Method Development for Isotope Analysis of Trace and Ultra-Trace Elements in Environmental Matrices

DOCTORAL T H E SIS Nicola Pallavicini Method Development for Isotope Analysis of Trace and Ultra-Trace Elements in Environmental Matrices

Department of Civil, Environmental and Natural Resources Engineering Division of Geosciences and Environmental Engineering

ISSN 1402-1544 Method Development for Isotope Analysis of ISBN 978-91-7583-719-2 (print) ISBN 978-91-7583-720-8 (pdf) Trace and Ultra-Trace Elements in Luleå University of Technology 2016 Environmental Matrices

Nicola Pallavicini

Applied Geology

Method development for isotope analysis of trace and Ultra-trace elements in environmental matrices

Nicola Pallavicini

Division of Geosciences and Environmental Engineering

Department of Civil, Environmental and Natural Resources Engineering

Luleå University of Technology

S-971 87 Luleå, Sweden Printed by Luleå University of Technology, Graphic Production 2016

ISSN 1402-1544 ISBN 978-91-7583-719-2 (print) ISBN 978-91-7583-720-8 (pdf) Luleå 2016 www.ltu.se

Abstract

The increasing load of toxic elements entering the ecosystems, as a consequence of anthropogenic processes, has grown public awareness in the last decades, resulting in a great number of studies focusing on pollution sources, transport, distribution, interactions with living organisms and remediation. Physical/chemical processes that drive the uptake, assimilation, compartmentation and translocation of heavy metals in biota has received a great deal of attention recently, since elemental concentrations and isotopic composition in biological matrices can be used as probes of both natural and anthropogenic sources. Further they can help to evaluate fate of contaminants and to assess bioavailability of such elements in nature. While poorly defined isotopic pools, multiple sources and fractionating processes add complexity to source identification studies, tracing is hindered mainly by poorly known or unidentified fractionating factors. High precision isotope ratio measurements have found increasing application in various branches of science, from classical isotope geochronology to complex multi-tracer experiments in environmental studies. Instrumental development and refining separation schemes have allowed higher quality data to be obtained and played a major role in the recent progress of the field. The use of modern techniques such as inductively coupled plasma sector field mass spectrometry (ICP-SFMS) and multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) for trace and ultra- trace element concentrations and isotope ratio measurements have given new opportunities. However, sources of errors must be accurately evaluated and avoided at every procedural step. Moreover, even with the utilization of sound analytical measurement protocols, source and process tracing in natural systems can be complicated further by spatial and temporal variability. The work described in the present thesis has been focused primarily on analytical method development, optimization and evaluation (including sample preparation, matrix separation, instrumental analysis and data evaluation stages) for isotopic and multi-elemental analyses in environmental samples at trace and ultra-trace levels. Special attention was paid to evaluate strengths and limitations of the methods as applied to complex natural environments, aiming at correct interpretation of isotopic results in environmental forensics. The analytical protocols covered several isotope systems of both stable (Cd, B, Cr, Cu, Fe, Tl and Zn) and radiogenic (Os, Pb and Sr) elements. Paper I was dedicated to the optimization and testing of a rapid and high sample throughput method for Os concentrations and isotope measurements by ICP-SFMS. If microwave (MW) digestion followed by sample introduction to ICP-SFMS by traditional solution nebulization (SN) offered unparalleled throughput important for processing large number of samples, high-pressure ashing (HPA) combined with gas-phase introduction (GPI) proved to be advantageous for samples with low (below 500 pg) analyte content. The method was applied to a large scale bio-monitoring case, confirming accumulation of anthropogenic Os in animals from an area affected by emissions from a stainless steel foundry. The method for Cr concentrations and isotope ratios in different environmental matrices was optimized in Paper II. A coupling between a high pressure/temperature acid digestion and a one pass, single column matrix separation allowed the analysis of chromites, soils, and biological matrices (first Cr isotope study in lichens and mosses) by ICP-SFMS and MC-ICP-MS. With an overall reproducibility of 0.11‰ (2Ȫ), the results suggested a uniform isotope composition in soil depth profiles. On the other hand a strong negative correlation found between ț53Cr and Cr concentrations in lichens and mosses indicates that airborne Cr from local anthropogenic source(s) is

depleted in heavy isotopes, therefore highlighting the possibility of utilization of Cr isotopes to trace local airborne pollution source from steel foundries. Paper III describes development of high-precision Cd isotope ratio measurement by MC-ICP-MS in a variety of environmental matrices. Several digestion methods (HPA, MW, ultrawave and ashing) were tested for sample preparation, followed by analyte separation from matrix using ion- exchange chromatography. The reproducibility of the method (2Ȫ for ț114Cd/110Cd) was found to be better than 0.1‰. The method was applied to a large number of birch leaves (n>80) collected at different locations and growth stages. Cd in birch leaves is enriched in heavier isotopes relative to the NIST SRM 3108 Cd standard with a mean ț114Cd/110Cd of 0.7‰. The fractionation is assumed to stem from sample uptake through the root system and element translocation in the plant and it exhibits profound between-tree as well as seasonal variations. The latter were compared with seasonal isotopic variations for other isotopic systems (Zn, Os, Pb) in the same trees to aid a better understanding of underlying processes. In Paper IV the number of isotope systems studied was extended to include B, Cd, Cu, Fe, Pb, Sr, Tl and Zn. The analytical procedure utilized a high pressure acid digestion (UltraCLAVE), which provides complete oxidation of the organic material in biological samples, and a two-column ion- exchange separation which represents further development of the separation scheme described in Paper III. Such sample preparation ensures low blank levels, efficient separation of matrix elements, sufficiently high analyte recoveries and reasonably high sample throughput. The method was applied to a large number of biological samples (n>240) and the data obtained represent the first combined characterization of variability in isotopic composition for eight elements in leaves, needles, lichens and mushrooms collected from a geographically confined area. To further explore the reason of variability observed, soil profiles from the same area were analyzed for both concentrations and isotopic compositions of B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn in Paper V. Results of this study suggest that the observed high variability can be dependent on operationally-defined fractions (assessed by applying a modified SEP to process soil samples) and on the typology of the individual matrix analyzed (assessed through the coupling of soil profile results to those obtained for other matrices: lysimetric waters, mushrooms, litter, needles, leaves and lichens). The method development conducted in this work highlights the importance of considering all possible sources of biases/errors as well as possibility to use overlapping sample preparation schemes for multi-isotope studies. The results obtained for different environmental matrices represent a starting point for discussing the role of natural isotopic variability in isotope applications and forensics, and the importance of in-depth knowledge of the multiple parameters affecting the variability observed.

Keywords: Isotope ratio measurements; ICP-MS; MC-ICP-MS; Bio-monitoring; natural variability; multi-tracer studies; fractionation

PREFACE

The thesis is based on the following papers hereafter referred to by their Roman numerals.

, Pallavicini, N., Ecke, F., Engström, E., Baxter, D.C., Rodushkin, I., 2013. A high-throughput method for the determination of Os concentrations and isotope ratio measurements in small- size biological samples. J. Anal. At. Spectrom. 28. doi:10.1039/c3ja50201e ,, 3RQWpU63DOODYLFLQL1(QJVWU|P(%D[WHU'&5RGXVKNLQ,&KURPLXP LVRWRSHUDWLRPHDVXUHPHQWVLQHQYLURQPHQWDOPDWULFHVE\0&,&306-$QDO$W6SHFWURP GRL&-$$ ,,, Pallavicini, N., Engström, E., Baxter, D.C., Öhlander, B., Ingri, J., Rodushkin, I., 2014. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS. J. Anal. At. Spectrom. 29. doi:10.1039/c4ja00125g ,9 Rodushkin, I., Pallavicini, N., Engström, E., Sörlin, D., Öhlander, B., Ingri, J., Baxter, D.C., 2016. Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation. J. Anal. At. Spectrom. 31, 220–233. doi:10.1039/C5JA00274E 9 Pallavicini, N.,(QJVWU|P( Baxter,'&gKODQGHU% Ingri, J., Hawley, S., Hirst, C., Rodushkina, K., Rodushkin, I., 2016. 5DQJHVRIB, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn FRQFHQWUDWLRQVDQG isotope ratios in environmentalmatrices from an urban area. Submitted to Science of the Total Environment.

Paper I, II, III and IV are reproduced by permission of the Royal Society of Chemistry

Paper V is reproduced by permission of Elsevier

Conference contributions

7th Nordic Conference on Plasma Spectrometry

A. Nicola Pallavicini, Emma Engström, Douglas C. Baxter and Ilia Rodushkin, 2014. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS. Presenting author: Nicola Pallavicini

B. Nicola Pallavicini, Emma Engström, Douglas C. Baxter, and Ilia Rodushkin, 2014. Variability in osmium, lead, zinc and cadmium isotope composition in birch leaves from Sweden. Presenting author: Nicola Pallavicini

C. Ilia Rodushkin, Nicola Pallavicini, Emma Engström and Douglas C. Baxter, 2014. Isotope analysis at trace and ultra-trace levels in environmental matrices Presenting author: Prof. Ilia Rodushkin

ICC 2014 The International Carbon Conference

D. Nicola Pallavicini, Emma Engström, Douglas C. Baxter and Ilia Rodushkin, 2014. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS. Presenting author: Nicola Pallavicini E. Nicola Pallavicini, Emma Engström, Douglas C. Baxter, and Ilia Rodushkin, 2014. Variability in osmium, lead, zinc and cadmium isotope composition in birch leaves from Sweden. Presenting author: Nicola Pallavicini

8th Nordic Conference on Plasma Spectrometry

F. Ilia Rodushkin, Nicola Pallavicini, Emma Engström, and Douglas C. Baxter, 2015. Variability in trace element isotope composition in environmental matrices. Presenting author: Prof. Ilia Rodushkin

Contents

1. INTRODUCTION ...... 1

1.1. SCOPE OF THE THESIS ...... 3 1.2. BASICS OF STABLE AND RADIOGENIC ISOTOPES ...... 4 1.3. ISOTOPE RATIOS IN ENVIRONMENTAL STUDIES ...... 5 1.4. METHOD DEVELOPMENT ...... 8 1.4.1. Sampling and digestion of samples ...... 9 1.4.2. Separation and interferences ...... 10 1.4.3. Concentrations and isotopic measurements ...... 11 1.4.4. Data collection and processing ...... 11 1.4.5. Quality control ...... 12 2. SUMMARY OF RESULTS ...... 13

2.1. OS CONCENTRATIONS AND ISOTOPE RATIO MEASUREMENTS IN BIOLOGICAL SAMPLES .. 13 2.1.1. SN vs GP introduction systems in ICP-SFMS ...... 15 2.1.2. Pollution source tracing: Os concentrations and isotope ratios method for large scale bio-monitoring ...... 16 2.2. CR ISOTOPE RATIO MEASUREMENTS IN ENVIRONMENTAL MATRICES ...... 17 2.2.1. Cr separation scheme ...... 17 2.2.2. Cr concentrations and isotope ratios for airborne pollution bio-monitoring ...... 18 2.3. CD ISOTOPE MEASUREMENTS AND NATURAL VARIABILITY: AN APPROACH TO MULTI- ISOTOPE STUDIES ...... 19 2.3.1. Purification of Cd, Pb and Zn through ion-exchange chromatography ...... 20 2.3.2. Single vs Multi-collector ...... 21 2.4. NATURAL VARIABILITY IN BIOLOGICAL SAMPLES: OPTIMIZATION OF A MULTI-ISOTOPE METHOD ...... 23 2.4.1. Purification of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn through ion-exchange chromatography ...... 23 2.5. NATURAL AND ARTIFICIAL VARIABILITY IN ENVIRONMENTAL SAMPLES ...... 26 2.5.1. Sequential extraction procedure ...... 26 2.5.2. Multi-element patterns...... 27 2.5.3. Single Element Patterns ...... 28 2.5.4. Landfills and industrial wastes: a case study...... 28 3. OVERALL CONCLUSIONS ...... 30 4. FUTURE STUDIES ...... 32 5. ACKNOWLEGMENTS ...... 33 6. REFERENCES ...... 34

1. Introduction

Biogeochemical cycling in the natural environment involves both essential and non-essential elements. More than 30 elements are considered to be essential for life (Schlesinger and Bernhardt, 2013), others are considered non-essential at a present level of knowledge and some are toxic. Nevertheless toxicity of an element can be identified as the capacity of the material to adversely affect biological functions (Smith and Huyck, 1999) and for many elements essentiality or potential toxicity depends on cumulative uptake/exposure. Public awareness has grown in recent decades, reflecting increased level of heavy metals in the environment caused by their ever growing industrial uses. Since the industrial revolution, emission sources have multiplied resulting in a significantly increased discharge of many potentially harmful elements with profound negative effects on ecosystems (De Vleeschouwer et al., 2007; Jarup, 2003; Thevenon et al., 2011). Sources of heavy metals in the environment include natural (e.g. geogenic), and anthropogenic [e.g. industrial, pharmaceutical, domestic effluent and atmospheric sources (Tchounwou et al., 2010)].

Metals (major, trace and ultra-trace elements) enter ecosystems through different pathways (e.g. weathering, dry and wet atmospheric deposition, desorption from soil surfaces etc.). Their bioavailability depends on a combination of individual elements characteristics, on prevailing physical and chemical conditions, and speciation. Mobilization from suspended particles, soils and sediments, for example, is driven by adsorption/desorption reactions and dissolution due to weathering. When available for uptake, metals are assimilated by biological systems through two main mechanisms: active transport through cellular membranes (mediated by carrier proteins) and passive diffusion. Once incorporated into biota, such elements undergo specific metabolic processes (that vary depending on the individual organism) and are eventually returned to the background environment (Driscoll et al., 1994). Proper regulation of uptake, assimilation, compartmentation and translocation of trace metals is of vital importance for the life of an organism (Lobinski et al., 2006). Content and isotope ratios of such metals can vary due to biological effects (Becker, 2007a). Furthermore, element isotopic composition recorded in biological matrices can be used as a probe of natural and anthropogenic sources in the environment they inhabit and aid studies of uptake processes. Such applications have grown dramatically and the development is reflected in recent literature (e.g. Bullen, 2012; Font et al., 2012; Irrgeher and Prohaska, 2015; Wiederhold, 2015). 1

Natural and anthropogenic metal fluxes cause temporal and spatial variability and modify elemental pools in terms of concentrations and isotope ratios. A differentiation between the two sources is important when it comes to quantification of pollution load. Assuming a stable endogenous (baseline) level and a single predominant and distinctive local emission point for the element of interest, identification/confirmation of the pollution source is relatively straightforward and concentration data obtained by appropriate analytical methodology can be sufficient for the task. Unfortunately, in real life situations, spatial and temporal variability in baseline levels, the existence of different potential local sources and contributions from long-range pollution, physicochemical variables and multitude of processes involved, combined with often low analyte concentrations, complicates definitive source identification.

Elements are subject to numerous isotope fractionating processes and their isotopic composition represents a unique record of such processes (Bullen, 2012). The differences in the relative abundances of isotopes have potentials to show source and fate of e.g. a contaminant in the environment (Rehkämper et al., 2008). Isotope tracing/authentication/provenance are currently well-recognized powerful tools in environmental forensics (e.g. Resano and Vanhaecke, 2012). Studies combining the use of more than two isotopic systems have exponentially grown in recent years, endorsing the potentials of multi-isotope studies (e.g. Jaouen et al., 2013; Liu et al., 2014; Rodrigues et al., 2011; Ruhl et al., 2014; Sherman et al., 2015).

Relatively recently, modern mass spectrometry instrumentation and method developments have opened the possibilities for the adoption of stable isotopes as environmental tracers (Irrgeher and Prohaska, 2015).

From the early days of gas source isotope ratio mass spectrometry, used for isotope ratio measurements of light elements (H, C, O, N, S), advancements in analytical instrumentation and refined methodological protocols have allowed for highly precise and reproducible measurements for a growing suite of elements and at lower concentrations, thus expanding the possibilities to new isotopic systems (including trace and ultra-trace elements) (Bullen, 2012; Cloquet et al., 2005; Gao et al., 2008; Kersten et al., 2014; Rodushkin et al., 2007b; Shiel et al., 2009). Nowadays, it is possible to perform isotope analysis with improved sensitivity and high precision/accuracy due to the refinement of techniques such as thermal

ionization mass spectrometry (TIMS) and more importantly, multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS). ICP-MS (both single and multi- collector) is at present time the most frequently employed analytical technique in routine isotope ratio analysis due to the flexibility attainable with different instrumentation and sample introduction systems (Becker, 2007b).

Nevertheless, obtaining accurate and precise isotopic information is a complex exercise since number of potential sources of error and/or biases have to be taken into account and addressed prior to, during and after mass spectrometric measurements. Thorough analytical work and method developments have to be performed in systematical way to produce reliable measurement methods. A complete validation of isotope ratio measurements is often a very complicated task because of the lack of matrix-matched certified reference materials (CRMs) with known isotopic composition(s), preventing straightforward accuracy check. All the individual stages of the method (sampling, sample preparation, analyte separation, measurements and data processing) have to be thoroughly controlled for analytical protocols enabling correction for error sources and estimation of combined uncertainties for data produced (Irrgeher and Prohaska, 2015).

For source identification, common limitations stem from poorly identifiable mixed pools, input from multiple sources and complicated fractionating processes, while process tracing finds its main challenge in poorly known or yet unidentified intermediate processes and associated fractionating factors (Wiederhold, 2015). Moreover, even when using analytically sound measurement protocols for source or process tracing in natural systems, further complication may arise from temporal and spatial variability in isotopic composition that may occur at different scales and need to be properly understood/addressed.

1.1. Scope of the thesis

The motivations for this work have been the following:

- To develop and optimize analytical protocols for precise and accurate isotope ratio measurements for a large number of trace and ultra-trace elements (osmium (Os), cadmium (Cd), boron (B), chromium (Cr), copper (Cu), iron (Fe), lead (Pb), strontium (Sr), thallium (Tl) and zinc (Zn) by ICP-MS (in both single and multi-

collector configuration) in various environmental and geological matrices with a focus on a multi-elemental approach. - to assess the potential of multi-isotope studies by applying the developed methods to large batches of environmental matrices. - to assess the spatial and temporal variability in isotope data in a limited geographic areas

1.2. Basics of stable and radiogenic isotopes

Isotopes of an element are defined as nuclides with the same number of protons but different number of neutrons, therefore being characterized by different masses.

In nature, two main processes are responsible for the variations in isotopic composition of elements: radioactive decay from a parent to a daughter nuclide, and stable isotope fractionation. Isotopic fractionation is defined as an alteration of the isotope ratio of an element, most frequently as a result of a geochemical or biogeochemical process (natural or anthropogenic). Radiogenic isotopes are formed when a parent nuclide decays to a daughter element (defined radiogenic isotope), thus changing the isotopic budget in a target material. Such variations in isotopic compositions of Sr, Pb, Nd, Os, Hf, Th, etc. have been widely exploited for assessing time of formation (geological dating) as well as identifying geological processes and source materials. Sr, Pb and Os systems (Rodushkin et al., 2007a; Wiederhold, 2015) found important uses in environmental and authentication/provenance (e.g in archeology, forensics and provenancing) studies.

Stable isotopes are nuclides that do not undergo radioactive decay and therefore do not disintegrate at measurable rates. Isotope abundance measurements are based on the fact that the sum of the abundances of all the isotopes with corresponding number of protons (i.e. belonging to the same element) is 100% (Becker, 2007b). Inorganic mass spectrometry is applied for determination of isotopic abundances and isotope ratio measurements. Results of stable isotope measurements are commonly reported as deviation in isotope ratios from so called ‘δ-zero’ isotopic standards, since absolute ratios (abundances) are difficult to measure with high precision and accuracy when variations in between isotopes abundances are minimal.

For this reason stable isotope ratios are reported as delta values (δ) through normalization of measured ratios in samples to the same ratio in a standard material (Wiederhold, 2015):

࢞ ࢟ ࢞Ȁ࢟ ሺࡹȀࡹሻ࢙ࢇ࢓࢖࢒ࢋ ( ૚૙૙૙ ( 1 כ ࡹൌ൤ ࢟ െ૚൨ ઼ ሺࡹȀ࢞ ࡹሻ ઼૙ܛܜ܉ܖ܌܉ܚ܌

Where xM and yM correspond to the two different isotopes of the element of interest, the (xM/yM) value refers to the measured ratio and (xM/yM) is the isotope ratio of the sample δ0standard bracketing standard used as δ-zero. The factor 1000 is used to convert the δ-values to per mil notation. When the δ-value refers to a ratio of a heavier to a lighter isotope, a positive δ-value corresponds to an enrichment in the heavier isotope compared to the standard which corresponds to the 0‰.

Existence of differences (mainly mass-dependent) in chemical and/or physical behavior of isotopes in reaction/processes results in isotopic changes even for ‘stable’ elements. This phenomenon is called isotope fractionation (Mook, 2001). At atomic level, fractionation occurs as a consequence of both physical (e.g. diffusion) and chemical (e.g. making and breaking of bonds) reactions (Newton, 2010). There are two kinds of mass-dependent isotope fractionation: kinetic and equilibrium fractionation. In the former, the result of the process is irreversible (one-way physical or chemical reactions as for example the evaporation of water) and is mainly determined by the binding energies of the compound of interest. Isotopically light molecules generally have higher velocities, smaller binding energies and react more rapidly compared to heavier ones. Equilibrium fractionation instead corresponds to the isotope effect involved in an equilibrium reaction. Therefore a condition for the existence of isotopic equilibrium between two compounds is the presence of an isotopic exchange mechanism (Mook, 2001).

1.3. Isotope ratios in environmental studies

Isotope information can be used in a wide range of applications, spanning from environmental geochemistry, to biological process analysis, clinical and metabolic studies, pollution source tracing and forensics and geographical provenance tracing (Degryse et al., 5

2010; Liu et al., 2014). Biogeochemical processes cause detectable isotopic variations in the environment. For example, redox reactions may cause high degree of isotopic fractionation in several systems [e.g. variations of about 3‰ in Cr, Cu and Fe systems (Bullen, 2012)]. Moreover biologically mediated processes such as nutrient uptake in living organisms, have shown to result in isotopic fractionating effects, even though less pronounced (Rehkämper et al., 2008). Therefore isotopic information have an important role to play in the characterization and study of biogeochemical cycling of elements. Additionally, radiogenic isotope systems carry information of age and origin of minerals.

This together (stable isotope fractionation as a result of geochemical and biogeochemical processes, and radiogenic alteration of isotope ratios as a function of time), provides a thorough basis for utilization of stable and radiogenic isotope information within environmental forensics. Since the isotope ratio of an element has the potential of carrying information of the origin, age and geochemical and/or biogeochemical processes the element has undergone, an accurate selection of isotope systems has the potential of providing the link needed for accurate emission source identification. Isotope information can be used as a fingerprint during provenance and source identification studies.

At the same time, distinct isotopic composition in host rocks and ore-forming intrusions can be used to trace pollution derived from for example mining activities and isotopically light elements produced in vapors by industrial processes can be traced against a heavy background signature.

Industrial processes may change isotope signatures in different directions and to a variable extent depending on the individual production technologies (Shiel et al., 2010) and the typology of matrix analyzed (e.g. fumes and slags (Cloquet et al., 2006; Wen et al., 2015) etc.).

Such modifications in isotope abundances therefore can be used to trace and quantify both natural and anthropogenic processes (Rehkämper et al., 2008).

As stated by Wiederhold et al.(2015), the important, major advantage of isotope methods is the relative independence of dilution effects. As a rule, the most promising systems are those influenced by a limited number of processes. Indeed, in contrast to experiments relying upon the use of artificially enriched tracers, isotopic composition in natural environments represents an intrinsic mark that results from a mixture of several source and processes. 6

When such sources are clearly discernible, mixing calculations represent a powerful tool to identify the weighed contribution for each source for a target point. As displayed in Figure 1, the method is relatively straightforward and fully efficient for those cases where anthropogenic loads carry an isotopic signal significantly different from the background composition.

Unfortunately in real life situations, spatial and temporal variability, the existence of different potential sources and contributions from long-range pollution sources together with often low analyte concentrations complicate definitive source attribution.

Therefore the challenges related to environmental isotope tracing can be summarized as follows (Wiederhold, 2015):

Figure 1 Mixing model calculation for tracing studies. Modified from Wiederhold (2015). The relative fractions of the two different metal sources (natural and anthropogenic) can be quantitatively identified in the river thanks to the metal isotope signatures.

- Anthropogenic source isotopic composition is often not well-defined (can be variable) or overlapping with the range of natural ratios - During and after element deposition/uptake/exposure, original ratios of stable isotopes can be altered by numerous intermediate processes - There are often a number of sources with variable or overlapping isotopic compositions - Variations in isotopic composition of natural background

Relative significance of these limitations varies between different elements (radiogenic vs stable) and different study objects, but as consequence, simple two end-point mixing model (Figure 1) is not always directly applicable.

Last but not least, deciphering isotope signals in natural systems requires sensitive, selective and precise analytical methods, since at trace- and ultra-trace analyte concentrations in often complex matrices and given limited sample amount (e.g. in biological samples), analytical performance (resolution) of available instrumentation may be insufficient to detect minor variations in isotope ratios.

1.4. Method development

Mass spectrometry is nowadays the method of choice for determining the composition of almost any sample matrix and can be considered as a routine technique for multi elemental trace- and ultra-trace analysis (Beauchemin, 2006) capable of providing isotopic information. In the last decades advances in technology have allowed significant improvements in the technique and since its introduction in the 1980s, ICP-MS has become the most versatile element-specific detection technique. Despite the wide range of instrumental types, designs and hardware configurations available, modern ICP-MS techniques share common ‘building blocks’ and the very general operation of any mass spectrometer can be described as follows: the sample is introduced into an ion source (ICP) where it is sequentially vaporized, atomized and ionized; positively charged ions are then extracted from the source into the vacuum region through continuously pumped interface and accelerated towards the mass analyzer where separation of ions by mass or by mass-to-charge ratio takes place. After separation, ions reach the detector(s) and signals proportional to ion beam intensity are amplified and recorded by a detection system either simultaneously (multi-collector

configurations) or sequentially (single collector set up) for subsequent data treatment. To obtain reliable isotopic data, rigorous elimination/control of all potential sources of bias/errors needs to be implemented throughout the entire analytical protocol.

A few important requirements for a correct stable isotope analysis are (De Groot, 2009):

- the preparation steps must avoid artificial fractionation, as a consequence of incomplete yield or else, biases thus introduced have to be corrected for in the data treatment stage; - contamination, memory effects or isotopic exchange are a consequence of external addition of material to the analyte of interest and must be minimized; - mass spectrometer measurements must be performed, when possible, in absence of spectral and non-spectral interferences; - instrumental mass bias during data evaluation must be carefully corrected for.

At the planning stage prior to the development of new methods, attention must be paid to optimize not only factors affecting quality of data generated, but also efficiency, throughput and consumption of reagents and consumables, factors that are fundamental for the achievement of “greener” and economically advantageous analytical schemes.

In the last decades, growing focus has been devoted to the development of analytical methods targeted at the implementation of the principles of Green Chemistry (Bendicho et al., 2012). As a rule such procedures strive to obtain a greening of the entire analytical steps. In the present work this criterion has been applied, as explained later in the text, at several scales.

1.4.1. Sampling and digestion of samples

The first step for a sound analytical investigation is the proper planning of sampling. For example the choice of items and consumables to be utilized should ensure a contamination- free handling/storage of samples and as a rule is driven by analytical requirements as sample matrix and analyte/s of interest (Irrgeher and Prohaska, 2015). In particular, for trace and ultra-trace studies it is of crucial importance to proceed with a thorough decontamination of transportation containers and lab wares. All the experiment included in this thesis were

performed in clean laboratory areas (Class 10000) by personnel wearing clean room gear and following all general precautions to reduce contaminations (Rodushkin et al., 2010a). Prior to sampling, a consideration must be given to ensure representativeness. Planning of sampling plays a key role when concentration and isotopic data of interest may exhibit natural variations. Isotope studies are no exception and variations in element content and isotopic composition in several environmental matrices depend upon different parameters such as the type of matrix, age of sample, location, sampling period, climatic conditions etc. (Pallavicini et al., 2014, Rodushkin et al., 2016, Chaussidon & Albarède, 1992; Tarricone, Wagner, & Klein, 2015). Once the material needed is collected, the first step in sample treatment in the laboratory is often to identify a suitable digestion process, which must ensure effective liberation of analyte element(s) from the sample matrix in a form suitable for subsequent separation steps. Mandatory yield checks should be included to prevent potential losses and contamination.

1.4.2. Separation and interferences

Complications in isotopic measurements, for trace- and ultra-trace studies in particular, derive from the presence of isobaric (overlapping in the signal registered in the mass spectrum of the analyte of interest with elements sharing the same mass) or polyatomic (combination of two or more elements having the same mass of the analyte of interest) interferences. Spectral interferences represent a serious limitation of ICP-MS based techniques, since they lead to inaccurate calculations of analyte concentrations, deriving from overestimations of measured values compared to the real ones (Lum and Sze-Yin Leung, 2016). Therefore, to minimize such hindrance, analyte element(s) should be separated from sample matrix and interfering elements. It is mandatory (with few exceptions) for isotopic analysis by MC-ICP-MS and desirable for sector field mass spectrometry (ICP- SFMS). The selection of an appropriate approach providing efficient analyte purification depends on the nature of the element studied and the sample composition.

Testing of separation effectiveness during present studies was accomplished by multi- elemental analysis of all fractions by ICP-SFMS. This provides (I) direct assessment of analyte recovery, (II) information on separation efficiency from matrix elements, (III) information on analyte concentration, needed for preparation of concentration and acid strength matched solutions for isotope ratio measurements and (IV) information on the 10

potential presence of spectrally interfering elements and isobars either from the sample matrix or from contamination during sample preparation.

1.4.3. Concentrations and isotopic measurements

After purification the analyte fractions are ready (after proper concentration and acid strength matching) to be introduced into the mass spectrometer for sample analysis. In the present study depending on the task to be achieved, ICP-SFMS, ELEMENT XR, Thermo Scientific, Bremen, Germany and MC-ICP-MS, NEPTUNE PLUS, Thermo Scientific, Bremen, Germany were employed. Both machines can operate in low-, medium- and high- mass resolution modes (with resolution defined asܴ ൌ ݉ȀǼ݉, with Ǽ݉ as the smallest mass difference where two masses ݉ and ݉൅Ǽ݉ are resolved). The use of a double- focusing configuration (the pairing in series of magnetic and electric field) is mandatory for the separation of ion beams with high energetic spread (Becker, 2007c). In conventional single collector instruments each isotope is measured sequentially. With MC-ICP-MS, a separate detector is dedicated for each isotope of interest and therefore detection occurs simultaneously, improving precision compared to ICP-SFMS, as small fluctuations in the ion beam are cancelled out (Wieser and Schwieters, 2005). Thus required level of precision dictates the choice of double-focusing sector field instruments for isotope ratio measurements. ICP-SFMS has been used in the present work mainly for concentration assessment in allowing for a complete screening of the elemental content in each sample. Moreover, the precision offered by the instrument is often sufficient for isotope ratio measurements of Os, Pb, B and Sr, while MC-ICP-MS was the method of choice when high precision isotopic data are the ultimate objective. Therefore MC-ICP-MS was used as routine isotope measurements technique in all the works described in this thesis.

1.4.4. Data collection and processing

Isotope data evaluation was performed offline, through spreadsheet calculations to correct for blanks, spectral interferences and instrumental mass bias. The majority of the isotope measurements included in the present thesis was performed by combination of internal standardization and bracketing calibration where δ-values were calculated against a bracketing δ-zero solutions (see par.1.2 for more details) following the

revised exponential correction model by Baxter et al. (2006) in order to correct for instrumental mass-bias. The blocks cyclically repeated during a measurement session incorporated standards and samples in the order: standard 1 – sample 1 – sample 2 – sample 3 – standard 2. The calculation of mean δ-values and σ for each sample was performed with the assumption of a linear variation in mass-bias through the measurement cycle; therefore ratios were calculated for sample 1 and 3 against standard 1 and 2 respectively, while sample 2 was calculated against the mean of the two standards.

1.4.5. Quality control

Method development and optimization should include a proper assessment of reproducibility and accuracy. Sample matrix, analyte content and all stages of the measurement procedure might affect method reproducibility. Thus accurate assessment of the overall reproducibility of the method would require replicate preparation, separation and analyses of all samples that might be impractical or even unfeasible for large studies. As a more cost and time efficient approach, typical method reproducibility can be estimated using a set of test samples, representative for matrices studied, prepared and analyzed in different analytical sessions.

In order to assess repeatability and immediate precision of the methods developed prior to applications to natural samples CRMs of matching matrix were generally prepared and analyzed several times during the course of the experiments. On the other hand, in-run instrumental repeatability was estimated as twice the standard deviation (SD) of duplicate consecutive measurements of a single sample preparation.

The accuracy of concentration data for most of the analytes was verified by analyses of various CRMs and comparison with certified values.

On the other hand verification of isotopic accuracy through a comparison with certified values, for a large part of the analytes investigated in the present work, was not possible due to the scarcity of certified isotopic standards. Therefore such evaluation was performed through other means. For some matrices tested (CRMs and samples with high analyte content in Paper II and δ-zero standards in Paper IV) it was possible to confirm the absence of column-induced fractionation with isotopic analysis prior to and after column separation. For instrumental quality control, sets of in-house isotope standards were analyzed in every 12

measurement sessions in Paper IV. In Paper III, the comparison between isotopic results found for several matrices (biological material and standards) processed through different preparation procedures allowed to test the influences of different preparation schemes on the obtained results. Furthermore method accuracy evaluation was performed through a comparison between data obtained with those previously published for similar matrices, where such data exist. This highlights the urgent need for inter-laboratory exercises to develop and validate suitable matrix-matched isotopic standards (as emphasized in Papers II- III and IV).

The next section provides a summary of the results obtained through the analytical method development leading to the composition of this thesis. For this purpose individual subsections will be dedicated to an overall description of the main experimental procedures supported by the most important findings.

2. Summary of results

2.1. Os concentrations and isotope ratio measurements in biological samples

Paper I describes development and validation of an optimized analytical procedure for multi- element characterization, as well as for reproducible Os concentration determination and isotope ratio measurement in various biological matrices suitable for large-scale bio- monitoring programs.

The study on which Paper I is based was performed using small tissue samples of bank vole - the most common vole species in the studied area - and it represents a continuation of a pilot study performed on a limited number of animals.

ICP-SFMS, equipped at different stages with traditional solution nebulization (SN) or gas- phase (GP) introduction systems, was used for both concentration and isotope ratio measurements of about 350 individual organs/tissues. Prior to concentration and isotope measurements, tissue samples need to be transformed into solution. Therefore the original matrix must undergo a digestion providing efficient elimination of sample matrix, sufficiently low blanks and high recoveries of analyte element(s). When dealing with trace- and ultra-trace analytes an optimized digestion should provide (Becker, 2007d):

- complete decomposition of sample matrix - high and reproducible analyte recovery - manageable method blanks and carry-over effects - high throughput and minimal labor intensity - the use of disposable lab ware

The method of choice for studies involving large number of samples is microwave-assisted digestion (MW, MDS-81D, CEM Corporation, Matthews, USA). The method was proven to be suitable for multi-elemental concentration analysis and Os isotope ratio measurements (more effective when ultimate isotopic precision is not a mandatory requirement (Pallavicini et al., 2013). Quantification by isotope dilution (ID) was used, accomplished by addition of isotopic spike prepared from Os metal enriched in the 190Os isotope (>97% enrichment, Oak Ridge, USA) to all samples prior to digestion. The main advantages of MW digestion performed using disposable vessels are the speed of preparation and high throughput of the method, though some gas-phase losses of Os tetroxide (OsO4) were observed. The latter is not an issue while using high-pressure asher (HPA-S, Anton Paar, Malmö, Sweden) technology, as this approach ensures complete sample/spike equilibration and quantitative analyte recovery. HPA closely resembles Carius tube digestion, a method widely utilized in the application of ID in Os studies (Birck and Barman, 1997; Qi et al., 2013; Rodushkin et al., 2007b; Shirey and Walker, 1995; Walker, 1988). On the other hand, HPA’s low throughput and long procedural time may limit applicability of such method in studies requiring rapid analysis of large sample batches. From recoveries of Os spike concentrations a few conclusions could be drawn. Namely, HPA digestion provides the highest recoveries with the potential to provide the most accurate data, while part of the Os spike can be lost during MW digestion. Nevertheless, due to a good reproducibility in Os spike recovery (though not complete) and significant (approximately 10-fold higher compared to HPA) throughput, potential losses in accuracy can be tolerated in large scale monitoring studies. Therefore the MW digestion procedure, coupled with solution nebulization (SN) introduction system, was deemed ‘fit for purpose’.

2.1.1. SN vs GP introduction systems in ICP-SFMS

For Os measurements, two different configurations of sample introduction system were tested. For samples with higher analyte content, both Os concentrations and 187Os/188Os isotopic ratios (together with multi-elemental data) were obtained by a traditional SN approach, feeding MW-assisted digests directly into the ICP, without matrix separation. The results obtained from ICP-SFMS measurements were corrected for blank, isobaric interferences from 187Re and Os spike contributions using tabulated isotope abundances. When concentrations of Os were below 500 pg g-1 such mathematical corrections were found insufficient for obtaining reliable isotopic information. Therefore an on-line distillation technique slightly modified from previous studies (Malinovsky et al., 2002; Rodushkin et al., 2007b) was adopted: a GP introduction system, which exploits the high tendency of Os to partition into the vapor phase as OsO4 at elevated temperatures and in oxidized conditions. This approach provides a simple and efficient way to increase instrumental sensitivity and to separate analyte from sample matrix and interfering elements on-line. However, this is achieved at the expense of lower sample throughput and higher (in many cases complete) consumption of sample digests.

Both concentration assessment and isotope ratio measurements were performed using ICP- SFMS in Paper I, using (depending on analyte concentration) various combinations of introduction system configurations and digestions. The outcomes suggested that even though SN allowed running a large number of sample analyses in a short time span, for high quality Os quantification, if no multi-elemental analysis is required or Os concentration is below a critical level (<10 pg g-1), GPI would be preferred, possibly coupled with HPA. Higher concentrations obtained for Os by HPA digestion compared to MW-assisted digestion during parallel analytical sessions proved incomplete recovery of the latter method. Precision of 187Os/188Os ratio measurements was better than 2-4% and 1.5% RSD when employing SN and GPI respectively.

The method proposed in Paper I offered significantly shorter procedural times compared to previously developed methodology (Rodushkin et al., 2007d), from about 30-60 to 8–10 minutes per solution analyzed, allowing a sensible reduction in operator hands-on time and Ar gas consumption and thus lowering operating cost and increasing throughput.

2.1.2. Pollution source tracing: Os concentrations and isotope ratios method for large scale bio- monitoring

Figure 2 depicts a typical source mixing situation: on one side (low concentrations/high radiogenic ratios) the main contribution is driven by a background source which becomes less predominant the closer we approach to the main pollution source (higher concentrations/lower anthropogenic ratios), with values matching those of the material processed in the smelter (Kemi chromite). These findings confirm emissions from ferrochrome smelting as the primary exposure source in the area (Rodushkin et al., 2010b, 2007c, 2007d). Interestingly, correlations between different vole’s body compartments as well as with many trace and ultra-trace elements were found (Figure 2, Paper I). Such information can be useful to delineate exposure pathways, metabolism, accumulation and potential toxicity.

Figure 2 Os isotope ratios in vole tissues/organs as a function of Os concentration. Isotope ratio measurements by GPI (open circles) and SN (closed circles)

2.2. Cr isotope ratio measurements in environmental matrices

2.2.1. Cr separation scheme

Paper II is based on the optimization of an analytical method for Cr concentrations and isotope ratios analysis in different environmental matrices. Several matrices were tested for a single column Cr separation procedure, adapted from previously proposed schemes (Ball and Bassett, 2000; Schoenberg et al., 2008). The modification of such methods allowed to obtain an efficient removal of matrix elements with consistently high Cr recoveries. The drawbacks deriving from the presence at trace level of a few interfering elements in Cr elution fraction, as for example Fe, could be obviated coupling such separation method with themulti- elemental schemeODWHUGLVFXVVHG 3DSHU,9 .

Samples with low analyte content were introduced into the mass spectrometer through a desolvating unit. Aridus II &HWDF 2PDKD 1( 86$ proved to be unsuitable for our purpose, due to the presence of severe intereferences of 40Ar12C+ on 52Cr, presumably originating from organics released by the membrane material. Consequently, the Aridus was replaced by an Apex desolvating nebulizer as a more sensitive introduction system.

The separation scheme is based on the property of Cr(VI) to form oxyanions in weak hydrochloric acid matrix. At load stage Cr is retained in the column, while most other cations/neutral elements are instead released. To maintain Cr in the oxidized form it is necessary to employ an oxidizing agent. For this purpose a coupling of (NH4)2S2O8 and NH3 was added to the sample solution. The addition of NH3 proved to be crucial for the obtainment of satisfactory recovery rates.

In approximately 20% of samples, Cr fractions were significantly contaminated by S demonstrating incomplete and variable separation of the latter.

In-run instrumental repeatability was as a rule better than 0.04‰ or 0.09‰, respectively, for SN and Apex introduction systems. Method reproducibility assessed by including in-house soil control samples corresponded to a 2σ of 0.11‰.

As in all the experiment performed during this PhD work, accuracy assessment was hampered by the absence of reference materials, certified for isotope ratios and in this specific case for δ53Cr.

2.2.2. Cr concentrations and isotope ratios for airborne pollution bio-monitoring

The method was applied to real life samples. Six soil profiles (the same batch of samples used in the work described in Paper V) were analysed together with lichens and moss samples. The latter were collected at different distances from Torneå steelworks (a chromite ore processing facility) on the Swedish/Finnish border, along a transect stretching from that same location to the town of Luleå. While negligible variations in isotopic composition were found for soil samples both in terms of depth and location, interesting results were obtained for the biological matrices.

In this work the first ever Cr isotope data for lichens and mosses suggested that these species can serve as bio-indicators for tracing air-borne Cr pollution emitted from stainless steel smelters.

As displayed in Figure 3 there is a clear inverse correlation between concentrations and δ53Cr in all the samples analysed. Closer to Luleå a markedly heavier δ53Cr signature was found in lichens compared to local soils (+0.4‰). This result can be explained as a predominant wet deposition contribution.

Closer to the Torneå steelworks, Cr concentrations increase and the isotope signature tends to lighter values (0.2‰) in both lichen and moss samples. The highest concentration is found at approximately 2 km from the steelwork, where primarily negative δ53Cr are found. The predominantly light composition of lichens and mosses can be a result of smelting and refining processes producing airborne light Cr isotopes as reported in literature for other isotopic systems (Rehkämper et al., 2012; Shiel et al., 2010).

Figure 3 δ53Cr in chromites and environmental samples as a function of inverse Cr concentration

2.3. Cd isotope measurements and natural variability: an approach to multi- isotope studies

The process of optimization of analytical procedures for large scale bio-monitoring programs proceeds in Paper III. A reproducible Cd isotope ratio measurement methodology was evaluated and modified where necessary to be suitable to large scale analysis of biological matrices (birch leaves/litter, mushrooms and lichens). Cd is a toxic element entering the environment through several industrial activities [mining (Larison et al., 2000), smelting/refining (Cloquet et al., 2006; Gao et al., 2008), waste incineration and coal combustion]. The limited advancements in the field of Cd isotope measurements are primarily due the low natural abundance of the element and the lacking of appropriate technology/methodology for precise measurements. In Paper III the addition of an ashing step for carbon-rich matrices prior to digestion was found to provide significant improvements in sample intake capacity, thus allowing isotope ratio measurements at lower analyte concentrations.

A detailed analysis of the suitability of the different digestion methods to the processing of biological matrices has been performed and several preparation schemes were tested including MW, HPA, ashing and UltraWave digestions/dissolutions using different acid/acid mixtures. Advantages offered by single cavity MW digestion system over the rest of the sample preparation methods have resulted in the adoption of UltraClave (Milestone, Sorisole, Italy) as preferable method for the digestion of materials in the majority of the studies included in this thesis. The approach combines high temperature and high pressure digestion conditions with a relatively high throughput (up to 40 samples per batch). The preparation time totals to three hours comprising 30 minutes ramp to 220°C followed by 20 minutes holding time at the same temperature, cooling and transfer plus dilution of sample digests to the final volume.

2.3.1. Purification of Cd, Pb and Zn through ion-exchange chromatography

In the literature several purification schemes have been proposed targeted at different analytes. Such schemes vary in complexity from a single pass through one column to very elaborative, multiple-stage schemes depending on the sample matrix (Ball and Bassett, 2000; Ferreira et al., 2007; Inagaki et al., 2007; Li et al., 2016; Wei et al., 2014; Yi et al., 1998).

The separation scheme optimized and applied to real life samples (Figure 4) used in Paper III is based on evolution and refinement of the protocol originally proposed by Cloquet et al. (2005) for Cd isolation. The protocol which involves the utilization of chromatographic columns packed with anion exchange resin AG-MP-1 (macroporous, 100–200 dry mesh size, 75–150mm wet bead size, Bio-Rad Laboratories AB, Solna, Sweden) results in efficient isolation of the main analytes Cd (recovery >95%), Pb (>90%) and Zn (>99%), with efficient decontamination from potentially interfering elements (Pd, Zr, Nb, Th). From the results obtained in Paper III, the proportion of the elements co-eluting with Cd varies depending on digestion method, sample matrix and even on each individual column used. Workarounds can be employed for example at the measurement stage, through the coupling of a desolvating unit (Aridus II WRWKH,&3V\VWHPZKLFKLQVDPSOHVZLWKKLJK0R

20 Figure 4 Detail of the ion-exchange separation procedure optimized in Paper I,I

FRQWHQWUHGXFHVWKHformation of interfering polyatomic species as MoO+. The utilization of a desolvating nebulizer lowers the requirements for the Cd concentration required for precise isotope ratio measurements down to 10 μg l-1 in 3-4 ml of solution and reduces the formation of oxides. On the other hand, in-run precision is degraded approximately two-fold while signal and wash-out times are significantly longer than with the standard configuration of the MC-ICP-MS sample introduction system. Thus the latter is advisable for processing of samples with relatively high analyte concentrations.

2.3.2. Single vs Multi-collector

In Paper III, ICP-SFMS was used exclusively for element concentrations assessment. Isotopic ratio measurements were performed using MC-ICP-MS equipped with either a

21 standard or a high sensitivity sample introduction setup (the previously mentioned Aridus II). In terms of quality of the results, the best long term reproducibility achieved for the entire method corresponded to 0.02‰ (2σ). One of the highlights of accuracy assessment is the absence of matrix-matched materials with certified isotopic information at present time available for the scientific community. Interestingly, parallel analysis of two different quality control standards (QCS) with same common supplier and catalogue number, but different amount of starting material, resulted in a 0.6‰ difference in δ114Cd/110Cd. Common δ-zero standards are vital for meaningful inter-laboratory comparisons. An important result was that the Cd in birch leaves had a composition enriched in heavy isotopes (mean δ114Cd/110Cd value of 0.7‰, range 0.3–1.3‰). Such results are somehow surprising, since the preferential accumulation of lighter metal isotopes in leaves is far more common in nature (Bullen, 2012; Pérez Rodríguez et al., 2014; Ryan et al., 2013; Weinstein et al., 2011; Viers et al., 2007). Birch leaves can be utilized as a bio-indicator and act like a record of their exposure to heavy metals, thanks to their tolerance levels and fast growing rate (Samecka-Cymerman et al., 2009). δ-values showed no significant correlation with concentration levels, independently of sampling location or plant height (with a tendency towards lighter isotopes at the top of the crown). For a better understanding of the origins of Cd fractionation in birch leaves and to propose a potential explanation for the observed values, concentrations and δ-values found were compared with other isotopic systems, namely Os, Pb and Zn, in the same matrices (birch leaves, lichens, mushrooms and litter). The results suggested that while for Os and Pb in plants an aerial contribution could be the predominant source, Cd and Zn isotope behavior was more likely to be attributable to a soil solution pool. Interestingly broad variations in Cd isotopic composition were found in trees growing in close proximity to one another even in samples collected on the same sampling date. This observation needed to be supported by further investigations, based on extended number of samples and including a broader range of isotopic systems. This has been the focus in Paper IV.

22 2.4. Natural variability in biological samples: optimization of a multi-isotope method

In Paper IV an analytical procedure allowing multi-elemental analyses and isotope ratio measurements of eight of these (B, Cd, Cu, Fe, Pb, Sr, Tl and Zn) in matrices relevant for bio-monitoring using a single high-pressure acid digestion was developed. The method was used to assess the natural variability of concentrations and isotopic compositions in bio- indicators (tree leaves, needles and mushrooms, over 240 samples) primarily in the city and suburbs of Luleå. Ranges found from leaves and needles were compared with data obtained for limited numbers of samples collected in Spain, Italy, France, United Kingdom and Iceland.

2.4.1. Purification of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn through ion-exchange chromatography

The core of Paper IV was the development of a two-column purification procedure providing low blank levels, efficient separation of matrix elements, sufficiently high analyte recoveries and relatively high sample throughput. The scheme utilized for ion-exchange separation in Paper III was complemented with the addition of supplementary steps resulting in an amalgamation of several published separation procedures merged to maximize separation efficiency and reduce procedural time. The reasoning behind such modification was to selectively elute a wide number of analytes of interest in a single pass from a single sample digest. Therefore the elution scheme for the isolation of Fe, Cu, and Zn proposed by Maréchal et al. (1999) was added as an initial step (same ion-exchange resin). With the introduction of a few prudent modifications it has been possible to extend the number of elements isolated from the single sample digest. The adapted protocol allows the elution of Cu, Fe, Zn and Cd in one single pass through the first ion exchange column. The implementation of a further column (filled with approximately 2 mL Sr-Spec resin, Eichrom Technologies, IL, USA) allowed the isolation of Sr and Pb through an adaptation of previously published methods (Rodushkin et al., 2011; Smet et al., 2010). Furthermore, as displayed in Figure 5, mixed load and matrix wash fraction contains all original Ag and Cr and can be used for separation of these analytes. Details about the validation of the Cr separation method are provided in Paper II.

23 Part of the work described in Paper IV included a temporal variability assessment test that consisted in a three year monitoring session of a specific set of samples and was initially performed to test the long-term reproducibility of the method. Four individual birch trees were monitored for elemental concentrations and isotope ratios at three stages (the last week of May of three consecutive years, 2013-2015), collected in a geographically limited area. The highlight of the investigation was the existence of significant between-year differences in terms of concentrations and isotope ratios in foliage from the same trees found for several elements. It was established that the observed elements behavior was due to differences in climatic conditions between the sampled years, due to fluctuations in the temperatures registered between years. Such results could indicate a variation in the isotopic source pools that supply elements to birch leaves at the early stage of growth and differences in fractionation effects during uptake or translocation within the plant. Our results and interpretation support recently published findings on Zn isotope behavior in larch needles. In their study Viers et al. (2015) interpreted the observed heavier δ66Zn signals in needles as an increased rooting depth together with a progressive decrease in organic carbon concentration due to soil thawing. The outcomes of the exercise described here demonstrate that foliage samples provide highly spatially- and temporally-resolved snapshots of elemental and isotopic interactions with deciduous plants on the individual scale. Therefore care must be given to interpretations based on temporally limited sampling campaigns, since the picture provided by the results could represent only a detail of a larger perspective.

24 Figure 5 Flow chart of the multi-elemental purificationprocedure developed in Paper I9

2.5. Natural and artificial variability in environmental samples

To understand the regularities in the isotopic variability obtained for bio-indicators in Paper IV, it was concluded that a better understanding of the composition of the individual variables within the system was needed. For this reason Paper V was focused on presenting a comprehensive dataset of concentrations and isotopic compositions for nine elements (B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn) in a variety of environmental samples. The latter included bio-indicator organisms (mushrooms, litter, needles, leaves and lichens) and 6 soil profiles sampled in two different locations selected on the base of elements content found in bio-indicators in the previous work. Soil solution was collected through drain-gauge passive-capillary lysimeters. A modified sequential extraction procedure (SEP) was also used as important compliment to bulk concentration and isotope ratio measurements, to obtain information about the sub-pools in which elements are present in soils.

Furthermore, the use of the multi-isotope approach in environmental studies was exemplified by presenting a case study involving attribution of contamination sources in two landfills contaminated by tailings from Fe production, Fe and Cu slag, and fly ash.

2.5.1. Sequential extraction procedure

One of the most relevant factors driving the rate of incorporation and fate of metals into biological organisms is the form or sub-pool in which elements are present in soil, i.e. dissolved, exchangeable, included in the mineral lattice and insoluble. Obtaining information related to metal speciation in soils, in addition to bulk concentrations, represents a valuable aid for the study of elements behavior in natural environment. For this purpose, in Paper V we used a SEP to process the collected soil profiles. The SEP utilized in the present work is a modification of a scheme developed in the Standards, Measurements and Testing program (SM & T–formerly BCR) of the European Union (Quevauviller et al., 1997, 1993). The reasoning behind a SEP is to mimic the extracting action of different leaching media in natural settings in order to differentiate metals on the basis of mobility in soils/sediments. The main steps involved in the procedure utilized in the present study can be summarized as shown in Table 1.

Table 1 Sequential extraction scheme utilized for soil samples in Paper V Code Extracted phase Solution Volume (mL)

F1 Exchangeable Distilled water 40

F2 Carbonates 0.11 M CH3COOH (acetic acid) 40 F3 Reducible 0.1 M NH2OH HCl (hydroxylamine hydrochloride) 40 F4 Oxidizable 8.8 M H2O2 (hydrogen peroxide) 10+10 1 M CH3COONH4 (ammonium acetate)pH 2 50 F5 Residual I Aqua regia 20 F6 Residual II HF 10

2.5.2. Multi-element patterns.

F1 fractions (defined as batch-type water) and lysimetric waters showed similar concentrations values for all the elements except for Cu, Pb and Zn (lower in lysimetric waters). The same applied to isotopic patterns for most of the elements. Lysimetric waters were found to be enriched in the heavy isotopes of Cr, Cu and Zn relative to the batch-type waters. These observations could be explained by elemental portioning/sorption processes between rainwater and soil particles, which can induce stable isotope fractionation, in combination with preferential incorporation of heavier isotopes in secondary minerals (Ziegler et al., 2005).

To differentiate chemical variability derived from biologic processes and from aerial deposition we compared the results obtained in the different biologic sample types. Lichens (rootless organisms) directly reflect aerial deposition patterns while mushrooms and vascular plants should be less affected by airborne sources (limited exposure and lower surface area/volume ratio). Lichens were found to have significantly higher concentrations of Cr, Pb and Tl compared to leaves and needles, but lower B and Sr. Narrow isotopic ranges found in lichens for most of the elements could be a reflection of a more homogenous airborne pool compared opposite to processes of root uptake and translocations in plants (Houben et al., 2014; Jouvin et al., 2012; Kiczka et al., 2010). Many elements were found to be isotopically similar in lichens and top soils suggesting dry deposition as an important process for local soil formation. High Cd, Cu, Zn accumulation was found in mushrooms confirming reported (Gast et al., 1988) high accumulating capacity for metals in these

organisms. Notable enrichment of lighter isotopes of Cu and Fe and heavier ones for Zn occurs in mushrooms compared to lysimetric waters. Such difference was attributed to redox processes potentially affecting uptake of former two elements. Fractionation during element uptake for Cu and Zn was confirmed by the fact that none of the soil fractions showed isotopic composition resembling those of mushrooms.

The wide isotopic ranges found in leaf and needle data support the assumption that biologic processes strongly overprint the ranges of isotopic composition found in the literature for the same matrices.

2.5.3. Single Element Patterns

Overall isotopic outcomes suggested that for Cu, Fe and Zn, the soil-plant transition resulted in a shift towards a lighter isotopic composition, while for B, Cd, Cr and Tl, the opposite was observed. This could be a reflection of differences in uptake strategies between different biological systems or attributable to the presence of a dominant aerial contribution with different isotopic signature compared to those of native soils. Interestingly the broad isotope ranges in isotopic results found in Paper IV were here confirmed.

2.5.4. Landfills and industrial wastes: a case study

Concentration and isotopic information for selected elements in samples of wastes from local industries as well as in two heavily contaminated soils (landfills) were presented in Paper V and are shown in Table 2.

Complementing such dataset with the results obtained for bio-indicators and soil profiles helped differentiating the key pollution contributors to the landfills and identifying Fe slag as key input into one of the landfills.

Cd cannot help to clearly trace pollution source for the landfills, but suggests either the presence of an unknown, significantly fractionated source or a significant post-depositional change. Cr concentrations appeared significantly elevated in both landfills and slags, but the narrow isotopic range found makes it unusable for this case study. Cu content is high in both landfills but the light isotopic signature of both landfills is not matching any of the potential sources unless substantial alteration of original ratios has occurred.

Fe concentration in the landfills is five times that found in natural soils and both wastes and landfills are characterized by heavier δ56Fe when compared to local topsoil. Specifically Fe concentrations in landfill B and slag production from Fe ore processing are similar and their isotopic signatures are identical (within uncertainty).

The Zn isotopic composition of landfill A is heavier than any of the potential source materials, despite being higher in Zn concentrations compared to local background soil.

Low total Zn concentrations in Fe slags are consistent with those found in landfill B and therefore further exclude Cu slag as dominant input. No enrichment is found for Pb in either landfill compared to local soils even though 206Pb/207Pb ratios in landfill B are close to those for both Fe and Cu slags. Therefore, the available body of data points at Fe slags as best match for material deposited in landfill B.

Table 2 isotopic and concentrations data for several elements found in Paper V in different industrial matrices

Cd δ114Cd Cr, δ53Cr Cu, δ63Cu Fe, δ56Fe Zn, δ66Zn Pb, 206Pb/207Pb

(μg g-1) (‰) (μg g-1) (‰) (μg g-1) (‰) (%) ‰ (μg g-1) (‰) (μg g-1)

Landfill A 1.5 0.0 450 0.2 2700 -1.0 15 0.4 1100 1.4 40 1.32

Landfill B 0.6 -1.5 2600 0.1 170 -1.4 10 1.0 180 -0.2 170 1.17

Fe slag 0.2 0.1 4100 0.1 100 0.7 12 0.9 150 -0.7 250 1.16

Cu slag 11 -0.2 650 0.1 16000 0.1 35 0.4 12000 0.0 200 1.17

Fly ash 1.1 1.7 50 0.2 25 0.3 2.7 0.2 50 0.6 15 1.21

3. Overall conclusions

This thesis focuses on development and optimization of analytical protocols aimed at precise, reproducible and high-throughput concentration and isotope ratio measurements in large batches of environmental samples. The methods developed are aimed at the analysis of a broad range of stable isotopic systems (radiogenic and non-radiogenic) using both single- and multi-collector ICP-MS.

The initially employed MW assisted digestion using individually loaded and pressurized vessels has been gradually abandoned in the course of the work, to be substituted by preparation based on single cavity UltraClave technology, combining a relatively high throughput with high temperature and high pressure conditions ensuring complete oxidation of the organic material.

The ion-exchange analyte separation scheme has been developed and calibrated from a single column separation originally targeted to the isolation of Cd (together with Zn and Pb) to an efficient yet relatively straightforward multi-elemental separation method for the isolation of Ag, B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn (extendable also to other analytes with minor due modifications).

Significant improvements in signal intensities for ultra-trace elements during isotope ratio measurements were achieved with help from the use of desolvating nebulizers. As a rule, system with desolvation occurring in cooled condenser was preferable to those based on the use of a heated semipermeable membrane, at least for some of the analytes studied (e.g. Cd and Cr). However, higher signal intensity was often accompanied by deterioration in stability with worsening of in-run precision as result. Therefore the use of traditional high stability SN introduction system with double spray chamber configuration for MC-ICP-MS is advisable when analyte concentration is not a limiting factor. Highly accurate Os measurements requires the use of analyte specific GPI system combined with ICP-SFMS and HPA digestion, in particular when multi-elemental information is not mandatory.

The absence of commercially available reference materials, specifically certified in isotope ratios for similar biological matrices, constituted a constant drawback limiting accuracy assessment. There is an urgent need for inter-laboratory exercises aimed at comparing results to fill the gap until appropriate CRMs become commercially available.

Nevertheless in the present work the potential of isotopic data as a powerful tool to study both sources of metals and processes affecting their fate in the environment has been confirmed.

Both Papers I and II have explored the potential of isotopic information in differentiating pollution sources. In the former, Os results by ICP-SFMS helped confirming the presence of anthropogenic Os exposure from an emission point source in biological samples, while variations found in δ53Cr have provided insight into the possible source of local airborne Cr pollution. A further examples of isotope tracing has been provided in Paper V, where multi- element concentrations and isotope data reveals correlations between emission source and target matrix. On the other hand, Papers III and IV have explored how variations in isotopic compositions of biota can be caused by various natural processes, with often unexpected outcomes (e.g. the broad variations in the isotopic ranges found for almost all the elements). Furthermore, the results obtained from the application of the developed multi-isotope method to environmental samples indicates the existence of surprisingly broad ranges of isotopic composition for the majority of the isotopic systems analyzed, despite confined sampling area. Such high degree of variability can stem from both natural and anthropogenic processes as seen in recent literature (Martinková et al., 2016).

The amplitude of natural isotopic variations of the analyte/s of interest needs to be rigorously assessed in order to improve planning of representative sampling in environmental studies relying upon isotopic information.

4. Future Studies

Isotope applications are becoming an essential tool for environmental studies. Further refinement in technology will certainly expand the possibilities in terms of elements range and measurement performances (precision and sensitivity) allowing the identification of more subtle variations.

The optimization of analytical protocols with focus on reliability and overall efficiency of methods combined with developments in instrumentation (e.g. element-specific configurations of introduction system) will broaden the list of isotopes to be used and matrices to be studied. As an example, an extension of the hereby optimized ion-exchange separation methodology to cover more analytes (e.g. Ca, Mg, Hg, Ga, Re) could be explored. Increasing the number of analytes separated/purified from a single sample would improve throughput, utilization of sample material, reagents and chemicals and, most importantly, increase the amount of information available for environmental investigations. Such development would certainly be beneficial for multi-elemental and multi-isotope studies. The implementation of statistical and modeling tools, as for example Multivariate Analysis (MVA) and End Member Mixing Analysis (EMMA), would further boost the potentials of multi-isotope studies.

Further long-term investigations should be addressed towards the identification and assessment of variability (both spatial and temporal) in element’s concentrations and isotopic compositions in various matrices (both natural and anthropogenic). Creating a comprehensive database of the substrate compositions and variations within substrates would be extremely valuable as for understanding of process(es) involved as well as for efficient planning of tracing studies based on isotope measurements. Though there is increasing availability of materials with known isotopic compositions (e.g. commonly used δ-zero standards, single-element QC solutions or geological materials), they are still very scarce in the field of environmental sciences relying upon isotope ratio measurements in fungi, plants and animals. Developing a set of matrix-matched reference materials with known isotopic compositions for as many elements as possible will aid straightforward accuracy assessment for analytical methods and thus benefit isotopic studies using bio-indicators. The latter can most realistically be accomplished by organizing inter- laboratory comparisons using commercially available CRMs.

5. AcknowleGgHments

First of all, I would like to express my most sincere gratitude to my supervisor Prof. Ilia Rodushkin. His guidance and constant availability throughout my PhD studies have been vital for me to achieve what is now contained in this thesis. I also want to thank him for the time spent in meticulously reviewing all my written works. I also want to thank Dr. Emma Engström for her scientific support and availability to listen to my doubts and questions.

I would like to acknowledge Prof. Björn Öhlander, Prof. Johan Ingri and Dr. Douglas Baxter for critical reviewing of my works.

ALS Scandinavia AB is gratefully acknowledged for technical support. I especially would like to thank the following people: Dieke Sörlin; Isabelle Ström; Peter Nordlund; Hans Waara; Peter Silverplatz; Erik Burman; Katerina Rodiouchkina; Simon Pontér and Milan Vnuk. Colleagues at ALS Scandinavia AB and the Division of Geosciences and Environmental Engineering at Luleå University of Technology are all acknowledged for their support. MetTrans Initial Training Network (funded by the European Union under the Seventh Framework Programme) is acknowledged for financial support.

Non ci sono parole per esprimere la gratitudine nei confronti della mia famiglia, Mamma, Papá ed Elena. Senza il vostro aiuto nulla di questo sarebbe stato possibile.

To Io, for always understanding me and being the patient partner and friend that you are.

Last but not least a big thank you to all my friends, both “new” (in particular Giuseppe, Damiano, Elena and Marco) and “old” ones (Davide, Luca, Bubu, Fede, Gabry, Gian, i Paolo, Manuel, Nox, Ale, Zeno and all the friends in Genova and away), who have helped me through these years and never made me feel alone.

Grazie a tutti di cuore.

6. References

Ball, J.W., Bassett, R.L., 2000. Ion exchange separation of chromium from natural water matrix for stable isotope mass spectrometric analysis. Chem. Geol. 168, 123–134. doi:10.1016/S0009-2541(00)00189-3

Baxter, D.C., Rodushkin, I., Engström, E., Malinovsky, D., 2006. Revised exponential model for mass bias correction using an internal standard for isotope abundance ratio measurements by multi-collector inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 21, 427–430. doi:10.1039/b517457k

Beauchemin, D., 2006. Inductively coupled plasma mass spectrometry. Anal. Chem. 78, 4111–4135. doi:10.1021/ac060712t

Becker, J.S., 2007a. Biology, in: Inorganic Mass Spectrometry - Principles and Applications. John Wiley & Sons, Ltd, Chichester, UK, pp. 317–319. doi:10.1002/9780470517222

Becker, J.S., 2007b. Isotope ratio measurements and their applications, in: Inorganic Mass Spectrometry - Principles and Applications. John Wiley & Sons, Ltd, Chichester, UK, pp. 223–228.

Becker, J.S., 2007c. Introduction to Mass Spectrometry, in: Inorganic Mass Spectrometry - Principles and Applications. John Wiley & Sons, Ltd, Chichester, UK, pp. 1–30.

Becker, J.S., 2007d. Analytical and practical considerations, in: Inorganic Mass Spectrometry - Principles and Applications. John Wiley & Sons, Ltd, Chichester, UK, pp. 177–212.

Bendicho, C., Lavilla, I., Pena-Pereira, F., Romero, V., 2012. Green chemistry in analytical atomic spectrometry: a review. J. Anal. At. Spectrom. 27, 1831. doi:10.1039/c2ja30214d

Birck, J.L., Barman, M.R., 1997. Re-Os Isotopic Measurements at the Femtomole Level in Natural Samples 20, 19–27.

Bullen, T.D., 2012. Stable isotopes of transition and post-transition metals as tracers in environmental studies, in: Baskaran, M. (Ed.), Handbook of Environmental Isotope Geochemistry. Springer, Berlin, Heidelberg, pp. 177–203. doi:10.1007/978-3-642- 10637-8

Chaussidon, M., Albarède, F., 1992. Secular boron isotope variations in the continental crust: an ion microprobe study. Earth Planet. Sci. Lett. 108, 229–241. doi:10.1016/0012-821X(92)90025-Q

Cloquet, C., Carignan, J., Libourel, G., Sterckeman, T., Perdrix, E., 2006. Tracing source pollution in soils using cadmium and lead isotopes. Environ. Sci. Technol. 40, 2525– 2530.

Cloquet, C., Rouxel, O., Carignan, J., Libourel, G., 2005. Natural Cadmium Isotopic Variations in Eight Geological Reference Materials (NIST SRM 2711, BCR 176, GSS- 1, GXR-1, GXR-2, GSD-12, Nod-P-1, Nod-A-1) and Anthropogenic Samples, Measured by MC-ICP-MS. Geostand. Geoanalytical Res. 29, 95–106. doi:10.1111/j.1751-908X.2005.tb00658.x

De Groot, P.A., 2009. Handbook of Stable Isotope Analytical Techniques. Elsevier B.V.

De Vleeschouwer, F., Gérard, L., Goormaghtigh, C., Mattielli, N., Le Roux, G., Fagel, N., 2007. Atmospheric lead and heavy metal pollution records from a Belgian peat bog spanning the last two millenia: human impact on a regional to global scale. Sci. Total Environ. 377, 282–295. doi:10.1016/j.scitotenv.2007.02.017

Degryse, P., Shortland, A., De Muynck, D., Van Heghe, L., Scott, R., Neyt, B., Vanhaecke, F., 2010. Considerations on the provenance determination of plant ash glasses using strontium isotopes. J. Archaeol. Sci. 37, 3129–3135. doi:10.1016/j.jas.2010.07.014

Driscoll, C.T., Otton, J.K., Iverfeldt, Å., 1994. Trace Metals Speciation and Cycling, in: Biogeochemistry of Small Catchments: A Tool for Environmental Research. pp. 299– 322.

Ferreira, S.L.C., de Andrade, J.B., Korn, M.D.G. a, Pereira, M.D.G., Lemos, V. a, dos Santos, W.N.L., Rodrigues, F.D.M., Souza, A.S., Ferreira, H.S., da Silva, E.G.P., 2007. Review of procedures involving separation and preconcentration for the determination of cadmium using spectrometric techniques. J. Hazard. Mater. 145, 358–67. doi:10.1016/j.jhazmat.2007.03.077

Font, L., van der Peijl, G., van Wetten, I., Vroon, P., van der Wagt, B., Davies, G., 2012. Strontium and lead isotope ratios in human hair: investigating a potential tool for determining recent human geographical movements. J. Anal. At. Spectrom. 27, 719– 732. doi:10.1039/c2ja10361c

Gao, B., Liu, Y., Sun, K., Liang, X., Peng, P., Sheng, G., Fu, J., 2008. Precise determination of cadmium and lead isotopic compositions in river sediments. Anal. Chim. Acta 612, 114–120. doi:10.1016/j.aca.2008.02.020

Gast, C.H., Jansen, E., Bierling, J., Haanstra, L., 1988. Heavy metals in mushrooms and their relationship with soil characteristics. Chemosphere 17, 789–799. doi:10.1016/0045-6535(88)90258-5

Houben, D., Sonnet, P., Tricot, G., Mattielli, N., Couder, E., Opfergelt, S., 2014. Impact of root-induced mobilization of zinc on stable Zn isotope variation in the soil-plant system. Environ. Sci. Technol. 48, 7866–7873. doi:10.1021/es5002874

Inagaki, K., Narukawa, T., Yarita, T., Takatsu, A., Okamoto, K., Chiba, K., 2007. Determination of cadmium in grains by isotope dilution ICP-MS and coprecipitation using sample constituents as carrier precipitants. Anal. Bioanal. Chem. 389, 691–6. doi:10.1007/s00216-007-1396-7

Irrgeher, J., Prohaska, T., 2015. Application of non-traditional stable isotopes in analytical ecogeochemistry assessed by MC ICP-MS - A critical review. Anal. Bioanal. Chem. doi:10.1007/s00216-015-9025-3

Jaouen, K., Pons, M.L., Balter, V., 2013. Iron, copper and zinc isotopic fractionation up mammal trophic chains. Earth Planet. Sci. Lett. 374, 164–172. doi:10.1016/j.epsl.2013.05.037

Jarup, L., 2003. Hazards of heavy metal contamination. Br. Med. Bull. 68, 167–182. doi:10.1093/bmb/ldg032

Jouvin, D., Weiss, D.J., Mason, T.F.M., Bravin, M.N., Louvat, P., Zhao, F., Ferec, F., Hinsinger, P., Benedetti, M.F., 2012. Stable isotopes of Cu and Zn in higher plants: Evidence for Cu reduction at the root surface and two conceptual models for isotopic fractionation processes. Environ. Sci. Technol. 46, 2652–2660. doi:10.1021/es202587m

Kersten, M., Xiao, T., Kreissig, K., Brett, A., Coles, B.J., Rehkämper, M., 2014. Tracing anthropogenic thallium in soil using stable isotope compositions. Environ. Sci. Technol. 48, 9030–9036. doi:10.1021/es501968d

Kiczka, M., Wiederhold, J.G., Kraemer, S.M., Bourdon, B., Kretzschmar, R., 2010. Iron isotope fractionation during Fe uptake and translocation in alpine plants. Environ. Sci. Technol. 44, 6144–6150. doi:10.1021/es100863b

Larison, J.R., Likens, G.E., Fitzpatrick, J.W., Crock, J.G., 2000. Cadmium toxicity among wildlife in the Colorado Rocky Mountains. Nature 406, 181–3. doi:10.1038/35018068

Li, W., Wang, C., Gao, B., Wang, Y., Jin, X., Zhang, L., Sakyi, P.A., 2016. Determination of multi-element concentrations at ultra-low levels in alternating magnetite and pyrite by HR-ICP-MS using matrix removal and preconcentration. Microchem. J. 127, 237– 246. doi:10.1016/j.microc.2016.03.018

Liu, H.-C., You, C.F., Chen, C.Y., Liu, Y.C., Chung, M.T., 2014. Geographic determination of coffee beans using multi-element analysis and isotope ratios of boron and strontium. Food Chem. 142, 439–445. doi:10.1016/j.foodchem.2013.07.082

Lobinski, R., Moulin, C., Ortega, R., 2006. Imaging and speciation of trace elements in biological environment. Biochimie 88, 1591–1604. doi:10.1016/j.biochi.2006.10.003

Lum, T.-S., Sze-Yin Leung, K., 2016. Strategies to overcome spectral interference in ICP- MS detection. J. Anal. At. Spectrom. 31, 1–11. doi:10.1039/C5JA00497G

Malinovsky, D., Rodushkin, I., Baxter, D., Öhlander, B., 2002. Simplified method for the Re–Os dating of molybdenite using acid digestion and isotope dilution ICP-MS. Anal. Chim. Acta 463, 111–124. doi:10.1016/S0003-2670(02)00372-0 37

Maréchal, C.N., Télouk, P., Albarède, F., 1999. Precise analysis of copper and zinc isotopic compositions by plasma-source mass spectrometry. Chem. Geol. 156, 251–273. doi:10.1016/S0009-2541(98)00191-0

Martinková, E., Chrastný, V., Francová, M., Šípková, A., Curík, J., Myška, O., Mižic, L., 2016. Cadmium isotope fractionation of materials derived from various industrial processes. J. Hazard. Mater. 302, 114–119. doi:10.1016/j.jhazmat.2015.09.039

Mook, W.G., 2001. Environmental Isotopes in the Hydrological Cycle - Principles and Applications. International Atomic Energy Agency and United Nations Educational, Scientific and Cultural Organization.

Newton, J., 2010. Stable isotope ecology, Encyclopedia of Life Sciences. doi:10.1002/9780470015902.a0021231

Pallavicini, N., Ecke, F., Engström, E., Baxter, D.C., Rodushkin, I., 2013. A high- throughput method for the determination of Os concentrations and isotope ratio measurements in small-size biological samples. J. Anal. At. Spectrom. 28, 1591–1599. doi:10.1039/c3ja50201e

Pallavicini, N., Engström, E., Baxter, D.C., Öhlander, B., Ingri, J., Rodushkin, I., 2014. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS. J. Anal. At. Spectrom. 29, 1570–1584. doi:10.1039/C4JA00125G

Pérez Rodríguez, N., Langella, F., Rodushkin, I., Engström, E., Kothe, E., Alakangas, L., Öhlander, B., 2014. The role of bacterial consortium and organic amendment in Cu and Fe isotope fractionation in plants on a polluted mine site. Environ. Sci. Pollut. Res. 21, 6836–6844. doi:10.1007/s11356-013-2156-1

Qi, L., Gao, J.-F., Zhou, M.-F., Hu, J., 2013. The Design of Re-usable Carius Tubes for the Determination of Rhenium, Osmium and Platinum-Group Elements in Geological Samples. Geostand. Geoanalytical Res. 37, 345–351. doi:10.1111/j.1751- 908x.2012.00211.x

Quevauviller, P., Rauret, G., López-Sánchez, J.F., Rubio, R., Ure, A., Muntau, H., 1997. Certification of trace metal extractable contents in a sediment reference material (CRM 601) following a three-step sequential extraction procedure. Sci. Total Environ. 205, 223–234. doi:10.1016/S0048-9697(97)00205-2

Quevauviller, P., Ure, A., Muntau, H., Griepink, B., 1993. Speciation of heavy Metals in Soils and Sediments. An Account of the improvement and harmonization of extraction techniques undertaken under the auspices of the BCR of the Commission of the European Communities. Int. J. Environ. Anal. Chem. 51, 135–151. doi:10.1080/03067319308027619

Rehkämper, M., Mclaughlin, M., Kirby, J., Arnold, T.I.M., Paulo, S.Ã.O., Azil, B.R., 2008. Application of Nontraditional Stable-Isotope Systems to the study of sources and fate of metals in the environment. Environ. Sci. Technol. 42, 655–664.

Rehkämper, M., Wombacher, F., Horner, T.J., Xue, Z., 2012. Natural and anthropogenic Cd isotope variations, in: Baskaran, M. (Ed.), Handbook of Environmental Isotope Geochemistry. Springer, Berlin, Heidelberg, pp. 125–154. doi:10.1007/978-3-642- 10637-8

Resano, M., Vanhaecke, F., 2012. Forensic Applications, in: Isotopic Analysis. Wiley-VCH Verlag GmbH & Co. KGaA, pp. 391–418. doi:10.1002/9783527650484.ch14

Rodrigues, C., Brunner, M., Steiman, S., Bowen, G.J., Nogueira, J.M.F., Gautz, L., Prohaska, T., Máguas, C., 2011. Isotopes as tracers of the Hawaiian coffee-producing regions. J. Agric. Food Chem. 59, 10239–10246. doi:10.1021/jf200788p

Rodushkin, I., Baxter, D.C., Engström, E., Hoogewerff, J., Horn, P., Papesch, W., Watling, J., Latkoczy, C., van der Peijl, G., Berends-Montero, S., Ehleringer, J., Zdanowicz, V., 2011. Elemental and isotopic characterization of cane and beet sugars. J. Food Compos. Anal. 24, 70–78. doi:10.1016/j.jfca.2010.05.005

Rodushkin, I., Bergman, T., Douglas, G., Engström, E., Sörlin, D., Baxter, D.C., 2007a. Authentication of Kalix (N.E. Sweden) vendace caviar using inductively coupled plasma-based analytical techniques: evaluation of different approaches. Anal. Chim. Acta 583, 310–318. doi:10.1016/j.aca.2006.10.038

Rodushkin, I., Engström, E., Baxter, D.C., 2010a. Sources of contamination and remedial strategies in the multi-elemental trace analysis laboratory. Anal. Bioanal. Chem. 396, 365–377. doi:10.1007/s00216-009-3087-z

Rodushkin, I., Engström, E., Baxter, D.C., 2007b. Evaluation of Simultaneous Analyte Leaching/Vapour Phase Introduction for Direct Osmium Isotope Ratio Measurements in Solid Samples by Double-Focusing Sector Field ICP-MS. Geostand. Geoanalytical Res. 31, 27–38. doi:10.1111/j.1751-908X.2007.00835.x

Rodushkin, I., Engström, E., Sörlin, D., Baxter, D., Hörnfeldt, B., Nyholm, E., Ecke, F., 2010b. Uptake and Accumulation of Anthropogenic Os in Free-Living Bank Voles (Myodes glareolus). Water, Air, Soil Pollut. 218, 603–610. doi:10.1007/s11270-010- 0671-y

Rodushkin, I., Engström, E., Sörlin, D., Pontér, C., Baxter, D.C., 2007c. Osmium in environmental samples from Northeast Sweden. Part II. Identification of anthropogenic sources. Sci. Total Environ. 386, 159–168. doi:10.1016/j.scitotenv.2007.06.012

Rodushkin, I., Engström, E., Sörlin, D., Pontèr, C., Baxter, D.C., 2007d. Osmium in environmental samples from Northeast Sweden. Part I. Evaluation of background status. Sci. Total Environ. 386, 145–158. doi:10.1016/j.scitotenv.2007.06.011

Rodushkin, I., Pallavicini, N., Engström, E., Sörlin, D., Öhlander, B., Ingri, J., Baxter, D.C., 2016. Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation. J. Anal. At. Spectrom. 31, 220– 233. doi:10.1039/C5JA00274E

Ruhl, L.S., Dwyer, G.S., Hsu-kim, H., Hower, J.C., Vengosh, A., 2014. Boron and Strontium Isotopic Characterization of Coal Combustion Residuals: Validation of New Environmental Tracers. Environ. Sci. Technol. 48, 14790–14798.

Ryan, B.M., Kirby, J.K., Degryse, F., Harris, H., McLaughlin, M.J., Scheiderich, K., 2013. Copper speciation and isotopic fractionation in plants: uptake and translocation mechanisms. New Phytol. 199, 367–378. doi:10.1111/nph.12276

Samecka-Cymerman, A., Kolon, K., Kempers, a. J., 2009. Short shoots of Betula pendula Roth. as bioindicators of urban environmental pollution in Wrocław (Poland). Trees 23, 923–929. doi:10.1007/s00468-009-0334-z

Schlesinger, W.H., Bernhardt, E.S., 2013. Chapter 2 - Origins, in: Biogeochemistry (Third Edition). pp. 15–48. doi:10.1016/B978-0-12-385874-0.00002-9

Schoenberg, R., Zink, S., Staubwasser, M., von Blanckenburg, F., 2008. The stable Cr isotope inventory of solid Earth reservoirs determined by double spike MC-ICP-MS. Chem. Geol. 249, 294–306. doi:10.1016/j.chemgeo.2008.01.009

Sherman, L.S., Blum, J.D., Dvonch, J.T., Gratz, L.E., Landis, M.S., 2015. The use of Pb , Sr , and Hg isotopes in Great Lakes precipitation as a tool for pollution source attribution. Sci. Total Environ. 502, 362–374. doi:10.1016/j.scitotenv.2014.09.034

Shiel, A.E., Barling, J., Orians, K.J., Weis, D., 2009. Matrix effects on the multi-collector inductively coupled plasma mass spectrometric analysis of high-precision cadmium and zinc isotope ratios. Anal. Chim. Acta 633, 29–37. doi:10.1016/j.aca.2008.11.026

Shiel, A.E., Weis, D., Orians, K.J., 2010. Evaluation of zinc, cadmium and lead isotope fractionation during smelting and refining. Sci. Total Environ. 408, 2357–2368. doi:10.1016/j.scitotenv.2010.02.016

Shirey, S.B., Walker, R.J., 1995. Carius Tube Digestion for Low-Blank. Anal. Chem. 67, 2136–2141.

Smet, I., De Muynck, D., Vanhaecke, F., Elburg, M., 2010. From volcanic rock powder to Sr and Pb isotope ratios: a fit-for-purpose procedure for multi-collector ICP–mass spectrometric analysis. J. Anal. At. Spectrom. 25, 1025–1032. doi:10.1039/b926335g

Smith, K.S., Huyck, H.L.O., 1999. An Overview of the Abundance, Relative Mobility, Bioavailability, and Human Toxicity of Metals. Rev. Econ. Geol. Environ. Geochemistry Miner. Depos. Volumes 6A and 6B.

Tarricone, K., Wagner, G., Klein, R., 2015. Toward standardization of sample collection and preservation for the quality of results in biomonitoring with trees – A critical review. Ecol. Indic. 57, 341–359. doi:10.1016/j.ecolind.2015.05.012

Tchounwou, P.B., Yedjou, C.G., Patlolla, A.K., Sutton, D.J., 2010. Molecular, Clinical and Environmental Toxicology, Exs. doi:10.1007/978-3-7643-8338-1

Thevenon, F., Graham, N.D., Chiaradia, M., Arpagaus, P., Wildi, W., Poté, J., 2011. Local to regional scale industrial heavy metal pollution recorded in sediments of large freshwater lakes in central Europe (lakes Geneva and Lucerne) over the last centuries. Sci. Total Environ. 412–413, 239–247. doi:10.1016/j.scitotenv.2011.09.025

Walker, R.J., 1988. Low-Blank Chemical Separation of Rhenium and Osmium from Gram Quantities of Silicate Rock for Measurement by Resonance Ionization Mass Spectrometry. Anal. Chem. 11, 1231–1234.

Wei, H.Z., Jiang, S.Y., Gary Hemming, N., Yang, J.H., Yang, T., Wu, H.P., Yang, T.L., Yan, X., Pu, W., 2014. An improved procedure for separation/purification of boron from complex matrices and high-precision measurement of boron isotopes by positive thermal ionization and multicollector inductively coupled plasma mass spectrometry. Talanta 123, 151–160. doi:10.1016/j.talanta.2014.02.009

Weinstein, C., Moynier, F., Wang, K., Paniello, R., Foriel, J., Catalano, J., Pichat, S., 2011. Isotopic fractionation of Cu in plants. Chem. Geol. 286, 266–271. doi:10.1016/j.chemgeo.2011.05.010

Wen, H., Zhang, Y., Cloquet, C., Zhu, C., Fan, H., Luo, C., 2015. Tracing sources of pollution in soils from the Jinding Pb-Zn mining district in China using cadmium and lead isotopes. Appl. Geochemistry 52, 147–154. doi:10.1016/j.apgeochem.2014.11.025

Wiederhold, J.G., 2015. Metal stable isotope signatures as tracers in environmental geochemistry. Environ. Sci. Technol. 49, 2606–2624. doi:10.1021/es504683e

Viers, J., Oliva, P., Nonell, A., Gelabert, A., Sonke, J., 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, 124–137. doi:10.1016/j.chemgeo.2007.01.005

Viers, J., Prokushkin, A.S., Pokrovsky, O.S., Kirdyanov, A. V, Zouiten, C., Chmeleff, J., Meheut, M., Chabaux, F., Oliva, P., Dupré, B., 2015. Zn isotope fractionation in a pristine larch forest on permafrost-dominated soils in Central Siberia. Geochem. Trans. 16, 1–15. doi:10.1186/s12932-015-0018-0

Wieser, M.E., Schwieters, J.B., 2005. The development of multiple collector mass spectrometry for isotope ratio measurements. Int. J. Mass Spectrom. 242, 97–115. doi:10.1016/j.ijms.2004.11.029

Yi, W., Halliday, A.N., Lee, D., Rehkämper, M., 1998. Precise Determination of Cadmium , Indium and Tellurium Using Multiple Collector ICP-MS. Geostand. Newsl. 22, 173–179.

Ziegler, K., Chadwick, O.A., Brzezinski, M.A., Kelly, E.F., 2005. Natural variations of δ30Si ratios during progressive basalt weathering, Hawaiian Islands. Geochim. Cosmochim. Acta 69, 4597–4610. doi:10.1016/j.gca.2005.05.008

A high-throughput method for the determination of Os concentrations and isotope ratio measurements in small-size biological samples

Nicola Pallavicini, Frauke Ecke, Emma Engström, Douglas C. Baxter and Ilia Rodushkin

Journal of Analytical Atomic Spectrometry

www.rsc.org/jaas Volume 28 | Number 10 | October 2013 | Pages 1533–1668

ISSN 0267-9477

PAPER Nicola Pallavicini et al. A high-throughput method for the determination of Os concentrations and isotope ratio measurements in small-size biological samples 0267-9477(2013)28:10;1-7

JAAS

View Article Online PAPER View Journal | View Issue

A high-throughput method for the determination of Os concentrations and isotope ratio measurements in Cite this: J. Anal. At. Spectrom., 2013, 28,1591 small-size biological samples

Nicola Pallavicini,*ab Frauke Ecke,cd Emma Engstrom,¨ ab Douglas C. Baxterb and Ilia Rodushkinab

An analytical method allowing multi-element characterization by external calibration, osmium (Os) concentration determination by isotope dilution (ID) and 187Os/188Os isotope abundance ratio measurement from a single sample preparation was developed. The method consists of microwave- assisted, closed-vessel acid digestion of small (0.01–0.4 g dry weight) biological samples spiked with Os solution enriched in a 190Os isotope followed by concentration and Os isotope ratio measurements using double-focusing, sector field inductively coupled plasma mass-spectrometry (ICP-SFMS) operated with methane addition to the plasma and solution nebulization (SN) sample introduction. For samples with Os content below 500 pg, complementary analysis using gas-phase introduction (GPI) on the remaining sample digests was performed. The use of disposable plastic lab ware for sample digestion and analysis by SN ICP-SFMS circumvents Os carry-over effects and improves the sample throughput and cost-efficiency of the method. For a 0.1 g dried sample, Os method limits of detection (MLODs) of 2 pg g 1 and 0.2 pg g 1 were obtained using SN or GPI, respectively. Long-term reproducibility of 187Os/188Os isotope abundance ratio measurements using the GPI approach was better than 1.5% RSD for our in-house control sample Received 19th June 2013 (moose kidney) with an Os concentration of approximately 5 pg g1. Os data for several commercially Accepted 5th August 2013 available reference materials of biological or plant origin (not certified for Os) are presented. The method DOI: 10.1039/c3ja50201e was used in the large scale bio-monitoring of free-living bank voles from an area affected by www.rsc.org/jaas anthropogenic Os emissions.

Introduction concentrations in environmental compartments (e.g. sediments from estuarine, lacustrine areas and a peat bog core)2–7 and the 8 While osmium (Os) is one of the least abundant elements on highly toxic nature of gaseous osmium tetroxide (OsO4). Earth, the last two decades have witnessed a signicant number However, the available information on Os concentrations of studies devoted to identication of various anthropogenic and isotope abundances in biological materials that has sources of this element – hospital emissions from biological potential to be used in bio-monitoring is very scarce, which may waste incineration and sludge discharge, automobile catalyst largely be attributable to the considerable analytical challenges emissions, and smelters.1 Moreover, variations in Os isotope associated with accurate determinations at environmentally composition stemming from the radioactive decay of Re have relevant concentrations. In a pilot study based on a sampling found increasing use in environmental studies where source- campaign conducted in autumn 2007,10 we have combined the specic isotopic signatures have the potential to shed light multi-element detection capability of ICP-SFMS operated with upon possible anthropogenic sources of Os, other platinum either solution nebulization (SN) sample introduction or high- group elements (PGEs, i.e. Pt, Pd, Ir, Ru, Rh) or elements orig- sensitivity, element-specic, gas-phase introduction (GPI) of 2–4 inating from matrices containing Os. This interest can be OsO4 for Os isotope ratio measurements in a limited number explained by well-documented recent signicant increases in Os (n ¼ 22) of free-living voles, a common herbivore rodent of the boreal forest in northern Sweden, snap-trapped along a spatial gradient from a known, local, anthropogenic Os source. In spite aDivision of Geosciences, Lulea˚ University of Technology, S-971 87 Lulea,˚ Sweden. of unambiguous demonstration of anthropogenic Os accumu- E-mail: [email protected] lation in wild herbivores, it was concluded that, to further b ˚ ALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Lulea, Sweden ﬀ c explore potential toxicity e ects caused by the element, studies Department of Wildlife, Fish and Environmental Studies, Swedish University of  Agricultural Sciences, SE-901 83 Umea,˚ Sweden based on a signi cantly larger number of samples would be dDepartment of Aquatic Sciences and Assessment, Swedish University of Agricultural needed. Moreover, to test the hypothesis that seasonal foraging Sciences, Box 7050, SE-750 07 Uppsala, Sweden shis may alter Os accumulation in bank voles, animals

This journal is ª The Royal Society of Chemistry 2013 J. Anal. At. Spectrom., 2013, 28, 1591–1599 | 1591 View Article Online JAAS Paper collected during different seasons need to be analyzed to Table 1 Operation conditions and measurement parameters for multi- provide information at the organ or tissue level. elemental and Os analysis with SN and GPI introduction systems For such a large-scale investigation, the analytical approach Rf power (W) 1450 used in the pilot study needed to be stream-lined, while Sample uptake rate (ml min 1) 0.3 retaining the requirements for reproducible concentration data Argon gas ow rates (l min 1) for Os and other relevant major, minor, trace and ultra-trace Coolant 15 elements in different organs/tissues as well as Os isotopic Auxiliary 0.85 Nebulizer 0.85–0.92 information in samples low in Os. When the amount of available material is below 0.5 g, e.g. with individual organs of small Variable Specication animals or biopsies, pre-concentration of analytes by ashing11 is not a viable option, while the alternative of pooling material SN m/z from several animals will impede the detection of potential 7 9 11 82 85 88 ff Low resolution mode (LRM) Li, Be, B, Se, Rb, Sr, di erences between specimens and is not feasible for large 89Y, 93Nb, 95,98Mo, 107,109Ag, studies. The time requirements for Os measurements using GPI 111,114Cd, 115In, 118,120Sn, (approximately 30 min per sample) severely limit sample 121,123Sb, 125,126Te, 133Cs, 138Ba, 139 140 178,180 184 throughput and cost efficiency, in terms of instrumental and La, Ce, Hf, W, 185,187 189,190,192 191,193 operator ‘hands-on’ time, as well as consumption of Ar. Due to Re, Os, Ir, 194,195Pt, 197Au, 201,202Hg, 205Tl, the chemical properties of Os, the uncertainty of concentrations 206,207,208Pb, 209Bi, 232Th, 238U produced by ICP-SFMS equipped with either SN or GPI using Medium resolution mode 23Na, 24Mg, 27Al, 28Si, 31P, 32S, external calibration can be relatively high.11,12 (MRM) 42,44Ca, 45Sc, 47Ti, 51V,52Cr, 55 56 59 60 63 The aim of this work was to develop and validate an analyt- Mn, Fe, Co, Ni, Cu, 64 115 ical methodology optimized for multi-element characterization, Zn, In High resolution mode (HRM) 39K, 69Ga, 72,74Ge, 75As, 77,78Se, as well as reproducible Os concentration and isotope ratio 115In, measurement in various biological matrices suitable for large- Acquisition mode E-scan scale bio-monitoring programs using small tissue samples No. of scans 6 For each resolution a excised from wild animals. Acquisition window (%) 50 in LRM; 100 in MRM and HRM Search windowa (%) 50 in LRM; 80 in MRM and HRM Experimental Integration window (%) 50 in LRM; 60 in MRM and HRM Instrumentation Dwell time per sample (ms) 10–50 in LRM; 20 in MRM, 50 All analyses were performed using a double-focusing sector eld in HRM  No. of samples per nuclide 30 in LRM, 25 in MRM and ICP-MS ELEMENT XR (Thermo Scienti c, Bremen, Germany) HRM equipped at different stages with introduction systems for traditional SN or GPI. Common to both systems were the GPI 185 187 187 190 demountable quartz torch with a 1.5 mm i.d. sapphire injector, m/z Re, Os + Re, Os, 192 a platinum capacitive de-coupling shield, a nickel sampler cone, Os ‘ ’ Mass, search, acquisition 10,10,10 a high sensitivity X-type skimmer cone and a PFA spray window, % chamber with two gas inlet ports (Cetac Technologies, Omaha, Samples per peak 150 NE, USA). For SN, samples were delivered to a micro-concentric Sample time, ms 20 PolyPro nebulizer using a FAST SD2 auto-sampler (ESI, Perkin- Replicates/Runs/Passes 6 3 20 Load 6–20 ml digest + 1 ml H2O2 Elmer, Santa Clara, USA) equipped with a six-port valve and a  1 –  Gas ow through, l min 0.35 0.45 2 ml sample loop lled and rinsed by vacuum suction. Methane Mantle temperature 140 C addition to the plasma was used to decrease the formation of Purge delay, min 2.5–3.5 oxide-based spectral interferences, improve sensitivity for Introduction system Fassel torch, 1.5 mm i.d., elements with high rst ionization potentials, and minimize MicroMist AR40-1-F02 matrix effects.13 Operating conditions and measurement nebulizer, Scott type (double- pass) spray chamber, nickel parameters for concentration measurements were the same as cones 12 in previous studies, although the sampling time for masses Ion lens settings Adjusted daily to obtain 187, 190 and 192 was increased to 50 ms (Table 1). The total maximum signal intensity measurement time per sample, including stabilization and a Percent of peak width. rinse, was 3.5 min with 4 ml sample solution consumed. Details of the GPI distillation system consisting of a 60 ml Pyrex glass reaction vessel mounted in an electric heating mantle can be found elsewhere.11,12 Measurement parameters A laboratory microwave oven (MDS-81D, CEM Corporation, for isotope ratio measurements are summarized in Table 1, Matthews, USA) and a high pressure asher (HPA-S, Anton Paar, providing a total measuring time of 8–9 min per sample. Malmo,¨ Sweden) were used for sample digestion.

1592 | J. Anal. At. Spectrom., 2013, 28, 1591–1599 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper JAAS

Chemicals and reagents of anthropogenic Os in Northeast Sweden.15 All voles were trapped within an area of less than 1 km2.Aer dissection, Nitric acid (HNO3), hydrochloric acid (HCl), hydrogen peroxide organs and tissues were freeze-dried and stored at 18 Cin (H2O2, $30%, all from Sigma-Aldrich Chemie Gmbh, Munich, 1.5 ml plastic tubes prior to analysis. Germany) and hydrogen uoride (HF, 48%, Merck, Darmstadt, Germany) used in this work were all of analytical grade. Water used in all experimental procedures was de-ionized Milli-Q Sample preparation water (Millipore, Bedford, MA, USA) puried by reverse osmosis followed by ion-exchange cartridges. Samples were weighed directly into plastic digestion vessels Osmium spike stock solution was prepared from Os metal (polypropylene tubes with screw caps, 12 ml volume 101 16.5 enriched in a 190Os isotope (>97% enrichment, Oak Ridge, USA) mm or 30 ml volume 84 30 mm, Sarstedt AG & Co., Numbrecht,¨ Germany) with the choice of vessel dictated by the by Na2O2/Na2CO3 fusion (960 C for 60 min) in a glassy carbon vessel.14 Aer cooling, the melt was dissolved in hot MQ-water, amount of organic material available. Small vessels were used acidied to 2.5 mol l 1 HCl and stored in a glass bottle. The Os for samples up to 100 mg weight. For CRMs and in-house control, the sample weight was 100 5 mg while all available concentration in the spike was determined by reverse isotope 1 material was used for vole organs and tissues, thus eliminating dilution (ID) using 1000 mg l Os standards of natural isotope 1 190 composition from three producers (Merck, Darmstadt, Ger- the need for homogenization. An aliquot of 100 ng l Os many; Promochem, Ulricehamn, Sweden; and Inorganic spike solution was weighed in vessels (the amount of the spike providing approximately 2 pg 190Os spike per 100 mg of sample) Ventures, Christiansburg, VA, USA). A working spike solution 1 – with an Os concentration of 100 ng l 1 was prepared daily by followed by addition of a HNO3 HF mixture (99 : 1 ml ml , serial dilution of stock solution in 1.0 mol l 1 HCl. Aer 1 ml for samples up to 100 mg weight and 5 ml for larger completing the study, these working spike solutions were samples). Tubes were tightly capped and mounted in a 60 (for 12 ml vessels) or a 24 (for 30 ml vessels) position rack placed in a analyzed using ICP-SFMS against the freshly prepared standard, large plastic container. Milli-Q water was added to the container and no measurable losses of Os during 3 months of storage were found (concentrations in all solutions were the same within 5% to approximately ¼ of the vessel height, thus creating water- RSD), suggesting that the frequency of diluted spike prepara- bath-like conditions. The container was placed on the rotating tion can be decreased. turntable of the MW oven and digestion was performed by applying 200 W power for 25 min followed by 300 W power for 25 min. Water surrounding the digestion vessels serves as a heat Samples sink, which together with the low initial MW power setting For method development, a range of certied reference mate- prevents the oxidation of organic matter from occurring too rials (CRM) of animal and plant origin, ERM BB184 Bovine rapidly, thus reducing the risk of overpressure with losses of the Muscle, ERM BB186 Pig Kidney, ERM BB422 Fish Muscle (all sample as a result. Aer digestion, tubes were rapidly cooled from the Institute for Reference Materials and Measurements, down by lling the container with cold tap water before placing Geel, Belgium), GBW 07605 Tea (Institute of Geophysical and in a refrigerator at a temperature of ca. +5 C for at least 20 min. Geochemical Exploration, Langfang, China), SRM 1547 Peach In lipid-rich matrices, yellowish white deposits may form on Leaves, SRM 1577A Bovine Liver, SRM 1571 Orchard Leaves (all tube walls during cooling. Aer cooling, MQ-water was added to from the National Institute of Standards and Technology, Gai- each vessel resulting in pale-yellow digests of approximately 1 thersburg, MD, USA), TORT-1 Lobster Hepatopancreas (Insti- 1.4 mol l HNO3 with traces of HF. A set of at least three prep- tute for Environmental Chemistry, Ottawa, Canada), IAEA-A-13 aration blanks and an in-house control sample was prepared Animal Blood (International Atomic Energy Agency, Vienna, with each digestion batch. An internal standard (In, at 2 mgl 1 Austria) and GBW 07601 Human Hair (China National Analysis nal concentration) was added to digests for samples with initial Center for Iron and Steel, Beijing, China) as well as an in-house weights of 50 mg or less thus making them ready for ICP-SFMS control sample (freeze-dried kidney collected from moose analysis using the SN approach. The other digests were further 1 hunted in Northeast Sweden) were used. In order to limit diluted with 1.4 mol l HNO3 to a nal dilution factor of exogenous contamination, the latter was prepared and approximately 200 (ml g 1) before internal standard addition. All homogenized without using stainless-steel tools. Note that solutions were kept tightly closed and refrigerated prior to and none of the materials mentioned above has a certied Os between instrumental analyses. Even under such conditions, Os concentration or isotopic composition. losses from acidic solutions stored in polypropylene vessels Aer optimization, the robustness of the analytical meth- occur at a rate of approximately 30% per week. Therefore storing odology was tested by analyses of more than 350 individual digests for prolonged time periods should be avoided. organs and tissues (kidney, liver, lungs, spleen and muscle) of For HPA digestion, approximately 100 mg of CRM or in- common herbivore species (bank vole [Myodes glareolus], n ¼ 51 house control sample and 190Os spike solution were weighed and eld vole [Microtus agrestis], n ¼ 13). Voles were snap- into 100 ml quartz digestion vessels before addition of 2 ml  trapped in the spring (April) and autumn (September) of 2011 HNO3. Vessels were closed with quartz lids, sealed using Te on from the nature reserve of Riekkola, south of the town of tape and loaded into the HPA digestion chamber, which Haparanda (2490E,65470N) in close proximity (approximately accommodated ve vessels (four samples and one preparation ˚ 4–5 km distance) to the steelworks in Tornea – the major source blank). Digestion was performed under >100 bar N2 pressure

This journal is ª The Royal Society of Chemistry 2013 J. Anal. At. Spectrom., 2013, 28, 1591–1599 | 1593 View Article Online JAAS Paper with a temperature program comprised of 30 min ramp to are severe contamination risks from sampling and homogeniza- 220 C, 30 min hold at this temperature followed by 30 min tion equipment made of stainless steel, as well as blank contri- ramp to and 90 min hold at 300 C. Aer cooling to below 30 C, butions from reagents used for sample preparation. Secondly, pressure from the digestion chamber was slowly released and sample preparation at elevated temperatures and under oxidizing the vessels were transferred to a refrigerator and kept for at least conditions results in severe volatilization losses of the element

30 min. Digests were colorless and transparent for all biological through formation of OsO4 vapor. Owing to its high affinity for and plant materials prepared by this method. Aer diluting organic materials and permeability, this compound will even be solutions 10-fold with MQ-water, Os isotope ratio measure- lost from closed plastic containers via diffusion through walls and ments were performed by ICP-SFMS using the GPI approach absorption/adsorption at plastic surfaces. The latter may cause within 4 h of sample digestion. Quartz vessels and lids were severe carry-over if digestion vessels are used repeatedly for cleaned in a sequence with hot tap water and hot aqua regia preparation of samples with variable Os content unless a very followed by an MQ-water rinse and drying at 100 C between rigorous cleaning regime is implemented. Thirdly, at the analysis each digestion batch. stage, accurate quantication may be jeopardized by varying transport efficiency between samples and standards, prolonged ff Results and discussion memory e ects and spectral interferences (isobars of W, Re and Pt, as well as oxides and argides of rare earth elements), Multi-element analysis though in animal matrices the latter is of lower signicance Multi-element characterization of biological materials was done compared to geological or even plant samples. using sample preparation by closed vessel MW-assisted acid As no homogenization was performed on vole organs and digestion followed by ICP-SFMS with a combination of internal tissues analyzed in this study and freeze drying was done in standardization and external calibration according to an plastic containers, pre-analytical contamination may only be analytical protocol described in detail previously.16 Slight introduced during desiccation. As the same desiccation proce- 10 modications of the method included the use of disposable dure was used during the pilot study with the majority of vole polypropylene vials instead of jacketed digestion vessels made of samples from remote areas having Os concentrations below 1 Teon, a newer generation of ICP-SFMS instrument (ELEMENT MLOD of 2 pg g , the exogenous contribution is certainly XR versus ELEMENT 2) with the conguration of the introduction below this level. system offering approximately 2.5 fold higher sensitivity, and the Adding an enriched Os isotope spike prior to sample diges- use of a FAST auto-sampler with double probe rinse stations and tion compensates for potential element losses during acid sample loop rinsing and lling by vacuum suction. This excludes digestion assuming complete sample/spike equilibration. The direct contact between the analytical solution and peristaltic latter requirement could be violated during MW-assisted pump tubing, thus decreasing contamination and carry-over digestion in plastic vessels if Os from the isotopic spike pref- effects, and providing higher throughput due to shorter sample- erentially escapes from the acid solution before digestion of the uptake and washout times. biological matrix is completed, resulting in apparently higher The use of disposable vessels eliminates any risk of analyte Os concentrations quantied by ID. In order to assess the carry-over and the need for elaborative cleaning between diges- impact of incomplete equilibration, a set of acid blanks, repli- tion batches. Since digestion and analysis of samples weighing cates of CRMs (SRM 1547, SRM 1571 and TORT-1) and the in- 190 below 100 mg are performed using a single vessel, the risks for house control sample, all spiked with 2 pg Os, were digested handling contamination and blank contributions from auto- using HPA in quartz vessels closed by Teon-taped quartz lids sampler tubes are reduced as well. During MW-assisted diges- held in place by >100 bar external N2 pressure. In spite of tion, the vessel material releases Ca, P, Al, Ti and Ba. However, as signicantly higher temperature and pressure conditions this contribution is relatively uniform, it can be effectively cor- compared with MW-assisted digestion, material losses of Os rected by blank subtraction using digestion blanks. Method can only occur during post-digestion opening due to the use of limits of detection (MLODs), calculated as three times the stan- gas-tight, impermeable vessels, thus ensuring very efficient dard deviation for analyte concentrations measured in digestion sample/spike equilibration. In fact, sample preparation using blanks prepared together with biological materials and corrected HPA resembles Carius tube digestion, the reference method 17,22–25,28 for dilution corresponding to 100 mg sample weight, as well as a widely used for ID analysis of Os in geological studies. summary of concentration data for different organs and tissues, Unfortunately, HPA is not well suited to large-scale studies are presented in Table 2. Analyte recovery was assessed because of very limited throughput. For example, approximately using SRM1577A Bovine Liver and was acceptable (in the range of three months of preparation time would be needed for diges- 85–108%) for all elements with certied information or previ- tion of the 400 samples analyzed during the course of this study. ously published concentration data available (Table 2). Comparing Os concentrations in samples prepared by HPA and MW-assisted digestion (Table 3), it is obvious that the former provides approximately 25–30% higher values. This Osmium concentrations contradicts the possibility of preferential losses of spike from Determination of Os concentrations in biological samples, espe- plastic vessels and instead points out incomplete recovery of cially at environmentally relevant levels, represents a signicant endogenous Os from the sample matrix in MW-assisted diges- analytical challenge. Firstly, at ultra-trace concentrations, there tion as the most probable reason for the discrepancy. Useful

1594 | J. Anal. At. Spectrom., 2013, 28, 1591–1599 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper JAAS

Table 2 Element concentration in voles for diﬀerent tissues (median (range min–max)) and in SRM 1577a reference material

SRM 1577a mean, MLOD Kidney, n ¼ 64 Liver, n ¼ 64 Lung, n ¼ 64 Melt, n ¼ 64 Muscle, n ¼ 64 (SD, n ¼ 6)/certied

P (mg g 1) 0.005 13 (0.9–15) 14 (1.1–17) 12 (1.0–13) 16 (3.0–22) 9.4 (2.1–22) 11.3(0.5)/11.1 K (mg g 1) 0.006 11 (1.0–13) 11 (1.2–14) 12 (1.1–15) 16 (3.0–22) 12 (1.6–14) 10.6(0.5)/9.96 S (mg g 1) 0.020 11 (1.2–15) 11 (0.8–13) 10 (0.7–12) 8.8 (0.9–12) 9.3 (1.3–12) 8.2(0.6)/7.8 Na (mg g 1) 0.020 4.4 (0.5–6.0) 4.2 (0.4–4.9) 4.6 (0.4–5.7) 5.5 (0.5–7.3) 4.2 (0.6–5.2) 2.40(0.10)/2.43 Ca (mg g 1) 0.001 0.54 (0.19–1.3) 1.0 (0.2–8.9) 0.62 (0.14–0.95) 1.1 (0.1–2.9) 2.1 (0.34–24) 0.123(0.007)/0.12 Mg (mg g 1) 0.001 0.9 (0.11–1.2) 1.1 (0.64–5.6) 0.85 (0.08–1.1) 1.3 (0.13–2.4) 0.87 (0.16–1.4) 0.597(0.034)/0.60 Fe (mg g 1) 0.0003 0.4 (0.15–1.1) 0.68 (0.3–2.0) 0.91 (0.19–1.3) 1.2 (0.12–5.7) 0.15 (0.04–0.24) 0.187(0.011)/0.194 Rb (mgg 1) 0.03 62 (30–203) 61 (27–188) 70 (36–238) 86 (8.0–260) 61 (16–198) 12.2(0.6)/12.5 Mn (mgg 1) 0.01 40 (7.0–122) 28 (10–97) 6.0 (2.0–23) 90 (3.0–337) 32 (2.0–151) 10.1(0.5)/9.9 Zn (mgg 1) 0.10 95 (10–131) 96 (11–124) 74 (6.0–90) 99 (10–135) 49 (7.0–64) 125(8)/123 Si (mgg 1) 5.0 25 (8.0–49) 26 (7.0–81) 22 (8.0–55) 47 (<5–232) 30 (8.0–50) 7(2) Cu (mgg 1) 0.02 18 (3.0–27) 16 (2.0–20) 7.0 (1.0–11) 8.0 (1.0–26) 9.0 (2.0–16) 155(9)/158 Al (mgg 1) 0.10 4.3 (0.53–19) 2.8 (0.31–13) 1.5 (0.41–13) 12 (0.46–71) 3.3 (0.56–11) 1.4(0.3) Ba (mgg 1) 0.01 0.99 (0.18–4.3) 0.60 (0.06–2.5) 0.39 (0.09–1.0) 3.9 (0.2–19) 1.5 (0.12–10) 0.044(0.004) Sr (mgg 1) 0.004 0.79 (0.13–2.0) 0.63 (0.11–2.1) 0.38 (0.11–0.92) 1.9 (0.19–7.0) 1.5 (0.26–7.5) 0.133(0.007)/0.138 Mo (mgg 1) 0.001 2.0 (0.40–3.1) 4.2 (0.57–5.5) 1.0 (0.26–2.2) 1.1 (0.06–4.8) 0.3 (0.05–0.77) 3.55(0.20)/3.5 Se (mgg 1) 0.020 3.5 (0.78–5.3) 1.9 (0.54–3.5) 1.3 (0.27–1.7) 1.3 (0.29–2.1) 0.49 (0.13–0.75) 0.67(0.03)/0.71 Cr (mgg 1) 0.003 0.44 (0.04–1.2) 0.28 (0.02–1.0) 0.16 (0.02–1.1) 1.1 (0.02–8.6) 0.70 (0.09–3.0) 0.24(0.03) Ni (mgg 1) 0.005 0.74 (0.07–1.6) 0.42 (0.02–1.3) 0.17 (0.02–0.54) 1.2 (0.07–3.5) 0.90 (0.08–2.7) 0.285(0.025) Cs (mgg 1) 0.0001 0.24 (0.05–1.7) 0.22 (0.05–1.5) 0.28 (0.07–1.9) 0.32 (0.02–2.3) 0.25 (0.04–1.8) 0.008(0.002) Cd (mgg 1) 0.0003 1 (0.05–5.5) 0.18 (0.01–0.8) 0.06 (<0.0003–0.27) 0.14 (0.01–0.75) 0.02 (<0.0003–0.05) 0.40(0.02)/0.44 Ti (mgg 1) 0.010 0.21 (0.02–0.59) 0.16 (0.04–0.91) 0.12 (0.03–0.92) 0.65 (0.01–4.2) 0.15 (0.02–0.88) 0.10(0.01) Pb (mgg 1) 0.001 1.3 (0.22–3.3) 0.18 (0.05–0.39) 0.11 (0.03–0.21) 0.48 (0.02–1.9) 0.11 (0.02–0.54) 0.125(0.009)/0.135 B(mgg 1) 0.020 0.30 (0.06–0.95) 0.24 (0.09–0.73) 0.18 (0.06–0.71) 0.47 (0.01–2.9) 0.39 (0.08–1.6) 0.98(0.08) Hg (ng g 1) 2 755 (18–1624) 160 (5–455) 57 (6–123) 167 (2–451) 54 (4–109) 3.6(0.7)/4 Co (ng g 1) 0.3 259 (72–510) 217 (56–367) 118 (16–205) 239 (29–623) 160 (24–392) 202(15)/210 As (ng g 1) 3 67 (13–375) 46 (13–143) 27 (6–114) 76 (7–410) 78 (12–696) 49(5)/47 Li (ng g 1) 20 62 (20–312) 50 (<20–217) 45 (<20–211) 65 (<20–384) 25 (<20–201) 200(30) V (ng g 1) 0.40 53 (3–155) 26 (2–76) 14 (1–64) 89 (2–483) 26 (2–110) 92(6)/99 W (ng g 1) 0.70 25 (1–103) 17 (<0.7–64) 19 (<0.7–52) 50 (1–612) 12 (<0.7–29) 3.7(1.1) Tl (ng g 1) 0.02 40 (9–235) 17 (6–83) 12 (2–125) 19 (4–84) 6 (1–36) 2.5(0.2) Sn (ng g 1) 5 <5 (<5–36) <5 (<5–73) <5 (<5–114) 6 (<5–157) 5 (<5–121) 11(4) Zr (ng g 1) 5 9 (<5–71) 5 (<5–44) 6 (<5–66) 21 (<5–144) 11 (<5–72) 8(2) Ce (ng g 1) 0.08 3.3 (0.30–16) 5.4 (0.69–27) 1.7 (0.27–11) 19 (0.30–222) 2.2 (0.26–11) 18.7(1.6) Bi (ng g 1) 0.07 29 (2.3–102) 13 (0.6–38) 7.1 (0.3–19) 15 (0.3–64) 3.9 (0.2–12) 0.33(0.05) Sb (ng g 1) 1 11 (1–54) 7 (<1–24) 3 (<1–7) 19 (1–73) 12 (1–35) 3.0(0.5)/3 Ge (ng g 1) 2 37 (8–53) 20 (5.6–36) 14 (2.7–19) 13 (<2–20) 5 (<2–8) <2 La (ng g 1) 0.05 2.2 (0.19–9.8) 3.8 (0.58–16) 1.2 (0.2–7.7) 13 (0.23–139) 1.5 (0.17–7.4) 10.8(0.8) Nb (ng g 1) 0.05 2.5 (0.14–11) 1.9 (0.17–12) 1.3 (0.20–17) 8.5 (<0.05–87) 2.5 (0.20–13) 1.1(0.2) Y (ng g 1) 0.12 2.3 (0.30–8.5) 2.4 (0.50–6.9) 0.64 (0.16–4.5) 8.8 (0.32–46) 1.4 (0.14–6.2) 1.3(0.2) Te (ng g 1) 0.2 12 (0.6–29) 0.9 (<0.2–2.8) 0.7 (<0.2–1.5) 1.5 (<0.2–3.6) 1.2 (0.3–2) 3.6(0.5) Ga (ng g 1) 0.02 1.2 (0.45–4) 1.1 (0.41–3.6) 0.91 (0.50–4.6) 3 (0.11–19.2) 1 (0.25–2.7) 2.1(0.3) U (ng g 1) 0.04 1.1 (0.15–3.9) 2.9 (0.12–9.1) 0.18 (<0.04–1.1) 1.7 (0.06–5.8) 0.75 (0.09–3) 0.79(0.07) Sc (ng g 1) 0.04 0.45 (<0.04–2.2) 0.39 (0.04–2.3) 0.19 (0.03–1.6) 2.0 (<0.04–11.8) 0.68 (0.08–6.8) 4.5(0.6) Th (ng g 1) 0.02 0.58 (0.06–3.3) 0.36 (0.04–5.8) 0.23 (0.02–1) 1.3 (0.03–10.4) 0.21 (0.02–0.63) 0.57(0.09) Hf (ng g 1) 0.1 0.32 (0.11–1.7) 0.29 (0.10–0.92) 1.6 (0.47–6.7) 0.73 (0.03–3.9) 0.35 (0.10–0.66) 0.25(0.05) Be (ng g 1) 0.20 0.75 (<0.2–3.2) 0.71 (0.20–2.6) <0.2 (<0.2–0.34) 1.2 (<0.2–3.9) 0.31 (<0.2–1.2) <0.2 Au (pg g 1) 200 <200 <200 <200 <200 <200 <200 Os (pg g 1) 2 577 (18–1380) 285 (5–937) 187 (5–898) 1485 (20–7725) 925 (13–2945) <2 Ag (pg g 1) 2 <2 <2 <2 <2 <2 <2 Re (pg g 1) 1 30 (5–71) 29 (5–77) 25 (5–74) 36 (1–90) 35 (6–85) 32(3) Pt (pg g 1) 10 <10 (<10–26) <10 (<10–27) <10 (<10–80) <10 (<10–44) <10 (<10–164) <10 Ir (pg g 1) 1 9 (<1–26) 4 (<1–15) 5 (1–59) 15 (<1–41) 15 (<1–37) <1 observations can be made by comparing ICP-SFMS 190Os signals digestion provided that the time between sample preparation integrated over a 6 min acquisition period with GPI, available and analysis is the same within a few hours. from the experiments leading to the data summarized in Signals decrease with increasing time elapsed aer the Table 3: preparation of the digest. For a given sample type prepared by a given method, The highest signals are observed in HPA digests with little reproducibility of integrated signals for replicate digestions is (<15% RSD) diﬀerences between spiked acid blanks, plants and better than 5–10% RSD for HPA and 10–20% for MW-assisted biological samples.

This journal is ª The Royal Society of Chemistry 2013 J. Anal. At. Spectrom., 2013, 28, 1591–1599 | 1595 View Article Online JAAS Paper

Table 3 Os concentrations and isotope ratios for reference and control samples

Test sample Digestion, sample introduction Conc. (SD) pg g 1 187Os/188Os (SD)

NIST1577a MW, GPI, January 2013, n ¼ 3 0.23(0.05) 0.449(0.078) Bovine Liver MW, GPI, April 2013, n ¼ 3 0.25(0.05) 0.438(0.053) ERM BB186 MW, GPI, January 2013, n ¼ 3 0.47(0.06) 0.366(0.058) Pig Kidney MW, GPI, April 2013, n ¼ 3 0.45(0.07) 0.354(0.044) GBW 07601 MW, GPI, January 2013, n ¼ 3 0.61(0.06) 1.231(0.057) Human Hair MW, GPI, April 2013, n ¼ 3 0.65(0.04) 1.219(0.035) NIST1547 MW, GPI, January 2013, n ¼ 3 1.29(0.15) 0.954(0.041) Peach Leaves MW, GPI, April 2013, n ¼ 3 1.23(0.09) 0.945(0.034) HPI, GPI, n ¼ 3 1.69(0.03) 0.905(0.031) NIST1571 MW, GPI, n ¼ 3 1.46(0.06) 1.137(0.063) Orchard Leaves HPI, GPI, n ¼ 3 1.88(0.03) 1.104(0.051) TORT-1 MW, GPI, January 2013, n ¼ 3 7.76(0.21) 1.169(0.023) Lobster MW, GPI, April 2013, n ¼ 3 7.53(0.13) 1.183(0.025) Hepatopancreas HPI, GPI, n ¼ 3 9.16(0.18) 1.062(0.031) Moose Kidney MW, SN, n ¼ 12 5.9(2.0) Not available MW, GPI, n ¼ 12 5.29(0.39) 0.1345(0.0022) HPI, GPI, n ¼ 3 7.21(0.25) 0.1344(0.0018) a Os concentrations in CRM ERM BB184 Bovine Muscle, ERM BB422 Fish Muscle, GBW 07605 Tea and IAEA-A-13 Animal Blood were below MLOD of 0.2 pg g 1.

Signals in MW-assisted digests are signicantly lower than higher than the 115In intensity for the same conguration in HPA digests, by approximately 50% in biological matrices conrming that even while using SN, the dominant mechanism and 75% in plant matrices. of Os transport to the ICP is by volatile species released from the The signals for acid blanks prepared by MW-assisted aerosol and deposited droplets inside the spray chamber. digestion are approximately only 5% of those for HPA Though benecial for enhancing sensitivity, this process is preparations. unfortunately accompanied by pronounced memory effects These ndings indicate that up to 95% of Os spike can be with 15–20% of the signal from the preceding sample still lost during MW-assisted digestion. The process is fairly repro- persisting during the analysis of the next one. When no a priori ducible and continues even during storage of digests aer information on the expected Os content is available, the accu- preparation, but is inhibited by the sample matrix in animal racy of results for samples low in Os analyzed aer samples high and plant materials. The latter effect is probably due to partial in Os will suffer. Fortunately, the existence of a strong positive 18 reduction of OsO4 by reactions with organic materials, with correlation between Os concentrations in various vole organs/ differences observed between samples of plant and animal tissues described in the pilot study10 was conrmed early in the origin most likely being due to variable lipid content. Assuming course of this project, allowing sample solutions to be arranged that ID results obtained aer HPA preparation are the most in the order of ascending Os concentrations within each accurate ones, data underestimating the actual Os content by measurement sequence. 30% at most should be expected while using MW-assisted The formation of volatile Os species and hence severe carry- digestion in polypropylene vessels. However, relatively good over effects can be effectively inhibited by converting acidic reproducibility (Table 3), ability to use inexpensive disposable digests to alkaline conditions by dilution with ammonia + digestion vessels and more than 10-fold higher sample Triton X + ethylenediaminetetraacetic acid diammonium salt throughput compensate for the losses in accuracy. mixture,12,19 but this approach requires a dedicated preparation At the analysis stage, the 190Os/192Os ratio in the sample and analysis sequence in order to avoid interferences incurred digest required to calculate the Os concentration by ID can, in in the determination of other elements. Moreover, extra dilu- theory, be obtained directly from the same measurement tion combined with almost ve-fold lower instrumental Os sequence used for multi-element characterization, as both sensitivity in such a matrix signicantly affects the detection major Os isotopes are monitored. In spite of the limited number capabilities to the point when the approach is not practically of scans, this ratio can be measured in the range 0.5–10 with a viable for the purposes of this study. typical precision of better than 5% RSD for Os signals for the Though the SN introduction system was thoroughly cleaned most abundant isotope above 1000 counts per s, or better than between measurement sessions, the instrumental blank for 3% RSD for signals above 10 000 counts per s. The typical 192Os has increased from less than 10 counts per s before the instrumental sensitivity for Os, expressed as the sum of the start of the study to 50–100 counts per s as the study progressed, 187Os, 188Os, 190Os, 189Os and 192Os intensities to compensate indicating that building-in an instrument background rather for abundance differences between measured solutions, with than impurities of chemicals or contamination during prepa- SN and the conguration of the introduction system used, is ration limits detection capabilities. Consequently, the MLOD of 20 000 counts per s per ng per l. This is approximately four-fold 2pgg 1 (Table 2) is not better than in the previous study10 in

1596 | J. Anal. At. Spectrom., 2013, 28, 1591–1599 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper JAAS spite of using a more sensitive ICP-SFMS instrument. Conse- 187Os/188Os ratio measurements quently, the uncertainty of results for samples with Os ff 1 In order to di erentiate between potential sources of exposure, concentrations below 20–100 pg g (depending on the mass of 187 information on the abundance of the radiogenic Os isotope, organs or tissues available) is relatively high. Using the SN- 187 188 frequently expressed by the Os/ Os ratio, is needed as a based method for CRMs, Os was only detected in one (TORT-1) useful compliment to the Os concentration. Apart from blank making this approach insufficiently sensitive for the analysis of 187 correction, in the SN approach isobaric interference from Re biological samples unaffected by anthropogenic emissions. has to be corrected mathematically using the signal monitored  With the help of GPI speci cally optimized for gaseous OsO4 on mass 185 and the tabulated abundances of Re isotopes.20 production, the detection capabilities of ICP-SFMS for ultra- Then the contributions from 187Os and 188Os introduced with the trace and isotopic analyses can be greatly enhanced. For spike have to be corrected using the known isotope abundances example, 2 pg of 190Os spike produce an intensity maximum of of the spike solution and the signal monitored on mass 190 approximately 100 000 counts per s or an integrated intensity of 190 which in turn needs to be subtracted with the Os contribution over 400 000 counts for six acquisitions of 1 min duration. The from the sample. However, for samples with Os concentrations typical integrated intensity for a preparation blank is below 1 below approximately 500 pg g the magnitudes of such correc- 1000 counts. Using a single distillation unit with single MQ- tions prevent any meaningful isotopic information from being ff water rinse between samples, the memory e ect is below 2% of obtained. Therefore measurements of 187Os/188Os ratios in bio- the preceding sample, and as Os concentrations in digests are logical samples by SN should be reserved for samples with known from SN analyses, an optimized order of analysis makes elevated Os concentrations. As Os is efficiently separated from Re these effects easily manageable. 187 188 in the GPI approach,21 Os and Os signals obtained by this Though GPI generates transient signals of different patterns method, available from measurements used to calculate Os for solutions prepared by HPA or MW-assisted digestion (Fig. 1), concentrations by ID, need to be corrected only for contributions the intensity maximum occurs during the rst few minutes aer from the isotope spike and the preparation blank. For 2 pg spike sample introduction. This allows the total analysis time per addition such corrections are less than 10% of uncorrected – 11 – solution to be shortened from 30 60 min to 8 10 min. This signals for samples containing above 0.5 pg of Os with non- was achieved due to the application of GPI to already digested radiogenic isotopic composition. Combined with much higher sample solutions, as well as by using a higher initial H O 187 188 2 2 sensitivity this allows a reliable estimation of Os/ Os ratios content in the reaction vessel and a higher stripping gas ow 190 at much lower Os levels than possible using SN. This makes through the solution (Table 1). With 2 pg Os spike, a preci- 187 188 preparation of unspiked digest dedicated for Os/ Os isotope sion of 3–5% RSD can be obtained for 190Os/192Os ratios of 20– ratio measurements unnecessary. However, for samples with Os 30 and better than 2% RSD for ratios below 20 improving the Os content below 0.2 pg uncertainties of isotopic data may be MLOD to approximately 0.02 pg, corresponding to 0.2 pg g 1 for considerable. 100 mg sample weight. Taking into account losses of Os during MW-assisted digestion and storage of digests prior to analysis, and the fact that approximately 40% of digest in this study was Quality control consumed during SN measurements, even better MLODs can be The accuracy of Os concentrations and ratio data is difficult to reached using dedicated sample digestion by HPA, especially assess as, to the best of our knowledge, there are no commer- when the sample amount is not a limiting factor (quartz cially available reference materials of biological origin that are digestion vessels can accommodate up to 500 mg dry organic certied for these parameters. In order to assess repeatability material which is ve-fold more than that was used during this and immediate precision of the method prior to its application study). to vole samples, several CRMs of biological and plant origin were prepared and analyzed in triplicate. For a few of these materials, triplicate preparations and analyses were repeated at the end of the project. A summary of the experimental results for these materials alongside data for the in-house control sample, analyzed with each analytical batch of vole samples, is presented in Table 3. Only TORT-1 and moose kidney exhibited Os concentrations above the MLOD of the SN-based approach. Os in bovine muscle, sh and animal blood was undetectable even using GPI, while sub-pg g 1 Os concentrations in human hair, pig kidney and bovine liver indicate the typical concentration range that can be expected in biological matrices originating from relatively uncontaminated areas. Mean Os concentrations determined in two analytical Fig. 1 Normalized Os signal obtained with the GPI system, as a function of sessions using samples prepared by MW-assisted digestion reaction time for HPA (filled squares) and MW-assisted (filled diamonds) digests. The first minute of measurement is offset from the sample load by 2.5–3.5 min agree well with measurement uncertainty, demonstrating good purge delay time needed to ramp gas through the reaction vessel and to heat the reproducibility of the method. However, signicantly (by reaction mixture. approximately 20%) higher results were obtained using HPA,

This journal is ª The Royal Society of Chemistry 2013 J. Anal. At. Spectrom., 2013, 28, 1591–1599 | 1597 View Article Online JAAS Paper again suggesting that Os recovery using MW-assisted prepara- concentrations in all samples tested were above the MLOD for tion is incomplete. Both in-run and between-run precision of the SN-based approach and there are wide concentration ranges 187Os/188Os ratio measurements are better than 2–4% RSD for in all organs or tissues with the maximum to minimum content samples with Os concentrations above 1 pg g 1. The interme- ratios in excess of 100. Where ID Os data were available from diate precision of 187Os/188Os isotope abundance ratio both SN and GPI approaches, the agreement between concen- measurement using the GPI approach was better than 1.5% trations determined by these methods was usually better than RSD for the in-house control sample (moose kidney, Table 3). 5–10% RSD for concentrations above 100 pg g 1. When applied to a large number of vole tissues and organs, Measured Os concentrations in specimens from individual the throughput of the method is on average 120–140 samples animals exhibit strong correlations between diﬀerent body per week. This estimate includes sample preparation by MW- compartments as well as with many trace and ultra-trace assisted digestion, simultaneous multi-element characteriza- elements, with several typical examples being shown in Fig. 2. tion and Os concentration determination by ID using ICP-SFMS This information is particularly helpful for delineating exposure with SN and Os concentration determination and 187Os/188Os pathways, metabolism, accumulation and potential toxicity. At isotope ratio measurements for samples containing below Os concentrations above 200 pg g 1, 187Os/188Os isotope ratios 500 pg g 1. Though a detailed discussion of the vole results is in all samples fall into a narrow 0.125–0.15 range (Fig. 3) con- outside the scope of this study, several observations concerning rming emissions from a ferrochrome foundry as the primary the performance of the method can be mentioned here. Os exposure source in the area.9,10,15 At lower concentrations, a shi

Fig. 2 Correlations between Os concentrations in various organs and between Os and other elements.

1598 | J. Anal. At. Spectrom., 2013, 28, 1591–1599 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper JAAS

2 G. Ravizza and M. Bothner, Geochim. Cosmochim. Acta, 1996, 60, 2753–2763. 3 G. R. Helz, J. M. Adelson, C. V. Miller, J. C. Cornwell, J. M. Hill, M. Horan and R. J. Walker, Environ. Sci. Technol., 2000, 34, 2528–2534. 4 S. Rauch, H. F. Hemond and B. Peucker-Ehrenbrink, Environ. Sci. Technol., 2004, 38, 396–402. 5 B. K. Esser and K. K. Tureklan, Environ. Sci. Technol., 1993, 27, 2719–2724. 6 G. Williams, F. Marcantonio and K. K. Turekian, Earth Planet. Sci. Lett., 1997, 148, 341–347. 7 S. Rauch, H. F. Hemond and B. Peucker-Ehrenbrink, – 187 188 J. Environ. Monit., 2004, 6, 335 343. Fig. 3 Os/ Os ratios in vole tissues/organs as a function of Os concentra- – tion. Isotope ratio measurements by GPI (open circles) and SN (closed circles). 8 OSHA, 1993, 29 CFR Part 1926, Federal Register #58:35076 35306. 9 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontèr and towards more radiogenic isotope compositions is obvious, D. C. Baxter, Sci. Total Environ., 2007, 386, 145–158. ‘ ’ suggesting a contribution from natural , soil-borne Os. 10 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ D. C. Baxter, B. Hornfeldt,¨ E. Nyholm and F. Ecke, Water, Air, Soil Conclusions Pollut., 2011, 218, 603–610. 11 I. Rodushkin, E. Engstrom¨ and D. C. Baxter, Geostand. The optimized analytical protocol developed and validated in the Geoanal. Res., 2007, 31,27–38. course of this study allows reproducible, sensitive, relatively 12 D. Malinovsky, I. Rodushkin, D. C. Baxter and B. Ohlander,¨ rapid, cost-efficient and high-throughput determination of Os Anal. Chim. Acta, 2002, 463, 111–124. concentrations in animal matrices that may potentially be 13 I. Rodushkin, P. Nordlund, E. Engstrom¨ and D. C. Baxter, expanded also for plant materials. In investigations when no J. Anal. At. Spectrom., 2005, 20, 1250–1255. multi-elemental characterization of samples is necessary or for 14 J. W. Morgan and R. J. Walker, Anal. Chim. Acta, 1989, 222, samples with expected Os concentrations below 10 pg g 1,SN 291–300. analyses can be omitted, allowing the entire sample digest to be 15 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontér and used for GPI. For highly accurate Os quantication, digestion D. C. Baxter, Sci. Total Environ., 2007, 386, 159–168. using HPA is to be preferred. Detection capabilities of the method 16 E. Engstrom,¨ A. Stenberg, S. Senioukh, R. Edelbro, D. C. Baxter can be further improved for larger samples as 5–10-fold higher and I. Rodushkin, Anal. Chim. Acta, 2004, 521, 123–135. weight can be used in both types of digestions which should make 17 R. Walker, Anal. Chem., 1988, 60, 1231–1234. applications for samples far from anthropogenic sources 18 I. C. Smith, Environ. Health Perspect., 1974, 8, 201–213. possible. The method will be helpful in obtaining data for 19 T. Cho, I. Akabane, and Y. Murakami, in Plasma source mass improving current knowledge on anthropogenic Os exposure and spectrometry: the proceedings of the third surrey conference on metabolism as well as facilitating the wider use of the Os isotope plasma source mass spectrometry. ed. K. E. Jarvis, A. L. Gray, system in forensic applications (e.g. for food authentication).26,27 J. G. Williams, Royal Society of Chemistry, London, 1990, – Acknowledgements pp. 94 119. 20 J. R. D. E. de Laeter, H. Hidaka, H. S. Peiser, K. J. R. Rosman – ALS Scandinavia AB is gratefully acknowledged for technical and P. D. P. Taylor, Pure Appl. Chem., 2003, 75, 683 800. – support. MetTrans Initial Training Network (funded by the 21 K. Tagami and S. Uchida, Anal. Chim. Acta, 2000, 405, 227 European Union under the Seventh Framework Programme) is 229. acknowledged for nancial support. We wish to thank Fredrik 22 J. Birck, M. Barman and F. Capmas, Geostand. Newsl., 1997, – Lindgren for eld assistance and Dieke Sorlin¨ for help with the 21,19 27. preparation of the in-house control material. The research 23 L. Qi, J.-F. Gao, M.-F. Zhou and J. Hu, Geostand. Geoanal. leading to these results has received funding from the People Res., 2013, 37, DOI: 10.1111/j.1751-908x.2012.00211.x. – Programme (Marie Curie Actions) of the European Union's 24 S. B. Shirey and R. J. Walker, Anal. Chem., 1995, 67, 2136 Seventh Framework Programme FP7 2007–2013, under the REA 2141. – grant agreement no. 290336. The views expressed in this article 25 C. M. Brauns, Chem. Geol., 2001, 176, 379 384. are those of the author and may not necessarily reect those of 26 I. Rodushkin, T. Bergman, G. Douglas, E. Engstrom,¨ the European Union. D. Sorlin¨ and D. C. Baxter, Anal. Chim. Acta, 2007, 583, 310–318. References 27 M. Resano and F. Vanhaecke, in Isotopic Analysis, Wiley-VCH Verlag GmbH & Co. KGaA, 2012, pp. 391–418. 1 C. Chen, P. N. Sedwick and M. Sharma, Proc. Natl. Acad. Sci. 28 T. Meisel, J. Moser, N. Fellner, W. Wegscheider and U. S. A., 2009, 106, 7724–7728. R. Schoenberg, Analyst, 2001, 126, 322–328.

This journal is ª The Royal Society of Chemistry 2013 J. Anal. At. Spectrom., 2013, 28, 1591–1599 | 1599

Chromium isotope ratio measurements in environmental matrices by MC-ICP-MS

Simon Pontér, Nicola Pallavicini, Emma Engström, Douglas C. Baxter and Ilia Rodushkin

JAAS

View Article Online PAPER View Journal | View Issue

Chromium isotope ratio measurements in environmental matrices by MC-ICP-MS Cite this: J. Anal. At. Spectrom.,2016, 31, 1464 a bc bc c Simon Ponter,´ * Nicola Pallavicini, Emma Engstrom,¨ Douglas C. Baxter and Ilia Rodushkinbc

An analytical procedure consisting of high pressure/temperature acid digestion using an UltraCLAVE system and a one pass, single column matrix separation using DOWEX AG 1X8 anion exchange resin was applied to the determination of Cr concentrations and d53Cr in chromites, soils, and biological matrices (epiphytic lichens and mosses) using a combination of ICP-SFMS and MC-ICP-MS. The overall reproducibility of the method was assessed by replicate preparation and Cr isotope ratio measurements performed by different operators in multiple analytical sessions over a few months and was found to be 0.11& (2s). The accuracy was evaluated using commercially available reference materials for which measured data were compared with certified values (for Cr concentrations) and previously published results (for isotope Received 19th April 2016 data). The results demonstrate a uniform Cr isotope composition in soil depth profiles sampled in Accepted 9th May 2016 different urban environments. A strong negative correlation between d53Cr and Cr concentrations in DOI: 10.1039/c6ja00145a lichens and mosses indicates that airborne Cr from local anthropogenic source(s) is depleted in heavy www.rsc.org/jaas isotopes.

1. Introduction Hexavalent chromium, Cr(VI), being a known carcinogen as well as highly mobile and soluble under oxidized conditions in the 2 Isotopic measurements have emerged as essential tools in form of anions CrO4 and HCrO4 is of particular interest. environmental sciences, providing valuable information on Trivalent chromium, Cr(III), on the other hand, is innocuous contamination sources and pathways, quantication of element and very inert. Previous studies on reduction mechanisms have transport and mixing, directions and rates of environmental shown that Cr isotope fractionation occurs during the reduction processes, etc.1 Continuous improvements in measurement of Cr(VI) to Cr(III), with lighter isotopes becoming enriched in the precision and rened separation methods have widened the product.1,9,10 Fractionation has also been proposed to take place eld continuously over the last few decades adding heavier, during the oxidation of Cr(III) to Cr(VI).11 Therefore, there is stable isotopes to the isotope ‘toolbox’.1 However, broader potential for tracing specic Cr sources and transformations, application of isotope signatures remains limited by high such as those occurring in chrome-plating processes or in – instrumentation cost and by lengthy, labor-intensive separation mining and ore renement industries.8,10 13 schemes. Since targeted variations in the Cr isotope composition are Chromium is widely dispersed in the earth's crust as a trace expected to be minor, the separation of Cr from a sample matrix element in the 200 mgg 1 range and around 0.3 mgL 1 in the is crucial for accurate isotope ratio determination. As chromium oceans,2 and occurs as four stable isotopes 50Cr, 52Cr, 53Cr, and, occurs as either cationic Cr(III) or anionic Cr(VI) species, the 54Cr with representative relative abundances of 4.345%, behavior of each during ion exchange separation is very diﬀerent. 83.789%, 9.501%, and, 2.365%, respectively.3 Earlier studies This dichotomy may be resolved either by converting all Cr to relying on Cr isotope ratios have been focused on identifying a chosen valence state and adapting the separation accordingly, – processes during formation of the solar system.4 7 Due to or by separating Cr(III) and Cr(VI) individually and later merging instrument development during the last few decades, Cr isotope them prior to making measurements. Both approaches have their measurements in samples with lower concentrations are now advantages and disadvantages. A number of procedures to isolate possible with recent studies focusing on the minor variations Cr have been proposed, which are oen very time consuming and – in abundances occurring in environmental matrices.8–10 elaborate, including several purication steps.3 16 Many published procedures use a double spike technique assuming that isotopically enriched Cr and naturally occurring aDepartment of Environmental Science and Analytical Chemistry, Stockholm Cr behave identically (complete sample/spike equilibration), University, SE-11418 Stockholm, Sweden. E-mail: [email protected] and therefore isotope fractionation caused by incomplete ana- b ˚ ˚ Division of Geosciences, Lulea University of Technology, S-971 87 Lulea, Sweden lyte recovery during the matrix separation process as well as cALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Lulea,˚ Sweden

1464 | J. Anal. At. Spectrom.,2016,31, 1464–1471 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS instrumental mass bias are corrected for and non-quantitative The MC-ICP-MS instrument used for Cr isotope ratio recovery can be tolerated.17 However, the approach requires measurements was a Neptune PLUS (Thermo Fisher Scientic, isotope ratio measurements using at least three interference- Bremen, Germany). The instrument was equipped with a plat- free Cr isotopes, which can be a challenge, especially for multi- inum guard electrode and nine Faraday cups (eight movable and collector inductively coupled plasma mass spectrometry one xedcentercup)andwasoperatedinhighresolutionmode (MC-ICP-MS). (providing a pseudo-resolution of 8500M/DM derived from the In this study, Cr was separated from a wide range of envi- peak slope of the rising edge measured at 5% and 95% relative ronmental matrices (Cr-bearing minerals, soils, lichens and peak height). The instrument can be coupled to three different mosses) from Northern Sweden using a single column proce- introduction system congurations: ‘standard’ consisting of dure modied from previously published work.15,16 Since double a cyclonic/Scott spray chamber, as well as Aridus II (Teledyne spiking was not used, quantitative recovery of analytes during CETAC technologies, Omaha, Ne, USA), and Apex (Elemental separation was of paramount importance to prevent articially Scientic, Omaha, Ne, USA) high efficiency desolvating nebulizers. introduced fractionation. Special care was taken to minimize all Instrumental operating conditions and measurement sources of spectral interferences and matrix effects. Cr isotope parameters are summarized in Table 1. ratios in puried fractions were measured by MC-ICP-MS using Due to the instrument's hardware limitations of cup posi- a combination of internal standardization and bracketing tioning limits for a given mass range, 52Cr, 53Cr, and 54Cr as well standards for instrumental mass bias correction. as 56Fe were measured simultaneously with the main cup conguration, while 60Ni, 61Ni and 62Ni were measured at another magnet setting (Table 1). The use of the so-called 2. Experimental dynamic mode certainly is less time effective than true simul- 2.1. Chemicals and reagents taneous measurements, but it allows for on-line correction of dri and matrix-induced changes in mass bias. Nitric acid (HNO3), hydrochloric acid (HCl), hydrogen peroxide Aer plasma ignition, the instrument was allowed (H2O2) (all from Sigma-Aldrich Chemie GmbH, Munich Ger- many) and hydrogen uoride (HF, 48% Merck, Darmstadt, a minimum of one hour to stabilize while aspirating 0.14 M Germany) were all of analytical grade. MilliQ water (Millipore, HNO3 blank solution prior to performing the daily optimization  Bedford, MA, USA) used in this study was puried by reverse of operational parameters (gas ow, torch position and lens osmosis followed by ion exchange. Anion exchange resin, settings) and mass calibration. The best signal stability and DOWEX AG 1X8 dry mesh size 200–400 mm (Sigma-Aldrich isotope ratio in-run precision were obtained by self-aspiration, Chemie GmbH), was suspended in MilliQ water and loaded into but a peristaltic pump was used since this decreased sample 2 mL columns. The oxidizing agent, ammonium perox- uptake and wash-out times considerably. All measurement solutions were always diluted to equal concentrations, odisulfate, (NH4)2S2O8 (Riedel-de Haën AG, Seelze-Hannover), was weighed and dissolved in MilliQ water to the desired providing concentration matching between samples and d concentration on the day of use. bracketing -zero standards for each individual run. With Chromium NBS SRM 979 was used as the d-zero standard. a standard introduction system, the Cr concentrations in 1 Chromium stock solution (Ultra Scientic Analytical Solutions, samples were adjusted to 2 mg L in 0.14 M HNO3 (providing 52 Lot. T00606, 10 000 mg L 1) was used as the quality control Cr intensity of approximately 40 V) and spiked with Ni to 1 mg 1 sample (QCS). Nickel stock solution (Ultra Scientic Analytical L . For samples with low Cr concentrations, an Aridus des- 40 12 + Solutions, Lot. M00866, 1000 mg L 1) was used to prepare the olvating nebulizer was tested initially, but severe Ar C 52 internal standard for MC-ICP-MS measurements. interference on Cr, presumably originating from organics released by the membrane material, made further application impossible. Consequently, the Aridus was replaced by an Apex 2.2. Instrumentation desolvating nebulizer. For solutions analyzed with this more Sample digestion was performed by using an UltraCLAVE sensitive introduction system, the Cr concentration was system (Milestone, Sorisole, Italy) that offers efficient micro- adjusted to 0.4 mg L 1 and spiked with Ni to 0.2 mg L 1. wave (MW) assisted digestion combining high temperature and All solutions were analyzed in duplicate giving a total high pressure conditions with a relatively high throughput. measurement time per sample of approximately 12 minutes All element concentrations in sample digests and column including uptake and washout. Duplicate measurements allow fractions were measured by using a single-collector, double- the detection of memory effects in the introduction system, focusing, inductively coupled plasma sector-eld mass spec- which proved to be non-existent. Outlier elimination was acti- trometer (ICP-SFMS, ELEMENT XR, Thermo Fisher Scientic, vated, using the 2s criterion in the resident Neptune soware. Bremen, Germany). It was equipped with a demountable quartz Signal intensities were transferred to commercially available torch, a nickel sampler cone, a high sensitivity X-skimmer cone, spreadsheet soware for further off-line calculations. The a PFA spray chamber and a SD2 auto-sampler (ESI, Perkin- isobaric interference from 54Fe on 54Cr was corrected mathe- Elmer, Santa Clara, USA) equipped with a six-port valve and matically using the monitored 56Fe signal together with tabu- 1.5 mL sample loop lled and rinsed utilizing vacuum suction. lated Fe isotope abundances and computed instrumental mass Details about instrumental parameters and measurement bias for each individual Neptune session. Instrumental mass conditions can be found elsewhere.18 bias was corrected off-line by using the revised exponential

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 1464–1471 | 1465 View Article Online

JAAS Paper

Table 1 MC-ICP-MS parameters for Cr isotope ratio measurements

Rf power (W) 1400 Ion lens settings Optimized daily for maximum sensitivity and signal stability Zoom optic parameters Adjusted daily for the highest sensitivity and peak shape Coolant gas (L min 1)16 Sample gas (L min 1) 0.9–1.25 Auxiliary gas (L min 1) 0.8 Sample uptake rate (mL min 1) 0.05–0.25 Integration time, s 1.262 Number of integrations 3 Number of blocks 9 Cycles per block 5 Amplier rotation Le

Cup conguration

Main L3 L2 L1 H1 H2 H3 52Cr 53Cr 54Cr 56Fe (54Fe) Sub-cong. 60Ni 61Ni 62Ni

correction model by Baxter et al.19 using the internal standard steelworks. The area has been studied previously and surveys (nickel ratio) and the Cr d-values were calculated against focusing on heavy metal deposition patterns on the Finnish bracketing d-zero solution. Three samples were analyzed side of the border have been carried out for more than two between two standards, together forming a block (block: stan- decades covering many elements including Cr using Hyloco- dard 1 – sample 1 – sample 2 – sample 3 – standard 2). The mean mium splendens and Pleurozium schreberi mosses.20 The distri- value of the two consequent measurements of the sample ratio bution of Cr in this area and potential inuence from was calculated against ratios for standards in each block. anthropogenic sources have been mentioned previously in the Assuming a linear change in mass bias, ratios for samples 1 and study devoted to the use of the Os isotope composition for 3 were calculated relative to those for standards 1 and 3, anthropogenic assessment.21 respectively, while sample 2 was calculated against the mean Epiphytic lichen (Usnea spp. and Bryoria spp.) and moss ratio for both standards. Results from the two measurements samples (Pleurozium schreberi) were collected from sampling were used to calculate mean d-values and s for each sample. sites at diﬀerent distances from Torne˚a steelworks during 2005 Chromium d-values were calculated using the following formula and 2006. Exact sampling locations can be seen in the study by " # 21 ðxCr=yCrÞ Rodushkin et al. In order to increase representativeness, dx=yCr ¼ sample 1 1000 pooled samples consisting of >20 individual lichens or mosses ðxCr=yCrÞ NBS 979 were collected from several trees at each sampling location where xCr and yCr represent two diﬀerent isotopes of interest, wearing powder-free gloves, dried at 50 C and stored in zip-lock (xCr/yCr) is the measured ratio of sample solution and plastic bags marked with the location and sampling date. sample  x y d Six soil pro les, approximately 60 cm deep, divided vertically ( Cr/ Cr)NBS 979 is the bracketing standard selected as -zero. The factor 1000 is used to convert ratios to per mil notation. into four sub-samples each, were collected in the city and ˚ 0 00 0 00 When the d-value refers to a ratio of a heavier isotope compo- suburbs of Lulea (65 34 57 N228 47 E, Northern Sweden). ˚ sition, a positive d-value corresponds to an enrichment in the The area surrounding Lulea is heavily industrialized, with SSAB heavier isotope compared to the standard. steelworks as the predominant local industry. The soil consists mainly of clay and silt loam overlying 1.9 Ga granitic bedrock with minor meta-sedimentary constituents.22 2.3. Samples and sampling area During initial stages of method development and testing, three certied reference materials were used, NIST 1547 Peach The study region in northeast Sweden, stretching from the leaves, NIST 2701 Hexavalent Chromium in Contaminated Soil Swedish/Finnish border at the northeast of the sampling tran- (both from the National Institute of Standards and Technology, sect to the town of Lule˚a at the southwest, is sparsely populated Gaithersburg, MD, USA) and basalt JB-1 (Geological Survey of but heavily industrialized. Japan, Tokyo, Japan), as well as one contaminated soil sample Major industries include metal foundries processing chro- obtained as part of a Swedish environmental monitoring mium, iron, copper, lead, and zinc ores as well as paper mills. program. The latter was used as an in-house control and was Of major interest for the present study are the mining of prepared and analyzed repeatedly as part of all sample batches chromium ores from the open pit mine in Kemi on the Finnish with recoveries ranging from 80%–101% (Table 2). side of the border and the processing of the ore in Torne˚a

1466 | J. Anal. At. Spectrom.,2016,31, 1464–1471 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

Table 2 Cr concentrations and d53Cr in reference and control samplesa

Test sample Preparation Cr, mgg 1, found s (certied s) d53Cr (2s), &

QCS T00606 Dilution, n ¼ 28 (10 000) 0.396 (0.072) Control soil Digestion/separation, n ¼ 18 425 21 0.351 (0.110) NIST 1547 Digestion/separation, n ¼ 3 0.80 0.04 (1*)NA** NIST SRM 2701 Digestion/separation, n ¼ 2 42 300 2700 (42 600 1200) 0.091 (0.044) NIST SRM 2701 Digestion/dilution, n ¼ 2 0.127 (0.066) JB-1 Digestion/separation, n ¼ 3 415 13 (417) 0.091 (0.052) a *Information values. **Too low Cr content for isotope ratio measurements.

Chromites (black and light-greenish varieties) originating oxyanions where most other elements form cationic or neutral from the open mine pit in Kemi were obtained from the complexes that pass through the column. However, aer geological collection of Lule˚a University of Technology. sample digestion the majority Cr is present as Cr(III) and an efficient oxidation step is mandatory to quantitatively convert 2.4. Sample digestion Cr(III) to Cr(VI). To achieve this, 1 mL 0.2 M (NH4)2S2O8 with 0.07 mL 6.7 M NH was added to 3 mL 0.2 M HCl sample 0.05–0.5 g (limited by the recommended maximum sample 3 solution. Teon vessels were tightly sealed and le on a hot amount for the digestion vessel used) dried material from each plate for a minimum of 2 h at 160 C for sufficient oxidation. sample location was weighed into a 12 mL Teon vial. Then Columns loaded with 2 mL DOWEX AG 1X8 anion resin 5 mL 14 M HNO was added and aer one hour of initial 3 were sequentially cleaned using 2 mL 14 M HNO and 10 mL oxidation, samples were stirred to ensure complete mixing, 3 5 M HNO and rinsed with 15 mL MilliQ following by resin before another 1 mL HNO and 0.05–0.2 mL HF were added to 3 3 preconditioning with 10 mL 0.2 M HCl. Aer cooling to room wash down samples adhering to vial walls. Teon vials were temperature, sample solutions were loaded onto the anion placed into a carousel which was inserted into a Teon-coated exchange column. The sample matrix was eluted using 20 mL UltraCLAVE reaction chamber lled with de-ionized water and 0.2 M HCl followed by 10 mL MilliQ water. Aer the matrix H O (10 : 1 v/v). The chamber was pressurized and the pre- 2 2 wash, 2 mL 5 M HNO was added to the column and the ow programmed digestion cycle (30 min ramp to 230 C followed by 3 was arrested by plugging the column to allow reduction of Cr(VI) 20 min hold at temperature and pressure) was initiated. Pro- to Cr(III) which no longer adheres to the resin. Aer at least cessing time, including cooling time and transfer to evapora- 2 hours, the column was opened and Cr was eluted by tion vessels, was approximately 2.5 h per digestion batch. sequentially adding another 28 mL 5 M HNO . To ensure Sample digests were transferred into 6 mL Teon screw-cup 3 quantitative elution of Cr, 2 mL 14 M HNO followed by 8 mL vials (Savillex, Minnetonka, Minnesota, USA) and dried down on 3 5 M HNO were added in a nal separate fraction; see Fig. 1 for a ceramic-top hot plate. Then 2 mL of 12 M HCl were added to 3 the full elution scheme. the residue and evaporated again to remove an excess of uo- The separation effectiveness was evaluated by multi- rine. The latter procedure was repeated twice followed by residue elemental analysis (70 elements measured) of all fractions by dissolution in 3 mL 0.2 M HCl for direct Cr separation on anion ICP-SFMS. This provides (I) direct assessment of Cr recovery, (II) exchange columns. Small (50 mL) aliquots were taken for pre- information on separation efficiency from matrix elements, (III) analysis of sample elemental compositions by ICP-SFMS. information on Cr concentration, needed for preparation of concentration- and acid strength matched solutions for isotope 2.5. Matrix separation ratio measurements and (IV) information on the potential An anion exchange chromatography procedure was adopted presence of spectrally interfering elements and isobars either and modied from previously published methods.15,16 Cr(VI)in from the sample matrix or from contamination during sample a weak hydrochloric acid matrix forms negatively charged preparation.

Fig. 1 Flow chart of the Cr separation method.

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 1464–1471 | 1467 View Article Online

JAAS Paper

All puried fractions containing more than 1% of total Cr were >93% was achieved in the rst 20 mL of the elution fraction with pooled in a 40 mL Teon beaker and evaporated to dryness on the 2–5% lost during sample loading and matrix wash steps hot plate at 160 C, effectively removing residual chloride from (Fig. 2b). Samples with recoveries below a 90% threshold were the matrix wash stages. An aliquot of 14 M HNO3 was pipetted re-prepared and re-analyzed or results for such samples were directly onto the residue, le to react for 10–15 min and then excluded from evaluations. Addition of NH3 proved crucial for diluted to the desired concentration for analysis on the Neptune. quantitative recovery. As evident from initial test separations Handling of samples and solutions was performed in clean during method development, Cr recoveries were signicantly laboratory facilities (class 10000) by personnel wearing clean lower and resulted in positively biased d53Cr when attempting to room gear, following general precautions to reduce contami- oxidize Cr in the absence of NH3. nation.23 All plastic labware was acid cleaned and rinsed with When the negatively charged Cr-oxyanion in the form of MilliQ-water before use. HCrO4 attaches to the strong-base anion exchange resin, the majority of major matrix elements (and Ni) pass through the column during sample loading and matrix washing, while 3. Results and discussion <0.1% of the original content co-elutes with Cr. It should be 3.1. Blank contribution stressed though that even 0.1% of the original Fe in a typical soil  ff The average Cr method blank for the entire procedure, assessed sample present in the puri ed Cr fraction may a ect accurate 54 by applying all preparation and separation steps to a set (n ¼ 14) measurements of ratios involving the Cr isotope. No complete of digestion blanks handled as samples, was 17 5 ng. This Cr separation from V and Ti can be achieved by the single- corresponds to a Cr contribution of less than 3% for samples column separation (Fig. 2b), with up to 8% of the original containing the minimum analyte content in this study. Re-use content still present in Cr fractions, preventing the use of the 50 of columns gives signicantly higher blanks (59 29 ng, n ¼ 9). Cr isotope.  As a result of this ‘column memory’, the use of new columns is In approximately 20% of samples, Cr fractions were signi - recommended for samples with low Cr concentrations. cantly contaminated by S demonstrating incomplete and variable separation of the latter. Sulfur is added to the oxidizing agent at concentrations of up to a thousand times higher than ffi 3.2. Separation e ciency that of Cr. In the majority of separations approximately 98% of S Typical elution proles for major matrix elements, Cr, Ni and is removed during sample loading and matrix washing, but for potentially interfering elements, are shown in Fig. 2. In the unknown reasons in some samples up to 20% of the S co-elutes majority of samples tested in the present study, Cr recovery with Cr.

Fig. 2 Elemental recoveries in diﬀerent column fractions.

1468 | J. Anal. At. Spectrom.,2016,31, 1464–1471 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

3.3. Interferences including both QCS solution (no separation) and separations of in-house soil control materials in all measurement sessions Apart from V and Fe isobars, Cr isotopes are also subject to providing 2s of 0.07& (n ¼ 28) for the former and 0.11& numerous polyatomic interferences from oxides, hydrides, and (n ¼ 16) for the latter. Since measurements of in-house soil carbon- and nitrogen-based sources.24 However, the latter replicate digestions, separations and analyses were performed interferences are unlikely to be present in the plasma while during numerous analytical sessions conducted by various analyzing puried Cr fractions or can be separated in high operators, this reproducibility estimate is considered to provide resolution mode. For example, spiking Cr standards with increasing concentrations of HCl (up to 0.04 M) did not affect Cr a valid assessment of the entire method. & ff d53 ratios conrming efficient mass separation of ClO(H)+ An almost 0.4 di erence in Cr between our QCS solution and NBS CRM 979 (Table 2) contradicts the presumption of interferences. Ellis et al.11 that Cr isotopes are not expected to be fractionated The effect of residual contaminant elements in the puried during industrial purication, and thus most or all supplies of Cr fraction on isotope measurements was evaluated by spiking industrial Cr should have values close to 0&. Similarly, standard solutions with increasing concentrations of Ti, V, Fe Schoenberg et al.16 observed an isotopic composition of d53Cr ¼ and S. The highest ratios of contaminant element to Cr were one 0.39& for a Merck Cr(III) standard solution. These ndings for Ti and V, ve for Fe and 250 for S. Ti impurities in the Cr  d fraction prevent the use of the 50Cr isotope as well as interfere serve to con rm the importance of having common -zero with the 62Ni isotope used for instrumental mass-bias correc- standards for all research groups studying Cr isotope 46 16 + fractionation. tion, because of interference from Ti O , although the 61 60 Verication of accuracy for Cr concentration measurements unaffected Ni/ Ni ratio can be used instead. Mathematical 54 54 by ICP-SFMS was accomplished by analysis of CRMs (Table 2). correction for the Fe isobar on Cr fails at Fe/Cr concentra- No matrix reference material with a certied Cr isotopic tion ratios of >0.1, which would efficiently prevent the use of composition exists, making comparison between the obtained this isotope in the majority of samples. Moreover, the efficiency d-values and previously published data the only available option of the mathematical correction can be further affected rstly by d53 & & uncertainty in actual instrumental mass bias deduced from to evaluate accuracy. The Cr for JB-1 ( 0.091 0.052 )is measured and tabulated Ni (or Cr) ratios and secondly by the in acceptable agreement with previously published results by Ellis et al.11 (0.04&, with no stated uncertainty) and Schoen- risk of Fe fractionation (either in the original sample or intro- berg et al.16 (0.178& 0.048&) given the aforementioned duced during column separation). At a S/Cr concentration ratio 53 precision estimate. It should be stressed that these literature of 250, d Cr in the spiked standard was slightly positively values were obtained using very different separation schemes biased (0.11& 0.03&). For the rest of the spiked standards, 53 and measurement techniques. Due to the high Cr concentration d Cr was statistically indistinguishable from zero within in NIST 2701 and the chromites, it was possible to measure measurement uncertainty conrming negligible levels of spec- d53  tral interferences and matrix effects using the proposed Cr a er simple dilution of digests and to compare data thus  method. obtained with those a er column separation. No statistically

3.4. Throughput Using the UltraCLAVE, up to 40 digestion vessels may be accommodated in the reaction chamber. This limits the batch size to 36 samples along with two method blanks and two reference materials. The entire procedure from sample weighing, digestion, evaporation of digests, separation on anion exchange columns, evaporation of puried fractions, ICP-SFMS concentration determinations, isotope ratio measurements by MC-ICP-MS to data evaluation can be done by one chemist in approximately three days, assuming that evaporations can be done overnight.

3.5. Precision and accuracy In-run instrumental repeatability in d53Cr can be estimated as twice the standard deviation (2s) of results from duplicate consecutive measurements of each sample and as a rule is better than 0.04& or 0.09& for measurements performed with the standard introduction system or Apex, respectively. Repro- ducibility provides a more realistic assessment of the developed method's ability to detect minor variations in isotope compo- Fig. 3 Cr concentrations and d53Cr as a function of soil depth. The sitions than repeatability.25 Reproducibility was assessed by arrow represents the typical reproducibility of isotope data.

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 1464–1471 | 1469 View Article Online

JAAS Paper signicant differences were found, effectively conrming the heavier (d53Cr approximately +0.4&) than that in local soils absence of column-induced fractionation. indicating different isotope compositions in airborne sources. Recent estimates show that all surface waters are enriched in 3.6. Cr in soil proles heavy Cr isotopes27 suggesting that wet deposition may explain d53 ˚ d53  the positive Cr values in lichens from Lulea. Cr concentrations and Cr in 6 soil pro les as a function of ˚ – m At approximately 30 km distance from the Tornea steel- depth are shown in Fig. 3. Concentrations vary in the (10 35) g 53 works, as Cr in both lichen and moss samples increases, d Cr g 1 range with a notable trend towards lower levels in topsoil. decreases to approximately 0.2&. Bio-indicators with the Similar patterns are also observed for elements such as Al and highest Cr content, sampled at approximately 2 km from the Ti and are caused by differences in organic matter content, the steelworks, have mainly negative d53Cr values and a high scatter latter containing less Cr than more predominantly inorganic of isotope compositions (Fig. 4). Average d53Cr for Kemi chro- sections. mites is 0.08&, which is identical to the mean value reported There is very little variation in the Cr isotopic composition for chromites by Farkaˇs et al.28 Therefore, ne Cr-rich dust from between different locations and depths (Fig. 3b), with the chromite mining and transport activities, unlikely to induce Cr majority of results being within the 0.08–0.14& range. It might isotope fractionation, can hardly be the major source of local appear that topsoils are slightly enriched in heavier Cr isotopes airborne Cr pollution. It seems that smelting and rening but given a typical method with a reproducibility of 0.11& this processes result in predominant airborne release of lighter Cr tendency is not signicant. No correlation was observed isotopes as reected in the Cr isotopic composition found in between Cr concentrations and d53Cr (R2 ¼ 0.2). lichens and mosses. It must also be mentioned that d53Cr values No notable differences were observed for proles taken from for Torne˚a lichens and mosses are reminiscent of dust, the city and from suburbs. assumed by Bonnand et al.27 to have the same composition as It should be noted that the in-house control soil sample used the continental crust (0.13&). for method development and performance testing has a significantly heavier isotope composition (0.35&).

3.7. Cr in lichens and mosses 4. Conclusion Concentrations of Cr in lichens increase from approximately 1.5 Digestion using the UltraCLAVE reaction chamber provides 1 1 mgg in the vicinity of Lule˚a to >30 mgg near the border complete oxidation of the organic material as well as digestion between Sweden and Finland. Concentrations in moss samples of chromite samples, while a one pass column separation allows 1 follow a similar pattern, increasing from approximately 4 mgg eﬃcient removal of matrix elements and shows consistently 1 in Lule˚a to almost 50 mgg when approaching Torne˚a steel- high Cr recoveries. However, Ti, V and Fe are still present at works. These data reaﬃrm previously published ndings on the trace levels, hindering accurate measurements of 50Cr and 54Cr anthropogenic impact of chromite mining and stainless steel isotopes. Incorporation of the present Cr separation procedure production in the area.19 The higher degree of Cr accumulation into multi-elemental purication schemes shows interesting in mosses than that in lichens also substantiates data acquired potential. For example, Cr is quantitatively recovered in the using other species in a Nigerian study.26 sample loading and matrix wash steps during a column sepa- 2 Interestingly, there is a strong (R ¼ 0.7) correlation between ration designed for B, Cu, Cd, Fe, Pb, Sr, Tl and Zn separation.29 53 d Cr in lichens or mosses and inverse Cr concentrations, as Therefore, using these fractions for Cr separation would shown in Fig. 4. In lichen samples from Lule˚a, Cr is signicantly signicantly reduce the amount of Fe present and enable the use of 54Cr. Although the Apex desolvation system allows isotope ratio measurements in samples with lower Cr concentrations and reduces oxide formation, in-run precision is inferior while signal stabilization and wash-out times are longer than those with the standard conguration. Therefore, the latter is to be preferred when the Cr concentration is not a limiting factor. The analytical protocol presented has proved to be suitable for precise 53Cr/52Cr isotope ratio measurements in various environmental matrices although the absence of commercially available matrix-matched reference materials with certied isotope ratios for Cr hampers the assessment of true method accuracy and emphasizes the need for such products to be developed and validated. The rst ever Cr isotope data obtained for lichens and mosses indicate the potential of using this approach for tracing Fig. 4 d53Cr in chromites and environmental samples as a function of and quantifying airborne Cr pollution caused by stainless steel inverse Cr concentration, error bars are 2s. foundries.

1470 | J. Anal. At. Spectrom.,2016,31, 1464–1471 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

Acknowledgements 11 A. S. Ellis, T. M. Johnson and T. D. Bullen, Science, 2002, 295(5562), 2060–2062. We acknowledge ALS Scandinavia AB for technical and nancial 12 J. W. Ball and J. A. Izbicki, Appl. Geochem., 2004, 19, 1123– support of this project. The Neptune was purchased through 1135. a grant from Kempestielsen to the Centre for Isotopic and 13 J. A. Izbicki, J. W. Ball, T. D. Bullen and S. J. Sutley, Appl. Trace Element Measurements, Lule˚a University of Technology. Geochem., 2008, 23(5), 1325–1352. We also wish to thank Katerina Rodiouchkina for help with 14 A. Makishima, K. Kobayashi and E. Nakamura, Geostand. sampling and sample preparation. The authors also wish to Newsl., 2013, 26,41–51. express their gratitude to the Stable Isotopes Laboratory of the 15 J. W. Ball and R. L. Bassett, Chem. Geol., 2000, 168(1–2), 123– Department of Earth Sciences of the University of Oxford for 134. providing NBS 979 reference solution. We also thank Bjorn¨ 16 R. Schoenberg, S. Zink, M. Staubwasser and F. von Ohlander,¨ Johan Ingri, Dimitry Malinovsky, and Lennart Blanckenburg, Chem. Geol., 2008, 249(3–4), 294–306. Widenfalk for help in obtaining geological samples. 17 A. D. Anbar, J. E. Roe, J. Barling and K. H. Nealson, Science, 2000, 288(5463), 126–128. 18 A. Stenberg, D. Malinovsky, I. Rodushkin, H. Andrén, References C. Pontér, B. Ohlander¨ and D. C. Baxter, J. Anal. At. Spectrom., 2003, 18(1), 23–28. 1 T. D. Bullen, in Handbook of Environmental Isotope 19 D. C. Baxter, I. Rodushkin, E. Engstrom¨ and D. Malinovsky, J. Geochemistry, ed. M. Baskaran, Springer, Berlin, Anal. At. Spectrom., 2006, 21(4), 427. Heidelberg, 2012, pp. 162–220. 20 J. Poikolainen, E. Kubin, J. Piispanen and J. Karhu, Sci. Total 2 K. S. Smith and H. L. O. Huyck, The environmental Environ., 2004, 318(1–3), 171–185. geochemistry of mineral deposits, 1999, vol. 6. 21 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontér and 3 J. R. De Laeter, J. K. Bohlke,¨ P. De Bievre, H. Hidaka, D. C. Baxter, Sci. Total Environ., 2007, 386(1–3), 159–168. H. S. Perser, K. J. R. Rosman and P. D. P. Taylor, Pure Appl. 22 I. Rodushkin, F. Odman¨ and H. Holmstrom,¨ Sci. Total Chem., 2011, 83, 397–410. Environ., 1999, 231,53–65. 4 A. Trinquier, J. L. Birck, C. J. Allègre, C. Gopel¨ and D. Uleck, 23 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontér and Geochim. Cosmochim. Acta, 2008, 72, 5146–5163. D. C. Baxter, Anal. Bioanal. Chem., 2010, 396, 365–377. 5 A. Shukolyukov and G. W. Lugmair, Geochim. Cosmochim. 24 T. W. May and R. H. Wiedmeyer, Anal. At. Spectrosc., 1998, Acta, 2004, 68, 2875–2888. 19(5), 150–155. 6 L. Qin, C. M. O. D. Alexander, R. W. Carlson, M. F. Horan and 25 N. Pallavicini, E. Engstrom,¨ D. C. Baxter, B. Ohlander,¨ J. Ingri T. Yokoyama, Geochim. Cosmochim. Acta, 2010, 74(3), 1122– and I. Rodushkin, J. Anal. At. Spectrom., 2014, 29(9), 1570– 1145. 1584. 7 M. Rotaru, J. L. Birck and J. C. Allègre, Nature, 1992, 26 A. E. Ite, I. I. Udousoro and U. J. Ibok, Am. J. Environ. Prot., 358(6386), 465–470. 2014, 2,22–31. 8 C. Wanner, U. Eggenberger, D. Kurz, S. Zink and U. Mäder, 27 P. Bonnand, R. H. James, I. J. Parkinson, D. P. Connelly and Appl. Geochem., 2012, 27, 644–654. I. J. Fairchild, Earth Planet. Sci. Lett., 2013, 382,10–20. 9 S. Zink, R. Schoenberg and M. Staubwasser, Geochim. 28 J. Farkaˇs, V. Chrastný, M. Novák, E. Cadkova,ˇ J. Paˇsava, Cosmochim. Acta, 2010, 74(20), 5729–5745. R. Chakrabarti and T. D. Bullen, Geochim. Cosmochim. 10 M. Novak, V. Chrastny, E. Cadkova, J. Farkas, T. D. Bullen, Acta, 2013, 123,74–92. J. Tylcer, Z. Szurmanova, M. Cron, E. Perchova, J. Curik, 29 I. Rodushkin, N. Pallavicini, E. Engstrom,¨ D. Sorlin,¨ M. Stepanova, J. Pasava, L. Erbanova, M. Houskova, B. Ohlander,¨ J. Ingri and D. C. Baxter, J. Anal. At. Spectrom., K. Puncochar, L. A. Hellerich and R. Hill, Environ. Sci. 2015, 31(1), 220–233. Technol., 2014, 48(11), 6089–6096.

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 1464–1471 | 1471 III

Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS

Nicola Pallavicini, Emma Engström, Douglas C. Baxter, Björn Öhlander, Johan Ingri and Ilia Rodushkin

Volume 29 Number 9 September 2014 Pages 1507–1728 JAAS

Journal of Analytical Atomic Spectrometry www.rsc.org/jaas

ISSN 0267-9477

HOT PAPER Nicola Pallavicini et al. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS

JAAS

View Article Online PAPER View Journal | View Issue

Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS Cite this: J. Anal. At. Spectrom.,2014, 29,1570 ab ab b a Nicola Pallavicini,* Emma Engstrom,¨ Douglas C. Baxter, Bjorn¨ Ohlander,¨ Johan Ingria and Ilia Rodushkinab

Various stages of an analytical method for high-precision cadmium (Cd) isotope ratio measurements by MC-ICP-MS (sample preparation, matrix separation, instrumental analysis and data evaluation) were critically evaluated and optimized for the processing of carbon-rich environmental samples. Overall reproducibility of the method was assessed by replicate preparation and Cd isotope ratio measurements in various environmental matrices (soil, sediment, Fe–Mn nodules, sludge, kidney, liver, leaves) and was found to be better than 0.1& (2s for d114Cd/110Cd) for the majority of samples. Cd isotope ratio data for several commercially-available reference materials are presented and compared with previously published results where available. The method was used in a pilot study focusing on the assessment of factors aﬀecting Cd isotope composition in tree leaves. A summary of results obtained for a large Received 16th April 2014 number (n > 80) of birch (Betula pubescenes) leaves collected from diﬀerent locations in Sweden and Accepted 20th June 2014 through the entire growing season is presented and potential reasons for observed variability in Cd DOI: 10.1039/c4ja00125g isotope composition are discussed. Seasonal dynamics of element concentrations and isotope www.rsc.org/jaas compositions in leaves were also compared for Os, Pb, Zn and Cd.

Introduction low natural Cd abundance in many matrices, have, until relatively recently, hindered advancement in the area, primarily in Cadmium is present in terrestrial materials as a trace element terrestrial samples.8 Instrumental developments and the with average concentrations in nature at the mgkg 1 level and rening of analytical methods during the last decade have having eight isotopes covering a 10 amu range from 106Cd to resulted in signicantly improved precision in Cd isotopic 116Cd. Human activities such as ore mining,1 smelting/ analyses3 as well as the capability to analyze samples with lower rening2,3 including the industrial production of nickel– analyte contents.9,10 In their pioneering work, Wombacher et al.6 cadmium batteries,4 waste incineration and coal combustion, reported on the rst isotopic analyses of Cd in terrestrial one of the major anthropogenic Cd sources,5 contribute to the materials by multiple collector inductively coupled plasma anthropogenic burden of this element. In living beings Cd mass spectrometry (MC-ICP-MS) achieving good long term accumulates in vital organs with toxic and carcinogenic effects.6 reproducibility and high precision allowing the resolution of For example, it has been reported that accumulation can take minor isotopic variations of ca. 0.1&. place in vertebrate's kidneys with serious consequences such as Measured isotopic variations in samples of different nature, irreversible renal tubular damage.1 viz. meteorites, sea water, geological and environmental Pilot studies of the Cd isotopic system date back to the 1970s, matrices, highlighted the possibility of directly linking Cd when large variations were measured in meteoritic and extra- isotopic composition with source.6 Two main mechanisms were terrestrial materials.7 This started a discussion involving reported to inuence Cd isotopic fractionation: biological different laboratories, interested in exploring the potential of activity and partial evaporation/condensation.6 Although most Cd isotopic variations in environmental source tracing. Cd isotope ratio studies performed to date have focused on The eld of science focusing on the use of Cd isotopes is still cosmological and geological samples, recent research shows in its infancy compared with other more commonly used stable how Cd isotopic composition can be a valuable aid in dis- heavy isotopic systems such as Fe, Zn, Cu or Mo. Insufficient tinguishing potential pollution sources in a wide range of – measurement precision, typically about 0.5& to 1.0&, limited environmental applications.2 4 Overall, even though reported by both the lack of appropriate analytical technology and the Cd isotopic variations, deriving from both natural processes and anthropogenic activity, exceed 0.5& for 114Cd/110Cd ratio, several major terrestrial environments, e.g. silicate Earth, are aDivision of Geosciences, Lulea˚ University of Technology, S-971 87 Lulea,˚ Sweden. characterized by very stable isotope composition with variability E-mail: [email protected] within 0.05&.6,8 bALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Lulea,˚ Sweden

1570 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

Further investigations have proven Cd isotopic variation in Table 1 MC-ICP-MS operating parameters for Cd isotope ratio biological matrices to be inuenced by both natural and measurements anthropogenic processes.9–11 Ripperger et al.11 for example were Rf power (W) 1450  able to demonstrate the in uence of biological activity on Cd Ion lens settings Optimized for maximum sensitivity and 114 110 isotopic fractionation of d Cd/ Cd z +0.3& to 3.8& in signal stability oceans. Correlations between Cd concentrations and isotopic Coolant gas (l min 1)14 Sample gas (l min 1) 0.9–1.2 composition in near-surface waters were attributed to the Auxiliary gas (l min 1) 1.1 processes associated with the uptake of dissolved Cd by Sample uptake rate (ml min 1) 0.04–0.1 phytoplankton. On the other hand several contemporary studies have Variable Specication assessed the environmental impact of smelting and rening 107 108 109 processes with the help of Cd isotopic abundances and the Isotopes (cup) Ag(L4), Cd(L3), Ag(L1), 110Cd(Centre), 112Cd+112Sn(H1), results suggest that, even though isotope fractionation for a 114Cd+114Sn(H2), 116Cd+116Sn(H3), single industrial process generally does not exceed 1& 117Sn(H4) (d114Cd/110Cd), the anthropogenic signature in exposed objects Integration time, s 1.049 is still clearly distinguishable.2–4,6,12 There is also a growing Number of integrations 5 number of environmental applications relying upon informa- Number of blocks 5 2,3 Cycles per block 5 tion for several isotopes. For example, in a multi-tracer study, Amplier rotation Le Cd, Zn and Pb isotopic compositions and elemental concen- ICP parameters Adjusted daily for maximum sensitivity trations were used to distinguish between natural and anthro- and signal stability pogenic sources of these metals in bivalves.9,10 Zoom optic parameters Adjusted daily for maximum sensitivity Interest in bio-monitoring using natural accumulator and peak shape organisms has grown in the past few decades. Analyzing diﬀerent type of samples of plant origin can provide detailed spatial and temporal records of pollution.13,14 Trees growing in urban or industrial proximities constitute the easiest and least demountable quartz torch with 1.5 mm i.d. sapphire injector 14,15 expensive tool for trace metal contamination assessment and a platinum capacitively de-coupling shield using a combi- while surplus isotopic information (e.g. Pb, U, Os) can be useful nation of internal standardization (In) and external calibration.  16,17 in identi cation of pollution sources. Birch (Betula pubes- The sample introduction system consisted of a PFA spray cenes) can be utilized as bioindicators, being particularly useful chamber with two gas inlet ports (Cetac), a micro-concentric for such studies owing to their tolerance to high levels of heavy PolyPro nebulizer and a FAST SD2 auto-sampler (ESI, Perkin- 15 metals and fast growing rate. Elmer, Santa Clara, CA, USA) equipped with a six-port valve and The aim of this work was to evaluate and where required a 2 ml sample loop lled and rinsed by vacuum suction. modify existing analytical methodology for reproducible Cd Methane addition to the plasma was used to decrease formation isotope ratio measurement in various biological matrices suit- of oxide-based spectral interferences, improve sensitivity for able for bio-monitoring programs and to assess the natural elements with high rst ionization potentials, and to minimize variability of Cd isotope composition of birch leaves. matrix eﬀects.18 Operating conditions and measurement parameters for concentration measurements were as in Experimental previous studies19. Ashing of samples was performed at 550 C in a laboratory Instrumentation oven (Nabertherm GmbH, Lilienthal, Germany). A laboratory Cd and Pb isotope ratio measurements were performed by a microwave (MW) oven (MARS 5, CEM Corporation, Matthews,  NEPTUNE PLUS (Thermo Scienti c, Bremen, Germany) MC- USA), a high pressure asher (HPA-S, Anton Paar, Malmo,¨ Swe- ICP-MS instrument operated in low resolution mode. Two den) and an UltraWave single reaction chamber MW digestion  sample introduction con gurations were used, namely stan- system (Milestone, Sovisole, Italy) were used for sample dard (consisting of PFA nebulizer with approximately 50 ml digestions. min 1 sample uptake, cyclonic/Scott double spray chamber arrangement and H-skimmer cone) and high sensitivity setups (comprising an Aridus II desolvating nebulizer system from Chemicals and reagents

Cetac, Omaha, NE, USA and MC-ICP-MS interface equipped Nitric acid (HNO3) and hydrochloric acid (HCl) (both from with either standard cones or ‘Jet’ sampler and ‘X-type’ skimmer Sigma-Aldrich Chemie Gmbh, Munich, Germany) and hydrogen cone). The cup conguration and operating conditions are uoride (HF, 48%, Merck, Darmstadt, Germany) used in this given in Table 1. work were all of analytical grade. Water used in all experimental All measurements of element concentrations were per- procedures was de-ionized Milli-Q water (Millipore, Bedford, formed by an ELEMENT XR (Thermo Scientic) double- MA, USA) puried by reverse osmosis followed by ion-exchange focusing sector eld ICP-MS instrument equipped with a nickel cartridges. AG-MP-1M ion-exchange resin (macroporous, 100– sampler cone, a high sensitivity ‘X-type’ skimmer cone, a 200 dry mesh size, 75–150 mm wet bead size, Bio-Rad

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1571 View Article Online

JAAS Paper

Laboratories AB, Solna, Sweden) was cleaned by overnight during 2005–2013 from different locations in Sweden. Most of 2 soaking in 0.7 M HNO3 followed by multiple rinses with Milli-Q these samples were taken within an area of less than 0.5 km water and loading as slurry into 2 ml columns. north of the town of Lule˚a (22802500E, 653703700N) in Northeast NIST SRM3108 Cd solution Lot 130116 (National Institute of Sweden during the 2012 and 2013 growing seasons (mid-May to Standards and Technology, Gaithersburg, MD, USA) was used as October). Either leaves from single trees (not more than two ‘d-zero’ Cd isotope bracketing standard, as suggested by Abou- leaves from a single branch on each sampling occasion), or chami et al.,20 in all measurement sessions. Three commercially from a group of trees in the same location (2–3 leaves from each available 1000 mg l 1 Cd standards were used to prepare three tree, hereaer referred to as pooled samples) were collected by quality control samples (QCS), namely QCS A (Spectrapure hand wearing powder-free laboratory gloves into Zip-Lock Standards, Oslo, Norway, Lot 111 issued February 2012, with Cd plastic bags marked with tree size, location and sampling date. metal as starting material), QCS B (Ultra Scientic, North Almost all samples were collected from the ground limiting the Kingstown, USA, Lot K00951 issued September 2009, with Cd maximum sampling height to approximately 2.5 m. Hence only nitrate hydrate Lot BH01228 as starting material) and QCS C lower branches were sampled for large trees. On one occasion, it (Ultra Scientic, Lot M00538A issued May 2011, with Cd nitrate was possible to collect leaves representing different heights hydrate Lot BH01820 as starting material). The rst solution from birch freshly felled in a storm. The amount of collected was analyzed at the beginning and at the end of every analytical material was chosen to provide approximately 2 g of dried leaves session, while the remaining two were analyzed in random for individual or 6–8 g for pooled samples. Samples of litter, order through the duration of the study. An aliquot of so called consisting from fall-out and partly decomposed birch leaves, ‘UM-Munster-Cd¨ ’ standard solution (20.6 mg l 1) was provided mushrooms (Boletus edulis) and large pendulous lichens (Alec- by the Institute for Geology and Mineralogy of the University of toria sarmentosa) were sampled in 2012–2013 at the Lule˚a Cologne. Two solid Cd chemicals, oxide (Spex Industries Inc., location. Within 24 h of collection, samples were transported to Metuchen, NJ, USA, Lot 7821N) and acetate dihydrate (Riedel De the laboratory, dried at 50 C overnight, homogenized and Haën AG, Hannover, Germany, Lot 32308), dissolved in stored in closed bags at room temperature prior to analysis. concentrated HNO3 were used to test the range of Cd isotope A ow chart depicting all steps of the analytical method is composition in chemicals used in the laboratory. shown in Fig. 1. Working solutions of the abovementioned standard or 1 chemicals with Cd concentrations of 5, 20 or 200 mgl , Sample preparation depending on conguration of MC-ICP-MS introduction system All laboratory ware coming into contact with samples or sample used, were prepared daily by serial dilution in 0.14 M HNO3. digests was soaked in 0.7 M HNO (>24 h at room temperature) Internal standard Ag, prepared from 1000 mg l 1 stock standard 3 and rinsed with MQ water prior use. All sample manipulations from Ultra Scientic, was added to all measurement solutions at were performed in clean laboratory areas (class 10 000) by half of the respective Cd concentration. personnel wearing clean room gear and following all general precautions to reduce contamination.21 Samples For method optimization and testing, a range of certied refer- Sample digestion ence materials (CRM), ERM BB186 Pig Kidney (Institute for To bring solid samples into solution, several preparation Reference Materials and Measurements, Geel, Belgium), NIST schemes were tested (Table 2): SRM 2711 Montana Soil (provided by CRPG – Centre de - MW-assisted digestion in closed Teon vials using a 22 Recherches Pétrographiques et Géochimiques, Vandœuvre les HNO3–HF acid mixture (Procedure A); Nancy, France), SRM 2709 San Joaquin Soil (NIST), VKI-QC - MW-assisted digestion in closed Teon vials using a municipal sludge (Eurons A/S, Vallensbæk Strand, Denmark), HCl–HNO3–HF acid mixture (Procedure B); LGC6187 river sediment (LGC, Teddington, Middlesex, UK), - MW-assisted digestion in closed Teon vials using a

GBW07311 stream sediment (National Research Centre for HCl–HNO3–HF acid mixture preceded by ashing of the sample Certied Reference Materials, Beijing, PR China), PACS-2 marine at 550 C (Procedure C); sediment and TORT-1 Lobster Hepatopancreas (National - HPA or UltraWave digestion using only HNO3 (Procedure D); Research Council of Canada, Ottawa, Canada), NOD-P-1 and - Ashing at 550 C followed by solubilization of ash in NOD-A-1 manganese nodules (United States Geological Survey, concentrated HCl (Procedure E). Denver, CO, USA), as well as in-house control samples, namely The rst procedure was used exclusively for preparation of freeze-dried kidney collected from moose (Alces alces) hunted in carbon-rich matrices for determination of element concentra- Northeast Sweden, freeze-dried herring (Clupea harengus from tions. Procedures B and C were used for soil, sludge and sedi- Bothnian Bay) liver, pooled dried mushroom (Boletus edulis) and ment CRMs. Procedures D and E were used for preparation of two pooled samples of birch (Betula pubescenes) leaves were carbon-rich matrices for isotope ratio measurements. used. Note that none of the materials mentioned above has a MW-assisted digestion is a mature and broadly available certied Cd isotopic composition. technique for digestion of various environmental matrices Aer optimization, the analytical methodology was applied using acid or acid mixtures23,24. Up to 40 samples can be to analyses of more than 80 birch leave samples collected prepared simultaneously in less than 3 hours, including sample

1572 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

Fig. 1 Flow chart of the analytical method.

Table 2 Details of sample preparation procedures

Variable Procedure A Procedure B Procedure C Procedure D Procedure E

Max. sample weight, g 0.05 0.5 0.5 0.5 5 Vessel volume and material 25 ml Teon 60 ml Teon 60 ml Teon 15 ml Teon 70 ml quartz 20 ml porcelain Number of vials per rack 40 40 40 15/4 NA

Acid mixture HNO3–HF HCl–HNO3–HF HCl–HNO3–HF HNO3 HCl v/v ratio 99/1 6/2/2 6/2/2 NA NA Volume, ml 2 10 10 10 1 Digestion system MARS 5 MARS 5 MARS 5 UltraWave/HPA-S Nabertherm Max. temp., C 170 170 170 250/300 550 Ramp time, min 15 15 15 25/30 120 Hold time, min 25 25 25 10/40 300

weighing, adding reagents, temperature ramping, holding and though the cost of instrumentation is signicantly higher and cooling times, as well as transferring digests to storage or sample throughput (especially for the HPA) is somewhat lower evaporation vessels. However, for dried organic matrices the than for MW oven-based methods. maximum recommended sample size per digestion vessel is Ashing of samples provides efficient mineralization of limited to 0.3–0.5 g. Therefore it may require several parallel organic-rich matrices with the ability to handle large initial digestions in order to obtain a representative sample or to weights, has the lowest equipment costs and allows high prepare a sufficient amount of analyte needed for isotope ratio throughput. Simple room temperature leaching of the solid measurements. Moreover, incomplete oxidation of carbon may sample residue in a low volume of mineral acids can be suffi- interfere with ion-exchange separation at later stages and cient for quantitative analyte recovery thus limiting blank preceding ashing of organic-rich matrices might be necessary. contributions from reagents and non-disposable digestion HPA and UltraWave digestions occur at much higher pressures vessels.21 Potential volatilization losses of analytes having and temperatures thus ensuring more efficient oxidation,

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1573 View Article Online

JAAS Paper signicant vapour pressures can, however, limit the applica- For digests prepared by all other procedures and in both bility of the latter approach and needs to be carefully checked. tested matrices, >98% of major matrix elements (K, Ca, Al, Mg, An aliquot of digests, including those intended primarily for P, S, Na, Fe, Si, Mn, etc.) pass through the column during isotope ratio measurements, was diluted and analyzed by sample loading (4 ml) and the subsequent 2 M HCl matrix wash ICP-SFMS. Cd recoveries for all CRMs tested in this study and (8 ml). The next fraction (12 ml of 0.3 M HCl) contains >90% of using all digestion approaches were in 94–102% range with no the initial Pb and can be used for Pb isotope ratio measure- statistically signicant differences between procedures with or ments by ICP-SFMS or MC-ICP-MS if needed, with a minor without ashing step. Hence, volatilization losses of Cd during fraction, approximately 0.5–1.0% of the matrix elements ashing procedure (if any) are below 6% conrming similar initially present remaining. Zn is quantitatively (>99%) recov- observations from studies employing Cd radioisotopes.25–27 It ered in 12 ml of 0.012 M HCl that can be used for Zn isotope should be noted that signicant losses of Hg, Te, Tl, Sb and Se ratio measurements by MC-ICP-MS if needed, followed by Cd were observed during ashing. The rest of sample solutions were elution (>98% recovery) in 0.0012 M HCl (24 ml). Finally, 12 ml evaporated to dryness in 25 ml Teon beaker at 95 Con of 6 M HNO3 containing traces of HF remove the remaining Ag, ceramic-top hot plate, followed by dissolution in 4 ml of 2 M Bi, Tl, Nb, Sb, Sn, Hg, Zr and U that are strongly bound to the HCl thus ready for Cd purication. For samples prepared by resin. The fact that almost all Ag elutes in this last fraction Procedure E, ashed material was transferred to a 50 ml poly- contradicts the ndings of Gao et al.3 who reported that Ag was ethylene, conical-bottom vial followed by addition of 1 ml 10 M recovered during sample load and matrix wash. The entire HCl, mechanical agitation at room temperature for 1 hour and separation procedure, including column pre-conditioning and – addition of 4 ml H2O. Remaining undissolved phases (residual passing 10 ml of H2O at the end of separation, takes 5 6 h and carbon particles and silicates) were allowed to settle and upper up to 40 columns have been used in parallel. The performance 4 ml of clear phase was used for Cd purication step. of the columns was unchanged aer ve consecutive separation cycles. Apart from efficient Cd separation from the sample matrix,  Cd puri cation manifest by <0.02% of the total dissolved solids load present in A number of procedures have been proposed for the separation the original digests eluting in the Cd fraction, and quantitative and purication of Cd, varying in complexity from a single pass recovery, contamination by elements that could result in spec- through one column to very elaborative, multiple-stage tral interferences on Cd and Ag isotopes to be monitored by MC- schemes,28–31 depending on the sample matrix. The majority of ICP-MS6 need to be assessed. Concentrations of Pd (source of recent studies on high-precision Cd isotope ratio measure- isobaric interferences on 108Cd and 110Cd), Zr (source of ZrO+ ments in environmental samples rely upon the separation interferences on 107Ag, 108Cd, 110Cd and 112Cd), Nb (source of procedure using AG-MP-1M ion-exchange resin originally NbO+ interference on 109Ag) and Th (source of Th2+ interference proposed by Cloquet et al.,8 in some cases with minor modi- on 116Cd) in the analyte fraction were below respective limits of cations according to Gao et al.3 The applicability of this proce- detection ensuring that Cd to interfering element concentration dure for the matrices analyzed in the present study was ratios are >100 000 and therefore contribute negligible level of evaluated for mushroom (prepared using Procedures A, D and spectral interferences. The Cd to Zn concentration ratio in the E) and soil (prepared using Procedures B and C) digests sepa- analyte fraction was >100 and given the low argide formation rated using columns lled with approximately 2 ml of resin rate in the ICP,4 the interferences from ZnAr+ on 107Ag, 108Cd slurry. This was done by sampling load, matrix wash and all and 110Cd can be neglected. Both Mo (source of MoO+ inter- elution fractions with 1 ml resolution followed by ICP-SFMS ferences on 108Cd, 110Cd, 112Cd, 114Cd and 116Cd) and Sn (source analysis, thus obtaining detailed elution proles for practically of isobaric interferences on 112Cd, 114Cd and 116Cd) were all elements present in these samples. Very similar elution present in the Cd fraction at levels approaching 1% of the Cd proles for major matrix elements and for Cd were obtained concentration in some samples. Addition of an extra elution regardless of sample matrix and preparation procedure, except step (12 ml 0.06 M HCl) between the Zn and Cd fractions, as that much broader elution peaks and therefore signicantly recommended by Gao et al.,3 helps to decrease Sn concentration higher elution volumes needed to be collected for complete Cd in the latter by approximately 30–50%, but as this also results in recovery from the separation of digests prepared by Procedure larger volumes required for complete Cd elution and prolongs A. Interference from high concentrations of residual non- the separation procedure by almost 20%, no overwhelming oxidized carbon, the presence of which was obvious from the benet was evident. distinctly deep-brown colour of load solutions prepared by this It should be noted that the mushroom and soil samples used procedure as opposed to the transparent or light-yellow tinged in the aforementioned tests contain relatively high Cd concen- digests typical of other procedures, affecting the column sepa- trations (>10 mg kg 1) and the severity of problems originating ration process is the most probable explanation for this effect. from concomitant elements in the analyte fraction may be Consequently sample preparation by Procedure A was not used considerably higher for matrices with sub mg kg 1 Cd for Cd isotopic analyses. It is possible that an undigested concentrations. Moreover, there is a risk that additional residue of organic matrix contributes to matrix effects previ- contamination might be introduced during the next prepara- ously attributed solely to resin-derived organic compounds and tion step – namely evaporation of the Cd fraction to dryness in inorganic elements.20,32 25 ml Teon vials at 95 C on a ceramic-top hot plate and

1574 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

dissolving the residue in 3 ml 0.14 M HNO3. This was necessary d-values for all Cd ratios were calculated against bracketing in order to prevent effects from remaining chloride ions on the NIST SRM 3108 standards: " # Ag internal standard, for close matching of the acid strength ðxCd=yCdÞ between all measurement solutions in the MC-ICP-MS analyt- dx=yCd ¼ sample 1 1000 ðxCd=yCdÞ ical sequence and to increase the Cd concentration four-fold, NIST3108 which is advantageous for samples low in Cd. For all samples where xCd and yCd correspond to the two different Cd isotopes, prepared in the course of this study, a 0.2 ml aliquot of this x y the ( Cd/ Cd)sample value refers to the measured ratio and solution was diluted 25-fold in 0.7 M HNO3 and analyzed by x y ( Cd/ Cd)NIST3108 is the isotope ratio of the standard. The factor ICP-SFMS. In combination with concentrations obtained from 1000 is used to convert the d-values to per mil notation. When the analysis of digests prior to separation and volumes of cor- the d-value refers to a ratio of a heavier to a lighter isotope, a responding fractions this enabled assessment of Cd recovery positive d-value corresponds to an enrichment in the heavier and the quantitation of interfering elements. isotope compared to the standard. Mean ratios from two consecutive measurements of the rst Isotope ratio measurements and data evaluation and the third samples (or QC) in each analytical block (stan- dard1–sample1–sample2–sample3–standard2) were calculated For MC-ICP-MS measurements, the Cd concentration was against ratios for standards 1 and 2 respectively. For the second m 1 adjusted to 200 gl with 0.14 M HNO3 and spiked with Ag to sample, mean ratios from bracketing standard were used m 1 50 gl before analysis with the standard introduction system assuming linear changes in instrumental mass bias persisting m 1 m 1 or to 20 gl and spiked with Ag at 5 gl for the high aer internal standard correction. Results from two consecutive sensitivity introduction system. When the latter system was measurements of each sample allow calculation of mean used, the Aridus was turned on approximately 30 min before d-values and respective standard deviations for all Cd isotope starting the MC-ICP-MS device. The instrument was allowed to ratios that are less affected by variations caused by imperfect stabilize for at least 1 h with the plasma lit while aspirating 0.14 amplier gain calibration. As an additional aid to check for M HNO3 blank solution prior to performing the optimization of internal consistency of isotope data, all the d-values for each  the operational parameters (gas ows, torch position, and lens sample were normalized by mass difference (e.g. Table 3), a – settings) and mass calibration in the 107 120 amu range. data presentation that resembles the frequently used three 114 m 1 Typical Cd intensity was 5 V for 200 gl using the standard isotope plot.32 m 1 introduction system, and 4 V for 20 gl using the Aridus and Pb and Zn isotope ratio measurements in separated fractions standard cones. Similar instrumental sensitivity was reported (Fig. 1) were performed using internal standardization with Tl 33 by Wombacher et al. for a Nu Plasma MC-ICPMS instrument. and Cu, respectively, and bracketing standards, NIST SRM 981 ‘ ’ ‘ Using the Aridus in combination with the Jet sampler and X- Common Lead and IRMM 3702, respectively, as described in ’ type skimmer cone, our NEPTUNE PLUS typically generated detail elsewhere.36,37 114 m 1 12 V intensity for the Cd isotope at a concentration of 20 gl . For Os isotope measurements a detailed method description Samples were analyzed using bracketing isotope standards is available in previous studies.16,17 (NIST SRM 3108) with matching Cd and Ag concentrations and acid strength. Three samples or QCS solutions were analyzed between each pair of standards. Two consecutive measure- Results and discussion ments were performed for each solution in the sequence. The soware option of excluding pass, run and block outliers was Performance of separation procedure deactivated as it was found that this improved correlation The average Cd method blank for the entire procedure, as between instrumental mass bias for Cd and Ag isotope ratios. assessed by applying all preparation and separation steps to a Even without using this option the typical in-run precision in set of reagent blanks handled as samples, was 0.14 0.09 ng isotope ratios was in the 0.02–0.04& range and 0.06–0.20& (n ¼ 17). This corresponds to <0.2% contribution for samples range for measurements performed with standard or high with the lowest Cd content tested and therefore has negligible sensitivity introduction systems, respectively. Higher signal impact on measured ratios. The Cd recovery from all samples uctuations observed using the latter system are the most likely separated during this study (n > 100) was above 95%. Levels of root cause for the deterioration in measurement precision. common contaminant-prone elements (Na, Ca, K, Mg, S) in Intensity data for all monitored isotopes (Table 1) were analyte fraction prepared for MC-ICP-MS were, as a rule, below transferred to spreadsheet soware for calculations. Firstly, 100 mgl1. In some matrices, mgl1 concentrations of Fe contributions from Sn isobaric interferences on 112Cd, 114Cd (Mn–Fe nodules), Zn, Tl, and Sb were also present. Such low and 116Cd were corrected mathematically using the 117Sn signal, levels of impurities are unlikely to cause signicant matrix tabulated Sn isotope abundances34 and computed instrumental effects or spectral interferences. As far as elements that can mass bias for each analytical session. Corrected intensities were spectrally interfere with Cd and Ag isotopes are concerned, the then used to calculate een Cd isotope ratios and the concentrations of Pd, Nb, Th and Zr were below 0.01 mgl 1 in all 109Ag/107Ag ratio. Instrumental mass bias was corrected in two Cd fractions. For birch leaves, mushroom, liver and kidney steps. Firstly, revised exponential correction35 using the internal samples, Mo and Sn levels in analyte solution were also below standard (Ag ratio) was applied to all Cd ratios. Secondly, 1% of respective Cd concentrations. However, in some

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1575 View Article Online

JAAS Paper

Table 3 Delta values (in &) for Cd isotope ratios normalized by respective mass diﬀerence for birch leaves, ﬁsh liver and NIST SRM 3108 spiked with Mo and Sn

Cd/Mo ¼ 4 Cd/Mo ¼ 10 Cd/Mo > 100 Cd/Mo > 100 Cd/Mo > 100 Cd/Mo ¼ 4 Cd/Sn > 100 Cd/Sn ¼ 2 Cd/Sn > 100 Cd/Sn > 100 Cd/Sn ¼ 10 Cd/Sn > 100 NIST SRM NIST SRM Birch leaves Fish liver NIST SRM 3108 NIST SRM 3108 3108 Aridus 3108 Aridus

(d110Cd/108Cd)/2 0.129 0.177 0.012 5.915 0.770 0.238 (d111Cd/108Cd)/3 0.130 0.180 0.003 3.823 0.560 0.160 (d112Cd/108Cd)/4 0.135 0.181 0.005 2.969 0.462 0.115 (d114Cd/108Cd)/6 0.136 0.182 0.008 1.959 0.344 0.081 (d116Cd/108Cd)/8 0.133 0.184 0.012 1.425 0.288 0.009 d111Cd/110Cd 0.131 0.185 0.014 0.357 0.063 0.005 (d112Cd/110Cd)/2 0.140 0.185 0.019 0.023 0.016 0.009 (d114Cd/110Cd)/4 0.139 0.185 0.020 0.019 0.009 0.008 (d116Cd/110Cd)/6 0.137 0.187 0.011 0.169 0.066 0.051 d112Cd/111Cd 0.150 0.185 0.011 0.403 0.058 0.022 (d114Cd/111Cd)/3 0.142 0.185 0.019 0.094 0.029 0.009 (d116Cd/111Cd)/5 0.139 0.187 0.021 0.012 0.003 0.086 (d114Cd/112Cd)/2 0.138 0.185 0.043 0.061 0.021 0.025 (d116Cd/112Cd)/4 0.136 0.188 0.028 0.150 0.039 0.102 (d116Cd/114Cd)/2 0.134 0.190 0.099 0.170 0.034 0.229 Mean (SD) 0.137 (0.005) 0.184 (0.003)

environmental CRMs, signicantly higher concentrations of solutions with Cd/Sn # 10 (Table 3). Assessment of potential these interfering elements were found, sometimes at levels natural or articially-introduced isotopic fractionation of the Sn exceeding 20% of the Cd content. In contrast to the majority of eluted in the Cd cut would require either a separate measure- elements, the elution behaviour of Sn and Mo varies signi- ment session for Sn isotope ratios or obtaining such informa- cantly depending on the sample matrix, the preparation tion by performing MC-ICP-MS measurements in the dynamic procedure, and even on the individual column used. For mode. Neither of these approaches was tested in the present example, during the separation of digests prepared without an study, but it should be considered for matrices where the Cd/Sn ashing step, most (>95%) Mo elutes in the last HNO3 fraction, concentration ratio in separated fractions is below 10. while when the same matrix is prepared by ashing, up to 50– Spectral interferences caused by MoO+ affect all Cd isotopes 60% of Mo elutes in the Pb and Zn fraction with partial tailing monitored, but to signicantly different degrees. For a given Cd/ into the Cd cut, and a secondary elution peak in the last frac- Mo concentration ratio and assuming stable oxide formation in tion. Hence, the share of the total loaded amounts of these the ICP or MC-ICP-MS interface, the severity of these interfer- elements co-eluting in the Cd fraction may differ by a factor of ences will depend on the relative abundances of the Mo isotopes 2–3 even for replicate preparations and separations. causing Mo16O+ interferences and those of the Cd isotopes In the case of Sn, the efficiency of mathematical corrections affected. For example, interference will be signicantly more might be affected by the uncertainty in the actual instrumental pronounced on the 108Cd (0.89% natural abundance) affected mass bias deduced from measured and tabulated34 Cd or Ag by 92Mo (14.8%) than for 110Cd (12.6%) affected by 94Mo (9.2%), ratios and also by Sn fractionation, either in the original sample and this in turn will affect the Cd isotope ratios very differently. or introduced during column separation. The severity of both Apparent deviations in d-values for different Cd ratios calcu- factors will increase with decreasing Cd to interferent ratio in lated for NIST SRM 3108 solutions spiked with increasing Mo the measurement solution. Instrumental mass bias per mass concentrations show strong positive correlation (R2 > 0.94) with unit calculated from measured (in bracketing standards) and the factor ((x 16)Mo yCd)/(xCd (y 16)Mo), conrming that calculated (using tabulated isotope abundances) Ag or Cd the contributions from Mo17O+,Mo18O+ or MoN+ are negligible isotope ratios varies by up to 0.3% depending on the ratio used, to a rst approximation. These tests also demonstrate that for possibly reecting uncertainties in tabulated abundances or solutions with Cd/Mo concentration ratios of 4 or above, there limitations of the correction model. A set of tests, using the are a few ratios (112Cd/110Cd, 114Cd/110Cd and 116Cd/111Cd) that measurement protocol described above and applied to NIST are not signicantly affected (Table 3). The formation of MoO+ SRM 3108 solutions spiked with increasing Sn concentrations, decreases almost 10-fold in a sample introduction system shows that the best agreement between corrected ratios and incorporating a desolvating nebulizer compared to one using a those for unspiked standards was obtained when employing standard conguration due to the lower solvent vapour loading mass bias deduced from the tabulated 112Cd/111Cd ratio. This to the ICP. Therefore the former sample introduction system yielded efficient correction, evident from d-values for affected offers important advantages for Cd isotope ratio measurements ratios aer correction in the 0.02& < d < 0.02& range, for the in samples containing Mo. majority of ratios except those involving the 116Cd isotope in

1576 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

The manner of presenting normalized d-values exemplied approach, typical method reproducibility can be estimated in Table 3 makes identication of results affected by spectral using a set of test samples prepared and analyzed in different interferences from MoO+ or from Sn very straightforward, being analytical sessions. Therefore either just the measurement and manifested by large uncertainties in mean normalized d-values data evaluation steps for synthetic Cd solutions and Cd chem- calculated using all 15 isotope ratios, and by distinct repro- icals prepared by dissolution, or the entire analytical procedure ducible deviation patterns characteristic of each interfering for solid environmental and geological matrices, was applied to species. Deviating d-values for ratios involving the 116Cd isotope a set of reference and control samples with variable matrix are indicative of inadequate correction for Sn interference and composition and Cd concentrations38 (Table 4). A minimum of imply that caution should be exercised if using results for other two aliquots of each material was prepared and each Cd solu- ratios that can be affected, viz., those involving 114Cd and 112Cd tion analyzed in duplicate during at least two measurement isotopes. Negative d-values for ratios with the 108Cd isotope as sessions. For ‘zero-matrix’ samples (QCS A–C and solid Cd denominator, monotonically increasing from 110Cd/108Cd to chemicals) with neither extensive sample preparation nor 116Cd/108Cd are caused by MoO+ interference and only the least column separation necessary, long-term reproducibility (2s)is affected Cd ratios (d114Cd/110Cd and d116Cd/111Cd) should be approximately 0.02& (Table 3) that is slightly better than the used. It should be noted that, for almost all birch leaves range of (0.02–0.1)& per amu quoted by Wombacher et al. using analyzed in this study, Mo and Sn contamination of the Cd either external standardization or bracketing standards fraction was negligible, providing typical spread for mean approaches.6 This denes the highest level of precision that can normalized d-values below 0.01&. This also suggests the be achieved for Cd isotope ratios measured by MC-ICP-MS in absence of measurable mass independent fractionation.6 solutions given perfect concentration and matrix matching with isotope standards in the absence of spectral interferences. Reproducibility Reproducibility depreciates to approximately 0.05& for the In theory, sample matrix, analyte content and all stages of the majority of solid environmental matrices, though in some cases & measurement procedure might affect method reproducibility. it may deteriorate to 0.1 (Table 4). As a rule, poorer repro- Thus accurate assessment of the overall reproducibility of the ducibility was noted for samples with either initially low Cd method would require replicate preparation, separation and content or when there are restrictions on the sample intake per – analyses of all samples that might be impractical or even preparation (Procedures B D). Therefore MC-ICP-MS measure- unfeasible for large studies. As a more cost and time efficient ments had to be made on undiluted Cd fractions (higher risk for

Table 4 d114Cd/110Cd for reference and control samples

Cd, mg kg 1 found s Test sample Preparation (certied s) d114Cd/110Cd (2s), &

QCS A Dilution, n ¼ 34 (1000) 0.020 (0.024) QCS B Dilution, n ¼ 17 (1000) 0.008 (0.020) QCS C Dilution, n ¼ 17 (1000) 0.584 (0.023) UM-Munster-Cd¨ Dilution, n ¼ 8 (20.6) 4.437 (0.042) ¼ CdO Dissolution in HNO3, n 8 (875 000) 0.007 (0.016) Cd(CH3COO)2$2H2O Dissolution in HNO3, n ¼ 8 (422 000) 0.018 (0.022) Moose kidney Procedure E, n ¼ 8 12.3 0.4 0.635 (0.034) Procedure D(HPA), n ¼ 4 12.5 0.3 0.597 (0.049) Fish liver Procedure D (UltraWave), n ¼ 6 1.82 0.10 0.789 (0.048) Mushroom Procedure E, n ¼ 10 10.1 0.6 0.365 (0.037) Procedure D (HPA), n ¼ 6 10.4 0.3 0.347 (0.089) Leaves A Procedure E, n ¼ 4 1.75 0.09 0.525 (0.068) Procedure D (UltraWave), n ¼ 4 1.73 0.12 0.500 (0.031) Leaves B Procedure E, n ¼ 4 1.14 0.07 0.733 (0.099) Procedure D (UltraWave), n ¼ 4 1.19 0.07 0.772 (0.102) ERM BB186 Pig kidney Procedure E, n ¼ 4 1.04 0.06 (1.09 0.05) 0.465 (0.062) TORT-1 Lobster Hepatopancreas Procedure E, n ¼ 4 25.7 1.3 (26.3 2.1) 0.123 (0.025) NIST SRM 2711 Procedure C, n ¼ 8 42.5 1.9 (41.70 0.25) 0.803 (0.071) Montana soil Procedure D (HPA), n ¼ 8 40.9 1.6 (41.70 0.25) 0.711 (0.106) NIST SRM 2709 soil Procedure C, n ¼ 4 0.37 0.02 (0.38 0.01) 0.007 (0.089) GBW 07311 stream sediment Procedure C, n ¼ 4 2.19 0.10 (2.3 0.1) 0.305 (0.054) VKI-QC municipal sludge Procedure C, n ¼ 4 1.19 0.09 (1.34 0.17) 0.067 (0.036) LGC6187 river sediment Procedure C, n ¼ 4 2.77 0.14 (2.7 0.3) 0.313 (0.048) PACS-2 marine sediment Procedure C, n ¼ 4 2.04 0.08 (2.11 0.15) 0.204 (0.040) NOD-A-1 manganese-nodule Procedure B, n ¼ 4 7.28 0.31 (7.5 0.16a) 0.086 (0.031) NOD-P-1 manganese-nodule Procedure B, n ¼ 4 21.5 1.1 (22.6 0.3a) 0.120 (0.038) a Non certied values from Axelsson et al.47

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1577 View Article Online

JAAS Paper minor matrix effects) with high sensitivity setup. Apart from term reproducibility. The fact that similar Cd isotope fraction- higher in-run precision (see Experimental), the stability of ations in favour of heavy isotopes were obtained for in-house instrumental mass bias using the Aridus is also notably inferior moose kidney control sample and for ERM BB186 Pig kidney is to that of the standard introduction system, and memory effects also reassuring. As moose diet mainly consists of terrestrial for Cd, Mo and Sn are more pronounced requiring much longer vegetation, the observed similarities between mean wash-out times. All these factors most probably contribute to d114Cd/110Cd values in moose kidney and in birch leaves poorer reproducibility and the use of Aridus should be avoided (Table 5) are to be expected. when ultimate precision is a major goal, e.g. in certication or The isotopically lighter Cd in Lobster Hepatopancreas and inter-laboratory comparison campaigns. Differences in repro- sh liver (Table 4) falls within the published range for bivalves ducibility between pure standard solutions and samples of collected from western Canada, Hawaii, and the US East Coast.9 biological origin have previously been noted in other stable Four out of ve chemicals and commercial standard solu- isotope studies39 as well. tions tested in this study have Cd isotope compositions indistinguishable from those for NIST SRM 3108 (Table 3) with d114Cd/110Cd in the range from 0.02& to 0.02&, while Cd in Accuracy QCS C is enriched in light isotopes. Interestingly, both QCS B As to the best of our knowledge no reference materials with and QCS C have a common supplier, the same catalogue certied Cd isotope composition are available, the only means number and the only difference is the lot of starting material, to evaluate method accuracy is to compare d-values obtained in Cd nitrate hydrate. An almost 0.6& difference in d114Cd/110Cd our study (Table 3) with previously published data where such between two standard solutions from the same supplier high- exist. This was possible for one Cd solution (UM-Munster-Cd)¨ lights the importance of having a common ‘d-zero’ Cd isotope and three reference materials (NOD-P-1, NOD-A-1 and NIST standard20 for all research groups studying Cd isotope 2711) as shown in Table 4. The degree of agreement between fractionation. data found here and previously reported for these materials was rather mixed: - UM-Munster-Cd¨ d-values obtained in this study (4.44 Cd isotope composition in birch leaves 0.04)& proved to be in good agreement with those proposed by A summary, of Cd data birch leaves and lichens collected in Cloquet et al.8 (4.48 0.04)&. This is despite differences in the Sweden and analyzed in the course of this study, is presented in “d-zero” reference materials used (NIST SRM 3108 and Cd Spex). Table 5. The rst general observation that can be made from the Our value is also within the range obtained for this standard in entire set of data is that the Cd in birch leaves has a composi- an inter-laboratory exercise.20 It must be noted though that this tion enriched in heavy isotopes (mean d114Cd/110Cd value of solution consists of articially fractionated Cd and that no 0.7&, range 0.3–1.3&). This is slightly surprising as the pref- sample pre-treatment other than dilution was needed; erential accumulation of lighter isotopes in leaves is far more - Acceptable correspondence was also found for NOD-P-1 d- common (Fe,38 Zn,40 Cu 41-43). values between this study (0.12 0.04)& and literature data With reservation for the relatively limited number of (0.13 0.12)&,8 which should be considered more than satis- sampling locations tested, there are no obvious site-specic factory given measurement uncertainties; differences between mean concentrations and the ranges of Cd - Agreement was less impressive between the results for the isotope compositions observed for 18 sampling sites. The latter second Fe–Mn nodule (NOD-A-1), i.e. (0.09 0.03)& versus span over almost 1250 km along the Swedish coast line, from in (0.07 0.12)&;8 the vicinity of the town of Hogan¨ äs in the south-west to the - The poorest agreement was between d-values for NIST 2711 neighbourhood of Haparanda in the north-east, as well as one (Montana soil), with found values of (0.80 0.07)& and (0.71 very conned sampling area in the vicinity of Lule˚a. For the 0.11)& deviating from published data, (0.51 0.02)&.8 entire data set, there is no signicant correlation between Cd All the mentioned Cd isotopic data from the literature were concentrations and d-values (R2 ¼ 0.08). Birch leaves with the obtained using a MC-ICP-MS Micromass® Isoprobe™ and highest Cd concentrations (>1 mgkg 1) found in some locations instrumental details can be found elsewhere.8 do not exhibit isotope compositions deviating signicantly The exact reasons for such discrepancies between results from the mean. obtained in different laboratories is unknown and current There are no differences in Cd concentrations in leaves difficulties in the assessment of method accuracy call for the growing at different heights (Table 5), although the Cd isotope preparation of reference materials that have certied Cd isotope composition tends to become slightly lighter at the top of the composition. crown, with fractionation during diffusion44 in sap solution No signicant differences were found for d114Cd/110Cd in being a plausible explanation. As only based on data from a kidney, mushroom, and leaves prepared by Procedures D and E single birch, no far-reaching conclusions can be drawn from (Table 4, mean difference <0.03&) thus assuring that the ashing these ndings. step does not notably affect the Cd isotope composition for On two sampling occasions, statistically signicant differ- carbon-rich matrices. Agreement is somewhat poorer (0.09& ences in d114Cd/110Cd were observed between pooled samples difference) for NIST SRM 2711 prepared by Procedures C and D, representing lower branches of old trees or all branches of but is still acceptable if allowing for a 0.1& estimation of long- young trees (Table 5). As these leaves were sampled at the very

1578 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

Table 5 Cd concentrations and d114Cd/111Cd in birch leaves and lichens from Sweden

Cd, mg kg 1 d114Cd/110Cd

All locations, 2005–2013 Mean SD Min O Max Mean SD Min O Max

All samples (n ¼ 83) 0.40 0.23 0.07 O 1.70 0.70 0.20 0.30 O 1.28 Lule˚a(n ¼ 65) 0.36 0.15 0.07 O 0.71 0.73 0.19 0.34 O 1.28 Rest of Sweden (n ¼ 18) 0.59 0.41 0.23 O 1.70 0.59 0.17 0.30 O 0.99

Lule˚a, 2013-08-02 Low branches (H ¼ 4m)(n ¼ 2) 0.68 0.07 0.622 0.070 Medium branches (H ¼ 16 m) (n ¼ 2) 0.65 0.06 0.454 0.055 High branches (H ¼ 24 m) (n ¼ 2) 0.63 0.06 0.430 0.049

Lule˚a, 2013-05-19 Old trees, pooled sample A (n ¼ 2) 0.31 0.03 0.810 0.088 Young trees, pooled sample A (n ¼ 2) 0.26 0.02 0.689 0.021

Lule˚a, 2013-05-25 Old trees, pooled sample B (n ¼ 2) 0.30 0.03 0.837 0.079 Young trees, pooled sample B (n ¼ 2) 0.35 0.03 0.662 0.043

Lule˚a, 2012–2013 2013-05-19 (n ¼ 5) 0.46 0.17 0.61 0.05 2013-05-25 (n ¼ 5) 0.44 0.16 0.66 0.10 2012-05-27 (n ¼ 5) 0.40 0.18 0.59 0.11 2013-07-17 (n ¼ 5) 0.27 0.11 0.81 0.18 2012-08-05 (n ¼ 5) 0.33 0.13 0.73 0.14 2013-08-31 (n ¼ 5) 0.28 0.11 0.81 0.19 2013-09-15 (n ¼ 5) 0.36 0.13 0.84 0.15 2013-10-05 (n ¼ 5) 0.37 0.19 0.87 0.13 Pooled litter samples (n ¼ 3) 0.74 0.29 0.46 0.19 Pooled lichen samples (n ¼ 4) 0.07 0.02 0.09 0.14 beginning of the growing season, it might be that Cd in samples dynamics of isotope composition of other elements, results from large trees originates from a pool accumulated in the stem obtained during this study were complimented by concentra- or shoots, while in young trees a higher proportion is derived tion and isotope data for Os, Pb and Zn in leaves, lichens and from the soil solution, and that there are isotopic differences litter. Obtained as a part of a study on the Os baseline status in between these pools. the environment,16,17 the Os concentration and isotope During the rst part of the growing season, changes in foliar (187Os/188Os ratio) data represent a very detailed temporal Cd concentrations strongly resemble the seasonal dynamics of record of the 2006 growing season using one of the birch trees such nutritional elements as P and Cu (R2 > 0.9). Namely, the sampled later on for the current work. Pb and Zn concentrations highest concentrations occur at the onset of leaf growth, fol- were available from analyses done to assess Cd concentrations lowed by a decline towards June–July probably due to ‘dilution’ (using preparation Procedure A) while corresponding isotope by organic matter. From late August until autumn, Cd concen- data were obtained using MC-ICP-MS and relevant fractions trations in leaves increase somewhat (Table 5), exhibiting from the column separations (Fig. 1). Seasonal changes in positive correlations with the levels of Ca, Pb, Sb and many element concentrations and isotope ratios of these four other elements during this period. This might be caused either elements in birch leaves are shown in Fig. 2 and 3. by accumulation through the concurrent supply of element with Common features of the seasonal dynamics of Os and Pb the ow of nutrients via the root system no longer being (Fig. 2) in birch leaves include the following: counterbalanced by growth dilution, or by surface absorption - Increases in concentrations from spring to autumn, being from aerial sources, or indeed a combination of both. In all ve particularly pronounced for Os; trees sampled to study seasonal effects, there is a clear trend - Isotope composition shis from more to less radiogenic towards heavier isotope composition from May to October. during the growing season; Indeed, d114Cd/110Cd correlates with sampling date having R2 > - Similarities in the isotopic composition of foliage and 0.8 in all trees. Both Cd concentration and isotope composition pendulous lichens at the end of growing season; reproduce well in leaves sampled in 2012 and 2013 (Table 5). - The isotopic composition of mushrooms, used here to estimate isotope composition of soil solution available to Multi-tracer information plants, is signicantly more radiogenic than in leaves with 187 188 206 204 In order to gain a better understanding of observed trends and mean Os/ Os ratio of 1.10 0.08 and mean Pb/ Pb to nd eventual similarities or differences in the seasonal ratio of 19.6 0.2 (not shown in Fig. 1);

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1579 View Article Online

JAAS Paper

Fig. 2 Changes in Os and Pb concentrations and 187Os/188Os and 206Pb/204Pb isotope ratios in birch leaves with squares and diamonds representing leaves from two diﬀerent trees, lichens (open circle) and litter (open triangle). Samples for Os isotope ratio measurements were collected during the 2006 growing season. Samples for Pb isotope ratio measurements were collected during the 2012 (open squares and diamonds) and 2013 (ﬁlled squares and diamonds) growing seasons.

- The isotopic composition of litter is more radiogenic than fractionation effects occurring during either uptake through the of leaves at the end of the growing season. root system or element translocation between different tree A very simplistic model explaining these observations can be compartments will not be seen against this changing radiogenic proposed. Young leaves have isotopic compositions that are a ‘background’. product of element mixing from soil solution (most radiogenic Birch leaves have mostly negative d66Zn/64Zn values (Fig. 3) in a system) and accumulated reserves in the tree from the which is in agreement with most previously published obser- preceding season. Any contribution from a third, aerial, less vations.40,45,46 The lightest Zn was observed during the very rst radiogenic component is low because of the brief exposure few weeks of leaf growth, and accompanied by the lowest period and low surface area of young leaves. As exposure of the measured Zn concentrations. During the rest of season, growing leaf surface to airborne contaminants continues d66Zn/64Zn remains relatively constant. There are reproducible through the entire season, the contribution of this component differences in Zn isotope composition between trees from the to elemental isotopic compositions in foliage increases and same area with birch leaves having higher Zn contents exhib- nally becomes dominant in the autumn. This is conrmed by iting lower degrees of fractionation. Zn in litter is also light the very similar isotope compositions of leaves and lichens – (0.06&), falling into the range of isotopic compositions found organisms with predominantly aerial supplies of both nutrients for birch leaves from the same location, while lichens have and contaminants (Fig. 2). Concentrations of both elements in heavier Zn than tree leaves (0.12&). A very similar Zn isotope lichens are, by several orders of magnitude, higher than in composition was reported in BCR-CRM 482 Lichen by Viers leaves due to much longer exposure time and much higher et al.40 Mushrooms have even heavier Zn isotope composition surface area of the former. (0.26&). For litter, a combination of soil-born contamination and The isotopic compositions of Cd in leaves are heavier than losses of the surface layers most affected by aerial input through those for either mushrooms (0.36&) or lichens (0.07&) at all organic material decay will result in shis back towards the times (Fig. 3). Unlike the situation for the uptake of Os and Pb ‘spring’ isotopic signatures. Due to signicant (more than 10%) radiogenic isotopes, potential Cd isotope fractionation during differences in isotope ratios for end-members, the ‘mixing incorporation in mushrooms cannot be neglected. Since process’ occurring in foliage can be followed through either Os vascular plants and fungi may have mycorrhizal association in or Pb isotopic data that nowadays can be obtained without any their shared root system, the direction, but not necessarily the major analytical challenge. However, potential mass dependent extent, of such fractionation should not be different for plants

1580 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS

Fig. 3 Changes in Zn and Cd concentrations and d66Zn/64Zn and d114Cd/110Cd in birch leaves with squares and diamonds representing leaves from two diﬀerent trees, lichens (open circle) and litter (open triangle). Samples were collected during the 2012 (open sq squares and diamonds) and 2013 (ﬁlled squares and diamonds) growing seasons.

and mushrooms. Hence, neither soil solution nor aerial Cd growing at various heights and between young and old trees isotope signatures are ‘visible’ in foliage, suggesting that (Table 5). signicant isotope fractionation occurs during element incor- Notable differences in Cd (and Zn) isotope composition for poration in birch leaves. It is highly unlikely that the process of leaves collected on the same day from different trees within a airborne Cd uptake through adsorption of particulate matter is very constrained area (Table 5) exceed the overall reproduc- accompanied by changes in isotopic composition. Indeed, if ibility of the analytical method and suggests that the extent of this source plays any notable role in increasing the element fractionation varies between individual birch trees. concentration during the autumn (Fig. 3) this should be man- Fig. 4 summarizes Cd and Zn isotope fractionation ifested by an isotopic shi favouring lighter isotopes. As this is (normalized to one amu) in the various carbon-rich matrices clearly not the case (Fig. 3), an aerial source of Cd (and Zn) can analyzed during this study. A detailed elaboration of the be neglected in birch leaves from this location. The lower Cd concentrations in lichens compared to leaves presents additional conrmation of this conclusion (Table 5). Therefore, soil solution is likely to be the only major Cd source in foliage and the observed fractionation in favour of heavier isotopes must occur in the organism itself, either during root uptake or element translocation through different tree compartments. Similarly, the amelioration of heavy isotope enrichment observed in litter (Table 5 and Fig. 3) cannot be solely explained by soil-borne adulteration, as this would require an unrealisti- cally high contamination load. Therefore it seems that heavier isotopes are preferentially leached during the initial stages of litter decomposition. The available experimental data are insufficient for clarication of the mechanisms responsible for an increasing proportion of heavier isotopes being incorporated during growing season (Fig. 3). It is possible that this phenomenon has common underlying mechanisms with those Fig. 4 Normalized Cd and Zn isotope fractionation in carbon-rich causing differences in isotope composition between leaves matrices.

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1581 View Article Online

JAAS Paper observed similarities (e.g. for lichens, mushrooms and sh for synthetic Cd solutions or chemicals (Table 4), a (0.05–0.1)& liver) or differences (e.g. for birch leaves, kidney or lobster) in range provides a more realistic assessment of overall repro- isotope shis is outside the scope of this paper. However, in ducibility for the entire analytical procedure as applied to spite of chemical similarities between these two metals, there environmental samples. The agreement between d114Cd/110Cd are clear indications of the existence of element-specic biotic found during this study with previously published results varies fractionation mechanisms or isotopic differences in element from very good (for ‘UM-Munster-Cd¨ ’ and NOD-P-1) to less pools available for living organisms. For example, the difference satisfactory (NOD-A-1 and NIST 2711). Accuracy assessment for between normalized isotope fractionation in Fe–Mn nodules for Cd isotope ratio measurements is hampered by the current Cd (+0.03& per amu) and Zn (+3.2& per amu) is almost iden- absence of matrix-matched materials with certied isotopic tical to that in lobster, which certainly deserves further study. information, preferably presented as d-values against a widely accepted isotope standard, and there is an urgent need for such CRMs to be developed and validated. Conclusions Cd in birch leaves fractionated in favour of heavier isotopes, d114 110 The optimized analytical protocol tested in the course of this the mean Cd/ Cd for >80 single-tree and pooled samples & – & study allows precise, reproducible and high-throughput being 0.7 with a range of (0.3 1.3) . This fractionation most measurements of Cd isotope ratio measurements in a wide probably occurs during Cd uptake through the root system and  range of environmental matrices. Different sample preparation element translocation in the birch. There is a shi towards approaches were tested and all but those based on conventional heavier isotope composition in leaves from spring to autumn  MW-assisted digestion were found compatible with the Cd (Fig. 3), an observation con rmed by regular sampling of 5 separation procedure. When available sample size is not a birch trees from the same location. Even when collected on the limiting factor, an ashing step, as an integral component of same sampling date, foliage displays relatively broad variations sample preparation for carbon-rich matrices, improves material in Cd isotope composition between trees growing in close representativeness and allows measurements to be performed proximity to one another. Moreover, there are indications that on samples with Cd concentrations below 0.1 mg kg 1. Cd fractionation in leaves depends upon growing height and  Column separation using the AG-MP-1M ion-exchange resin tree age, though these ndings are based on very limited allows efficient de-contamination from matrix elements and experimental data and should be treated with caution. The shows consistently high Cd (as well as Pb and Zn) recoveries. magnitude of the observed natural variability relative to source However, Sn and Mo – elements causing spectral interferences signatures should be carefully considered before using Cd affecting Cd isotopes – partly co-elute in the analyte fraction. isotopic information in birch leaves, or other bio-indicators, for Though usually negligible in carbon-rich matrices, such inter- environmental exposure assessment, a task that in many situ- ferences may severely limit the accuracy of Cd isotope ratio ations can be better accomplished using traditional radiogenic measurements in soils, sediments and sludge. Presentation of isotope systems. 38 isotope data in the form of normalized d-values (Table 3) helps Quoting Bullen, there clearly remains much to be done to to identify and discard potentially affected ratios. For samples understand the causes of transition and post-transition metal, analyzed in the present study the use of an additional column to stable-isotope fractionation in living systems, and certainly Cd separate Cd from Sn and Mo6 was found to be unnecessary. deserves at least a thorough reconnaissance for a variety of  The use of a desolvating nebulizer extends the concentration species and eld situations. With greater understanding of the range in measurement solution at which reproducible observed variations in isotope composition in terms of known  and accurate Cd isotope ratios can still be guaranteed down to processes, future work will gradually shi toward using isotopic 10 mgl 1 whilst reducing oxide formation that is useful for the signatures to identify as yet unknown or unconstrained analysis of sample solutions containing Mo. As in-run precision processes in plants and other biological systems. is degraded approximately two-fold while signal stabilization and wash-out times are signicantly longer than with the Acknowledgements standard conguration of the MC-ICP-MS sample introduction system, the latter is to be recommended for samples containing ALS Scandinavia AB is gratefully acknowledged for technical relatively high analyte concentrations. Though long term support. MetTrans Initial Training Network (funded by the reproducibility (2s for d114Cd/110Cd) can be as good as 0.02& European Union under the Seventh Framework Programme) is

1582 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014 View Article Online

Paper JAAS acknowledged for nancial support. We wish to thank Katerina 20 W. Abouchami, S. J. G. Galer, T. J. Horner, M. Rehkämper, Rodiouchkina for eld assistance and Dieke Sorlin¨ for help with F. Wombacher, Z. Xue, M. Lambelet, M. Gault-Ringold, sample preparation. The authors also wish to express their C. H. Stirling, M. Schonb¨ ächler, A. E. Shiel, D. Weis and gratitude to Dr Frank Wombacher (Institute for Geology and P. F. Holdship, Geostand. Geoanal. Res., 2013, 37,5–17. Mineralogy of the University of Cologne, Cologne, Germany) 21 I. Rodushkin, E. Engstrom¨ and D. C. Baxter, Anal. Bioanal. and Prof. Cristophe Cloquet (CRPG – Centre de Recherches Chem., 2010, 396, 365–377. Pétrographiques et Géochimiques, Vandœuvre les Nancy, 22 I. Rodushkin, E. Engstrom,¨ D. Sorlin¨ and D. Baxter, Sci. Total France) for providing us respectively with UM-Munster-Cd¨ Environ., 2008, 392, 290–304. standard and NIST SRM 2711reference material. The research 23 F. E. Smith and E. A. Arsenault, Talanta, 1996, 43, 1207–1268. leading to these results has received funding from the People 24 H. M. Skip Kingston and S. J. Haswell, Microwave-Enhanced Programme (Marie Curie Actions) of the European Union's Chemistry: Fundamentals, Sample Preparation, and Seventh Framework Programme FP7 2007–2013, under the REA Applications, American Chemical Society, 1997. grant agreement no 290336. The views expressed in this article 25 S. R. Koirtyohann and C. A. Hopkins, Analyst, 1976, 101, 870– are those of the authors and may not necessarily reect those of 875. the European Union. 26 M. Blanussa and D. Breski, Talanta, 1981, 28, 681–684. 27 J. G. Van Raaphorst, A. W. Van Weers and H. M. Haremaker, References Fresenius' Z. Anal. Chem., 1978, 293, 401–403. 28 W. Yi, A. N. Halliday, D. Lee and M. Rehkämper, Geostand. 1 J. R. Larison, G. E. Likens, J. W. Fitzpatrick and J. G. Crock, Newsl., 1998, 22, 173–179. Nature, 2000, 406, 181–183. 29 K. Inagaki, T. Narukawa, T. Yarita, A. Takatsu, K. Okamoto 2 C. Cloquet, J. Carignan, G. Libourel, T. Sterckeman and and K. Chiba, Anal. Bioanal. Chem., 2007, 389, 691–696. E. Perdrix, Environ. Sci. Technol., 2006, 40, 2525–2530. 30 S. L. C. Ferreira, J. B. de Andrade, M. D. G. a. Korn, 3 B. Gao, Y. Liu, K. Sun, X. Liang, P. Peng, G. Sheng and J. Fu, M. D. G. Pereira, V. a. Lemos, W. N. L. dos Santos, Anal. Chim. Acta, 2008, 612, 114–120. F. D. M. Rodrigues, A. S. Souza, H. S. Ferreira and 4 A. E. Shiel, D. Weis and K. J. Orians, Sci. Total Environ., 2010, E. G. P. da Silva, J. Hazard. Mater., 2007, 145, 358–367. 408, 2357–2368. 31 R. Q. Thompson and S. J. Christopher, Anal. Methods, 2013, 5 J. O. Nriagu and J. M. Pacyna, Nature, 1988, 333, 134–139. 5, 1346. 6 F. Wombacher, T. J. Horner and Z. Xue, in Handbook of 32 A. E. Shiel, J. Barling, K. J. Orians and D. Weis, Anal. Chim. Environmental Isotope Geochemistry, ed. M. Baskaran, Acta, 2009, 633,29–37. Springer Berlin Heidelberg, Berlin, Heidelberg, 2012, pp. 33 F. Wombacher, M. Rehkämper, K. Mezger and C. Munker,¨ 125–154. Geochim. Cosmochim. Acta, 2003, 67, 4639–4654. 7 K. J. R. Rosman, J. R. De Laeter and M. P. Gorton, Earth 34 J. R. D. E. de Laeter, H. Hidaka, H. S. Peiser, K. J. R. Rosman Planet. Sci. Lett., 1980, 48, 166–170. and P. D. P. Taylor, Pure Appl. Chem., 2003, 75, 683–800. 8 C. Cloquet, O. Rouxel, J. Carignan and G. Libourel, Geostand. 35 D. C. Baxter, I. Rodushkin, E. Engstrom¨ and D. Malinovsky, J. Geoanal. Res., 2005, 29,95–106. Anal. At. Spectrom., 2006, 21, 427–430. 9 A. E. Shiel, D. Weis and K. J. Orians, Geochim. Cosmochim. 36 C. R. Quétel, E. Ponzevera, I. Rodushkin, A. Gerdes, Acta, 2012, 76, 175–190. R. Williams and J. Woodhead, J. Anal. At. Spectrom., 2009, 10 A. E. Shiel, D. Weis, D. Cossa and K. J. Orians, Geochim. 24, 407–412. Cosmochim. Acta, 2013, 121, 155–167. 37 A. Stenberg, H. Andrén, D. Malinovsky, E. Engstrom,¨ 11 S. Ripperger, M. Rehkämper, D. Porcelli and A. N. Halliday, I. Rodushkin and D. C. Baxter, Anal. Chem., 2004, 76, Earth Planet. Sci. Lett., 2007, 261, 670–684. 3971–3978. 12 B. Gao, H. Zhou, X. Liang and X. Tu, Environ. Pollut., 2013, 38 T. D. Bullen, in Handbook of Environmental Isotope 181, 340–343. Geochemistry, ed. M. Baskaran, Springer Berlin Heidelberg, 13 B. Markert and V. Weckert, Sci. Total Environ., 1989, 86, 289– Berlin, Heidelberg, 2012, pp. 177–203. 294. 39 F. Larner, M. Rehkämper, B. J. Coles, K. Kreissig, D. J. Weiss, 14 S. C.ˇ Alagić, S. S. ˇSerbula, S. B. Toˇsić, A. N. Pavlović and J. V B. Sampson, C. Unsworth and S. Strekopytov, J. Anal. At. Petrović, Arch. Environ. Contam. Toxicol., 2013, 65, 671–682. Spectrom., 2011, 26, 1627–1632. 15 A. Samecka-Cymerman, K. Kolon and A. J. Kempers, Trees, 40 J. Viers, P. Oliva, A. Nonell, A. Gélabert, J. E. Sonke, 2009, 23, 923–929. R. Freydier, R. Gainville and B. Dupré, Chem. Geol., 2007, 16 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontèr and 239, 124–137. D. C. Baxter, Sci. Total Environ., 2007, 386, 145–158. 41 C. Weinstein, F. Moynier, K. Wang, R. Paniello, J. Foriel, 17 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontér and J. Catalano and S. Pichat, Chem. Geol., 2011, 286,266– D. C. Baxter, Sci. Total Environ., 2007, 386, 159–168. 271. 18 I. Rodushkin, P. Nordlund, E. Engstrom¨ and D. C. Baxter, J. 42 N. Pérez Rodr´ıguez, F. Langella, I. Rodushkin, E. Engstrom,¨ Anal. At. Spectrom., 2005, 20, 1250–1255. E. Kothe, L. Alakangas and B. Ohlander, Environ. Sci. Pollut. 19 D. Malinovsky, I. Rodushkin, D. Baxter and B. Ohlander,¨ Res. Int., 2014, 21, 6836–6844. Anal. Chim. Acta, 2002, 463, 111–124.

This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom.,2014,29,1570–1584 | 1583 View Article Online

JAAS Paper

43 B. M. Ryan, J. K. Kirby, F. Degryse, H. Harris, 45 F. Moynier, S. Pichat, M.-L. Pons, D. Fike, V. Balter and M. J. McLaughlin and K. Scheiderich, New Phytol., 2013, F. Albarède, Chem. Geol., 2009, 267, 125–130. 199, 367–378. 46 E. Couder, PhD thesis, Université catholique de Louvain, 2011, 44 I. Rodushkin, A. Stenberg, H. Andrén, D. Malinovsky and http://hdl.handle.net/2078.1/76412, (accessed June 2014). D. C. Baxter, Anal. Chem., 2004, 76, 2148–2151. 47 M. D. Axelsson, I. Rodushkin, J. Ingri and B. Ohlander,¨ Analyst, 2002, 127,76–82.

1584 | J. Anal. At. Spectrom.,2014,29,1570–1584 This journal is © The Royal Society of Chemistry 2014

Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation

Ilia Rodushkin, Nicola Pallavicini, Emma Engström, Dieke Sörlin, Björn Öhlander, Johan Ingri and Douglas C. Baxter

JAAS

View Article Online PAPER View Journal | View Issue

Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic Cite this: J. Anal. At. Spectrom.,2016, 31, 220 compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation†

ab ab ab b Ilia Rodushkin,* Nicola Pallavicini, Emma Engstrom,¨ Dieke Sorlin,¨ a a b Bjorn¨ Ohlander,¨ Johan Ingri and Douglas C. Baxter

An analytical procedure allowing multi-elemental analyses and isotope ratio measurements of eight of these (B, Cd, Cu, Fe, Pb, Sr, Tl and Zn) in matrices relevant for bio-monitoring using a single high- pressure acid digestion was developed. Method blanks, separation efficiency of matrix elements, repeatability and reproducibility were evaluated using sets of preparation blanks, certified reference materials and duplicate samples prepared and analyzed over a period of several months. The method Received 14th July 2015 was used to assess the natural variability of concentrations and isotopic compositions in bio-indicators Accepted 9th September 2015 (tree leaves, needles and mushrooms, over 240 samples) collected mainly from a confined area in DOI: 10.1039/c5ja00274e North-East Sweden. Ranges found from leaves and needles were compared with data obtained for www.rsc.org/jaas limited numbers of samples collected in Spain, Italy, France, United Kingdom and Iceland.

1. Introduction the last decade.16 For example, in environmental studies involving bio-indicators, Pb11,17 and Sr18 isotopes have been – Isotopic information can be used to aid a wide range of scien- relied upon for at least the past three decades, whereas Os,19 22 tic disciplines, including environmental geochemistry and Cd23 and Tl have been relatively new additions contributing to plant sciences. Such uses include tracing metal contamination the expanding array of investigations. sources/pathways, studying biological processes (nutrient and Another relatively new development is the application of – anthropogenic uptakes/cycles, within plant transport mecha- multi-tracer studies,2,24 28 as source tracing in two or three- nisms),1 remediation and geographical provenance.2–5 Vari- dimensional space potentially allows distinguishing samples ability in the isotopic composition of radiogenic elements, e.g. having overlapping isotope signatures for a single element.16 In lead (Pb), strontium (Sr) and osmium (Os), has been frequently an exemplary study, Sherman et al.24 used Pb, Sr and Hg isotopes utilized in environmental studies6–12 and application niches in precipitation to identify the signature of coal combustion. continue to grow. Mass-dependent fractionation of boron (B) d11B coupled with 87Sr/86Sr ratios has also proved to be effective and other light elements (carbon, oxygen and nitrogen) has now in determining the origin of coal combustion residuals.25 been used for provenance studies, tracing pollution sources and Apart from anthropogenic assessment, a great deal of atten- water mixing for decades.2,13–15 tion in recent studies has been given to the study of the biological Enhancements in measurement precision due to develop- processes responsible for variations in isotopic compositions in – ments in analytical instrumentation, e.g. the advent of multiple plants and animals.27,29 35 In the plant sciences, the major focus collector inductively coupled plasma mass spectrometry (MC- has been on elements essential for plant growth. With the help of ICP-MS), as well as continual rening of preparation/separation isotopic data, Rosner et al.32 concluded that B assimilation by and pre-concentration methods1 have allowed inclusion of plants is directly inuenced by the local conditions (both natural heavier stable elements in the ‘isotope toolbox’ and the number and anthropogenic). This has been further conrmed in another of published stable isotope studies has grown exponentially in recent multi-isotope study where the coupling of Sr and B isotope ratios has been used to trace the geographic origins of coffee beans.2 Jouvin et al.31 described two models for fractionation of a ˚ ˚ Division of Geosciences, Lulea University of Technology, S-971 87 Lulea, Sweden the micro-nutrients Cu and Zn during uptake by plants, bALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Lulea,˚ Sweden. proposing different fractionation patterns for different uptake E-mail: [email protected] † Electronic supplementary information (ESI) available. See DOI: strategies. Uptake mechanisms for Cu, Fe and Zn by plants have 29–31,36,37 10.1039/c5ja00274e been the focus of a number of studies.

220 | J. Anal. At. Spectrom.,2016,31,220–233 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

Wider applications of isotope signatures in environmental hydrogen uoride (HF, 48%, Merck, Darmstadt, Germany) used geochemistry and plant sciences are oen hampered by the in this work were all of analytical grade. Water used in all high cost of instrumentation, the need for oen tedious, elab- experimental procedures was de-ionized Milli-Q water (Milli- orative and time-consuming sample preparation and analyte pore, Bedford, MA, USA) puried by reverse osmosis followed by separation schemes,38,39 as well as challenges to verify the ion-exchange cartridges. For dilution of sample digestion accuracy of analytical methods.16,40–42 As a result, many studies aliquots intended for B isotope ratio measurements, water was are still based on very limited numbers of samples and oen further puried by sub-boiling distillation in Teon stills consider only a single isotope system, which may affect the (Savillex, Minnetonka, MN, USA). AG MP-1M ion-exchange resin transferability of any conclusion drawn. Clearly, a massive (macroporous, 100–200 dry mesh size, 75–150 mm wet bead size, amount of isotopic information would need to be acquired for a Bio-Rad Laboratories AB, Solna, Sweden) was cleaned by soak- meaningful assessment of the natural variability in various eco- ing in 0.7 M HNO3 followed by rinsing with Milli-Q water and systems and to identify the factors responsible for such vari- loading as slurry into 2 mL columns. Pre-packed 2 mL columns ability. In approaching an investigation, special importance with Sr-Spec resin (Eichrom Technologies, IL, USA) were used as must also be given to the geographical scale of the study or else supplied. important local details may resist detection.11 Thus the possi- The following chemicals were used as ‘d-zero’ standards: bility to obtain isotope data for many elements coupled to NIST SRM 3108 Cd solution Lot 130116, NIST SRM 976 Cu concentration information in a reasonable time and without the standard solution, NIST SRM 951a – boric acid, NIST SRM 981 need to use several types of instruments is of extreme value. common lead, and NIST SRM 987 (NBS-987) – strontium The aim of this work is to provide a detailed description of an carbonate (all from the National Institute of Standards and analytical procedure allowing for MC-ICP-MS isotope ratio Technology, Gaithersburg, MD, USA); IRMM 3702 zinc solution measurements of at least eight elements from a single digestion and IRMM-014 Fe metal (both from the Institute for Reference of a limited amount (approximately 0.5 g) of plant material. The Materials and Measurements, Geel, Belgium). For Ag and Tl analytical procedure was applied to a substantial number of isotopic analyses, commercial standards (1000 mg L 1 mono- leaves, needles and mushrooms (frequently used bio-indica- elemental solutions supplied by Ultra Scientic, North Kings- tors11,43,44) collected from urban environments in several Euro- town, RI, USA; Ag: Lot M00474; and Tl: Lot L00709) were used as pean countries to assess the extent of the natural variability of B, ‘d-zero’ standards. Cd, Cu, Fe, Pb, Sr, Tl and Zn isotope compositions as well as seasonal variations. 2.3. Samples A large part of the samples was collected in the city and suburbs 2. Experimental of Lule˚a (northern Sweden), a medium-sized town (population 2.1. Instrumentation approximately 75 000) located in the province of Norrbotten. All isotope ratio measurements except for B in those samples The study area lies almost entirely on a 1.9 Ga granitic bedrock 47 low in the analyte were performed using a NEPTUNE PLUS with a minor metasedimentary constituent. Clay and silt loam 48 (Thermo Scientic, Bremen, Germany) MC-ICP-MS instrument are the main soil constituents even though it is important to operated with variously congured introduction systems, note that some of the soil components in the urban areas can be ˚ including Aridus II (Teledyne CETAC Technologies, Omaha, NE, non-native. The area surrounding the town of Lulea is heavily USA) and Apex (Elemental Scientic, Omaha, NE, USA) des- industrialized, with steelworks as the dominant local industry. – olvating nebulizers. Cup congurations used, operating condi- The sets of biological samples collected during 2013 2015 tions and measurement parameters are given in Table 1. from approximately 50 individual locations included common B isotope ratio measurements in some samples and all birch (Betula pubescens) leaves, Norway spruce (Picea abies) measurements of elemental concentrations were performed by needles and fruit bodies of edible mushrooms (Boletus edulis, double-focusing sector eld ICP-MS (ICP-SFMS; ELEMENT XR, Leccinum scabrum, Leccinum versipelle, Leccinum aurantiacum Thermo Scientic).23 Methane addition to the plasma was used and Suillus variegatus). At a few locations leaves of oak (Quercus), to decrease formation of oxide-based spectral interferences, aspen (Populus tremula) and rowan (Sorbus aucuparia) were also improve sensitivity for elements with high rst ionization sampled. The amounts of material from each sampling location – potentials, and to minimize matrix effects.45 Operating condi- (corresponding to approximately 0.5 1.5 g dry weight per – tions and measurement parameters for concentration matrix) consisted of 10 50 leaves (depending on the growth ff measurements were as described in a previous study.46 stage) collected from di erent branches/trees, needles from the A laboratory UltraCLAVE single reaction chamber microwave last year grown on parts of lower branches and mushrooms digestion system (Milestone, Sovisole, Italy) was used for collected under sampled trees (where available). Sampling was sample digestions. performed either at the beginning of the growing season (May- early June, birch leaves only) or just before senescence (early September), though in the majority of locations samples were 2.2. Chemicals and reagents taken from different trees in spring and autumn. All samples

Nitric acid (HNO3) and hydrochloric acid (HCl), both from were collected wearing powder-free laboratory gloves into zip- Sigma-Aldrich Chemie GmbH (Munich, Germany) and lock plastic bags marked with geographic coordinates, the type

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 220–233 | 221 View Article Online

JAAS Paper

Table 1 MC-ICP-MS operating parameters and measurement conditions for isotope ratio measurementsb

Sample Cup conguration Conguration of the Resolution Integration uptake rate Elementa introduction system mode time (s) (L min 1) L4 L3 L2 L1 C H1 H2 H3 H4

Cd/Ag Aridus/Apex desolvating Low 0.524 0.04–0.06 107Ag 108Cd 109Ag 110Cd 111Cd 112Cd 114Cd 116Cd 117Sn systems, self-aspirating (108Pd) (110Pd) (112Sn) (114Sn) (116Sn) micro-concentric PFA nebulizer, X-type skimmer cone Zn/Cu Pumped Micromist Medium 0.262 0.20–0.25 — 63Cu 64Zn 65Cu 66Zn 67Zn 68Zn 70Zn — nebulizer, double spray (64Ni) (70Ge) chamber, H-type skimmer cone Fe/Ni Pumped Micromist Medium 0.262 0.20–0.25 54Fe — 56Fe 57Fe — 60Ni 61Ni 62Ni — nebulizer, double spray (54Cr) chamber, H-type skimmer cone Sr/Zr Pumped Micromist Low 0.262 0.20–0.25 82Kr 83Kr 84Sr 85Rb 86Sr 87Sr 88Sr 90Zr 91Zr nebulizer, double (84Kr) (86Kr) (87Rb) spray chamber, H-type skimmer cone B Pumped Micromist Low 0.524 0.40–0.50 —— 10B ———11B —— nebulizer, mini cyclonic spray chamber, X-type skimmer cone Pb/Tl Aridus/Apex desolvating Low 0.524 0.04–0.06 — 202Hg 203Tl 204Pb 205Tl 206Pb 207Pb 208Pb — systems, self-aspirating (204Hg) micro-concentric PFA nebulizer, X-type skimmer cone a One element is used as the internal standard for the second element, except for Zr which is only used as the internal standard. b RF power: 1400–  1  1  – 1  1450 W. Coolant gas ow: 15 L min . Auxiliary gas ow: 1.4 L min . Sample gas ow: 0.9 1.25 L min . Additional gas ow (N2, Aridus and Apex) 0.01–0.02 L min 1. Ion lens settings: adjusted daily to obtain maximum sensitivity and signal stability. Zoom optic settings: adjusted daily to obtain maximum resolution. Number of blocks: 9. Number of cycles per block: 5. Number of integrations: 3–5. Amplier rotation: le.

of sample and collection date. Sampling locations were chosen (Table 2). Note that none of the materials mentioned above has on the base of a sampling grid of 1 km2 mesh size, covering a a certied isotopic composition. total area of ca. 200 km2. Sampling height was limited to roughly 2.5 m from the ground for all leaf and needle samples. 2.4. Sample preparation Samples from other geographic locations – Genoa (Italy), suburbs of Barcelona (Spain), the city of Reykjavik and location All sample manipulations were performed in clean laboratory Þorsm´ ork¨ (Iceland), the city of Paris (France) and suburbs of areas (Class 10000) by personnel wearing clean room gear and Birmingham (United Kingdom) – were collected during autumn following all general precautions to reduce contaminations.49 2014 and spring 2015 though in signicantly less numbers. All laboratory ware coming into contact with samples/sample When no birch or spruce trees were found (Italy and Spain), digests was soaked in 0.7 M HNO3 (>24 h at room temperature) leaves and needles of other tree species, oak (Quercus), olive and rinsed with MQ water prior to use. (Olea europaea), and pine (Pinus sylvestris), were collected. Mushrooms were mechanically cleaned from external exog- 3 For verifying the method, a set of certied reference mate- enous material and divided into approximately 1 cm pieces rials (CRMs) has been included and processed in parallel to using a ceramic knife on a Teon plate. Samples were then “natural samples” throughout the procedure (digestion, sepa- dried at 50 C to constant weight, homogenized by crushing in ration and analysis): ERM BB186 pig kidney (Institute for plastic bags and stored air-tight packed at room temperature. Reference Materials and Measurements), TORT-1 lobster 2.4.1. Sample digestion. About 0.5 g of dried material from hepatopancreas and NASS-4 open ocean water (National each sample bag was accurately weighed into a 12 mL Teon  Research Council of Canada, Ottawa, Canada), NIST SRM 1547 vial before addition of 5 mL 14 M HNO3.Aer the initial peach leaves (National Institute of Standards and Technology) oxidation of organic matter subsided, vials were gently agitated and NJV 94-5 wood fuel (Swedish University of Agricultural and any solid material adhering to the walls was washed down Sciences, Sweden), providing representative variability in the by an additional 1 mL of HNO3. Vials (up to 40 per batch) were concentrations of analytes and ranges of isotope compositions placed into a carousel with numbered slots, which was then loaded into the Teon-coated UltraCLAVE reaction chamber

222 | J. Anal. At. Spectrom.,2016,31, 220–233 This journal is © The Royal Society of Chemistry 2016 Paper hsjunli h oa oit fCeity2016 Chemistry of Society Royal The © is journal This

Table 2 Figures of merit for the analytical method and results for CRMsa

Total procedural Minimum concentration for Element LOD (mgg 1 ) blank (ng) isotope measurements (mgg 1 ) NIST 1547 (n ¼ 16) IRMM-BB186 (n ¼ 8) NJV 94-5 (n ¼ 10) NRCC-TORT-1 (n ¼ 8)

B 0.07 40 2 29.1(1.8) 29(2) 0.58(0.12) – 5.32(0.20) – 5.21(0.27) – Cd 0.001 0.15 0.05 0.0254(0.0011) 0.026(0.003) 1.04(0.04) 1.09(0.05) 0.269(0.013) 0.27(0.028) 26.2(1.0) 26.3(8.0) Cu 0.01 8 2 3.45(0.16) 3.7(0.4) 34.4(1.8) 36.5(1.8) 2.02(0.15) 2.2(0.30) 421(22) 439(5.0) Fe 0.05 450 20 209(13) 218(14) 244(15) 255(13) 71.8(3.7) 70(16) 196(9) 186(5.9) Pb 0.002 0.25 0.025 0.795(0.066) 0.87(0.03) 0.036(0.004) 0.040(0.005) 0.652(0.040) 0.68(0.025) 9.32(0.42) 10.4(1.9) Sr 0.02 20 2 53.5(2.5) 53(4) 0.204(0.034) – 11.6(0.4) – 108(6) 113(4.4) Tl 0.0001 0.08 0.01 0.021(0.002) 0.015(0.001) – 0.166(0.006) – 0.005(0.001) – Zn 0.1 90 5 17.2(0.8) 17.9(0.4) 128(7) 134(5) 37.7(1.9) 38(8.5) 168(8) 177(5.6) Ag 0.0004 0.2 0.025 0.0016(0.0004) 0.0009(0.0003) – 0.020(0.002) – 6.91(0.41) – Hg 0.001 0.05 0.5 0.034(0.003) 0.031(0.007) 0.017(0.002) 0.023(0.011) 0.021(0.002) – 0.309(0.013) 0.330(0.006)

Mean instrumental repeatability at the 2 SD level for Reproducibility at the 2 SD level for NIST 1547 IRMM-BB186 NJV 94-5 NRCC-TORT-1 Isotopes n > 240 sample duplicates (&) n > 10 sample duplicates (&) (n ¼ 16) (n ¼ 8) (n ¼ 10) (n ¼ 8)

d11B 1.8 2.5 40.3(2.6) 9.1(7.2) 13.8(4.4) 26.1(4.1) d114 Cd 0.047 0.158 0.068(0.148) 0.455(0.065) 0.052(0.084) 0.143(0.053) d65Cu 0.045 0.130 0.433(0.205) 2.669(0.245) 0.395(0.265) 0.083(0.097) d56Fe 0.038 0.072 0.304(0.054) 2.066(0.144) 0.160(0.077) 0.200(0.094) 208 Pb/206 Pb 0.079 0.330 2.482(0.010) 2.478(0.048) 2.424(0.011) 2.464(0.005) 206 Pb/207 Pb 0.094 0.360 1.213(0.006) 1.191(0.034) 1.159(0.007) 1.192(0.004) 87 Sr/86Sr 0.036 0.210 0.71339(0.00009) 0.71003(0.00109) 0.73151(0.00013) 0.70925(0.00009)

.Aa.A.Spectrom. At. Anal. J. d205 Tl 0.054 0.115 0.269(0.153) NA 0.244(0.123) NA d66Zn 0.032 0.084 0.337(0.144) 0.608(0.105) 0.075(0.101) 0.616(0.099) d109 Ag 0.048 NA NA NA NA NA a The uncertainty given in parentheses is either the reproducibility expressed as twice the standard deviation (SD) for n replicates performed by at least two diﬀerent operators or the condence interval for the certied value. Data for CRMs are presented in the order: experimental mean values (2 SD) certied value (95% condence interval). ,2016, 31 View ArticleOnline 220 , – 3 | 233 JAAS 223 View Article Online

JAAS Paper

containing a deionized water–H2O2 mixture (10 : 1 v/v). The this column contain >95% of the original Ag and can be used for chamber was pressurized with compressed argon and the pre- purication of this element by loading in 4 mL of 2 M HCl onto programmed digestion cycle (30 min ramp to 220 C followed by AG MP-1M resin-containing columns and eluting with 14 M 23 20 min holding time at that temperature) was initiated. The HNO3. All columns can be re-used several times, although the total processing time, including cooling and subsequent efficiency of matrix separation gradually deteriorates aer 5–6 transfer and dilution of sample digests to a nal volume of 10 cycles with the matrix/Cu and Zn/Cd cut-off affected the most. It mL into storage polypropylene tubes, was approximately three should be noted that approximately 0.1% of the initial Sr and hours per digestion batch. 0.2–0.4% of the initial Pb remain on Sr-specic columns and In some samples, minor quantities of white precipitates of therefore may affect subsequent separations for samples with siliceous material were formed. Rapid dissolution of the much lower analyte concentrations or grossly different isotopic precipitate was achieved aer addition of 30 mL of 16 M HF and compositions. manual agitation for a few minutes. Sets of method blanks and All separated analyte fractions except those for Sr and Hg CRMs were prepared with each batch of samples. were evaporated to dryness and dissolved in 2–10 mL of 0.3 M

As for any of the wide variety of methods used for the prep- HNO3. (An aliquot of 14 M HNO3 was pipetted directly onto the aration of biological matrices for subsequent ICP-SFMS analysis solid residue as a rst step, allowed to react for 15–25 min, and in the laboratory, e.g. ashing, hot-block and microwave diges- then diluted appropriately by addition of MQ water.) 0.1 mL 10,50,51 tions, and high pressure ashing, digestion using the aliquots of separates were diluted 50-fold with 1.4 M HNO3 and UltraCLAVE has its merits and limitations. The former include analyzed by ICP-SFMS (same approach as for sample digests). complete oxidation of carbonaceous material thus ensuring This provides: (I) information on analyte contents needed to negligible eﬀects of undigested organics on the subsequent prepare concentration- and acid strength-matched solutions for separation procedure, applicability to all matrixes tested in this isotope ratio measurements; (II) direct assessment of analyte study, ease of sample handling/loading (limited material recovery; (III) control over separation eﬃciency from matrix manipulation and thus lowered risk of contamination), and elements; and (IV) a test for the presence of potentially spec- relatively high throughput. The major limiting factor is the trally interfering elements either from the sample matrix or amount of material that can be digested in a 12 mL vessel from handling contamination. (approximately 0.5 g dried material), which may require pro- The absence of articially introduced fractionation during cessing parallel digestions for samples low in some analytes, separation/evaporation and analysis stages was tested by sepa- though this approach was not required in the present work. rating a mixture of ‘d-zero’ standards with concentrations

Aliquots of digests were diluted 50-fold with 1.4 M HNO3, typical for birch leaves. providing a total digestion factor of approximately 1000 v/m, 2.4.3. Isotope ratio measurements and data evaluation. As and analyzed by ICP-SFMS using a combination of internal pre-analysis of puried fractions by ICP-SFMS provided proof of standardization and external calibration.50 Portions of diluted the absence of notable concentrations of elements forming digests remaining aer this analysis (approximately 6 mL) were isobaric interferences with analyte isotopes (e.g. Cr or Pd), no used for B isotope ratio measurements either directly or aer Faraday cup was used to monitor these interferences on-line additional dilution. The rest of the original sample digest was (Table 1). At least 1.5 h instrumental heating-up and stabiliza- evaporated to dryness in a 25 mL Teon beaker at 95 Cona tion time with plasma on was allowed before starting the opti- ceramic-top hot-plate, followed by dissolution in 4 mL of 9.6 M mization of operation parameters and mass-calibration. HCl, thus being ready for subsequent purication. For B, MC-ICP-MS measurements were performed on 2.4.2. Analyte purication. Briey, aer loading evaporated unseparated digests which can be diluted to provide at least 20 sample digests taken up in 4 mL of 9.6 M HCl onto AG MP-1M mgL 1 concentrations in measurement solution yielding >1 V resin-containing columns and rinsing out matrix elements with intensity for 11B. Profound B memory eﬀects in the introduction the same acid, Cu, Fe, Zn, Cd + Tl and Hg are quantitatively system were minimised by using a low volume spray chamber,  eluted from the resin using rst HCl of decreasing molarities diluting all samples, standards and blanks in 1.4 M HNO3, and then a mixture of 6 M HNO3 containing traces of HF. In employing higher sample uptake rates and increasing the contrast to matrix separation in geological/industrial materials, washing time between samples and standards to 200 s. This there is no risk of overloading the resin capacity with any of the ensures that the instrumental 11B blank is below 0.02 V before analytes and therefore the entire digest volume can be used. measurement of the next solution is started. Compared to a Cd separation procedure,23 sample loading in For the remaining elements, separated fractions were more concentrated HCl allows separation of Cu, Fe and Zn diluted to equal concentration levels; because of limited analyte using the same column, while neither Ag nor Pb is eﬃciently content in some biological samples, several xed measurement retained by the resin. The sample load and matrix wash frac- concentrations were used with minimum requirements listed in tions (collected into a 25 mL Teon beaker) contained >99.5% Table 2. A set of bracketing standards matching samples in of initial Sr and >85% of initial Pb. Aer evaporation and re- terms of analyte and internal standard concentrations as well as dissolution in 4 mL of 7 M HNO3, Sr and Pb were separated acid strength was prepared for each isotope system. High using Sr-specic columns, by selective elution with 0.05 M degrees of concentration matching between samples and

HNO3 and 0.1 M ethylenediaminetetraacetic acid (EDTA), bracketing standards (better than 10%) are needed for accu- respectively. The sample load and matrix wash fractions from rate measurements of Cd, Ag, Cu and Zn. Measuring samples

224 | J. Anal. At. Spectrom.,2016,31, 220–233 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS against standards with halved or doubled analyte concentra- Three sample solutions were analyzed between two stan- tions will result in up to 0.3& errors. The matching tolerance dards. Two consecutive measurements were performed for each for Sr, Pb, Tl and Fe isotope ratio measurements is signicantly solution in the sequence. The MC-ICP-MS soware option of broader, where up to 30% concentration difference between excluding pass, run and block outliers was deactivated as it was samples and standards will not affect results notably. found that the presence of some outliers actually improves the The best signal stability and in-run precision in isotope correlation between instrumental mass bias levels for analyte ratios is obtained with an introduction system consisting of a and internal standard isotope ratios. self-aspirating PFA nebulizer with approximately 0.05 mL Data evaluation, including correction for blanks, spectral min 1 sample uptake, cyclonic/Scott double spray chamber interferences and instrumental mass bias, was performed off- arrangement and H-skimmer cone. Sample throughput can be line using commercially available spreadsheet soware. increased by almost 40% by increasing the sample uptake Instrumental mass bias was corrected in a two-step procedure four-fold using a peristaltic pump (due to much shorter using rst exponential correction by internal standardization solution-in, signal stabilization and wash-out times) with less using an algorithm proposed by Baxter et al.55 followed by pronounced matrix effects as an extra benet. However this standard–sample bracketing (SSB). Mean ratios from two option requires replacing PFA with a Micromist (Glass consecutive measurements of the rst and the third samples in Expansion Ltd, West Melbourne, Australia) nebulizer. For each analytical block (standard 1–sample 1–sample 2–sample ultra-trace elements, the intensity provided by such a cong- 3–standard 2) were calculated against ratios for standards 1 and uration is insufficient and the use of desolvating nebulizers 2, respectively. For the second sample, mean ratios from the and the X-skimmer cone is mandatory for Cd, Ag, Pb and Tl in bracketing standards were used assuming linear changes in the majority of samples, providing, depending on the analyte, instrumental mass bias persisting aer internal standard seven- to 25-fold intensity gains. Intensity can be increased correction. Results from two consecutive measurements of each even further (by a factor of 3–4) by increasing the sample sample allow calculation of mean d-values and respective uptake of desolvating nebulizers to 0.15–0.20 mL min 1, standard deviations for all isotope ratios that are less affected by though this results in increasingly unstable and ‘spiky’ variations caused by imperfect amplier gain calibration. signals. Even with 0.05 mL min 1 sample uptake typical in- For Ag, B, Cd, Cu, Fe and Zn d-values were calculated using run precision is almost three times poorer than with the the general formula widely adopted in isotopic studies:  " # standard introduction system con guration. ðxM=yMÞ The use of Aridus was found to be unsuitable for Hg isotope dx=yM ¼ sample 1 1000 ðxM=yMÞ ratio measurements because Hg vapour is lost from the system d0standard while passing the desolvating membrane. As it was impossible where xM and yM correspond to the two different isotopes of the to pre-concentrate Hg by evaporation of the puried fraction x y element of interest, the ( M/ M)sample value refers to the (again because of analyte losses) and the need to further dilute 7 x y d measured ratio and ( M/ M) 0standard is the isotope ratio of the M HNO3 matrix of this fraction prior to analysis, even the use of bracketing standard used as delta-zero. When the d-value refers the APEX did not allow reliable Hg isotope ratio measurements to a ratio of a heavier to a lighter isotope, a positive d-value in samples collected for this study. Though other means of Hg corresponds to an enrichment in the heavier isotope compared introduction to MC-ICP-MS have been suggested (e.g. cold to the standard. Calculated d-values for different Cd, Fe and Zn 52–54 vapour and purge-and-trap ), these were not tested here and isotope ratios in each sample were normalized by the respective plans to measure Hg isotopes were abandoned. mass difference providing an additional aid to check for m 1 For samples with B concentrations <20 gg , isotope ratio internal consistency of isotope data.23 measurements were performed by ICP-SFMS using the same Further details on isotope ratio measurement, data process-  con guration of the introduction system as for MC-ICP-MS and ing and corrections can be found in previous studies.23,40,56,57 paying special attention to avoid tailing from Ar4+ spectral interference appearing on the low-mass side of the 10B isotope in the 5% acquisition window. Though the in-run precision of 3. Results and discussion single collector instrumentation is inferior to that of MC-ICP- MS by a factor of 5–10, it nevertheless allows isotope ratio 3.1. Performance of the separation procedure measurements at much lower B concentrations in measure- Wider use of isotope ratio measurements by MC-ICP-MS, ment solutions. B isotope ratio measurements in NIST 1547 and especially in multi-tracer studies, has resulted in the develop- NASS-4 CRMs were performed during every measurement ment of numerous matrix separation/analyte pre-concentration sequence, with results from both techniques agreeing to within schemes evaluated for various elements and sample types. The measurement uncertainty. Additionally, Sr and Pb isotope ratio complexity of the schemes varies from the very simple, i.e. a measurements in samples low in analytes (<10 mgg 1 and <0.5 single pass through a commercially available pre-packed mgg 1, respectively) were repeated by ICP-SFMS using solution column containing a specic ion-exchanger (e.g. Sr,58 U, Th59,60), remaining from MC-ICP-MS measurement sessions and an to the very time-consuming and elaborate, consisting of several introduction system consisting of a self-aspirating PFA nebu- different purication steps (e.g. Cd, Mo38,39,61). As the number of lizer, a cyclonic/Scott double spray chamber arrangement and steps increases, so do the risks associated with potential sample an H-skimmer cone. contamination and/or articially induced isotope fractionation.

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 220–233 | 225 View Article Online

JAAS Paper

Fig. 1 Flowchart of the puriﬁcation procedure.

Application of published procedures to new types of samples blanks handled as samples, are listed in Table 2. In spite of oen requires re-validation because the sample matrix, the extensive sample handling, method blanks, with few excep- concentrations of analyte and interfering elements, as well as tions, correspond to <2% contribution to concentrations found the particulars of the sample digestion approach may severely in samples containing the minimum analyte content required affect separation performance. for isotope ratio measurements (Table 3) and therefore have a The analyte purication scheme used in this study is an negligible effect on measured ratios. Analyte recovery was above amalgamation of several published separation procedures 95% from the majority of samples and CRMs separated during merged to maximize separation efficiency and reduce proce- this study with the sole exception of Pb (above 85%). Though dural time, while ensuring high analyte recoveries and low the lower recovery of Pb implies a risk for articially introduced contamination levels. Based on published isolation procedures mass-dependent fractionation, the bias introduced can be for Cu, Fe and Zn,62–64 Cd,7,23 as well as Sr and Pb,5,65 introduc- tolerated given the range of radiogenic Pb ratios found. Samples tion of a few prudent modications extended the number of with recoveries below these thresholds were either re-prepared elements isolated from the single sample digest. Volumes and re-analyzed (when the amount of sample collected was needed at each step were obtained through replicate calibration sufficient) or results for affected analytes were excluded from of columns by collecting the sample load, matrix wash and all the following evaluations. elution fractions with 1 mL resolution before analysis by ICP- SFMS to obtain detailed elution proles for all elements present in these samples. A ow chart depicting all steps of the puri- 3.2. Precision cation procedure is shown in Fig. 1. Instrumental repeatability was estimated as twice the standard Average method blanks for the entire procedure, assessed by deviation (SD) of duplicate consecutive measurements of a applying all preparation and separation steps to a set of reagent single sample preparation. The mean instrumental

226 | J. Anal. At. Spectrom.,2016,31, 220–233 This journal is © The Royal Society of Chemistry 2016 Paper hsjunli h oa oit fCeity2016 Chemistry of Society Royal The © is journal This Table 3 Statistical summary (mean(standard deviation)median min. O max.) for concentration and isotope data in various bio indicatorsa

Birch leaves Birch leaves Spruce needles Mushrooms Birch leaves Leaves Lule˚a, May– Lule˚a, Lule˚a, Lule˚a, Leaves Genoa, Iceland, Barcelona, Birch leaves June September September September May–June Leaves Paris, Needles Paris, August October Birmingham, Parameter (n ¼ 54) (n ¼ 58) (n ¼ 31) (n ¼ 30) (n ¼ 10) June (n ¼ 9) June (n ¼ 4) (n ¼ 8) (n ¼ 7) June (n ¼ 3)

B(mgg 1 ) 13(6)11 29(15)25 14(6)14 1.7(1.6)1.0 48(26)43 32(10)29 30(20)23 50(16)44 62(53)45 17(8)19 7 O 40 7 O 76 3 O 31 0.2 O 615O 88 18 O 46 13 O 58 36 O 75 13 O 160 8 O 25 d11B(&) 7.7(6)9.4 7.9(8)8.0 18(9)20 1(10)–4 21(15)18 6.4(5.5)6.0 25(8.6)22 13(10)16 22(9)18 16(5)16 4 O 21 7 O 25 2 O 37 12 O 29 4 O 40 1 O 16 18 O 37 6 O 26 14 O 36 12 O 22 Cd (ng g 1 ) 380(150)330 280(130)240 57(43)40 1600(1500)1100 490(470)420 560(550)330 60(80)30 150(70)120 61(59)28 43(15)34 190 O 820 74 O 630 13 O 160 50 O 5100 1 O 1100 30 O 1600 5 O 180 6 O 180 6 O 180 31 O 65 d114 Cd (&) 0.42(0.21)0.44 0.50(0.15)0.49 0.10(0.25)0.11 0.16(0.22)0.23 0.13(0.13)0.10 0.09 (0.11)0.07 0.15 (0.06)–0.15 0.51(0.29)0.52 0.05 (0.14)–0.01 0.40 (0.33)0.25 0.09 O 0.73 0.17 O 0.95 0.42 O 0.56 0.52 O 0.43 0.01 O 0.49 0.14 O 0.21 0.21 O 0.09 0.11 O 0.96 0.23 O 0.09 0.18 O 0.79 Cu (mgg 1 ) 10(3.4)10 4.7(0.8)4.6 3.2(0.9)3.1 46(27)42 16(15)13 11(7.1)10 13(9.1)12 5.7(1.3)5.1 11(9.0)6.4 7.0(3.3)6.9 4.0 O 17 3.1 O 6.8 2.1 O 5.2 10 O 120 4 O 58 4.5 O 28 3.3 O 24 4.6 O 7.9 3.3 O 24 3.3 O 11 d65Cu (&) 0.44(0.24)–0.42 0.55(0.27)–0.50 1.17(0.43)–1.16 1.17(0.71)–1.26 0.50(0.52)–0.59 0.26(0.22)–0.34 0.24(0.34)–0.33 0.30(0.41)–0.26 0.14(0.50)–0.14 0.28(0.27)–0.19 1.0 O 0.05 1.3 O 0.09 2.0 O 0.41 2.3 O 0.15 1.3 O 0.28 0.55 O 0.07 0.51 O 0.10 0.82 O 0.19 1.1 O 0.27 0.58 O 0.06 Fe (mgg 1 ) 150(60)140 310(210)240 150(110)120 260(550)50 250(210)210 94(31)95 440(310)420 760(440)680 240(230)160 120(50)110 70 O 360 110 O 1500 40 O 440 20 O 3500 110 O 1500 50 O 150 150 O 750 290 O 1300 55 O 750 77 O 180 d56Fe (&) 0.30(0.14)–0.28 0.21(0.18)–0.20 0.35(0.34)–0.27 0.35(0.35)–0.35 0.10(0.14)–0.13 0.33(0.13)–0.30 0.12(0.15)0.12 0.09(0.10)–0.06 0.04(0.10)–0.04 0.32(0.23)–0.26 0.58 O 0.03 0.84 O 0.09 1.32 O 0.05 1.11 O 0.25 0.31 O 0.09 0.53 O 0.12 0.05 O 0.26 0.31 O 0.01 0.17 O 0.09 0.57 O 0.12 Pb (ng g 1 ) 170(130)110 540(330)480 160(110)130 190(150)130 250(150)210 570(300)520 800(1500)580 73(70)55 2800(5400)750 290(150)260 60 O 590 120 O 2100 47 O 540 15 O 510 40 O 480 340 O 1400 470 O 2200 25 O 240 220 O 15 000 130 O 470 208 Pb/206 Pb 2.44(0.02)2.44 2.43(0.02)2.42 2.42(0.03)2.42 2.44(0.02)2.43 2.46(0.02)2.47 2.45(0.01)2.44 2.44(0.02)2.45 2.45(0.02)2.46 2.44(0.01)2.44 2.43(0.01)2.42 2.386 O 2.477 2.396 O 2.464 2.344 O 2.472 2.405 O 2.480 2.431 O 2.489 2.435 O 2.459 2.417 O 2.466 2.423 O 2.477 2.433 O 2.466 2.417 O 2.435 206 Pb/207 Pb 1.19(0.03)1.19 1.17(0.02)1.17 1.18(0.02)1.18 1.18(0.02)1.18 1.18(0.02)1.18 1.18(0.01)1.18 1.16(0.01)1.16 1.18(0.02)1.18 1.17(0.02)1.17 1.16(0.01)1.17 1.134 O 1.252 1.126 O 1.220 1.145 O 1.259 1.155 O 1.240 1.163 O 1.210 1.164 O 1.199 1.154 O 1.167 1.149 O 1.223 1.158 O 1.218 1.154 O 1.170 Sr (mgg 1 ) 29(14)26 36(12)35 18(16)15 0.53(0.38)0.37 73(38)64 150(110)150 110(90)90 71(23)69 56(60)23 21(20)12 12 O 70 17 O 70 2 O 63 0.1 O 1.8 29 O 150 14 O 380 25 O 230 49 O 110 11 O 160 6 O 46 87 Sr/86Sr 0.729(0.003) 0.734(0.005) 0.735(0.007) 0.731(0.009) 0.710(0.002) 0.708(0.001) 0.708(0.001) 0.707(0.001) 0.710(0.001) 0.710(0.001) 0.729 0.733 0.733 0.730 0.710 0.708 0.708 0.707 0.710 0.710 0.723 O 0.734 0.722 O 0.748 0.724 O 0.753 0.714 O 0.763 0.708 O 0.711 0.708 O 0.711 0.708 O 0.709 0.706 O 0.709 0.709 O 0.712 0.710 O 0.711 Tl (ng g 1 ) 4.9(5.0)2.6 7.6(6.8)6.0 15(18)10 17(17)11 4.7(3.5)6.4 3.9(2.8)3.2 7.9(4.9)7.4 1.6(1.8)0.9 5.7(4.1)4.4 1.5(0.9)1.0 1 O 18 2 O 48 0.8 O 90 2 O 70 0.2 O 9 0.6 O 83O 15 0.6 O 5.7 2 O 14 0.9 O 2.6 .Aa.A.Spectrom. At. Anal. J. d205 Tl (&)NA NA 0.40(0.17)–0.38 0.39(0.13)–0.40 NA NA 0.29(0.17)–0.27 NA NA NA 0.62 O 0.18 0.62 O 0.14 0.49 O 0.11 Zn (mgg 1 ) 140(70)120 320(150)280 52(20)51 120(50)100 120(130)60 160(80)160 64(40)50 360(150)340 50(40)30 91(50)82 35 O 380 75 O 820 20 O 96 40 O 240 10 O 360 60 O 320 38 O 120 190 O 570 12 O 120 38 O 150 d66Zn (&) 0.15(0.22)–0.12 0.09(0.15)–0.12 0.07(0.24)–0.07 0.53(0.35)0.55 0.05(0.33)0.11 0.23(0.20)–0.18 0.11(0.04)–0.10 0.20(0.16)0.22 0.13(0.46)–0.22 0.15(0.06)–0.12 0.94 O 0.13 0.62 O 0.24 0.95 O 0.47 0.02 O 1.21 0.61 O 0.47 0.59 O 0.01 0.18 O 0.07 0.03 O 0.46 0.59 O 0.54 0.21 O 0.10 Ag (ng g 1 ) 23(15)20 18(36)9 21(10)19 3000(2300)2500 5.5(3.8)4.6 13(3.8)12 11(3.7)11 3.8(1.9)3.1 18(22)13 8.5(0.7)8.3 ,2016, 3 O 54 2 O 220 5 O 39 280 O 9300 0.7 O 11 6 O 18 8 O 15 2 O 73O 67 7.9 O 9.4 d109 Ag (&)NANANA0.21(0.13)–0.20 NA NA NA NA NA NA 31 0.31 O 0.07 View ArticleOnline 220 , Hg (ng g 1 ) 3.4(1.6)3.0 12(2.8)11 7.1(1.7)7.1 510(460)260 17(7.8)13 17(8.4)16 52(33)47 8.7(1.0)8.8 38(31)23 9.4(6.1)5.9 O O O O O O O O O – – 0.5 5.1 5 20 4 11 18 1500 9 31 6 36 17 96 7 10 5 95 5 17 3 | 233 a Results are presented in the format: average (standard deviation) median, minimum O maximum; note that n is the number of diﬀerent samples of the specied material collected and averaged JAAS

227 in each dataset. View Article Online

JAAS Paper repeatability for isotopic measurements, averaged over all through the membrane increases, which was indeed observed samples (n > 240), was as a rule <0.05& (Table 2). The slightly in the test. To avoid this undesirable effect, the Apex desolvating poorer repeatability for Pb ratios is due to the high proportion nebulizer was used for all subsequent Cd isotope ratio of samples in the datasets containing too little Pb for optimum measurements, with desolvation occurring by passing through MC-ICP-MS measurement. B repeatability represents both MC- a cooled condenser rather than a heated membrane. ICP-MS (approximately 2/3 of all results) and ICP-SFMS data. These gures are by a factor of 2–2.5 times better than instru- 3.3. Accuracy mental, between-block SDs of individual measurements due to The accuracy of concentration determinations by ICP-SFMS was the fact that the contribution from imperfect amplier gain veried by analyses of various CRMs (Table 2). For the majority calibration has been cancelled out. of analytes, recovery is within the 90–110% range. For accurate assessment of the overall reproducibility of the Though a set of in-house isotope standards (seawater CRM entire method, duplicate digestions and separations were per- for B, isotope standard solutions, commercial mono-element formed for 14 samples analyzed during different analytical standards or dissolved pure salts/metals with isotope compo- sessions conducted by various operators. Data for the four sition different from the respective d-zero standards for the rest CRMs that were a part of each analytical batch can also be used of analytes) was analyzed in every measurement session, such for this purpose. Results are summarized in Table 2 and quality controls only apply to the instrumental stage and are demonstrate generally that reproducibility values are two- to thus unsuitable for assessment of the overall accuracy of the ve-fold poorer than those of repeatability, reecting cumula- method. An absence of measurable fractionation in d-zero tive effects arising from minor differences in the efficiency of standards passed through the separation procedure assures separation, blanks, spectral interferences and mass-bias sufficiently good analyte recoveries, but as these standards are corrections, instrumental mass calibration stability and the both matrix- and interfering element-free, they are not the quality of instrument optimization. This reproducibility perfect solution to quality assurance either. As to the best of our provides a more realistic assessment of the developed method's knowledge no matrix-matched CRMs with certied isotopic ability to detect minor variations in isotope compositions than compositions for the elements encompassed by this study are repeatability. It should be stressed though that reproducibility available, the only means to evaluate method accuracy is to gures presented in Table 3 are valid for sample types analyzed compare data obtained in our study (Table 2 and 3) with those in this study and may not be applicable to other matrices with previously published for similar matrices, where such data higher (or lower) concentrations of analytes and interfering exist. Though only two of four CRMs used were of plant origin, elements. the inclusion of IRMM-BB186 and NRCC TORT-1 provided One observation made during the precision assessment for heavy isotope extremes for Cu and Zn, respectively. Cd isotope ratio measurements deserves special note. It was Roux et al.67 have reported d11B of (40.12 0.21)& in NIST found that agreement between duplicates analyzed using Ari- 1547 using purication of B by cation exchange chromatog- dus was signicantly inferior to that obtained with the standard raphy and micro-sublimation followed by MC-ICP-MS sample introduction system. In extreme cases d114Cd between- measurements, identical within uncertainties to our result. run variations exceeded 0.3&, even when the same separated Isotopically heavy Cu in animal kidney has been explained by digest fractions were re-analyzed. In order to investigate the fractionation during the breakdown of ascorbate into oxalate.68 reason for poor precision, a comprehensive set of experiments Maréchal et al.69 have reported d66Zn of +0.51& in NRCC TORT- was performed by measuring Cd isotopes in standard solutions: 2 (lobster hepatopancreas) and Balter et al.70 lighter Zn isotope (I) with variable acid strength; (II) at different plasma sampling composition in sheep kidney compared to other organs. Cd and depths; (III) with variable sample ow rate; and (IV) with vari- Zn isotope data for IRMM-BB186 and NRCC TORT-1 agree well able dry gas (Ar) ow through the Aridus desolvating with our previously published results23 conrming if not high membrane. It was found that changes in instrumental mass accuracy at least good reproducibility for datasets generated in bias for Cd caused by the aforementioned variations were large analytical campaigns two years apart. Measured d-values adequately corrected using Ag for internal standardization in for different Cd, Fe and Zn isotope ratios normalized by tests (I), (II) and (III), while even minor changes in Ar ow respective mass differences agree well (correlation coefficient through the desolvating membrane resulted in severe uncor- >0.98) for samples with analyte contents above the minimum rected effects, pointing to decoupling of the mass-bias correla- required concentrations. tion between Cd and Ag. Most probably, this is due to partial Cd For some of the samples, B, Pb and Sr isotope ratios were losses through the membrane that could be explained by measured by both MC-ICP-MS and ICP-SFMS. Sr and Pb data- the evaporation/sublimation of Cd from cadmium nitrates sets obtained by these techniques are compared in Fig. 2 Cd(NO ) $XH O in the Aridus because the desolvating 3 2 2 demonstrating very satisfactory agreement. module reaches temperatures above the boiling point of 66 Cd(NO3)2$XH2O. The effect would result in preferential losses of lighter Cd isotopomers that have higher degrees of volatili- 3.4. Throughput zation and diffusion rates through the membrane. A combina- Given a batch size of 36 samples plus two preparation blanks tion of these effects would be expected to lead to an isotope shi and two CRMs (as limited by the maximum of 40 digestion towards higher apparent d114Cd as the ow of drying gas vessel positions in the UltraCLAVE), unrestricted availability of

228 | J. Anal. At. Spectrom.,2016,31, 220–233 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

Fig. 2 Comparison of Pb and Sr isotope ratios obtained in bio-indicators by ICP-SFMS and MC-ICP-MS. The ordinary linear regression equations are: (top left) y ¼ (0.0045 0.0175) + (1.0027 0.0148)x; (top right) y ¼ (0.0480 0.0913) + (0.9807 0.0374)x; and (bottom) y ¼ (0.0036 0.0155) + (0.9950 0.0215)x where the uncertainties correspond to 99% conﬁdence intervals.

hot plate(s) for evaporation of digests and puried fractions were less than ve samples in each category, and because the (performed mostly overnight), and simultaneous access to ICP- results for these samples fall into ranges found for birch leaves SFMS and MC-ICP-MS, the entire procedure from sample or spruce needles from the same collection period. weighing to data evaluation can be done by two chemists in In the ESI† data for four similarly sized birch trees sampled approximately two weeks. This is approximately three-fold more annually for three years are discussed in more detail. The key time-consuming than complete isotopic analysis of a single nding from the latter dataset is that the individual trees, element (e.g. Cd), but because many operations in the proce- growing within a radius of 100 m, exhibit signicant spatial and dure can be performed in parallel, this is signicantly more temporal variations in concentrations and isotopic composi- time-effective (and less sample consuming) compared to using tions. This strongly suggests that the elemental and isotopic individual preparation/separation schemes for all eight signatures of deciduous plants are as unique as the trees elements. themselves and bear little connection to events on a regional scale. 3.5. Variations in concentrations and isotopic compositions 3.5.1. Boron. B concentrations vary from sub-mgg 1 levels in leaves, needles and mushrooms in mushrooms to over 50 mgg 1 in leaves. Leaves and needles ˚ Element content and isotope composition in leaves can (and from Lulea have a lower B content compared to samples from does) reect different accumulation pathways from a variety of other locations and there is a pronounced increase in the leaves' natural and anthropogenic sources. Variations in the same B concentrations throughout the growing season. Except for depend upon the type of tree (different accumulation mecha- mushrooms, all other matrices are enriched in the heavy B nisms, nutrition supply strategies, leaf morphology), age/size, isotope with needles having heavier B than leaves from the ˚  tree location (on continental as well as local scales; soil type and same location. Leaves from Lulea have a signi cantly lighter B composition, sub-soil geology, proximity to local contamination isotope composition than leaves from France, Italy and Spain, d11 & z & sources, etc.), location of leaves on the tree (height, orientation but are still heavier than the average crust ( B 7 (ref. of branches), sampling period, weather preceding sampling 72)), with a clear temporal pattern from spring to fall. The entire d11 & & occasion, etc.23,44,71 The ability to detect such variability will span of observed B-values is 52 (ranging from 12 to & entirely depend on the ‘resolution’ (overall reproducibility) of +40 ) which agrees well with the isotopic range of B in plant 32 the analytical procedure used and data in Table 3 demonstrate tissues reported by Rosner et al. the extent of such effects. 3.5.2. Cadmium. Cd concentrations increase in the order – – A statistical summary of concentrations and isotope data for needles leaves mushrooms. The lowest mean Cd concen- the majority of samples analyzed in the course of this study is trations were found in leaves from Birmingham and Barcelona, presented in Table 3. Data for oak, aspen and rowan leaves, as while the highest were observed in leaves from Genoa and Paris. well as for pine needles collected in Lule˚a are omitted as there The latter also exhibit the widest range of concentrations

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31,220–233 | 229 View Article Online

JAAS Paper covering 2–3 orders of magnitude, probably reecting differ- ranges for each individual matrix, location and sampling ences in accumulation between different tree species since not occasion are signicantly narrower. The RSD for mean exclusively birch leaves were sampled from these locations. Cd 208Pb/206Pb and 206Pb/207Pb ratios for all sample groups is concentrations in birch leaves are highest in spring, decreasing 0.48% and 0.74%, respectively. In birch leaves from the same approximately two-fold during the growing season due to dilu- location (Lule˚a) the range of measured Pb isotope ratios tion with organic materials. Cd in birch leaves from Sweden, decreases considerably from spring to fall sampling. Iceland and UK is signicantly isotopically heavier than in 3.5.6. Strontium. Mushrooms have the lowest Sr concen- leaves from France, Italy and Spain, and falls into the same trations (sub- to low mgg 1 range) while 40- to 300-fold higher range as reported for birch leaves analyzed during a previous levels are typical for the rest of the matrices with samples from study.23 Wei et al.38 have reported a d114Cd range from 0.39& Paris (both needles and leaves) having the highest Sr contents. to 0.08& for biomass collected from four plant species There are minor differences in mean 87Sr/86Sr ratios between (Solanum nigrum, Ricinus communis, Cyperus alternifolius and different matrices from the same location, becoming more Pteris vittata) from China, similar to that found for leaves from radiogenic in the order: leaves from Iceland (0.707), leaves and Barcelona. Needles have a distinctly lighter isotopic composi- needles from Paris (0.708), leaves from Genoa, Barcelona and tion than leaves from the same location. The total span of Birmingham (0.710) and all samples collected in Lule˚a (0.729– observed d114Cd is approximately 1.4& (0.42& to +0.96&). 0.735). The ratios reect the ages of the underlying bedrock at 3.5.3. Copper. The Cu concentration pattern in different each location, with the oldest being found in the Lule˚a area matrices follows that for Cd, increasing in the order needles – ranging from 1.8 to 2.8 Ga.77 Except for the latter location, leaves – mushrooms. As for Cd, the highest levels are found in ranges of 87Sr/86Sr ratios observed elsewhere overlap signi- young leaves, followed by sharp decreases by a factor of 2–3as cantly, potentially complicating the use of this isotope system leaves grow. Concentration ranges for Cu in birch leaves for conrmation of the geographical origin of the plant mate- collected from different locations and at the same growing stage rial.3,4,18 Highly radiogenic mean 87Sr/86Sr ratios found in Lule˚a are very similar. Light mean Cu isotopic composition was samples agree well with gures published by Aberg˚ et al.78 for typical for all types of matrices tested in this study with the plant samples collected from the central part of Sweden. Except lightest Cu found in needles and mushrooms from Lule˚a. The for the Lule˚a location where wide ranges of observed 87Sr/86Sr measured d65Cu varies from 2.30& to +0.41& with relatively ratios can be caused by heterogeneity of the 87Sr/86Sr in the consistent spans of 0.6& to 1.4& found for birch leaves from granitic bedrocks,79 variations in Sr ratios in plant samples from individual locations. the same group are seldom >0.3% RSD. 3.5.4. Iron. Mean Fe concentrations vary from approxi- As the use of Zr for mass-bias correction80 allows for mately 100 mgg 1 in Parisian leaves to over 750 mgg 1 in leaves assessment of mass-dependent fractionation of the 88Sr/86Sr from Iceland. Concentrations as high as almost 0.4% were found ratio, it can be stated that none of the samples shows frac- in Suillus variegatus mushrooms. In contrast to temporal regu- tionation outside the 0.04& to +0.04& range. Therefore, larities in Cu and Cd concentrations, the Fe content of birch 88Sr/86Sr ratios can safely be used for mass bias correction of leaves increases from spring to fall. The mean d56Fe is negative 87Sr/86Sr ratios in plant samples, providing identical results to for all groups of samples except needles from Paris, conrming Zr corrected data (Fig. 3) with considerably less effort. earlier ndings,34,73 with very little variations for birch leaves 3.5.7. Thallium. Tl was present in all matrices tested in the from Lule˚a, Paris and Birmingham. As for Cu isotopes, the low ng g 1 range, with the lowest extreme being found for birch lightest Fe was found in needles and mushrooms from Lule˚a. leaves from Iceland and UK, and the highest for spruce needles The total span in the observed d56Fe is approximately 1.6& and mushrooms from Lule˚a. From >240 samples, only approx- (1.32& to +0.26&). Light Fe isotope composition in plants was imately 40 had sufficient Tl content for MC-ICP-MS isotope ratio previously reported in a number of publications34,73–75 with observed d56Fe ranges in leaves from 1.3& to +0.09&. 3.5.5. Lead. Variability in mean Pb concentrations is amongst the highest of the elements studied, spanning from 73 ng g 1 in leaves from Iceland to over 2800 mgg 1 in Spanish leaves. The average concentration of Pb in birch leaves from Lule˚a increases more than three-fold during the growing season, conrming earlier data.23 Reimann et al.76 reported median Pb concentrations of 590 ng g 1 in birch leaves and 250 ng g 1 in spruce needles collected from 40 sites along a 120 km transect cutting through the city of Oslo, Norway, thus approximately double that observed from the results of the fall sampling in Lule˚a. Although the extensive variations observed for 208Pb/206Pb 206 207 Fig. 3 Comparison between the two approaches for MC-ICP-MS Sr and Pb/ Pb ratios (from 2.344 to 2.489 or 5.9% and from mass bias correction. The ordinary linear regression equation is y ¼ 1.126 to 1.259 or 11.3%, respectively) encompass the majority of (0.0011 0.0015) + (0.9984 0.0021)x where the uncertainties all ratios reported for biological samples from Europe,11,12,76 correspond to 99% confidence intervals.

230 | J. Anal. At. Spectrom.,2016,31,220–233 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS measurements, and meaningful statistics could only be derived origin. There is thus a growing need for inter-laboratory exer- for two matrices (Table 3). Needles and mushrooms from the cises, preferably using commercially available CRMs repre- Lule˚a location have Tl isotope compositions enriched in the senting various matrices, to ll this gap and ensure data lighter isotope with a total range in observed d205Tl values of transferability until appropriate CRMs become available. 0.8& (0.62& to +0.18&). Kersten et al.81 recently reported a In general the proposed method represents a starting point narrow d205Tl range in green cabbage (Brassica oleracea) from for further development in the direction of multi-tracer studies. China of 0.54& to 0.25&. With some relatively modest modications, the list of separated 3.5.8. Zinc. Mean Zn concentrations vary from 50 mgg 1 in elements can be extended to include Ca, Mg and Ga. Vapor- leaves from Barcelona to >300 mgg 1 in birch leaves from Ice- phase introduction of Hg can overcome the insuﬃcient sensi- land and Lule˚a (fall sampling). Needles contain signicantly tivity of the current system. The possibility to obtain precious lower Zn levels than leaves and mushrooms from the same information on several isotope systems from a single sample locations, and Zn concentrations increase through the growing will aid the interpretation of natural processes and enable more season. The mean d66Zn for the majority of matrices falls into a reliable pollution source attribution. relatively narrow range around zero (0.23& to +0.20&), while The majority of results obtained for eight isotope systems in signicantly heavier Zn isotopic compositions are found in more than 240 samples agree well with previously published mushrooms. Viers et al.35 reported a range in d66Zn from data where they exist. For some bio-indicators (e.g. needles and 0.50& to +0.08& in larch (Larix gmelinii) needles (n ¼ 12) mushrooms), our data represent the rst combined character- from Central Siberia, which is approximately half of that found ization of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn isotopic compositions. for spruce needles from Lule˚a. The total spread in the observed Even aer removing some known variation sources such as the d66Zn values is approximately 2.2& (0.95& to +1.21&). type of bio-indicator, sampling height and sampling period, 3.5.9. Silver and mercury. Ag was present in the low ng g 1 very broad ranges in isotopic compositions of many elements range in almost all matrices tested, although mushrooms could were found in samples collected from relatively conned exhibit concentrations in some species of Boletus edulis as high geographical areas (Table 3) or even from within 100 m (see ESI as 9 mgg 1. Consequently Ag isotope ratios were only deter- Table S1†), signicantly exceeding method reproducibility. The mined in the latter matrix with slightly negative mean d109Ag observed degree of isotopic variability, irrespective if caused by and relatively low variability (total range 0.31& to +0.07&). natural or anthropogenic factors, may complicate such isotope Temporal changes in the concentrations of Ag birch leaves applications as source tracing, geographical origin authentica- resemble those for Cd and Cu, being highest in young leaves. tion, studying plant metabolism, etc., and should be carefully The opposite temporal trend was found for Hg with concen- considered for each given study object. trations increasing three- to four-fold between May and Better understanding of regularities in observed isotope September. variability in leaves and needles would require acquiring isotope proling of diﬀerent soil compartments and soil solution as a 4. Conclusions function of soil depth – work which is currently underway.

The analytical protocol tested in this study has proved to be suitable for isotope ratio measurements of at least eight Acknowledgements elements in leaf, needle and mushroom samples. Digestion ALS Scandinavia AB is gratefully acknowledged for technical using the UltraCLAVE device provides complete oxidation of the support. MetTrans Initial Training Network (funded by the organic material, while two-column separation ensures low European Union under the Seventh Framework Programme) is ffi ffi blank levels, e cient separation of matrix elements, su ciently acknowledged for nancial support. We wish to thank Katerina high analyte recoveries and relatively high sample throughput. Rodiouchkina for eld assistance and help with sample prep- It was shown that losses of Cd may occur in membrane aration. We would also like to show our gratitude to Enzo desolvation systems, which may result in poor precision. Stranieri, Maria Grazia and Sergio Pallavicini for eld assis- Therefore, utilization of an Apex desolvation system can be a tance. The research leading to these results has received fund- 82,83 better alternative than the frequently used Aridus for Cd ing from the People Programme (Marie Curie Actions) of the isotope ratio measurements. European Union's Seventh Framework Programme FP7 2007– Regular use of duplicate sample preparations and analyses 2013, under the REA grant agreement no. 290336. The views ff performed in di erent batches or measurement sessions is a expressed in this article are those of the authors and may not must for overall reproducibility assessment. The use of necessarily reect those of the European Union. synthetic, matrix-free isotope standards or replicate analyses within the same measurement session results in articially optimistic estimates of precision. The absence of commercially available CRMs with certied isotopic compositions continues to hamper straightforward accuracy assessment. In contrast to the geological eld, there are still only a very few published datasets containing information on the isotopic composition of CRMs of biological

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 220–233 | 231 View Article Online

JAAS Paper

References 24 L. S. Sherman, J. D. Blum, J. T. Dvonch, L. E. Gratz and M. S. Landis, Sci. Total Environ., 2015, 502, 362–374. 1 T. D. Bullen, in Handbook of Environmental Isotope 25 L. S. Ruhl, G. S. Dwyer, H. Hsu-kim, J. C. Hower and Geochemistry, ed. M. Baskaran, Springer, Berlin, A. Vengosh, Environ. Sci. Technol., 2014, 48, 14790–14798. Heidelberg, 2012, pp. 177–203. 26 T. Fujii, F. Moynier, J. Blichert-To and F. Albarède, 2 H.-C. Liu, C. F. You, C. Y. Chen, Y. C. Liu and M. T. Chung, Geochim. Cosmochim. Acta, 2014, 140, 553–576. Food Chem., 2014, 142, 439–445. 27 K. Jaouen, M. L. Pons and V. Balter, Earth Planet. Sci. Lett., 3 M. V. Baroni, N. S. Podio, R. G. Badini, M. Inga, H. A. Ostera, 2013, 374, 164–172. M. Cagnoni, E. A. Gautier, P. P. Garc´ıa, J. Hoogewerff and 28 C. Rodrigues, M. Brunner, S. Steiman, G. J. Bowen, D. A. Wunderlin, J. Agric. Food Chem., 2015, 63, 4638–4645. J. M. F. Nogueira, L. Gautz, T. Prohaska and C. Máguas, J. 4 P. Degryse, A. Shortland, D. de Muynck, L. van Heghe, Agric. Food Chem., 2011, 59, 10239–10246. R. Scott, B. Neyt and F. Vanhaecke, J. Archaeol. Sci., 2010, 29 D. Houben, P. Sonnet, G. Tricot, N. Mattielli, E. Couder and 37, 3129–3135. S. Opfergelt, Environ. Sci. Technol., 2014, 48, 7866–7873. 5 I. Rodushkin, D. C. Baxter, E. Engstrom,¨ J. Hoogewerff, 30 Y. T. Tang, C. Cloquet, T. Sterckeman, G. Echevarria, P. Horn, W. Papesch, J. Watling, C. Latkoczy, G. van der J. Carignan, R. L. Qiu and J. L. Morel, Environ. Sci. Peijl, S. Berends-Montero, J. Ehleringer and V. Zdanowicz, Technol., 2012, 46, 9972–9979. J. Food Compos. Anal., 2011, 24,70–78. 31 D. Jouvin, D. J. Weiss, T. F. M. Mason, M. N. Bravin, 6 A. E. Shiel, D. Weis, D. Cossa and K. J. Orians, Geochim. P. Louvat, F. Zhao, F. Ferec, P. Hinsinger and Cosmochim. Acta, 2013, 121, 155–167. M. F. Benedetti, Environ. Sci. Technol., 2012, 46, 2652–2660. 7 C. Cloquet, J. Carignan, G. Libourel, T. Sterckeman and 32 M. Rosner, W. Pritzkow, J. Vogl and S. Voerkelius, Anal. E. Perdrix, Environ. Sci. Technol., 2006, 40, 2525–2530. Chem., 2011, 83, 2562–2568. 8 G.-X. Sun, X.-J. Wang and Q.-H. Hu, Environ. Pollut., 2011, 33 T. Deng, C. Cloquet, Y. Tang and T. Sterckeman, Environ. Sci. 159, 3406–3410. Technol., 2014, 48, 11926–11933. 9 M. Rehkämper, F. Wombacher, T. J. Horner, and Z. Xue, in 34 M. Guelke and F. Von Blanckenburg, Environ. Sci. Technol., Handbook of Environmental Isotope Geochemistry, ed. M. 2007, 41, 1896–1901. Baskaran, Springer, Berlin, Heidelberg, 2012, pp. 125–154. 35 J. Viers, A. S. Prokushkin, O. S. Pokrovsky, A. V. Kirdyanov, 10 I. Rodushkin, T. Bergman, G. Douglas, E. Engstrom,¨ C. Zouiten, J. Chmeleff, M. Meheut, F. Chabaux, P. Oliva D. Sorlin¨ and D. C. Baxter, Anal. Chim. Acta, 2007, 583, and B. Dupré, Geochem. Trans., 2015, 16,1–15. 310–318. 36 B. M. Ryan, J. K. Kirby, F. Degryse, H. Harris, 11 J. Sucharová, I. Suchara, C. Reimann, R. Boyd, P. Filzmoser M. J. McLaughlin and K. Scheiderich, New Phytol., 2013, and P. Englmaier, Appl. Geochem., 2011, 26, 1205–1214. 199, 367–378. 12 M. J. M. Notten, N. Walraven, C. J. Beets, P. Vroon, J. Rozema 37 C. R. Quétel, E. Ponzevera, I. Rodushkin, A. Gerdes, and R. Aerts, Appl. Geochem., 2008, 23, 1581–1593. R. Williams and J. Woodhead, J. Anal. At. Spectrom., 2009, 13 M. A. Smith and S. Pell, J. Archaeol. Sci., 1997, 24, 773–778. 24, 407–412. 14 M. Perini, F. Camin, L. Bontempo, A. Rossmann and 38 R. Wei, Q. Guo, H. Wen, J. Yang, M. Peters, C. Zhu, J. Ma, E. Piasentier, Rapid Commun. Mass Spectrom., 2009, 23, G. Zhu, H. Zhang, L. Tian, C. Wang and Y. Wan, Anal. 2573–2585. Methods, 2015, 7, 2479–2487. 15 D. Xue, B. de Baets, O. van Cleemput, C. Hennessy, 39 V. Migeon, B. Bourdon, E. Pili and C. Fitoussi, J. Anal. At. M. Berglund and P. Boeckx, Environ. Pollut., 2012, 161,43– Spectrom., 2015, 30, 1988–1996. 49. 40 J. Aggarwal, F. Bohm,¨ G. Foster, S. Halas, B. Honisch,¨ 16 J. G. Wiederhold, Environ. Sci. Technol., 2015, 49, 2606–2624. S.-Y. Jiang, J. Kosler, A. Liba, I. Rodushkin, T. Sheehan, 17 J. Klaminder, R. Bindler and I. Renberg, Appl. Geochem., J. Jiun-San Shen, S. Tonarini, Q. Xie, C.-F. You, Z.-Q. Zhao 2008, 23, 2922–2931. and E. Zuleger, J. Anal. At. Spectrom., 2009, 24, 825–831. 18 B. Song, J. Ryu, H. S. Shin and K. Lee, J. Agric. Food Chem., 41 B. C. Reynolds, J. Aggarwal, L. Andre, D. Baxter, C. Beucher, 2014, 62, 9232–9238. M. A. Brzezinski, E. Engstrom, R. B. Georg, M. Land, 19 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontér and M. J. Leng, S. Opfergelt, I. Rodushkin, H. J. Sloane, D. C. Baxter, Sci. Total Environ., 2007, 386, 159–168. S. H. J. M. van den Boorn, P. Z. Vroon and D. Cardinal, J. 20 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ C. Pontèr and Anal. At. Spectrom., 2007, 22, 561–568. D. C. Baxter, Sci. Total Environ., 2007, 386, 145–158. 42 J. F. Carter and B. Fry, Anal. Bioanal. Chem., 2013, 405, 2799– 21 N. Pallavicini, F. Ecke, E. Engstrom,¨ D. C. Baxter and 2814. I. Rodushkin, J. Anal. At. Spectrom., 2013, 28, 1591–1599. 43 M. Lodenius, Environ. Res., 2013, 125, 113–123. 22 I. Rodushkin, E. Engstrom,¨ D. Sorlin,¨ D. Baxter, B. Hörnfeldt, 44 K. Tarricone, G. Wagner and R. Klein, Ecol. Indic., 2015, 57, E. Nyholm and F. Ecke, Water, Air, Soil Pollut., 2011, 218, 341–359. 603–610. 45 I. Rodushkin, P. Nordlund, E. Engstrom¨ and D. C. Baxter, J. 23 N. Pallavicini, E. Engstrom,¨ D. C. Baxter, B. Ohlander,¨ J. Ingri Anal. At. Spectrom., 2005, 20, 1250–1255. and I. Rodushkin, J. Anal. At. Spectrom., 2014, 29, 1570–1584. 46 D. Malinovsky, I. Rodushkin, D. Baxter and B. Ohlander,¨ Anal. Chim. Acta, 2002, 463, 111–124.

232 | J. Anal. At. Spectrom.,2016,31, 220–233 This journal is © The Royal Society of Chemistry 2016 View Article Online

Paper JAAS

47 I. Rodushkin, F. Odman¨ and H. Holmstrom,¨ Sci. Total 65 I. Smet, D. de Muynck, F. Vanhaecke and M. Elburg, J. Anal. Environ., 1999, 231,53–65. At. Spectrom., 2010, 25, 1025–1032. 48 L. Ericson, M. Fabricius, E. Danielsson, B. Hultman, H. Juto, 66 R. C. Weast, in CRC Handbook of Chemistry and Physics, CRC and C. Huhtasaari, De odlade jordarna i norrbottens och Press, Inc., Cleveland, 58th edn, 1977–1978, 1978, p. B–97. vasterbottens¨ lan.¨ The cultivated soils of Norrbotten and 67 P. Roux, D. Lemarchand, H. J. Hughes and T. Marie-Pierre, Vasterbotten¨ , 1985. Geostand. Geoanal. Res., 2015, DOI: 10.1111/j.1751- 49 I. Rodushkin, E. Engstrom¨ and D. C. Baxter, Anal. Bioanal. 908X.2014.00328.x. Chem., 2010, 396, 365–377. 68 T. Fujii, F. Moynier, M. Abe, K. Nemoto and F. Albarède, 50 E. Engstrom,¨ A. Stenberg, S. Senioukh, R. Edelbro, Geochim. Cosmochim. Acta, 2013, 110,29–44. D. C. Baxter and I. Rodushkin, Anal. Chim. Acta, 2004, 521, 69 C. N. Maréchal, E. Nicolas, C. Douchet and F. Albarède, 123–135. Geochem., Geophys., Geosyst., 2000, 1,1–15. 51 I. Rodushkin, T. Ruth and A.˚ Huhtasaari, Anal. Chim. Acta, 70 V. Balter, A. Zazzo, A. P. Moloney, F. Moynier, O. Schmidt, 1999, 378, 191–200. F. J. Monahan and F. Albarede, Rapid Commun. Mass 52 R. Sun, M. Enrico, L.-E. E. Heimburger,¨ C. Scott and Spectrom., 2010, 24, 605–612. J. E. Sonke, Anal. Bioanal. Chem., 2013, 405, 6771–6781. 71 B. Markert, Fresenius' Z. Anal. Chem., 1989, 335, 562–565. 53 H. Lin, D. Yuan, B. Lu, S. Huang, L. Sun, F. Zhang and 72 M. Chaussidon and F. Albarède, Earth Planet. Sci. Lett., 1992, Y. Gao, J. Anal. At. Spectrom., 2015, 30, 353–359. 108, 229–241. 54 Q. Huang, Y. Liu, J. Chen, X. Feng, W. Huang, S. Yuan, H. Cai 73 M. Kiczka, J. G. Wiederhold, S. M. Kraemer, B. Bourdon and and X. Fu, J. Anal. At. Spectrom., 2015, 30, 957–966. R. Kretzschmar, Environ. Sci. Technol., 2010, 44, 6144–6150. 55 D. C. Baxter, I. Rodushkin, E. Engstrom¨ and D. Malinovsky, J. 74 N. Pérez Rodr´ıguez, F. Langella, I. Rodushkin, E. Engstrom,¨ Anal. At. Spectrom., 2006, 21, 427–430. E. Kothe, L. Alakangas and B. Ohlander,¨ Environ. Sci. Pollut. 56 L. van Heghe, E. Engstrom,¨ I. Rodushkin, C. Cloquet and Res., 2014, 21, 6836–6844. F. Vanhaecke, J. Anal. At. Spectrom., 2012, 27, 1327–1334. 75 F. Moynier, T. Fujii, K. Wang and J. Foriel, C. R. Geosci., 2013, 57 A. Olofsson and I. Rodushkin, Archaeometry, 2011, 53, 1142– 345, 230–240. 1170. 76 C. Reimann, B. Flem, A. Arnoldussen, P. Englmaier, 58 E. P. Horwitz, R. Chiarizia and M. L. Dietz, Solvent Extr. Ion T. E. Finne, F. Koller and Ø. Nordgulen, Appl. Geochem., Exch., 1992, 10, 313–336. 2008, 23, 705–722. 59 E. P. Horwitz, M. L. Dietz, R. Chiarizia, H. Diamond, 77 G. Gaal and R. Gorbatschev, Precambrian Res., 1987, 35,15– A. M. Essling and D. Graczyk, Anal. Chim. Acta, 1992, 266, 52. 25–37. 78 G. Aberg,˚ G. Jacks, T. Wickman and P. J. Hamilton, Catena, 60 E. P. Horwitz, in International Workshop on the Application of 1990, 17,1–11. Extraction Chromatography in Radionuclide Measurement, 79 T. Nakano and T. Tanaka, in Tracers in Hydrology, 1993, pp. IRMM, Geel, 1998, vol. E, pp. 27–37. 73–78. 61 B. Gao, Y. Liu, K. Sun, X. Liang, P. Peng, G. Sheng and J. Fu, 80 J. Irrgeher, T. Prohaska, R. E. Sturgeon, Z. Mester and Anal. Chim. Acta, 2008, 612, 114–120. L. Yang, Anal. Methods, 2013, 5, 1687–1694. 62 C. N. Maréchal, P. Télouk and F. Albarède, Chem. Geol., 1999, 81 M. Kersten, T. Xiao, K. Kreissig, A. Brett, B. J. Coles and 156, 251–273. M. Rehkämper, Environ. Sci. Technol., 2014, 48, 9030–9036. 63 F. Larner, M. Rehkämper, B. J. Coles, K. Kreissig, D. J. Weiss, 82 S. Ripperger, M. Rehkämper, D. Porcelli and A. N. Halliday, B. Sampson, C. Unsworth and S. Strekopytov, J. Anal. At. Earth Planet. Sci. Lett., 2007, 261, 670–684. Spectrom., 2011, 26, 1627–1632. 83 A. E. Shiel, D. Weis and K. J. Orians, Sci. Total Environ., 2010, 64 P. A. Sossi, G. P. Halverson, O. Nebel and S. M. Eggins, 408, 2357–2368. Geostand. Geoanal. Res., 2015, 39, 129–149.

This journal is © The Royal Society of Chemistry 2016 J. Anal. At. Spectrom.,2016,31, 220–233 | 233 Electronic Supplementary Material (ESI) for Journal of Analytical Atomic Spectrometry. This journal is © The Royal Society of Chemistry 2015

Electronic Supplementary Information for

Assessment of natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation

Ilia Rodushkin,a,b Nicola Pallavicini,a,b Emma Engström,a,b Dieke Sörlin,b Björn Öhlander,a Johan Ingria and Douglas C. Baxterb a Division of Geosciences, Luleå University of Technology, S-971 87 Luleå, Sweden b ALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Luleå, Sweden

Long-term variability in elemental concentrations and isotope ratios in birch leaves Factors known to affect the concentrations and isotopic compositions of elements present in foliar material broadly comprise species, age, location, soil characteristics, climate and proximity to sources of contamination, as discussed in section 3.5 of the main article. Table S1 provides both concentration and isotope data for leaves collected during the last week of May in three consecutive years, from four individual birch trees (of approximately the same size with stem diameter near ground of 30-35 cm) growing in close proximity (within 100 m) to each other at a sampling point situated approximately 5 km North-East of Luleå city center. On the basis of this experimental design it was assumed that the majority of the aforementioned sources of variation would be the same for all trees, and therefore the results for the three sampling campaigns were expected to agree within measurement uncertainty. However, this was clearly not the case for the majority of the elements studied (Table S1). B concentrations are similar for all trees and show a reproducible trend of decreasing concentrations from 2013 to 2015. δ11B is the same, within the long-term reproducibility of the analytical method, for three of the trees while significantly higher in one (birch B). No between- year significant variations in B isotope composition in leaves from the same tree were noted. Cd concentrations differ between individual birches with no clear between-year pattern. On the other hand, δ114Cd clearly shifts towards lighter isotopic composition between 2013 and 2015. The Cd isotopic composition of one tree (birch D) is significantly heavier than in the others. During the same period, Cu concentrations tend to increase while δ65Cu (and δ56Fe) became less negative. Fe concentrations, however, decrease similarly to those of B. Pb concentrations exhibit the largest between-year differences, decreasing from 2013 to 2015 almost three-fold (Birches A and B, Table S1) while 208Pb/206Pb and 206Pb/207Pb ratios increase. Ranges of Pb isotope ratios in leaves from the same vicinity are actually wider than those reported for birch leaves from 40 sampling locations in Oslo (2.421–2.451 208Pb/206Pb and 1.153– 1.188 for 206Pb/207Pb1). Sr concentrations differ by a factor of 2–3 in foliage from different trees and for some birches exhibit almost two-fold between-year variability. Foliar Sr isotopic composition becomes less radiogenic with between-year variability for a single tree as high as 0.6%. Temporal variations in

1 birch leave Zn concentrations follow the same pattern as that described for B, Fe, Pb and Sr, while δ66Zn in all four trees became significantly more negative over the duration of the study. This experiment was initially planned to provide an assessment of long-term reproducibility including contributions from the sampling process. Since findings demonstrating the existence of significant between-year differences in foliage from the same trees were highly unexpected, very thorough control of all method stages and re-analyses of all samples collected from a single birch were performed. No analytical errors could be detected. Profound changes in anthropogenic foliar uptake on both local and regional scales, for example due to annual differences in predominant wind directions during the exposure period though not unthinkable, can hardly explain the variability for elements such as Cu, Fe, Sr or Zn. In the search for potential explanations, it was established that there were very significant differences in May weather during the study period. May 2013 was uncharacteristically hot with May 30 actually being the warmest day of the year reaching 27oC with the historical average for the day being only 14oC. The average daytime temperature reached 10oC as early as the first week of the month, whereas May 2015 was unusually cold with the average daytime temperature approaching, but not reaching 10oC only at the end of the month.2 Therefore, though collected on almost the same annual date, birch leaves from 2013 represent a distinctly different growing stage than those from 2015. Firstly, there was an offset of at least two weeks in the appearance of the first leaves between these two years. Secondly, the frozen soil depth, which limits nutrient uptake from deep soil horizons through the root system, rhizosphere activity3 and affects sub-surface water drainage, was much shallower in 2013 than in 2015 when collection occurred. The direction of trends observed between 2013 and 2015 data (Table S1) resembles that for seasonal changes noted for the entire set of birch results (two first columns in Table S1), given that leaves collected in 2013 represent later stages of the growth cycle. These results indicate isotopic differences between source pools supplying elements to birch leaves during the first 2-4 weeks of growth (soil solution, sap stored in stem, aerial sources) and changes in fractionation effects occurring during either uptake through the root system or element translocation between different tree compartments. Our Zn isotope data reaffirm recently published findings4 that the increased rooting depth and the decreased organic carbon concentration in the root uptake zone resulting from progressively thawing soil leads to heavy isotopes becoming more and more available for larch roots. This results in a shift towards heavier Zn composition in larch needles during the course of the vegetative season. Using sequential leaching of soil, Song et al.5 have demonstrated that Sr in the exchangeable and carbonate fractions (bioavailable) has a lower 87Sr/86Sr ratio that that in the silicate fraction, consistent with a low 87Sr/86Sr ratio in the plant and therefore in the organic-rich topsoil horizon. This would explain the more radiogenic Sr ratios in leaves collected in 2013 (Table S1) as increased contributions from deeper thawed soil horizons. Variations of bioavailable Sr concentrations and 87Sr/86Sr with depth of root uptake were also reported by Poszwa et al.6 These results demonstrate that foliage samples provide highly spatially- and temporally-resolved snapshots of elemental and isotopic interactions with deciduous plants on the individual scale.

2 Table S1. Concentrations and isotopic composition of B, Cd, Cu, Fe, Pb, Sr and Zn in leaves collected from 4 individual birch trees collected last week of May 2013-2015

Sampling B δ11B Cd δ114Cd Cd δ65Cu Fe δ56Fe Pb Sr Zn δ66Zn Birch 208Pb/206Pb 206Pb/207Pb 87Sr/86Sr period (μg g-1) (‰) (ng g-1) (‰) (μg g-1) (‰) (μg g-1) (‰) (ng g-1) (μg g-1) (μg g-1) (‰)

A May 2013 14,1 6.2 547 0.463 8.5 -0.440 171 -0.488 203 2.396 1.164 15.1 0.7310 179 -0.219 May 2014 11.9 7.2 490 0.367 9.3 -0.502 137 -0.368 122 2.421 1.179 14.9 0.7300 169 -0.359 May 2015 11.2 7.2 535 0.278 11.8 -0.188 125 -0.366 71 2.470 1.213 12.9 0.7276 125 -0.457

B May 2013 16.1 9.6 395 0.402 9.7 -0.617 165 -0.457 200 2.404 1.168 53.9 0.7296 382 0.016 May 2014 9.9 9.8 285 0.344 10.7 -0.404 137 -0.313 146 2.407 1.174 49.9 0.7270 296 -0.123 May 2015 8.7 10.7 347 0.306 11.1 -0.301 103 -0.315 69 2.450 1.208 28.3 0.7255 137 -0.307

C May 2013 10 5.9 444 0.411 12.1 -0.580 163 -0.482 142 2.418 1.177 43.4 0.7296 288 0.104 May 2014 7.7 6.8 289 0.351 13.5 -0.471 114 -0.287 112 2.432 1.186 37.1 0.7269 179 -0.022 May 2015 7.4 7.6 403 0.288 14.1 -0.308 94 -0.255 64 2.449 1.202 35.1 0.7268 144 -0.198

D May 2013 13.4 4.9 357 0.650 9.3 -0.452 188 -0.288 104 2.425 1.182 31.8 0.7284 186 -0.112 May 2014 10.9 6.6 223 0.621 10.0 -0.328 146 -0.264 84 2.428 1.184 28.8 0.7250 149 -0.301 May 2015 10.5 7.1 209 0.574 11.2 -0.241 125 -0.261 71 2.458 1.205 17.7 0.7252 63 -0.405 References

1. C. Reimann, B. Flem, A. Arnoldussen, P. Englmaier, T. E. Finne, F. Koller, and Ø. Nordgulen, Appl. Geochem., 2008, 23, 705–722.

2. Swedish Meteorological and Hydrological Institute, http://opendata-download- metobs.smhi.se/explore/#, (accessed July 2015).

3. D. Houben, P. Sonnet, G. Tricot, N. Mattielli, E. Couder, and S. Opfergelt, Environ. Sci. Technol., 2014, 48, 7866–7873.

4. J. Viers, A. S. Prokushkin, O. S. Pokrovsky, A. V Kirdyanov, C. Zouiten, J. Chmeleff, M. Meheut, F. Chabaux, P. Oliva, and B. Dupré, Geochem. Trans., 2015, 16, 1–15.

5. B. Song, J. Ryu, H. S. Shin, and K. Lee, J. Agric. Food Chem., 2014, 62, 9232–9238.

6. A. Poszwa, B. Ferry, E. Dambrine, and B. Pollier, Biogeochem., 2004, 67, 1–20.

4 V

Ranges of B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotope ratios in environmental matrices from an urban area

Nicola Pallavicini, Emma Engström, Douglas C. Baxter, Björn Öhlander, Johan Ingri, Scott Hawley, Catherine Hirst, Katarina Rodushkina and Ilia Rodushkin

Ranges of B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotope ratios in environmental matrices from an urban area

Nicola Pallavicinia,b, Emma Engströma,b, Douglas C. Baxterb, Björn Öhlandera, Johan Ingria, Scott Hawleyc, Catherine Hirstd, Katarina Rodushkinae and Ilia Rodushkina,b.

aDivision of Geosciences and Environmental Engineering, Luleå University of Technology, S-971 87 Luleå, Sweden bALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Luleå, Sweden cDepartment of Earth Sciences, Durham University, Durham, UK dDepartment of Geosciences, Natural History Museum, Stockholm, Sweden eDepartment of Chemistry, Uppsala University, Uppsala, Sweden

e-mail: [email protected]

Abstract

Isotopic compositional values provide powerful information for identifying, discriminating and tracing inputs to natural compartments. Further progress in the widespread use of isotopic techniques in natural sciences requires a better assessment of elemental and isotopic compositional variability in environmental matrices. This study assesses the current local- scale variability (ranges) of concentration and isotopic composition data for nine elements, boron (B), cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), lead (Pb), strontium (Sr), thallium (Tl) and zinc (Zn) in soils, lysimetric waters, mushrooms, litter, needles, leaves and lichens. Sequential extractions were performed on soil samples from 6 depth proﬁles providing operationally-deﬁned pools of the elements. The isotope ratios of the pools were then determined to complement the total elemental concentration data. Large variations in the isotope ratio data were found for most of the systems, spanning almost the natural ranges of values reported in the literature. These results represent a starting point for discussing the role of variability in isotopic applications (for example as a limiting factor in isotopic mixing models where relative contribution from isotopically inhomogeneous end-members can hardly be resolved) and a baseline for future in-depth studies on the multiple parameters affecting the variability observed.

Keywords: Isotope ratios, bio indicators, natural variability, fractionation, sequential extraction

1 1. Introduction

The development and utilization of isotope based techniques in environmental science has grown exponentially in last few decades (Arnold et al., 2015; Ilia Rodushkin et al., 2016; Shiel et al., 2010; Wiederhold, 2015). Studies are increasing adopting multi-isotope approaches to determine sample provenance (Baroni et al., 2015; Liu et al., 2014; Serra et al., 2005) to identify pollution sources (Cloquet et al., 2006; Geagea et al., 2008a; Hissler et al., 2008) and to identify processes in biological matrices (Deng et al., 2014; Jouvin et al., 2012). Ultimately, uncertainties in unambiguously interpreting isotope ratio data limits the power of all single and multi-elemental methods in biological and environmental sciences (Chrastný et al., 2015). Historically, analytic (im)precision was a signiﬁcant barrier to unambiguous data interpretation for stable (Cloquet et al., 2005; Maréchal et al., 1999) and radiogenic isotopic systems (Gwiazda et al., 1998). The advent and continued development of multiple collector inductively coupled plasma mass spectrometry (MC-ICP-MS) has greatly improved analytic precision. Today, the ability to conﬁdently interpret high-quality MC-ICP-MS data is increasingly limited by a lack of data on the isotopic variability of environmental samples. This uncertainty is manifested in issues of sample representativeness i.e. how well the sampled material is comparable to known standards and previous studies.

Sample representativeness is an important parameter that must be taken into account as natural isotopic variability, both spatial and temporal, represents a major source of ambiguity in isotopic studies (e.g. Bullen, 2012; Rodushkin et al., 2016; Wiederhold, 2015). In a previous study i.e Rodushkin et al. (2016), we reported the total concentrations and isotope ratios of a number of elements in leaves, needles and mushrooms. Even after removing some known sources of variation such as sampling height (for leaves and needles) and sampling period, very broad ranges in isotopic compositions of many elements were found across samples collected over a conﬁned geographic area. The variability was attributed to a combination of differences in accumulation pathways, soil type, sub-soil geology, proximity to local contamination sources, sampling period and weather preceding sampling occasion (Markert, 1989; Pallavicini et al., 2014; Tarricone et al., 2015). Given such a large number of confounding factors may contribute to the isotopic composition of environments samples it was concluded that, while foliage samples may provide highly spatially- and temporally-resolved snapshots of elemental and isotopic interactions, a better understanding of the composition of the individual variables with the system was needed to understand the observed isotopic variability in bio-indicators.

Therefore, the aim of this work is to present a comprehensive dataset of concentrations and isotopic compositions for nine elements in a variety of environmental samples (topsoil, lysimetric waters, mushrooms, litter, needles, leaves and lichens). As the potential fate of an element in a biogeochemical cycle is defined by the form or sub-pool in which it is present i.e. dissolved, exchangeable, included in the mineral lattice or insoluble (Bielicka-Giełdo et al., 2013), distinguishing the sub-pools is an important compliment to bulk concentration and isotope ratio measurements. The utilization of sequential extraction procedures (SEP) with-in this study facilitates the determination of 6 functionally defined elemental pools from 6 soil profiles from urban and suburban locations.

The targeted elements were selected because they are representative of elements which are essential for plant’s grow (Zn, Fe, Cu, and B), can be of environmental concern (Cd, Pb, Cr,

2 Cu, Zn and Tl), are commonly used tracers in provenance studies (Sr, B, and Pb), are reflective of both radiogenic (Sr and Pb) and stable isotopic systems (B, Cd, Cr, Cu, Fe, Sr, Tl and Zn) and can be analysed using a single multi-element/multi-isotopic analytical procedure (Ilia Rodushkin et al., 2016). The observed isotopic ranges are used to compare the magnitude of local scale variability for the elements to published assessments of global isotopic variability. In order to illustrate the significance of such variability assessments, a case study is presented using multi-isotope data for landfills contaminated by tailings from Fe production, Fe and Cu slag, and fly ash.

2. Materials and methods

2.1. Study site All samples in this project were collected in and around the city of Luleå, a medium-sized town situated in northern Sweden in the province of Norrbotten (Figure 1). The original village of Luleå dates back to 1649 (Lundberg and Peterson, 2005) and has more recently become industrialized including a major steelworks. Samples collected within approximately 5 km direct distance from local steelworks and harbours will for the remainder of this study be referred to as ‘city’ samples. Samples from a broader area, approximately 10 km from the major local industries, are referred to as ‘suburb’ samples. The suburban locality was relatively recently a seabed [The study area, as well as entire northern part of the Nordic region is rising due to post-glacial up-lift almost 9 mm a year, due to isostatic glacial readjustment (Poutanen and Steffen, 2014)] and has been involved in the urbanization process only from the late 1980’s.

The local soil consists mainly of clay and silt loam overlying 1.9 Ga granitic bedrock with minor meta-sedimentary constituents. In terms of soil type the suburban location is characterized by a rather homogeneous sandy, well sorted soil with little variations in grain size. The sampling locations in the urban (city) area are till soils characterized by coarser and more heterogeneous grain sizes and formed at the time the land emerged from a sea some 400 years ago.

2.2. Sample collection and preparation The sampling procedures for bio-indicator organisms (mushrooms, n>60; one pooled litter sample; leaves, n>200; needles, n>40; lichens, n>30) are reported in Pallavicini et al. (2014) and Rodushkin et al. (2016). Individual soil samples were collected at the end of July 2015 following the same scheme (n>150, approximately 10 g of soil from upper 2-3 cm layer sampled by acid-washed plastic spoon). To supplement the individual soils samples six soil profiles, two from the city centre [where Rodushkin et al. (2016) found the highest concentrations of many elements in leaves and lichens] and four from the suburban area [where Rodushkin et al. (2016) found the lowest elemental concentrations in bio-indicators) hereafter referred to as “city” and “suburb” respectively, were also collected. The profiles were collected using a vertical soil core sampler to a maximum sampling depth of 60 cm (consistent with typical rooting depth for tree species sampled (Mauer and Palátová, 2003). Once collected each soil core was divided into four equal 15cm horizontal sub samples and stored in 50 mL polypropylene screw-capped tubes. The soil subsamples do not directly corresponds to different soil horizons. Two heavily contaminated soil samples (landfills) from the same region were also added to this study. In order to compliment the soil samples, pore-water solutions (approximately 250 mL) were collected from nine locations in the town using drain-gauge passive-capillary lysimeters (n=15) installed at 20 cm depth.

3 Rutvik Biological samples Luleå Soil samples

Bensbyn Gammelstaden

Storheden Porsön Björkskatan

Mjölkudden Notviken

Bergviken

Karlsvik

Hertsön

Luleå Skurholmstaden

Bergnäset

Svartöstaden SSAB

SSAB

Luleå Airport

0 1 2 km

Figure 1 Map of the sampling locations. Filled circles represent the sampling point where biological sam- SOHVKDYHEHHQFROOHFWHG OHDYHVQHHGOHVDQGPXVKURRPV ¿OOHGVWDUVUHSUHVHQWWKHORFDWLRQVZKHUHVRLO samples have been collected

Details on sample preparation procedure including: chemicals and reagents used, sample digestion and sequential extraction procedure (SEP), analyte/matrix separation, element concentration and isotopic analyses fully reported in the ESI. The SEP followed method involved six extractions according to methods developed by Tokalioglu et al. (2003) and has been summarized in Table 1.

2.3. Analytic Performance and Reporting Methods Information on operating parameters and measurement conditions for MC-ICP-MS can be found in Table 2 (see ESI for more detailed description). All stable isotope systems are reported in standard delta (b) notation while radiogenic systems are reported as ratios in accordance with standard practices. Stable isotopic differences between samples are reported as capital delta

4 Table 1 Sequential extraction scheme

Volume Code Extracted phase Solution (mL)

F1 Exchangeable Distilled water 40

F2 Carbonates 0.11 M CH3COOH (acetic acid) 40

F3 Reducible 0.1 M NH2OH HCl (hydroxylamine hydrochloride) 40

F4 Oxidizable 8.8 M H2O2 (hydrogen peroxide) 10+10

1 M CH3COONH4 (ammonium acetate) pH 2 50 F5 Residual I Aqua regia 20

F6 Residual II HF 10

(6) notation denoting the difference between two b values. For the stable isotope systems as many of the individual isotopes were measured as possible (see Table 2) to validate the mass dependent nature of the measurements, but for easy of reporting only a single b ratio will be used consistently for each system (Table 2).

The analytical results for reference standards [QC Loam Soil (VIK) (soil powder) (Eurofins A/S Denmark, Vallensbæk Strand, Denmark), GBW 07410 (soil) (National Research Centre of Geoanalysis, Beijing, China) and IRMM ERM-CC141 (soil powder)] along with typical repeatability and intermediate precision figures for soils, mushrooms and leaves are reported in Table 3. All reference standards were processed and analyzed in parallel with the samples. It should be noted that results for GBW 07410 and ERM-CC141 are calculated using concentrations and isotope ratios for individual sequential extraction fractions and therefore more uncertain. A full list of certified materials used to assess performance of the method for bio-indicators is available in the ESI.

2.3.1. Precision The intermediate precision, defined as the reproducibility of an analytic technique from digestion to measurements on multiple occasions by different analysts for the reference standards BCR- 146R and CRM052 is <7% for most elements (Table 3). Intermediate precision is about two times greater than repeatability (Table 3) due to varying element concentrations, slight variations in column yield, efficiency of matrix separation and contamination during column chemistry, instrumental optimization, etc. Intermediate precision, defined in this manner, is therefore considered to provide a valid assessment of the overall method uncertainty or the analytical ‘resolution’ of the procedure. For example, due in large part to low absolute B concentrations, B isotope ratio measurements in mushrooms have a mean uncertainty of approximately 5‰ if performed on different occasions and natural variations in isotopic composition below this level cannot be discerned.

5 . -1 Pb Pb; Sr obtain Tl B 207 86 Cd Fe Cr 207 Zn Cu 11 56 53 66 56 205 114 į į į Sr/ į į į Pb/ į Ratios Pb/ 87 Reported 208 206 elemental -1

$X[LOLDU\JDVÀRZ/PLQ solution -1 Lot 130116 Lot L00709 (NBS-987) Sr carbonate Tl:1000 mg L VWDQGDUG8OWUD6FLHQWL¿F NBS SRM 979 Cr standard µį]HUR¶UHIHUHQFHVWDQGDUGV North Kingstown, RI, USA; NIST SRM 976 Cu standard NIST NIST SRM 981 common Pb NIST Sr carbonate NIST SRM 987 Sr carbonate NIST NIST SRM 3108 Cd solution NIST Sn - - - 117 ODQWJDVÀRZ/PLQ Pb - Sn Cd Ni Zn 62 Ni - IRMM-014 Fe metal 70 116 208 116 62 B - - SRM 951a boric acid NIST Pb Sn Sr - - Cd Ni Ni Zn 11 88 61 61 68 114 207 114 Pb Sn Sr Cd Ni Ni Fe - - Zn Rb 87 56 60 60 67 87 112 206 112 Tl Sr - Cd Kr Zn 7DEOHPRGL¿HGIURP5RGXVKNLQHWDO 86 86 b 66 205 111 5)SRZHU±:&RR b Pb Pd Cd Hg Cr Fe Fe - Rb Cu 54 54 57 85 65 &XSFRQ¿JXUDWLRQ 110 204 110 204 Tl B--- Sr Ag Ni)Cr Fe IRMM-3702 Zn solution Kr Zn 10 84 64 53 56 84 64 203 109 ( Pd Cd - Hg Cr Kr Cu 52 83 63 108 108 202 ------Ag Fe Cr Kr L4 L3 L2 L1 C H1 H2 H3 H4 54 54 82 . Ion lens settings: adjusted daily to obtain maximum sensitivity and signal stability. Zoom optic settings: adjusted daily to . Ion lens settings: adjusted daily to obtain maximum sensitivity and signal stability. -1

107 mode Dynamic 6XEFRQ¿J FRQ¿JXUDWLRQ ) -1 Ni and Ag which were only used as the internal standards. Ni and

Sample (mL min (mL uptake rate ORFN1XPEHURILQWHJUDWLRQV±$PSOL¿HUURWDWLRQOHIW

1.262 0.05-0.25 main - time (s) , Aridus and Apex) 0.01–0.02 L min Apex) 0.01–0.02 L Aridus and , 2 Integration Low 0.262 0.20-0.25 Low 0.524 0.04-0.06 Low 0.524 0.40-0.50 - - Low 0.524 0.04-0.06 High mode Medium 0.262 0.20-0.25 Medium 0.262 0.20-0.25 - Resolution $GGLWLRQDOJDVÀRZ 1 -1 cone cone cone system Pumped Pumped Pumped Pumped Pumped skimmer cone skimmer cone skimmer cone skimmer cone/ Apex desolvating microconcentric PFA microconcentric PFA microconcentric PFA microconcentric PFA microconcentric PFA microconcentric PFA X-type skimmer cone spray chamber, H-type spray chamber, spray chamber, H-type spray chamber, spray chamber, H-type spray chamber, spray chamber, H-type spray chamber, systems, self-aspirating systems, self-aspirating systems, self-aspirating cyclonic spray chamber, cyclonic spray chamber, Aridus/Apex desolvating Aridus/Apex desolvating Micromist nebulizer, mini Micromist nebulizer, nebulizer, X-type skimmer nebulizer, nebulizer, X-type skimmer nebulizer, nebulizer, X-type skimmer nebulizer, Micromist nebulizer, double Micromist nebulizer, Micromist nebulizer, double Micromist nebulizer, Micromist nebulizer, double Micromist nebulizer, Micromist nebulizer, double Micromist nebulizer, &RQ¿JXUDWLRQRILQWURGXFWLRQ MC-ICP-MS operating parameters and measurement conditions for isotope ratio measurements a B Sr Cd Fe/Ni Pb/Tl Cr/Ni Zn/Cu Elements One element is used as the internal standard for second element, except 6DPSOHJDVÀRZ±/PLQ PD[LPXPUHVROXWLRQ1XPEHURIEORFNV1XPEHURIF\FOHVSHUE a Table 2 Table

6 Soil, n=20 &HUWL¿HG&, 5HFRYHU\ -1 Leaves, n=24 Mushrooms, n=16 n=8 n=6 PbPb 1.1496±0.0002 2.4288±0.0003 1.2078±0.0004 2.4650±0.0004 Sr 0.70805±0.00003 0.70914±0.00004 207 207 86 Tl. ‰ -0.50±0.04 -0.41±0.09 Cd. ‰ 0.11±0.09 -0.01±0.09 Cr. ‰Cr. Cu. ‰Fe. ‰ 0.04±0.09 0.22±0.06Zn. ‰ 0.18±0.05 -0.12±0.05 0.14±0.07 0.84±0.14 0.80±0.07 -0.18±0.05 B. ‰ -3.40±1.53 7.21±1.09 Pb/ Pb/ Sr/ 11 114 53 65 56 205 66 Soil, n=10 į į į į 206 208 87 į &HUWL¿HG&, 5HFRYHU\ )RXQGJJ į į -1 -1 Leaves, n=12 JJ &HUWL¿HG&, )RXQGı )RXQGı -1 5HSHDWDELOLW\ın=8 ,QWHUPHGLDWHSUHFLVLRQı n=6 -1 JJ &HUWL¿HG&, )RXQGıJJ &HUWL¿HG&, 5HFRYHU\ )RXQGJJ -1 -1 n=8 GBW 07410GBW Found ERM-CC141. Found Mushrooms, )RXQGJJ )RXQGıJJ Concentrations and isotope information for CRMs PbPb 2.4452 1.1781 2.4431 1.1913 0.0037 0.0022 0.0053 0.0034 0.0021 0.0011 0.0040 0.0028 0.0058 0.0041 0.0024 0.0016 Sr 0.7122 0.7291 0.0014 0.0001 0.0002 0.0018 0.0002 0.0002 207 207 86 Tl. ‰ -0.53 -0.67 0.089 0.106 0.031 0.112 0.122 0.059 Cd. ‰ -0.07 -0.11 0.022 0.032 0.074 0.041 0.072 0.116 Cr. ‰Cr. Cu. ‰Fe. ‰ 0.11 0.34 0.16Zn. ‰ 0.02 0.06 0.28 0.06 NA 0.18 0.074 0.077 0.048 NA 0.078 0.056 0.062 0.052 0.078 0.034 0.034 0.097 NA 0.086 0.088 0.086 0.081 NA 0.095 0.073 0.059 0.107 0.061 B. ‰ -6.0 4.5 3.2 0.7 1.6 5.4 1.2 2.4 Pb/ Pb/ Sr/ 11 114 53 65 56 205 66 į BCdCrCuFe 42.1Pb 0.117SrTl 57 23.7Zn 32200 29.9 175 0.66 38.3±4,8 71.7 0.086±0,015 66±4,5 23.2±2,2 32200 136 110 29.2±3,2 188±9 0.62±0,16 69.1±2,7 102 86 100 102 106 93 0.28 104 13 30 19800 34.4 0,14 0.25±0.04 12 52 NA NA 12.4±0.9 112 31±4 32.2±1.4 NA NA NA 50±4 105 107 NA 0.31 97 NA 104 16 NA 14.4 42.1 24800 0.35±0.05 66 0.52 60 68 14.4±1.4 NA 89 NA 41±4 86±8 NA 100 57±4 NA NA 103 NA 77 NA 105 NA CdCuFePb 17.7±0.4 798±25 13600±800 548±43 18.4±0.4Element NA 831±16 07410 GBW 583±17 263±9 130±7 4610±320 115±11 256±6 4280±208 130±3 108±3 Aqua Regia ERM-CC141, Total ERM-CC141, B 17.3±1.6 NA 177±11 NA Sr 1070±99 NA 243±13 113±11 Tl 0.469±0.033 NA 75.5±6.2 71.6±2.7 Zn 3050±170 3040±60 512±34 516±12 Cr 175±10 174±7 346±19 332±5 į į į į į į Element BCR-146R, Agua Regia CRM052, Aqua Regia BCR-146R208 206 87 CRM052 Table 3 Table

7 2.3.2. Accuracy Elemental concentrations in the aqua regia digests were within 90%-110% range of the certified values for the CRMs except for Sr in CRM052 (Table 3). Re-analysis of the digests by an alternative technique (ICP-optical emission spectrometry) confirmed the double- focusing sector field ICP-MS (ICP-SFMS) results, and the reason for our high reported SR concentrations remains unclear. The overall accuracy of the SEP was calculated by summing the elemental concentrations in all the fractions in GBW 07410 and ERM-CC141. The majority of elements were within 10% of the certified values, but Cr concentrations samples are around 80% of the certified values.

The reference materials utilized in this study are not certiﬁed for isotopic composition, and, to the authors’ knowledge, there is no other published isotope data for the measured CRMs. Therefore, the accuracy of the presented isotope ratio data cannot be directly evaluated. Isotopic information for CRMs presented in Table 3 can aid future inter-laboratory calibrations of new reference standards for such matrices.

3. Results and discussion

3.1. Element concentrations and isotope ratios in environmental samples Isotope and concentration data for each element studied is presented in Tables 4-12. The tables are formatted to show the soil profile information for all six SEP fractions in the city and suburb locations (with the reported concentrations representing an average of the individual profiles). For some analytes only pooled isotope ratio data is available the F1+F2+F3 and F4+F5+F6 fractions due to insufficient concentrations in the single extractions to make the measurements. Concentration and isotope ratio data for the top soils, lysimetric waters, mushrooms, needles, leaves, lichens, and litter sample are reported, for comparison, below the profile data.

Discussion in this section will start with an examination of multi-element patterns which help identify major processes effecting all the samples. The multi-element patterns will be followed by sections focused on element specific patterns including how each element behaves in the city relative to the suburban samples and how each element behaves as a function of soil depth. Comparing the data between the city and the suburb locations (and between upper and lower soil layers) gives an indication of the isotopic and elemental signature of anthropogenic activity. It is important to point out that differences in soil characteristics/morphology at each location may play an important role in differences observed. Depth-related differences in the isotope ratios of mobile SEP fractions within the soil profiles highlights the potential impact of gradual soil thawing during spring-early summer on the isotopic composition of the element pool available to plant root system (Engström et al., 2010; Ilia Rodushkin et al., 2016; Viers et al., 2015). The final area of discussion will highlight a case study involving environmental pollution.

3.1.1. Multi-element patterns. 3.1.1.1 Batch-type water and Lysimeritc Waters The batch-type water leaches (fraction F1) and lysimetric waters have similar elemental concentrations, with the exception of Cu, Pb and Zn, assuming average ratio between volume

8 of water and soil mass of 10. The concentrations for the latter analytes are several orders of magnitude lower in lysimetric waters than in water leach tests. The concentration patterns are mirrored by the isotopic pattern of most of the elements. Isotope ratios of radiogenic elements in the F1 fractions from upper soil layers overlap the compositions of lysimetric waters. The differences in B, Cd and Fe isotopic compositions between these two matrices are also very minor. However, lysimetric waters were found to be preferentially enriched in the heavy isotopes of Cr, Cu and Zn relative to the batch-type waters.

The overall isotopic and elemental pattern is most readily explained as the product of exchange between rainwater and soil particles. In natural systems weakly sorbed elements such as Cu and Pb likely under-go reversible sorption and their retention is strongly depended on meteoric percolation dynamics.

At least for Cu and Zn this portioning/sorption process induces stable isotope fractionation (Balistrieri et al., 2008).

3.1.1.2 Comparing Biologic Sample Types. Comparing the compositions of the biologic samples i.e. lichen, mushrooms and vascular plant tissue provides a basis for differentiating chemical variability due to biologic processes from variability due to depositional patterns. Lichens do not contain roots and therefore their composition provides information on aerial (precipitation, sea aerosols, long range pollution) sources of elements (Conti and Cecchetti, 2001). Unlike lichens, fruit bodies of mushrooms and vascular plant tissue have much less direct influence from airborne sources, both because of very short duration of exposure and significantly lower surface area/volume ratio. Thus, the radiogenic element composition of mushrooms should resemble lysimetric waters while stable isotopic difference between the two matrices will indicate biologic fractionation. The influence of biologic fractionation on the composition of environmental samples is expected to be most apparent in the tree leaves and needles.

Compared to leaves and needles lichens have significantly higher concentrations of Cr, Pb and Tl and lower concentrations of B and Sr. In general lichens have a significantly narrower range of isotopic variability than mushrooms, leaves and needles for elements including: Cd, Cu, Fe, Pb and Zn. This is probably a reflection of the homogenous nature of Aeolian inputs combined with limited biologic fractionation within lichen (Houben et al., 2014; Jouvin et al., 2012; Kiczka et al., 2010). Top soils can be an important component of Aeolian material, but the Fe concentration data suggests this is not true across our sample area. The isotopic similarities between lichens and top soils is more likely a product of the dry deposition of exogenous material representing an important process for local soil formation.

Mushrooms are clearly accumulating Cd, Cu and Zn relative to the water-leachable element fraction of soils. With the exception of Cd the isotopic composition of these metals also differs between the mushrooms and lysimeric waters. The fact that none of the soil fractions shows Cu and Zn isotopic composition resembling those of mushrooms conﬁrms that the fractionation occurs during element uptake. This is likely due to active uptake of heavy metals by siderophilic root exudates. Lithophilic and chalophilic elements (B, Cd, Pb, Sr and Tl) do not appear to be signiﬁcantly fractionated by mushrooms. The inclusion of a broader array of biochemical

9 processes in higher plants may therefore explain the wide range of inter-species and inter-tree isotopic differences in the leaf and needle data.

The litter layer was sampled to try and detect isotopic shift related to nutrient re-absorption back into tree during late stages of senescence (Wright and Westoby, 2003). However, preliminary analysis of Fe and Pb concentration data demonstrate that the litter has become signiﬁcantly contaminated by soil particles and therefore bears the isotope signature of leaves in late stage of development overlapped with the isotopic composition of the top soils.

3.1.2. Single Element Patterns Boron Total B concentrations in the soil proﬁle decrease with depth in the city samples, but not in the suburban samples (Table 4). Between 80% and 95% of the total B content was associated with the refractory phases (F5-F6) consistent with mineral bound B being the main source of B in the soils (Tabelin et al. 2014). Secondary mineral formation is associated with the preferential incorporation of 10B into the secondary minerals (Louvat et al., 2014) resulting in a heavier residual B composition in solution. This process is the appears be the best explanation of the 11 composition of the lysimetric waters in the present study (6 Bwater-soil 5+11‰). The B isotopic composition of the mobile fraction, despite being characterized by consistently negative b11B values, tended towards a heavier signature in the upper soil layers, probably reﬂecting input from 11 vegetation with plant biomass having distinctly heavier values (6 Bleaves/needles-soil 5+20‰).

The b11B values of our samples vary by around 64‰ (Table 4). This is corresponds to almost half of the published span of b11B values (Xiao et al., 2013). More interestingly, the results for our bio-indicator samples (mushrooms, needles, leaves and lichens) show similar variability to coffee beans collected from vastly different geographic areas i.e. (Serra et al., 2005; Wieser et al., 2001) (Table 4). This indicates it may be difﬁcult to differentiate the biotic and abiotic controls of B fractionation in natural biologic matrices.

There is distinct trend towards heavier B isotopic signatures in the direction soil

Cadmium Significantly higher soil-borne Cd concentrations were measured at the city locations (Table 5) than the suburban location. Such differences are more pronounced in the surface soil layers compared to deeper profiles. For example, the uppermost layers of the city soil profiles contain, on average, about 10 times the concentration found in the deepest inorganic layer indicating that contamination is confined to the upper 30-40 cm. The stratification is significantly less pronounced in the suburb. Overall this suggests anthropogenic activity is a major source of Cd to the urban soil samples.

10 -9 -4 -4 -9 -16 -22 -21 -14 Min. Max. Range ı 2 -9 -17 -12 -12 -7 -3-9-4 -11 -11 -6 -16-7 -10-6 -13 -7 -13 -15 -2 -10 -15 -8 -16 -19 -5 -12 -11 -24 -23 12 -15 -14 -2 -8 -25-10 -24 -4 -17 -2 -2 10 -17 -14 -6 -4 B, ‰ B, ‰ 11 11 į į 7.91 4.38 2.56 2.47 6.47 5.94 6.17 5.56 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 Mean Median &RQFHQWUDWLRQJJ Boron concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.57 0.24 0.48 1.05 1.71 3.87 0-15 cm 0.16 0.06 0.10 0.17 0.83 5.15 16-30 cm31-45 cm46-60 cm 0.21 0.04 0.0416-30 cm 0.1631-45 cm 0.1146-60 cm 0.15 0.15 0.11 0.04 0.06 0.02 0.35 0.05 0.06 0.09 0.11 0.06 0.62 0.09 0.06 0.38 0.04 0.34 2.88 0.04 0.10 1.90 0.07 1.86 0.05 0.60 0.80 0.58 5.01 5.10 4.75 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) o ol875 -7-87-171229 17199-23840 soils877253 Top watersLysimetric MushroomLitter 10Needles14136631 Leaves 7Lichens 1.3 0.6 12 21 21 1.5 2.7 5 NA 0.2 18 2.4 NA 54 6.8 NA 13 1.1 NA 1.6 7 6.6 76 4 -8 4 8 -10 3 13 32 NA 10 NA -2 -24 37 10 NA 3 7 9 9 NA 27 9 13 -7 NA 40 34 27 41 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 4 Table

11 -0.02 -0.10 -0.15 -0.09 Min. Max. Range ı ------Cd, ‰ Cd, ‰ 0.080.170.25 0.040.26 0.40 0.26 0.08 0.330.23 0.09 -0.02 -0.19 -0.06 -0.69 -0.91 -0.19 0.00 -0.31 -0.24 -0.91 - - - - 0.19 114 114 į į 0.756 0.074 0.289 0.081 0.065 0.069 0.061 0.056 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 Mean Median &RQFHQWUDWLRQJJ Cadmium concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.006 0.016 0.427 0.149 0.122 0.036 0-15 cm 0.002 0.010 0.027 0.010 0.013 0.012 16-30 cm31-45 cm46-60 cm 0.003 0.001 0.001 0.012 0.012 0.016 0.137 0.019 0.005 0.056 0.015 0.010 0.049 0.016 0.015 0.032 0.019 0.017 16-30 cm 0.001 0.027 0.012 0.008 0.009 0.011 31-45 cm46-60 cm 0.001 0.001 0.026 0.025 0.008 0.003 0.005 0.006 0.010 0.009 0.010 0.012 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric MushroomLitter 0.04NeedlesLeaves 0.66 0.03Lichens 1.6 0.51 0.05 0.30 1.2 0.06 0.32 0.01 0.33 0.13 NA 0.04 0.13 1.4 0.22 0.29 0.12 0.04 NA 2.40 0.03 0.16 0.03 0.01 NA 5.0 0.06 0.09 0.16 NA 0.83 0.18 0.22 0.08 0.19 0.09 0.07 0.15 0.09 0.14 0.14 0.17 0.23 0.46 0.08 0.10 -0.32 0.34 NA 0.28 0.48 0.07 0.42 0.26 0.20 -0.71 NA 0.29 0.03 -0.52 0.74 0.52 -0.60 NA 0.56 0.05 1.23 1.31 NA 1.08 0.18 1.91 NA 0.13 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 5 Table

12 The sum of the ﬁrst four fractions (F1-F2-F3-F4) contains about 70% of the total soil Cd demonstrating the high mobility of this element in all layers of the soil proﬁles. The abundance of readily mobile Cd in soil and thus its accessibility to plant root uptake was noted by Petit and Rucandio (1999), who suggested that Cd mobility in soil is dictated by variations in pH and redox conditions. Compared to deep soil layer, the largest pool (by almost two orders of magnitude in city locations) of Cd concentration in soil surface is the reducible fraction (F3). This fraction has an isotopic composition of 0‰. Thus, isotope composition of anthropogenic Cd accumulated in the soil is 0‰ which is within the overall reproducibility of method. Biologic fractionation appears the be the best explanation of the heavier Cd isotopic composition in bio-indicators (Pallavicini et al., 2014; Ilia Rodushkin et al., 2016). Higher plants rely on active transport (carrier-mediated) to uptake Cd and Zn, and active transport been associated to a preferential translocation of heavier isotopes (Deng et al., 2014). Preferential incorporation of the heavy isotope would explain the observed shift between soil and leave signatures (average 114 114 6 Cdsoil-leaves5 -0.4‰) as well as the negative b Cd for the residual/immobile Cd pool in the soils.

Published variations of b114Cd values in rock and mineral samples do not exceed 1‰, excluding the extreme fractionation values published for extra-terrestrial material (Rehkämper et al., 2012). This relatively narrow fractionation span (corresponding to 0.25‰ per mass unit) is small compared to light elements such as Li and B, reﬂecting less relative mass difference between Cd isotopes. However, there is limited data on the Cd isotope ratios of non-geologic material. The mean b114Cd values for all our samples, except leaves, are within ± 0.1‰ and exhibit ranges similar to those published for geological samples (Rehkämper et al., 2012). The range in b114Cd values found in leaves from the studied area (almost 2‰, Table 5) demonstrates variability far exceeding both the intermediate precision (Table 3) and variations observed in soil and lysimetric waters (Table 5) thus clearly indicating signiﬁcant isotopic fractionation during Cd uptake.

Chromium There are minor differences in soil Cr concentrations between the sample locations (Table 6). Top soils have slightly lower Cr contents which is probably due to dilution by Cr poor detrital inorganic material. The mobile fractions contain only a minor share of the total Cd pool irrespective of location or depth; on average 83-88% of the total Cr is associated with residual phases. Cr isotopic compositions are very homogenous with practically all ratios in soils being within the uncertainty of the method (Table 3).

The Cr content of mushrooms, needles and leaves was too low for isotope ratio measurements. b53Cr values increased from soil to soil solution to lichens. The latter suggesting differences in the isotopic composition of Cr most likely originate from the deposition of exogenous airborne material. Wet deposition could be the explanation for the heavy signature found for b53Cr in lichens (Pontér et al., 2016). There is a positive correlation between Cr concentrations and increasing proximity to a chromite processing steelworks, and a negative correlation between b53Cr values and increasing proximity to the steelworks implicating the smelter as the primary airborne Cr contributor in the area.

13 - - 0.11 0.04 0.03 0.04 0.04 0.12 Min. Max. Range d bottom layers for F1 and F4 correspond to averaged ı ------Cr, ‰ Cr, ‰ Cr, 0.050.01 0.10 0.12 0.21 0.28 0.04 0.11 0.02 0.01 0.11 0.12 0.22 - - 0.11 - - 0.16 - - 0.12 - - -0.13 -0.12 0.37 0.02 0.03 0.09 -0.16 -0.11 0.38 0.04 0.04 0.11 53 53 į į 27 23 26 25 22 18 21 19 LRQV Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median (n=4) a Chromium concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.09 0.18 0.14 7.5 7.7 9.5 0-15 cm 0.08 0.13 0.13 2.4 8.5 11 16-30 cm31-45 cm46-60 cm 0.09 0.09 0.04 0.10 0.03 0.04 0.12 0.06 0.05 4.3 2.8 1.7 11 11 15 11 8.8 9.2 16-30 cm31-45 cm46-60 cm 0.09 0.08 0.10 0.06 0.04 0.02 0.13 0.14 0.16 2.0 2.0 1.9 5.8 8.6 7.8 10 10 9.0 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb Top soilsTop watersLysimetric MushroomLitter 4.3NeedlesLeaves 21 4.4Lichens 0.12 18 0.4 0.10 1.8 0.23 0.26 16 0.12 3.9 0.21 2.2 NA 0.22 0.01 5.1 0.19 4 NA 1.6 0.15 0.63 0.05 NA 68 1.1 0.10 0.97 NA 1.5 0.95 4.2 0.26 NA 0.26 0.10 NA NA 0.01 0.07 NA NA NA 0.25 NA 0.143 NA NA 0.36 NA 0.27 -0.16 NA NA 0.36 NA 0.02 NA 0.41 NA 0.11 NA NA 0.57 NA NA 0.21 NA NA NA 0.57 NA NA 0.36 6RLOSUR¿OHV &RQFHQWUDWLRQJJ a Values reported as empty entries are due to pooling of several fractions. The isotopic values reported for the outmost top an reported as empty entries are due to pooling of several fractions. Values a YDOXHVRIUHVSHFWLYHO\VRLOSUR¿OHV)))DQG)))IUDFW Table 6 Table

14 Copper Similar to Cd, Cu concentrations found for the top soil horizons from the city locations are signiﬁcantly higher than those from the suburbs (Table 7). Copper concentrations decrease with increasing depth. The smallest Cu pool was found within the F3 fraction in all depths and at all sample locations. Hence, Cu is not associated with Mn and Fe-oxides over the entire soil proﬁle. The exchangeable and reducible phases also constitute minor fractions of the total Cu in the soil, with averages for all the sampling locations and depths of 4% and 1% respectively. Most of the Cu in top soil resides in the oxidizable fraction (F4), generally associated with organic matter, and has a b65Cu value of approximately 0.2‰.

In the present study, the Cu isotopic ratios shows no notable depth-dependent change with heavy signature common in all city soils, while b65Cu of approximately -0.1‰ was found in the suburb. This is consistent with what has been found in other soil systems which receive Cu from an external exogenous source (Fekiacova et al., 2015). In this case we presume the urban samples reﬂect anthropogenic activity while the suburban soils preserve a less perturbed system (Bigalke et al. 2010).

There is signiﬁcant overlap in the Cu isotopic composition of the biologic samples (Table 7), but there does appear to be a shift towards a lighter Cu isotopic in the needles and leaves relative to the soils. A shift towards lighter b65Cu in vegetation is consistent with the previous reports of Cu fractionation in plants e.g. Weinstein et al. (2011).

There are a number of similarities between the behavior of Cu and Fe (discussed in the next section). The isotope fractionation patterns of the two systems follow the same trend of isotopically heavy composition in soils which shift towards lighter values in biological samples. This pattern is characteristic of metal nutrients, with respect to the preferential incorporation of lighter isotopes from the metal source pool (Bullen, 2012; Pallavicini et al., 2014). Biochemical reduction in the rhizosphere, associated with organic exudates, (Ryan et al., 2013) results in the fractionation of the bioavailable pools of Fe and Cu towards lighter values. At least for Fe this fractionation appears to be directly linked the redox changes, rather than the plants themselves, (Skulan et al., 2002) such that the fractionation of Cu in plants may also not be a strictly biological process. This does not inherently conﬂict with studies which have concluded organic siderophores preferentially bind the heavier isotopes of Fe i.e. (Wiederhold et al., 2006; Wu et al., 2011), but rather reﬂects the balance of redox vs biologic fractionation mechanisms.

Iron The total concentrations of Fe in city soil samples are approximately double those found in the suburbs (Table 8). Total Fe concentrations do not change significantly with depth , and most Fe was bound on the F5 and F6 fractions, consistent with primary mineral bound iron. The next most abundant phase was the oxidizable one (F4), consistent with nearly all mobile Fe existing as organic-bound Fe and/or crystalline Fe (oxhydr)oxides (Poulton and Raiswell, 2005). In the city locations, the Fe concentrations in fractions F2, F3 and F4 are significantly higher in top soil layers compared to deep soil. A similar stratification pattern was not found in profiles from suburban samples. This difference is probably a result of either the longer duration of pedogenic formation i.e. the more extensive weathering driven transformation of primary oxides into secondary (oxhydr)oxides or local anthropogenic influence in the city location.

15 - - 0.26 0.31 0.26 0.33 -0.06 -0.10 Min. Max. Range ı ------Cu, ‰ Cu, ‰ 0.150.27 0.30 0.410.35 0.92 1.040.06 - 0.16 0.68 0.25 - 0.22 - 0.51 -0.47 -0.14 - - -0.11 - - - 0.28 0.30 0.11 0.40 0.22 -0.23 0.10 0.05 0.38 0.23 0.29 0.06 65 65 ɷ ɷ 16 39 7.2 6.6 6.4 4.8 4.9 3.8 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median Copper concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.13 0.50 0.06 2.1 3.0 0.58 0-15 cm 0.33 0.58 0.08 26 11 1.2 16-30 cm31-45 cm46-60 cm 0.18 0.16 0.14 0.2116-30 cm 0.1131-45 cm 0.1546-60 cm 0.06 0.03 0.12 0.05 0.09 9.4 0.08 0.17 3.1 0.07 2.0 5.7 0.05 0.06 3.5 0.07 3.9 0.06 0.46 1.9 0.31 1.5 0.34 1.2 2.1 2.6 2.0 0.37 0.58 0.39 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric MushroomLitter 0.6NeedlesLeaves 15 0.5Lichens 44 11 0.2 39 3.2 7 11 7.1 0.4 4.6 3.0 25 NA 5.3 3 1.1 4.5 0.9 NA 10 3.4 84 1.0 2.1 NA 120 3 3.1 NA 5.2 17 7.1 0.76 0.72 0.06 -0.78 0.17 0.01 -0.56 -0.08 -1.09 0.64 0.21 0.74 -1.15 NA -0.35 0.05 0.99 -0.51 -2.4 -0.33 0.58 0.05 NA 0.35 0.51 0.35 0.62 -2.0 0.20 NA 1.02 -1.34 3.02 0.05 -0.28 NA 0.83 0.30 2.08 NA 2.15 0.58 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 7 Table

16 - - 0.18 0.17 0.16 0.04 0.16 0.19 Min. Max. Range ı ------Fe, ‰ Fe, ‰ 0.070.07 -0.67 -1.4 -1.3 -1.1 0.88 0.75 0.18 0.19 0.15 0.11 0.10 0.24 -1.8 0.36 0.20 0.16 0.21 0.21 -2.0 0.33 0.22 0.17 -0.49 - - 0.20 - - -0.87 - - 0.09 - - 56 56 ɷ ɷ 18000 21000 13000 14000 13000 14000 22000 26000 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median Iron concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 45 66 560 530 7000 6100 0-15 cm 26 110 420 2700 16000 6800 31-45 cm46-60 cm 75 3916-30 cm31-45 cm 1546-60 cm 17 73 200 54 160 54 33 740 10 330 7 12000 620 14000 630 5300 780 580 6000 550 5800 460 7400 5100 7000 5500 5400 16-30 cm 48 31 290 2100 14000 6000 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric Mushroom 5200LitterNeedles 19000 5400Leaves 17000Lichens 130 900 9000 40 1180 3600 140 4100 200 NA 5900 72000 330 250 110 160 NA 17 210 110 150 NA 2100 110 33 61 NA 150 440 0.19 0.07 1500 500 0.15 0.10 0.43 0.13 -0.48 -0.20 -0.33 -0.47 0.07 -0.41 0.91 0.45 0.35 -0.29 NA -0.35 -0.09 1.11 0.78 -1.16 -0.26 -0.09 0.41 NA 0.26 0.20 0.04 -1.4 NA -0.84 1.42 -0.14 0.05 NA 0.09 -0.02 1.45 NA 0.93 0.12 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 8 Table

17 All the leachate fractions, with the exception of the hydroxylamine hydrochloride (F3) and acetic acid (F2) phases from deeper layers, from the soils have positive b56Fe values. The F2 phase, containing iron carbonated minerals (Poulton and Raiswell, 2005) exhibits a distinctly light isotopic composition through the entire soil profile, which is consistent with the known fractionation factors between Fe carbonates and other Fe minerals (Blanchard et al., 2009). The F3 phase, corresponding primarily to amorphous iron (oxyhydr)oxides (Poulton and Raiswell, 2005), is consistent with kinetic iron reduction in soils (Skulan et al., 2002; Wiederhold et al., 2006). As mentioned in the preceding paragraph, plant exudates drive iron reduction under eH/ pH conditions where ferrous iron is highly unstable. This appears to create a situation in many soils where iron fractionation near root interfaces adheres to kinetic fractionation [e.g. Kiczka et al. (2010) and Guelke et al., (2010)]. The gradient we observed in the F3 fraction within the soil profiles could be partly explained by changes in root activity with depth. These results indicate that b56Fe dynamics in the soil profiles may be highly dependent on local vegetation. The aforementioned link between root exudates and iron reduction is only applicable to dicots and non-grass monocots (strategy I plants). Strategy II plants utilize a different biochemical process to uptakes iron and therefore do not necessarily cause iron reduction in the soils. Trends in Fe isotopic composition in different matrices are similar to those observed for Cu. Slightly positive b56Fe values are typical for lysimetric waters while isotopically lighter Fe was found in mushrooms, leaves and needles.

The most refractory phases (F5-F6) were characterized by a positive and rather constant b56Fe value throughout the soil profile (Table 8). The deepest soil layer, between 46-60 cm below the surface, represents the least weathered section of the profile and therefore most closely reflecting the original bedrock isotopic composition. The overall average b56Fe values (0.16‰), agrees well with the values reported for granite reference materials (0.13±0.12‰, Chapman et al., 2009). Ancient hydrothermal alteration has however altered the Fe isotopic composition of some Swedish granites (Dideriksen et al. 2010). The amount of Fe found in the F5 fraction in the present study is consistent with the presence of a large pool of hematite (Poulton and Raiswell, 2005). Hematite is the key indicator of hydrothermal alteration in Swedish granites (Dideriksen et al. 2010) indicating that hydrothermal alteration is almost certainly responsible for the heavy isotopic composition measured.

Lead

Pb in the soils appears to reflect endmember relationship between natural bedrock Pb and anthropogenic Pb. Among the elements studied, Pb displays one of the highest concentration differences between top soils from the city and suburbs (Table 9). In deeper soil layers (depths >30 cm), Pb concentrations are constant at both locations suggesting an atmospheric sources for the excess Pb. The absence of significant Pb stratification in the suburban samples suggests lower contributions from long-range atmospheric Pb pollution than what might be expected based on other locations in Northern Sweden (Klaminder et al., 2005). This might be explained by the relatively short time interval since the area emerged from the sea.

The isotopic composition of the presumed atmospheric component of the soil is around 206Pb/207Pb=1.15 and 208Pb/207Pb= 2.39. This was calculated using the Pb composition of the suburban soils and immobile Pb pools in the urban soils, which have a consistent Pb composition (Table 9), as the signature for nature Pb. The presumed atmospheric Pb isotopic composition

18 1.15/2.41 1.17/2.41 1.27/2.44 1.32/2.47 1.24/2.44 1.22/2.43 1.21/2.42 1.21/2.41 Min. Max. Range ı Pb Pb 207 207 Pb/ Pb/ 208 208 Pb/ Pb/ 207 207 Pb/ Pb/ 1.14/2.37 1.16/2.41 1.15/2.39 1.15/2.42 1.17/2.43 1.10/2.34 1.17/2.43 1.20/2.43 1.16/2.43 1.17/2.43 1.20/2.44 1.09/2.33 1.28/2.50 1.23/2.46 1.22/2.49 1.28/2.51 1.35/2.49 1.13/2.34 1.42/2.61 1.35/2.54 1.42/2.60 1.42/2.58 1.54/2.62 1.13/2.34 1.32/2.57 1.35/2.58 1.29/2.54 1.32/2.56 1.38/2.55 1.08/2.32 1.30/2.55 1.25/2.53 1.26/2.48 1.30/2.51 1.37/2.55 1.08/2.32 1.33/2.56 1.23/2.50 1.29/2.52 1.38/2.64 1.42/2.58 1.08/2.32 1.44/2.64 1.22/2.50 1.46/2.61 1.54/2.69 1.55/2.64 1.08/2.32 206 206 11 11 91 35 14 14 12 9.4 for waters) -1 Min. Max. Mean Median ': J/ DW ı -1 -1 Mean Median &RQFHQWUDWLRQJJ Lead concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.31 0.02 3.2 52 30 5.9 0-15 cm 0.04 0.04 0.54 1.2 5.9 5.9 16-30 cm 0.17 0.02 0.75 9.4 18 6.6 31-45 cm 0.08 0.02 0.27 1.3 7.7 4.3 46-60 cm 0.02 0.03 0.08 0.15 4.3 4.9 16-30 cm 0.04 0.08 0.34 0.26 4.6 5.5 31-45 cm 0.04 0.08 0.19 0.52 4.2 6.8 46-60 cm 0.02 0.07 0.07 0.05 3.2 7.4 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Top soilsTop watersLysimetric Mushroom 0.50Litter 0.40Needles 18Leaves 0.17Lichens 0.40 16 0.11 2.3 0.15 0.03 0.14 21 0.29 0.12 NA 1.9 1.1 0.21 0.01 0.11 5 1.8 NA 0.29 0.51 0.02 470 0.4 NA 0.04 0.54 NA 1.3 2.1 1.17/2.45 1.17/2.44 2.6 0.03/0.02 1.10/2.42 1.18/2.45 1.20/2.47 1.17/2.45 0.10/0.05 1.35/2.49 0.04/0.03 1.34/2.50 1.10/2.41 1.18/2.42 0.15/0.01 1.30/2.56 1.18/2.42 1.05/2.30 1.16/2.42 0.20/0.15 0.03/0.03 1.18/2.44 1.77/2.61 1.18/2.43 1.11/2.34 0.72/0.31 NA 1.26/2.47 0.03/0.02 1.14/2.38 0.15/0.21 1.11/2.39 1.14/2.38 1.25/2.49 NA 0.01/0.01 0.14/0.10 1.14/2.37 1.15/2.40 0.01/0.03 NA NA NA Suburb (n=4) 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 9 Table

19 is also consistent with the Pb composition of modern lichen. Furthermore the composition of this natural Pb pool is within the range of agricultural soils reported for northern Sweden (Reimann et al., 2012).

The dominance of the anthropogenic Pb within the critical zone is evident in the mean Pb ratios of soils, lysimetric waters and bio-indicators. All of these matrixes have a signiﬁcantly less radiogenic isotopic composition that the presumed natural Pb baseline. Hence natural Pb does not appear to be a major source of bioavailable Pb.

Figure 2 represents a comparison between Pb isotope ratios in birch leaves and pine needles collected in Luleå (present study) and in Oslo (Reimann et al., 2008). The mean ratios agree well, but there is a marked differences in the spread of the data. The present work has been designed to average temporal heterogeneities (several sampling sessions through each growth period during approximately 5 years) while the needles and leaves in Oslo were all sampled on one single occasion, and therefore only consider spatial variability.

ϮϬϴWďͬϮϬϳWď ϮϬϲWďͬϮϬϳWď

Figure 2 Comparison between Pb isotope ratios obtained leaves and needles in the present study and those proposed by Reimann et al. (2008).

The selective extractions also highlight the influence preferential mineral weathering can have on the isotopic compositions of Pb pools within soils. The residual fraction F6 (HF digestion) of the SEP used in the present study contains a uniform Pb concentrations at all locations and depths which is significantly less radiogenic Pb than the other fractions (Table 9). Similarly, in a study of sediments from 31 lakes in Sweden, Renberg et al. (2002) found lower 206Pb/207Pb ratios in samples prepared using HF digestion compared to those treated using HNO3+HClO4 (10:1 v/v). They attributed this to the Pb contribution from feldspars, which are more resistant to chemical weathering than most other minerals. In the present study potassium feldspar is the main Pb carrier in the silicate soil fractions (Mellqvist et al., 1999). The majority of radiogenic Pb, formed in silicates by 238U, 235U or 232Th decay, might be efficiently leached in the first five

20 SEP fractions This leaves the ﬁnal HF fractionation with a less radiogenic isotope composition reﬂecting the time of mineral formation approximately 2 Ga years ago.

Strontium Total Sr concentrations display little variability across the dataset (Table 10). Top soils have slightly lower Sr concentrations, most likely reﬂecting dilution effects by organic matter. Most of the Sr is bound in the residual fractions (F5-F6), suggesting most of the element is bound in refractory mineral phases within the soil.

Mean 87Sr/86Sr ratios in all environmental matrices from Luleå fall into a relatively narrow range (0.729-0.734) with little differences in isotopic composition between soil and bio-indicators. Thus, unlike B, Sr in needles, leaves and lichens appears to be of a terrestrial rather than marine origin; seawater aerosols have less radiogenic Sr isotope ratios (Rodushkin et al., 2007). To a significant degree the homogeneity of the 87Sr/86Sr ratios in the mean samples is expected given the composition of the local bedrock (Mellqvist et al., 1999; Åberg and Wickman, 1987). Sr isotope ratios, primary variations in radiogenic87Sr, have been extensively used for authentication and provenance studies (Wiederhold, 2015). Waters, crops, livestock, a variety of food products and ultimately man are known to reflect the Sr isotope ratios typical for the geological setting in which they occur (Song et al., 2014). Strontium isotopic systematics have been also used, in combination with other systems, to trace anthropogenic sources in the environment (Geagea et al., 2008a, 2008b; Hissler et al., 2008). However, the analytic variability of the Sr ratios from individual extractions, 0.721 to 0.754 (Table 10), may reflect the impact of small scale heterogeneity on single isotope measurements. Some of this variability is likely due to mineral processes. The exchangeable and carbonate fractions of sequential leached soil, have lower 87Sr/86Sr ratios than that in the silicate fraction Song et al. (2014). This is consistent with a less radiogenic Sr vlaues in plants and therefore in the organic-rich topsoil horizon, as observed for soil profiles from the suburb in this study. Overall heterogeneities on the scale we found have been identified in a number of environmental studies i.e. Degryse et al. (2010) and Chang et al. (2016) where they represented serious challenges in link isotope measurements to sample provenance.

Thallium Total Tl concentrations in soils from city and suburban locations vary over a very limited range of approximately 0.4-0.6 μg g-1 at all depths (Table 11). Thallium predominantly resides in the residual fractions (98% of Tl recovered in leachates F5-F6) most likely in the crystalline structure of silicate minerals such as potassium feldspar and quartz (Vanek et al. 2010). The remaining Tl, associated with reducible and oxidizable phases, is signiﬁcantly higher in top soil from the city than in the suburb. Our ﬁndings are in very close to Tl speciation data for the rhizospheric soils reported by Jia et al. (2013) who utilized an almost identical SEP.

There is very limited data on Tl isotopes in environmental samples. Kersten et al. (2014) reported Tl isotope ratio data from contaminated soils quoting a b205Tl value of 5-0.3‰ relative to the natural background [there is no international zero-b reference standard for Tl], with average Tl concentrations of 0.75 μg g-1. In soil proﬁles, they found a lighter Tl signature with increasing depth in both the extractable and the residual fractions which is similar to our data (Table 11). In our soils the F1 and F2 fractions have positive b205Tl values relative to both our reference

21 - - 0.727 0.732 0.727 0.728 0.732 0.733 Min. Max. Range ı Sr Sr 86 86 ------0.727 -0.729 - - 0.732 - - 0.733 - - - 0.7300.727 0.7340.724 0.7340.727 0.734 0.734 0.733 0.743 0.733 0.728 0.735 0.724 0.732 0.726 0.733 0.727 0.727 0.735 0.731 0.747 0.727 0.727 Sr/ Sr/ 87 87 146 179 188 167 154 171 180 148 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median Strontium concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.3 0.6 1.2 0.62 6.5 146 0-15 cm 0.5 2.3 11 5.4 6.4 141 31-45 cm46-60 cm 0.1 0.1 1.5 1.4 0.41 0.32 0.33 0.31 6.0 5.1 179 188 16-30 cm31-45 cm46-60 cm 0.3 0.1 0.1 1.716-30 cm 1.2 1.2 5.1 0.2 1.0 0.60 3.0 0.70 1.1 0.50 6.0 7.0 0.69 7.7 138 0.42 161 170 4.8 148 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric MushroomLitter 12NeedlesLeaves 25Lichens 10 0.5 23 0.3 2 28 18 15 34 8.3 0.2 NA 15 8 31 3 7.4 0.1 NA 16 12 16 150 3.9 NA 1.8 2 2.5 NA 6 63 16 120 0.733 0.731 0.729 0.733 0.731 0.727 0.001 0.731 0.004 0.006 0.732 0.734 NA 0.723 0.733 0.730 0.721 0.733 0.733 0.739 NA 0.731 0.730 0.750 0.010 0.007 0.016 0.006 0.005 0.039 NA 0.724 0.725 0.721 0.754 NA 0.742 0.742 0.030 NA 0.017 0.024 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 10 Table

22 - - -0.42 -0.46 -0.50 -0.52 -0.51 -0.62 Min. Max. Range ı ------Tl, ‰ Tl, ‰ 0.360.360.20 0.100.18 0.12 -0.10 -0.02 -0.07 0.00 -0.21 -0.32 -0.26 -0.39 -0.28 -0.45 -0.38 -0.40 -0.27 -0.47 -0.28 -0.49 -0.54 -0.57 -0.08 --0.12 - - -0.52 - - -0.62 - - - 205 205 ɷ ɷ 0.45 0.50 0.36 0.39 0.57 0.56 0.62 0.61 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median Thallium concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 0.004 0.0020-15 cm 0.024 0.046 0.001 0.058 0.001 0.32 0.004 0.007 0.081 0.48 16-30 cm31-45 cm46-60 cm 0.001 0.0003 0.001 0.0002 0.000716-30 cm 0.0006 0.009 0.00231-45 cm 0.00246-60 cm 0.024 0.006 0.001 0.004 0.0007 0.070 0.051 0.0008 0.0005 0.0007 0.064 0.40 0.003 0.0006 0.30 0.002 0.31 0.002 0.005 0.004 0.003 0.054 0.074 0.056 0.50 0.53 0.55 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric MushroomLitter 0.04Needles 0.26Leaves 0.04Lichens 0.015 0.20 0.024 0.009 0.030 0.010 0.002 0.26 0.016 0.006 0.010 NA 0.07 0.023 0.03 0.002 0.004 0.020 0.009 0.081 NA 3.2 0.007 0.001 0.024 0.001 NA 0.090 0.004 0.043 NA 0.060 -0.14 -0.17 -0.18 -0.61 -0.28 0.07 -0.61 -0.31 0.30 -0.22 -0.30 -0.38 0.17 -0.53 -0.41 -0.05 -0.62 -0.39 -0.54 0.21 -0.95 NA 0.17 0.42 0.16 -0.20 0.10 -0.63 NA -0.42 1.04 0.70 -0.72 0.16 -0.09 NA -0.41 0.89 0.33 NA 0.31 NA 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 11 11 Table

23 standard and the average environmental values. . This is then reﬂected in signiﬁcantly higher b205Tl in lysimetric waters than in top soils, which in turn contributes to the heavier Tl ratios found in mushrooms, needles and leaves (Table 11).

Zinc Zinc concentrations follow a similar pattern to one described for Pb. The top layers of soil proﬁles from the city are signiﬁcantly enriched in Zn while deeper soil samples are characterized by uniform concentrations (Table 12).

For all samples, Zn was mainly found in the ﬁrst four fractions (80-90% leached in F1-F4), suggesting a relatively high mobility of the element in soils. This agrees with previous suggestions that Zn is preferentially mobilized in soils (Beesley et al., 2010; Moreno-Jiménez et al., 2009). The Zn depth distribution in soils from both locations correlates with Fe oxyhydroxides (Table 8). Zn might be distributed between inner and outer complexes on the Fe oxyhydroxides, with the weaker sorbed pool leaching in the F2 fraction and the stronger one in the F3 fraction following oxyhydroxide reduction. The Zn concentration peaks at different depths in the soil proﬁles (Table 12) because oxyhydroxide concentrations (fraction F3) also peak at different depths (Table 8).

The Zn isotope ratios of the leachate fractions with the highest Zn content exhibit a marked difference between surficial and deep layers (F2-F3). This isotopic offset is potentially consistent with the deposition of an atmosphere Zn, with a composition equal to or slighter lighter than the bulk top soils. This is significantly lighter than the Zn composition of the deep soil layers (46-60cm), which have a values within the Zn composition of the local bedrock (-0.05- 1.5‰; converted from the “Lyon” JMC 3-0749 L standard) (Fekiacova et al., 2015). Bigalke et al. (2010) found a similar pattern of Zn isotopic compositional change with soils depth and attributed the pattern to the deposition of dust-derived Zn. In a region which does not receive significant pollution inputs, Viers et al. (2015) found the Zn isotopic composition of soils was constant and that there was no measurable Zn isotopes fractionation between parent rocks and soils. However, atmospheric dust deposition is not the only potential explanation for the pattern. Fekiacova et al. (2015) attributed Zn fractionation in soil horizons with sorption onto the Fe oxyhydroxides.

The mean b66Zn of top soils and lysimetric waters is approximately +0.2‰, while a signiﬁcantly higher value of +0.6‰ was found in mushrooms indicating preferential uptake of heavier isotopes. The mechanism(s) of uptake of heavy metal ions by fungi can be via either an active or passive (Kapoor et al. 1999), and the prevailing heavier Zn signature in mushrooms compared to the soil pool suggests an active uptake strategy. On the contrary, Zn in needles and leaves is isotopically lighter most likely as a result of translocation through the shoot(Fujii et al., 2014).

24 - - 0.15 0.03 -0.04 -0.14 -0.01 -0.09 Min. Max. Range ı ------Zn, ‰ Zn, ‰ 0.01 -0.07 -0.09 -0.06 -0.01 0.34 0.01 0.02 0.04 0.00 0.30 0.37 -0.17 -0.17 -0.14-0.23 -0.11 - -0.31 0.32 - 0.11 - - -0.13 - - 0.10 - - -0.13 -0.07 0.01 0.02 -0.24 0.30 66 66 ɷ ɷ 91 84 110 140 560 140 160 560 Min. Max. Mean Median for waters) -1 ı ': J/ DW -1 -1 &RQFHQWUDWLRQJJ Mean Median Zinc concentrations and isotopic composition in environmental samples from Luleå 0-15 cm 7.9 75 27 5.6 15 12 0-15 cm 7.2 230 180 96 38 13 16-30 cm31-45 cm46-60 cm 17 5.4 4.916-30 cm 10031-45 cm 4946-60 cm 35 260 10 35 7.6 120 12 10 49 18 110 45 7.4 28 24 23 13 14 22 14 5.5 3.4 11 11 3.6 11 14 15 11 10 10 Fractions F1 F2 F3 F4 F5 F6 Total F1 F2 F3 F4 F5 F6 Total City (n=2) Suburb (n=4) Top soilsTop watersLysimetric MushroomLitter 5.0NeedlesLeaves 93 4.6Lichens 120 83 3.2 110 220 51 220 72 1.3 50 NA 57 50 180 28 11 NA 40 60 19 140 640 NA 260 8 22 11 NA 36 800 96 70 0.15 0.18 0.14 0.58 0.16 0.13 0.00 0.55 -0.05 0.20 -0.13 -0.09 0.36 NA -0.11 0.30 -0.47 -0.08 -0.16 0.04 NA 0.21 0.69 0.35 0.30 1.25 0.07 NA -0.94 1.16 -0.96 1.41 0.06 NA 0.24 0.67 -0.07 1.18 NA 1.64 0.11 0.17 6RLOSUR¿OHV &RQFHQWUDWLRQJJ Table 12 Table

25 3.2. /DQG¿OOVDQGLQGXVWULDOZDVWHV Pollution source tracing using isotopic data can be challenging for a number of reasons: < the potential source compositions are often not well deﬁned end-members;

< the potential source compositions may overlap natural environmental variability;

< even where end-member compositions are distinct and well-constrained there might be a large number of potential sources;

< for stable isotopes, numerous post-deposition fractionating processes may alter the original ratios (e.g. Wiederhold, 2015). In order to assess the severity of these complicating factors in the area studied, concentration and isotopic data on waste samples from local industries were compared to the composition of two heavily contaminated soils (landﬁlls) (Table 13).

Table 13 į-values found in this study for different industrial matrices (bold text) compared to values reported in literature for ashes and fumes Pb, μg Cd, μg g-1 į114Cd Cr, μg g-1 į53Cr Cu, μg g-1 į63Cu Fe, % į56Fe Zn, μg g-1 į66Zn 206Pb/207Pb g-1

/DQG¿OO$ 1.5 0.0 450 0.2 2700 -1.0 15 0.4 1100 1.4 40 1.32

/DQG¿OO% 0.6 -1.5 2600 0.1 170 -1.4 10 1.0 180 -0.2 170 1.17

)HVODJ 0.2 0.1 4100 0.1 100 0.7 12 0.9 150 -0.7 250 1.16

&XVODJ 11 -0.2 650 0.1 16000 0.1 35 0.4 12000 0.0 200 1.17

)O\DVK 1.1 1.7 50 0.2 25 0.3 2.7 0.2 50 0.6 15 1.21

Both landfills have Cd concentrations within range found for top soils, but landfill B has very light Cd isotopic composition. The highest Cd level, found in Cu slag, has an isotopic composition undistinguishable from the mean for top soils. Therefore using Cd isotopes to trace this pollution source in soils is problematic. Fly ash, on the contrary, is significantly enriched in heavier isotopes (well above higher extremes for ranges in environmental matrixes tested) , but such isotopic signature cannot be seen in lichens – bio indicator that should primarily be affected by fly ash. Thus the Cd in landfill B either originated from unknown, very fractionated source, or its original isotopic composition was changed significantly in process(es) that occurred after deposition.

Compared to local soils, Cr concentrations are signiﬁcantly elevated in both landﬁlls and slags. However, the very narrow span of isotopic compositions (b53Cr of 0.1‰-0.2‰) overlaps the range found for top soils, making Cr isotopic information useless for pollution sourcing in this case.

26 Both landfill sites are enriched in Cu, which has a light isotopic composition, relative to the surrounding soil. The highest Cu concentration is in Cu slag. However, similarly to the case of Cd, the Cu isotopic composition of Cu slag is undistinguishable from the mean isotopic composition for top soils. Both fly ash and Fe slag are enriched in heavy Cu isotopes and the very light isotopic composition of landfills cannot be caused by any of these sources unless substantial alteration of original ratios has occurred.

The landfill soils contain about 5 times the concentrations of Fe of the surrounding natural soils. All wastes and landfills have heavier Fe isotopic composition than mean value for local top soils. The Fe concentrations in landfill B and slag production are similar and isotopic signatures of the two are identical within measurement uncertainty. Mass balance constraints restrict the post-deposition fractionation of elements presented in solid samples at concentration level in the order of 10%. Thus Fe slag can be identified as potentially significant pollution source in landfill B.

Soils from landfill A are contaminated with Zn relative to the background soil Zn concentrations. b64Zn in the local waste materials ranges from -0.7‰ in Fe slag to +0.6‰ in fly ash, with Cu slag – the material highest in Zn – having value of 0.0‰. The Zn isotopic composition of Landfill A is heavier than any of the potential contamination sources tested. The lack of significant Zn enrichment in landfill B helps distinguish between Cu and Fe slag as a dominant input into the region. Cu slag is highly enriched in Zn, while Fe slag is poor in Zn, such that only the deposition of Fe slag would be consistent with low total Zn concentrations.

Neither of Landfill locations are significantly enriched in Pb relative to the top soils. Landfill A, with lowest Pb concentrations, has a 207Pb/206Pb ratio that is very close to mean value for top soils probably indicating mixing/dilution of deposited wastes deriving from background soils. The 207Pb/206Pb isotope ratio in Landfill B is close to those for both Fe and Cu slags therefore both slags can be potential sources of Pb in the landfill with insignificant mixing with local soils. However, as identified in the preceding paragraphs, Cu slag can be excluded based on the other elemental patterns. Consequently, from the available body of concentration and isotopic data, Fe slag seems to be the best match for material deposited in landfill B, though origin of Landfill A is much less certain.

4. Conclusions

All the samples in this study were collected from spatially-limited area, but a number of the analytes still displayed a signiﬁcant range of concentration and isotopic values. Hence this study highlights the need to consider natural variability, which often far exceeds the intermediate precision of the analytical method, when interpreting results of environmental studies. For example, the use of isotopic information in bio-indicators requires accounting for temporal and spatial variations in exposure patterns well as biologic fractionation within the organisms. The sequential extraction scheme, combined with the isotope ratio measurements of separated fractions. provides a useful additional dimension for identifying and isolating drivers of these variations. In the future sequential extractions could be used to help understand the dynamic process which cause isotopic fractionation over various temporal and special scales. Examples include: changes in the isotopic composition of leaves during the growing season (Kiczka et al., 2010; Ilia Rodushkin et al., 2016; Viers et al., 2015); isotopic variations in river water

27 (Engström et al., 2010); and better understanding of the fate of pollutants in soils. In particular, the elemental form(s) and associated isotopic compositions, as well as soil compartments in which pollutants become accumulated, can be delineated. The effect of soil-plant cycling on elements in botanical matrixes resulted in shifts towards lighter isotopic compositions for Cu, Fe and Zn and heavier ones for B, Cd, Cr and Tl. This is most likely a reflection of how differences in uptake mechanisms and/or the presence of pronounced aerial contributions alter the isotopic signatures of material relative to native soils. Expanding the number of elements utilized in isotope tracing provides a powerful way to decipher sources and fate of environmental exposure by adding degrees of freedom to the process. What is now needed is a more comprehensive database of the substrate compositions and natural variations within substrates to fully utilize the additional information. From both a concentration and isotope standpoint Cd and Cu do not appear to be useful indicators of the pollution inputs into the landfills. Fe and Pb isotopes and Zn concentrations potentially preserve a more reliable way to differentiate the key pollution contributors to the landfills. Specifically Fe slag appears to be a key input into one of the landfills but not the other. The data underscores the need to collect data on a wide number of elements in order to confidently trace pollution sources in the environment. Studies on post-depositional isotopic aberrations/alterations of stable isotope ratios in waste materials are also needed to reduce the ambiguity associated with multi-element tracing studies.

28 References Arnold, T., Markovic, T., Kirk, G.J.D., Schönbächler, M., Rehkämper, M., Zhao, F.J., Weiss, D.J., 2015. Iron and zinc isotope fractionation during uptake and translocation in rice (Oryza sativa) grown in oxic and anoxic soils. Comptes Rendus Geosci. 347, 397–404. doi:10.1016/j. crte.2015.05.005 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, 311–328. doi:10.1016/j. gca.2007.11.013 Baroni, M. V., Podio, N.S., Badini, R.G., Inga, M., Ostera, H.A., Cagnoni, M., Gautier, E.A., García, P.P., Hoogewerff, J., Wunderlin, D.A., 2015. Linking Soil, Water, and Honey Composition To Assess the Geographical Origin of Argentinean Honey by Multielemental and Isotopic Analyses. J. Agric. Food Chem. 63, 4638–4645. doi:10.1021/jf5060112 Beesley, L., Moreno-Jiménez, E., Clemente, R., Lepp, N., Dickinson, N., 2010. Mobility of arsenic, FDGPLXPDQG]LQFLQDPXOWLHOHPHQWFRQWDPLQDWHGVRLOSUR¿OHDVVHVVHGE\LQVLWXVRLOSRUH water sampling, column leaching and sequential extraction. Environ. Pollut. 158, 155–160. doi:10.1016/j.envpol.2009.07.021 %LHOLFND*LHáGRĔ$5\áNR(ĩDPRMü.'LVWULEXWLRQELRDYDLODELOLW\DQGIUDFWLRQDWLRQ of metallic elements in allotment garden soils using the BCR sequential extraction procedure. Polish J. Environ. Stud. 22, 1013–1021. Bigalke, M., Weyer, S., Kobza, J., Wilcke, W., 2010. Stable Cu and Zn isotope ratios as tracers of sources and transport of Cu and Zn in contaminated soil. Geochim. Cosmochim. Acta 74, 6801–6813. doi:10.1016/j.gca.2010.08.044 Blanchard, M., Poitrasson, F., Méheut, M., Lazzeri, M., Mauri, F., Balan, E., 2009. Iron isotope IUDFWLRQDWLRQEHWZHHQS\ULWH )H6 KHPDWLWH )H2 DQGVLGHULWH )H&2 $¿UVWSULQFLSOHV density functional theory study. Geochim. Cosmochim. Acta 73, 6565–6578. doi:10.1016/j. gca.2009.07.034 Bullen, T.D., 2012. Stable isotopes of transition and post-transition metals as tracers in environmental studies, in: Baskaran, M. (Ed.), Handbook of Environmental Isotope Geochemistry. Springer, Berlin, Heidelberg, pp. 177–203. doi:10.1007/978-3-642-10637-8 Chang, C.-T., You, C.-F., Aggarwal, S.K., Chung, C.-H., Chao, H.-C., Liu, H.-C., 2016. Boron and strontium isotope ratios and major/trace elements concentrations in tea leaves at four major tea growing gardens in Taiwan. Environ. Geochem. Health 38, 737–748. doi:10.1007/s10653-015- 9757-1 Chapman, J.B., Weiss, D.J., Shan, Y., Lemburger, M., 2009. Iron isotope fractionation during leaching of granite and basalt by hydrochloric and oxalic acids. Geochim. Cosmochim. Acta 73, 1312– 1324. doi:10.1016/j.gca.2008.11.037 &KUDVWQê9ýDGNRYi(9DQČN$7HSHU/&DEDOD-.RPiUHN0&DGPLXPLVRWRSH IUDFWLRQDWLRQZLWKLQWKHVRLOSUR¿OHFRPSOLFDWHVVRXUFHLGHQWL¿FDWLRQLQUHODWLRQWR3E±=QPLQLQJ and smelting processes. Chem. Geol. 405, 1–9. doi:10.1016/j.chemgeo.2015.04.002

29 Cloquet, C., Carignan, J., Libourel, G., Sterckeman, T., Perdrix, E., 2006. Tracing source pollution in soils using cadmium and lead isotopes. Environ. Sci. Technol. 40, 2525–2530. Cloquet, C., Rouxel, O., Carignan, J., Libourel, G., 2005. Natural Cadmium Isotopic Variations in Eight Geological Reference Materials (NIST SRM 2711, BCR 176, GSS-1, GXR-1, GXR- 2, GSD-12, Nod-P-1, Nod-A-1) and Anthropogenic Samples, Measured by MC-ICP-MS. Geostand. Geoanalytical Res. 29, 95–106. doi:10.1111/j.1751-908X.2005.tb00658.x Conti, M.E., Cecchetti, G., 2001. Biological monitoring: lichens as bioindicators of air pollution assessment--a review. Environ. Pollut. 114, 471–492. Degryse, P., Shortland, A., De Muynck, D., Van Heghe, L., Scott, R., Neyt, B., Vanhaecke, F., 2010. Considerations on the provenance determination of plant ash glasses using strontium isotopes. J. Archaeol. Sci. 37, 3129–3135. doi:10.1016/j.jas.2010.07.014 Deng, T., Cloquet, C., Tang, Y., Sterckeman, T., 2014. Nickel and zinc isotope fractionation in hyperaccumulating and nonaccumulating plants. Environ. Sci. Technol. 48, 11926–11933. doi:10.1021/es5020955 Dideriksen, K., Christiansen, B.C., Frandsen, C., Balic-Zunic, T., Mørup, S., Stipp, S.L.S., 2010. Paleo-redox boundaries in fractured granite. Geochim. Cosmochim. Acta 74, 2866–2880. doi:10.1016/j.gca.2010.02.022 Engström, E., Rodushkin, I., Ingri, J., Baxter, D.C., Ecke, F., Österlund, H., Öhlander, B., 2010. Temporal isotopic variations of dissolved silicon in a pristine boreal river. Chem. Geol. 271, 142–152. doi:10.1016/j.chemgeo.2010.01.005 Fekiacova, Z., Cornu, S., Pichat, S., 2015. Tracing contamination sources in soils with Cu and Zn isotopic ratios. Sci. Total Environ. 517, 96–105. doi:10.1016/j.scitotenv.2015.02.046 Foster, G.L., Pogge Von Strandmann, P.A.E., Rae, J.W.B., 2010. Boron and magnesium isotopic composition of seawater. Geochemistry, Geophys. Geosystems 11, 1–10. doi:10.1029/2010GC003201 Fujii, T., Moynier, F., Blichert-Toft, J., Albarède, F., 2014. Density functional theory estimation of isotope fractionation of Fe, Ni, Cu, and Zn among species relevant to geochemical and biological environments. Geochim. Cosmochim. Acta 140, 553–576. doi:10.1016/j.gca.2014.05.051 Geagea, M.L., Stille, P., Gauthier-Lafaye, F., Millet, M., 2008a. Tracing of industrial aerosol sources in an urban environment using Pb, Sr, and Nd isotopes. Environ. Sci. Technol. 42, 692–698. doi:10.1021/es071704c Geagea, M.L., Stille, P., Gauthier-Lafaye, F., Perrone, T., Aubert, D., 2008b. Baseline determination of the atmospheric Pb, Sr and Nd isotopic compositions in the Rhine valley, Vosges mountains (France) and the Central Swiss Alps. Appl. Geochemistry 23, 1703–1714. doi:10.1016/j. apgeochem.2008.02.004 Guelke, M., von Blanckenburg, F., Schoenberg, R., Staubwasser, M., Stuetzel, H., 2010. Determining the stable Fe isotope signature of plant-available iron in soils. Chem. Geol. 277, 269–280. doi:10.1016/j.chemgeo.2010.08.010 Gwiazda, R., Woolard, D., Smith, D., 1998. Improved lead isotope ratio measurements in environmental and biological samples with a double focussing magnetic sector inductively coupled plasma mass spectrometer (ICP-MS). J. Anal. At. Spectrom. 13, 1233–1238. doi:10.1039/a804062a

30 Hissler, C., Stille, P., Krein, A., Geagea, M.L., Perrone, T., Probst, J.L., Hoffmann, L., 2008. Identifying the origins of local atmospheric deposition in the steel industry basin of Luxembourg using the chemical and isotopic composition of the lichen Xanthoria parietina. Sci. Total Environ. 405, 338–344. doi:10.1016/j.scitotenv.2008.05.029 Houben, D., Sonnet, P., Tricot, G., Mattielli, N., Couder, E., Opfergelt, S., 2014. Impact of root- induced mobilization of zinc on stable Zn isotope variation in the soil-plant system. Environ. Sci. Technol. 48, 7866–7873. doi:10.1021/es5002874 Jia, Y., Xiao, T., Zhou, G., Ning, Z., 2013. Thallium at the interface of soil and green cabbage %UDVVLFDROHUDFHD/YDUFDSLWDWD/ 6RLO±SODQWWUDQVIHUDQGLQÀXHQFLQJIDFWRUV6FL7RWDO Environ. 450–451, 140–147. doi:10.1016/j.scitotenv.2013.02.008 Jouvin, D., Weiss, D.J., Mason, T.F.M., Bravin, M.N., Louvat, P., Zhao, F., Ferec, F., Hinsinger, P., Benedetti, M.F., 2012. Stable isotopes of Cu and Zn in higher plants: Evidence for Cu reduction at the root surface and two conceptual models for isotopic fractionation processes. Environ. Sci. Technol. 46, 2652–2660. doi:10.1021/es202587m Kapoor, A., Viraraghavan, T., Cullimore, D.R., 1999. Removal of heavy metals using the fungus Aspergillus niger. Bioresour. Technol. 70, 95–104. doi:10.1016/S0960-8524(98)00192-8 Kersten, M., Xiao, T., Kreissig, K., Brett, A., Coles, B.J., Rehkämper, M., 2014. Tracing anthropogenic thallium in soil using stable isotope compositions. Environ. Sci. Technol. 48, 9030–9036. doi:10.1021/es501968d Kiczka, M., Wiederhold, J.G., Kraemer, S.M., Bourdon, B., Kretzschmar, R., 2010. Iron isotope fractionation during Fe uptake and translocation in alpine plants. Environ. Sci. Technol. 44, 6144–6150. doi:10.1021/es100863b Klaminder, J., Bindler, R., Emteryd, O., Renberg, I., 2005. Uptake and recycling of lead by boreal forest plants: Quantitative estimates from a site in northern Sweden. Geochim. Cosmochim. Acta 69, 2485–2496. doi:10.1016/j.gca.2004.11.013 Liu, H.-C., You, C.F., Chen, C.Y., Liu, Y.C., Chung, M.T., 2014. Geographic determination of coffee beans using multi-element analysis and isotope ratios of boron and strontium. Food Chem. 142, 439–445. doi:10.1016/j.foodchem.2013.07.082 Louvat, P., Hartman, J., Hosono, T., Kiyoshi, I., Amann, T., Bouchez, J., Gaillardet, J., 2014. Behaviour of boron isotopes in the streams and springs of Aso caldera , Kyushu , Japan. Lundberg, C., Peterson, L., 2005. Land use history of central Luleå a case study in the use of historical maps together with modern geographic municipal information. Appl. GIS 1, 1–30. Maréchal, C.N., Télouk, P., Albarède, F., 1999. Precise analysis of copper and zinc isotopic compositions by plasma-source mass spectrometry. Chem. Geol. 156, 251–273. doi:10.1016/ S0009-2541(98)00191-0 Markert, B., 1989. Multi-element analysis in ecosystems: basic conditions for representative VDPSOLQJRISODQWPDWHULDOV)UHVHQLXV¶=HLWVFKULIWIU$QDO&KHPLH±GRL BF00474249 0DXHU23DOiWRYi(7KHUROHRIURRWV\VWHPLQVLOYHUELUFK %HWXODSHQGXOD5RWK GLHEDFN in the air-polluted area of Krušné hory Mts . Stand 2003, 191–199.

31 Mellqvist, C., Ohlander, B., Skiold, T., Wikstrom, A., 1999. The Archaean–Proterozoic Palaeoboundary LQWKH/XOHnDUHDQRUWKHUQ6ZHGHQ¿HOGDQGLVRWRSHJHRFKHPLFDOHYLGHQFHIRUDVKDUSWHUUDQH boundary. Precambrian Res. 96, 225–243. doi:10.1016/S0301-9268(99)00011-X Moreno-Jiménez, E., Peñalosa, J.M., Manzano, R., Carpena-Ruiz, R.O., Gamarra, R., Esteban, E., 2009. Heavy metals distribution in soils surrounding an abandoned mine in NW Madrid 6SDLQ DQG WKHLU WUDQVIHUHQFH WR ZLOG ÀRUD J. Hazard. Mater. 162, 854–859. doi:10.1016/j. jhazmat.2008.05.109 Pallavicini, N., Engström, E., Baxter, D.C., Öhlander, B., Ingri, J., Rodushkin, I., 2014. Cadmium isotope ratio measurements in environmental matrices by MC-ICP-MS. J. Anal. At. Spectrom. 29, 1570–1584. doi:10.1039/C4JA00125G Petit, M.D., Rucandio, M.I., 1999. Sequential extractions for determination of cadmium distribution LQFRDOÀ\DVKVRLODQGVHGLPHQWVDPSOHV$QDO&KLP$FWD±GRL6 2670(99)00487-0 Pontér, S., Pallavicini, N., Engström, E., Baxter, D.C., Rodushkin, I., 2016. Chromium isotope ratio measurements in environmental matrices by MC-ICP-MS. J. Anal. At. Spectrom. 31, 1464-1471. doi:10.1039/C6JA00145A Poulton, S.W., Raiswell, R., 2005. Chemical and physical characteristics of iron oxides in riverine and glacial meltwater sediments. Chem. Geol. 218, 203–221. doi:10.1016/j.chemgeo.2005.01.007 Poutanen, M., Steffen, H., 2014. Land uplift at Kvarken archipelago and High Coast UNESCO World Heritage area 16, 16179. Rehkämper, M., Wombacher, F., Horner, T.J., Xue, Z., 2012. Natural and anthropogenic Cd isotope variations, in: Baskaran, M. (Ed.), Handbook of Environmental Isotope Geochemistry. Springer, Berlin, Heidelberg, pp. 125–154. doi:10.1007/978-3-642-10637-8 Reimann, C., Flem, B., Arnoldussen, A., Englmaier, P., Finne, T.E., Koller, F., Nordgulen, Ø., 2008. The biosphere: A homogeniser of Pb-isotope signals. Appl. Geochemistry 23, 705–722. doi:10.1016/j.apgeochem.2007.12.002 Reimann, C., Flem, B., Fabian, K., Birke, M., Ladenberger, A., Négrel, P., Demetriades, A., Hoogewerff, J., 2012. Lead and lead isotopes in agricultural soils of Europe – The continental perspective. Appl. Geochemistry 27, 532–542. doi:10.1016/j.apgeochem.2011.12.012 Renberg, I., Bindler, R., Emteryd, O., 2002. Stable lead isotopes and lake sediments — a useful combination for the study of atmospheric lead pollution history. Sci. Total Environ. 292, 45–54. doi:10.1016/S0048-9697(02)00032-3 Rodushkin, I., Bergman, T., Douglas, G., Engström, E., Sörlin, D., Baxter, D.C., 2007. Authentication of Kalix (N.E. Sweden) vendace caviar using inductively coupled plasma-based analytical techniques: evaluation of different approaches. Anal. Chim. Acta 583, 310–318. doi:10.1016/j. aca.2006.10.038 Rodushkin, I., Pallavicini, N., Engström, E., Sörlin, D., Öhlander, B., Ingri, J., Baxter, D.C., 2016. Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two-column matrix separation. J. Anal. At. Spectrom. 31, 220–233. doi:10.1039/C5JA00274E

32 Rodushkin, I., Pallavicini, N., Engström, E., Sörlin, D., Öhlander, B., Ingri, J., Baxter, D.C., 2016. Assessment of the natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotopic compositions in leaves, needles and mushrooms using single sample digestion and two- column matrix separation. J. Anal. At. Spectrom. 31, 220-233. doi:10.1039/c5ja00274e Rosner, M., Pritzkow, W., Voerkelius, S., Vogl, J., 2009. Boron isotopes - a new tracer for the origin and authenticity of food, in: Boron Isotopes - a New Tracer for the Origin and Authenticity of Food. Ryan, B.M., Kirby, J.K., Degryse, F., Harris, H., McLaughlin, M.J., Scheiderich, K., 2013. Copper speciation and isotopic fractionation in plants: uptake and translocation mechanisms. New Phytol. 199, 367–378. doi:10.1111/nph.12276 Serra, F., Guillou, C.G., Reniero, F., Ballarin, L., Cantagallo, M.I., Wieser, M., Iyer, S.S., Héberger, K., Vanhaecke, F., 2005. Determination of the geographical origin of green coffee by principal component analysis of carbon, nitrogen and boron stable isotope ratios. Rapid Commun. Mass Spectrom. 19, 2111–2115. doi:10.1002/rcm.2034 Shiel, A.E., Weis, D., Orians, K.J., 2010. Evaluation of zinc, cadmium and lead isotope IUDFWLRQDWLRQGXULQJVPHOWLQJDQGUH¿QLQJ6FL7RWDO(QYLURQ±GRLM scitotenv.2010.02.016 Skulan, J., Beard, B., Johnson, C., 2002. Kinetic and equilibrium Fe isotope fractionation between aqueous Fe (III) and hematite. Geochim. Cosmochim. Acta 66, 2995–3015. doi:10.1016/S0016- 7037(03)00266-7 Song, B., Ryu, J., Shin, H.S., Lee, K., 2014. Determination of the source of bioavailable Sr using tracers: a case study of hot pepper and rice. J. Agric. Food Chem. 62, 9232–9238. doi:10.1021/ jf503498r Tabelin, C.B., Hashimoto, A., Igarashi, T., Yoneda, T., 2014. Leaching of boron, arsenic and selenium from sedimentary rocks: II. pH dependence, speciation and mechanisms of release. Sci. Total Environ. 473–474, 244–253. doi:10.1016/j.scitotenv.2013.12.029 Tarricone, K., Wagner, G., Klein, R., 2015. Toward standardization of sample collection and preservation for the quality of results in biomonitoring with trees – A critical review. Ecol. Indic. 57, 341–359. doi:10.1016/j.ecolind.2015.05.012 7RNDOLRۜOXùࡤ.DUWDOùࡤ%LURO*$SSOLFDWLRQRIDWKUHHVWDJHVHTXHQWLDOH[WUDFWLRQSURFHGXUH for the determination of extractable metal contents in highway soils. Turkish J. Chem. 27, 333– 346. 9DQČN$*U\JDU7&KUDVWQê97HMQHFNê9'UDKRWD3.RPiUHN0$VVHVVPHQWRI the BCR sequential extraction procedure for thallium fractionation using synthetic mineral mixtures. J. Hazard. Mater. 176, 913–918. doi:10.1016/j.jhazmat.2009.11.123 Weinstein, C., Moynier, F., Wang, K., Paniello, R., Foriel, J., Catalano, J., Pichat, S., 2011. Isotopic fractionation of Cu in plants. Chem. Geol. 286, 266–271. doi:10.1016/j.chemgeo.2011.05.010 Wiederhold, J.G., 2015. Metal stable isotope signatures as tracers in environmental geochemistry. Environ. Sci. Technol. 49, 2606–2624. doi:10.1021/es504683e

33 Wiederhold, J.G., Kraemer, S.M., Teutsch, N., Borer, P.M., Halliday, A.N., Kretzschmar, R., 2006. Iron isotope fractionation during proton-promoted, ligand-controlled, and reductive dissolution of goethite. Environ. Sci. Technol. 40, 3787–3793. doi:10.1021/es052228y Viers, J., Prokushkin, A.S., Pokrovsky, O.S., Kirdyanov, A. V, Zouiten, C., Chmeleff, J., Meheut, M., Chabaux, F., Oliva, P., Dupré, B., 2015. Zn isotope fractionation in a pristine larch forest on permafrost-dominated soils in Central Siberia. Geochem. Trans. 16, 1–15. doi:10.1186/s12932- 015-0018-0 Wieser, M.E., Iyer, S.S., Krouse, H.R., Cantagallo, M.I., 2001. Variations in the boron isotope composition of Coffea arabica beans. Appl. Geochemistry 16, 317–322. doi:10.1016/S0883- 2927(00)00031-7 Wright, I.J., Westoby, M., 2003. Nutrient concentration, resorption and lifespan: Leaf traits of Australian sclerophyll species. Funct. Ecol. 17, 10–19. doi:10.1046/j.1365-2435.2003.00694.x Wu, L., Beard, B.L., Roden, E.E., Johnson, C.M., 2011. Stable iron isotope fractionation between aqueous Fe(II) and hydrous ferric oxide. Environ. Sci. Technol. 1847–1852. doi:10.1021/ es103171x Xiao, J., Xiao, Y.K., Jin, Z.D., He, M.Y., Liu, C.Q., 2013. Boron isotope variations and its geochemical application in nature. Aust. J. Earth Sci. 60, 431–447. doi:Doi 10.1080/08120099.2013.813585 Åberg, F., Wickman, F.E., 1987. Variations of 87Sr/86Sr in water from streams discharging into the Bothnian Bay, Baltic Sea. Hydrol. Res. 18, 33–42.

34 Electronic Supplementary Information for

Ranges of B, Cd, Cr, Cu, Fe, Pb, Sr, Tl and Zn concentrations and isotope ratios in environmental matrices from an urban area

Nicola Pallavicinia,b, Emma Engströma,b, Douglas C. Baxterb, Björn Öhlandera, Johan Ingria, Scott Hawleyc, Catherine Hirstd, Katarina Rodushkinae and Ilia Rodushkina,b.

aDivision of Geosciences and Environmental Engineering, Luleå University of Technology, S- 971 87 Luleå, Sweden

bALS Laboratory Group, ALS Scandinavia AB, Aurorum 10, S-977 75 Luleå, Sweden

cDepartment of Earth Sciences, Durham University, Durham, UK

dDepartment of Geosciences, Natural History Museum, Stockholm, Sweden

eDepartment of Chemistry, Uppsala University, Uppsala, Sweden

1. Chemicals and reagents

All sample preparation was conducted utilizing high purity reagents: de-ionized Milli-Q water (Millipore, Bedford, MA, USA) further purified by sub-boiled distillation in Teflon stills (Savillex, Minnetonka, Minnesota, USA), Nitric acid (HNO3) and hydrochloric acid (HCl), both from Sigma-Aldrich Chemie GmbH (Munich, Germany), acetic acid (CH3COOH, Merck

KGaA, Darmstadt, Germany), hydroxylamine hydrochloride (NH2OH·HCl, 98.0%, Sigma-

Aldrich Chemie GmbH, Munich, Germany), hydrogen peroxide (H2O2 ≥30%, Sigma-Aldrich

Chemie GmbH, Munich, Germany), ammonium acetate (CH3COONH4, Merck KGaA,

Darmstadt, Germany), aqua regia (HNO3 + 3HCl, both acids from Sigma-Aldrich Chemie GmbH, Munich, Germany) and hydrofluoric acid (HF, 48%, Merck Merck KGaA, Darmstadt, Germany). All laboratory ware coming into contact with samples/sample digests was soaked in 0.7 M HNO3 (>24 h at room temperature) and rinsed with MQ water prior to use.

2. Biological samples

All biological samples i.e. common birch (Betula pubescens) leaves, Norway spruce (Picea abies) needles and fruit bodies of edible mushrooms (Boletus edulis, Leccinum scabrum, Leccinum versipelle, Leccinum aurantiacum and Suillus variegatus) were collected in 2013– 2015 from approximately 50 individual locations. Each sample comprised of 0.5–1.5 g dry weight of leaves (10–50 depending on the growth stage) collected from different branches/trees, needles from the last year grown on parts of lower branches and mushrooms collected under sampled trees. Sampling height was limited to roughly 2.5 m from the ground for all leaf and needle samples. Sampling was performed either at the beginning of the growing season (May early June, birch leaves only) or just before senescence (early September). In the majority of locations samples were taken from different trees in spring and autumn. All samples were collected wearing powder-free laboratory gloves into zip lock plastic bags marked with geographic coordinates, sample type and collection date. Sampling locations were chosen utilizing a 1 km2 sampling grid across a total area of ca. 200 km2. For method verification a certified reference materials (CRMs) were included and processed in parallel to “natural samples” (digestion, separation and analysis): ERM BB186 pig kidney (Institute for Reference Materials and Measurements), TORT-1 lobster hepatopancreas and NASS-4 open ocean water (National Research Council of Canada, Ottawa, Canada), NIST SRM 1547 peach leaves (National Institute of Standards and Technology) and NJV 94-5 wood fuel (Swedish University of Agricultural Sciences, Sweden). None of the materials mentioned above has a certified isotopic composition, but they are do provide a representative natural range of analytes concentrations and isotope compositions (I. Rodushkin et al., 2016).

3. Soil samples preparation and digestion

The topsoil samples were dried at 100oC until they attained a constant weight and then 0.5 g was weighed into 50 mL polypropylene vessels before being dissolved in 20 mL of aqua regia. Vessels were loosely capped and placed in a graphite-topped heating block with cavities matching the vessels’ diameter. Digestion was carried out for 1 h at 110oC under reflux conditions. An aliquot of digest was preserved for element concentrations analysis by double- focusing sector field inductively coupled mass spectrometry (ICP-SFMS) as described in detail by Engstrom et al. (2004). The solid and liquid phases were then careful separated and the rest of the digest was evaporated to dryness. Two mL of H2O2 was then added and evaporated to oxidize remaining organic phases, and the residue was re-dissolved in 2 mL 10 M HCl for separation of individual elements prior to isotope ratio measurements.

4. Biological samples preparation and digestion

Mushrooms were mechanically cleaned, to remove external exogenous material, and divided into approximately 1 cm3 pieces using a ceramic knife on a Teflon plate. Samples were then dried at 50°C, until they attained a constant weight, and homogenized by crushing in plastic bags before being stored in air-tight packed at room temperature.

About 0.5 g of dried material from each sample bag was accurately weighed into a 12 mL

Teflon vial before the addition of 5 mL 14 M HNO3. After the initial oxidation of organic matter subsided, vials were gently agitated and any solid material adhering to the walls was washed down by an additional 1 mL of HNO3. Vials (up to 40 per batch) were placed into a carousel with numbered slots, which was then loaded into the Teflon-coated UltraCLAVE reaction chamber containing a deionized water–H2O2 mixture (10:1 v/v). The chamber was pressurized with compressed argon and a pre-programmed digestion cycle (30 min ramp to 220°C followed by 20 min holding time at that temperature) was initiated. The total processing time, including cooling and subsequent transfer and dilution of sample digests to a final volume of 10 mL into storage polypropylene tubes, was approximately three hours per digestion batch. In some samples, minor quantities of white precipitates of siliceous material were formed. Rapid dissolution of the precipitate was achieved after addition of 30 mL of 16 M HF and manual agitation for a few minutes. Sets of method blanks and CRMs were prepared with each batch of samples. As for any of the wide variety of methods used for the preparation of biological matrices for elemental analysis e.g. ashing, hot-block and microwave digestions, and high pressure ashing (Emma Engström et al., 2004; Rodushkin et al., 2007, 1999), digestion using the UltraCLAVE has its merits and limitations. The former include the complete oxidation of carbonaceous material which ensures undigested organics have a negligible impact on the subsequent separation procedure, the applicability of the method to all matrixes tested in this study, the ease of sample handling/loading (limited material manipulation and thus lowered risk of contamination), and a relatively high throughput. The major limiting factor is the amount of material that can be digested in a 12 mL vessel (approximately 0.5 g dried material). This can require processing parallel digestions for samples low in some analytes, though this approach was not required in the present work.

Aliquots of the digests were diluted 50-fold with 1.4 M HNO3, providing a total digestion factor of approximately 1000 v/m, and analyzed by ICP-SFMS using a combination of internal standardization and external calibration (Emma Engström et al., 2004). Portions of diluted digests remaining after this analysis (approximately 6 mL) were used for B isotope ratio measurements either directly or after additional dilution. The rest of the original sample digest was evaporated to dryness in a 25 mL Teflon beaker at 95°C on a ceramic-top hot-plate, followed by dissolution in 4 mL of 9.6 M HCl, in preparation for subsequent purification.

5. Sequential extraction procedure for soil samples

For the study of operationally defined fractions in soil we used a sequential extraction procedure (SEP). The SEP adopted for the present study is slightly modified version of a scheme developed in the Standards, Measurements and Testing program (SM & T–formerly BCR) of the European Union (Quevauviller et al., 1997, 1993). The original leaching scheme included four solutions (acetic acid, hydroxylamine hydrochloride, hydrogen peroxide and ammonium acetate). This study follows an enhanced scheme previously performed by Tokalioǧlu et al. (2003). The modifications consists of: 1) the addition of a distilled water leach at the beginning of the process; 2) the addition of a HF leach at the end of the process ; and 3) the merging of two originally separate phases (hydrogen peroxide and ammonium acetate). Details of the extraction procedure can be found in Table 2 in the main article.

Adding the distilled water extraction (step F1) before the acetic acid dissolution aims to mimic the percolated/saturated of rain or melted snow through soils. This should mobilize elements which are most liable to plant root uptake without major chemical or physical modification, e.g. by phytosiderophores aiding nutrient availability. Sample tubes containing 3 g - 5 g of soil were weighted against empty tubes and 40 mL of distilled (high purity) Milli-Q water was added. Tubes were then placed on an end-over turntable and left for 16 h. After centrifugation (2000 rev min-1 for 4 min), the water supernatant was carefully transferred to empty containers paying special attention not to transfer any solid material.

The second step (F2) preferentially targets elements bound to carbonates. Forty mL of 0.11 M acetic acid was added to the solid residue in the original sample tubes. The tubes were then placed on the end-over turntable for 16h before centrifugation and decanting of the leachate to a new tube.

The next stage (F3) was performed as the previous one, except with the reducing agent hydroxylamine hydrochloride. At this stage, elements bound to Fe-Mn oxides are leached. Such elements are affected by changes in oxic/anoxic conditions (Zimmerman and Weindorf, 2010) and need a reagent which is adequate for the reduction to the ferrous and manganous forms of the respective oxides (Tessier et al., 1979). Hence the use of hydroxylamine hydrochloride as reducing medium is a suitable choice.

The next phase of the extraction targeted elements bound to organic matter. This has traditionally involved the use of hydrogen peroxide (H2O2) as oxidation medium. Tessier et al. (1979) claimed that even though other more effective methods for the oxidation of organic matter could be employed, H2O2 is the most suitable in terms of specificity as the use of other oxidizers such as combinations of nitric, hydrochloric and perchloric acids could result in the liberation of some of the silicate-bound metals. The oxidation stage (F4) in the present study consisted of the following procedure: 10 mL of H2O2 was carefully added in small aliquots to the soil residue remaining after the F3 step to avoid violent oxidation reactions. The digest was left at room temperature for about 1 h with occasional manual shaking. The solution was subsequently heated at 85°C for 1 h. At this stage, the volume of the liquid phase was reduced to a few millilitres so a second aliquot of 10 mL of hydrogen peroxide was added to the residue and the extraction procedure was repeated. The solution was then heated to near dryness on a hot block and 50 mL of 1 M ammonium acetate solution was added to the residue, followed by 16 h extraction on the end-over turntable, centrifugation and separation of leachate. The final stages (F5 and F6) release elements from the most refractory mineral phases. In brief,

20 mL of aqua regia (15 mL HCl + 5 mL HNO3) was added to the residue after F4 step and left to react overnight at room temperature followed by treatment on a hot block at 110oC for 1 h under reflux conditions. The acid digest was decanted to a new tube, 20 mL of Milli-Q water were added to the residue, the content was thoroughly mixed and the tubes were centrifuged for 15 minutes at 1000 rpm. This Milli-Q solution was combined with the aqua regia digest. Traditionally the SEP schemes would have stopped at this stage. However, a substantial amount of undigested residue remains after the latter treatment due to the insolubility of many primary silicate minerals in any of aforementioned treatments. This is likely to cause incomplete analyte recoveries in silicate rocks. Therefore, 10 mL of HF was added to the washed residue after F5 step and left on the end-over rotator/shaker overnight. The contents were diluted to 40 mL with Milli-Q water, centrifuged and the supernatant decanted.

A set of reference materials and two procedural blanks were prepared in parallel with the soil samples for quality control purposes.

6. Separation

Element concentrations in all fractions were determined by ICP-SFMS using small aliquots of extracts while the rest of each leachate was evaporated to dryness for subsequent matrix separation. Owing to the significantly lower concentrations of many analytes in soil samples from the suburbs the first (F1+F2+F3) and the last (F4+F5+F6) three SEP fractions from each layer were pooled before evaporation in a number of cases. Prior to isotope ratio measurements, the analytes of interest and matrix interfering elements have to be chromatographically separated by ion-exchange purification procedures. The chromatographically procedure used in the present study for the isolation of all elements of interest, except for Cr, has been developed, optimized and tested on real life samples in our previous work (Ilia Rodushkin et al., 2016) and can be summarized as follows.

Sampled are loaded in in 4 mL of 9.6 M HCl onto AG MP-1M (macroporous, 100–200 dry mesh size, 75–150 mm wet bead size, Bio-Rad Laboratories AB, Solna, Sweden) resin- containing columns. Matrix elements are then washed through the column utilizing the same acid. Copper, Fe, Zn, Cd + Tl and Hg are quantitatively eluted from the resin using HCl loads of decreasing molarities followed by a mixture of 6 M HNO3 containing traces of HF. In contrast to matrix separation in geological/industrial materials, there is no risk of overloading the resin capacity with any of the analytes and therefore the entire digest volume can be used. Sample loading in more concentrated HCl allows separation of Cu, Fe and Zn using the same column, while neither Ag nor Pb is efficiently retained by the resin. The sample load and matrix wash fractions (collected into a 25 mL Teflon beaker) contained >99.5% of initial Sr and >85% of initial Pb. After evaporation and redissolution in 4 mL of 7 M HNO3, Sr and Pb were separated using Sr-specific columns (Eichrom Technologies, IL, USA), by selective elution with 0.05 M HNO3 and 0.1 M ethylenediaminetetraacetic acid (EDTA), respectively. The sample load and matrix wash fractions from this column contain >95% of the original Ag which can be purified by loading the eluent in 4 mL of 2 M HCl onto AG MP-1M resin- containing columns and eluting with 14 M HNO3 (Pallavicini et al., 2014). All columns can be re-used several times, although the efficiency of matrix separation gradually deteriorates after 5–6 cycles with the matrix/Cu and Zn/Cd cut-offs affected the most. It should be noted that approximately 0.1% of the initial Sr and 0.2–0.4% of the initial Pb remain on Sr-specific columns which may affect subsequent separations for samples with much lower analyte concentrations or grossly different isotopic compositions. All separated analyte fractions, except those for Sr and Hg, were evaporated to dryness and dissolved in 2–10 mL of 0.3 M HNO3. (An aliquot of 14 M HNO3 was pipetted directly onto the solid residue as a first step, allowed to react for 15–25 min, and then diluted appropriately by addition of MQ water.) 0.1 mL aliquots of the separates were diluted 50-fold with 1.4 M

HNO3 and analyzed by ICP-SFMS (same approach as for sample digests). This provides: (I) information on analyte contents needed to prepare concentration e.g. acid strength-matched solutions for isotope ratio measurements; (II) direct assessment of analyte recovery; (III) control over separation efficiency from matrix elements; and (IV) a test for potentially spectrally interfering elements either from the sample matrix or from handling contamination. Cr purification was performed using Dowex AG 1-X8 Resin (100–200 dry mesh size, 106–180 μm wet bead size, Bio-Rad Laboratories AB). For the details of the Cr separation have are reported in detail by Pontér et al., (2016). Analyte recovery and the efficiency of matrix separation were tested by analysis of load solutions and purified fractions by ICP-SFMS.

The potential of artificially introduced fractionation during separation/evaporation and analysis stages was tested by processing a mixture of ‘δ-zero’ standards with concentrations typical for birch leaves. Acceptance criteria were recoveries above 90%, except for Pb where lower recoveries can be tolerated (Ilia Rodushkin et al., 2016), and concentration ratio(s) of interfering elements to analyte below pre-defined, experimentally determined values ensuring manageable levels of spectral interferences and matrix effects. On the rare occasions when separation failed to meet these criteria, solutions were rejected and no further processing took place. Successfully purified fractions were evaporated to dryness in 40 mL Teflon beakers on a hot plate at 100°C. An aliquot of 14 M HNO3 was pipetted directly onto the residue, left to react for 10-15 min and then diluted to the desired concentration for isotope ratio measurements. AG MP-1M ion-exchange resin was used as supplied.

7. Element concentration and isotopic analyses

Element concentrations were determined by double-focusing sector field inductively coupled mass spectrometry (ICP-SFMS) on an ELEMENT XR, (Thermo Scientific, Bremen, Germany) with instrumental set-up, isotope selection and other relevant information as previously reported (Emma Engström et al., 2004). Isotope ratios in purified fractions were measured on a NEPTUNE PLUS (Thermo Scientific) MC-ICP-MS or ICP-SFMS (the latter being used for B and Pb in samples with low analyte concentrations) using a combination of internal standardization and bracketing standards for instrumental mass bias correction (Table 1 in the main article). All isotopic compositions are reported in standard delta (δ) notation, except for Pb and Sr that are reported as ratios. The following reference materials were used for as ‘δ-zero’ reference standards: NIST SRM 3108 Cd solution Lot 130116, NBS SRM 979 Cr standard, NIST SRM 976 Cu standard solution, NIST SRM 951a boric acid, NIST SRM 981 common Pb, and NIST SRM 987 (NBS-987) Sr carbonate (all from the National Institute of Standards and Technology, Gaithersburg, MD, USA); IRMM 3702 Zn solution and IRMM-014 Fe metal (both from the Institute for Reference Materials and Measurements, Geel, Belgium). For Tl isotopic analyses, commercial standards (1000 mg L-1 mono-elemental solutions supplied by Ultra Scientific, North Kingstown, RI, USA; Lot L00709) were used as ‘δ-zero’ standards. All sample manipulations were performed in clean laboratory areas (Class 10000) by personnel wearing clean room gear and following all general precautions to reduce contaminations (Rodushkin et al., 2010).

All solutions were analyzed in duplicate. Signal intensities were transferred to commercially- available spreadsheet software for further off-line calculations, including blank, isobaric interference(s) as well as instrumental mass bias corrections. The latter was corrected by the revised exponential correction model by Baxter et al. (2006) using the internal standard (for all isotopic systems but B) and the corrected ratios or δ-values were calculated against bracketing standard solutions. Three samples were analyzed between two standards, together forming a block (block: standard 1 – sample 1 – sample 2 – sample 3 – standard 2). The mean value of the two consequent measurements of the sample ratio was calculated against ratios for standards in each block. Assuming a linear change in mass bias, ratios for samples 1 and 3 were calculated relative to those for standards 1 and 3, respectively, while sample 2 was calculated against the mean ratio for both standards. Results from the two measurements were used to calculate mean ratios or δ-values and in-run repeatability for each sample.

References

Engström, E., Stenberg, A., Baxter, D.C., Malinovsky, D., Makinen, I., Ponni, S., Rodushkin, I., 2004. Effects of sample preparation and calibration strategy on accuracy and precision in the multi-elemental analysis of soil by sector-field ICP-MS. J. Anal. At. Spectrom. 19, 858–866. doi:10.1039/b315283a

Engström, E., Stenberg, A., Senioukh, S., Edelbro, R., Baxter, D.C., Rodushkin, I., 2004. Multi-elemental characterization of soft biological tissues by inductively coupled plasma– sector field mass spectrometry. Anal. Chim. Acta 521, 123–135. doi:10.1016/j.aca.2004.06.030

Pontér, S., Pallavicini, N., Engström, E., Baxter, D.C., Rodushkin, I., 2016. Chromium isotope ratio measurements in environmental matrices by MC-ICP-MS. J. Anal. At. Spectrom. 31, 1464-1471. doi:10.1039/C6JA00145A

Rodushkin, I., Bergman, T., Douglas, G., Engström, E., Sörlin, D., Baxter, D.C., 2007. Authentication of Kalix (N.E. Sweden) vendace caviar using inductively coupled plasma- based analytical techniques: evaluation of different approaches. Anal. Chim. Acta 583, 310–318. doi:10.1016/j.aca.2006.10.038

Rodushkin, I., Engström, E., Baxter, D.C., 2010. Sources of contamination and remedial strategies in the multi-elemental trace analysis laboratory. Anal. Bioanal. Chem. 396, 365–377. doi:10.1007/s00216-009-3087-z

Rodushkin, I., Ruth, T., Huhtasaari, Å., 1999. Comparison of two digestion methods for elemental determinations in plant material by ICP techniques. Anal. Chim. Acta 378, 191– 200. doi:10.1016/S0003-2670(98)00635-7

Tessier, A., Campbell, P.G.C., Bisson, M., 1979. Sequential extraction procedure for the speciation of particulate trace metals. Anal. Chem. 51, 844–851. doi:10.1021/ac50043a017

Tokalioǧlu, Ş.̧, Kartal, Ş.̧, Birol, G., 2003. Application of a three-stage sequential extraction procedure for the determination of extractable metal contents in highway soils. Turkish J. Chem. 27, 333–346.

Zimmerman, A.J., Weindorf, D.C., 2010. Heavy metal and trace metal analysis in soil by sequential extraction: a review of procedures. Int. J. Anal. Chem. 2010, 1–7. doi:10.1155/2010/387803