Analytical Performances of Nanostructured Gold Supported on Metal Oxide Sorbents for the Determination of Gaseous Mercury
Nanostructured gold supported TiO2, ZnO, and Al2O3 materials (1% w/w Au) were tested as sorbents for gaseous mercury (Hg) trapping and preconcentration. Their analytical performances were first compared with the one of traditional gold wool trap for the quantification of Hg standards injected into the argon flow followed by thermal desorption at 600°C and CVAFS detection. Good linearity and reproducibility were obtained, especially for Au/TiO2 material (; slope: 1.39) in the volume range of 10 to 60 µL (132–778 pg Hg). This latter even showed a better performance compared to pure Au in the volume range of 10 to 100 µL (132–1329 pg Hg) when the carrier gas flow was increased from 60 to 100 mL . The method detection limit (MDL) obtained with Au/TiO2 trap (0.10 pg ) was suitable for total gaseous mercury (TGM) determination. Au/TiO2 was, therefore, used in trapping and determining TGM in collected air samples. TGM values in the samples ranged from 6 to 10 ng . Similar results were obtained with the commercial gold-coated sand trap which showed an average TGM concentration of ng .
One of the most important environmental concerns of Hg is not only its toxicity but also its persistence and long-life in the atmosphere. Hg from point source emissions may remain localized in the environment or may be transported regionally and even globally . Thus, simple, rapid, sensitive, and selective detection of atmospheric Hg is required when assessing the potential human health risk from an exposure.
Ambient air may contain considerable amounts of Hg, which are generally present in the elemental form (Hg0) [2, 3]. Sampling and the subsequent analysis of atmospheric Hg, which generally occurs at the femtogram level, is often made as TGM and is performed with highly sensitive detection methods combined with preconcentration techniques . Gold and other precious metals are well known for their high efficiency in trapping Hg from the gas phase by amalgamation. Therefore, gold based collectors, which are used in many forms such as sand, wool, foil, wire, or deposits on different supports, play an important role in the preconcentration of Hg prior to detection [5–8].
In recent years, the catalytic properties of finely dispersed gold particles on oxide support materials have attracted much attention. Gold was rapidly recognized to be an extremely unique and highly active metal when prepared as supported nanoparticles [9, 10]. This is mainly due to the reduced dimensions of the gold particles and a strong interaction with the support [11, 12]. Thus, supports such as metal oxides are considered to be some of the materials adequate for preparing highly active supported gold catalysts . There is a strong feeling amongst catalyst researchers that in a number of specific areas gold has a potential to be applied commercially, including in pollution control.
Thus, different types of nanostructured gold (or nanogold) materials have been recently tested in many research laboratories as Hg probes and chemosensors in natural waters [2, 13–17] or as oxidation catalysts for Hg0 in flue gases [18, 19].
In this paper, three nanogold supported metal oxides materials (Au/TiO2, Au/ZnO, and Au/Al2O3) were used for the sampling, preconcentration, and determination of gaseous Hg and their performances were compared to those of commercially available, namely, gold wool and gold-coated sand, sorbents.
2. Materials and Methods
Gaseous Hg was determined by double amalgamation cold vapor atomic fluorescence spectrometry (DA-CVAFS) . Gaseous Hg species were trapped in a first column (sample trap) and desorbed to a second column (analytical trap) at 550–600°C. This latter usually contains pure gold that allows a more efficient thermal desorption (900°C) of the analyte to the detector. A transient signal was obtained with an increased signal/noise ratio. Detection limits were, therefore, lowered and the sensitivity was improved. The main advantages of the double amalgamation are found in the preconcentration of Hg at the second trap and in the reduction of interferences from water vapor and organic compounds during the first thermal desorption .
The calibration was done by injecting gaseous Hg standards directly to the analytical trap. The source of gaseous Hg was a drop of liquid Hg contained within a headspace bottle. The Hg drop was in equilibrium with the gaseous phase. A known volume of gas was collected with a microsyringe through a septum. The equilibrium concentration of Hg0 was, therefore, temperature dependent only and was measured with a thermocouple. The relationship between the vapor tension of Hg0 and the temperature is available in the literature . Therefore, the amount of injected Hg0 could accurately be determined.
Hg0 standards were injected and “trapped” using commercial gold wool (Au), provided by the Laboratoire de Chimie Analytique, Bio-Inorganique et Environnement (LCABIE-IPREM, France), or nanogold supported on metal oxides (Au/TiO2, Au/ZnO, and Au/Al2O3). The nanogold sorbents pellets were obtained from Mintek (South Africa). The analytical sorbents consisted of 0.1 g of 1% (w/w) gold particles (<10 nm clusters) dispersed on metal oxides support (Figure 1). The sorbents were initially grinded, weighed accurately, and then held securely in a length of quartz tube, which was connected to a temperature controlled heater.
A coil of resistance wire surrounded the quartz tube and was resistively heated with precise temperature control during the analysis step until the coil began to glow dull red.
The trap was heated to release Hg, which was carried to the AFS detector (Tekran 2500, Canada), in a flow of pure argon (60 mL min−1). Analytical performances of nanogold materials were first evaluated by injecting Hg0 standards and by comparing the obtained instrument calibration lines to the one of commercial pure gold. A triplicate injection was performed for a given volume of standard for statistical purposes.
The best performing analytical trap among the nanogold sorbents (i.e., Au/TiO2) was later used for the collection of gaseous samples and TGM analysis. The TGM determination was also carried out by double amalgamation cold vapor AFS (DA-CVAFS) and the performances of nanotraps were compared to those of commercial gold-coated sand traps. A summary of the analytical steps followed for both experiments is shown in Figure 2.
Simultaneous air collection for TGM measurements was done, as shown in Figure 3, with Au/TiO2 and gold-coated sand traps using a peristaltic pump (ASF THOMAS, Germany). The traps were connected in series to check on collection efficiency. The flow was controlled with a flowmeter (Bronkorst HiTech B.V. E-7000, Netherlands) in order to get an accurate measurement of the sample volume. The air sampling flow was 600 mL min−1 and a 0.1 μm quartz filter (Whatman) was used to prevent the intrusion of dust and aerosols in the traps. Samples were collected at the ground floor (F1), first floor (F2), second floor (F3), and the roof (R) of the laboratory building.
For the collection of ambient air, a volume gradient was established in order to optimize the sampling time (i.e., the sample volume) and therefore to insure the samples representativity. Short collection times of 10, 20, and 40 min were used which corresponded to sample volumes of 6, 12, and 24 L (at a flow of 600 mL min−1). The air sampler was conditioned every day prior to sampling for stabilization and also for minimizing Hg adsorption on the walls. Sampling traps were later hermetically closed, stored in a double plastic bag, and analyzed as soon as possible.
It was necessary to perform a quality control before analysis and thus minimize analytical errors that could affect the Hg determination at such low level (memory effect, contamination, etc.). Therefore, analytical and field blanks were processed through the entire series of sampling and analytical steps which allowed us to determine the different method detection limits (MDL) and the methods limit of quantification (MLQ). Blanks were obtained by either passing UHP grade argon to the traps (analytical blank) or by replacing Whatman filters with the sampling sorbent in the sampling line (field blank). For the latter, the air passing through the sampling trap was, therefore, considered “Hg-free.” Field blanks traps were treated in the same way as for actual air samples collection. The resulting “blank” values were subtracted from the analytical results.
3. Results and Discussion
3.1. Analytical Performances
An example of signals obtained during analyses is presented in Figure 4.
A good linearity was obtained on the three nanogold materials, especially for Au/TiO2, in the volume range of 10–60 μL Hg0, which corresponds to a concentration range of 131.8–777.5 pg Hg0 at an argon flow of 60 mL min−1 (Figure 5). In addition, signals obtained from triplicate injections of Hg0 standards when using Au/TiO2 as the analytical trap (Table 1) have demonstrated an excellent repeatability with RSD values always below 10%.
Other parameters such as retention time, number of theoretical plates, and full duration at half maximum (FDHM) height also have been studied and are presented in Table 2.
The linearity range of Au/TiO2 was even improved by increasing the carrier gas flow to 100 mL min−1. Calibrations with Au/TiO2 trap also exhibited a better slope, that is, better analytical response from the detector, under the above conditions compared to the one obtained with gold wool as shown in Figure 6.
It appears, based on the above results, that commercial Au wool presents the best characteristics: sharp peaks, best regression coefficient, and highest number of theoretical plates. Au wool was, therefore, considered to be the best analytical trap followed by Au/TiO2.
It is important to notice that, although chromatograms obtained with Au/ZnO and Au/Al2O3 had retention times similar to the one for Au wool, these materials showed lower linearity range compared to Au/TiO2 with lower slopes (Figure 5 and Table 2). They also have demonstrated poor baseline recovery after Hg desorption (Figure 7).
This implies a great retention of Hg vapor in the sorbents. Moreover, Au/Al2O3 materials exhibited hygroscopic properties (water retention) once exposed to the ambient air. Finally, Au/ZnO and Au/Al2O3 sorbents have shown different analytical responses (poor reproducibility) when the materials were roughly or finely ground. The difference in analytical performances observed between the nanosorbents could be attributed to the inner structure of the materials used, although this was not investigated in this study.
Numerous experimental and theoretical studies aimed at a better understanding of the properties of gold in the nanometer size regime have suggested that adsorption properties of gold catalysts are deeply influenced by the pore size and specific surface area of the material (see  and the references therein). Gold can be present as individual particles in the catalyst or can form agglomerations or clusters.
The size and the shape of the gold aggregates depend on the temperature and can be metastable once formed at low temperatures. These factors directly affect electronic, optical, and chemical characteristics of nanogold materials. In the case of the ultrathin oxide films, it was shown that the metal substrate lying underneath the film may affect the properties of gold particles, thus leading to an adsorption behavior that depends on the oxide thickness . This could explain the fact that different adsorption capacities were observed with some sorbents (e.g., Au/ZnO), depending on the physical pretreatment.
Finally, studies have also shown that water strongly interacts with oxygen atoms in gold supported metal oxides to make a water-oxygen complex or hydroxyls [23–25], although Quiller and coworkers  have suggested that isolated stable hydroxyls may not be formed and could be more transient in character. The susceptibility of nanogold supported metal oxides materials to act as a Brønsted base or a nucleophilic base , due to the presence of oxygen atoms, could explain the “wetting” observed with the Au/Al2O3 sorbent. A deeper characterization of these materials is, therefore, of importance in order to improve their performance.
It can be seen from the above observations that Au/TiO2 has demonstrated better analytical performances compared to Au/Al2O3 and Au/ZnO. This is the reason why Au/TiO2 was selected to be tested as a sampling trap under real environmental conditions.
3.2. TGM Sampling and Determination
As mentioned previously, in order to optimize the air sampling volume, a volume gradient was established using short collection times of 10, 20, and 40 min which corresponded to sample volumes of 6, 12, and 24 L at a flow of 600 mL min−1. The obtained Hg concentrations (mean value: ng m−3) were all beyond the method analytical performances (see Table 3). Due to sample variability caused by the air circulation in the building, the difference in Hg concentration (±8%) for the different sampling volumes was not significant (Figure 8). The optimized sampling time was, thus, chosen to be 20 min, that is, 12 L in volume.
The obtained methods detection limits (MDL) and methods limit of quantification (MLQ) are shown in Table 3.
Detection limit values were excellent and suitable for the detection of Hg0 at background level. It has to be recalled that TGM levels for background continental areas were reported to be in the range of 1.0 to 4.0 ng m−3 [28, 29].
Results of the TGM measurements performed on the collected air samples from the laboratory environment are presented in Table 4.
TGM concentrations measured with both Au-coated sand and Au/TiO2 traps were very similar with mean values of ng m−3 (range: 6.3–8.6 ng m−3) and ng m−3 (range: 6.0–10.1 ng m−3), respectively. Due to the range of TGM measured and the variability in samples (air recirculation in the building, climate changes, and activities in the laboratory), the obtained values were not considered to be significantly different. Moreover, statistical -test and test have been used to compare replicates measurements as well as the different standard deviations obtained with both gold-coated sand and Au/TiO2 traps. Results (Table 5) show that there is no significant difference between both replicates measurements and standard deviations for all the samples. Indoor (L1, L2, and L3) and outdoor (R) concentrations were almost similar, although the variability in R was higher than in L samples for both traps. This might be caused by changes in environmental conditions during sampling (wind direction and speed, air temperature, point sources, etc.).
It is well known that TGM exhibits an important diurnal variability with generally a maximum peak at midday . The obtained average TGM values were within the range of reported mean TGM concentrations from urban areas [31, 32]. It is important to recall, here, that the collection efficiency of TGM using Au-coated quartz sand has been reported to be ≥95% . Therefore, the closeness of TGM values measured with Au/TiO2 and the commercial Au-coated sand in the majority of the samples demonstrates the successful application of the nanogold trap under real environmental conditions.
Au/TiO2 has demonstrated a consistent response after being frequently used for about 3 months depending on storage conditions. A degradation of the adsorption efficiency (drop of signal intensity for a given concentration) was observed when the sorbents were exposed to the light for several hours.
This study has demonstrated the successful application of nanostructured gold supported in metal oxides materials for the trapping and preconcentration of Hg directly from the gaseous phase. Au/TiO2 has shown better analytical performances compared to Au/ZnO and Au/Al2O3, although it has also exhibited some retention problems. The superior performance of Au/TiO2 may relate to the grain morphology of TiO2, dispersion of gold particles, and the architecture of metal/oxide junctions.
The excellent MDL obtained with Au/TiO2 makes it suitable for background TGM determination. Environmental measurements of TGM with Au/TiO2 were similar to those obtained with traditional gold traps.
The development of an analytical procedure for TGM determination using Au/TiO2 sorbents was achieved in a performing and cost effective way compared to current methods since only 1% (w/w) of gold was required for the preparation of the materials. Finally, a deeper characterization of the inner structure of the studied materials is crucial to understand the different behaviors observed and to improve, where necessary, the performance of the materials.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors thank Mintek (South Africa) for providing the nanosorbents, NRF (South Africa), and CNRS-LCABIE-IPREM (France) for funding the project.
US EPA, “Mercury study report to Congress,” EPA 452/R-97-003, 1997.View at: Google Scholar
G. C. Bond and D. T. Thompson, “Catalysis by gold,” Catalysis Reviews: Science and Engineering, vol. 41, no. 3-4, pp. 319–388, 1999.View at: Google Scholar
B. Zhou, S. Hermanans, and G. A. Sormojai, Nanotechnology in Catalysis, vol. 1-2, Kluwer Academic, Boston, Mass, USA, 2004.
W. Seames, M. Mann, D. Muggli et al., “Mercury oxidation via catalytic barrier filters: phase II final report,” Award no. DE-FG26-04NT42188, Department of Chemical Engineering, University of North Dakota, Grand Forks, ND, USA, 2007.View at: Google Scholar
N. S. Bloom and E. A. Crecelius, “Determination of mercury in seawater at sub-nanogram per liter levels,” Marine Chemistry, vol. 14, no. 1, pp. 49–59, 1983.View at: Google Scholar
OSPAR Commission, “JAMP guidelines for the sampling and analysis of mercury in air and precipitation,” Monitoring Guideline 1997-8, 1997, http://www.ospar.org/documents/dbase/decrecs/agreements/97-08e.doc.View at: Google Scholar
D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, Fla, USA, 86th edition, 2005.
M. A. Lazaga, D. T. Wickham, D. H. Parker et al., “Reactivity of oxygen adatoms on the Au(111) surface,” in Catalytic Selective Oxidation, vol. 523 of ACS Symposium Series, pp. 90–109, 1993.View at: Google Scholar
D. A. Outka and R. J. Madix, “Broensted basicity of atomic oxygen on the gold(110) surface: reactions with methanol, acetylene, water, and ethylene,” Journal of the American Chemical Society, vol. 109, no. 6, pp. 1708–1714, 1987.View at: Google Scholar
W. F. Fitzgerald, G. A. Gill, and A. D. Heurot, “Air-sea exchange of mercury,” in Trace Metals in Seawater, C. C. S. Wang, E. Boyle, K. W. Bruland, J. D. Burton, and E. D. Goldberg, Eds., pp. 297–315, 1984.View at: Google Scholar
O. Lindqvist and H. Rodhe, “Atmosphere mercury—a review,” Tellus, vol. 37, no. 3, pp. 136–159, 1985.View at: Google Scholar
R. Ebinghaus, H. H. Kock, S. G. Jennings, P. McCartin, and M. J. Orren, “Measurements of atmospheric mercury concentrations in Northwestern and Central Europe—comparison of experimental data and model results,” Atmospheric Environment, vol. 29, no. 22, pp. 3333–3344, 1995.View at: Google Scholar