UV spectroscopic properties of principal inorganic ionic species in natural waters

UV spectroscopy is a well-established measurement method in instrument-based analytics. One of its main applications is quantitative analysis of natural and waste waters, in which it is often used to determine inorganic substances and monitor water quality. Guenther et al. (2001) investigated the potential of UV spectroscopy for determining total sulphide and iodide in natural waters. Johnson & Coletti (2002) designed an in-situ ultraviolet spectrophotometer to measure the concentration of nitrate, bisulphide and bromide at depth in the ocean. Lourenço et al. (2006) used UV spectroscopy for monitoring water quality from a wastewater plant. In the department of thermal process engineering at the University of Duisburg-Essen, Germany, UV spectroscopy was deployed to detect sulphur (IV) components in aqueous electrolyte solutions with high ionic strength (Cox et al. 2015, 2016).

A recurring problem in the investigation of individual species is interference due to absorption by other components in the systems investigated, and the resulting increased background absorption and possible exceeding of measurement limits. Typical natural waters include a complex matrix of constituents, of which inorganic components, particularly, absorb in the far ultraviolet wavelength range from 195 to 280 nm (Collos et al. 1999; Johnson & Coletti 2002). This results in a multicomponent UV spectrum, governed by the superposition of absorptions by the individual species. Analysis of these spectra is carried out using different methods. Collos et al. (1999) studied the individual contributions of absorbing species in seawater to identify a wavelength interval where nitrate could be determined without interference. Many approaches are based on complex multivariate data analysis (Lourenço et al. 2006) or multiple linear regression models (Guenther et al. 2001; Johnson & Coletti 2002). In the latter, the concentrations of the existing species are determined by fitting an equation for the approximation of the background absorption to the observed absorbance spectrum. This method requires knowledge of all molar absorptivities of the components present.

In this study, the UV spectroscopic properties of all common inorganic water species in the far UV region (195 to 280 nm) were investigated at 25 °C. Molar absorptivities were determined in calibration experiments over a wide range of concentrations. The publication of molar absorptivities in broad wavelength intervals allows readers to predict the UV spectra of complex ionic solutions, enabling the evaluation of UV spectroscopy as an analytical method for future applications based on a theoretical approach. The authors are not aware of any other publication carrying specific molar absorptivity values, especially at the absorption band edges. The approach was validated by comparing measured and calculated spectra for different natural waters, and dominant components were identified.

The experiments were carried out in a closed, double-shell, glass vessel at 25 °C. To guarantee isothermal conditions the vessel was coupled to a circulation thermostat (F-25ME, Julabo, Seelbach, Germany), enabling temperature settings with an accuracy of ±0.1 °C. A resistance temperature detector (PT100, Bola) monitored experimental temperature. To prevent the influence of atmospheric oxygen and carbon dioxide, nitrogen was passed into the vessel via a gas frit and discharged through a cooler to avoid evaporation losses from the experimental solution. A quartz immersion probe (Excalibur, Hellma, Müllheim, Germany) with an optical path length (d) of 20 mm was used to detect the UV spectra. This was connected to a UV-Vis spectrophotometer (Cary 60 UV-Vis, Agilent, Santa Clara, CA, USA) by fibre-optic cables. The spectrophotometer can analyse the wavelength range from 190 to 1,100 nm and uses a xenon flash light (80 Hz) to produce electromagnetic radiation, which is resolved to separate wavelengths by a Czerny-Turner monochromator. The spectrophotometer can compensate intensity fluctuations while operating measurements by using two silicon photodiodes.

The pH value was measured continuously with a pH electrode (InLab Pure Pro ISM, Mettler Toledo, Columbus, OH, USA) in addition to the spectrometric recordings.

A closable sample inlet was used to add analyte and the solution was homogenised with a magnetic stirrer during the study. All measuring instruments were locked with glass fittings and made of inert materials, glass or Teflon.

At the start of the experiment ultrapure water (Millipore, 18.2 MΩcm) was treated with nitrogen for at least four hours to remove dissolved oxygen and carbon dioxide. Subsequently, a gravimetrically determined mass of ultrapure water was transferred to the experimental vessel. After reaching thermal equilibrium (T = const.) the base spectrum was measured, comprising the extinctions E of the solvent water and optical system. Next, a defined mass of the electrolytic analyte was added to the solution and the analyte spectrum measured. Extinction values exceeding 3 are excluded due to the low intensity of passing electromagnetic radiation, which leads to the signal being disturbed by stray and false light (Figure 1). Finally, a difference spectrum was generated by subtracting the base spectrum from that of the analyte, to show the isolated extinction of the analyte.

The spectra were detected at a resolution of 1 nm. Using a flashlight frequency of 80 Hz and an averaging time of one second, each spectrum data point is the average of 80 measurements. Five spectra were detected and averaged for each measuring point.

In order to determine the molar absorptivities in the calibration experiments, a series of measurements with increasing analyte concentration c was performed. In addition to the upper measuring limit, a lower detection limit was introduced to exclude data points with high noise levels.

In accordance with DIN 32645 (DIN Deutsches Institut für Normung e.V. 2008), the determination limit describes the minimum signal level that can be used for quantitative analysis. In this work the determination limit is defined as 10 times the standard deviation. The latter is calculated in an absorption-free wavelength interval of 20 nm from the fluctuation of the signal around the nominal value of zero. For the calibrations below only those data points above the lower limit were used (Figure 2).

A set of difference spectra was generated using the procedure above for all analyte spectra. The concentration dependency of extinction can be characterised with a linear regression for each wavelength. According to Lambert Beer′s law (Thomas & Burgess 2007), the regression slope corresponds to the product of molar absorptivity, ε, and optical path length and thus enables the determination of ε. The spectral range differs for each measuring point, because the extinction and, therefore, the measuring limit position change with concentration. The numbers of measuring points available for the linear regression differ particularly near the detection limits. To ensure a reliable database for molar absorptivity determination, only regressions with at least six data points were used. As a consequence, there is a specific wavelength range for each species, depending on the maximum concentration, in which molar absorptivities can be determined. In previous studies the molar absorptivity error was determined by reproduction measurements as 5%.

The investigation of electrolytes to determine the absorption properties of single ionic species requires isolation of the extinctions of anions and cations. For this reason the extinctions of the monovalent chlorides NaCl, KCl and HCl were measured in the concentration range c = 0 to 350 mmol/l. There is strong similarity between solutions containing the cations Na⁺, K⁺, and H⁺ (Figure 3). Because of this it is assumed that the cations do not absorb in the spectral range investigated and do not influence it. The independence of the UV spectra from the monovalent cations is confirmed by several publications (Smith & Symons 1958; Blandamer & Fox 1970; Zuman & Szafranski 1976). In consequence, the anions were all studied on the basis of sodium compounds.

This systematic study included the principal ionic components of natural waters (Sigg & Stumm 2011; Millero 2013). The absorption of the trace components ammonium (NH₄⁺), iodide (I⁻) and nitrate (NO₃⁻) was also analysed. Table 1 lists the components detectable in the wavelength range investigated together with the maximum experimental concentration, c_max, and the wavelength interval, Δλ_ε, where molar absorptivities could be determined. The maximum concentrations were set to exceed typical seawater concentrations (Millero 2013), so interpolations are admissible.

In Figure 4 (upper part) the molar absorptivities of the halides bromide, iodide and chloride are plotted versus wavelength (Table 2). Iodide absorbs at the highest wavelengths. Starting from 252 nm the absorptivities increase to a maximum of 13,186 l/mol/cm at 226 nm. After that, the value decreases to a minimum at 209 nm and increases again towards 195 nm. This agrees with observations by Fox & Hayon (1977a), which identify a first peak at a wavelength of 226.4 nm and a second peak at 195.7 nm. The second peak is only indicated in this work but cannot be observed completely due to the proximity to the limit of the investigated spectral range. The bromide and chloride extinctions show similar characteristics. The absorptivities increase from high to low wavelengths, starting at λ_bromide = 225 nm and λ_chloride = 215 nm, and ending at 195 nm. No absorption peaks can be seen in the spectral range investigated and only the low-energy part of the absorption band is detected. Fox & Hayon (1977b) report a bromide absorption peak at 197.1 nm. The molar absorptivities of chloride are more than an order of magnitude lower than those of iodide and bromide over the whole spectral range. Fox et al. (1978) identify an absorption maximum at 174.5 nm, which is beyond the spectral range of this work. Fox & Hayon (1977a, 1977b) and Fox et al. (1978) characterise the absorption of halides as typical of charge-transfer-to-solvent (c.t.t.s.) spectra, where an electron is transferred from the absorbing molecule to the solvent. Halides form two distinct absorption bands, which shift to shorter wavelengths from iodide to fluoride (Blandamer & Fox 1970). The shift effect is confirmed in this work for the molar absorptivities of iodide, bromide and chloride.

Figure 4 (middle) shows the molar absorptivities of nitrate and hydroxide (Table 3). The first significant absorption of nitrate is detected at 244 nm. With decreasing wavelengths, molar absorptivities increase to a maximum of ε(NO₃⁻) = 9,053 l/mol/cm at 203 nm. After that, they decrease towards 195 nm. In the literature absorption by nitrate at T = 20 °C is characterised by two absorption peaks. Thomas & Burgess (2007) publish molar absorptivities of ε(NO₃⁻) = 8,630 l/mol/cm at 205.6 nm and ε(NO₃⁻) = 8.0 at 301.6 nm. Meyerstein & Treinin (1961) locate the maxima at 201 nm and 302 nm with values of ε(NO₃⁻) = 9,900 and ε(NO₃⁻) = 7.2 l/mol/cm. The low-energy band is assigned to an internal n → π* transition, and the high-energy band to an internal π → π* transition. The low-energy band could not be observed in this study because of the low concentrations measured. Hydroxide absorption starts at 210 nm and rises continuously to 195 nm. In the literature the hydroxide absorption maximum is at λ = 186 nm, with a molar absorptivity of ε(OH⁻) = 3,980 l/mol/cm by Ley & Arends (1929), and at λ = 187 nm with a molar absorptivity of ε(OH⁻) = 3,860 l/mol/cm by Fox et al. (1977). Therefore, no maximum is expected in the far UV region. The electron transition is categorised ‘c.t.t.s.’ (Fox et al. 1977), as in the case of the halides.

The molar absorptivities of sulphate, bicarbonate and carbonate are presented in Table 4 and plotted in Figure 4 (lower). Molar absorptivities for sulphate were determined in the range 207 to 195 nm. Absorptivities increase with decreasing wavelength to ε(SO₄²⁻) = 25.34 l/mol/cm. Ley & Arends (1932b) report molar absorptivity of ε(SO₄²⁻) = 21.38 l/mol/cm at 195 nm. Shapiro (1965) assigns the absorption to a c.t.t.s transition.

Due to the dissociation equilibrium between bicarbonate and carbonate it was necessary to adjust the pH in the experiments – the pH for bicarbonate calibration was set at 8.7. Molar absorptivity increases, starting from 215 nm, to a maximum of ε(HCO₃⁻) = 95.57 l/mol/cm at λ = 195 nm. Ley & Arends (1932a) observe a molar absorptivity of ε(HCO₃⁻) = 89.4 l/mol/cm at 195 nm, which is in good agreement with the results of this study. The absorption is ascribed to a forbidden n → π* transition by Mookherji & Tadon (1966).

Carbonate calibration was performed at pH = 13. Because of the high OH⁻ concentration at this pH and the resulting high extinction of OH⁻ ions, the experiments were carried out with a much shorter optical path length – d = 2 mm – to suppress OH⁻ extinction. Nevertheless, the contribution of the hydroxide extinction causes the upper detection limit to be exceeded at wavelengths below 207 nm, so no analysis of this range was possible. In the remaining spectral range, carbonate molar absorptivities could be determined up to 225 nm. An increase in molar absorptivity to a maximum of ε(CO₃²⁻) = 126.35 l/mol/cm at 209 nm was observed. After that the value decreases towards 207 nm. In contrast, Ley and Arends (Ley & Arends 1932a) describe a steady increase in molar absorptivity, with a value of ε(CO₃²⁻) = 141–177 l/mol/cm at 209 nm without observing a subsequent decrease. They report high fluctuations in their carbonate calibration, so that the published values can only be interpreted as guides.

The absorption spectra of natural waters in the far UV region were determined using the molar absorptivities determined in the calibration experiments. The specific extinctions of all absorbing components were calculated on the basis of given concentrations, for the respective systems, and then summed to give the total extinction. It is assumed that the base spectrum (extinctions of water and the optical system) equals zero over the whole spectral range. The given concentrations are average values, so that differences may occur in relation to specific water bodies. Analysing the total extinction enabled the dominant species to be identified and evaluation of the suitability of UV spectroscopic measurements for the systems investigated. The optical path length for the calculation was set to d = 20 mm, corresponding to that in the apparatus used. Synthetic solutions with equal anionic concentrations based on sodium compounds were also prepared and measured spectroscopically. The calibrations were validated by comparing the calculated (cal.) and measured (meas.) total extinctions. The method of additive superposition of the specific extinctions for the prediction of multicomponent ionic systems was also evaluated. Error bars were omitted for clarity.

As an example representing waters of low ionic concentrations, the UV absorption of groundwater was investigated. The inorganic component concentrations are taken from work by Shvartsev (2008) and give a global average value of different groundwaters from the leaching zone. The composition of this system can therefore be regarded as representative of groundwater. Specific systems only differ slightly. Because the pH is 6.75 it can be assumed that the carbonate and hydroxide concentrations are negligible. Figure 5 shows the ionic concentrations and calculated extinctions. It is evident that the total extinction is dominated largely by nitrate. In addition, extinctions by bicarbonate and chloride occur below 215 nm, where they increase with decreasing wavelength. The extinction of bicarbonate increases more strongly than that of chloride, so it provides the main contribution to the total extinction at 195 nm. The total extinction increases with decreasing wavelength over the whole spectral range, reaching a maximum of E = 0.98 at 195 nm, enabling groundwater systems to be analysed in the full range. The calculated and measured total extinctions agree very well (Figure 5). The bromide and sulphate extinctions can be neglected because of the low concentrations of those species (Figure 6).

The next system studied was typical river water, which contains only chloride, sulphate, bicarbonate and nitrate in relevant concentrations with respect to absorbing species. The concentration specification (Figure 7) is drawn from Millero (2013). Due to the even lower concentration of sulphate compared to groundwater the extinction can be neglected, so there will be no presentation at this point. As in groundwater, the total extinction of river water is influenced strongly by nitrate (Figure 7). At wavelengths below 215 nm an overlap with the absorption of chloride and bicarbonate is found. The total extinction reaches a maximum of E = 0.53 at 195 nm, considerably below the groundwater extinction. A comparison of calculated and measured absorbance shows high agreement, so that the superposition principle can be considered valid for this system (Figure 7).

Seawater was studied, as a system of high ionic strength. While the total salinity of seawater varies considerably, the ratio of ionic concentrations is almost constant (Millero 2013). Following Millero, the concentrations were set at those of a reference seawater with a salinity of S = 35 g/l. The concentrations and extinctions of the anions, as well as the calculated and measured total extinctions are plotted against wavelength in Figure 8. For the given concentrations, nitrate, bromide, chloride, sulphate and bicarbonate exhibit high extinctions, while those of carbonate and hydroxide are low (and their influence on the total extinction is negligible (Figure 9)).

In this context, only nitrate absorbs at long wavelengths between 244 and 225 nm, so the nitrate and total extinctions are identical. Bromide also absorbs below 225 nm and the extinctions of nitrate and bromide overlap. With decreasing wavelengths the extinction of bromide increases more quickly than that of nitrate and below 219 nm bromide dominates the total extinction.

The absorption of chloride and sulphate starts at 215 nm. Arising particularly from the high absorption of bromide, the total extinction increases rapidly with decreasing wavelength and exceeds the upper detection limit at wavelengths below 213 nm, rendering analysis of the spectral range from 195 to 212 nm infeasible. In other words, the main spectral regions for chloride, sulphate, and bicarbonate cannot be analysed in seawater.

The total extinction measured in the range from 235 to 225 nm is in good agreement with the calculated total extinction, so a good representation of the dominant nitrate extinction can be assumed. For wavelengths from 225 to 215 nm, the measured extinction values are lower than those calculated. In this spectral range the contributions of nitrate and bromide overlap. The differences can be attributed to the bromide extinction, and a possible cause could be the high ionic strength of the aqueous solution and the resulting ionic interactions. Blandamer & Fox (1970) describe the influence of interactions to species with c.t.t.s. transitions as a shift of the absorption band to shorter wavelengths, as observed for bromide. However, the deviations are slight so the calculation by additive superposition of the spectral contributions still gives a fair approximation of total absorption.

As shown in Figure 8, total extinction can only be measured down to 215 nm because of the use of fibre-optic cables in the experimental setup, which leads to significant absorption in the base spectrum. The sum of the contributions of the base spectrum and the ions results in a high measured total extinction, so the upper detection limit is reached at higher wavelengths than the calculated total extinction. Regardless of the experimental setup, it is clear that UV spectroscopy is limited in seawater with respect to the usable spectral range.

As can be seen, UV spectroscopy in the far UV region is subject to limitations particularly in high ionic strength systems. These arise from high extinction values caused by the given ion matrix, leading to exceedance of the maximum measurable extinction of current spectrophotometers and thus a reduction in the usable spectral range. Components that absorb only in this interval cannot be detected. Particular attention should be paid to nitrate, which absorbs strongly in all of the systems studied. Apart from nitrate, bromide absorbs strongly in seawater systems.

The spectral location of the detection limits may differ for different optical path lengths. Shorter paths result in the measurement of lower total extinctions so the available spectral range is expanded to shorter wavelengths. At the same time, however, the minimum detectable signal strength, due to the lower detection limit, is also shifted to shorter wavelengths, constricting the available spectral range. Operating with a longer optical path length leads to the opposite effect. Hence, the optical path length should be tailored to the given sample.

Comparison of the calculated and measured extinction spectra proves that calculation of total extinction by superposition of the specific extinctions of all ions present is valid, especially for low ionic strength systems. Only systems of high ionic strength – e.g., seawater – reveal differences, which can be explained by the interactions between ions. The calibration measurements thus seem reliable and can be used to predict the UV spectra of multicomponent ionic systems without experimental work.

To study the absorption behaviour of natural waters in the UV region from 195 nm to 280 nm, the molar absorptivities of their principal ionic components were determined. It was shown that inorganic cations neither absorb in the spectral range investigated nor influence the absorption behaviour of anions, in agreement with the literature. It could be shown that the far UV region is suitable for investigating the relevant inorganic anions. The absorption bands of all components are in the transition region from far UV to vacuum UV radiation. Because of this, only parts of the bands can be observed in the spectroscopic range 195 to 280 nm. Absorption maxima were detected only for iodide, nitrate and carbonate. The highest molar absorptivities found were for iodide, bromide, nitrate and hydroxide, so these species influence UV spectroscopic measurements strongly. The chloride, sulphate, bicarbonate and carbonate absorptivities are smaller by up to two orders of magnitude. The UV spectra of natural seawater, groundwater and river water were calculated on the basis of average ion concentrations using molar absorptivities. The individual contributions enable identification of the dominant ions at different wavelength intervals. In the case of seawater, extinction is influenced mainly by bromide and nitrate. Because of the high bromide concentration, the maximum measurable extinction was exceeded and the usable spectrum was limited to the short wavelength region. In ground- and river-waters total extinction is lower over the whole wavelength range and the spectrum is dominated by nitrate.

Comparison of the calculated and measured total extinctions validates the calibration measurements, and illustrates that additive superposition of specific extinctions of all ions present is an acceptable method of estimating UV spectra based on a theoretical approach. Therefore the UV spectra anticipated for ionic systems can be predicted using published molar absorptivities. Thus it is possible to evaluate the suitability of UV spectroscopy as a methodology for given measurement tasks, and to define limits and potential applications. In addition to inorganic constituents, natural waters contain organic species, which can influence the UV spectrum in high concentrations.

The authors acknowledge gratefully the financial support by AiF in the research project 19074N.