Chemical Geology 357 (2013) 203–214

Contents lists available at ScienceDirect

Chemical Geology

j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo

Teflon-HPLC: A novel chromatographic system for application to isotopegeochemistry and other industries

T.J. Ireland a,⁎, F.L.H. Tissot a, R. Yokochi a,b, N. Dauphas a

a Origins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, Chicago, IL 60637, United Statesb Department of the Earth and Environmental Sciences, University of Illinois at Chicago, Chicago, IL 60607, United States

⁎ Corresponding author at: Department of Earth and E685 Commonwealth Avenue, Room 130, Boston, MA 022422 6469.

E-mail address: tireland@bu.edu (T.J. Ireland).

0009-2541/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.chemgeo.2013.08.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 February 2013Received in revised form 1 August 2013Accepted 4 August 2013Available online 16 August 2013

Editor: L. Reisberg

Keywords:ChromatographyHPLCHDEHPRare-earth elementsNi isotopesIsotope geochemistry

Traditional column chromatography techniques are restricted by several factors, including limitations on columnlength, resin grain size, elution time, and a general inability to slowly ramp up or down reagent concentrationsalong a gradient. Likewise, most existing high-performance liquid chromatography (HPLC) systems are notamenable to certain column chromatography techniques that require highly concentrated acids.Here, we outline the development of a Teflon HPLC (T-HPLC) system for application to a wide variety of chroma-tography problems. The primary factors that set the T-HPLC system apart from any currently available chroma-tography procedure are the following: 1) a fluid flow path enclosed entirely by Teflon, 2) fully automatedelution schemes controlled by computer software, which allows for fresh mixing of reagents for each elutionstep, and fine scale gradient/ramp elutions, 3) temperature control of the entire system (up to 80 °C) forenhanced chemical separations and, 4) a modular design making the system easily adaptable to a variety ofseparation schemes.The effectiveness of the T-HPLC system is tested on two column techniques that are of particular interest to thegeochemistry/cosmochemistry communities. The first application involves the separation ofNi fromMg,which isrequired for high precision Ni isotopic studies and for investigating the abundance of the extinct 60Feradionuclide. The T-HPLC system greatly simplifies and improves upon the classical technique. In a single passon an 80 cm long column, we achieve excellent separation of Ni from Mg, with a much improved time frame(10 h versus 70 h). The second application is the separation of the individual rare earth elements from eachother. The isotopic compositions of the multi-isotopic REEs (La, Ce, Nd, Sm, Eu, Gd, Dy, Er, Yb and Lu) may holdimportant information about nucleosynthetic processes, cosmic-ray exposure effects in meteorites and airlessbodies, and mass fractionation effects. For this application, we also developed a computer simulation that usesexperimentally determined partition coefficients to simulate an elution curve in order to optimize the actualcolumn elution scheme. Overall, we were able to achieve excellent separation of the multi-isotopic REEs fromeach other.Although the two applications that we explore are in the fields of geochemistry/cosmochemistry, the modulardesign and adaptability make the T-HPLC system useful to a wide array of scientific fields and industries.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Column chromatography techniques have proven to be an invalu-able tool in the geosciences, as well as in other scientific fields and in-dustries, and are commonly employed in a wide variety of geologicalapplications. In particular, open-system, gravity-driven column chro-matography has been established as an essential step in many isotopegeochemistry methods, which require high purity separations of the el-ements of interest from matrix elements. These well-established tradi-tional column techniques are undoubtedly useful, but their application

nvironment, Boston University,15, United States. Tel.: +1 301

ghts reserved.

to challenging and new problems in isotope geochemistry can be limit-ed and some fine-scale separations still require complicated multi-stepand time-consuming protocols. The chief limitations of traditional col-umns pertain to the overall length of column, resin size and lackof temperature control, which can severely compromise separationefficiencies.

In contrast to traditional column chromatography techniques, high-performance liquid chromatography (HPLC) systems have evolved con-siderably. HPLC systems offer several advantages over traditionalcolumns, which include: 1) a closed-system setup, 2) the ability to pres-surize the system, which allows for longer columns and finer resin,resulting in better separation, and 3) automation of the elutionsequence. There are several HPLC systems that have been used forbulk geochemical analyses such as the Dionex DX-500, which can beused for measuring transition metals and REE concentrations in rock

204 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

and seawater samples (Dionex, 1992; Tachibana and Huss, 2003). How-ever, despite these advantages, HPLC systems have not been widely ap-plied to radiogenic and non-traditional stable isotope geochemistry. Asstated above, part of the reason for this trend is because of well-established traditional methods for many elements of interest — therehas been no impetus to change. Another reason is thatmost commercial-ly available HPLC systems suffer from significant limitations as well.These drawbacks are typically associated with the materials used in theconstruction of these systems. For example, many HPLC systems aredesigned to work with dilute acids and organic solvents, and may con-tain parts in the reagent flow path that can be corroded, dissolved orleached with the use of concentrated acids or certain organic solventsthat are commonly used for separating elements of geochemical rele-vance. PEEK (polyether ether ketone) is often used in HPLC systems(including theDionexDX-500mentioned above) to avoid contaminationwithmetals, but even at 20 °C, this material is not recommended for usewith hydrobromic, hydrofluoric, nitric, and sulphuric acids (Pritchard,1995). Another related issue is that the electronic controls and housingare often spatially associatedwith theHPLC unit, sometallic parts rapidlycorrode upon exposure to harsh chemicals and vapors, thereby limitingthe lifespan of the apparatus. For example, Sivaraman et al. (2002)changed their chemical approach for a HPLC system due to concernsover the potential corrosion of components in their system by highmolarity acids. For these reasons, application of HPLC systems to radio-genic and non-traditional stable isotope geochemistry is rare. Given theadvantages of HPLC systems, it is unfortunate that these systems havenot yet been widely adapted to the world of isotope geochemistry.

Here, we bring the benefits of HPLC to isotope geochemistry, byoutlining the development of a Teflon-HPLC (T-HPLC) system that wehave constructed at the University of Chicago. This development wasmade possible by the availability of Teflon-made components used inthe semi-conductor industry. Teflon has not been traditionally used inHPLC because of high costs, difficulties in machining to tight tolerancesand concerns about porosity. However, in isotope geochemistry wherelow analytical blanks and system integrity are critical, the advantagesof Teflon outweigh its drawbacks. Another aspect of our effort hasbeen to isolate the electronic controls from exposure to corrosive fumes.

The T-HPLC system has several distinctive features that improveupon existing technologies and allow it to be easily modified for appli-cations to a wide variety of geochemical and cosmochemical problems,which include: 1) a fluid flow path that is entirely enclosed in Teflon, 2)fully automated elution schemes controlled through LabView software,which allows for freshmixing of reagents for each elution step, and finescale gradient/ramp elutions. This automation should also improvereproducibility, 3) temperature control of the entire system (up to80 °C) for enhanced chemical separations, and 4) a modular designmaking the system easily adaptable to a variety of separation schemes.These features of the system can facilitate high element yields throughthe column and efficient separation of the analyte(s) of interest. Follow-ing a detailed description of our T-HPLC system,wewill discuss a coupleof focused applications that are of particular interest in isotope geo-chemistry, cosmochemistry, and nuclear chemistry.

The first application involves the separation of Ni from Mg, which isimportant for studying the abundance of extinct radionuclide 60Fe(60Fe → 60Ni; t1/2 = 2.62 Myr; Rugel et al., 2009) in the early solar sys-tem. Tang and Dauphas (2012) demonstrated that up to five repetitionsof a 14-hour long column chemistry procedurewere needed to adequate-ly separate Ni from Mg, with chemical reagents having to be constantlymixed due to acetone evaporation. Our goal is to improve upon the effi-ciency of this elution procedure. Our secondmajor focus is the separationof individual rare-earth elements (REEs) from each other. The REEs com-prise a suite of geochemically and cosmochemically important elementsknown for similar partitioning characteristics; hence, these elementshave proven to be very challenging to separate from one another (Nashand Jensen, 2001). The isotopic composition of the multi-isotopic REEsmay hold important information about nucleosynthetic processes

(McCulloch and Wasserburg, 1978a; McCulloch and Wasserburg,1978b; Andreasen and Sharma, 2006, 2007; Carlson et al., 2007;Gannoun et al., 2011; Qin et al., 2011), neutron capture and cosmic-rayexposure effects on meteorites (Eugster et al., 1970a; Eugster et al.,1970b; Burnett et al., 1971; Lugmair and Marti, 1971; Russ et al., 1971;Bogard et al., 1995; Hidaka et al., 2000a; Hidaka et al., 2000b; Hidakaand Yoneda, 2007; Hidaka et al., 2009) and mass fractionation effects(Albalat et al., 2012). Several previous studies have reported moderatesuccess in separating the individual REE by HPLC (Sisson et al., 1972;Yoshida and Haraguchi, 1984; Cassidy and Chauvel, 1989; Meisel et al.,2001; Sivaraman et al., 2002; Haley and Klinkhammer, 2003; Vermaand Santoyo, 2007), but these analytical developments have not yettransitioned into isotope geochemistry due to the lack of suitable HPLCsystems that meet the stringent standards of this field.

Although our focus is on radiogenic and non-traditional stableisotope geochemistry, we want to emphasize that this novel T-HPLCsystem is easily adaptable to many chromatographic problems andcan be extremely useful in a vast array of industries.

2. Description of the T-HPLC system

2.1. Brief background on column chromatography

In general, column chromatography techniques take advantage ofthe differing affinities of certain elements on specific resins, in order toisolate the elements of interest and remove any potential interferingor undesired elements. The partitioning behavior of an element on aresin is related to the resin properties, as well as to the liquid phaseadded to the column. This behavior can be changed by varying the prop-erties of this liquid phase (e.g., changing reagent molarity or reagenttype). To achieve quantitative separation of an element from other ma-trix elements, careful determination of elemental partitioning on theresin being used is essential. Several factors can impact the efficiencyof element separation on a column. These factors include the length ofcolumn, size of resin beads (thus, the porosity of the column and theheight of a theoretical plate), interconnectivity of resin, temperature,and the type/concentration of liquid used during elution. Gravity-driven column chromatography techniques, for example, are primarilyrestricted by the mesh-size of the resin, as well as the overall length ofthe column, which can have a prohibitive effect on elution time. Smallerresin mesh-sizes and longer column lengths will both cause an increasein elution time, or potentially halt elution. An increased elution timemay be accompanied by diffusion within the column, which is a detri-mental problem that will ultimately offset any gain obtained by usinglonger columns. Column systems that use a pumpor pressurizationme-dium can mitigate most of these diffusion effects. Pump or pressure-driven columns also allow one to use a finer resin mesh size, as wellas narrower or longer columns, which may increase column efficiencyby increasing the reactive surface area. In addition, elution flow ratescan be more precisely controlled by varying pump speeds or pressures.

Further, traditional column setups are generally run at room tem-perature; thus, open systems are not amenable to temperature controlbecause it is difficult to ensure that all parts of the column are at thesame temperature and that reagents are not lost by evaporation. How-ever, increased temperaturesmay have a significant effect on separationfactors and the sharpness of an elution peak (Cerrai and Testa, 1963;Siekerski and Sochacka, 1964; Horwitz et al., 2006; Pourmand andDauphas, 2010). Another impediment of traditional column setups isthe difficulty involved to achieve small-step gradient ramps in eluentconcentration as the eluents need to be pre-mixed prior to columnchemistry. Due to limitations on the amount of different eluents beingused, generally only coarse steps are utilized. Lastly, some reactantsused in elutions can be unstable and may break down over time. Forexample, H2O2 or acetone is occasionally used in certain column chem-istries (Strelow, 1988;Munker et al., 2001). Over a relatively short time,H2O2 will break down to H2O, or acetonewill evaporate. The challenges

205T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

outlined above can be addressed effectively by our proposed T-HPLCsystem.

A distinct advantage of column chromatography systems, especiallywith respect to precious samples, is that numerous elements can be sep-arated from a single sample digestion through the meticulous selectionof resins and a thorough collection of all material that passes through acolumn. For example, Pourmand and Dauphas (2010) outlined a meth-od for the separation of Ca, U, Hf and Lu from meteoritic material thatused two different types of resin (TODGA and Ln-resins) and severalelution steps to separate and collect the desired elements. In this man-ner, the amount of sample digested is minimized and a large amountof chemical and isotopic data may be acquired.

2.2. Description of the T-HPLC system

HPLC systems have the power to overcome several of the limitationsposed by traditional column setups. However, as stated previously,current commercially available systems are unsuitable for the harshchemicals and reagents required for some geochemical separations.The answer to this problem is to create a closed HPLC system with aflow path entirelymade of Teflon (thematerial of choice in isotope geo-chemistry for the last 40 years),where the electronic controls are isolat-ed from the corrosive chemicals and fumes used in the purification unit.Due to its superior material properties (i.e., chemically inert and resis-tant), Teflon is the most suitable substance for the wide variety ofreagents that may be used. In addition to this innovative flow path, itis desirable to have an HPLC system capable of separating a largerange of chemical elements, adept at easily switching from one applica-tion to another quickly and inexpensively, and able to provide a high

Fig. 1. i) Schematic diagramof the T-HPLC system showing the important design elements involkey sections of the instrument. A) Six reagent reservoirs that can store any combination of reagethat allow for water to circulate and maintain the temperature of the reagents. B) A custom-dcolumn.An insulating jacket surrounds themixing chamber to preserve a constant temperature.length and/or diameter can be easily adapted to any elemental systemweare interested in. D) Twelution cuts to different collection vessels. E) A water circulator pumps water up through the s80 °C. Thewater flow path proceeds from thewater jacket surrounding the column, to the insulinside the reagent reservoirs, before returning to the circulator.

degree of separation of elements in a single pass through the column.The key aspects to our T-HPLC are as follows:

1) A flow path entirely made of Teflon (PTFE/PFA);2) six reagent reservoirs;3) a custom-designed mixing chamber for reagent mixing;4) a modular and adjustable chromatographic column setup (both in

diameter and length);5) control over the thermal state of the entire system (up to 80 °C); and6) complete automation using LabView software and a code we

designed.

Theflowpath beginswith the reagent reservoirs (labeledA in Fig. 1),which can hold up to 300 mL of each reagent. There are six such reser-voirs that can be used for any combination of water, bases, acids (e.g.,HNO3, HCl, HF, H2SO4, oxalic acid), organic solvents (e.g., acetone) orany other reagent required for a particular column chemistry. Insideeach of these reservoirs is a coiled Teflon tube that allows for water toflow through the reservoirs to maintain the reagents at a set tempera-ture. These reservoirs are connected to six computer-controlled self-priming solenoid pumps (Cole-Parmer Part 73120-40), which haveTeflon-wetted interiors. These pumps deliver reagents into a custom-designed mixing chamber (B in Fig. 1) at a rate of 40 μL per stroke.Since the volumes of acid, water and organic solvents introduced tothemixing chamber are carefully regulated, we can very accurately con-trol the eluentmolarity (as demonstrated in Fig. 5). Themain advantageof this setup is the ability to vary acid molarities discretely on a smoothramp in a procedure known as gradient elution in HPLC, rather thanjumping from one acid molarity to another. A gradient elution typicallyprovides better separation of elements.

ved in the system. ii) Photograph showing the overall setup of the T-HPLC system, denotingnts required for a particular column chemistry. There are Teflon coils inside these reservoirsesigned mixing chamber ensures that reagents are well-mixed prior to passing onto theC) The column itself consists of a rigid length of Teflon tubing in awater jacket. The columnomanifold solenoid valves are present at thebase of the column,which can direct discreteystem to maintain the system temperature. The water can be heated to a temperature ofating jacket around themixing chamber, before circulating through the Teflon coils hosted

206 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

The custom-designed mixing chamber has a couple of key dis-tinguishing features. First, a computer-monitored Teflon level sensor,close to the base of the chamber, monitors the liquid level. The volumebelow this sensor marks the limit of the smallest liquid pulse out of thechamber and is on the order of 150–200 μL. Second, a Teflon magneticstirring bar is placed just above the level sensor to ensure that reagentsinside the chamber are thoroughly mixed and equilibrated prior to intro-duction on the column. The motor for the magnetic stirrer sits in a cavityon the side of the mixing chamber, but is not in contact with any liquiddirectly. Once the pumps finish filling the mixing chamber reservoir andthe reagents are thoroughly mixed, the chamber is pressurized via dryN2 gas and a three-way solenoid valve (Cole-Parmer Part 01540-11)just below the chamber opens, allowing the liquid to enter the column.

The advantages of using gas to push the liquid are that it is extremelyclean, offers constant pressure with time, is robust against corrosion,and minimizes the dead volume between the mixing chamber and col-umn. A drawback is that the gas dissolved in the mixing chamber canpotentially form bubbles in the column if there is a drop in pressure oran increase in temperature in the column relative to the mixing cham-ber. The issue of bubble formation is minimized by maintaining themixing chamber at the same temperature as the column. In the elutionsthat we performed, we did not find any noticeable bubble formationthat could hamper the flow. Another restriction in the current systemis that the maximum column pressure that can be reached is currentlylimited to the tolerance of the mixing chamber (~80 psi N2 in oursystem).

Once the liquid level passes the level sensor, the valve between themixing chamber and column closes, the N2 gas is vented and a newstage of pumping begins. Themixing chamber is designed to hold pres-sures up to ~80 psi, with the lid fitting snugly against a Teflon O-ring.The system is further sealed through an insulating jacket, which closelyfits around the mixing chamber. This insulating jacket also has a path-way for water flow to keep the mixing chamber at the same tempera-ture as the rest of the system.

The three-way solenoid valve at the base of the mixing chamberserves a couple of important functions. The first, as outlined above, isto provide a barrier from the mixing chamber to the column during re-agent mixing, which can then be opened for the injection of the mobilephase to the column. The second function, and use of the third port onthe valve, is for sample introduction through a Luer lock connectionand syringe. In this manner, the sample can be introduced directly tothe column without any concern about contaminating the mixingchamber.

The columnportion of the T-HPLC system represents another innova-tion (C in Fig. 1). Our design includes the capability of varying the columnlength, depending on the demands of the elemental system being stud-ied. In essence, the column in our system is a piece of Teflon tubingcontained in a water jacket. Thus, by adding or removing pieces fromthewater jacket, and changing the length of Teflon tubing, we can easilyadapt the column length and the column diameter to any elemental sys-temof interest. The column isfilledwith thedesired resin prior to attach-ment to the system, utilizing a vacuum box to ensure that the resin istightly packed. Ultimately, after passing through the column, the elutedvolumes are collected in Teflon beakers distributed through computer-controlledmanifold solenoid valves (D in Fig. 1). The use of two solenoidmanifolds, with 6 outlets each, allow us to independently collect a largenumber of elution cuts without the need of tending to changing vials.

A further enhancement is that the whole system can be thermallycontrolled. Using a circulating water heater (E in Fig. 1), we havedesigned an independent, closed loop water system that is in contactwith all of the components of the HPLC system and can support temper-atures of up to 80 °C. The circulation path goes from the circulator,through the water jacket surrounding the column, up to the insulatingjacket that surrounds themixing chamber (B in Fig. 1), to the six Tefloncoils that are isolated inside of the reagent reservoirs (A in Fig. 1), beforethe water is re-circulated back to the water heater.

Lastly, the proposed system is controlled via an external computer,running LabView software from National Instruments. LabView uses agraphical programming interface that enables the control of electriccomponents, as well as the ability to program additional commands.Through this computer systemwe are able to specify an elution scheme(i.e., the mixing parameters, including the types of reagents, volumesandmolarities desired), fromwhich the program calculates the amountof liquid needed fromeach liquid reservoir for each step. From there, thecomputer program controls the pumping of the solenoid pumps, themixing of reagents, the monitoring of the liquid level in the mixingchamber, the opening/closing of the valve that leads from the mixingchamber to the column, the N2 pressurization of the mixing chamber,and the distribution of eluted volumes from the solenoid manifoldsat the end. In this manner, a complex elution can be completelyautomated.

2.3. Future improvements

Weare currently exploring the option of upgrading to pneumaticallyactuated valves and pumps, whereby these components are operatedvia a compressed gas pressure instead of through electronics. Thesetypes of valves and pumps tend to have a longer lifetime (i.e., moreactuations) and can bemade almost entirely out of Teflon (the solenoidvalves used here have an important metal component). In this manner,the electronics, which would control the gas flows to the pneumaticpumps and valves, can be even further isolated from the HPLC system.An additional improvement involves housing all the electronic compo-nents inside an enclosure with positive N2 pressure, which is designedto isolate the electronics from potentially corrosive fumes.

An important limitation of the existing T-HPLC system is that onlyone separation can be performed at a time. With traditional columns,several samples can be separated concurrently from multiple columns.Thus, the gain in efficiency in a single separation can be offset by thelack of throughput. For some complex separations (i.e., the REE asdiscussed in Section 4), this is a fair tradeoff. To address this issue, weare considering the potential of hooking up multiple columns to thesystem in the future.

3. Applications of T-HPLC to Ni–Mg separation

3.1. Brief background on Ni isotope measurements for cosmochemicalapplications

The study of Ni isotopic anomalies and Ni stable isotope variationsby TIMS (thermo-ionization mass spectrometry) and MC-ICP-MS(multi-collector inductively coupled plasma mass spectrometry) hasbeen a very active area of research in cosmochemistry and isotope geo-chemistry (Birck and Lugmair, 1988; Shukolyukov and Lugmair, 1993a;Shukolyukov and Lugmair, 1993b; Quitté et al., 2007; Dauphas et al.,2008; Regelous et al., 2008; Quitté et al., 2010, 2011; Steele et al.,2011; Tang and Dauphas, 2012). In particular, discussions about theprevalence of 60Fe, an extinct radionuclide that decays to 60Ni (t1/2 =2.62 Myr; Rugel et al., 2009), at the time of solar system formationhave been quite intense (Shukolyukov and Lugmair, 1993a; Tachibanaand Huss, 2003; Mostefaoui et al., 2004; Guan et al., 2007; Marhas andMishra, 2012; Tang and Dauphas, 2012). This debate is centered aroundsubstantial differences (almost 2 orders of magnitude) in the estimateof the 60Fe / 56Fe ratio at the time of solar system formation, a parame-ter that has significant bearing on the astrophysical context of solar sys-tem birth. To resolve this problem, Tang and Dauphas (2012)measuredthe Ni isotopic composition of several meteorite samples, includingangrites, HEDs and unequilibrated ordinary chondrites. Tang andDauphas (2012) developed a 3-step column chromatography processfor the purification of Ni from matrix elements in these samples — inthe next section we outline the second step of this procedure, whichrepresents a significant time impediment to the overall process.

207T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

3.2. Difficulties with current methodology

A major difficulty faced by analysts in order to generate precise andaccurate measurements of Ni isotopes for application to 60Fe–60Ni ex-tinct radionuclide systematics is to separate Ni from Mg in silicaterocks (where the Mg/Ni ratio is high). One reason that a high degreeof separation between Ni and Mg is required is due to the presence ofan argide interference produced on 24Mg (24Mg40Ar), which can inter-fere with the low abundance isotope 64Ni. In addition, measurementprecision on the other Ni isotopes (58Ni, 60Ni, 61Ni and 62Ni) decreaseswith increasing matrix concentrations. To ensure a negligible matrix ef-fect, a ratio of Cmatrix/CNi b 0.5 is required.

The separation of Ni from Mg is achieved by using a 40 cm long(0.3 cm ID) Teflon column filled with Bio-Rad AG50-X12 200–400 mesh hydrogen-form cation exchange resin, which is attached toa vacuum box to facilitate the elution. The elution scheme is as follows(adapted from Strelow et al., 1971):

1) Condition columnwith 10 mLof 20%10 MHCl–80% acetonemixture.2) Load sample in 5 mL of 20% 10 M HCl–80% acetone mixture.3) Rinse column with 30 mL of 20% 10 M HCl–80% acetone mixture.4) Collect Ni with 150 mL of 20% 10 M HCl–80% acetone mixture.5) Repeat procedure 5 times to ensure adequate separation of Ni (10 to

100 ppm level) from Mg (% level).

This separation scheme was used by Tang and Dauphas (2012) ashigh-precision measurement of Ni isotopes was crucial for their work,making it essential to remove all matrix elements from the solution.This procedure represents a significant time inefficiency in the Ni sepa-ration process. Each elution lasts about 14 h— so thewhole second stepof the 3-step Ni purification technique takes a total of about a week'sworth of time. In addition to the time inefficiency, the HCl–acetonemixture becomes unstable over the course of an elution and acetonewill evaporate, so it needs to be freshly mixed every half hour.

3.3. Application of T-HPLC system to the separation of Ni from Mg

Two experiments were conducted to ascertain howwell our T-HPLCsystem functions in terms of separating Ni from Mg and the results areshown in Fig. 2. Both experiments had the same loading solution, whichcontained 1 μg/g of Fe, Ni andMg respectively, in a solution of 20% 10 MHCl and 80% acetone. The rinsing and collection solutions paralleledthose of Tang and Dauphas (2012) consisting of 20% 10 M HCl and80% acetone. Elution steps were collected in 4 mL intervals, with eachstep consisting of a fresh mixing of reagents in the mixing chamber.Following collection, each step was evaporated to near-dryness and

Fig. 2. Elution curves generated from two experiments involving the separation of Ni from Mgshows that there is significant overlap between the elution curves for Ni and Mg. B) A 80 cmthe elution peaks sharpen, and that there is excellent separation of Ni from Mg.

picked up in 0.46 M nitric acid and analyzed on a Neptune MC-ICP-MSat the Origins Lab of the University of Chicago.

The first experiment was conducted at room temperature on a40 cm long column (diameter of 0.3 cm) pressurized to produce aflow rate of about 1 mL/min for liquid passing through the column. Intotal, 260 mL of solution were collected in a time span of ~8 h. Thesecond experiment was conducted at an elevated temperature of65 °C on a longer, 80 cm column (diameter of 0.3 cm) pressurized toa flow rate of about 1 mL/min. For experiment 2, a total of 500 mL ofsolution were collected, with a total elapsed time of ~10 h.

The results of these experiments are shown in Fig. 2 (A and B). Forthefirst experiment, thefinal elution shows significant overlap betweenthe elution peaks of Ni and Mg from about 130 to 200 mL of solutioncollected. This curve is very similar to that observed by Tang andDauphas (2012) for the column parameters described in Section 3.2.This result is not surprising, given that the experiment was conductedat essentially the same conditions as the typical setup.

The elution curve for the second experiment shows an immense im-provement in the separation of Ni and Mg. The elution peaks for theseelements are much sharper and the separation between Ni and Mg isvery clean. Formost rockmatrices, one pass through the T-HPLC systemwould provide sufficient separation of these two elements. This is an ob-vious saving on time and it demonstrates the superiority of our T-HPLCsystem relative to conventional column chromatography.

4. Applications of T-HPLC to the separation of the REE

4.1. Brief background on REE separation for cosmochemical applications

As stated in the introduction, the REEs comprise a suite of geochem-ically and cosmochemically important elements known to have similarpartitioning characteristics. Thus, the REEs are of particular interest togeochemists and cosmochemists, but are also challenging to separatefrom each other at the level required for isotope geochemistry work.Severalwell-establishedmethods have been employed for the bulk sep-aration of REEs for geochemical applications by traditional columnchemistry (e.g., Gast et al., 1970; Strelow and Jackson, 1974; Hookeret al., 1975; Walsh et al., 1981; Crock et al., 1984; Henderson andPankhurst, 1984; Balaram, 1996; Baker et al., 2002; Pourmand et al.,2012) or by HPLC (see review by Verma and Santoyo, 2007). Thesemethods are effective for obtaining bulk concentration data andchondrite-normalized REE patterns, which are often used to provideinsight into a sample's history. However, these bulk methods are notsuitable for high-precision isotopic analyses of the individual REEsbecause of isobaric, oxide and molecular interferences that may bepresent during ICP-MS analysis.

on the T-HPLC system. A) A 40 cm long column (diameter 0.3 cm) at room temperaturelong column (diameter of 0.3 cm) at an elevated temperature (65 °C) demonstrates that

208 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

Over the past several years, it has become increasingly important ingeochemistry/cosmochemistry to address various questions involvingREE isotope systematics (e.g., Caro et al., 2003; Boyet and Carlson,2005; Andreasen and Sharma, 2006; Andreasen and Sharma, 2007;Bennett et al., 2007; Carlson et al., 2007; O'Neil et al., 2008; Thraneet al., 2010; Gannoun et al., 2011; Albalat et al., 2012). For instance, inthe geologic community, a lot of attention has been centered on separat-ing Nd and Sm from the rest of the REEs because of two commonlyused radiogenic isotope systems (147Sm → 143Nd t1/2 = 106 Gyr;146Sm → 142Nd t1/2 = 68 Myr) and chromatographic techniques forthese two elements are fairly well established (Lugmair et al., 1975;DePaolo and Wasserburg, 1976; Wasserburg et al., 1981; Jacobsen andWasserburg, 1984; Rehkämper et al., 1996; Boyet and Carlson, 2005;Caro et al., 2006; Debaille et al., 2007; Caro et al., 2008).

However, the ability to measure the isotopic compositions of all theREEs froma single sample digestion and columnprocedure could be a sig-nificant leap forward in terms of how we analyze these elements, withimplications for nucleosynthetic processes (McCulloch and Wasserburg,1978a; McCulloch and Wasserburg, 1978b; Andreasen and Sharma,2006, 2007; Carlson et al., 2007; Gannoun et al., 2011; Qin et al., 2011),neutron capture and cosmic-ray exposure effects on meteorites (Eugsteret al., 1970a; Eugster et al., 1970b; Burnett et al., 1971; Lugmair andMarti, 1971; Russ et al., 1971; Bogard et al., 1995; Hidaka et al., 2000a;Hidaka et al., 2000b; Hidaka and Yoneda, 2007; Hidaka et al., 2009) andmass fractionation effects (Albalat et al., 2012).

Methods of separating the individual REEs using α-HIBA (alpha-hydroxyisobutyric acid; e.g., Hidaka et al., 1995; Rehkämper et al.,1996; Meisel et al., 2001; Carlson et al., 2007) or Ln-resin (which usesHDEHP as the extractant phase; e.g., Pin and Zaldegui, 1997; Horwitzet al., 2006; Bouvier et al., 2008; Hidaka et al., 2009) have been refinedover the years, but these studies have mainly focused on a small subsetof the REEs or have not provided enough resolution between REE toallow for high-precision isotopic analysis. Although Ln-resin is a power-ful method for separating REEs, α-HIBA has been favored in HPLC sys-tems because the HDEHP chemistry requires the use of high molarityand corrosive acids that cannot be handled by commercial HPLC sys-tems. This is most clearly expressed by Sivaraman et al. (2002) whoconsidered the feasibility of using a HDEHP coated column in a HPLCsystem, but decided against this approach because of concern that thehigh molarity acids would damage the system.

A thorough study of REE separation using Ln-resin and the T-HPLCsystem was conducted to assess the effectiveness of this novel chro-matographic technique. This study involved the determination of parti-tion coefficients of the REEs on Ln-resin using both HCl and HNO3 atvarying acid molarities, a computer simulation of the elution curve(using the appropriate resin and column properties) in order to opti-mize the gradient elution scheme that was utilized, and the actual REEelution itself. Through this study, it is shown that the designed T-HPLCsystem is effective at addressing a very complex elution scheme andhas a remarkable potential to improve or facilitate other chromato-graphic problems.

4.2. Determination of REE partition coefficients on Ln-resin

Several previous studies have used Ln-resin for partial REE separa-tion (e.g., LREE: Pin and Zaldegui, 1997; Sm and Gd: Hidaka et al.,2009). On resins using HDEHP as an extractant phase, the affinity ofthe REE with HDEHP increases with atomic mass. Thus, the lighterREE, which have a lower affinity, can be eluted with weaker acids andthe heavier REE with stronger acids (Pierce et al., 1963; Horwitz andBloomquist, 1975). This elution sequence, from light to heavy REE, is op-posite to that of α-HIBA (Rehkämper et al., 1996). To optimize the elu-tion sequence, we ran elution simulations using plate theory (Martinand Synge, 1941), which requires the partition coefficient as an inputparameter.

The partition coefficients (Kd; i.e., the affinity of an element for a par-ticular resin) of the REE on Ln-resin, as a function of HCl and HNO3

molarity, were determined by a series of batch equilibration experi-ments following the procedure outlined by Pourmand and Dauphas(2010) conducted at room temperature. In brief, a series of 10 μg/gbulk REE standards were generated from commercially availablemono-elemental solutions, which were evaporated to dryness and sub-sequently re-dissolved in 5 mL of various HCl or HNO3molarities (rang-ing from 0.01 to 5 M) in clean 15 mL PFA beakers. For each respectivemolarity step, approximately 300 mg of dry Ln-resin was collectedfrom 2 mL pre-packed cartridges (50–100 μm bead size), which areavailable from Eichrom. The resin was added to the PFA beakers, andthe mixture was agitated by placing on a Vortex shaker for 5–10 minevery 2 h. After 10 h of equilibration, the resulting mixtures werepassed through 2 mL Bio-Rad columns fitted with frits, to separate theacid phase from the resin. To prepare for mass spectrometry, the acidphases were dried down and re-dissolved in 0.46 M HNO3.

The concentrations of the REE in each solution were determined bysample-standard bracketing on a Thermo Neptune MC-ICP-MS at theOrigins Lab of the University of Chicago, following the proceduresof Pourmand et al. (2012). An Apex-Q + Spiro desolvating nebulizerwas used for sample introduction to the ICP-MS, which helps to reduceoxide production in the plasma. Oxide interferences are a potentialproblem with bulk REE analyses, as oxide production on the light REEcan interfere with the heavier REE.

The Kd (concentration in mol/g of element in solid divided by theconcentration in mol/L of element in liquid) for each REE in the variousacid molarities was calculated by the following equation:

Kd ¼ CB−CAð ÞV=wCA

where CB and CA are the concentrations of REE in solution before andafter equilibration, w is the weight of dry resin and V is volume of acidused in the equilibration experiment.

The results of the Kd experiments are shown in Fig. 3 and the linearregression statistics for each curve (slope and intercept) are given inTable 1. These data are consistent with the observations of Pierce et al.(1963), Horwitz and Bloomquist (1975) and Pin and Zaldegui (1997),who also noted similar linear trends for comparable experiments. Agreater spread of Kd (i.e. higher separation factors) was observed forHCl relative to HNO3, which may reflect slight complexation of theREEs at elevated HNO3 molarities.

4.3. Computer simulation of elution curve

As the REEs have very similar partitioning characteristics, minorchanges in the elution schemewill result in drastically different separa-tion patterns. To avoid a lengthy and potentially unsuccessful trial anderror approach, it was important to optimize the elution schemebefore-hand. Thus, we developed a chromatography simulation code (seeMathematica code in the Supplementalmaterials) based on plate theory(Martin and Synge, 1941) and an earlier programwritten by one of theauthors (Dauphas, 2002).

The plate theory of chromatography states that a chromatographiccolumn can be divided into a finite number of theoretical plates of de-fined height.Within each plate, and at any point in time, equilibriumbe-tween the liquid phase and the solid phase is achieved. Let's note Cs andCl as the concentration of the element in the solid phase (in mol/g) andthe liquid phase (inmol/mL), respectively. The affinity of an element fora particular resin is given by the partition coefficient, Kd, which isdefined as:

Kd ¼ Cs

Cl: ð1Þ

Fig. 3. Partition coefficients (Kd) of the REE on Ln-resin as a function of acid molarity for A) hydrochloric acid and B) nitric acid. The two plots are broadly similar, which is con-sistent with Horwitz and Bloomquist (1975) and Pin and Zaldegui (1997); however, there is a greater spread of Kd (i.e., higher separation factors) using HCl. The Kd values wereregressed for use in the elution simulations (Table 1).

209T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

As the liquid moves down into the column from theoretical plate totheoretical plate, it equilibrates with the solid in the next plate. Whenthe liquid in plate i − 1 moves down to into plate i, the distribution ofthe element contained in the liquid (from plate i − 1) and in the solid(from plate i) is not at equilibrium anymore (i.e. Kd ≠ Cs / Cl). The distri-bution of the element between the liquid and the solid phase changes ac-cordingly to the Kd value imposed by the reagent type and molarity. Thefollowing mass balance equation can be written to describe this process:

mis þmi−1

l ¼ miseq

þmileq

ð2Þ

where the exponents refers to the plate, ms and ml to the mass of ele-ment, respectively, in the solid phase and the liquid phase, and the

Table 1Linear regression statistics for determination of REE Kd as a function of acid molarity at room t

Element Equation (log form) Slope Err slope

For HClLa y = −1.841log(x) − 0.250 −1.841 0.026Ce y = −1.933log(x) + 0.013 −1.933 0.035Pr y = −2.000log(x) + 0.047 −2.000 0.056Nd y = −2.017log(x) + 0.096 −2.017 0.071Sm y = −2.681log(x) + 0.236 −2.681 0.151Eu y = −2.796log(x) + 0.446 −2.796 0.164Gd y = −2.800log(x) + 0.600 −2.800 0.160Tb y = −3.231log(x) + 1.046 −3.231 0.122Dy y = −2.958log(x) + 1.602 −2.958 0.249Ho y = −3.109log(x) + 1.830 −3.109 0.194Er y = −3.084log(x) + 2.191 −3.084 0.143Tm y = −3.084log(x) + 2.712 −3.084 0.126Yb y = −3.038log(x) + 3.151 −3.038 0.097Lu y = −2.967log(x) + 3.360 −2.967 0.108

For HNO3

La y = −1.709log(x) − 0.717 −1.709 0.051Ce y = −1.969log(x) − 0.870 −1.969 0.030Pr y = −2.049log(x) − 0.869 −2.049 0.028Nd y = −2.105log(x) − 0.849 −2.105 0.082Sm y = −2.316log(x) − 0.508 −2.316 0.045Eu y = −3.120log(x) − 0.951 −3.120 0.249Gd y = −3.128log(x) − 0.696 −3.128 0.423Tb y = −3.707log(x) − 0.612 −3.707 0.071Dy y = −3.537log(x) − 0.111 −3.537 0.131Ho y = −3.926log(x) + 0.159 −3.926 0.282Er y = −3.975log(x) + 0.566 −3.975 0.251Tm y = −3.957log(x) + 1.105 −3.957 0.230Yb y = −3.646log(x) + 1.646 −3.646 0.198Lu y = −3.520log(x) + 1.939 −3.520 0.161

a Conversion factor calculated from taking the center of the peak volume at 70 °C (from Fig.

subscript eq to values after equilibrium has been reached (i.e. Kd = Cs /Cl). This equation can be rewritten as:

dCisV

is þ Ci−1

l V i−1l ¼ dCi

seqVis þ Ci

leqVil ð3Þ

where d refers to the density of extractant-loaded beads, and Vl and Vs,respectively, to the volume of liquid and solid in the plate. The volumesof liquid and solid are the same in each plate, and can be expressed as afunction of the porosity of the resin (ϕ) and the volume of the plate Vi, as:

Vil ¼ Vi � ϕ ð4Þ

emperature.

y-int Err y-int r2 Kd conversion factor for 70 °Ca

−0.250 0.034 1.000 1.380.013 0.046 0.999 1.200.047 0.067 0.998 1.060.096 0.086 0.996 1.050.236 0.116 0.987 0.930.446 0.126 0.986 0.900.600 0.122 0.987 0.901.046 0.065 0.993 0.901.602 0.100 0.966 0.831.830 0.077 0.981 0.842.191 0.046 0.991 0.822.712 0.049 0.990 0.833.151 0.041 0.994 0.843.360 0.046 0.993 0.83

−0.717 0.067 0.998−0.870 0.040 1.000−0.869 0.036 1.000−0.849 0.108 0.997−0.508 0.059 0.999−0.951 0.224 0.987−0.696 0.421 0.982−0.612 0.054 0.999−0.111 0.100 0.995

0.159 0.115 0.9840.566 0.103 0.9841.105 0.094 0.9871.646 0.081 0.9881.939 0.066 0.992

6) divided by the center of the peak volume at room temperature (from Fig. 4).

210 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

and

Vis ¼ Vi � 1−ϕð Þ: ð5Þ

By substituting Eqs. (1), (4) and (5) into Eq. (3), one can calculatethe equilibrium concentration of the element in the liquid movingdown into the column in each plate as:

Cileq

¼ d 1−ϕð ÞCis þ ϕCi−1

l

d 1−ϕð ÞKd þ ϕ: ð6Þ

And the corresponding equilibrium concentration of the element inthe solid becomes:

Ciseq

¼ Kd � Cileq: ð7Þ

Eqs. (6) and (7) are used in the chromatography simulation code tocalculate the equilibrium condition within each plate in the columnthroughout the entire elution.

The input parameters to the simulation code are: 1) the columnlength and the parameters needed to calculate the volume of resin ineach theoretical plate (i.e., column radius, porosity, and the HETP[height equivalent to a theoretical plate]), 2) the volume of each elutionstep, 3) the density of the extractant-loaded beads and 4) the partitioncoefficient of the element(s) for the resin during each elution step (i.e.,Kd corresponding to the reagent type and molarity). Determination ofthe HETP for Ln-resin (~0.35 ±0.05 cm) was evaluated using the Kd

values measured for this study, and by adjusting the HETP value untilsimulated elution patterns fit published elution curves obtained onLn-resin (Pourmand and Dauphas, 2010; Ali and Srinivasan, 2011).

Using the above value of HETP, the Kd values of the REEs, a resin den-sity of 1.15 g/mL, and a porosity of 67% (also called FCV — free columnvolume — and equal to the free volume in mL per mL of packed resin,Horwitz et al., 2006)we simulated the elution patterns of the REE for var-ious ramps of HCl molarity (see Fig. 4 caption for detail on the ramps).Fig. 4 shows a set of molarity ramps (left panels) and their correspondingsimulated elution of the 14 REE + Yttrium (right panels). The bottompanels show the best optimized elution ramp and separation pattern.This elution scheme was implemented to test the T-HPLC (see nextsection). It is worth noting that the chromatography simulation code pro-vided in this publication is applicable to any elution scheme on any resin,provided that the appropriate input parameters are known.

4.4. Application of the T-HPLC system to the separation of the REEs

To test the performance of our T-HPLC system, we prepared a multi-element REE solution, containing 10 μg/g of each REE, as well as Sc andY (which behave similarly to the REEs). A 70 cm long column (diameterof 0.3 cm) was filled with Ln-resin with 25–50 μm resin bead size (thefinest mesh size commercially available). The experiment was conductedat 70 °Cwith the pressure adjusted to provide aflow rate of ~0.5 mL/min.The resin was initially cleaned using 50 mL of 10 M HCl and then condi-tioned with 30 mL of 0.05 M HCl. The standard solution was loadedonto the column in a 10 mL volume in 0.05 M HCl. Utilizing the elutioncurve simulation program to optimize our separation, we devised anelution scheme that would slowly increase HCl molarity along a pre-defined ramp starting at 0.1 M HCl and increasing to 10 M HCl over 183steps. Fig. 5 shows a plot of the desired molarity ramp as defined by theelution curve simulation, along with the calculated molarity from thesteps in the LabView program and several intervals where we titratedthe eluted solution to check its molarity. This figure demonstrates thatthe T-HPLC system is very effective and accurate at controlling reagentmixing and molarities.

The gradient rampwas programmed into the LabView program andthe whole elution scheme was automated. Each mixing step involved a

total of 4 mL of reagents, and the eluted solution was collected in either2 mL or 4 mL increments, depending onwhen elementswere predictedto elute. In total, 680 mL of solution were collected over a period of~15 h. These solutions were subsequently evaporated to near-drynessand then picked up in 0.46 M nitric acid for analysis on the NeptuneMC-ICP-MS. Admittedly, this is a long amount of time for a singlecolumn; however, most of the procedure is done automatically andthe quality of the separation would be difficult to achieve using a tradi-tional column setup.

The results of the elution are plotted in Fig. 6. There is some overlapof mono-isotopic REEs with neighboring multi-isotopic REEs (e.g., ProverlapswithNd), but this is not a big concern since there are no isobar-ic interferences between these isotopes. In addition, there is some dis-similarity between the elution curve from Fig. 6 and the simulatedelution curve from Fig. 4, although the broad shapes of each curve,and the relative placement of the elution peaks, are consistent witheach other. The slight discrepancy may be attributed to uncertaintiesin the experimentally determined Kd's and/or because the Kd's used inthe simulation were determined at room temperature, while the T-HPLC experiment was run at 70 °C. Conversion factors for the differ-ences between the Kd at 70 °C versus those at room temperature aregiven in Table 1, and range from 1.38 for La to ~0.83 for Dy to Lu.

Most importantly, however, the T-HPLC system achieved excellentseparation of the multi-isotopic REEs (La, Ce, Nd, Sm, Eu, Gd, Dy, Er, Yband Lu) from each other— a result that is difficult to obtainwith currenttechniques. Of particular note, there is strong separation of Nd fromboth Ce and Sm, which is a key factor for studies concerned withthe Sm–Nd isotopic systems. With some necessary refinement (seeSection 4.6), this result represents a major breakthrough for measuringREE isotopic abundances. All multi-isotopic REEs can now potentially beseparated and analyzed from a single sample digestion, helping to eluci-date various processes that affect the REE,while usingminimal amountsof sample. Overall, this result demonstrates the effectiveness of our sys-tem and its great potential to tackle even the most difficult of columnchromatography techniques.

4.5. Blank

A major concern with any new geochemical method, especiallythose that require large volumes of reagents, is the analytical blank.There are two main potential sources of blank contamination in aHPLC setup — the first from the reagents being used, and the secondfrom the resin. To minimize the blank from reagents, only acid thathad been distilled multiple times through quartz and Teflon stills wasutilized. Additionally, purifiedwater from aMilli-Q system,with a resis-tivity of 18.2 MΩ/cm, was used as the diluent. This careful preparationof reagents, as well as the internal monitoring of blank levels overtime, can help ensure that there is a minimal contribution from thesesources.

To address the potential blank contribution from the resin, a coupleof cleaning steps were implemented prior to the column chemistry.Firstly, loose resin (~25 to 50 g) was poured into a 1 L pre-cleanedPFA bottle, containing 6 M HCl. This solution was shaken vigorouslythroughout the day, and allowed to settle overnight. The followingmorning, the acid was decanted, the resin was rinsed with MQ water,and the process was repeated. After this step, the resin was stored inMQ until ready for use. After loading the resin into a column, an addi-tional cleaning step was employed (see Section 4.4) — 50 mL of 10 MHCl was eluted through the system, which should remove any REEthat may still be present because the Kd of the REE on Ln-resin at highHCl molarities is very low (e.g., Fig. 3).

A complete system blank was evaluated by running the elutionscheme outlined in Section 4.4 through the T-HPLC system, without aloading step. In this manner, a total of 670 mL of solution was collected,and was subsequently evaporated to near-dryness and picked upin 0.46 M HNO3 for analysis on the ICP-MS. Table 2 shows the

Fig. 4. Results of simulated elution of the REE on Ln-Spec resin obtained using various ramp of HCl molarity. The molarity ramps are calculated using the following arbitrary

function: D vð Þ ¼ Mr× 1− 1=e v=v fð Þp−1=e1−1=e

� �þMi where Vf is the total elution volume, Mr the range of molarity spanned during the elution, Mi the initial molarity of the ramp, and

V the eluted volume. p is a curvature parameter that can vary from 0 to ∞. Elution curves are computed using a Mathematica code based on plate theory (Martin and Synge,1941). The left panels show the curvature of the ramp as a function of the value of p, while the right panels show the corresponding simulated elution of the 14 REE + Yttrium.As can be seen, when separating the REE, the elution scheme (i.e. shape of the ramp) is a crucial parameter to optimize in order to obtain the best separation of the elements inthe minimum volume of elution. Such a result is achieved on the bottom panel for Mi = 0.1, Mr = 10, Vf = 700 mL and p = 4 (bottom panels).

211T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

concentration of REE blanks measured on the T-HPLC system, and acomparison of this blank with REE levels in CI chondrites (Pourmandet al., 2012). The measured system blanks range from 100 pg for Ndup to a high of 1481 pg for Ce, although blanks for individual elutionsteps would be reduced. We should be able to improve upon theseblanks in the future by cleaning further the reagents and decreasingthe diameter of the column to reduce the volume of acids and resinused. If 1 g of an average CI chondrite is dissolved and run throughthe T-HPLC system, this system blank would represent less than 0.5%for most of the REE.

4.6. Concern with Ln-resin and future efforts with REE separation

A noted concern with utilizing Ln-resin is a tailing effect of elutionpeaks that can often be observed as acid molarity is increased(Rehkämper et al., 1996; Pin and Zaldegui, 1997; Makishima et al.,2008; Hirahara et al., 2012). The small-step gradient ramp employedwith the T-HPLC system reduced this tailing significantly, compared toour previous attempts using an open system columnwith broadmolar-ity steps, but did not eliminate it. When the elution percent axis isviewed on a log scale, this tailing effect is readily apparent. The last

Fig. 5. Comparison of the desired elution ramp (gray line), asmodeled by the elution sim-ulation, and the calculatedmolarity from the LabView program (small green dots). Severalelution steps were titrated during the experiment to test the accuracy of our mixing andelution (open triangles). It is apparent that there is excellent agreement between thesecurves, illustrating the effectiveness and reliability of the acid pumping and mixingsystem. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

Table 2Evaluation of total system blank.

CI chondrite concentrationa Total system blank Blank as % of sampleb

ng/g pg

La 246.9 783 0.32Ce 632.1 1481 0.23Pr 95.9 149 0.16Nd 485.4 116 0.02Sm 155.6 330 0.21Eu 59.9 617 1.03Gd 209.3 882 0.42Tb 37.8 495 1.31Dy 257.7 514 0.20Ho 55.4 368 0.66Er 166.7 357 0.21Tm 26.1 238 0.91Yb 169.4 286 0.17Lu 25.6 169 0.66

a From Pourmand et al., 2012.b For 1 g of CI chondrite sample.

212 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

~0.25 to 0.75% of each REE has a long drawn-out tail that intersects theelution peaks of the other REE (Fig. 7).Whether this is a problem duringisotopic analysis remains to be tested. Note that the α-HIBA chemistryelutes REEs in the opposite order to Ln-resin (from heavy to light onα-HIBA; Lugmair et al., 1975; Rehkämper et al., 1996), so a secondpass of the REE cuts on cation-exchange resin using α-HIBA wouldvery effectively remove the tailing problem.

5. Conclusions

Here, we have developed and tested a novel T-HPLC system for usein addressing awide variety of geological and cosmochemical problems.Our new T-HPLC system represents an advancement over traditionalcolumn chromatographic techniques in a variety of ways:

1) All Teflon fluid flow path;2) fully automated elution schemes allowing for fresh mixing of re-

agents for each elution step and the ability to perform gradient/ramp elutions;

Fig. 6.Elution curve for the separation of theREE as a function of the volumeeluted. There is exceeach other, demonstrating the effectiveness of our T-HPLC system. The column was 70 cm lon

3) A modular design with the ability to vary column length, resin type,and flow rate;

4) temperature control over the entire system (up to 80 °C); and5) separation of electronic components from the HPLC system.

We tested our system on two separation schemes that are of partic-ular interest to the geochemical and cosmochemical communities. Inthe first test, we established that the separation of Ni from Mg is muchmore efficient and effective using a long column at elevated tempera-ture. Using these conditions, we effectively separated Ni from Mg inone column pass, as opposed to five passes that were utilized in priortraditional column chromatographic schemes.

In the second test, we demonstrated that individual REEs can be ef-fectively separated from each other, using a long column and elevatedtemperatures. This result may have important ramifications for howREE isotopic studies are performed in the future.While some challengesstill remain (i.e., tailing of the REEs), we are confident that our T-HPLCsystem will be able to address these obstacles.

The distinct attributes of the T-HPLC system gives the user muchmore power over the separationmethod and greatly simplifies the com-plex chemistry that is required for some separations. In addition, thissystem not only saves the user valuable time, but it also eliminatesmany of the potential issues with conventional column chemistry(e.g., inability to achieve fine-scale gradient ramps, human error). Ourmajor point of emphasis is that this T-HPLC system is readily adaptable

llent separation of all themulti-isotopic REE (La, Ce, Nd, Sm, Eu, Gd, Dy, Er, Yb, and Lu) fromg (0.3 cm diameter) and utilized Ln-resin at 25–50 mesh size, at a temperature of 70 °C.

Fig. 7. The same elution curve from Fig. 6 with the y-axis in log scale. Note the long tails of each REE, which intercept the elution curves of the other REE. This tailing accounts for 0.25 to0.75% of each element. Similar tailing behavior has been documented for the REEs on Ln-resin (e.g., Rehkämper et al., 1996; Pin and Zaldegui, 1997;Makishima et al., 2008; Hirahara et al.,2012).

213T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

to address a wide variety of chromatographic techniques across severalscientific fields and industries.

Acknowledgments

The authorswould like to thank Kevin Lynch andRichardDojutrek ofthe University of Illinois at Chicago, as well as Helmut Krebs of the Uni-versity of Chicago for their help and guidance with anything machineshop related. Marc-Alban Millet is also thanked for his help with theREE elution. This work greatly benefited from thorough reviews by T.Meisel and an anonymous reviewer, as well as from editorial remarksby L. Reisberg. This work was supported by NASA grant NNX12AH60Gand by a fellowship from the Packard Foundation to N.D. Some of thefeatures of this system are described in a U.S. Provisional PatentApplication (No. 61843509).

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.chemgeo.2013.08.001.

References

Albalat, E., Telouk, P., Albarède, F., 2012. Er and Yb isotope fractionation in planetarymaterials. Earth Planet. Sci. Lett. 355–356, 39–50.

Ali, A., Srinivasan, G., 2011. Precise thermal ionization mass spectrometric measurementsof 142Nd/144Nd and 143Nd/144Nd isotopic ratios of Nd separated from geological stan-dards by chromatographic methods. Int. J. Mass Spectrom. 299, 27–34.

Andreasen, R., Sharma, M., 2006. Solar nebula heterogeneity in p-process samarium andneodymium isotopes. Science 314, 806–809.

Andreasen, R., Sharma, M., 2007. Mixing and homogenization in the early solar system:clues from Sr, Ba, Nd and Sm isotopes in meteorites. Astrophys. J. 665, 874–883.

Baker, J., Waight, T., Ulfbeck, D., 2002. Rapid and highly reproducible analysis of rare earthelements by multiple collector inductively coupled plasma mass spectrometry.Geochim. Cosmochim. Acta 6, 3635–3646.

Balaram, V., 1996. Recent trends in the instrumental analysis of rare earth elements ingeological and industrial materials. Trends Anal. Chem. 15, 475–486.

Bennett, V.C., Brandon, A.D., Nutman, A.P., 2007. Coupled 142Nd–143Nd isotopic evidencefor Hadean mantle dynamics. Science 318, 1907–1910.

Birck, J.L., Lugmair, G.W., 1988. Nickel and chromium isotopes in Allende inclusions. EarthPlanet. Sci. Lett. 90, 131–143.

Bogard, D.D., Nyquist, L.E., Bonsal, B.M., Garrison, D.H., Wiesmann, H., Herzog, G.F.,Albrecht, A.A., Vogt, S., Klein, J., 1995. Neutron capture 36Cl, 41Ca, 36Ar, and 150Sm inlarge chondrites: evidence for high fluences of thermalized neutrons. J. Geophys.Res. 100, 9401–9416.

Bouvier, A., Vervoort, J.D., Patchett, P.J., 2008. The Lu–Hf and Sm–Nd isotopic compositionof CHUR: constraints from unequilibrated chondrites and implications for the bulkcomposition of the terrestrial planets. Earth Planet. Sci. Lett. 273, 48–57.

Boyet, M., Carlson, R.W., 2005. 142Nd evidence for early (N4.53 Ga) global differentiationof the silicate Earth. Science 309, 576–581.

Burnett, D.S., Huneker, J.C., Podosek, F.A., Price Russ III, G., Wasserburg, G.J., 1971. The ir-radiation history of lunar samples. Proc. 2nd Lunar Sci. Conf, pp. 1671–1679.

Carlson, R.W., Boyet, M., Horan, M., 2007. Chondrite barium, neodymium and samariumisotopic heterogeneity and early earth differentiation. Science 316, 1175–1178.

Caro, G., Bourdon, B., Birck, J.L., Moorbath, S., 2003. 146Sm–142Nd evidence from Isuameta-morphosed sediments for early differentiation of the Earth's mantle. Nature 423,428–432.

Caro, G., Bourdon, B., Birck, J.L., Moorbath, S., 2006. High-precision 142Nd/144Nd measure-ments in terrestrial rocks: constraints on the early differentiation of the Earth's man-tle. Geochim. Cosmochim. Acta 70, 164–191.

Caro, G., Bourdon, B., Halliday, A.N., Quitté, G., 2008. Super-chondritic Sm/Nd in Mars,Earth, and the Moon. Nature 452, 336–339.

Cassidy, C.M., Chauvel, C., 1989. Modern liquid chromatographic techniques for the sepa-ration of Nd and Sr for isotopic analysis. Chem. Geol. 74, 189–200.

Cerrai, E., Testa, C., 1963. Separation of rare earths by means of small columns of Kel-F supporting di(2-ethylhexyl)orthophosphoric acid. J. Inorg. Nucl. Chem. 25,1045–1050.

Crock, J.G., Lichte, F.E., Wildeman, T.R., 1984. The group separation of the rare earth ele-ments and yttrium from geological materials by cation-exchange chromatography.Chem. Geol. 45, 149–163.

Dauphas, N., 2002. Cosmochimie isotopique du Molybdene. (PhD thesis) INPL.Dauphas, N., Cook, D.L., Sacarabany, A., Frölich, C., Davis, A.M., Wadwha, M., Pourmand, A.,

Rauscher, T., Gallino, R., 2008. Iron-60 evidence for early injection and efficientmixing of stellar debris in the protosolar nebula. Astrophys. J. 686, 560–569.

Debaille, V., Brandon, A.D., Yin, Q.Z., Jacobsen, B., 2007. Coupled 142Nd–143Nd evidence fora protracted magma ocean in Mars. Nature 450, 525–528.

DePaolo, D.J., Wasserburg, G.J., 1976. Nd isotopic variability and petrogenetic models.Geophys. Res. Lett. 3, 249–252.

Dionex, 1992. Determination of lanthanide metals in digested rock samples by chelationion chromatography. Technical note. 27.

Eugster, O., Tera, F., Burnett, D.S., Wasserburg, G.J., 1970a. The isotopic composition of Gdand the neutron capture effects in samples from Apollo 11. Earth Planet. Sci. Lett. 8,20–30.

Eugster, O., Tera, F., Burnett, D.S., Wasserburg, G.J., 1970b. Isotopic composition of gadolin-ium and neutron-capture effects in some meteorites. J. Geophys. Res. 75, 2753–2768.

Gannoun, A., Boyet, M., Rizo, H., El Goresy, A., 2011. 146Sm–142Nd systematics measured inenstatite chondrites reveals a heterogeneous distribution of 142Nd in the solar nebula.Proc. Natl. Acad. Sci. 108, 7693–7697.

Gast, P.W., Hubbard, N.J., Weismann, H., 1970. Chemical composition and petrogenesis ofbasalts from Tranquility base. Proceeds Apollo 11 Lunar Sci. Conf. 2, 2, pp. 1143–1163.

Guan, Y., Huss, G.R., Leshin, L.A., 2007. 60Fe–60Ni and 53Mn–53Cr isotopic systems in sul-fides from unequilibrated enstatite chondrites. Geochim. Cosmochim. Acta 71,4082–4091.

Haley, B.A., Klinkhammer, G.P., 2003. Complete separation of rare earth elements fromsmall volume seawater samples by automated ion chromatography: method devel-opment and application to benthic flux. Mar. Chem. 82, 197–220.

Henderson, P., Pankhurst, R.J., 1984. Chapter 13: analytical chemistry. In: Henderson, P.(Ed.), Rare Earth Element Geochemistry. Elsevier, New York, New York.

Hidaka, H., Yoneda, S., 2007. Sm and Gd isotopic shifts of Apollo 16 and 17 drill stemsamples and their implications for regolith history. Geochim. Cosmochim. Acta71, 1074–1086.

Hidaka, H., Ebihara, M., Shima, M., 1995. Determination of the isotopic compositions ofsamarium and gadolinium by thermal ionization mass spectrometry. Anal. Chem.67, 1437–1441.

Hidaka, H., Ebihara, M., Yoneda, S., 2000a. Neutron capture effects on samarium,europium and gadolinium in Apollo 15 deep drill-core samples. Meteorit. Planet.Sci. 35, 581–589.

214 T.J. Ireland et al. / Chemical Geology 357 (2013) 203–214

Hidaka, H., Ebihara, M., Yoneda, S., 2000b. Isotopic study of neutron capture ejects on Smand Gd in chondrites. Earth Planet. Sci. Lett. 180, 29–37.

Hidaka, H., Yoneda, S., Nishiizumi, K., 2009. Cosmic-ray exposure histories of Martian me-teorites studied from neutron capture reaction of Sm and Gd isotopes. Earth Planet.Sci. Lett. 288, 564–571.

Hirahara, Y., Chang, Q., Miyazaki, T., Takahasi, T., Kimura, J.-I., 2012. Improved Nd chemicalseparation technique for 143Nd/144Nd analysis in geological samples using packed Lnresin columns. JAMSTEC Rep. Res. Dev. 15, 27–33.

Hooker, P.J., O'Nions, R.K., Pankhurst, R.J., 1975. Determination of rare-earth elements inUSGS standard rocks bymixed-solvent ion exchange and mass-spectrometric isotopedilution. Chem. Geol. 16, 189–196.

Horwitz, E.P., Bloomquist, C.A.A., 1975. Chemical separations for super-heavy elementsearches in irradiated uranium targets. J. Inorg. Nucl. Chem. 37, 425–434.

Horwitz, E.P., McAlister, D.R., Dietz, M.L., 2006. Extraction chromatography versus solventextraction: how similar are they? Sep. Sci. Technol. 41, 2163–2182.

Jacobsen, S.B., Wasserburg, G.J., 1984. Sm–Nd isotopic evolution of chondrites and achon-drites, II. Earth Planet. Sci. Lett. 67, 137–150.

Lugmair, G.W., Marti, K., 1971. Neutron capture effects in lunar gadolinium and the irra-diation histories of some lunar rocks. Earth Planet. Sci. Lett. 13, 32–42.

Lugmair, G.W., Scheinin, N.B., Marti, K., 1975. Sm–Nd age and history of Apollo 17 basalt75075: evidence for early differentiation of the lunar exterior. Proc. 6th Lunar Sci.Conf, pp. 1419–1429.

Makishima, A., Nath, B.N., Nakamura, E., 2008. New sequential separation procedure forSr, Nd and Pb isotope ratio measurement in geological material using MC-ICPMSand TIMS. Geochem. J. 42, 237–246.

Marhas, K.K., Mishra, R.K., 2012. Fossil record of 60Fe in QUE97008 chondrule. 75th Met.Soc, p. #5273.

Martin, A.J.P., Synge, R.L.M., 1941. A new form of chromatogram employing two liquidphases. 1. A theory of chromatography. 2. Application to the micro-determinationof the higher monoamino-acids in proteins. Chromatogr. Amino Acids 151,1358–1368.

McCulloch, M.T., Wasserburg, G.J., 1978a. Barium and neodymium isotopic anomalies inthe Allende meteorite. Astrophys. J. 220, L15–L19.

McCulloch, M.T., Wasserburg, G.J., 1978b. More anomalies from the Allende meteorite:samarium. Geophys. Res. Lett. 5, 599–602.

Meisel, T., Schöner, N., Paliulionyte, V., Kahr, E., 2001. Determination of rare earth ele-ments, Y, Th, Zr, Hf, Nb and Ta in Geological Reference Materials G-2, G-3, SCo-1and WGB-1 by sodium peroxide sintering and inductively coupled plasma-massspectrometry. Geostand. News. 26, 53–61.

Mostefaoui, S., Lugmair, G.W., Hoppe, P., El Goresy, A., 2004. Evidence for live 60Fe inmeteorites. New Astron. Rev. 48, 155–159.

Munker, C., Weyer, S., Scherer, E., Mezger, K., 2001. Separation of high field strength ele-ments (Nb, Ta, Zr, Hf) and Lu from rock samples of MC-ICPMS measurements.Geochem. Geophys. Geosyst. 2, 1–19.

Nash, K.L., Jensen, M.P., 2001. Analytical-scale separations of the lanthanides: a review ofthe techniques and fundamentals. Sep. Sci. Technol. 36, 1257–1282.

O'Neil, J., Carlson, R.W., Francis, D., Stevenson, R.K., 2008. Neodymium-142 evidence forHadean mafic crust. Science 321, 1828–1831.

Pierce, T.B., Peck, P.F., Hobbs, R.S., 1963. The separation of the rare earths by partitionchromatography with reversed phases — part II: behaviour of individual elementson HDEHP-Corvic columns. J. Chromatogr. 12, 81–88.

Pin, C., Zaldegui, J.F.S., 1997. Sequential separation of light rare-earth elements, thoriumand uranium by miniaturized extraction chromatography: application to isotopicanalyses of silicate rocks. Anal. Chim. Acta 339, 79–89.

Pourmand, A., Dauphas, N., 2010. Distribution coefficients of 60 elements of TODGA resin:application to Ca, Lu, Hf, U and Th isotope geochemistry. Talanta 81, 741–753.

Pourmand, A., Dauphas, N., Ireland, T.J., 2012. A novel extraction chromatography andMC-ICPMS technique for rapid analysis of REE, Sc, and Y: revising CI chondrite abun-dances. Chem. Geol. 291, 38–54.

Pritchard, G., 1995. Anti-corrosion polymers: PEEK, PEKK and other polyaryls. Rapra Rev.Rep. 7.

Qin, L., Carlson, R.W., Alexander, C.M.O'D., 2011. Correlated nucleosynthetic isotopicvariability in Cr, Sr, Ba, Sm, Nd and Hf in Murchison and QUE 97008. Geochim.Cosmochim. Acta 75, 7806–7828.

Quitté, G., Halliday, A.N., Meyer, B.S., Markowski, A., Latkoczy, C., Gunther, D., 2007.Correlated iron-60, nickel-62 and ziconcium-96 in refractory inclusions and theorigin of the solar system. Astrophys. J. 655, 678–684.

Quitté, G., Markowski, A., Latkoczy, C., Gabriel, A., Pack, A., 2010. Iron-60 heterogeneity andincomplete isotope mixing in the early solar system. Astrophys. J. 720, 1215–1224.

Quitté, G., Latkoczy, C., Schönbächler, M., Halliday, A.N., Gunther, D., 2011. 60Fe–60Nisystematics in the eucrite parent body: a case study of Bouvante and Juvinas.Geochim. Cosmochim. Acta 75, 7698–7706.

Regelous, M., Elliott, T., Coath, C.D., 2008. Nickel isotope heterogeneity in the early SolarSystem. Earth Planet. Sci. Lett. 272, 330–338.

Rehkämper, M., Gärtner, M., Galer, S.J.G., Goldstein, S.L., 1996. Separation of Ce from otherrare-earth elements with application to Sm–Nd and La–Ce chronometry. Chem. Geol.129, 201–208.

Rugel, G., Faestermann, T., Knie, K., Korschinek, G., Poutivtsev, M., Schumann, D., Kivel, N.,Günther-Leopold, I., Weinreich, R., Wohlmuther, M., 2009. New measure of the 60Fehalf-life. Phys. Rev. Lett. 103, 072502.

Russ III, G.P., Burnett, D.S., Lingenfelter, R.E., Wasserburg, G.J., 1971. Neutron capture on149Sm in lunar samples. Earth Planet. Sci. Lett. 13, 53–60.

Shukolyukov, A., Lugmair, G.W., 1993a. Live iron-60 in the early Solar System. Science259, 1138–1142.

Shukolyukov, A., Lugmair, G.W., 1993b. 60Fe in eucrites. Earth Planet. Sci. Lett. 119, 159–166.Siekerski, S., Sochacka, R.J., 1964. Reversed-phase partition chromatography with

di-(2-ethylhexyl)orthophosphoric acid as the stationary phase. J. Chromatogr.16, 385–395.

Sisson, D.H., AlanMode, V., Campbell, D.O., 1972. High-speed separation of the rare earthsby ion exchange part II. J. Chromatogr. 69, 129–135.

Sivaraman, N., Kumar, R., Subramaniam, S., Vasudeva Rao, P.R., 2002. Separation of lantha-nides using ion-interaction chromatography with HDEHP coated columns.J. Radioanal. Nucl. Chem. 252, 491–495.

Steele, R.C.J., Elliott, T., Coath, C.D., Regelous, M., 2011. Confirmation of mass-independentNi isotopic variability in iron meteorites. Geochim. Cosmochim. Acta 75, 7906–7925.

Strelow, F.W.E., 1988. Distribution coefficients and ion-exchange selectivities for 46 ele-ments with a macroporous cation-exchange resin in hydrochloric acid–acetonemedium. Talanta 35, 385–395.

Strelow, F.W.E., Jackson, P.F.S., 1974. Determination of trace and ultra-trace quantities ofrare-earth elements by ion exchange chromatography-mass spectrography. Anal.Chem. 46, 1481–1486.

Strelow, F.W.E., Victor, A.H., van Zyl, C.R., Eloff, C., 1971. Distribution coefficients andcation exchange behavior of elements in hydrochloric acid–acetone. Anal. Chem.43, 870–876.

Tachibana, S., Huss, G.R., 2003. The initial abundance of Fe-60 in the Solar System.Astrophys. J. Lett. 588, L41–L44.

Tang, H., Dauphas, N., 2012. Abundance, distribution and origin of 60Fe in the solar proto-planetary disk. Earth Planet. Sci. Lett. 359–360, 248–263.

Thrane, K., Connelly, J.N., Bizzarro, M., Meyer, B.S., The, L.-S., 2010. Origin of excess 176Hf inmeteorites. Astrophys. J. 717, 861–868.

Verma, S.P., Santoyo, E., 2007. High-performance liquid and ion chromatography: separa-tion and quantification analytical techniques for rare earth elements. Geostand.Geoanal. 31, 161–184.

Walsh, J.N., Buckley, F., Barker, J., 1981. The simultaneous determination of rare earth el-ements in rocks using inductively coupled plasma source spectrometry. Anal. Chem.74, 141–148.

Wasserburg, G.J., Jacobsen, S.B., DePaolo, D.J., McCulloch, M.T., Wen, T., 1981. Precise de-termination of Sm/Nd ratios, Sm and Nd isotopic abundances in standard solutions.Geochim. Cosmochim. Acta 45, 2311–2323.

Yoshida, K., Haraguchi, H., 1984. Determination of rare earth elements by liquid chroma-tography/inductively coupled plasma atomic emission spectrometry. Anal. Chem. 56,2580–2585.