Monitoring Atmospheric Atomic Mercury by Optical Techniques

Abstract

Mercury is a serious neurotoxic agent, and the control and monitoring of emissions are important. Optical spectroscopy is a powerful technique for measurement of mercury, which in the atmosphere predominantly appears in atomic form. The mercury resonance line close to 254 nm can be utilized in long-path absorption measurements of average concentrations or in light detection and ranging (lidar) studies, where range-resolved concentration values can be obtained. In addition, point monitors often use optical detection, frequently after pre-concentration, as well as for mercury compounds in conjunction with denuders, which transfer the compounds into atomic mercury. The present review discusses mercury measurement methods with respect to merits and sensitivity to interference. The main focus is on remote-sensing techniques, and many examples from industrial and mining monitoring are given. Further, mercury emissions related to the extraction of geothermal energy are discussed. Finally, an example from archaeology—the Qin tomb in Xi’an—is presented. Advanced measurement techniques can help in shaping an environment largely free from mercury contamination. Further, the aspect of mercury being an important geophysical tracer gas can also be exploited.

1. Introduction

The monitoring of atmospheric mercury pollution by optical spectroscopic techniques will be covered in this review, and special emphasis will be placed on non-intrusive remote-sensing approaches.

Mercury is a serious neurotoxic pollutant that during recent years has attracted significant international attention (see, e.g., [1,2,3,4]). Mercury has had many uses in instrumentation and devices, such as thermometers and electrical relays [5], and has been used extensively in chlor-alkali electrochemical plants as a liquid electrode in the production of sodium hydroxide and chlorine [6]. A further major source of mercury pollution is the combustion of coal [7]. Industrial mercury mining and non-ferrous metal production activities as well as the extraction of geothermal energy are also associated with the release of mercury. However, the largest amount of mercury presently enters the environment through artisanal gold mining. Heating in primitive ovens of the main ore, cinnabar (HgS), and subsequent condensation of the released mercury vapor lead to major atmospheric releases, while extraction of gold from sand using liquid mercury for amalgamation also strongly affects water and soil [8]. Historically, mercury compounds were also used as anti-fungal agents for seeds in agriculture. Serious intoxication events, mostly related to methyl mercury (CH₃)₂Hg in, e.g., Iraq [9] and Japan [10] with numerous fatalities have drawn special attention to the mercury hazards. In view of the obvious detrimental effects on human health and the biosphere in general, the Minamata Convention from 2013, which regulates the phasing out of the use of mercury, has been adopted by a large number of states [11]. However, mercury pollution is still a matter of major concern. Much research is performed worldwide regarding mercury and its detrimental effects, and a biennial international conference on mercury pollution is being convened, with its latest version in 2022 [12].

The release of mercury into the environment is discussed in many publications with global and regional estimates [13,14,15]. The speciation of mercury compounds and the conversion in the environment between different mercury compounds are also the subject of much investigation [16,17]. Regional studies of mercury pollution are reported, e.g., in [18,19,20,21]. Transboundary transport makes the mercury pollution a true global problem [22].

A special feature of mercury, and in strong contrast to other pollutants, is the fact that the atmospheric contents to a very high degree (90% or more) is present in atomic form (gaseous elemental mercury (GEM) - Hg⁰), while molecular compounds (reactive gaseous mercury (RGM)), such as HgO, HgCl₂, HgBr₂ and Hg(OH)₂, and particle-bound mercury (total particulate mercury (TPM)) only account for a small fraction. Metallic mercury has an extraordinarily high vapor pressure, leading to substantial evaporation into the atmosphere even at room temperature. Filled d- and s-shells, and the absence of a resulting atomic spin in the ¹S₀ ground state (similarly as in the inert gases), explain its low inherent reactivity. The slow process of transformation into the most hazardous methyl mercury clearly is of major interest [23,24,25].

Being a free atom makes atomic mercury (Hg⁰) very suitable for extremely sensitive optical monitoring using the strong resonance line in the transition 6s6p ³P₁ - 6s^{2 1}S₀ close to 254 nm. The present review will address different aspects of optical spectroscopy in performing the monitoring tasks. General methods for measurements of atmospheric mercury are reviewed in [26], including mass spectroscopy, chromatography, X-ray fluorescence in addition to optical techniques. RGM and TPM require specific monitoring methods, as discussed, e.g., in [1,2,17,27,28] and fall outside the scope of the present paper. However, it should be noted that instrumentation for such purposes frequently employ so-called denuders to split the molecules to free mercury atoms, which again finally are detected by optical techniques (e.g., mercury vapor fluorescence spectroscopy in TEKRAN instruments).

We will in the next section describe optical in-situ monitoring techniques and approaches to remote sensing. Our focus is on relative merits and sensitivity to interference. Then, in Section 3, examples from mercury monitoring campaigns, mostly using remote-sensing techniques, will be given for different application fields, including industrial, geophysical and archaeological monitoring. Conclusions will then be drawn in a final section.

2. Optical Techniques for Atomic Mercury Monitoring

2.1. Introduction

Optical spectroscopic techniques are powerful in wide areas of science, including fundamental atomic physics, astronomy, environmental remote sensing and analytical chemistry. High specificity, high sensitivity and real-time data delivery are prime attractive features, which certainly strongly pertain to the context of mercury monitoring and often make them methods of choice in comparison to non-optical approaches. Because of comparatively scarce line structures of free atoms in comparison with the very complex vibrational–rotational spectra of molecules, the analysis of free atoms is particularly straightforward and is widely used in commercial atomic absorption spectroscopy instrumentation. The many facets of atomic and molecular spectroscopy in numerous applications are illuminated in monographs such as [29,30,31,32].

2.2. The Atomic Mercury 6s6p ³P₁ - 6s^{2 1}S₀ Transition Close to 254 nm

We will now discuss the detailed internal structure of the predominantly employed atomic mercury transition close to 254 nm between the ground state 6s^{2 1}S₀ and the excited state 6s6p ³P₁. The transition probability and associated oscillator strength of the transition are largely determined by the undisturbed atom upper-state lifetime, which is about 125 ns [33]. At atmospheric pressure, collisional effects cause a reduction in the excited state lifetime and associated broadening of the spectral line and a reduction in the fluorescence yield due to quenching (non-radiative transitions).

Natural mercury consists of seven stable isotopes, of which ²⁰⁰Hg and ²⁰²Hg are the most abundant ones (23 and 30%, respectively). Isotopic shifts slightly separate the transitions of the constituents, and, in the presence of a non-zero nuclear spin (1/2 and 3/2 for ¹⁹⁹Hg and ²⁰¹Hg, respectively, with abundances 17 and 13 %), there is also an additional hyperfine structure. These structures were early investigated by classical Fabry-Pérot interferometry and more recently by high-resolution laser spectroscopy [34,35,36]. Figure 1 shows the resulting structures in laser absorption spectroscopy. Spectra of mercury gas in an evacuated cell are shown for mercury with the natural isotopic composition as well as for separated isotope samples. The resulting strongly broadened profile for mercury in ambient air is displayed [34].

Environmental monitoring and in particular remote-sensing applications are generally performed in ambient atmospheric conditions, and the pressure-broadened composite line is the one of interest. It is also important to consider spectral interference with extremely weak (forbidden) atmospheric molecular oxygen lines, as discussed in [37,38] and indicated in Figure 2. For low mercury concentrations measured over long path lengths, such interference may cause appreciable measurement errors if not accounted for. The influence of linewidth of the interrogating spectral light (laser source or filtered classical light source) in generating a spurious background mercury concentration is elucidated in [37].

2.3. In Situ/Point Monitoring

Most measurements of atmospheric mercury are performed by in-situ monitors, where ambient air is extracted through a gas inlet into the measurement device. Clearly, it is then very important that the gas transport system does not influence the gas composition. In contrast, remote-sensing devices interrogate the free gas over ranges extending to 1 km or more. Highly sensitive monitoring devices are needed, since the global background concentration of mercury is only 1–2 ng/m³ [21,39].

2.3.1. Absorption Approach

Optical absorption measurements are governed by the Beer–Lambert law (see, e.g., [29], Section 6.5.1), stating that the attenuation of the interrogating optical radiation is exponentially decreasing and dependent on the product of the (uniform) gas concentration c and the path length l:

$$\frac{I_t}{I_0}=e^{-\sigma cl}$$

(1)

Here, I₀ is the initial light beam intensity, while I_t is the resulting one after passage of the sample. We realize that a lower concentration can be compensated by a longer path length. The size of the spectral imprint (fractional attenuation) depends on the spectral absorption coefficient σ(λ). σ has a very high value for mercury, which allows the measurements of very low concentrations. The law is strictly valid only if the spectral linewidth of the light source is very small compared to the absorption profile. This is certainly not the case for the low-pressure mercury spectral lamps used in practical point-monitoring devices. This can be handled by calibration towards known concentrations using supplied standard concentrations of mercury to the instrument. This in turn requires a precision knowledge of the mercury vapor pressure, which is determined by the temperature.

Measurements of ambient mercury concentrations using devices with short optical path lengths require pre-concentration of the mercury gas, also to avoid the molecular oxygen interference problem. Pre-concentration is frequently achieved by amalgamation in gold foils, over which known amounts of ambient gas is pumped. After accumulation in the gold material, strong and well-defined ramped-up electrical heating causes all the mercury to escape in a “puff”, the presence of which is recorded by the instrument. For calibration, saturated air from a mercury-containing flask of well-defined temperature can instead be pumped into the system in a similar way, and the absorption transients are compared for evaluation of the unknown concentration. Commercial equipment, e.g., GARDIS instruments, operate on the atomic absorption principle.

2.3.2. Fluorescence Approach

Mercury atom absorption of the excitation light derived from a low-pressure mercury lamp is followed by the release of fluorescence light at the same wavelength, around 254 nm, back to the ground state. A fluorescence instrument is described in [40]. At low pressures, the process is 100% efficient. However, since fluorescence is partly quenched by ambient air collisions, calibration against gas of known concentration is again needed. Since absorption acts on the majority of the atoms, which reside in the ground state, rather than on the tiny fraction of atoms in the excited state, giving rise to fluorescence, absorption measurements should be the most sensitive approach. However, since low concentrations give rise to only a very tiny intensity reduction in a high light flux and since there are always instabilities due to technical aspects, it can be preferable to instead use the fluorescence process, where the signal rises up from an ideally zero background. Stray light can in practice cause a non-wanted background offset. We note that fluorescence is less sensitive to oxygen interference, since molecular fluorescence occurs at multiple wavelengths. Fluorescence monitoring is employed, e.g., in commonly used TEKRAN instruments, which can be equipped with different denuders, allowing the measurement of also RGM and TPM in addition to the dominant TGM (Hg⁰).

Both the absorption and emission approaches, often referred to as cold vapor atomic spectrometry, require accurate mercury vapor pressure data.

2.3.3. Correlation Approach

Spectral correlation instrumentation utilizes specific modulation in light passing the gaseous sample. The modulation is directly correlated with the presence of the studied gas, which absorbs part of the radiation (see, e.g., [29], Section 6.6.1). Spectrally resolving instruments utilize a mask, which periodically blocks absorbed and unabsorbed wavelengths (COSPEC Instruments, [41]). They utilize ambient solar or sky radiation, which is available down to the stratospheric ozone cutoff at 300 nm. The molecular gases SO₂ and NO₂ can conveniently be measured. However, the techniques are not directly applicable to Hg, which requires a very high spectral resolution and light available in the deep UV wavelength range. Gas correlation instrumentation operates on the principle that “nobody knows the absorption spectrum of a gas better than the gas itself”. The incoming light is alternatingly passed directly to a detector and through a gas cell with sufficient cl product (cf., the Beer–Lambert law, Equation (1)) to make the gas opaque for the specific absorbing wavelengths of the gas under study. If the system is first balanced out for the case of clean air (no specific absorptive imprint by the pollution gas in the received radiation), there is an imbalance (modulation) in the light intensity in the presence of external pollution, since the external gas absorption does not influence the signal through the correlation cell, which already fully blocks those spectral components. Line-of-sight gas correlation instruments of this type have been developed [42,43], and the principle was also extended to imaging [44], which is of particular interest in the infrared spectral region with mapping of, e.g., methane in the natural-gas-handling facilities and the petrochemical industry [45].

The gas correlation principles also apply to atomic mercury monitoring, but the short-wavelength radiation at 254 nm must then be supplied by a suitable source. Multimode diode lasers can be mixed to the 254 nm region in a similar way as employed in the work described in Figure 1 [46,47]. Spectral and temporal fluctuation patterns can be employed for specific gas identification [48].

Gas correlation light detection and ranging (lidar) with a broad-band laser system has even been demonstrated for mercury. The recorded light transients from the ambient atmosphere are detected directly and through a blocking mercury cell, and the signals are compared [49]. Such techniques were also demonstrated for methane [50].

A special correlation technique dedicated to atomic mercury utilizes the Zeeman effect, which splits up a zero-nuclear-spin atom in the ³P₁ state into three components resulting in the emission line from a mercury lamp in a strong magnetic field to split up into two equally strong circularly polarized components of opposite helicity when observed in the magnetic field direction. A separated isotope introduced in the lamp is spectrally displaced with regard to the pressure-broadened composite absorption profile of naturally occurring mercury. Only one strongly displaced Zeeman component is affected by the absorption by mercury, if present. By using a polarization modulator in combination with lock-in electronic techniques, the presence of mercury in the gas sucked into a multi-pass absorption cell is manifested as a modulation [51,52] and can be detected with high sensitivity. The technique, commercially denoted LUMEX, is capable of real-time monitoring of mercury without any pre-concentration, once molecular oxygen is compensated for.

2.4. Remote Sensing

While the local gas concentration at a particular place can be measured by in situ/point monitoring, information on gas concentrations at locations other than the place of the measurement equipment (remote sensing) can be obtained. Basically, two different approaches can be taken: long-path optical absorption or differential absorption lidar (DIAL).

2.4.1. Long-Path Optical Absorption

The long-path absorption method uses an optical transmitter and a receiver, recording light that has travelled over the chosen path length, which can be 1 km or more. The light source and the receiver can be separated by the path distance or, more conveniently, placed next to each other, with the transmitted light beam being returned back via a retroreflector placed at half the distance of the effective path length. Since all photons travel the same path, it is not possible to obtain range-resolved path-length information—only the average concentration c_av is obtained as evaluated using the Beer–Lambert law, Equation (1).

The required spectral resolution can be obtained using classical light sources (normally a high-pressure xenon lamp) in combination with a high-resolution spectrometer for more forgiving cases, where the absorptive structures are not very sharp. This is the case for gases such as SO₂ and NO₂. Because of atmospheric turbulence, special means have to be adopted to be able to distinguish minute absorption imprints. This can be achieved by very fast repetitive spectral scans using a flying-slit arrangement, or by using a CCD or CMOS detector. The method is then normally referred to as differential optical absorption spectroscopy (DOAS) [53,54]. Commercial equipment is available (e.g., OPSIS), and numerous measurements have been performed on urban pollution, industrial emissions, etc. The measurement approach and monitoring examples are presented in some detail in [55].

Because of the very narrow atomic mercury absorption line and the presence of partly overlapping molecular oxygen lines (see Figure 2), DOAS monitoring of ambient mercury concentrations becomes difficult. The situation is quite different in strongly polluted areas, where the oxygen interference becomes more tractable. Measurements with the DOAS technique at chlor-alkali plants with strong mercury emissions are described in [56].

Long-path absorption measurements of low ambient concentrations can instead be performed using narrow-band laser radiation, where the wavelength can be spectrally adjusted within the complex line profile, where the interference is negligible (Figure 2). The setup would then basically be as the one used for achieving the data in Figure 1, but with a considerably longer path length. Clearly, efficient non-linear generation of the relevant radiation is very helpful, and technology for this purpose is swiftly maturing.

2.4.2. Differential Absorption Lidar (DIAL)

Light detection and ranging (lidar) [57,58] is the optical counterpart to radar and is widely used, e.g., in distance measurement and monitoring of atmospheric particulates. Compact lidar systems are now making a major impact in the development of autonomously driving vehicles [59]. A short laser pulse is transmitted into the atmosphere and an echo of a target at distance l is received after a time

$$t=\frac{2l}{c_{light}}$$

(2)

where c_light is the velocity of light. A shorter pulse results in a more accurate distance measurement. Monitoring of atomic mercury by lidar techniques was first demonstrated by the Swedish group in the early 1980s [60,61]. Data from lidar recordings of mercury across a valley in Toscana are included in Figure 3 [62]. Monitoring the height of the distinct echo from the vegetation-covered side of the valley as a function of wavelength can be used to evaluate a mean concentration value, basically as in long-path absorption measurements but now without the need for a physical retroreflector. At closer range, Mie scattering from particulates and Rayleigh scattering from major molecular constituents show up as a distributed, range-resolved signal. It can be considered that there is a minute back-reflecting mirror at any range, causing a lidar echo. The atmospheric signal reduction over shorter paths can likewise be used to evaluate average concentrations as indicated in the Figure 3.

The full use of atmospheric backscattering for range-resolved measurements is illustrated in Figure 4. The received atmospheric signal P(R, ΔR) from a range interval ΔR at range R is given by the general lidar equation, which can be written as

$$P\left(R,\Delta R\right)=CW{N}_b\left(R\right){\sigma }_b\left(R\right){\sigma }_b\left(R\right)\frac{\Delta R}{R^2}exp\left[-2{{\displaystyle \int }}_0^R\left[\sigma \left(\lambda \right)N\left(r\right)+K_{ext}\left(r\right)\right]dr\right]$$

(3)

Here, C is a system constant, W the transmitted power and N_b(R) the number of backscattering particles at range R with backscattering cross-section σ_b(R). The final expression describes the beam attenuation over the interrogation range R due to molecules of concentration N(r) and wavelength-dependent cross-section σ(λ). Finally, K_ext(r) describes the attenuation due to particles.

Figure 4. Schematic representation of the differential absorption lidar (DIAL) technique. (a) Measurement scenario with two industrial plumes containing a pollutant gas, which emerge into a uniform atmosphere of backscattering particles. (b) Corresponding lidar backscattering curves recorded for on- and off-resonance wavelengths for the gas under study. (c) Ratio (DIAL) curve for on- and off-resonance recordings. Gas plumes manifest themselves as pronounced reduction in the ratio value. Note that if the atmosphere contained localized clouds of particles, the excess signal would still not show up in the DIAL curve (see Equation (4)). (d) Evaluated range-resolved gas concentration curve obtained by applying Equation (1) over chosen range-integration intervals ([63], reproduced with permission from OPTICA).

Figure 4. Schematic representation of the differential absorption lidar (DIAL) technique. (a) Measurement scenario with two industrial plumes containing a pollutant gas, which emerge into a uniform atmosphere of backscattering particles. (b) Corresponding lidar backscattering curves recorded for on- and off-resonance wavelengths for the gas under study. (c) Ratio (DIAL) curve for on- and off-resonance recordings. Gas plumes manifest themselves as pronounced reduction in the ratio value. Note that if the atmosphere contained localized clouds of particles, the excess signal would still not show up in the DIAL curve (see Equation (4)). (d) Evaluated range-resolved gas concentration curve obtained by applying Equation (1) over chosen range-integration intervals ([63], reproduced with permission from OPTICA).

When there is a distributed absorptive gas in the atmosphere, the backscattered light intensity is sensitive to whether the laser source is tuned to the absorptive wavelength λ_on or at a neighboring less-absorbed one, λ_off. This is reflected in the wavelength-dependent absorption coefficient σ(λ) in Equation (3). Actually, by sequentially tuning the transmission laser to the on-resonance wavelength λ_on and the reference wavelength λ_off and dividing the two lidar signals (after accumulating a large number of transients, with fast switching back and forth between the two wavelengths λ to reduce the influence of atmospheric turbulence), the DIAL dimension-less expression is obtained from Equation (3):

$$\frac{P_{{\lambda }_{on}}\left(R,\Delta R\right)}{P_{{\lambda }_{off}}\left(R,\Delta R\right)}=exp\left[-2\left[\sigma \left({\lambda }_{on}\right)-\sigma \left({\lambda }_{off}\right)\right]{{\displaystyle \int }}_0^{R}N\left(r\right)dr\right]$$

(4)

Here, σ(λ_off) can be zero if the reference wavelength is totally non-absorbing. We notice that the influence of the largely unknown parameters is eliminated in this process (the wavelength dependence of N_b(R), σ_b(R) and K_ext(r) is negligible for a small value of Δλ = λ_on − λ_off). In particular, the varying backscattering from particulates, which might be quite non-uniformly distributed, is cancelled out by forming the ratio. The slope of the DIAL curve indicates the presence of the gas. Finally, the range resolved concentration of the particular pollutant matching the λ pair is obtained by differentiation. The full procedure is illustrated in Figure 4. DIAL techniques are discussed in [64,65].

Range-resolved concentration measurement combined with easy change in measurement direction makes it possible to assess the total flux from an area/installation by combining with wind velocity data, as will be further discussed below. This is a unique capability of the DIAL technique, as is the possibility to achieve vertical concentration profiles, which are not readily accessible with other methods.

Clearly, the lidar recordings are subject to noise, and in order to obtain significant results, the evaluation is performed over certain ΔR intervals. The DIAL curve is clearly a reflection of the Beer–Lambert law, evaluated over appropriate range intervals, where the initial and final intensity are provided by the corresponding backscattering ratio values. The accuracy of the evaluated concentration value depends on the size of ΔR. The range interval over which the evaluation is performed must be increased for low concentrations c.

As is clear from Equations (1) and (3), the evaluated mercury concentrations directly depend on the applied value of σ, since only the product σc l is experimentally measured. The parameter σ can be determined in careful absorption measurements performed in the laboratory, where then the mercury vapor pressure (the gas concentrations) must be accurately known. Experimental calibration efforts are described in [35,38]. Concentration values obtained by DIAL were on a few occasions directly compared with readings from commercial monitoring equipment, which was arranged with the gas inlet hose placed directly under the DIAL laser beam [66]. The results of such measurements are illustrated in Figure 5. Satisfactory agreement was obtained. Clearly, the commercial mercury instruments then again need to be calibrated against well-known mercury concentrations, a process which is equivalent to determining the value of the absorption cross-section.

DIAL systems for atomic mercury monitoring can be arranged on mobile platforms. The Swedish group constructed different generations of mobile lidar systems based on tunable lasers. These were organized with Nd-YAG-pumped dye lasers [63,67] or OPO systems [68]. Pulse repetition rates were typically 10 Hz, and practical measurement ranges for mercury were of the order of 1 km. Based on the Swedish work, a mobile DIAL system was constructed in China [69], where again a dye laser was employed because of its sturdiness compared to an OPO system. The construction of the Chinese system is shown in Figure 6 [69]. The system was arranged to be conveniently docked to the fixed laboratory when not deployed for field work, as illustrated in the Figure 6.

3. Atomic Mercury Monitoring

Since commercial mercury monitoring instruments based on atomic absorption or fluorescence are routinely being utilized in wide-spread point monitoring, we will in this section of examples of monitoring campaigns concentrate rather on the less commonly used techniques, which are largely of the remote-sensing type. Lidar monitoring always requires some eye-safety considerations, but since the relevant wavelength, 254 nm, is in the deep UV region where the eye structures are opaque, such considerations are strongly relaxed for the case of mercury monitoring. Further, since the ozone layer cuts off solar radiation below 300 nm, monitoring can be performed in low-background conditions, even at daytime.

3.1. Clean Air Monitoring

A very high sensitivity for low mercury levels can be obtained with laser cavity ring-down spectroscopy [70]. This is a type of intra-cavity measurement, where the effective path length achieved is long, and the regenerative nature of stimulated amplification is fully utilized. DIAL measurements resulting in very low concentration values were performed in the geothermal fields of Iceland in anticipation of detecting elevated mercury levels. However, due to geochemical conditions specific for Iceland, no indication of excess mercury was found. Rather, very low values were measured 2 ng/m³—typical Atlantic background levels [71].

3.2. City Monitoring

Due to various human activities, the atmospheric mercury concentrations in cities are elevated, but normally not excessively. Thus, DIAL monitoring requires integration over distances of typically 100 m. A small-size European city, such as Lund, Sweden, has quite low concentrations [61], while major cities clearly reach considerably higher levels. Since monitoring campaigns using DIAL are extended over quite a limited time period, the values obtained are by no means representative. Values reaching up to 10 ng/m³ were observed in the Chinese cities of Hangzhou [37], Guangzhou [38], Zhengzhou [72] and Xi’an [73].

A detailed study of mercury contents in the air in Guangzhou was performed with a commercial LUMEX Zeeman modulation correlation instrument [74]. The construction of such an instrument was discussed in Section 2.3.3. Concentrations along highways crossing the city were measured in different traffic situations from a personal car, as illustrated in Figure 7. Mercury levels are known to depend on traffic intensities [75,76]. Indoor concentrations, in particular in different departments of a major hospital, were measured with the hand-carried instrument. Results from the measurements are illustrated in Figure 8. Mercury pollution related to health-care procedures have been studied quite intensely; see [77,78,79].

3.3. Industrial Monitoring

Chlor-alkali plants produce sodium hydroxide (NaOH), which is an extremely important base chemical in the chemical industry. Traditionally, a brine solution is subject to electrolysis with liquid mercury as the (negative) cathode, where the Na⁺ ions are amalgamated into the mercury, which can be pumped to a different section, where clean water is added to wash out the sodium. Since the electrolytic cells become hot and there are considerable spills of the metal in maintenance, the production is associated with considerable release of the pollutant. DIAL is a particularly suitable method to assess the total flux of the pollutant out of an area. The laser beam is scanned in a cross-section of the gas plume. The result of such a scan recorded at Rosignano Solvay is shown in Figure 9 [80].

Figure 7. Atomic mercury concentrations along a 36 km long highway traverse across the city of Guangzhou, China. Measurements were performed with a Zeeman correlation monitor (Section 2.3.3) operated from a taxi with the gas-extraction hose protruding from a window. During rush-hour traffic, the mercury levels are significantly increased and reflect local traffic flow situation. ([74], reproduced under the CC BY License, MPDI).

Figure 7. Atomic mercury concentrations along a 36 km long highway traverse across the city of Guangzhou, China. Measurements were performed with a Zeeman correlation monitor (Section 2.3.3) operated from a taxi with the gas-extraction hose protruding from a window. During rush-hour traffic, the mercury levels are significantly increased and reflect local traffic flow situation. ([74], reproduced under the CC BY License, MPDI).

Figure 8. Atomic mercury concentrations in different departments of a major Chinese hospital (a,b). A Zeeman correlation instrument was hand-carried through departments during a 95-minute walk. Note the scale multiplication factor in (a), where maximum values close to 1400 ng/m³ were recorded ([74], reproduced under the CC BY license, MPDI).

Figure 8. Atomic mercury concentrations in different departments of a major Chinese hospital (a,b). A Zeeman correlation instrument was hand-carried through departments during a 95-minute walk. Note the scale multiplication factor in (a), where maximum values close to 1400 ng/m³ were recorded ([74], reproduced under the CC BY license, MPDI).

Figure 9. Cross-section through the atomic mercury plume downwind from the chlor-alkali plant at Rosignano Solvay, Italy. Local concentrations values are shown color coded, and the horizontally and vertically integrated loads are also indicated. Integration in both directions and multiplication with the perpendicular wind-speed component yields a total flux of 74 g/h ([80], reproduced with permission of Elsevier).

Figure 9. Cross-section through the atomic mercury plume downwind from the chlor-alkali plant at Rosignano Solvay, Italy. Local concentrations values are shown color coded, and the horizontally and vertically integrated loads are also indicated. Integration in both directions and multiplication with the perpendicular wind-speed component yields a total flux of 74 g/h ([80], reproduced with permission of Elsevier).

The local concentration values (kg/m³) are then integrated in the two spatial dimensions of the plume (m × m) to obtain a value of the mercury load over the cross-section (kg/m) as illustrated in Figure 9. Finally, we multiply with the vertical wind speed component (m/s) to obtain the total flux, with unit kg/s. Lidar measurement campaigns have been performed at electrolytic plants in Bohus, Sweden [81], Rosignano Solvay, Italy [82], and Tarnow, Poland [80]. The total flux of mercury out of the three plants is compared in Figure 10 for summer (with stronger outgassing from spills) and winter operation.

Mercury pollution of the atmosphere also results from the metal processing plants [83].

3.4. Geophysical Monitoring

There are several sources of geophysical gases [84]. Figure 11 gives an overview of gas emissions of geophysical origin and indicates approaches for their detection [85]. Clearly, the mining and processing of mercury produce pollution. The extraction of geothermal energy can be associated with large releases of mercury, especially in areas where mercury minerals are deposited. Finally, volcanic emissions can contain mercury.

3.4.1. Geothermal Fields

DIAL mapping of the air surrounding geothermal plants has been performed at Castelnuovo di Val di Cecina, Pian Castagnaio and Larderello and in the Laguni Rossi geothermal field, all in Tuscany, Italy [62]. Concentrations around one of the cooling towers at Larderello (Europe’s largest geothermal plant) are shown in Figure 12. Again, total flux can be assessed by combining lidar data with wind-field data.

Iceland is well known for extensive geothermal resources. However, as mentioned, the search in the extracted hot gases did not reveal any atomic mercury at the Svartsengi, Krysuvik or Nesjavellir fields [71]. This illustrates that the geochemical conditions can be very different at different geographical locations.

3.4.2. Volcanoes

Volcanic gases contain, besides H₂O, mostly CO₂, SO₂ and H₂S but also mercury, as investigated with point monitoring, e.g., at Mt. Etna [86] and in sub-surface fumaroles off the Eolean island of Panarea [87]. DIAL monitoring at Italian volcanoes has been attempted in conjunction with extensive shipborne measurements of SO₂ emissions on the Etna, Stromboli and Vulcano volcanoes [88,89]. As mentioned earlier, the lidar range for mercury measurements is of the order of 1 km, which limited the possibilities at Etna (3360 m). However, at Stromboli [930 m] and Vulcano (500 m) serious attempts for mercury monitoring were made. Concentrations were below or at detection level, and it was concluded that upper limits of mercury flux from the two volcanoes were 24 and 2.5 kg/day, respectively.

Figure 13 shows a photo of the Islands of Panarea and Stromboli. An echo-sounder recording of the rising gas plumes from the sea bottom (50 m down) off Panarea is included in the figure, as recorded from the bridge of the Italian research vessel N/O Urania in conjunction with the campaigns described in [88,89]. Multiple gas plumes as discussed in [87] are observed. The operating principle of the echo-sounder is similar to that of lidar, but now instead using (ultra)sound waves, propagating at about 1500 m/s.

DIAL recordings at the crater of Solfatara (Phlegraean fields) north of Naples revealed very low total fluxes of mercury as well as SO₂ [90]. It has been speculated whether mass extinction of triassic fauna is due to volcanic emissions of mercury [91].

3.4.3. Mercury Mining Areas

We have performed mapping of mercury emissions from several of the most prominent mercury mines in the world: Abbadia S. Salvatore, Italy [92], Almaden, Spain [93], Idrija, Slovenia [94] and Wanshan, China [66]. By combining lidar and wind data, unique estimates of the total flux out from the areas can be achieved. In spite of the cessation of active mining of the cinnabar (HgS) ore due to the implementation of the Minamata convention, the mining sites (where now clean-up efforts are being pursued) release substantial amounts of mercury to the atmosphere. Examples of experimental lidar recordings and the resulting DIAL (ratio) curve as measured at the Abbadia S. Salvatore site are given in Figure 14. The reason why the expected 1/R² intensity fall-off (see Equation (3)) of the recordings is not observed is that the amplification at the photomultiplier detector is ramped up for larger time delays to handle the associated large dynamic range in these types or recordings. We note that such a ramping is normalized away in the DIAL ratio curve. Mean concentration values over the path lengths indicated are shown. Very high values are observed in the vicinity of the cinnabar roasting plant, where the metal is released from the ore (HgS + O₂ --> Hg + SO₂).

A monitoring campaign was recently performed at the Wanshan mining site in the Guizhou province of China. A cross-section shows concentrations up to 40 ng/m³ (Figure 15) at the measurement site, which was about 2 km away from the main mining activities [66].

Since mercury is frequently found in conjunction with precious metals such as gold, silver and copper, efforts to investigate the use of the very volatile Hg gas as a geo-marker for exploration have been pursued [95].

3.5. Archeological Monitoring

Historical records indicate that liquid mercury was sometimes introduced in ancient tombs as an honorary gift from the ruler to prominent men. Mercury monitoring was included in the study of the Spring/Autumn tomb in Hansheng, China [96]. The historian Sima, covering the life of Qin Shi Huang, who unified China around 200 BC, describes [97] that large amounts of mercury were deposited in the emperor’s tomb, which is “protected” by the famous 8000-man terracotta army, next to the so far non-excavated mound. A recent DIAL study revealed that, indeed, a halo of mercury gas can be observed over the tomb, presumably due to evaporation from the underground tomb chamber through cracks developed over time [98]. The measurement scene and results are shown in Figure 16. The project attracted considerable interest, e.g., from popular science publications; see, e.g., [99,100].

4. Conclusions

Mercury is the only pollutant gas which is dominantly present as free atoms in the atmosphere. The monitoring of neurotoxic mercury can conveniently be performed using optical spectroscopy with high selectivity, sensitivity and real-time data delivery. Monitoring is clearly the first step in the quest to reduce/eliminate mercury from the environment. The element mercury has an unusually high vapor pressure and can serve as an interesting geophysical tracer gas. A sensitivity down to the rural background level can be achieved with optical techniques. A particularly high sensitivity can be achieved with pre-concentration procedures in conjunction with cold vapor atomic spectrometry and by Zeeman modulation correlation techniques. We have reviewed point-monitoring and remote-sensing analysis techniques and paid particular attention to molecular oxygen interference. The unique advantages of differential absorption lidar techniques in providing range-resolved mercury mapping, which allows total flux determinations and vertical concentration profile assessment, have been stressed. Mercury monitoring related to industrial activities, mining and geothermal energy extraction are illustrated. An example from archeology further highlighted the many aspects of mercury monitoring.