Sw

ND

h

•

•

•

•

a

ARRAA

KMNTH

1

com

f

(

0h

Analytica Chimica Acta 813 (2014) 56–62

Contents lists available at ScienceDirect

Analytica Chimica Acta

jou rn al h om epage: www.elsev ier .com/ locate /aca

olid phase extraction of trace amount of mercury from naturalaters on silver and gold nanoparticles

. Panichev, M.M. Kalumba ∗, K.L. Mandiwanaepartment of Chemistry, Tshwane University of Technology, P.O. Box 56208, Arcadia 0007, Pretoria, South Africa

i g h l i g h t s

Ag/Au nanoparticle membrane filterswere used to extract Hg extractionfrom natural water.The LOD of Hg determination waslower (0.4 ng L−1) than the traditionalmethod of cold vapour generation(200 ng L−1).Sample contamination was mini-mized by thermal desorption of Hgdetermination.Hg collected on nanoparticle mem-brane filters were kept for at least 5months without Hg loss.

g r a p h i c a l a b s t r a c t

r t i c l e i n f o

rticle history:eceived 19 March 2013eceived in revised form 5 December 2013ccepted 5 January 2014vailable online 13 January 2014

a b s t r a c t

Silver (Ag) and gold (Au) nanoparticles impregnated in nylon membrane filters have been proposed asa new solid phase for preconcentration of mercury from natural waters. Water samples were treatedwith KMnO4 to convert all mercury species to inorganic Hg2+ and this was followed by the reduction ofHg2+ with NaBH4 to elemental Hg0. The determination of Hg was carried out by thermal evaporation ofmercury from membrane filters using Zeeman mercury analyzer RA–915+ (Lumex, Russia). This process

eywords:ercuryanoparticleshermal evaporationg analyzer

does not involve any additional sample treatment and sharply reduces risk of samples contamination. Thelimit of detection (LOD) was found to be 0.04 ng (absolute mass). Relative LOD was 0.4 ng L−1 for 100 mL ofwater. The method was validated through the analysis of CRM NRCC Tort–2 (Lobster hepatopancreas) andthe found value (0.30 ± 0.07 �g g−1) was in good agreement with the certified value (0.27 ± 0.06 �g g−1).High efficiency of Hg accumulation from aqueous phase to membrane filters can be attributed to a largesurface area of nanoparticles.

. Introduction

The determination of mercury (Hg) in natural waters using

old vapour (CV) generation technique became a routine methodf trace analysis for several decades [1]. The advantages of thisethod lie in the minimisation of interferences from matrix due to

∗ Corresponding author. Tel.: +27 12 382 6233/+27 73 661 6061 (mobile);ax: +27 12 382 6286.

E-mail addresses: Kalumba.merime1@gmail.com, panichevn@tut.ac.zaM.M. Kalumba).

003-2670/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.aca.2014.01.011

© 2014 Elsevier B.V. All rights reserved.

complete separation of the analyte (Hg) from the matrix in gaseousstate [2]. The detection of Hg is carried out at room temperaturein atomic absorption or atomic fluorescent mode at a wavelengthof 253.7 nm [3]. Generally, before Hg measurements by CV, watersamples should be chemically treated. The first step of sample treat-ment is the oxidation of all Hg species present in the sample toone chemical form (Hg2+) [4]. In the second step, Hg2+ is reducedto its elemental form either by the addition of mild reducing agent

(SnCl2) or more reactive substance (NaBH4). The elemental mercury(Hg0) is purged from the solution by the carrier gas and transportedto a measurement cell. The limit of detection (LOD) of Hg deter-mination in natural waters using CV is approximately 0.2 �g L−1

Chim

[nep

ifHdvfioLol

owbbdeeaimtcnsa(

pcAfiabaafm

2

2

Rstp7aastoaZoTac

formed. The concentration of Au nanoparticles in obtained solutionwas found to be 1980 ± 30 �g L−1, which corresponds to predicted

N. Panichev et al. / Analytica

5–7]. This LOD is not low enough for Hg to be determined at theatural environmental level. The determination of Hg at trace lev-ls would be possible by analytical methods in which a step of Hgreconcentration is involved [8–10].

Solid phase preconcentration of Hg on thin films of noble metalsncluding gold, palladium and silver are favourable solid sorbentsor trapping mercury vapour due to their high affinity towards theg. At present, two methods of Hg preconcentration have beeneveloped. One of them is to trap Hg0 which was formed during coldapour generation on electrothermal atomizers covered with thinlms of noble metals, like Au, Pd or Rh [11–14]. The main advantagef trapping Hg0 on thin layers of noble metals is the reduction ofOD. The disadvantages of Hg preconcentration from gaseous staten thin films of noble metals are higher risk of contamination andonger cycle time [15].

The other novel solid-phase preconcentration of Hg is basedn the sorption of Hg0 on gold nanoparticles directly from naturalaters. In this case, the stage of cold vapour generation is avoided

ecause Hg from the aqueous phase is collected on the solid sor-ent, which consists of silica coated with gold nanoparticles. Theetermination of Hg is carried out by atomic fluorescent spectrom-try (AFS) after thermal desorption of Hg from the sorbent. Leopoldt al. [16–19] found that all Hg species can be preconcentrated bymalgamation of Hg0 on the surface of silica due to catalytic activ-ty of the nano–gold particles, which provides decomposition of

ercury species and some complexes in aqueous sample at roomemperature. Subsequently, the liberation of trapped mercury wasarried out by Hg analyzer after thermal desorption and as a result,o reagents were needed for species conversion, preconcentration,ample storage and desorption, resulting in reduced contaminationnd lower blank signal. Consequently, a very impressive LOD for Hg180 pg L−1) have been achieved [19].

In the present work, the method of solid phase extraction andreconcentration of mercury from natural waters with low con-entration of Hg is described. It is based on the sorption of Hg0 ong or Au nanoparticles impregnated into membrane filters duringltration of water samples. To avoid the problems connected withdditional experiments to prove that gold nanoparticles are capa-le of converting different Hg species in acidified waters withoutdditional reagents, all water samples were treated with KMnO4ccording to “classical” procedure of water samples preparationor Hg determination [5,6]. The measurements were carried out by

ercury analyzer after thermal desorption of Hg0 from filters.

. Experimental

.1. Instrumentation

All mercury measurements were carried out with a ModelA–915+ Zeeman Mercury analyzer (Lumex, St. Petersburg, Rus-ia) fitted with PYRO–915+ attachment. The working principal ofhe instrument is based on the thermal desorption of Hg from sam-les placed in pyrolysis tubes, which can be heated from 300 ◦C to00 ◦C. The measurements of Hg concentration are carried out in annalytical cell, constantly heated at 800 ◦C. In this work the temper-ture of preheated tube was set at 700 ◦C. When known amount ofolid sample placed in the sampling boat is inserted in preheatedube, Hg vapour and smoke formed after combustion of matrix’srganic materials are transported into analytical cell. Backgroundbsorption in the analytical cell is eliminated by the high-frequencyeeman correction system. The high total multi-path optical length

f analytical cell (0.4 m) allows detecting Hg on picogram level.he concentration of Hg in a sample is determined from integratednalytical signals, using calibration graph, plotted with differentertified reference materials (CRMs). The PYRO–915+ enables Hg

ica Acta 813 (2014) 56–62 57

determination in samples with complex–matrices, such as soils,sediments, oil products and foodstuffes using pyrolysis techniquewithout samples pretreatment [20,21]. With no chemical samplepretreatment and without addition of chemical modifiers, the riskof contamination is minimized.

The distribution of Ag and Au nanoparticles on membrane filtershave been studied with JSM 7500 F Scanning Electron Microscope(JEOL, Ltd., Tokyo, Japan). The size of the AuNPs and AgNPs preparedwere investigated using a JEM-2100 (HR) transmission electronmicroscopy (JEOL, Ltd., Tokyo, Japan) operated at an acceleratingvoltage of 200 kV. The concentration of Ag and Au in solutions withnanoparticles and the amount of adsorbed Ag and Au on filterswere determined by electrothermal atomic absorption spectrome-try (ETAAS), using Perkin Elmer Model AAnalyst 600 spectrometer(Perkin Elmer, USA).

2.2. Reagents and reference materials

All solutions were prepared using ultra-pure water with aresistivity of 18.2 M� cm, obtained from Milli-Q Plus water purifi-cation system (Millipore, Bedford, MA, USA). Mercury standardsolutions in the range of 0.10–2.0 �g L−1 were prepared by sequen-tial dilution of 1000 mg L−1 Hg stock solution (Merck, Darmstadt,Germany). Silver nitrate, AgNO3 (Saarchem, Merck, Germany) andgold as HAuCl4 stock standard solution of 1000 �g L−1, (Spec-troscan, Teknolab, Norway) were used for preparation of Ag andAu nanoparticles. Sodium borohydride, NaBH4 3% (m/v) solutionin 0.3% (m/v) NaOH was used as a reducing agent and glucose asthe stabilizer were prepared from analytical grade reagents (Merck,Darmstadt, Germany). After dissolution of the reagents, the solu-tions were filtrated through Millipore hydrophilic PVDF 0.45 �mfilters to remove undissolved solids.

The following CRMs were used in this study to calibrate RA–915+Zeeman Mercury analyzer: SRM 1515, Apple leaves–Trace elements(National institute of Standards and Technology, USA), certifiedvalue Hg 0.044 �g g−1, CMI 7002 Light Sandy soil–Trace elements,(Analytika Co Ltd., Czech Republic, certified value Hg 0.090 �g g−1

and CRM NCS DC 73308 Stream sediment (China National AnalysisCenter for Iron and Steel, China), certified value: Hg 0.28 �g g−1.

2.3. Preparation of silver and gold nanoparticles

The preparation of Ag nanoparticles was carried out accordingto the procedure recommended by A. Zielinska et al. [22]. For thispurpose, 2.0 mL of 0.005 M AgNO3 was added to 500 mL deionizedwater. The calculated concentration of Ag ions in obtained solu-tion was 2000 �g L−1 but the actual concentration using ETAASwas found to be 2220 ± 60 �g L−1. After addition of 10 mL of 5 g L−1

NaBH4 and 0.05 g of glucose, the colourless mixture was slowlyturned pale yellow, indicating that Ag-nanoparticles were formed.Nanoparticles of Au were prepared by the addition of 6 mL of lemonjuice to 50 mL of hydrogen tetrachloroaurate solution (HAuCl4)[23]. The mixture was brought to a boil with vigorous stirring. Thecolour of the solution changed from colourless to purple to ruby redwithin 10 min. The colloidal solution was stirred for an additional20 min, cooled at room temperature, transferred to an amber bottleand stored at room temperature. The colour of the solution changedfrom yellowish to ruby red, indicating that gold nanoparticles have

value of 2000 �g L−1. The last step to impregnate nanoparticles intomembrane filters have been done by filtration of solutions with Agor Au nanoparticles. The mass of Ag and Au nanoparticles anchoredon filters was determined by ET AAS.

58 N. Panichev et al. / Analytica Chimica Acta 813 (2014) 56–62

3500

3550

3600

3650

3700

3750

3800

0 2 4 6 8 10 12 14 16

Pea

k a

rea

(Arb

itra

ry u

nit

)

2c

dHGbfsp(hBoiaps

3

3

cd1istcc

3

tNo

3500

3550

3600

3650

3700

3750

3800

0 2 4 6 8 10 12 14 16

Pea

k a

rea

(Arb

itra

ry u

nit

)

nanoparticles. These amounts could be considered as the total sorp-tion capacity of membrane filters.

An estimate of particle size had been performed using the TEMimages of AuNPs and AgNPs on the membrane surfaces (Fig. 4) were

5

10

15

20

25

Mas

s of

Ag a

nd A

u o

n f

ilte

r (

µg)

NaBH4 concentration, % (m/v )

Fig. 1. Effect of NaBH4 concentration on 5 ng Hg peak area.

.4. Samples preparation for the determination of total Hgoncentration

For sample collection, all glassware used were cleaned withetergent, thoroughly rinsed with tap water, soaked in a 5% (v/v)NO3 solution overnight and finally rinsed with Milli-Q water.lass bottles were rinsed three times with the river water beforeeing filled. Samples were drawn through by vacuum suction, byorce of a pump, or simply rely on gravity to pass through theolid sorbent. Organic and inorganic Hg compounds in water sam-les were decomposed with 2% (m/v) KMnO4. The excess oxidantpermanganate) was destroyed by the addition of few drops ofydroxylammonium chloride until the solution was colourless.efore filtration, Hg2+ was reduced to Hg0 by the addition of 1 mLf 0.001 M NaBH4 per each 100 mL of water sample. In all cases its imperative that sufficient contact time of 1 min was allowed fornalytes to be adsorbed onto the membrane filters, without beingrolonged to the extent of taking several hours simply to pass theample through [24].

. Results and discussions

.1. Effect of NaBH4 concentration

The first parameter to be examined was the effect of NaBH4oncentration on the production of the mercury vapour. Fig. 1emonstrates the effect of increasing concentration of NaBH4 from

to 15% on the peak area of 5 ng Hg. The results showed the analyt-cal signal of Hg was constant after addition of 3–15% (m/v) NaBH4olutions. Hence, 3% (m/v) NaBH4 was used throughout the studyo minimize deterioration of NaBH4 concentration in solution. Thisoncentration was also thought to be the optimum in terms ofost-effectiveness.

.2. Effect of volume of NaBH4

In order to determine the appropriate volume of NaBH4 requiredo achieve optimum peak area of Hg, various volumes of 3% (m/v)aBH4 were added to the solution containing an absolute amountf 5 ng Hg2+. It was found that 1–15 mL of NaBH4 are capable to

Volume of NaBH4 (mL)

Fig. 2. Effect of the volume of NaBH4 3% (m/v) on 5 ng Hg peak area.

produce the absorption signals of Hg with constant peak area(Fig. 2). On the base of these studies, an addition of 5 mL of NaBH4was chosen as an optimal volume.

3.3. Adsorption capacity of membrane filters for Ag and Aunanoparticles

The amount of Ag or Au nanoparticles adsorbed by membranefilters and their total sorption capacity were determined after thefiltration of different volumes of solutions with known concentra-tion of Ag and Au as quantified by ETAAS analysis of filters. Forthis purpose, filters were dissolved in 1.0 mL of freshly preparedaqua regia, and diluted to 50 mL with deionized water. The resultsof ETAAS determination (Fig. 3) showed that after filtration of dif-ferent volumes of AuNPs and AgNPs solutions, membrane filtersaccumulated stable amount of Ag (11 ± 1 �g) and Au (20 ± 0.7 �g)

0

0 20 40 60 80 100 120

Volumes of Ag ( ) and Au ( ) n anop articl es so lut ion ( mL)

Fig. 3. Sorption of Ag (�) and Au (�) nanoparticles on 0.20 �m membrane filters.

N. Panichev et al. / Analytica Chimica Acta 813 (2014) 56–62 59

rticles

fstrsm

tamtecpsm

3

gttbat1ors

Fig. 4. TEM images of seed nanopa

ound to be approximately 20 nm. Clearly, the fact that they areome 10 fold smaller than the membrane filter pore size means thatheir presence on the membrane is not due to a filtering action butather adsorption onto the surface where they appear to be quitetrongly retained, as evidenced by the fact that once prepared, theseembranes can be used for months with good efficiency.The different loadings pertain to mass can be accounted for by

he relative atomic weight differences for Au and Ag. In point of fact,pproximately the same molar loadings are achieved with bothetals and since the resultant nanoparticles are approximately

he same size, then the surface areas presented are approximatelyqual, thus accounting for the nearly identical sorption kinetics andapacities for mercury uptake [25]. It was concluded that for thereparation of membrane filters for Hg sorption, 50 mL of stockolution with concentration of Ag or Au nanoparticles of approxi-ately 2000 �g L−1 could be taken.

.4. Adsorption of Hg on membrane filters

The kinetics of adsorption that describes the solute uptake rateoverning the contact time of the adsorption reaction was one ofhe important characteristics that define the efficiency of adsorp-ion. Hence, in the present study, the kinetics of Hg adsorption onlank nylon membrane filters and filters impregnated with AuNPsnd AgNPs was studied. The results, presented in Fig. 5 showed thathe effect of contact time of Hg solutions with concentrations of

500 ng L−1 (A) and 2500 ng L−1 (B) on Hg sorption by nanoparticlesn membrane filters. From these data follows that the adsorptioneaches its maximum value in 60 s, which corresponds to 100% of Hgorption and remains stable up to 240 s, while only 7.9% of Hg has

of Au (A and C) and Ag (B and D).

been absorbed by the blank membrane filter. The process observedis a heterogeneous phase separation of the mercury from the solu-tion to the surface of the nanoparticle. Assuming that “intimate”mixing was achieved throughout the filtering process, wherein thesolution flows past the immobilized particles, it was clear that thedegree of interaction of the solution with the adsorbent was thesame whether the sample volume is 1 mL or 500 mL. It was onlythe time of interaction that was important.

3.5. Adsorption of Hg on filters with Ag and Au nanoparticles

The study of Hg adsorption on filters impregnated with Agnanoparticles has been carried out after filtration of different vol-umes of 1 ng mL−1 Hg standard. The tested volumes varied between5 and 300 mL, which correspond to the range between 5 and 300 ng(absolute mass) of Hg. The results showed that filters with poresize of 0.2 and 0.22 �m have the same efficiency for Hg sorption,because no statistical difference between amounts of Hg adsorbedhave been found at 95% level of confidence (Fig. 6A).

The adsorption of Hg on filters with Ag and AuNPs was car-ried out by filtration of 50–300 mL of 0.1 ng mL−1 Hg standardsolutions. The results indicated that both kinds of nanoparticlesadsorb Hg with the same efficiency in the range of 0.5–50 ng(Fig. 6B). The mechanism of Hg interaction with AuNPs and AgNPsand thus their effect of the adsorption of mercury is possibly dueto the 5d10(Hg2+)–4d10(Ag+) high metallophilic interaction which

allowed Hg amalgamation on Ag [26], and due to the big surfacearea of AgNPs and AuNPs [27].

The recovery of Hg adsorbed on filters with impregnated Ag(Au) NPs (Fig. 6C) had been studied after filtration of 50 mL of

60 N. Panichev et al. / Analytica Chimica Acta 813 (2014) 56–62

A

B

0

200

400

600

800

1000

1200

1400

1600

0 50 10 0 15 0 20 0 250

Conce

ntr

atio

n, ng/g

Time, sec

0

500

1000

1500

2000

2500

3000

0 50 10 0 15 0 20 0 250

Conce

ntr

atio

n, ng/g

Time, sec

Fn2

s5(fs

3fi

ssm(ww

secSmaimowc(

C

0

20

40

60

80

100

120

100 50 0 100 0 200 0 3000 400 0 5000

Rec

over

y o

f m

ercu

ry (

%)

Concentration of nanop article s so luti on, µg L-1

A

B

Ag: 0.22 µm : y = 618.1x + 650.4

R² = 0.993

Ag: 0.2 µm: y = 604.4x + 347.5

R² = 0.992

0

40000

80000

120000

160000

200000

0 100 200 300 400

Pea

k a

rea

(arb

itra

ry u

nit

s)

Absolute mass of mercury (ng) on 0.2 µ m ( ) and 0.22 µm ( )

membrane filter s

Ag: y = 600.7x + 316.6

R² = 0.9989

Au: y = 581.4x + 275.8

R² = 0.9984

0

10000

20000

30000

40000

0 10 20 30 40 50 60

Pea

k a

rea

(Arb

itra

y u

nit

s)

Absolute mass of collected Hg (ng)

Fig. 6. Absolute mass of Hg (ng) collected on 0.20 �m (�) and 0.22 �m (�) fil-ters impregnated with Ag nanoparticles (A); Absolue mass of Hg (ng) collectedon 0.20 �m membrane filters impregnated with Ag (�) and Au (�) nanoparticles

ig. 5. Adsorption kinetics of Hg onto blank membrane filter (�) and filters impreg-ated with Au (�) and Ag (�) nanoparticles from Hg solutions 1500 �g L−1 (A) and500 �g L−1 (B).

olutions with nanoparticles concentration from 100 to000 �g L−1. It was found that complete adsorption of Hg100 ± 2%) was achieved with solutions Ag (Au)NPs concentrationsrom 2000 �g L−1 to 5000 �g L−1. For practical application, theolutions of Ag (Au)NPs 2000 �g L−1 was used.

.6. Distribution of Ag, Au nanoparticles and Hg on membranelters

The scanning electron microscope (SEM) and energy dispersivepectroscopy (EDS) images of the surface of the membrane arehown in Fig. 7, where (A) is a view of the blank nylon, (B) modifiedembranes with AuNPs, (C) modified membranes with AuNPs + Hg,

D) modified membranes with AgNPs and (E) modified membranesith AgNPs + Hg. The maps of Hg distribution demonstrate that itas attached to specific sites, to individual Ag or Au nanoparticles.

The bond between Hg and nanoparticles of both elements wastable because the amount of Hg on the filters did not changeven after 6 months of filter storage at room temperature. Typicalross-sectional areas of the blank membrane were viewed usingEM as shown by the porous structures (Fig. 7A), however, theacrovoids do not span the entire width of the membrane as the

dsorbent–adsorbate interactions depend both on diffusion andnteractive parameters [28]. The presence of porous structure in the

embrane indicates that there is uniform distribution of the pores

ver the surface. The presence of AuNPs and AgNPs in the filtersas confirmed by ETAAS analysis and EDS measurements which

onfirmed the peaks of Au (Fig. 7B), amalgamation of Au with HgFig. 7C), Ag (Fig. 7D) and Ag (Fig. 7E).

(B); Comparison of the efficiency of Hg adsorption on 0.20 �m filters from differentconcentrations of Au (�) and Ag (�) nanoparticles solutions (C).

3.7. Analytical characteristics of the method

The analytical parameters of the method of Hg determination innatural water based on the adsorption of Hg on Ag or Au nanoparti-cles impregnated in membrane filters were carefully investigated.

It was found that the preconcentration of Hg can be carried outeither on Ag or on Au nanoparticles with the same efficiency asthe slope of the calibration curves of Hg standards in Au and Ag

N. Panichev et al. / Analytica Chimica Acta 813 (2014) 56–62 61

F S) spem E), mo

nt0c

itW

ig. 7. Scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDodified membranes with AuNPs + Hg; (D), modified membranes with AgNPs and (

anoparticles were equal at 95% level of confidence. The calibra-ion curve plotted for small concentrations of Hg in the range of.5–50 ng (absolute mass) or 5–500 ng L−1 is linear with regressionoefficients of R2 = 0.9989.

The LOD is defined as the analyte concentration correspond-ng to three (3) standard deviations of integrated absorbance forhe blank solution divided by the slope of the calibration function.

hen blank signal cannot be measured, the calibration data and

ctra of products: (A); Blank membrane, (B), modified membranes with AuNPs; (C),dified membranes with AgNPs + Hg

regression statistics can be used instead [29]. The limit of detection(LOD) was calculated from the equation of the calibration curve dueto the absence of samples of water free of Hg that could be used asblanks [29].

For this purpose, ten (10) different masses of Hg were loadedon filters after filtration of different volumes of 0.1 ng mL−1 Hg.The calibration curve was plotted as absolute mass of Hg (ng)versus peak area (arbitrary units) and was defined by simple

62 N. Panichev et al. / Analytica Chim

Table 1Analytical results for the determination of trace mercury in natural water samplescollected in Apies river.

Samples Unspiked(ng L−1)

Spiked(ng L−1)

Found (ng L−1) Recovery (%)

Location 1 7.5 ± 1.3 2.0 9.5 ± 1.5 98.8Location 2 15 ± 2.5 2.0 16.9 ± 1.8 99.0Location 3 120 ± 5 2.0 122 ± 8 100.2

litwspt0TTtv

crc

3

itcrtstfl

4

ipwtnt

[[

[

[

[

[[[

[[[[

[

[

[

[

[[

Location 4 35 ± 3 2.0 37 ± 1.3 100.1Location 5 19 ± 1 2.0 20.9 ± 2.4 99.1Location 6 14 ± 1 2.0 15.9 ± 1.4 98.7

inear regression equation, y = 490.8x + 10.9, R2 = 0.9987, resultingn the absolute LOD of 0.040 ng. This value is slightly higher thanhe LOD for Hg determination in solid sediments (0.013 ng), whichas obtained using the same RA–915+ mercury analyzer [30]. It

hould be noted that the LOD in concentration units is inverselyroportional to the sample volume processed. Because in this studyhe typical volume of water sample used for Hg determination was.1 L, the LOD in concentration unit was found to be 0.4 ng L−1.he method was validated by analysis of dissolved CRM NRCCort–2 (Lobster hepatopancreas) with good agreement betweenhe found Hg concentration (0.30 ± 0.07 �g g−1) and the certifiedalue (0.27 ± 0.06 �g g−1).

To check the performance of the mercury analyzer, three repli-ates of CRM CMI 7002 Light Sandy soil were analyzed daily. Theesults of Hg determination were within the confidence interval ofertified value of 0.090 ± 0.012 �g g−1.

.8. Analysis of natural water samples

The developed method was applied for the determination of Hgn natural waters from Apies River in Pretoria. Water samples wereaken on six (6) different sampling sites in the central part of theity. Samples were treated with KMnO4 in acid media and Hg wasestored to Hg0 with NaBH4. The results showed that the concen-ration of Hg in Apies River is relatively low, but varied from oneampling site to the other (Table 1). The variability of Hg concen-ration on different sampling location could be connected with theow of effluent from numerous drain pipes into river.

. Conclusions

It was shown that silver (Ag) and gold (Au) nanoparticlesmpregnated in nylon membrane filters can be used as a new solidhase for preconcentration of small amounts of Hg from natural

aters. High efficiency of adsorption of Hg from aqueous phase

o membrane filters can be attributed to the large surface area ofanoparticles. The additional advantage of the proposed method ishe thermal desorption of Hg from membrane filters using Zeeman

[[

[

ica Acta 813 (2014) 56–62

mercury analyzer RA–915+ which eliminate the sample prepara-tion step thereby sharply reducing the risk of contamination. TheLOD for the determination of Hg in water was 0.4 ng L−1 much lowerthan 200 ng L−1 achieved by cold vapour generation technique. Theresults of this study clearly show the potential of this method whichcould be applied to monitor Hg in water with trace levels of Hg.Furthermore, water samples can be sampled directly to the impreg-nated filters from water bodies thereby minimizing contaminationduring transport and storage.

References

[1] W.L. Clevenger, B.W. Smith, J.D. Winefordner, Crit. Rev. Anal. Chem. 27 (1997)1–26.

[2] J. Dedina, D. Tsalev, Hydride Generation Atomic Absorption Spectrometry, JohnWiley and Sons, Chichester, 1955.

[3] B. Welz, M. Sperling, Atomic Absorption Spectrometry, third ed., WILEY VerlagGmbH, Weinhem, 1999.

[4] S.M. Ulrich, T.W. Tanton, S.A. Abdashitova, Crit. Rev. Environ. Sci. Tech. Sci. 31(2001) 241–293.

[5] US EPA, Method 245.1, Determination of Mercury in Water by Cold VaporAtomic Absorption Spectrometry, 1979.

[6] Y. Zhang, S.C. Adeloju, Talanta 74 (2008) 951–957.[7] D.P. Torres, D.L.G. Borges, V.L.A. Frescura, A.J. Curtius, J. Anal. At. Spectrom. 24

(2009) 1118–1122.[8] K. Leopold, M. Foulkes, P. Worsfold, Trends Anal. Chem. 28 (2009) 426–435.[9] K. Leopold, M. Foulkes, P. Worsfold, Anal. Chim. Acta. 663 (2010) 127–138.10] Y. Gao, Z. Shi, Z. Long, P. Wu, C. Zheng, X. Hou, Microchem. J. 103 (2012) 1–14.11] E.M.M. Flores, B. Welz, A.J. Curitus, Spectrochim. Acta. Part B 56 (2001)

1605–1614.12] V. Cerveny, P. Rychlovsky, J. Netoliska, J. Sima, Spectrochim. Acta Part B 62

(2007) 317–323.13] V. Romer, I. Costa–Mora, I. Lavilla, C. Bendicho, Spectrochim. Acta Part B 66

(2011) 156–162.14] Method 1631E, Mercury in Water by Oxidation, Purge and Trap, and Cold Vapor

Atomic Fluorescence Spectrometry, U.S. Environmental Protection Agency,Office of Water, Washigton, DC, 2002.

15] T. Labatzke, G. Shlemmer, Anal. Bioanal. Chem. 378 (2004) 1075–1082.16] K. Leopold, M. Foulkes, P. Worsfold, Anal. Chem. 81 (2009) 3421–3428.17] A. Zierhut, K. Leopold, L. Harwardt, P. Worsfold, M. Schuster, J. Anal. At. Spec-

trom. 24 (2009) 767–774.18] A. Zierhut, K. Leopold, L. Harwardt, M. Schuster, Talanta 81 (2010) 1529–1535.19] K. Leopold, A. Zierhut, J. Huber, Anal. Bioanal. Chem. 403 (2012) 2419–2428.20] C. Bin, W. Xiaoru, F.S.C. Lee, Anal. Chim. Acta 447 (2001) 161–169.21] S. Sholupov, S. Pogarev, V. Ryzhov, N. Moshynov, A. Stroganov, Fuel Process.

Technol. 85 (2004) 473–485.22] A. Zeielinska, E. Skwarek, A. Zaleska, M. Gazda, J. Hupka, Proc. Chem. 1 (2009)

1560–1566.23] M.V. Sujitha, S. Kannan, Spectrochim. Acta Part A, Mol. Biomol. Spectrosc. 102

(2013) 15–23.24] H.L. Lord, Solid phase microextraction for drug analysis, in: J. Paivlizyn, H.L. Lord

(Eds.), Handbook of Sample Preparation, Google Books, 2010, pp. 365–379.25] R.J. Sime, Physical Chemistry: Methods, Techniques and Experiment, Saunders

College Publishing, Philadelphia, 1990, pp. 528–532.26] C. Guo, J. Irudayaraj, Anal. Chem. 83 (2011) 2883–2889.27] M. Levlin, E. Ikävalko, T. Laitinen, Fresenius J. Anal. Chem. 365 (1999) 577–586.

28] M.L. Ferreira, M.E. Gschaider, Macromol. Biosci. 1 (2001) 233–248.29] J.N. Miller, J.C. Miller, Statistics and Chemometrics for Analytical Chemistry,

fifth ed., Pearson Education Limited, England, 2005.30] K.N. Mekonnen, A.A. Ambushe, B.S. Chandravanshi, M.R. Abshiro, R.I. McCrindle,

N. Panichev, Toxicol. Environ. Chem. 94 (2012) 1678–1687.