Controls on the Valence Species of Arsenic in Tobacco Smoke:XANES Investigation with Implications for Health and RegulationRobert C. J. Campbell,† William E. Stephens,* and Adrian A. Finch

Department of Earth & Environmental Sciences, University of St. Andrews, Irvine Building, North Street, St. Andrews, Fife, KY169AL, U.K.

Kalotina Geraki

Diamond Light Source, Harwell Science and Innovation Campus, Didcot, Oxfordshire, OX11 0DE, U.K.

*S Supporting Information

ABSTRACT: Arsenic (As) is one of four metals/metalloids intobacco being considered for regulation. In vitro toxicologicalresponse to As varies substantially, determined primarily byvalence and compound speciation, and inorganic arsenite(As(III)) compounds are the most toxic to humans. This studyuses X-ray absorption near edge structure (XANES) todetermine valence states of As from the tobacco plant to thecrucial combustion stage that creates respirable smoke.Samples studied include cultivated plants (some burdenedwith additional As), reference standards, and commercialproducts, along with smoke condensate and ash from thesesamples. The relative contributions of As(III) and As(V) tothe XANES spectra are analyzed, and a consistent pattern ofredox changes emerges. Tobacco leaf and manufacturedproducts tend to be dominated by As(V) whereas combustion produces respirable smoke invariably in As(III) form and ashinvariably as As(V). The valence state of precursor tobacco is not a controlling factor because all the As mobilized in smoke isreduced during combustion. This study concludes that tobacco combustion exposes smokers to potentially the most toxic formsof arsenic, and this exposure is magnified in regions where arsenic is present in tobacco crops at relatively high concentrations.

■ INTRODUCTION

According to the World Health Organisation (WHO), tobaccosmoke contributes to the deaths of about half its users1 yet theindividual and combined effects of the components of smokeresponsible for the various smoking-related diseases are notfully understood. A recent EU-sponsored review identified 98compounds in tobacco smoke that exceed “thresholds oftoxicological concern”2 among which are seven metals ormetalloids including arsenic (As). The US Food and DrugAdministration, recently given responsibility for the regulationof tobacco products, included arsenic in a slightly differentgroup of elements in its list of “harmful and potentially harmfulconstituents in tobacco products and tobacco smoke”.3 TheTobacco Product Regulation Study Group (TobReg), an expertpanel advising the WHO, reviewed the available scientificevidence of harm caused by metals and metalloids in tobaccosmoke before recommending that “tobacco purchased fromeach new agricultural source [be tested] for levels of arsenic,cadmium, lead and nickel” and that tobacco blends offered forsale be tested for concentrations of these elements.4 Thesereviews invariably identify arsenic in tobacco as hazardous;however, due to the paucity of available studies none of these

reviews takes the speciation of arsenic in smoke into account,whether it is transferred in elemental or compound form,inorganic or organic, and the valence state(s) (trivalent/pentavalent) on combustion and transfer to the respiratorytract. These factors are fundamental to assessing the hazard,given that a large body of toxicological research identifiesinorganic As(III) species as highly toxic to humans based oningestion studies.5 Exposure to arsenic through inhalation is lesswell studied, and the effects of species on the risk to humans isnot yet known, although there is evidence that As(III) causestumors in rats. 5

The concentration of As in tobacco ranges from less than 1μg g−1 to a few μg g−1, while up to a few tens of nanograms percigarette are present in mainstream smoke.1 Estimates of masstransference of As from tobacco to smoke fall in the range of7−18%.6 Such levels could pose a risk to smokers’ health if As

Received: September 3, 2013Revised: November 25, 2013Accepted: February 12, 2014Published: February 12, 2014

Article

pubs.acs.org/est

© 2014 American Chemical Society 3449 dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−3456

is dominantly present at the point of human exposure in itsinorganic and/or reduced state.The aim of the present study is to determine the speciation

of As delivered to the smoker, to inform those assessing therisks of individual smoke components. More specifically theobjective is to characterize the valence of As along its pathwayfrom the harvested plant, through curing to the smokingproduct, and in particular through the combustion event duringsmoking that creates the respirable aerosol. The analyticalapproach uses X-ray absorption near edge structure (XANES)spectra obtained using synchrotron radiation. Unlike HPLC-ICPMS and most other speciation techniques XANES can beapplied in situ without requiring chemical extraction and theassociated risk of modifying speciation can be monitored.Past attempts to achieve this aim using XANES have been

largely thwarted by insufficient sensitivity of availablesynchrotrons to characterize the valence state of As in smokeproducts at low As concentrations. Liu and colleagues in seriesof papers linked their XANES results on tobacco and smokecondensate with speciation studies using HPLC-ICPMS.7−9

The weak As response in some of their XANES spectra led touncertain conclusions concerning the valence of As in smoke.Notwithstanding, the impression given by the most recentliterature is that the reduced and potentially more toxic formsof As are minor and transient species in tobacco smokeproducts, and that any As(III) present is rapidly oxidized soonafter generation to As(V) forms.8 If this is correct, then As intobacco may not be sufficiently hazardous to engage theinterest of regulators; however, there is currently sufficientuncertainty to warrant the collection of higher resolutionXANES spectra on a much wider variety of samples.Here we adopt a 2-fold approach to overcoming the technical

difficulties besetting earlier studies. First, use of a thirdgeneration synchrotron (Diamond, the UK’s national facility)offers a considerably brighter source of X-rays generating muchstronger signals at lower concentrations of As than earlierstudies. Second, tobacco burdened with elevated concentrationsof As has been successfully cultivated for use in smokingexperiments, taking care to ensure verisimilitude with productsavailable for public consumption.

■ EXPERIMENTAL SECTION

Synchrotron Analysis. All standards and samples wereanalyzed on beamline I18 of the Diamond third generationsynchrotron facility10 at Didcot, Oxfordshire, UK (www.diamond.ac.uk). Diamond is a 3 GeV synchrotron, andmeasurements were taken with a typical beam current of 250mA. Beamline I18 receives X-rays from an insertion device, andthe beamline employs a Si(111) monochromator, the radiationfrom which was focused by Kirkpatrick−Baez mirrors into a ∼3μm diameter spot. The surface of the sample was 45° to theincoming beam, and the fluorescence detector was perpendic-ular to the beam (see Supporting Information). XAS data werecollected on the As K-edge to include the near edge (XANES)region only. The response of the sample was measured as As K-α X-ray fluorescence using a Vortex four-channel X-ray detectorwith dedicated fast-counting electronics, and the final responseis the sum of all channels. The monochromator energycalibration was checked with a Au foil (11 919 eV), and themeasurement of an As2O5 standard was repeated at thebeginning of every session to check for reproducibility. Thefinal data are the averages of up to 50 cycles (depending on

concentration), and each cycle typically took 5−9 min toacquire.The samples were leaf fragments, pelleted powders (ground

leaf, shredded leaf, and ground ash), condensates, and filters.These were presented to the beam on Kapton film oroccasionally on high-purity quartz slides. Analyses of As(III)standards at room temperature showed modification of thespectra with time (20−30 min), inferred to be due to X-rayinduced oxidation; therefore, experiments were performed atcryogenic temperatures. Initially this was performed in acontinuous flow nitrogen cryostat (Oxford Instruments Micro-statHiResII) with the sample mounted in vacuum and cooledby conduction and a Kapton window for the X-rays. Undersuch conditions, no sample modification of As(III) standardswas observed.11 Similarly, smoke condensate with very low Asconcentrations showed no observable changes in edge positionwith exposure to the beam (experiments described later). Wealso compared data from the continuous flow cryostat withresults of analysis under a Cryojet (Oxford Instruments) inwhich a flowing jet of cold nitrogen gas close to 77 K is directedat the exposed sample surface. The temperature of the sampledropped to ∼90 K under those conditions, and the absence offrosting in the analysis area indicated that the continuousnitrogen flow kept air away from the analysis area. Analyses ofsamples using the cryojet provided for far faster sample changesand, as As(III) standards showed no sample modification, thecryojet became our preferred method. Analysis of a naturaldiamond (C) procedural blank showed no evidence for Asresponses from scattered X-rays coming from the beamlinetable and experimental hutch. The background subtraction andnormalization of data for XANES was performed usingATHENA.12 The edge position is determined using ATHENAfrom the maximum of the first derivative of the raw data.X-ray absorption edge profiles of As were obtained using

XANES analyses of tobacco leaf (As burdened and unbur-dened), manufactured cigarettes, and smoked products (ash,filter, and particulates). Details of reference and samplematerials, and descriptions of experimental conditions includingcultivation and machine smoking, are available in SupportingInformation. In summary, As-burdened tobacco was grownfrom KT209 tobacco seeds using seeding compost and variousconcentrations of sodium arsenate and sodium arsenite insolution, added together and separately. XANES spectra wereobtained from inorganic compounds for edge energy calibration(arsenic(III) oxide, arsenic(V) oxide, sodium arsenite, sodiumarsenate, and arsenopyrite) and from tobacco leaf grown fromKT209 seed with and without As burdening. Those selected forXANES investigation were unburdened leaf samples RCS115(in vivo), RCS123 (in vivo), RCS102 (0.7 μg/g As), andRCS133 (0.3 μg/g As), burdened leaf RCS108 (10.3 μg/g As),RCS120 (8.6 μg/g As), RCS123 (13.5 μg/g As), RCS125 (246μg/g As), and RCS132 (10.1 μg/g As), international referencestandards (CTA-VTL-2 (0.97 μg/g As) and 1R4F (0.31 μg/gAs), and commercial tobacco products (STA336 (0.7 μg/g As)and STA486 (3.3 μg/g As)). Note that all concentrations arequoted as dry mass.

■ RESULTSCultivation of As-Burdened Tobacco Plants. Glass-

house cultivation of As-burdened tobacco leaf resulted in up to3 orders of magnitude increases in As concentration. The driedleaf of unburdened control sample of KT209 contains 0.7 μg/gAs whereas intermediate burdening (RCS132) created tobacco

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563450

leaf containing 5−15 μg/g As. More substantial burdening ledto tobacco leaves with 100−250 μg/g As, although there weremany instances where plants failed to develop under burdeningconditions, particularly when As was added solely as As(V).Only outwardly healthy plants were preserved through theextent of the growing period (typically 120 days). XANESspectra were collected across this range of burdened tobacco,but for smoking experiments sampling was confined to thelower levels of burdening.

XANES Spectra of Standards (Figure 1). Edge energy,defined as the maximum in the first derivative of the data, lies∼11 868 eV for As(III) from arsenic trioxide (As2O3) andsodium arsenite (NaAsO2), whereas the As(V) absorption edgelies ∼11 873 eV in As pentoxide (As2O5) and sodium arsenate(Na2HAsO4.7(H2O)) (Figure 1). The edge for arsenopyrite(FeAsS, Apy in Figure 1) is significantly lower at 11 866 eV, inreasonable agreement with interpretations of this feature asindicating the dominance of As(I) in arsenopyrite13,14

XANES Spectra of Fresh and Cured Tobacco Leaf(Figure 2A). Control sample RCS115 was inserted as freshlycut green leaf into the beam, and its profile representsunburdened live leaf. Its edge energy shows As(III) to bedominant with the hint of a shoulder consistent with minorAs(V). RCS102 is a similar sample that has additionally beenfermented, and it shows the development of a more prominentfeature attributable to As(V). Note that these unburdenedsamples have <1 μg/g As when dried, yet credible XANESspectra can be generated using a third generation synchrotron.Burdened sample RCS123IV was inserted as freshly cut greenleaf into the beam, and its profile represents burdened live leaf.

Its edge energy shows As(V) to be dominant with the hint of ashoulder consistent with minor As(III).The spectra for RCS120, RCS123, RCS125, and RCS108

represent leaf grown in As-burdened conditions. In all cases adistinct As(III) edge is present and is even dominant in the caseof RCS120. Of these burdened samples only RCS123 was notfermented, and it is noteworthy that As(V) dominates itsXANES spectrum. There is no systematic response to Asburden; for example, RCS108 was cultivated with As(III), yetAs(III) in the resultant tobacco is less prominent than As(V)(Figure 2A, spectrum 11). Similarly, there is little correlationbetween valence balance and total As uptake. RCS125 is themost As-rich sample, and As(III) and As(V) appear to beapproximately equivalent in its XANES spectrum. Lessburdened samples such as RCS123 (13.5 μg/g) and RCS120(8.6 μg/g) show very different balances. It should be noted thatsome of the peaks (particularly RCS120 and RCS1250) occur

Figure 1. XANES spectra for inorganic reference compounds.Compound abbreviations and spectral interpretations are given inTable 1. Gray bands are centered on the first derivative maxima inAs2O3 and As2O5, taken to indicate the edge energies of inorganicallybound As(III) and As(V), respectively, and are reproduced in all otherXANES spectra for reference. Error is typically ±0.25 eV.

Figure 2. XANES spectra for (A) KT209 control tobacco leaf andKT209 grown with various levels of total As burdening (as annotated),and (B) a selection of cured and processed tobaccos from diversesources. Sample abbreviations and spectral interpretations for eachnumbered spectrum are given in Table 1.

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563451

at higher energies than associated with the As(V) edge in theinorganic reference compounds (Figure 2A), and this shift mostprobably reflects different organic molecular environments.In summary, these spectra indicate that fresh leaf may be

dominated either by As(III) or As(V), and burdening does notlead to a predictable balance between the valence states.Fermentation leads to spectra with both As(III) and As(V);again the relative valence balance is not predictable,XANES Spectra of Cigarette Tobacco (Figure 2B).

As(V) is the dominant valence state in four cigarette tobaccos.Smoking machine reference standard 1R4F, trace elementstandard CTA-VTL-2, and US popular brand STA336 with alow concentration of As all indicate strong edges correspondingwith As(V) with only slight indicators of As(III). A Chinesebrand with elevated As (STA486, 3.3 μg/g As) has a greaterfraction of As(III) although As(V) still dominates. The onlysample to depart from this general pattern is RCS132, tobaccogrown under burdened conditions, fermented, cased withpropylene glycol, and then shredded. This sample has 10.1 μg/g As and shows As(III) to be dominant although a shoulderattributable to As(V) is discernible. These data show that thesecommercial tobacco products are dominated by the oxidizedform with only minor As(III), unless the sample has beengrown in an As-rich environment.XANES Spectra of Smoke Condensate and Ash

(Figure 3). Smoke condensate collected around perforations(as described in Experimental Section) include numerous As-

rich points even in samples with very low bulk Asconcentrations. The edge energy of the control tobacco(RCS133 with 0.7 μg/g As), the medium burdened leaf(RCS132 with 10.1 μg/g As), and the reference standard 1R4Fare essentially identical with peak positions indicative of As(III)(11 868 ± 1 eV). RCS132 is a much smoother profile because ithas a higher concentration. All three samples were created andtransferred at room temperature and analyzed under the lowtemperature conditions of the cryojet.XANES spectra for ash generated from the same tobaccos are

also consistent, composed predominantly of oxidized As(V)species (Figure 3). The edge energies are 11 872 eV (controlsample RCS102), 11 875 eV (As-burdened samples RCS132),and 11 872 eV (1R4F reference standard), within the As(V)edge energy defined by inorganic standards (11 873 eV). Theseash samples contain little or no As(III), and we infer that thecombustion processes producing the ash effectively oxidize allthe unmobilised As.

XANES Spectra of Aging Experiments on TobaccoCondensate (Figure 4). Much has been made in earlier

papers about the transient character of the As(III) componentof tobacco smoke particles.7,8 In the present study no change inoxidation state was observed in room temperature experimentson smoke condensate carried out in a normal atmosphere sometens of minutes after completing experiments using thesmoking machine (Figure 4). If As(III) is the species posingthe greater risk in smoke, then the role of smoke aging could beimportant in reducing this risk by oxidation. This promptedfurther investigations into potential changes in As valence stateby repeating the experiments on smoke condensate withdifferent time lags and storage conditions prior to acquiringXANES spectra. Although these experiments do not replicatereal smoking conditions, particularly within the lungs, anyevidence for rapid and complete oxidation of As couldpotentially be misused in claims that the risk from arsenicexposure is negligible.

Figure 3. XANES spectra for smoke condensate from the 1R4Freference standard, KT209 control and one As-burdened tobacco, ablank filter butt, and ash collected from a selection of experiments.Sample abbreviations and spectral interpretations for each numberedspectrum are given in Table 1.

Figure 4. XANES spectra for smoke condensate under differentconditions of aging (lag time after smoking, with or without the cryojetduring experiment). Sample abbreviations and spectral interpretationsfor each numbered spectrum are given in Table 1.

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563452

Arsenic-burdened tobacco RCS132 was machine smokedwith the filter pad encased in dry ice, and the assembly was kepttogether after the experiment while relocating to thesynchrotron beam where it was transferred to the cryojet.The sample was prevented from warming up to roomtemperature to limit alterations in “electrothermal potential”.8

The XANES spectrum for RCS132 (29 in Figure 4) from thesmoke condensate experiment (Figure 3) is included forcomparison. The spectra from the smoking of one cigarette(26) and three cigarettes (27) are essentially the same, althoughthe data are noisier in the former, given that fewer cigaretteswere smoked to avoid aging effects. No significant shift in edgeenergy reflecting oxidation is observed. The three cigarettesexperiment was repeated and the condensate stored for 30 minat room temperature before insertion into the beam. TheXANES spectrum (28) shows no significant change, and theexperiment fails to demonstrate any oxidation of As(III) withaging. The reference standard 1R4F was smoked, and XANESspectra were collected without (29) and with (30) the cryojetto check the extent of any reduction or oxidation under beamconditions during approximately 30 min of exposure to thebeam.11,15 When compared to unaged spectra, Figure 3, nosuch effect is observed over the time taken to transfer thesample to the beamline (a few tens of minutes).

■ DISCUSSION

The XANES experiments demonstrate that tobacco combus-tion generates smoke that condenses on a filter pad withdominantly reduced As(III) species and ash with entirelyoxidized As(V) species. This outcome is observed regardless ofthe balance of As(III):As(V) in the precursor tobacco.Human exposure to As occurs primarily through ingestion of

contaminated food or water. Typical daily intakes from foodand beverages in nonoccupational settings range between 20and 300 μg/day (the higher figures associated with largeconsumption of seafood). Inhalation may contribute 1 μg/dayin a nonsmoker but estimates for a smoker’s additionalexposure range from about 1 to 10 μg/day.5 Arsenic exposurevia tobacco smoke may be even higher in heavy smokers and/or in parts of the world where high concentrations of As areoften found in tobacco.16 Thus, smoking can be a significantsource of As exposure, although these exposure estimates arebased on total As and do not take account of the potentiallydifferent toxicities of various As species based on valence stateand inorganic and organic compounds. The most recent IARCevaluation of inorganic As compounds implicates arsenictrioxide, As(III), and As(V) in causing cancers of the lung,urinary bladder, and skin and positively associates As and itsinorganic compounds with cancers of the kidney, liver, andprostate.5 The evaluation also recognized that As(III) species

Table 1. List of Samples and Their XANES Spectra with Interpretation of Valence Balancea

XANESspectrum figure sample remarks

XANESpattern

1 1 As(V) inorganic reference n/a2 1 Na−As(V) inorganic reference n/a3 1 As(III) inorganic reference n/a4 1 Na−As(III) inorganic reference n/a5 1 Apy arsensopyrite, inorganic reference n/a6 2A RCS102 KT209 tobacco leaf (unburdened control) III < V7 2A RCS115 IV KT209 tobacco leaf (in vivo) III ≫ V8 2A RCS120 KT209 tobacco leaf with 8.6 μg/g As III > V9 2A RCS123 IV KT209 tobacco leaf (in vivo) with 13.5 μg/g As (dry weight) III ≪ V10 2A RCS123 KT209 tobacco leaf with 13.5 μg/g As III < V11 2A RCS108 KT209 tobacco leaf with 103 μg/g As III ≪ V12 2A RCS125 KT209 tobacco leaf with 248 μg/g As III ≈ V13 2B RCS132 KT209 tobacco leaf, As-burdened, fermented, shredded, and cased tobacco (10.1 μg/g As) III > V14 2B 1R4F tobacco from low tar reference cigarette III ≪ V15 2B CTA-VTL-2 trace element reference tobacco III ≪ V16 2B STA336 tobacco from a major US brand III ≪ V17 2B STA486 tobacco from a major Chinese brand III < V18 3 RCS133 smoke condensate from five cigarettes made with KT209 fermented, shredded, and cased

tobacco (0.3 μg/g As)III ≫ V

19 3 RCS132 smoke condensate from five cigarettes made from KT209 fermented, shredded, and casedtobacco (10.1 μg/g As)

III ≫ V

20 3 1R4F smoke condensate from one smoked (low tar reference) cigarette III ≫ V21 3 RCS133 ash from KT209 fermented, shredded, and cased tobacco (0.3 μg/g As) III ≪ V22 3 RCS132 ash from KT209 fermented, shredded, and cased tobacco (10.1 μg/g As) III ≪ V23 3 1R4F ash from one smoked (low tar reference) cigarette III ≪ V24 3 filter butt (procedural

blank)filter from Roll Your Own brand of tube + filter tip, unsmoked III < V

25 4 RCS132 five cigarettes smoked, no CO2(s) cryojet III ≫ V26 4 RCS132 one cigarette smoked, filter stored in CO2(s), no cryojet III ≫ V27 4 RCS132 three cigarettes smoked, filter stored in CO2(s), no cryojet III ≫ V28 4 RCS132 three cigarettes smoked, 30 min delay at room temp, no CO2(s), cryojet III ≫ V29 4 1R4F smoke condensate analyzed with cryojet III ≫ V30 4 1R4F smoke condensate analyzed without cryojet III ≫ V

aMore details on the samples are provided in Table S1 of the Supporting Information.

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563453

are more toxic and bioactive than As(V) species as aconsequence of greater chemical reactivity of As(III) andbecause As(III) enters cells more readily. Thus, in tobaccosmoke, a dominant inorganic As in trivalent state might beexpected to be the most harmful, although studies ofcarcinogenesis in humans due to inhaled As species are lacking5

To facilitate interpretation, an ordinal sequence of fivecombinations of valence states is defined, namely As(III) ≪As(V), As(III) < As(V), As(III) ≈ As(V), As(III) > As(V), andAs(III) ≫ As(V), identified by qualitative assessments of therelative contributions of As(III) and As(V) to the XANESspectra based on edge positions and the shape of the spectralprofile. These are listed for each sample of tobacco smokecondensate and ash in Table 1. The same data are plotted inFigure 5, summarizing all results in context from tobacco leaf tocommercial product, through combustion to smoke and ash.

As(III)−As(V) Variation in Tobacco Leaf and SmokingProducts. Very high As concentrations in tobacco have beencommonplace historically; for example, US commercial tobaccoaveraged 57 μg/g in 1951 when arsenical pesticides werelegal.17 The As burdening experiments reported above resultedin leaf with As concentrations up to 246 μg/g but beyond about1015 μg/g plants began to show signs of stress. Onlysamples with As-burdened concentrations at the lower end ofthis range were used in smoking experiments.Figure 2 clearly indicates that the tobacco plant has a

complex response to As burdening, and the As(III)−As(V)

relationship does not simply reflect the amount of added As.Indeed a sample to which only As(III) was applied has thestrongest As(V) edge, indicating that the plant mediates the Asoxidation state. However, some general features are apparent.Unburdened and unfermented KT209 tobacco is dominated byAs(V), but As(III) appears to increase with fermentation.Addition of different mixes of As(III) and As(V) to KT209resulted in leaf grown with burdens of 8.6 (RCS120) and 13.5μg/g As (RCS123) with significant As(III), but again As(V)dominates in the unfermented leaf whereas As(III) dominatesin the fermented sample. Exceptional burdening of KT209 withthe same As(III)−As(V) mix resulted in leaf with 246 μg/g As,and the fermented sample has approximately equivalent As(III)and As(V). Burdening with As(III) only, followed byfermentation, resulted in leaf with 103 μg/g As which isdominated by As(V) but with significant As(III).Tobacco cultivation occurs in an aerobic environment. The

redox conditions in the growth medium may favor As(V) sothat higher plants take As(V) via phosphate transporters,though As(III) would be transported through aquaporins.18,19

As(V) is reduced to As(III) in the roots, complexed withpeptides such as glutathione As(GS)3 and then potentiallytransported to root cells vacuoles or translocated to theshoot.19,20 Reactive oxygen species, generated throughreduction of inorganic As introduced within plants initiates acomplex constitutive detoxification mechanism in higher plantssuch as tobacco, involving transcription of phytochelatins oreven metallothioneins.11,15,19,21,22 The system is even morecomplex in that the leaves are in a state of senescence once thetobacco plant is harvested; it is known that metallothionein isexpressed in the phloem of tobacco for long distance transportof metals.23

An HPLC-ICPMS study of reference tobacco 3R4F9

indicated the presence principally of As(V), broadly consistentwith our XANES study of 1R4F, a sample prepared from asimilar blend.These data indicate that As(V) dominates in fresh tobacco

leaf but that fermentation involves the reduction of some (ormost) of the arsenic to As(III), a well-established process ofdetoxification by the plant.24,25 This process had not gone tocompletion in any of the leaf samples studied, but an increasedrelative proportion of As(III) is seen in all fermented samples.Tobacco intended for smoking, including reference materialsand commercial products, all indicate that As(V) is prevalentwith minor As(III) (Table 1, Figure 2). As(III) dominates onlyin one burdened and fermented tobacco sample (RCS132,Figure 2B).Combining all the XANES spectra of unsmoked tobacco it

becomes clear that reduction of As(V) to As(III) is adetoxification feature that applies when levels of As are highin the plant. Unburdened leaf samples, reference materials, andcommercial products all have bulk As concentrations below 1μg/g, and these samples remain dominated by As(V) even afterfermentation. Significant quantities of As(III) are found only inburdened tobacco leaf that in our experiments range from 8 to236 μg/g As. These data suggest that oxidized forms of As willtend to dominate in the product globally, except possibly inregions where tobacco relatively enriched in As is grown duepolluted environments and/or fertilizers.26

Valence State Changes Due to Combustion. Combus-tion has a marked effect on the valence state of As. This is clearfrom Figure 5 which emphasizes the division of As betweenAs(III) in smoke condensate and As(V) in residual ash.

Figure 5. Summary of changes in valence speciation of arsenic alongthe pathway from raw tobacco through curing and manufacture tocombustion and the creation of smoke products (ash and respirablesmoke condensate). Numbers refer to the XANES spectra detailed inTable 1 and shown in Figures 2−4. Broken lines join differentproducts of the same sample to emphasize the effect of combustion ineffectively partitioning As(III) and As(V) between ash and smoke,whatever the valence balance in precursor tobacco.

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563454

Notable is the absence of obvious mixtures of As(III) andAs(V), and this outcome is consistent regardless of the valencebalance of As in precursor tobacco. Earlier longer duration,lower resolution XANES studies7−9 concluded that 3R4Fsmoke condensate is a mix of As(III) and As(V), but ourshorter duration studies found no evidence for significantamounts of As(V) in smoke condensate, although we concur infinding that ash contains arsenic primarily in its oxidized state.It is clear that the environment generating the smoke aerosol

behind the burning coal is strongly reducing, consistent withfindings on the tobacco combustion and pyrolysis micro-environments.15,27 Thus, smoking As(V)-dominated tobaccogenerates an As(III)-dominated aerosol along with a residualash dominated by As(V), as summarized in Figure 5,contrasting strongly with the “mixed” valence findings of Liuand co-workers.7 Whether some of the ash has been through anintermediate stage of reduction cannot be determined from theXANES data. These results are consistent whether burdened orunburdened tobacco is used in the experiments.Smoke Aging and Oxidation. From smoke formation to

inhalation takes up to 2 s during which temperature falls from>700 °C to ambient. Inhaled and subsequent exhaled smokeundergo further aging during which redox changes continueover about 15 min.15 It has been argued that any As(III)produced during smoking and preserved in the smokecondensate fraction is transient and will rapidly oxidize duringthis interval.8 The XANES spectra of those studies wereobtained after 1−7 days and indeed are strongly oxidized exceptfor samples that were stored under dry ice and appear topreserve significant amounts of the reduced species.7,9 Here wealso use XANES to examine the effects of smoke aging byobserving shifts in edge energy that may indicate oxidation butover much shorter time scales. No significant shift related tooxidation was observed at room temperature over approx-imately 60 min that it took to smoke the cigarettes and toacquire their XANES spectra, 1 order of magnitude longer thanit normally takes to smoke a cigarette.7,8 These results areconsistent with the prediction of dominant As(III) usingpublished pH27 and Eh28 values projected in a Pourbaix plot forthe stability of arsenic species.7 We conclude that the oxidationof reduced arsenic species is slow relative to the time scale ofsmoking and indeed the rate of bronchial absorption (seebelow).Health Implications. Compared with dietary sources,

smoking tends to expose humans to smaller amounts ofarsenic, but this study indicates that it is all essentially in thereduced state, and published HPLC studies indicate that thelarge majority is inorganic.8 Thus, even low concentrations ofarsenic in smoke may represent a significant source of exposureto reduced inorganic arsenic species, especially for heavysmokers. These findings are summarized in Figure 5. Exposurewill be magnified in environments, most common in Asia,where As levels in tobacco may be significantly higher than istypical of major western manufacturers.16

Using volunteers to smoke cigarettes impregnated with asolution of radioactive sodium arsenite (i.e., reduced arsenic) inthe form of 74As, it has been shown that nearly 90% of inhaledAs is absorbed by the body and much of this occurs via thebronchial tree within the first day following exposure.29 If thesmoke condensate used in these XANES experiments faithfullyrepresents the arsenic in the smoke aerosol delivered to thelungs, then exposure will initially involve primarily inorganicAs(III) species. Further research is required to gain a better

understanding of the oxidation rates and the response ofspeciation to lung fluids and tissue.

Regulatory Implications. Regardless of precursor speciesit appears that combustion invariably generates the mosthazardous form of As in smoke, and XANES studies ofunburdened and As-burdened tobacco indicate that the processof reducing arsenic during combustion is effective over a farwider range of concentrations than might be expected incommercial tobacco crops. Thus, these findings on As incigarette smoking (but not oral use) add strong support toTobReg’s proposal that regulatory authorities should considerrequiring manufacturers to test cured tobacco and commercialproducts for levels of As. The findings also suggest that Asconcentration in tobacco, relatively easily determined in mostlaboratories, may be a sufficient indication of relative changes inthe risk of exposure to arsenic.

■ ASSOCIATED CONTENT*S Supporting InformationFurther details of the materials used in this study, cultivationprocedures, smoking experiments, XRF analysis, and thematerials used in experiments. This material is available freeof charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Tel: +44 (0) 1334 463947. Fax: +44 (0) 1334 463949. E-mail:Ed.Stephens@st-andrews.ac.uk.

NotesThe authors declare no competing financial interest.†Deceased August 12, 2013.

■ ACKNOWLEDGMENTSThis paper is dedicated to the memory of Rob Campbell, a finescientist and good friend who died as the manuscript was beingfinalized.We are grateful to The Netherlands National Institute for

Public Health and the Environment (RIVM) for providingsupport for RCJC and to Professor Antoon Opperhuizen fororganizing this support and his interest in discussing the results.Drs. Reinskje Talhout and Walther Klerx of RIVM are thankedfor discussions and practical assistance on smoke generationand smoking and for the loan of a single channel smokingmachine for use in experiments. We appreciate discussions onAs speciation with Professor Andy Meharg (Queen’s Universityof Belfast) and help with plant cultivation from Harry Hodgeand Dave Forbes. We thank Diamond Light Source for accessto beamline i18 through projects SP6601 and SP7744 thatcontributed to the results presented here and in particular theadvice and assistance of Professor Fred Mosselmans and Dr.Paul Quinn of DLS.

■ REFERENCES(1) WHO. WHO Report on the Global Tobacco Epidemic - TheMPOWER Package; World Health Organization: Geneva, 2008; p 14.(2) Talhout, R.; Schultz, T.; Florek, E.; van Benthem, J.; Wester, P.;Opperhuizen, A. Hazardous compounds in tobacco smoke. Int. J.Environ. Res. Public Health 2011, 8, 613−628.(3) FDA. Harmful and Potentially Harmful Constituents in TobaccoProducts and Tobacco Smoke: Established List. http://www.fda.gov/TobaccoProducts/GuidanceComplianceRegulatoryInformation/ucm297786.htm (accessed November 14, 2013).

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563455

(4) WHO. Report on the scientific basis of tobacco product regulation:fourth report of a WHO study group; World Health Organization:Geneva, 2012; pp 8−9.(5) IARC. Arsenic, metals, fibres, and dusts. IARC monographs on theevaluation of carcinogenic risks to humans. In World HealthOrganization, International Agency for Research on Cancer: Lyon,France, 2012; Vol. 100C, p 79.(6) Lugon-Moulin, N.; Martin, F.; Krauss, M. R.; Ramey, P. B.; Rossi,L. Arsenic concentration in tobacco leaves: A Study on threecommercially important tobacco (Nicotiana tabacum L.) types.Water, Air, Soil Pollut. 2008, 192, 315−319.(7) Liu, C.; Hu, J.; McAdam, K. G. A feasibility study on oxidationstate of arsenic in cut tobacco, mainstream cigarette smoke andcigarette ash by X-ray absorption spectroscopy. Spectrochim. Acta2009, B64, 1294−1301.(8) Liu, C.; Wright, C. G.; McAdam, K. G.; Taebunpakul, S.; Heroult,J.; Braybrook, J.; Goenaga-Infante, H. Arsenic speciation in tobaccoand cigarette smoke. Beitr. Tabakforsch. Int. 2012, 25 (2), 375−380.(9) Taebunpakul, S.; Liu, C.; Wright, C.; McAdam, K.; Heroult, J.;Braybrook, J.; Goenaga-Infante, H. Determination of total arsenic andarsenic speciation in tobacco products: From tobacco leaf and cigarettesmoke. J. Anal. At. Spectrom. 2011, 26, 1633−1640.(10) Mosselmans, J. F. W.; Quinn, P. D.; Dent, A. J.; Cavill, S. A.;Moreno, S. D.; Peach, A.; Leicester, P. J.; Keylock, S. J.; S.R., G.;Atkinson, K. D.; Rosell, J. R. I18 - the microfocus spectroscopybeamline at the Diamond Light Source. J. Synchrotron Radiat. 2009, 16,818−824.(11) Maher, W.; Foster, S.; Krikowa, F.; Donner, E.; Lombi, E.Measurement of inorganic arsenic species in rice after nitric acidextraction by HPLC-ICPMS: Verification using XANES. Environ. Sci.Technol. 2013, 47, 5821−5827.(12) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS:Data analysis for X-ray absorption spectroscopy using IFEFFIT. J.Synchrotron Radiat. 2005, 12, 537−541.(13) Huffman, G. P.; Huggins, F. E.; Shah, N.; Zhao, J. M. Speciationof arsenic and chromium in coal and combustion ash by XAFSspectrometry. Fuel Process. Technol. 1994, 39, 47−62.(14) Nesbitt, H. W.; Muir, L. J.; Praw, A. R. Oxidation of arsenopyriteby air and air-saturated, distilled water, and implications formechanism of oxidation. Geochim. Cosmochim. Acta. 1995, 59,1773−1786.(15) Wooten, J. B. Gas-phase radicals in cigarette smoke: A re-evaluation of the steady-state model and the Cambridge filter pad.Mini-Rev. Org. Chem. 2011, 8, 412−426.(16) O’Connor, R. J.; Li, Q.; Stephens, W. E.; Hammond, D.; Elton-Marshall, T.; Cummings, K. M.; Giovino, G. A.; Fong, G. T. Cigarettessold in China: Design, emissions and metals. Tobacco Control 2010, 19,47−53.(17) Tso, T. C. Micro- and secondary-elements in tobacco. Bot. Bull.Acad. Sin. 1966, 7, 28−63.(18) Wojas, S.; Clemens, S.; Sklodowska, A.; Antosiewicz, D. M.Arsenic response of AtPCS1- and CePCS-expressing plants - Effects ofexternal As(V) concentration on As-accumulation pattern and NPTmetabolism. J. Plant Physiol. 2010, 167, 169−175.(19) Zhao, F. J.; Ma, J. F.; Meharg, A. A.; McGrath, S. P. Arsenicuptake and metabolism in plants. New Phytol. 2009, 181, 777−794.(20) Pickering, I. J.; Prince, R. C.; George, M. J.; Smith, R. D.;George, G. N.; Salt, D. E. Reduction and coordination of arsenic inIndian mustard. Plant Physiol. 2000, 122, 1171−1177.(21) Mahmood, Q.; Ahmad, R.; Kwak, S. S.; Rashid, A.; Anjum, N. A.Ascorbate and glutathione: Protectors of plants in oxidative stress. InAscorbate-Glutathione Pathway and Stress Tolerance in Plants; Anjum,N. A.; Chan, M. T.; Umar, S., Eds.; Springer: Dordrecht, 2010; pp209−229.(22) Meharg, A. A.; Hartley-Whitaker, J. Arsenic uptake andmetabolism in arsenic resistant and nonresistant plant species. NewPhytol. 2002, 154, 29−43.

(23) Cobbett, C.; Goldsbrough, P. Phytochelatins and metal-lothioneins: Roles in heavy metal detoxification and homeostasis.Annu. Rev. Plant Biol. 2002, 53, 159−182.(24) Garg, N.; Singla, P. Arsenic toxicity in crop plants: Physiologicaleffects and tolerance mechanisms. Environ. Chem. Lett. 2011, 9 (3),303−321.(25) Webb, S. M.; Gaillard, J. F.; Ma, L. Q.; Tu, C. XAS speciation ofarsenic in a hyper-accumulating fern. Environ. Sci. Technol. 2003, 37(4), 754−760.(26) Stephens, W. E.; Calder, A. Analysis of non-organic elements inplant foliage using polarised X-ray fluorescence spectrometry. Anal.Chim. Acta 2004, 527 (1), 89−96.(27) Baker, R. R. Smoke Chemistry. In Tobacco: Production,Chemistry and Technology; Davis, D. L.; Nielsen, M. T., Eds.; BlackwellScience: Oxford, 1999; pp 398−439.(28) Schmeltz, I.; Tosk, J.; Jacobs, G.; Hoffmann, D. Redox potentialand quinine content of cigarette smoke. Anal. Chem. 1977, 49, 1924−1929.(29) Holland, R. H.; McCall, M. S.; Lanz, H. C. A study of inhaledarsenic-74 in man. Cancer Res. 1959, 19, 1154−1156.

Environmental Science & Technology Article

dx.doi.org/10.1021/es4039243 | Environ. Sci. Technol. 2014, 48, 3449−34563456