Field determination of inorganic mercury in seawaters by a portable dual-channel and purge-and-trap system with atomic fluorescence spectrometry

ABSTRACT An on-line flow injection mercury analyser (FIMA) for inorganic mercury (Hg) analysis in environmental waters is presented. The FIMA is a portable dual-channel and purge-and-trap system combining aqueous reduction with stannous chloride (SnCl2), two-stage gold (Au) amalgamation, thermal desorption of Hg0 and final detection with cold vapour atomic fluorescence spectrometry (CVAFS). Simultaneous on-line operations of sample loading and analysis make analytical throughput and accuracy improved and contamination eliminated. Results of system optimisation and reliability regarding the determination of inorganic Hg2+ are described. Analytical performances of both channels were the same in terms of data reliability. Low procedural blanks were obtained from laboratory and field trip (≤5 pg, n = 30). Excellent calibrations (r2 > 0.99) with a high precision and good stability (RSD <5%, n = 30) were obtained. Detection limits and reproducibility of the methods have been estimated as ≤0.05 ng L−1 and <5% (0.5 ng L−1, n = 30). The system was finally validated by measurements comparison of the two channels (r2 > 0.98), and analysis of a certified reference material (coastal seawater BCR-579, recoveries 101 ± 4%, n = 10). Overall, a sample throughput of ~10 samples per hour with two-channel detections of FIMA can be achieved. The onboard analysis of inorganic Hg species (e.g. elemental, reactive, and total Hg) in surface seawaters of the East China Sea (ECS) are also successfully achieved. The portable FIMA is fast, easy, and robust for reliable monitoring of Hg in natural seawaters.


Introduction
Mercury (Hg) is a highly persistent, bioaccumulative and neurotoxic pollutant.Globally, environmental Hg pollution has increased over the past two centuries due to anthropogenic sources, mostly through fossil fuel burning, mining and industrial activities [1][2][3].These additional sources of Hg pose a threat to mankind through the consumption of fish and seafood [4,5].The tissue concentrations of Hg in many marine and freshwater fish easily exceeds guidelines (e.g.0.5 µg g −1 w.w.) suggested by the World Health Organization (WHO) to protect public health [6].The U.N. Minamata Convention on Hg was therefore initiated in Jan. 2013 and came into force in 2017 to prevent Hg pollution by controlling and mitigating the anthropogenic Hg release (http://www.mercuryconvention.org/).Regarding public health and ecological problems, the monitoring of Hg in environmental matrices is therefore of great importance.
To obtain accurate environmental Hg measurements is, however, analytically challenging due to very low levels of Hg species, species transformation and sampling-related contamination issues [7][8][9].Additionally, the distribution of Hg species varies temporally and spatially so that real-time, field measurements are greatly desired.Methods using derivatisation and volatilisation of Hg coupled with the purge-and-trap technique provide excellent sensitivity and are broadly applicable to environmental issues [10,11] The United States Environment Protection Agency (US EPA) standard method #1631 based on dual-stage gold (Au) amalgamation approach for inorganic Hg determination in natural waters was, for instance, worldwide used [12,13].It comprises manual handling and off-line purging and analysis over 20 minutes per sample.Many improvements have been nevertheless made in analytical techniques/systems, coupled to atomic absorption spectrometry (AAS) [14][15][16], atomic fluorescence spectrometry (AFS) [17][18][19][20][21], inductively coupled plasma (ICP) [22][23][24], for Hg speciation analysis in natural waters during past decades.However, to fulfill the field Hg determination onboard in simple, efficient, reliable and cost-effective manners is practically demanding.
An on-line determination of Hg with flow analysis technique is well useful for field/ shipboard Hg measurements in natural waters [16,25,26].It involves all analytical processes from sample introduction, chemical reaction (e.g.reduction, derivatisation), purge-and-trap, thermal desorption to detection/acquisition take place in a closed flow system.Once sample gas Hg is trapped (e.g.Hg 0 on a gold (Au)-coated, nano-Au trap), final detection is subsequently achieved by cold vapour (CV) atomic detectors (e.g.AAS, AFS) after on-line thermal desorption.The CVAFS due to high sensitivity, selectivity, and easy handling is recommended for the ultra-trace Hg analysis in natural waters by standard methods of the EN and US EPA [27].Therefore, coupling approaches/systems, based on the flow injection analysis with the CVAFS, availably provide the simple and reliable ultra-trace Hg analysis in natural waters for field measurements [28][29][30][31].
This paper aims to provide a cheap, simple coupling system that can be operated onboard for reliable ultra-trace Hg analysis in environmental waters.We hence developed the dual-channel flow injection mercury analyser (FIMA) for Hg analysis, which is an online flow injection purge-and-trap system combining aqueous reduction with stannous chloride (SnCl 2 ), two-stage gold (Au) amalgamation, and thermal desorption of Hg 0 together with cold vapour atomic fluorescence spectrometry (CVAFS).The FIMA with dual-channel detections effectively permits alternating field sample loading and analysis.Quantitative analysis of inorganic Hg species (e.g.dissolved elemental Hg (DEHg), reactive Hg (RHg) and total Hg (THg)) were reliably undertaken, after optimisation of the FIMA and validation by a certified reference material (coastal seawater BCR-579).The results of analytical performance were consistent between dual analytical channels and method practicability in the field demonstrated.Field results were further obtained from the onboard analysis of seawater samples from the East China Sea (ECS).

FIMA setup
The general scheme of the field FIMA is presented in Figure 1 which design is based on a coupling of the two-stage amalgamation approach (i.e.US EPA Standard Method 1631) with flow injection analysis (Figure 1).The FIMA is a portable dual-channel and purge-andtrap hyphenated system combining the following analytical steps: (1) sample introduction, (2) aqueous reduction of Hg 2+ with SnCl 2 , and (3) stripping of Hg 0 from the solution with a gas-liquid separator (GLS); (4) dual Au-amalgamation and (5) thermal desorption with a dual-channel trapping system, and (6) detection of Hg 0 with a cold vapour atomic fluorescence spectrometer (CVAFS) and ( 7) data acquisition with an integration software with a laptop/computer.All analytical stages of the FIMA are on-line connected in a closed system (mostly 3.2 mm outside diameter Teflon PFA tubing and fitting).All tubing and transfer line lengths were minimised to prevent condensation and increase detection efficiency.The FIMA consisted of two sampling/analytical channels that two 'sampling traps' (ST) and an 'analytical trap' (AT) were coupled by three six-way injection valves.The dual-channel systems, therefore, allow alternating sample loading and analysis.One detection channel #1 for Hg analysis is, for instance, operated in the Ar gas train of a twostage Au amalgamation (ST1 in injection-v1 and AT3-v3, Figure 1(a)).During the analysis, the next sample can be loaded up for subsequent trapping in channel #2 (i.e.ST2 in injection-v2, Figure 1(b)).Simultaneous and easy operation online for continuous measurements increases sample throughput, improves analytical reliability, and reduces contamination risk.Detailed information regarding design principle, used material and operational procedures of the FIMA can be found in those similar systems for elemental Hg analysis in environmental air and water samples [18,26].

Sample collection and processing
The field samplings in the ECS shelf were carried out on the R/V Ocean Researcher-I.The ultraclean sampling techniques and protocols were closely followed during the sample collection and analysis [7][8][9].Two 2-L acid-cleaned Teflon bottles for surface seawater (<1 m) were collected off the front of the ship while slowly moving forward into the flow.One 2-L aliquot of seawater was for the analysis of dissolved elemental Hg (DEHg, i.e.Hg 0 ) [18]; another one was for the rest analyses of Hg species, e.g.reactive Hg 2+ (RHg, i.e. easily reducible Hg, so-called labile Hg) [32], and total Hg (THg).
Standard operations of clean seawater samples after sampling for Hg species analyses were described as below: Firstly, a ~ 1-L unamended sample was immediately in-line transferred to an aluminium foil-wrapped borosilicate Gas-Liquid Separator (GLS, ~1.2 L volume) with N 2 purging for the DEHg determination in a closed circuit to avoid the ambient air Hg 0 contamination [18].Then, a ~ 100-mL of unamended sample for the RHg analysis, which was directly transferred to a smaller GLS (~250 mL volume) for purging, was then carried out together right after sample collection [30][31][32].Finally, ~100-mL unfiltered seawaters (1% v/v HCl) were analysed for the TM after the UV-assisted BrCl oxidation, which can quantitatively release organic bound-Hg from coastal and shelf seawaters [33][34][35].Briefly, unfiltered samples were oxidised with 0.05 mL BrCl solution and irradiated under 1200 W UV light for 1 hr at a temperature of ~80°C (18 subsamples at once).The oxidised sample was then neutralised with 0.2 mL NH 2 OH•HCl (20% w/v) to destroy extra halogen (Br 2 ) within 30 minutes before TM analysis.

Inorganic Hg species analysis
All Hg determinations were carried out on board by the FIMA with two Gas-Liquid Separators (Figure 1), using cold vapour atomic fluorescence spectroscopy (CVAFS).Briefly, surface seawater samples (~100 mL), following UV-assisted BrCl oxidation and NH 2 OH•HCl neutralisation for THg, or without reagents amended (~100 mL) for RHg analyses were transferred to a GLS.For either kind of sample, an aliquot of 0.1 mL of the acidic SnCl 2 solution (20% w/v in HCl 10% v/v) was then added to reduce Hg 2+ in solution to Hg 0 .Generated Hg 0 was purged from the GLS vessel by Hg-free N 2 gas (6 minutes at 0.3 L min −1 ) routing through a glass frit (~20 μm porosity, Figure 1) onto the Au-sand (~0.2 g, 40-80 mesh quartz sand) sample trap (ST1 or ST2, quartz tubing, 3.2 mm i.d., 13 cm length), which was wrapped with a Ni-Cr wire attached to an adjustable voltage transformer for heating.The Hg 0 was then thermally desorbed (~600°C) from the ST1 (or ST2) and onto an analytical trap (AT3) and from that detected finally by the FIMA detector (Tekran 2500 at a fluorescence of 253.7 nm) at an Ar flow of 30 mL min −1 for 4 min.The whole analytical process of Hg 0 determination from sample delivery to final detection took 10 minutes and was automatically controlled by Quick Chrom (SISC Inc.) software.As for the DEHg, a 1-L surface seawater sample without the addition of SnCl 2 was directly purged by N 2 at a rate of ~0.8-1 L min −1 for a complete purging of sample Hg 0 gas about 20 minutes [18].The subsequent thermal desorption and detection procedures for the DEHg determination by the FIMA are then the same as the analyses of THg and RHg mentioned above.

Standards
A stock standard solution of 1000 µg mL −1 Hg 2+ in 5% nitric acid, traceable to the U.S. National Institute of Standards and Technology (NIST), was purchased.The secondary Hg standard solution (~1 µg mL −1 Hg in 5% nitric acid) was prepared by diluting the stock Hg standard solution with Milli-Q water.Working Hg standard solution (1 ng mL −1  Hg in 1% nitric acid) was prepared regularly by quantitatively diluting the secondary Hg standard.All standard dilutions were performed on a volume basis and solutions were stored at 4°C.Standards for calibration curves were prepared by adding known amounts of Hg 2+ into acidified Hg-free seawaters after N 2 purging.In addition, the certified coastal seawater (BCR-579 from the Institute for Reference Materials and Measurements for Europe, certified value: 1.9 ± 0.5 ng L −1 ) was analysed during the THg analysis.

Blanks and calibrations
For trace-level Hg analysis, it is important to ensure that the reagents and materials used have low Hg blanks.All chemicals used were therefore traced metal grade and ultrapure Milli-Q water (18 MΩ, Millipore) was used throughout to clean bottles and prepare all reagents.The trace-metal purified reagent-grade concentrated nitric (67%) and hydrochloric (37%) acids (Superpure, Seastar, USA) were used.Solutions of bromine monochloride (BrCl, 1.5% KBr w/v + 1.1% KBrO 3 w/v,) were prepared according to Bloom and Crecelius (1983) [33] and hydroxylamine hydrochloride (NH 2 OH•HCl, 30% w/v) and stannous chloride (SnCl 2 , 20% w/v with 10% v/v conc.HCl) were prepared according to US EPA Method 1631 [13].The SnCl 2 and NH 2 OH•HCl (adding 0.1 mL SnCl 2 ) solutions were purged with Hg-free N 2 to remove any residual Hg 0 at 300 mL min −1 for 1 hour.Both the KBr and the KBrO 3 were muffled overnight at ~250°C to reduce the Hg content of the reagents.
All glassware and Teflon tubing, fitting, and vials/bottles were cleaned with the diluted Micro-90 detergent (alkaline cleaner, 1% v/v), thoroughly rinsed with DI water, soaked in a 10% HNO 3 solution for a few days, and finally rinsed with Milli-Q water before use.Cleaned sample tubing, vials, and bottles were placed in our Hg-free class 100 clean hood to air dry and then double bagged in new zip-lock bags before use.
For Au trap conditioning, before starting each analysis session, the Au traps were blanked initially by heating at 600°C to bring their blanks down to an acceptable level (≤ 1 pg) for 3 hours at an Ar flow of 10 mL min −1 .The soda-lime traps were blanked overnight by heating at ~100°C at an N 2 purging of 10 mL min −1 (blanks ≤ 1 pg).Once the blanks of Au and soda-lime traps were low and stable, analysis of calibration and real samples may begin.Generally, blanks from all analytical procedures were determined at the beginning of each run and between every five samples analysed.
Routine calibration was performed through gas Hg standards and aqueous working Hg standards.A known mass of Hg 0 gas standard before and during sample analysis was, for instance, injected into the Ar carrier stream through a Teflon injection tee just upstream of the sample or analytical traps.Calibration curves were also obtained by adding known amounts of Hg 2+ into acidified Hg-free seawaters after SnCl 2 reduction by the FIMA.Spiked recovery tests of the sample/analytical traps were conducted with injections in the carrier stream and with known Hg additions in Hg-free seawaters once per five samples.All total Hg analyses were validated against the BCR-579 certified coastal seawater and dual-channel analysis comparison.

Optimisation of the FIMA
The on-line FIMA analyser was originally custom-made with a compact and careful design.Before using in the field, we conducted optimisation experiments focused on the following areas: Thermal desorption efficiency, carrier flow rate effect, and purging efficiency, and evaluation of analytical performance in terms of inorganic Hg analysis.

Thermal desorption efficiency
Complete trapping efficiencies of Au-coated sand for Hg 0 have been well evaluated in environmental samples (e.g.air, waters, extracts) under different sampling/purging intervals (up to ~2 L min −1 ) [18,26].Thermal desorption needed to release Hg 0 trapped on Aucoated sand into the detection system via an Ar stream.The heating temperature is one of the important factors affecting the quality and quantification of the Hg 0 peak.Thermal desorption efficiency of the Au-sand trap was thus examined with a spike of ca.200 pg Hg 0 at a carrier Ar flow rate of 30 mL min −1 under the different heating temperatures controlled by a voltage-adjustable transformer.In brief, experimental results showed an elution peak of Hg 0 with a temperature profile, from room temperature to the final ~600°C kept, in 2 minutes (Figure 2(a)).Hg 0 signal was detected in about 30 seconds at heating to ~400°C and reached a maximum in 40 seconds.The Hg 0 peak looked sharp and nearly symmetric.
Heating the sand at different temperatures (from 170 to ~600°C) show that larger signals were obtained with a temperature of above 500°C according to the observed maximum PH (peak height) and PA/PW 1/2 (peak area/peak width at half peak height) ratio and low PW 1/2 (Figure 2(b); peak parameters defined in the supporting materials and Figure S1a).At low heating temperatures, Hg 0 was slowly released so that PH and PA/PW 1/2 were small, but PA and PW 1/2 large.That is, the peak shape was flat and broad with tailing so that signal quality was not good for quantification.Instead, high heating temperatures, especially when heated above 500°C, caused complete and rapid desorption of Hg 0 , so that the PH and PA/PW 1/2 were high and consistent with peak symmetries for better quantifications.In addition, both PH and PA/PW 1/2 had the analogous trend during temperature increase since the PA integral area over PW 1/2 was similar to the PH proportionally.In summary, the optimum heating temperature to desorb Hg 0 from the Au-sand was set at ~600°C in this study.

Carrier flow rate effect
The effect of the carrier flow rate during Hg 0 thermal desorption by heating to ~600°C was examined under the different flow rates of Ar (from 5 to 200 mL min −1 ).The experiment was conducted by injecting ~100 pg of Hg 0 onto the ST1, then desorbing onto the analytical trap and subsequently heating of that column at ~600°C and detecting by the CVAFS.The flow rate of Ar affected the residence time of Hg 0 in the cuvette and prominently determined the peak quality (Figure S1).When the flow was small (less than 20 mL min −1 ) and therefore cuvette residence times were long, higher values of the peak area (PA) and peak width at half-height (PW 1/2 ) were obtained and the peaks exhibited a lot of tailing and asymmetry (Figure 3(a,b); Figure S1b).In contrast, higher flow rates and shorter residence times gave peaks with smaller PA, PW 1/2 , peak height (PH), and PA/PW 1/2 .The peak shapes were more symmetric, but sensitivities were lower since signals were smaller.The optimal results of relatively large responses with good peak shapes were observed at Ar flows of 20-30 mL min −1 according to the values of maximum PH, PA/PW 1/2 , asymmetry factor (Af), and tailing factor (Tf) (measures of peak shape in supporting materials; Figure 3(a,b)).The optimum Ar flow rate of the experiment was chosen at a flow of ~30 mL min −1 .The flow rate of carrier Ar could be, however, modified according to the measurement needs of environmental samples and analytical strategies.

Purging efficiencies
Another aspect of the analytical system that was optimised was the efficiency with which Hg 0 was purged from the GLS (Figure 4(a)).The GLS and associated valving are illustrated in Figure 4(b).Two major factors were examined concerning the purging efficiency of Hg 0 : 1) the purging time, and 2) the flow rate of the purging N 2 used.Purging efficiency for Hg 0 converted from Hg 2+ was tested by analysis of synthetic aqueous solution (i.e.Milli-Q water, containing ~100 pg of Hg 2+ in 100 mL).The recoveries presented in Figure 4 (a) were made by the yields normalised to the sample Hg 0 concentration.The results showed that a complete stripping efficiency of Hg 0 took 6 min at a fixed N 2 flow of 0.3 L min −1 for a 100-mL sample.Additionally, the purging flow required to get a complete recovery was ≥0.2 L min −1 with a constant purging time of 7 min.
To understand how much the gas flow rate and purging time can strip out the dissolved Hg 0 completely (i.e. a volume ratio of purging gas/water sample), the theoretic calculation of the equilibrium system in the GLS was formulated.Assuming that the Hg 0 between the gaseous and aqueous phases reaches an equilibrium, the amount of Hg 0 decreased in the aqueous phase with purging time can be, therefore, expressed as Where the C W is the Hg 0 concentration in the solution (ng L −1 ); C a 0 is the Hg 0 concentration (ng L −1 ) in the purging N 2 that enters the GLS; C a is Hg 0 concentration (ng L −1 ) in the N 2 leaving the GLS; r a is the N 2 gas flow rate (L min −1 ), V is the sample volume (L), t is the purging time (min).Since the Hg 0 in the N 2 gas is less and tends to zero after passing the Au-sand trap, r a x C a 0 can be eliminated.By Henry's Law, C a is equal to C W x H ' (H ' is dimensionless Henry's law constant).Equation ( 1) can be then arranged as below: Two sides of Equation ( 2) are integral from t = 0, C w 0 to t = t, C w .Finally, the solution is It is known that the concentration of Hg 0 in the aqueous phase is related to gas flow rate, purging time, and sample volume.In Milli-Q water or seawater at different temperatures (i.e.range from 278 to 308 K), Henry Constant H' was empirically inversed with water temperature as shown below [36]: where T is the temperature (K).Combined ( 3) and ( 4), the recovery (%), determined by two variables: the gas flow rate (r a ) and the purging time (t) in the fixed conditions (e.g.temperature 298 K, solution volume ~0.1 L), can be expressed as: Complete Hg 0 recoveries (99.7%) were obtained at a purging N 2 flow rate of 0.3 L min −1 and purging time of 6 min for a 100-mL of water sample validated through both theoretic and experimental data (Figure 4(a)).In brief, recoveries of ~99% can be theoretically achieved with a volume ratio of purging gas/water sample of ~15.The volume of purging N 2 gas used in this study (1.8 L) was above that estimated by a theoretical approach (1.5 L), indicating full stripping efficiency was achieved.

Summary of optimised conditions
Briefly, the optimisation revealed a complete purging of Hg 0 from solution takes 6 min for a 100 mL sample at an N 2 flow of 0.3 L min −1 and then measuring Hg 0 at an Ar flow of 30 mL min −1 for 4 min at the heating of 600°C.The whole analytical process for Hg 2+ analysis sample delivery to final detection takes ~10 minutes.All in all, ~10 samples can be analysed per hour with a two-channel detection of FIMA for inorganic Hg 2+ analysis, following the recommended operation procedure shown in Figure 5.As for the dissolved Hg 0 , a complete stripping of sample Hg 0 gas from a 1-L sample shall take over 15 minutes at an N 2 purging rate of ~1 L min −1 .As a whole, the optimal conditions for inorganic Hg species analysis are briefly summarised in Figure 6.

Analytical performance and validation
The analytical performance of the FIMA for inorganic Hg analysis under the optimum working conditions was evaluated in terms of blank, reproducibility, recovery, and detection limit and then briefly summarised in Table 1.Analytical performances of both channels were identical in terms of data reliability (Figure 7(a)).Low blank values were obtained from the in-series two-trap blanks (i.e.ST1+ AT3, or ST2+ AT3, respectively) of one channel analysis performed in laboratory and field trip (≤5 pg, n = 30).Excellent reproducibility (relative standard deviation or RSD ≤ 5%) was achieved during the analyses of spiked gas   standard (~50 pg, n = 30), working aqueous Hg 2+ standard (~0.5 ng L −1 Hg, n = 30), and duplicate field samples (n = 60) for both channel systems.Calibrations were performed with a range of 5 duplicate gas standard Hg 0 injections and working aqueous standards from 40 to 800 pg.Excellent calibration (r 2 > 0.99) with a high precision (RSD <5%) were obtained.Good stability (RSD <5%, n = 30) in the slopes of calibration curves were obtained between June and July 2017 for field and laboratory tests.The FIMA has, therefore, a system detection limit of 0.5 pg Hg, defined as three times the standard deviation of the procedural blank (n = 6), at a constant Ar carrier flow of 30 mL min −1 .
The accuracy of the proposed methods was finally validated through the proper quality control by data comparison of the two channels (Figure 7(a), r 2 > 0.98), and analysis of a certified reference material (coastal seawater BCR-579, 1.9 ± 0.5 ng L −1 ).Recoveries of the BCR-579 were between 96% and 105% (101 ± 4%, n = 10; Figure 7(b)).Excellent reproducibility achieved was valid for the seawater samples analysed.Estimated detection limits for a 100-mL water sample were ≤0.05 ng L −1 for total Hg.The results demonstrate that the FIMA is a robust device for reliable examining and monitoring of THg in natural seawaters and the proposed methods are accurate and precise.

Application to marginal seawater samples
We have used the FIMA during one of the LORECS sampling campaigns (Long-Term Research of the East China Sea) to track the effects of river discharge and coastal upwelling on Hg levels in the ECS (Figure 8(a)) [30].The distributions of inorganic Hg species concentrations (e.g.THg: 1.4 to 14.0 ng L −1 (average: 4.6 ± 3.2); RHg: 0.07 to 0.7 ng L −1 (0.2 ± 0.1); DEHg 13 to 90 pg L −1 (30 ± 12), n = 27) in the surface seawaters of the ECS shelf with surface seawater salinity (SSS) contour were therefore investigated (Figure 8(b,  c,d)).The distribution patterns of the Hg species were generally similar to those of THg.High Hg levels in the area close to Changjiang river mouth (SSS <31) and coastal upwelling area (SSS >33) were identified through river-sea mixing examinations.Higher concentrations of THg (7.6 ± 4.2 ng L −1 , n = 4) were, for instance, found in the vicinity of Changjiang river mouth, and lower levels in offshore Kuroshio Water regions (2.0 ± 0.5 (n = 4), SSS >34), reflecting the inputs of Hg from the rivers.High levels of THg (9.9 ± 3.9 ng L −1 , n = 5) were also observed in coastal sites south to the Chien-Tang river mouth, where the monsoon-driven upwelling introduced nutrients and Hg into the coastal surface in summer.Overall, total Hg levels (average: 4.6 ± 3.2 ng L −1 ) in the ECS shelf waters were much higher than those in surrounding seawaters (~1-2 ng L −1 ).It implies the ECS shelf environment is facing significant Hg pollution.

Conclusions
The proposed FIMA technique is robust and suitable for shipboard Hg analysis in natural seawaters due to high sensitivity, excellent reproducibility, complete recoveries, low blanks, and simple handling.Simultaneous on-line operations of sample loading and analysis further make sample throughput fast and contamination eliminated.Excellent analytical performance of methods proposed was determined and methods were well validated by comparison with analyses of certified reference material and dual-channel detection.The field application to the East China Sea shelf prominently reveals marine Hg pollution with dynamic distribution and source attribution of Hg.Advanced development of the FIMA could however include continuously automated flow analysis and be versatile for the methylmercury analysis to fulfill high spatiotemporal resolutions of Hg species in coastal marine surveys.

Figure 2 .
Figure 2. (a) Elution peak of Hg 0 with a temperature profile; Index parameters of peak signal, PA: Peak area, PH: Peak height, PW 1/2 : Peak width at half peak height.(b) Optimisation of desorption efficiency of Hg 0 .Errors are ±1 SD (n = 3).

Figure 4 .
Figure 4. (a) Effects of purging times (circle, red-theoretical line) and N 2 flow rates (square, blue dashed-theoretical line) on the purging efficiency of Hg 0 .Purging time tests were done with 0.1 L of an aqueous solution at 0.3 L min −1 ; purging flow rate tests were done with 0.1 L of the solution at a constant purging time of 7 min.(b) Schematic of continuous purging in the gas-liquid separator (GLS).C a 0 , C a is the Hg 0 concentration in the purging N 2 entering and leaving the GLS, respectively; r a is the N 2 gas flow rate (L min −1 ); v is sample volume.Errors are ±1 SD (n = 3).

Figure 5 .
Figure 5. Recommended operation procedure of the FIMS for inorganic Hg 2+ analysis in natural samples.

Figure 6 .
Figure 6.General scheme of the applied methods for the determinations of the THg, RHg, and DEHg in natural waters.

Figure 7 .
Figure 7. Analytical validation of the FIMA by (a) data comparison for the THg and RHg between the dual-channel detections (1:1 dashed red line for reference), respectively, with linear regression lines and (b) quality accuracy analysis of THg with the CRM BCR-579 recoveries.

Table 1 .
Analytical performance of the FIMA for inorganic Hg analysis.