ORIGINAL PAPER

Isotopic fractionation of mercury induced by reductionand ethylation

Lu Yang & Ralph E. Sturgeon

Received: 12 June 2008 /Revised: 6 August 2008 /Accepted: 12 August 2008 / Published online: 11 September 2008# Springer-Verlag 2008

Abstract Isotope ratio measurements characterizing 202Hg/200Hg in NIST SRM 3133 Mercury Standard Solution wereundertaken by multicollector inductively coupled plasmamass spectrometry employing NIST SRM 997 Tl for massbias correction by use of the slope and the intercept obtainedfrom a natural logarithmic plot of each session of measure-ments of 202Hg/200Hg against 205Tl/203Tl. The calculatedvalue of 1.285333±0.000192 (mean and one standarddeviation, n=40) for the mass bias corrected 202Hg/200Hgwas then used for mass bias correction of other Hg isotopepairs. Ratios of 0.015337±0.000011, 1.68770±0.00054,2.3056±0.0015, 1.3129±0.0013, 2.9634±0.0038, and0.67937±0.0013 (expanded uncertainty, k=2) were obtainedfor 196Hg/198Hg, 199Hg/198Hg, 200Hg/198Hg, 201Hg/198Hg,202Hg/198Hg, and 204Hg/198Hg, respectively. Reduction ofHg(II) to Hg0 in solutions of SRM 3133 was then undertakenusing SnCl2, NaBH4, UV photolysis in the presence offormic acid, and ethylation of Hg(II) using NaBEt4. Thesereactions induced significant isotope fractionation withmaximum values of 1.17±0.07, 1.08±0.09, 1.34±0.07,and 3.59±0.09‰ (one standard deviation, 1SD, n=5) for δ202/198Hg relative to the initial isotopic composition in thesolution following 85–90% reduction of the Hg by SnCl2,NaBH4, UV photolysis, and ethylation with NaBEt4,respectively. Mass-dependent fractionation was found to bedominant for all reduction processes.

Keywords Isotope ratios .Multicollector inductivelycoupled plasmamass spectrometry .Mass-independentfractionation . Isotopic fractionation .Mercury reduction .

Mercury ethylation

Introduction

Mercury is introduced into the environment through bothgeochemical and anthropogenic activities [1], resulting inwell-known environmental pollution. Isotope fractionation ofnontraditional elements has become a valuable tool forstudying the biogeochemical cycles of metals [2] and theintroduction of multicollector (MC) inductively coupledplasma mass spectrometry (ICP-MS) has allowed the highlyprecise determination of isotope ratios in a variety of samples.Recent publications from several research groups studyingHg have demonstrated per mil level variations in isotopiccomposition in both natural samples and induced duringlaboratory experiments [3–19], suggesting that isotopes of Hgcan provide a useful new tool for fingerprinting sources ofthis element in the environment and for studying a widevariety of chemical and biological processes in nature.

There is currently a lack of Hg standards characterizedfor isotopic composition at levels of sufficient accuracy andprecision to foster such studies for Hg [5, 11]. In an effortto facilitate comparison of data from independent laborato-ries, Blum and Bergquist [11] suggested that a commonstandard be adopted for Hg, noting that National Institute ofStandards and Technology (NIST) SRM 3133 (MercuryStandard Solution) would present a satisfactory candidatefor this purpose. Generally, Hg isotope ratio measurementscharacterizing isotopic fractionation are usually reported inδ notation relative to a Hg standard. Such δ values are

relative:sample samplem

std stdm

1 1000‰R K

R K= − ×δ , where Rsample

mand Rstd

mare

measured ratios in the sample and standard and Ksample andKstd are mass bias correction factors in the sample andstandard. Under ideal conditions, this approach obviates theneed for accurate mass bias correction (assuming it is samefor both sample and standard solutions under the measure-ment conditions, Ksample = Kstd) and thus measured ratios can

be used to derivesamplem

stdm

1 1000‰R

Rδ ≈ − × . Unfortunately, these

Anal Bioanal Chem (2009) 393:377–385DOI 10.1007/s00216-008-2348-6

L. Yang (*) :R. E. SturgeonInstitute for National Measurement Standards,National Research Council Canada,Ottawa, ON K1A 0R6, Canadae-mail: Lu.Yang@nrc-cnrc.gc.ca

conditions are frequently not met with real matrix samples andbecause nonstable mass discrimination within the ICP-MSsystem may be encountered. Alternatively, a double spiketechnique for mass bias correction can be considered if twoenriched Hg isotope spikes with known isotopic compositionare available. Accurate and precise ratio measurements [20]are achieved because the mass bias can be directly determinedin the sample matrix. To date, a double spike for determinationof Hg isotope ratios has not been established, possibly owingto the cost of high-purity enriched spikes, the significantefforts associated with accurate calibration of the compositionof the spikes, and possible cross-contamination betweenanalytical runs of unspiked and spiked samples. Thus,accurate mass bias correction remains a general impedimentto determination of isotope ratios and, as such, a referencematerial with known isotopic composition having adequateprecision and accuracy is still desired to provide validation ofanalytical methods for determination of Hg.

Isotopic fractionation of Hg can be induced by manyprocesses, both mass-dependent [3–13] and mass-independent[14–16]. Reduction of Hg species to Hg0 vapor is animportant pathway for the transfer of Hg from aquatic systemsto the atmosphere [14]. To distinguish the sources of atmo-spheric Hg and to better understand Hg cycling at the water/air interface, investigation of such fractionation occurringduring the reduction/volatilization process is needed. In thelaboratory, volatilization of Hg (vapor generation) from theaqueous phase can be easily achieved by chemical reaction ofHg(II) with SnCl2 and NaBH4 reductants, by alkylation withNaBEt4 [21, 22], or by UV photolysis [23]. In this study,isotopic fractionation of Hg induced during such reactionswas investigated.

Experimental

Instrumentation

AThermo Fisher Scientific Neptune (Bremen, Germany) MC-ICP-MS system equipped with nine Faraday cups and acombination of cyclonic and Scott-type spray chambers with aperfluoroalkoxy MCN50 self-aspirating nebulizer (ElementalScientific, Omaha, NE,USA) operating at 50μl min−1 was usedfor all measurements. The plug-in quartz torch with sapphireinjector was fitted with a platinum guard electrode. Low-resolution mode was used in this study with Rpower(5,95%) ∼300. Optimization of the Neptune system was performed asrecommended by the manufacturer; typical operating condi-tions are summarized in Table 1.

A 6-W UVC Hg pen lamp (model 81-1057-51, Analamp,Claremont, CA, USA, λmax 253.7 nm) having a 5-cm lightedlength and 0.5-cm outer diameter was used for UV photolysisof solutions of Hg(II) in dilute formic acid.

Reagents and solutions

Nitric acid and hydrochloric acid were purified in-house priorto use by subboiling distillation of reagent grade feedstock in aquartz still. High-purity (18 MΩ cm) deionized water wasobtained from a NanoPure mixed-bed ion-exchange systemfed with reverse osmosis domestic feed water (Barnstead/Thermolyne, Dubuque, IA, USA). A 0.2 M BrCl solution wasprepared in a fume hood by dissolving 27 g KBr (FisherScientific, Nepean, Canada) in 2.5 l HCl, then slowly adding38 g KBrO3 while stirring. Environmental grade ammonia(20–22%, v/v) and formic acid were obtained from Anache-mia Science (Montreal, Canada). A 2% (m/v) solution ofSnCl2 was prepared daily by dissolving stannous chloride(American Chemicals, Montreal, Canada) in 2% HCl.Sodium tetraethylborate solution, 0.5% (m/v), was prepareddaily by dissolving NaBEt4 (Strem, Bischeim, France) indeionized water. Sodium tetrahydroborate solution, 0.5% (m/v),was prepared daily by dissolution of NaBH4 pellets (AlfaAesar, Word Hill, MA, USA) in deionized water. SRM 3133and SRM 997 (Tl isotopic) were obtained from the NIST(Gaithersburg, MD, USA). A 1,000 μg g−1 stock solution ofTl was prepared by quantitatively dissolving SRM 997 in afew milliliters of HNO3, followed by dilution with deionizedwater to result in a matrix of 5% HNO3.

Sample preparation for isotope ratio measurements

Sample preparation was conducted in a class-100 clean room.Replicate 500 ng g−1 solutions of SRM 3133 used to establish

Table 1 Multicollector inductively coupled plasma mass spectrome-try operating conditions

Instrument settingsRF power 1,250 WPlasma Ar gas flow rate 15.0 l min−1

Auxiliary Ar gas flow rate 1.00 l min−1

Ar carrier gas flow rate 1.075 l min−1

Sampler cone (Ni) 1.1 mmSkimmer cone (Ni) 0.8 mmLens settings Optimized for maximum

analyte signal intensityData acquisition parametersScan type Static measurementsCup configuration (196Hg)L4, (198Hg)L3, (199Hg)L2,

(200Hg)L1, (201Hg)C, (202Hg)H1,(203Tl)H2, (204Hg)H3, (205Tl)H4,(195Pt)IC4, and (206Pb)IC5

Resolution ∼300Sensitivity 1.2 V/ppm for 201HgBlank 2% HCl, 0.002 N BrCl 0.003 V for 201HgIntegration time 4.194 sNumber of integrations 1Number of cycles/blocks 10/5

378 L. Yang, R. E. Sturgeon

mass bias corrected 202Hg/200Hg were prepared by dilutingSRM 3133 in a solution of 2% HCl containing 0.002 N BrCl,followed by spiking the solution with a 10 μg g−1 solution ofTl (diluted SRM 997), yielding a working concentration of100 ng g−1. For the final determination of all Hg isotoperatios in SRM 3133, 18 replicate 500 ng g−1 test sampleswere prepared by diluting the stock in a solution of 2% HClcontaining 0.002 N BrCl.

Sample preparation for isotope fractionation studies

All experiments were performed in a fume hood in a 150-ml-volume quartz beaker. A 100-ml volume of a 20 μg g−1

concentration of SRM 3133 in 2% formic acid, 2% HCl, 2%HCl, or deionized water was used for vapor generation by UVphotolysis, by chemical reduction using SnCl2 or NaBH4 or byethylation with NaBEt4. The volatile Hg species producedduring these reactions was removed by continuously spargingthe solution with a flow of argon gas at about 2 l min−1. ForUV photolysis, the irradiation time was varied from 30 s to5 min to reduce/remove 10-90% of the initial concentration ofHg(II) from the beaker. Similarly, when reduction/ethylationwas achieved using SnCl2, NaBH4, or NaBEt4 to result inconcentrations of Hg(II) remaining in the range 90–10% of theinitial concentration, incremental volumes of 30–120 μl of 2%SnCl2, 50–200 μl of 0.5% NaBH4, or 200–1,000 μl of 0.5%NaBEt4, respectively, were added to the test solution. After a5-min reaction/sparing time to ensure complete volatilization ofthe volatile Hg fraction, a 1-ml aliquot of the solution waspipetted into a precleaned 50-ml plastic vial, diluted to 20 ml in2% HCl containing 0.002 N BrCl, and a spike of the 10 μg g−1

Tl (SRM 997) standard solution was added to yield aconcentration of 100 ng g−1. Major matrix elements such asSn, B, and Na were assessed in the samples by a quickmeasurement using the MC-ICP-MS system. A bracketingstandard solution of SRM 3133 in 2% HCl containing 0.002 NBrCl was then prepared so as to contain similar concentrationsof Hg, Tl, Sn, Na, and B. The concentrations of samples andbracketing standards were matched to within 5%.

Analysis procedure and data processing

Samples and standards were introduced into the plasma usingself-aspiration mode at 50 μl min−1. The intensities of Hg andall other measured isotopes of interest (see below) obtainedfrom a blank solution of 2% HCl and 0.002 N BrCl weresubtracted from those of all samples and standards. For Hgfractionation studies, samples were introduced in the se-quence SRM 3133-sample-SRM 3133. A static run wasemployed collecting 196Hg, 198Hg, 199Hg, 200Hg, 201Hg,202Hg, 203Tl, 204Hg, and 205Tl isotopes on Faraday cups atlow 4, low 3, low 2, low 1, central, high 1, high 2 high 3, andhigh 4 positions, respectively. 195Pt and 206Pb were monitored

using ion counter 4 and ionization chamber 5. Dataacquisition parameters are summarized in Table 1. Tenmeasurements were made on each sample solution. Minorisobaric interferences from 196Pt on 196Hg, 198Pt on 198Hg,and 204Pb on 204Hg were automatically corrected for on thebasis of the intensities simultaneously measured for 195Pt and206Pb. Signals intensities for Pt and Pb were 1,000 cps indiluted solutions of SRM 3133 and thus exerted no detectableinfluence on corrected Hg isotope ratios. IUPAC values for195Pt/196Pt, 195Pt/198Pt, and 206Pb/204Pb were assumed andion counter and Faraday cups were cross-calibrated inaccordance with the manufacturer’s recommended proce-dures. No significant memory effect from Hg was evidentwhen using solutions of 2% HCl containing 0.002 N BrCl asthe sample matrix and wash solutions; the 201Hg signal(0.6 V for 500 ng g−1 Hg solution) was reduced to its baselinelevel (0.003 V) within several minutes of washing.

Mass bias correction

Use of Tl (205Tl/203Tl=2.38714), assuming it suffers anidentical mass bias correction factor as that for Hg, iscommon practice for implementing internal standardizationfor mass bias correction of Hg intensity ratios [10–14, 18].However, it has been recognized in recent years that differentelements can be isotopically fractionated differently in MC-ICP-MS [2] and, as a result, absolute isotope ratios derivedin this manner for Hg can be erroneous. With absolute valuesof isotope ratios also being of interest, an alternativeapproach was adopted. With use of an exponential law formass bias correction [24], the following relationships areevident for Hg and Tl, respectively:

RHg202=200

m ¼ RHg202=200

T

m202

m200

� �f Hg

ð1aÞand

RTl205=203

m ¼ RTl205=203

T

m205

m203

� �f Tl

; ð1bÞ

where subscripts m and T denote measured and mass biascorrected or true ratios, m202, m200, m205, and m203 are absolutemasses of the isotopes of interest, and f is the mass biascorrection factor. The following equation can be obtained bytaking the natural logarithm of Eqs. 1a and 1b and rearranging:

lnRHg202=200

m ¼ lnRHg202=200

T � f Hg

f Tl

ln M202M200

� �

ln M205M203

� � lnRTl205=203

T

24

35

þ f Hg

f Tl

ln M202M200

� �

ln M205M203

� � lnRTl205=203

m : ð2Þ

Isotopic fractionation of mercury induced by reduction and ethylation 379

From a plot of lnR202=200m versus lnR205=203

m , the slopeand the intercept derived during each 10–15-h measurementsession can be calculated and used to obtain RHg202=200

T :

s ¼ f Hg

f Tl

ln M202M200

� �

ln M205M203

� � ; ð3Þ

Intercept ¼ lnRHg202=200

T � s lnRTl205=203

T ; ð4Þ

where s is the slope. The NIST certified value of 2.38714 wasused for RTl205=203

T to obtain mass bias corrected RHg202=200

T . Thismass bias corrected 202Hg/200Hg ratio was subsequently usedfor internal mass bias correction of all other Hg isotope ratiosin SRM 3133. This method is similar to previously reportedapproaches successfully applied to the determination of Pbisotope ratios using Tl, of Cu isotope ratios using Zn and viceversa [25–27] and of Hg isotope ratios using Tl [8]. However,in the last of those studies [8], the raw data obtained during a20-month measurement period resulted in plots that exhibitedrather poor correlation coefficients (i.e., R2=0.84).

Results and discussion

Isotope ratios for SRM 3133

Good measurement precision for isotope ratios wasachieved using an integration time of 4.194 s acquiring

data with ten cycles and five blocks within a reasonablemeasurement time. One measurement requires about 8 minincluding uptake and rising time and consumes approxi-mately 0.3 ml of sample, corresponding to 150 ng Hg.

As noted earlier, Tl has been widely used for internal massbias correction of Hg data, but identical mass bias correctionfactors have been assumed for both elements [10–14, 18]. Toexamine whether this approach generates accurate data forHg ratios, different test solution concentrations of Hg wereprepared from SRM 3133 and spiked with a fixed Tlconcentration; as well, a fixed Hg solution concentrationwas examined containing different Tl spike levels. As evidentin Fig. 1, mass bias corrected 202Hg/200Hg varies, confirmingthat mass bias correction based on Eqs. 1a and 1b and theerroneous assumption that f Hg = f Tl is not adequate fordetermination of absolute ratios, as also noted by others [8].

Equation 3 expresses a linear relationship between thenatural logarithms of measured 202Hg/200Hg and 205Tl/203Tland, as evident from Fig. 2, results in a well-defined linearrelationship (R2>0.995, n=10) over a measurement session

1.285

1.286

1.287

1.288

1.289

1.290

1.291

1.292

0 200 400 600 800 1000 1200

Hg or Tl Concentration

202 H

g/20

0 H

Fig. 1 Effect of Hg and Tl concentrations on mass bias corrected202Hg/200Hg. Circles Tl fixed at 100 ng g−1 and squares Hg fixed at500 ng g−1

0.2648

0.2653

0.2658

0.2663

0.2668

0.2673

0.2678

0.8822 0.8828 0.8834 0.8840 0.8846 0.8852

0.2630

0.2636

0.2642

0.2648

0.2654

0.2660

0.2666

0.8810 0.8815 0.8820 0.8825 0.883 0 0.8835 0.8840

B

Aln

(202 H

g/20

0 Hg)

ln (Tl205/Tl203)

Fig. 2 ln-ln space plot for Hg and Tl ratios arising from eachmeasurement session. Error bars (internal precisions) are smaller thanthe plotted symbol. A circles 500 ng g−1 Hg and Tl fixed at 100 ng g−1,y ¼ 1:109857x� 0:714631, R2=0.9993; squares 100 ng g−1 Hg and Tlfixed at 100 ng g−1, y ¼ 1:078463x� 0:687437, R2=0.9960. B circles500 ng g−1 Hg on 27 November 2007, squares 500 ng g−1 Hg on 6December 2007, triangles 500 ng g−1 Hg on 5 March 2008

380 L. Yang, R. E. Sturgeon

of 10–15 h. Clearly, f Hg/f Tl obtained from a single measure-ment session in each solution is remarkably constant but, asexpected, variations in f Hg/f Tl occur as the concentration ofHg is changed because of, amongst other variables, matrixeffects in the plasma. Nevertheless, a mass bias correctedvalue of 1.2852±0.0030 (regression fit of data based onone measurement session, combined uncertainty, k=1) for202Hg/200Hg in the 100 ng g−1 solution of SRM 3133obtained using Eqs. 3 and 4 is in good agreement with avalue of 1.2854±0.0018 (combined uncertainty, k=1)determined using a 500 ng g−1 solution of SRM 3133.Day-to-day variations in plasma and mass spectrometerconditions are to be expected and thus f Hg/f Tl in a particularsolution of Hg may vary. Notably, owing to the greatstability of the Neptune system, although 202Hg/200Hg and205Tl/203Tl determined in a 500 ng g−1 solution of SRM3133 during this measurement period (November 2007 toMarch 2008) varied, all data contributed to the same linearrelationship, as shown in Fig. 2, plot B.

Although linear regression of ln(202Hg/200Hg) versus ln(205Tl/203Tl) is capable of generating accurate Hg ratio data,it inherently suffers from poor precision owing to the needto generate the slope and intercept; reducing the uncertaintyassociated with this process can only be achieved byincreasing the number of measurement sessions. A valueof 1.285333±0.000192 (mean and 1SD) for mass biascorrected RHg202=200

T in SRM 3133 was obtained from 40measurement sessions undertaken during the period fromNovember 2007 to early March 2008.

For the determination of the remaining Hg isotope ratiosin SRM 3133, 18 500 ng g−1 replicate solutions wereprepared and each was measured ten times under optimizedexperimental conditions. The averaged value of RHg202=200

T

reported above was used for subsequent internal mass biascorrection (exponential law) of all other Hg isotope ratios:

RHgx=198

T ¼ RHgx=198

m

m198

mx

� �lnRHg202=200

m �lnRHg202=200

Tlnm202�lnm200

: ð5Þ

Isotope ratios of 0.015337±0.000011, 1.68770±0.00054,2.3056±0.0015, 1.3129±0.0013, 2.9634±0.0038, and0.67937±0.0013 (mean and expanded uncertainty, k=2) wereobtained for 196Hg/198Hg, 199Hg/198Hg, 200Hg/198Hg,201Hg/198Hg, 202Hg/198Hg, and 204Hg/198Hg, respectively, inSRM 3133. Accounting for all possible sources of uncertaintyarising from the measurement process is of fundamentalimportance to ensuring traceability of a measurement. Detailsof the means by which standard uncertainties were estimatedfor all measurement parameters are beyond the scope of thisstudy, although the calculation procedures were in accordwith recommended international procedures [28, 29]. It is tobe noted that the certified value of 2.38714 for 205Tl/203Tlwas used and its associated uncertainty included in thecalculation of the expanded uncertainty of the Hg ratios (seeTable 3 for the example of 196Hg/198Hg). The results obtainedin this study show some disagreement with those reported byBlum and Bergquist [11], i.e., 0.0153685±0.0000092,1.68721±0.00007, 2.30468±0.00013, 1.31209±0.00010,2.96141±0.00031, and 0.68012±0.00019 (mean and twostandard deviations, 2SD) for 196Hg/198Hg, 199Hg/198Hg,200Hg/198Hg, 201Hg/198Hg, 202Hg/198Hg, and 204Hg/198Hg,respectively, for this SRM. Note that these authors [11] usedTl for mass bias correction but assumed f Hg = f Tl.Additionally, no contribution to uncertainty from the certified205Tl/203Tl ratio was included in their reported Hg ratios. Inan attempt to determine how much of an impact this approach

Table 2 Hg isotope fractionation arising from derivatization with NaBEt4

Sample δ 199/198Hg (‰) δ 200/198Hg (‰) δ 201/198Hg (‰) δ 202/198Hg (‰) δ 204/198Hg (‰)

1 (1 s, n=10) 0.016±0.030 0.048±0.063 0.099±0.069 0.129±0.094 0.21±0.142 (1 s, n=10) 0.078±0.027 0.166±0.038 0.272±0.069 0.325±0.082 0.53±0.123 (1 s, n=10) 0.160±0.050 0.326±0.061 0.480±0.091 0.644±0.092 1.09±0.123 (1 s, n=10) 0.458±0.037 0.947±0.027 1.42±0.11 1.844±0.090 2.93±0.124 (1 s, n=10) 0.468±0.054 0.852±0.049 1.296±0.073 1.707±0.072 2.51±0.105 (1 s, n=10) 0.558±0.050 1.125±0.038 1.696±0.049 2.283±0.075 3.47±0.106 (1 s, n=10) 0.89±0.12 1.84±0.10 2.70±0.14 3.596±0.092 5.51±0.13Sample δ′ 199/198Hg (‰) δ′ 200/198Hg (‰) δ′ 201/198Hg (‰) δ′ 202/198Hg (‰) δ′ 204/198Hg (‰)1 (1 s, n=10) 0.016±0.030 0.048±0.063 0.099±0.069 0.129±0.094 0.21±0.142 (1 s, n=10) 0.078±0.027 0.166±0.038 0.272±0.069 0.325±0.082 0.53±0.123 (1 s, n=10) 0.160±0.050 0.326±0.061 0.480±0.091 0.644±0.092 1.09±0.123 (1 s, n=10) 0.458±0.037 0.946±0.027 1.42±0.11 1.842±0.090 2.93±0.124 (1 s, n=10) 0.468±0.054 0.852±0.049 1.296±0.073 1.707±0.072 2.51±0.105 (1 s, n=10) 0.558±0.050 1.125±0.038 1.695±0.049 2.289±0.075 3.46±0.106 (1 s, n=10) 0.89±0.12 1.81±0.10 2.70±0.14 3.590±0.092 5.50±0.13

All data are relative to SRM 3133.

Isotopic fractionation of mercury induced by reduction and ethylation 381

would have on our data, our 196Hg/198Hg ratio was selectedfor recalculation assuming f Hg = f Tl of the combineduncertainty evaluated. As evident in Table 4, even accountingfor these discrepancies in data manipulation, the two sets ofratios of 0.0152992±0.0000042 (k=2) using Tl for mass biascorrection assuming f Hg = f Tl and 0.015337±0.000011 (k=2)using the mass bias correction method presented in this studyremain statistically disparate.

Isotope fractionation

As noted earlier, for many applications of stable isotopes,such as provenance and earth science studies, it is generallyof more interest to know the relative differences in isotopicratios between samples than to know their absolute isotopicamount ratios. As such, results are usually reported in δnotation relative to an agreed-upon standard (which for

these studies is SRM 3133, as suggested by Blum andBergquist [11]) with use of Eq. 6:

sample

/198

/198/198

SRM 3133

Hg 1x

xx

R

Rδ = − ×1000 ‰

ð6Þwhere x designates the isotope of interest. For the Hgsystem, both mass-dependent fractionation (MDF) andmass-independent fractionation (MIF) have been reported[5–18]. To evaluate whether isotope fractionation isdependent or independent, the following kinetic mass-dependent relationship [13, 30] was examined:

d0 i=198Hgd0 202=198Hg

¼ln m198

mi

� �

ln m198m202

� � ; ð7Þ

Table 3 Uncertainty components for 196Hg/198Hg in NIST SRM 3133 using the mass bias correction method developed

Parameter Typical value u(xi)@f@xi

� �@f@xi

� �u xið Þ

Uncertainty components for 202Hg/200Hg in SRM 3133Slope 1.066799 0.006719 1.285569 0.008637Intercept −0.677015893 0.006079 1.118568 0.006800RTl205=203T , SRM 997 2.387140 0.000721 0.574513 0.000414

RHg202=200

Ti, one session run 1.285569 0.011000

Final RHg202=200

T in SRM 3133 1.285333 0.001739Uncertainty components for 196Hg/198Hg in SRM 3133R196=198m 0.0150935 0.0000072 1.0159129 0.0000074

m198 197.966752 0.000003 0.0001204 0.00000000036m196 195.965814 0.000004 −0.0001216 −0.00000000049R202=200m 1.3053900 0.000007 0.0119769 0.00000008671

Final RHg202=200

T in SRM 3133 1.285333 0.001739 −0.0121638 −0.00002115665m202 201.970625 0.000003 −0.0002421 −0.00000000073m200 199.968309 0.000003 0.0002421 0.00000000073R196=198ni

, one sample 0.0153337 0.0000224Final R196=198

n in SRM 3133 0.015337 0.0000053

Equations 3 and 4 were used to derive the mass bias corrected value for RHg202=200

Tiobtained in each session of measurements. The final value of

202 Hg/200 Hg in NIST SRM 3133, used for mass bias correction of all other Hg isotope ratios using Eq. 5, was obtained from the mean value of40 such measurement sessions. Uncertainty associated with this value can be further reduced if the number of measurements is further increased.

Table 4 Uncertainty components for 196Hg/198Hg in NIST SRM 3133 using Tl mass bias correction, assuming f Hg = f Tl

Parameter Typical value u(xi)@f@xi

� �@f@xi

� �u xið Þ

R196=198m 0.0150935 0.0000072 1.0136255 0.0000073

m198 197.966752 0.000003 0.0001030 0.00000000031m196 195.965814 0.000004 −0.0001040 −0.00000000042R205=203m 2.4185594 0.000081 0.0065470 0.00000053317

RTl205=203T in SRM 997 2.387140 0.000721 −0.0066332 −0.00000478538

m205 204.974412 0.000003 −0.0002071 −0.00000000062m203 202.972329 0.000003 0.0002071 0.00000000062R196=198ni , one sample 0.0152992 0.0000088

Final R196=198n in SRM 3133 0.0152992 0.0000021

382 L. Yang, R. E. Sturgeon

where δ′ i/198Hg is defined as 1000� ln Ri=198Hgsample

.�R

i=198Hgstd Þ. When isotope fractionation is mass-dependent,

Eq. 7 gives a value 1; for MIF, Eq. 7 gives values of morethan 1 or less than 1. It is also worth noting that ifR

i=198Hgsample

.R

i=198Hgstd � 1, then di=198Hg � d0 i=198Hg. This ap-

proximation is usually met in practice since measured δvalues are generally in the range 0–10‰. As shown inTable 2, values in the range 0–5.5‰ for δ i/198Hg (largestfractionation range observed in this study) obtained fromexperiments using NaBEt4 are almost identical to those forδ′ i/198Hg. For these Hg fractionation studies, mass biascorrection was achieved with use of both standard-sample-standard bracketing and internal normalization, i.e., thedetermined value of 1.285333±0.000192 (mean and 1SD)for RHg202=200

n was used to implement mass bias correction on205Tl/203Tl in two adjacent SRM 3133 standards. Theiraverage value was then used to apply exponential mass biascorrections to Hg isotope ratios in the samples. This issimilar to our earlier approach used for determination of Srratios with use of Zr [31].

Reduction of Hg species to Hg0 vapor is an importantpathway for the transfer of Hg from aquatic systems to theatmosphere [14]. It is also of interest to examine the impactof laboratory-induced reduction/volatilization, as this is acommon approach taken for the analytical quantitation ofHg in samples and is widely used to enhance the sensitivityof measurement with ICP-MS instrumentation for the studyof possible fractionation in samples. In this study, threecommon means of reduction of Hg(II) to Hg0 vapor weretested, i.e., SnCl2, NaBH4, and UV photolysis, to examinepossible Hg fractionation arising from these processes.

Isotope fractionation during reduction of Hg(II) to Hgo

As shown in Figs. 3 and 4, significant fractionation of Hgisotopes occurs during reduction of Hg(II) to Hg0 with useof SnCl2, NaBH4, and UV photolysis. Experiments weredesigned to induce reduction followed by quantitativeremoval of the Hg0 from solution in the range 10–90% ofthe original concentration. This facilitated detection of

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

-0.50 0.00 0.50 1.00 1.50 2.00

-0.50 0.00 0.50 1.00 1.50 2.00

δ' x/

198 H

g (‰

)

Bδ204/198Hg (‰)

δ201/198Hg (‰)

∆ δ200/198Hg (‰)

δ199/198Hg (‰)

δ' 202/198Hg (‰), SnCl2 reduction

δ' x/

198 H

g (‰

)A

δ204/198Hg (‰)

δ201/198Hg (‰)

∆ δ200/198Hg (‰)

δ199/198Hg (‰)

δ' 202/198Hg (‰), UV irradiation

Fig. 3 Mass dependence of fractionation for all samples from Hgfractionation experiments using UV photolysis and SnCl2 reduction.Solid lines are the theoretically predicted mass-dependent fractionation(MDF) based on δ′ 202/198Hg using Eq. 7. Error bars are one standarddeviation of the mean of five measurements of each sample

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

-0.50 0.00 0.50 1.00 1.50 2.00

-0.50

0.50

1.50

2.50

3.50

4.50

5.50

6.50

-0.50 0.50 1.50 2.50 3.50 4.50

δ' x/

198 H

g (‰

)

B

δ' 202/198Hg (‰), NaBEt4 alkylation

δ' x/

198 H

g (‰

)

A

δ' 202/198Hg (‰), NaBH4 reduction

δ204/198Hg (‰)

δ201/198Hg (‰)

∆ δ200/198Hg (‰)

δ199/198Hg (‰)

δ204/198Hg (‰)

δ201/198Hg (‰)

∆ δ200/198Hg (‰)

δ199/198Hg (‰)

Fig. 4 Mass dependence of fractionation for all samples from Hgfractionation experiments using NaBH4 and NaBEt4. Solid lines arethe theoretically predicted MDF based on δ′ 202/198Hg using Eq. 7.Error bars are one standard deviation of the mean of five measure-ments of each sample

Isotopic fractionation of mercury induced by reduction and ethylation 383

isotope fractionation while retaining sufficient residual Hg(II) for reliable measurements. Overall, the heavier isotopesof Hg were enriched in the remaining solutions. Controlexperiments were conducted under identical conditions,with the exception that no SnCl2, NaBH4, or UV field wasadded/applied to the sample in order to eliminate possibleexperimental artifacts. No detectable loss of Hg from thetest solution was observed, indicating that simply spargingthe sample with Ar does not induce any change in Hgconcentration nor does it induce any detectable fraction-ation. The highest δ values of 1.1–1.3‰ for δ 202/198Hg inthe test samples relative to SRM 3133 were obtained duringUV photolysis experiments after 85–90% loss of theoriginal Hg concentration, as shown in Fig. 5.

For the even isotopes, strong MDF was observed duringreduction of Hg(II) using SnCl2, NaBH4, and UV photol-ysis. For the odd isotopes, the values of δ′ 199/198Hg and δ′201/198Hg deviate slightly from the theoretically predictedMDF lines shown in Figs. 3 and 4a, suggesting minor MIFmay occur during the processes of reduction by SnCl2 andNaBH4 and during UV photolysis. However, the deviationof δ′ 199/198Hg and δ′ 201/198Hg from the theoreticallypredicted MDF lines does not exceed 2SD, indicating MIFis insignificant. To further confirm this, Δ values of199Hg/198Hg and 201Hg/198Hg in the range −0.16±0.34(2SD) to 0.05±0.14 (2SD) were derived from the above-mentioned experiments using Eqs. 8a and 8b, which havebeen successfully used by others [11, 14] when δ is below10‰, confirming MIF is not significant:

Δ199 Hg ¼ d199 Hg� d202 Hg�0:2520� � ð8aÞ

and

Δ201Hg ¼ d201Hg� d202Hg� 0:7520� �

: ð8bÞAlthough a recent study [14] on the photoreduction of

Hg(II) by natural sunlight in the presence of dissolvedorganic carbon concluded that MIF occurred for oddisotopes of Hg (up to 2.5‰), our data generated using

photolysis of solutions spiked with formic acid do notreplicate this finding. It is possible that the fundamentalmechanisms underlying the photochemical reactions are notidentical, in that Bergquist and Blum [14] used fulvic acid,whereas formic acid was used in this study as the source ofdissolved organic carbon. In addition, the irradiationconditions were also different in that natural sunlight [14]may induce different characteristics compared with the UVsource used for this study. Considering that the magneticisotope effect is predicted to result in the most pronouncedMIF for odd isotopes and that radical reactions arefrequently involved in photolysis, differences in themechanisms of the ensuing reactions may account for thedegree of MIF observed in the two systems.

The three Hg reduction experiments described abovemay be considered Rayleigh systems and thus a fraction-ation factor, αaq-g, which is the ratio of isotope ratio in theaqueous phase to that in the gas phase can be obtainedusing the Rayleigh fractionation law (Eq. 9) [13]:

1000þ d1000þ di

¼ f1

aaq�g�1

� �R ; ð9Þ

where δ corresponds to the Hg isotope ratio in theremaining solution, δi corresponds to the initial value forthis isotope ratio, when fR=1, and fR is the mass fraction ofHg in the remaining solution. The fractionation factor, αaq-g,can be calculated from the slope of a linear regression ofthe natural logarithm of Eq. 9:

ln1000þ d1000þ di

� �¼ 1

aag�g� 1

� �ln fR: ð10Þ

Fractionation factors of 1.00055±0.00005, 1.00043±0.00004, and 1.00049±0.00004 (expanded uncertainty, k=2) were obtained 202Hg/198Hg reduction experiments basedon UV, SnCl2, and NaBH4. The fractionation factorobtained in this study with SnCl2 is in general agreementwith results of 1.00044±0.00003 and 1.00047±0.00002(two standard errors) reported by Zheng et al. [13] for theirreduction/volatilization experiments using SnCl2

Isotope fractionation during ethylation of Hg(II) to HgEt2

Recent studies [15, 16] have concluded that biologicalprocesses involved in the methylation of Hg(II) byorganisms in aquatic ecosystems mediate MIF of Hgisotopes, whereas only MDF has been reported in lakesediments. MIF is influenced by the effects of the nuclearspin and nuclear volume of Hg, as well as the penetration ofthe inner electron shells of Hg by valence electrons fromligands owing to the high nuclear charge of Hg and itsinefficient shielding by d and f electrons [16, 32]. Although

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

0.000.100.200.300.400.500.600.700.800.901.00

Fraction of remaining Hg

δ' 20

4/19

8 Hg

(‰)

Fig. 5 Development of Hg fractionation as the extent of reductionincreases during UV photolysis experiments in 5% formic acid

384 L. Yang, R. E. Sturgeon

the characteristics of CH3− and CH3CH2− functionalgroups are not completely identical, they should havesimilar effects when reacting with Hg. However, resultsobtained from laboratory experiments during progressiveethylation of Hg(II) show only very strong MDF of Hgisotopes, as evident in Fig. 4b. These experiments weredesigned to induce 10–90% ethylation and quantitativeremoval (volatilization) of the alkylated fraction from thetest solution. Noticeably higher fractionation of up to 3.59±0.09‰ (1SD, n=10) for δ 202/198Hg relative to SRM 3133was obtained in the solution after loss of 85% of the initialHg(II) concentration compared with MDF induced byreduction.

The fractionation factor calculated for ethylation was1.0012±0.0002 (expanded uncertainty, k=2), larger thanthat arising from reduction to Hg0 by UV photolysis, SnCl2,and NaBH4. These observations, in addition to resultsreported by others [10, 13], may provide an additional toolfor distinguishing biogeochemical processes involving Hg.

Conclusion

An accurate and precise method has been developed formeasurement of isotope ratios in NIST SRM 3133 by MC-ICP-MS. Although the mass bias correction factors for Tland Hg are not identical, fHg/fTl was found to be constantover a measurement session, which permitted mass biascorrection of RHg202=200

n to be achieved. A value of 1.285333±0.000192 (mean and 1SD) for mass bias corrected RHg202=200

n

in SRM 3133 was obtained from 40 measurement sessionsundertaken over a 5-month period. This value was thenused for internal mass bias correction of all Hg isotoperatios in this SRM. Isotope ratios of 0.015337±0.000011,1.68770±0.00054, 2.3056±0.0015, 1.3129±0.0013,2.9634±0.0038, and 0.67937±0.0013 (expanded uncer-tainty, k=2) were obtained for 196Hg/198Hg, 199Hg/198Hg,200Hg/198Hg, 201Hg/198Hg, 202Hg/198Hg, and 204Hg/198Hg,respectively.

Significant isotope fractionation was detected duringprogressive chemical and photolytic reduction of Hg(II) toHg0 with use of SnCl2, NaBH4, and UV photolysis as well asduring progressive ethylation of Hg(II). MDF was found tobe dominant in all reactions studied with relatively smallfractionation factors of 1.00043 to 1.00055 for the reductionexperiments and a larger fractionation factor of 1.0012 for theethylation process. MIF for odd Hg isotopes was found to beinsignificant during reduction of Hg(II) with SnCl2 andNaBH4. Additionally, unlike MIF arising from biologicalprocesses [15, 16], chemical ethylation with NaBEt4 did notinduce MIF in this study. The present work has provided newdata for Hg fractionation arising from reduction of Hg(II) toHg0 by NaBH4 and UV photolysis as well as during

ethylation by reaction with NaBEt4. These results may beof importance to our further understanding of a wide varietyof chemical and biological processes in nature.

References

1. Craig PJ (1986) Organometallic compounds in the environment:principles and reactions. Longman, Harlow

2. Johnson CM, Beard BL, Albaréde F (2004) In: Johnson CM, BeardBL, Albaréde F (eds) Geochemistry of non-traditional stable isotopes.Reviews in mineralogy and geochemistry, vol 55. MineralogicalSociety of America and the Geochemical Society, pp 1–24

3. Evans RD, Hintelmann H, Dillon PJ (2001) J Anal At Spectrom16:1064–1069

4. Lauretta DS, Klaue B, Blum JD Buseck PR (2001) GeochimCosmochim Acta 65:2807–2818

5. H Hintelmann, S Lu (2003) Analyst 128:635–6396. Jackson TA, Muir D, Vincent WF (2004) Environ Sci Technol

38:2813–28217. Smith CN, Kesler SE, Klaue B, Blum JD (2005) Geology 33:825–

8288. Xie Q, Lie S, Evans D, Dillon P, Hintelmann HJ (2005) Anal At

Spectrom 20:515–5229. Ridley WI, Stetson SJ (2006) Appl Geochem 21:1889–1899

10. Kritee K, Blum JD, Johnson MW, Bergquist BA, Barkay T (2007)Environ Sci Technol 41:1889–1895

11. Blum JD, Bergquist BA (2007) Anal Bioanal Chem 388:353–35912. Malinovsky D, Sturgeon RE, Yang L (2008) Anal Chem 80:2548–

255513. Zheng W, Foucher D, Hintelmann H (2007) J Anal At Spectrom

22:1097–110414. Bergquist BA, Blum JD (2007) Science 318:417–42015. Jackson TA (2006) Geochim Cosmochim Acta 70:A284–A28416. Jackson TA, Whittle DM, Evans MS, Muir DCG (2008) Appl

Geochem 23:547–57117. Sonke JE, Zambardi T, Toutain JP (2008) J Anal At Spectrom

23:569–57318. Epov VN, Rodriguez-Gonzalez P, Sonke JE, Tessier E, Amoureux

D, Bourgoin LM, Donard OFX (2008) Anal Chem 80:3530–353819. Dzurko M, Foucher D, Hintelmann H (2008) Anal Bioanal Chem

doi:10.1007/s00216-00802165-y20. Galer SJG (1999) Chem Geol 157:255–27421. Dědina J, Tsalev DL (1995) Hydride generation atomic absorption

spectrometry. Wiley, Chichester22. Sturgeon RE, Mester Z (2002) Appl Spectrosc 56:202A–213A23. Sturgeon RE, Willie SN, Mester Z (2006) Anal At Spectrom

21:263–26524. Russell WA, Papantastassiou DA, Tombrello TA (1978) Geochim

Cosmochim Acta 42:1075–109025. Maréchal N, Télouk P, Albarède F (1999) Chem Geol 156:251–27326. Albarède F, Télouk P, Blichert-Toft J, Boyet M, Agranier A,

Nelson B (2004) Geochim Cosmochim Acta 68:2725–274427. White WM, Albarède F, Télouk P (2000) Chem Geol 167:257–27028. International Organisation for Standardisation (1995) Guide to the

expression of uncertainty in measurement. International Organi-sation for Standardisation, Geneva

29. Ellison SLR, Rosslein M, Williams A (2001) Quantifyinguncertainty in analytical measurements, 2nd edn. EURACHEM/CITAC, pp 87–94

30. Young ED, Galy A, Nagahara H (2002) Geochim CosmochimActa 66:1095–1104

31. Yang L, Peter C, Panne U, Sturgeon RE (2008) J Anal AtSpectrom (in press)

32. Schauble EA (2007) Geochim Cosmochim Acta 71:2170–2189

Isotopic fractionation of mercury induced by reduction and ethylation 385