during T°C decrease (1.3143, M10 and 11, circles). This T°C change of 8 °C can induce a maximum δ66Zn variation of −0.15‰.

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

2002; Bermin et al., 2006). According to the reported partition coefficients (Pai, 1988), almost all metals and Ca should be retained on Chelex-100 resin at pH higher than 5.5. The theoretical capacity (total number of sites) of the resin is 0.7 meq/ml. With 0.5 ml of resin, hundreds of liters of water could potentially be purified given the average Zn concentrations in dilute river waters. We checked the Zn retention capacity by loading the Chelex-100 column with 1.8 l river water that had been previously stripped of Zn (with Chelex-100 resin), and then amended with 180 μg of Zn. All the Zn was fixed by the resin in this experiment. According to ion exchange chromatography theory, the necessary volume Velu to elute Zn (volume of maximum concentration) is given by: Velu = V0 + M×Kd

ð8Þ

where M is the dry mass of resin, Kd the partition coefficient (metal concentration on the wet resin divided by that in solution) and V0 the interstitial volume of the column (about 35% of the volume of the resin). At pH = 5.5, the calculated elution volume is about 5 · 104 ml according to the given Kd of Zn of about 105 (Pai, 1988), and the resin volume of 0.5 ml. This calculation shows that 50 l of river water can theoretically be passed through the column, before Zn breakthrough occurs. Our sample introduction flow rate of 1.6 ml/min allows a longer sample–resin interaction time compared to 2 ml/min (Pai, 1988), allowing a better retention (and thus recovery) of the elements

127

on this resin. As discussed above, metal partition coefficients, Kd, on Chelex-100 resin strongly depend on pH (Pai, 1988). We used this property to separate Zn from other elements also fixed on the resin (Ca, Mg, Cu, Fe, etc.). At pH = 3, Zn's Kd is about 100, but Ca and Mg have a Kd of about 10 and can therefore be removed from the resin. At pH = 2.5, Zn can be eluted using 10 ml acid, but Cu and Fe remain fixed as they have higher Kd. In order to check these theoretical considerations, elution curves have been established by systematically measuring the element concentrations in the different elution fractions using an ICP-MS. Results show that major elements such as Na and K are not retained by the resin (Fig. 8A). Ca, Mg, and Sr are detected only in the first 20 ml buffer solution, being 7.1%, 6.5%, and 7.4% of the total loaded amounts (the rest having already passed through during loading of the sample). A little of Fe is present in the first buffer fraction, while most of the Fe must be eluted after the last fraction (Fig. 8A). A first series of experiments shows that, in contrast to what was reported by Pai (1988), Zn is not eluted in 10 ml HNO3 at pH = 2.56 but in subsequent elution at pH = 2.10 (7.94 mM) or even 1.82 (15.14 mM). These differences are due to the extreme pH sensitivity of Chelex-100 resin and the probable influence of ionic strength on Kd or tailing buffering effect. A large volume (N30 ml) of HNO3 solution at pH 2.10 appears necessary to recover all Zn. The elution volume is smaller at pH 1.82 but Cu is already partially present in this fraction. A compromise between reducing the elution volume and quantitatively separating Zn and Cu is found by eluting Zn at pH 1.90. After

Fig. 8. Elution scheme of Zn pre-concentration on Chelex-100 resin (A) and purification on AG 1-X4 resin (B); reproducibility of δ66Zn for the pre-concentration, purification step and the whole separation procedure for AAS Zn–Cu standard in distilled water and/or purified river matrix (C). In the first step (Chelex-100), most of the major and trace elements are removed (A). The Zn fraction still contains Ni, which is removed in the second step (AG 1-X4) (B). The external reproducibility for both individual steps and total separation is of the same order of magnitude as our long-term external reproducibility for isotope MC-ICP-MS measurements of the Zn-AAS standard.

128

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

adjustment of volumes and HNO3 concentrations, the protocol of Zn elution from Ca, Cu, and other matrix elements is finally fixed (Fig. 8A). Trace element such as Cr and Ba are also found only in the 20 ml buffer elution; Ti and V are determined only in the following Cu elution fraction at pH = 0.88 (Fig. 8A). HNO3 was found to be better suited than HCl for sharp and well-separated elution peaks. After separation on Chelex-100, the Zn fraction is evaporated and further purified using a AG 1-X4 column procedure. This resin is used to separate Zn from Ni, as both have similar Kds on Chelex-100. Zn is finally eluted by 0.5 ml 0.5 M HNO3 (Fig. 8B). No Ni, Ba, La or Ce is found in this final Zn fraction. As shown by previous studies on Chelex-100, matrix can affect the Kd and hence the elution curves (Pai, 1988; Liu and Ingle, 1989; Jiann and Presley, 2002). In order to check matrix effects, we conducted 55 separations on synthetic distilled water solutions (n = 33) and columnpurified Seine River waters (n = 22) to test the separation reproducibility. In these synthetic solutions, precisely known amounts of Zn were added. Fractions collected before and after the Zn fractions were also analyzed on the Neptune ICPMS and recovery yields were systematically calculated. Results showed no significant Zn amounts in these neighbor fractions and Zn yields were of 100 ± 1% in the final Zn elution fraction. Zn isotopic measurements were carried out on 24 of the solutions generated during the separation tests at the different stages of the procedure (15 on Chelex-100 column alone, 5 on AG 1-X4 column alone and 4 for the whole two-step procedure, Fig. 8C). All δ66Zn of the final elution fractions varied from −0.07‰ to 0.07‰ with an average value of −0.01‰, confirming that no isotopic fractionation occurs during the separation procedure. The blanks for Chelex-100 and AG 1-X4 separation steps were respectively 6.8 ± 0.9 ng and 15 ± 5 pg. The total blank of the two-column separation was generally smaller than 7 ng. No contribution of the blank was found in MC-ICPMS Zn isotopic analysis on the Neptune, since at least 1 μg of Zn was separated from the raw aqueous solutions for isotopic measurement. Finally, we tested if possible isotopic fractionation of Zn occurs on the Chelex-100 resin when the recovery is not quantitative, since significant mass fractionations have been previously reported during ion-exchange chromatography (Marechal et al., 1999; Chang et al., 2003; Wombacher et al., 2003). Ten mg Zn-AAS standard was added to 0.05 M ammonium acetate solution (pH = 5.5) and loaded on the Chelex-100 column. After matrix elution, 10 sequential fractions of 1 ml HNO3 (pH = 1.90) were collected and analyzed on the Neptune. The results clearly showed that fractionation occurs on the column and that the first elution fractions are enriched in the heavy isotopes (Fig. 9), indicating that the light isotopes have a greater affinity for the resin as compared to the heavier Zn isotopes. A total δ66Zn variation of 0.18‰ was found between the beginning and the end of Zn elution (Fig. 9). A mass balance calculation gave an integrated δ66Zn value of 0.04‰ for all recovered 1 ml fractions, corresponding to the calibrated value of the original Zn-AAS standard (0.05‰). If 10% of Zn from the first elution is lost, then the calculated integrated isotopic ratio should be 0.02‰. This shows that a 100% recovery yield is absolutely necessary for accurate Zn isotopic determination. Preferential adsorption of light isotopes onto ion exchange resins has been reported for Mg and B isotopes (Lemarchand et al., 2002; Chang et al., 2003).

with 0.2 μm cellulose acetate or Millipore express membrane filter. All filtrates to be analyzed were collected after a filtration of 2 l of raw water and acidified to pH = 2 with distilled HNO3. Filtrated samples were stored in polypropylene bottles in a refrigerator at 4 °C. Previous studies have shown that a small fraction of Zn (the inert fraction) may perhaps not be retained by Chelex-100 resin when organic-matter-enriched water is passed through the column (Figura and McDuffie, 1979; Liu and Ingle, 1989; Jiann and Presley, 2002). In this study, 10 samples with variable dissolved organic carbon (DOC) concentrations (from 2 to 31 mg/l) were UV-irradiated before the chemical separation. Zn concentration and isotope composition determined on these irradiated samples were compared with those of the same non UV-irradiated samples. This experiment was done to test the possibility of losing isotopically distinct organic-matter-bound Zn during the separation protocol. The results show that slightly higher Zn concentrations were determined for eight UV-irradiated samples processed through the chemical procedure as compared to non-irradiated solutions. Two other samples showed lower columnpurified Zn concentrations in the UV-irradiated solutions (Table 2). The concentration difference (column 8 in Table 2) between irradiated and non-irradiated duplicates varied from 0.5% to 9%, with an average value of 2.7%, which is less than our average precision (5%) for Zn concentration measurement (by ICP-MS, X-series II, Thermo). These differences were not correlated to the DOC contents of the samples, and the absence of a systematic difference in Zn isotopic ratio between irradiated and non-irradiated samples further demonstrates that organic matter has no significant influence on the Zn isotope composition measured in the Seine River waters. The interpretation can be: 1) there are less colloids in our samples filtrated at 0.2 μm compared to previous results using 0.45 μm filters (Jiann and Presley, 2002) and up to 98% of Zn present as labile complexes and free ions at this pH condition (2–5.5) (Turner et al., 1981), or the complexed Zn is also retained by Chelex-100 resin due to its strong affinity; 2) if there is a Zn isotope fractionation between truly dissolved and organicmatter-bound Zn, the difference of δ66Zn between these two phases is probably smaller than analytical reproducibility (0.04%) and the amount of organic-Zn is too low; 3) the sample-loading pH (5.5) of chemical separation does not favor the adsorption of Zn on colloids, which have been desorbed during the sample conservation at pH = 2 (Gelabert et al., 2006). For the isotope measurements of natural water samples, the roof runoff sample, “RF5”, and the Seine River water sample, “S63”, were measured the most frequently (31 and 15 times, respectively, over numerous analytical sessions). These samples have a respective

5.5. Zn isotopic variation in the Seine River, France Seine River water was regularly sampled near the university campus in Paris and along a geographical transect from the headwater to the estuary (Chen et al., 2008). All river waters were collected with an acid-cleaned 2 L sampler (polypropylene) at a depth of 1.5 m below the water surface. Rainwater was collected on the university roofs with a Teflon-made rain-collector (0.1 m2 surface area). Sewage waste waters were collected directly in the sewage treatment system (SIAAP). Samples were filtered immediately after sampling using a Sartorius AG16540 Teflon-made filtration system (142 mm diameter)

Fig. 9. Zn isotope compositions of sequentially eluted 1 ml fractions collected from the first pre-concentration chemistry. Ten mg Zn-AAS was loaded on the Chelex-100 column in 0.05 M ammonium acetate solution. Zn isotopes fractionate during the column elution, with lighter isotopes being preferentially retained on the resin. The integrated value of 0.04‰ (solid line) is consistent with the original Zn-AAS value (0.05‰ ± 0.04, dotted line) calibrated by Zn-JMC and Cu-SRM976 international standards during two years.

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

129

Table 2 Results of comprising tests on direct measured and UV-irradiated samples Sample name S24 S45 S48 RW3 RF1 RF4 RF5 SW1 PTSW1 PTSW2

Sample type

The Seine River water The Seine River water The Seine River water Rain water Roof runoff Roof runoff Roof runoff Sewage water Plant-treated sewage water Plant-treated sewage water

D.O.C

Measurement result of non irradiated sample

Measurement result of UV-irradiated sample

Zn concentration deviation

(mg/l)

δ66Zn (‰) (n)

S.D. (2σ)

δ66Zn (‰) (n)

S.D. (2σ)

(%)

2.5 4.8 6.4 2.0 30.9 2.7 16.9 16.0 23.6 11.7

0.07 (12) 0.25 (9) 0.12 (12) 0.17 (12) −0.07 (10) −0.02 (17) −0.02 (31) 0.28 (10) 0.09 (12) −0.03 (12)

0.02 0.03 0.01 0.08 0.04 0.04 0.05 0.02 0.06 0.04

0.07 (7) 0.24 (3) 0.12 (6) 0.17 (6) −0.09 (4) −0.02 (6) −0.03 (6) 0.25 (4) 0.09 (4) −0.04 (4)

0.06 0.05 0.04 0.05 0.06 0.08 0.05 0.04 0.04 0.08

−5.4 −9.1 −1.8 −0.5 1.0 −3.1 −4.8 8.0 −4.8 −6.4

D.O.C.: dissolved organic carbon content; n, number of measurements; S.D.: standard deviation of all measurement results.

external 2σ uncertainty of 0.05 and 0.04‰ for δ66Zn (Chen et al., 2008). Thus, the long-term reproducibility is comparable to the shorter-term reproducibility that we present here. Zinc isotope compositions for 50 water samples from the Seine River basin are presented in a three isotope plot in Fig. 10. Two regression lines can be defined in Fig. 10 by the following equations:  δ68=64 Zn = 2:069×δ66=64 Zn−0:003 r 2 = 0:9972

ð9Þ

and  δ68=66 Zn = 1:155×δ66=64 Zn−0:005 r 2 = 0:9795 :

ð10Þ

The slopes of these two lines are consistent with the mass-dependent fractionation at the 95% confidence level (e.g., Young et al., 2002; Viers et al., 2007). This indicates that all sample purifications were sufficient and neither residual interference nor statistic counting problems during MC-ICP-MS measurement and mass bias corrections have occurred during our Zn isotopic analyses. A detailed presentation and interpretation of these Zn isotope data can be found in Chen et al. (2008). 6. Conclusion We have developed a two-step chemical separation procedure for isolating Zn from dilute aqueous solutions (i.e., Zn concentration b10 μg/l). The first pre-concentration step using a Chelex-100 resin allows the direct introduction of the natural water sample onto the column at pH 5.5 without evaporation. The resin is very sensitive to pH changes, so metals are selectively eluted with HNO3 solutions of increasing acidity. The second step of purification using the AG 1-X4 resin allows the separation

Fig. 10. Three-isotope plot for measurement results of 50 environmental water samples with Zn concentration varying from 0.1 to 1700 μg/l. Data points of δ68/66Zn (squares) and δ68/64Zn (circles) versus δ66/64Zn plot on the mass-dependent fractionation lines, attesting the quality of the chemical separation and the MC-ICP-MS measurement.

of Zn from Ni. The whole separation protocol has been thoroughly tested, and has been shown to be reproducible and satisfactory for Zn isotope measurements using the Neptune MC-ICP-MS. The instrumental mass bias correction was performed using the empirical external normalization (EEN) method with Cu as an internal dopant. The external reproducibility of δ66Zn is 0.04‰ (2σ) for both standard and sample measurements. During one measurement session, several regression lines for the standard measurements must be defined and used for the EEN method, as we observe “jumps” of the mass bias when natural samples are measured. Our matrix-addition experiments suggest that these jumps result probably from imperfect chemical separation, despite the fact that ICP-MS measurements suggest the separates are pure. Though numerous tests have been performed to decipher the factors that impact mass bias sensitivity, some of the matrix effects are not resolved yet. With this analytical procedure for Zn separation and MC-ICP-MS measurement, we were able to successfully evaluate the Zn budget of the Seine River and clearly identify natural and anthropogenic inputs to the river (Chen et al., 2008). Acknowledgments We thank A. Galy, J. Bouchez, E. Tipper and D. Jouvin for fruitful discussions. D. Borrok is greatly thanked for English corrections and thorough reviews. SIAAP is thanked for sample supply. The authors wish to acknowledge two anonymous reviewers who greatly improved the quality of the manuscript. J.-B. Chen benefited from a grant of Region Ilede-France. This is IPGP contribution number 2422. References Al-Ammar, A.S., Gupta, R.K., Barnes, R.M., 1999. Correction for non-spectroscopic matrix effects in inductively coupled plasma-mass spectrometry by common analyte internal standardization. Spectrochim. Acta, Part B: Atom. Spectrosc. 54 (13), 1849–1860. Albarede, F., Telouk, P., Blichert-Toft, J., Boyet, M., Agranier, A., Nelson, B., 2004. Precise and accurate isotopic measurements using multiple-collector ICPMS. Geochim. Cosmochim. Acta 68 (12), 2725–2744. Andrén, H., Rodushkin, I., Stenberg, A., Malinovsky, D., Baxter, D.C., 2004. Sources of mass bias and isotope ratio variation in multi-collector ICP-MS: optimization of instrumental parameters based on experimental observations. J. Anal. At. Spectrom. 19, 1217–1224. Archer, C., Vance, D., 2004. Mass discrimination correction in multiple-collector plasma source mass spectrometry: an example using Cu and Zn isotopes. J. Anal. At. Spectrom 19, 656–665. Balistrieri, L.S., Borrok, D.M., Wanty, R.B., Ridley, W.I., 2008. Fractionation of Cu and Zn isotopes during adsorption onto amorphous Fe(III) oxyhydroxide: experimental mixing of acid rock drainage and ambient river water. Geochim. Cosmochim. Acta 72 (2), 311–328. Bermin, J., Vance, D., Archer, C., Statham, P.J., 2006. The determination of the isotopic composition of Cu and Zn in seawater. Chem. Geol. 226, 280–297. Borrok, D.M., Wanty, R.B., Ridley, W.I., Wolf, R., Lamothe, P.J., Adams, M., 2007. Separation of copper, iron, and zinc from complex aqueous solutions for isotopic measurement. Chem. Geol. 242, 400–414. Borrok, D.M., Nimick, D.A., Wanty, R.B., Ridley, W.I., 2008. Isotopic variations of dissolved copper and zinc in stream waters affected by historical mining. Geochim. Cosmochim. Acta 72, 329–344.

130

J.-B. Chen et al. / Chemical Geology 259 (2009) 120–130

Chang, V.T.-C., Makishima, A., Belshaw, N.S., O'Nions, R.K., 2003. Purification of Mg from low-Mg biogenic carbonates for isotope ratio determination using multiple collector ICP-MS. J. Anal. At. Spectrom 18, 296–301. Chen, J.-B., Gaillardet, J., Louvat, P., 2008. Zinc isotopes in the Seine River water, France: a probe of anthropogenic contamination. Environ. Sci. Technol. 42, 6494–6501. Cloquet, C., 2005. Géochimie élémentaire et isotopique des métaux lourds. Applications environnementales en tissu urban et industriel. PhD dissertation, Institut National Polytechnique de Lorraine, France. Cloquet, C., Carignan, J., Lehmann, M., Vanhaecke, F., 2008. Variation in the isotopic composition of zinc in the natural environment and the use of zinc isotopes in biogeosciences: a review. Anal. Bioanal. Chem. 390 (2), 451–463. Deines, P., Goldstein, S.L., Oelkers, E.H., Rudnick, R.L., Walter, L.M., 2003. Standards for publication of isotope ratio and chemical data in Chemical Geology. Chem. Geol 202 (1–2), 1–4. Figura, P., McDuffie, B., 1979. Use of Chelex resin for determination of labile trace metal fractions in aqueous ligand media and comparison of the method with anodic stripping voltammetry. Anal. Chem. 51 (1), 120–125. Fraser, M.M., Beauchemin, D., 2000. Effect of concomitant elements on the distribution of ions in inductively coupled plasma-mass spectroscopy. Part 1. Elemental ions. Spectrochim. Acta, Part B: Atom. Spectrosc. 55 (11), 1705–1731. Gaillardet, J., Viers, J., Dupré, B., 2005. Trace element in river waters. Treatise on geochemistry 5. Surface and Ground Water Weathering and Soils, pp. 225–272. Gelabert, A., Pokrovsky, O.S., Viers, J., Schott, J., Boudou, A., Feurtet-Mazel, A., 2006. Interaction between zinc and freshwater and marine diatom species: surface complexation and Zn isotope fractionation. Geochim. Cosmochim. Acta 70 (4), 839–857. Ingri, J., Malinovsky, D., Rodushkin, I., Baxter, D.C., Widerlund, A., Andersson, P., Gustafsson, O., Forsling, W., Ohlander, B., 2006. Iron isotope fractionation in river colloidal matter. Earth Planet. Sci. Lett. 245 (3–4), 792–798. Jiann, K.T., Presley, B.J., 2002. Preservation and determination of trace metal partitioning in river water by a two-column ion exchange method. Anal. Chem. 74, 4716–4724. John, S.G., Genevieve Park, J., Zhang, Z., Boyle, E.A., 2007. The isotopic composition of some common forms of anthropogenic zinc. Chem. Geol. 245 (1–2), 61–69. John, S.G., Rouxel, O.J., Craddock, P.R., Engwall, A.M., Boyle, E.A., 2008. Zinc stable isotopes in seafloor hydrothermal vent fluids and chimneys. Earth Planet. Sci. Lett. 269 (1–2), 17–28. Kavner, A., John, S.G., Sass, S., Boyle, E.A., 2008. Redox-driven stable isotope fractionation in transition metals: application to Zn electroplating. Geochim. Cosmochim. Acta 72, 1731–1741. Lemarchand, D., Gaillardet, J., Göpel, C., Manhès, G., 2002. An optimized procedure for boron separation and mass spectrometry analysis for river samples. Chem. Geol. 182, 323–334. Liu, Y., Ingle, J.D.J., 1989. Automated two-column ion exchange system for determination of the speciation of metals in natural waters. Anal. Chem. 61, 525–529. Luck, J.M., Ben Othman, D., Albarede, F., 2005. Zn and Cu isotopic variations in chondrites and iron meteorites: early solar nebula reservoirs and parent-body processes. Geochem. Cosmochim. Acta 69 (22), 5351–5363. Malinovsky, D., 2004. Fractionation of Fe and Zn Isotopes: methods and some applications in aqueous geochemistry. PhD dissertation, Lulea University, Sweden. Malinovsky, D., Stenberg, A., Rodushkin, I., Andren, H., Ingri, J., Ohlander, B., Baxter, D.C., 2003. Performance of high resolution MC-ICP-MS for Fe isotope ratio measurements in sedimentary geological materials. J. Anal. At. Spectrom. 18, 687–693. Malinovsky, D., Dahlqvist, R., Baxter, D.C., Ingri, J., Rodushkin, I., 2005. Performance of diffusive gradients in thin films for measurement of the isotopic composition of soluble Zn. Anal. Chim. Acta 537 (1–2), 401–405. Marechal, C., Albarede, F., 2002. Ion-exchange fractionation of copper and zinc isotopes. Geochim. Cosmochim. Acta. 66, 1499–1509. Marechal, C., Telouk, P., Albarede, F., 1999. Precise analysis of copper and zinc isotopic compositions by plasma-source mass spectrometry. Chem. Geol. 156, 251–273. Marechal, C., Nicolas, E., Douchet, C., Albarede, F., 2000. Abundance of zinc isotopes as a marine biogeochemical tracer. Geochem. Geophys. Geosyst. 1 1999GC000029. Mason, T.F.D., Weiss, D.J., Horstwood, M., Parrish, R.R., Russell, S.S., Mullane, E., Coles, B.J., 2004a. High-precision Cu and Zn isotope analysis by plasma source mass spectrometry. Part 1. Spectral interferences and their correction. J. Anal. At. Spectrom. 19 (2), 209–217. Mason, T.F.D., Weiss, D.J., Horstwood, M., Parrish, R.R., Russell, S.S., Mullane, E., Coles, B.J., 2004b. High-precision Cu and Zn isotope analysis by plasma source mass spectrometry. Part 2. Correcting for mass discrimination effects. J. Anal. At. Spectrom. 19, 218–226. Moynier, F., Albarede, F., Herzog, G.F., 2006. Isotopic composition of zinc, copper, and iron in lunar samples. Geochem. Cosmochim. Acta 70, 6103–6117. Olesik, J.W., Hobbs, S.E., 1994. Monodisperse dried microparticulate injector: a new tool for studying fundamental processes in inductively coupled plasmas. Anal. Chem. 66, 3371–3378.

Pai, S.-C., 1988. Pre-concentration efficiency of Chelex-100 resin for heavy metals in seawater. Part 2. Distribution of heavy metals on a Chelex-100 column and optimization of the column efficiency by a plate simulation method. Anal. Chem. Acta 211, 271–280. Peel, K., Weiss, D., Chapman, J., Arnold, T., Coles, B., 2008. A simple combined samplestandard bracketing and inter-element correction procedure for accurate mass bias correction and precise Zn and Cu isotope ratio measurements. J. Anal. At. Spectrom. 23, 103–110. Petit, J.C.J., de Jong, J., Chou, L., Mattielli, N., 2008. Development of Cu and Zn isotope MC-ICP-MS measurements: application to suspended particulate matter and sediments from the Scheldt estuary. Geostand. Geoanal. Res. 0, 1–18. Pichat, S., Douchet, C., Albarede, F., 2003. Zinc isotope variations in deep-sea carbonates from the eastern equatorial Pacific over the last 175 ka. Earth Planet. Sci. Lett. 210, 167–178. Poitrasson, F., Freydier, R., 2005. Heavy iron isotope composition of granites determined by high resolution MC-ICP-MS. Chem. Geol. 222, 132–147. Pokrovsky, O.S., Viers, J., Freydier, R., 2005. Zinc stable isotope fractionation during its adsorption on oxides and hydroxides. J. Colloid Interface Sci. 291, 192–200. Rehkämper, M., Mezger, K., 2000. Investigation of matrix effects for Pb isotope ratio measurements by multiple collector ICP-MS: verification and application of optimized analytical protocols. J. Anal. At. Spectrom. 15, 1451–1460. SIAAP, Syndicat Interdépartemental pour Assainissement de l'Agglomération Parisienne. http://w1.siaap.fr/. Sonke, J.E., Sivry, Y., Viers, J., Feydier, R., Dejonghe, L., André, L., Aggarwal, J.K., Fontan, F., Dupré, B., 2008. Historical variations in the isotopic composition of atmospheric zinc deposition from a zinc smelter. Chem. Geol. 252, 145–157. Stenberg, A., Andren, H., Malinovsky, D., Engstrom, E., Rodushkin, I., Baxter, D.C., 2004. Isotopic variations of Zn in biological materials. Anal. Chem. 76, 3971–3978. Stewart, I.I., Olesik, J.W., 1999. Time-resolved measurements with single droplet introduction to investigate space-charge effects in plasma mass spectrometry. J. Am. Soc. Mass Spectrom. 10 (2), 159–174. Thirlwall, M., 2001. Inappropriate tail corrections can cause large inaccuracy in isotope ratio determination by MC-ICP-MS. J. Anal. At. Spectrom. 16, 1121–1125. Toutain, J.-P., Sonke, J., Munoz, M., Nonell, A., Polvé, M., Viers, J., Freydier, R., Sortino, F., Joron, J.-L., Sumarti, S., 2008. Evidence for Zn isotopic fractionation at Merapi volcano. Chem. Geol. 253, 74–82. Turner, D.R., Whitfield, M., Dickson, A.G., 1981. The equilibrium speciation of dissolved components in freshwater and sea water at 25 °C and 1 atm pressure. Geochim. Cosmochim. Acta 45 (6), 855–881. Vance, D., Archer, C., Bermin, J., Kennaway, G., Cox, E.J., Statham, P.J., Lohan, M.C., Ellwood, M.J., 2006. Zn isotopes as a new tracer of metal micronutrient usage in the oceans. Geochim. Cosmochim. Acta, Suppl. 1 70 (18), A666. Vanhaecke, F., Vanhoe, H., Dams, R., Vandecasteele, C., 1992. The use of internal standards in ICP-MS. Talanta 39 (7), 737–742. Vanhaecke, F., Dams, R., Vandecasteele, C., 1993. ‘Zone model’ as an explanation for signal behaviour and non-spectral interferences in inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 8, 433–438. Viers, J., Oliva, P., Nonell, A., Gelabert, A., Sonke, J.E., Freydier, R., Gainville, R., Dupre, B., 2007. Evidence of Zn isotopic fractionation in a soil–plant system of a pristine tropical watershed (Nsimi, Cameroon). Chem. Geol. 239 (1–2), 124–137. Weiss, D.J., Mason, Th.F.D., Zhao, F.J., Kirk, G.J.D., Ciles, B.J., Horstwood, M.S.A., 2005. Isotopic discrimination of zinc in higher plants. New Phytol. 165, 703–710. Weiss, D.J., Rausch, N., Mason, T.F.D., Coles, B.J., Wilkinson, J.J., Ukonmaanaho, L., Arnold, T., Nieminen, T.M., 2007. Atmospheric deposition and isotope biogeochemistry of zinc in ombrotrophic peat. Geochim. Cosmochim. Acta 71 (14), 3498–3517. Wilkinson, J.J., Weiss, D.J., Mason, T.F.D., Coles, B.J., 2005. Zinc isotope variation in hydrothermal systems: preliminary evidence from the Irish Midlands ore field. Econ. Geol. 100, 583–590. Wombacher, F., Rehkämper, M., Mezger, K., Munker, C., 2003. Stable isotope compositions of cadmium in geological materials and meteorites determined by multiple-collector ICPMS. Geochim. Cosmochim. Acta 67 (23), 4639–4654. Woodhead, J., 2002. A simple method for obtaining highly accurate Pb isotope data by MC-ICP-MS. J. Anal. At. Spectrom. 17, 1381–1385. Young, E.D., Galy, A., Nagahara, H., 2002. Kinetic and equilibrium mass-dependent isotope fractionation laws in nature and their geochemical and cosmochemical significance. Geochim. Cosmochim. Acta. 66 (6), 1095–1104. Zhu, X.K., O'Nions, R.K., Guo, Y., Belshaw, N.S., Rickard, D., 2000. Determination of natural Cu-isotope variation by plasma-source mass spectrometry: implications for use as geochemical tracers. Chem.Geol. 163 (1–4), 139–149.