THE MULTIWAVE IR-SPECTROMETRY AS THE TOOL FOR TOTAL HYDROCARBON ESTIMATIONS

Аннотация

 Problem. To evaluate the total concentration (с_Σ) of hydrocarbons (HСs) in natural or waste waters, it is possible to extract all HCs and to measure the generalized analytical signal in the certain IR spectral region. Then the total TPH index (с*) can be determined in terms of the certain standard substance X_st. The error of this estimation δс = (с* - с_Σ) / с_Σoften exceeds 50 % rel. Specialists believe that the transition to multiwavelength signal measurement using n analytical wavelength values (n > 1) improves the accuracy of с_Σ estimations. No systematic investigations of this problem were conducted earlier.

          Experiment. The model mixtures of alkanes, cycloalkanes and arenes in CCl₄ were prepared to simulate the composition of refined extracts. IR-spectra were registered with FT-801 Fourier-spectrometer. The absorbance of each mixture was measured  by three methods: (1) – only with 2930 cm^-1; (2) – with 2930, 2960 and 3030 cm^-1; (3) – with 2855, 2930, 2960 and 3030 cm^-1. The generalized signal was calculated for each mixture by summation of the measurands for different analytical wavelength values (AWV). Corresponding formulae with weighting factors were derived beforehand by the technique of least squares. The total index value (c*) was determined by the calibration curve А_Σ= f(с_x), and  Simard mixture was used as X_st  in all cases. Calculated δс values for standard methods (1) and (2) were compared with the data obtained using the new method (3). To interpret the results, the absorbance factors were compared for certain individual HCs.  

         Results. The precision of generalized signals slightly decreases when the number of AWV grows. Simultaneous LODs are raised and the absorbance factors for individual HCs are leveled. The errors of с_Σ evaluation via c* for all methods are statistically significant but they diminish when n grows. RMSEP parameter decreases with the methods line (1) > (2) > (3). Standard methods (1) and (2) do not provide the level of the result accuracy which is necessary according to Russian standards. The new multiwavelength method (3) provides the estimation accuracy which is required. For this method δс values do not exceed 15 % in modulo, even for mixtures with the high content of arenes.

Discussion. Our experiment showed that the simultaneous use of some AWVs really provides more reliable с_Σ estimations. The main reason for this effect is the leveling of the absorption coefficients for various HCs, particularly for alkanes and arenes. Subsequent reduction of systematic errors may be secured if we will not use the univariate calibration curves.  The multivariate calibration, obtained from the training set of the spectral data (method 3a), leads to accurate с_Σ  estimations. In this case δс values do not exceed 8 %, and for the most part of model mixtures systematic errors are not statistically significant. In “Conclusion” the fundamental significances of obtained results and perspectives of their application in aqueous analysis are discussed.   

         Key words: aqueous analysis, total indices, total content of hydrocarbons, multiwave IR spectrometry, systematic errors

(Russian)

DOI:http://dx.doi.org/10.15826/analitika.2015.19.1.011

S.V. Usova, М.А. Fedorova, S.V. Petrov, V.I. Vershinin

Dostoevsky Omsk State University (OmGU), Omsk, Russian Federation