The use of mid-infrared diffuse reflectance spectroscopy for acid sulfate soil analysis.

Good management of sulfide minerals and sulfuric acid in Acid Sulfate Soils (ASS) requires cost-effective rapid analytical data for their characterisation. However, the determination of properties in ASS samples using traditional laboratory techniques is expensive and time consuming. Excessive delays in analysis risks sample changes from oxidation. Mid-infrared (MIR) spectroscopy with multivariate regression offers a quicker and cheaper surrogate. This manuscript reports the prediction of some of the following key soil parameters in ASS characterisation using benchtop (Perkin Elmer) and handheld (ExoScan) diffuse reflectance MIR Fourier transform (DRIFT) spectrometers: Total Organic Carbon (TOC), Titratable Actual Acidity (TAA), Extractable Sulfate Sulfur (ESS), Reduced Inorganic Sulfur (RIS), Retained Acidity (RA), Acid Neutralising Capacity (ANC), and Lime Calculation (LC). Three sets of representative ASS soil profiles, comprising 132 samples from hyposulfidic, hypersulfidic and sulfuric materials, and covering a wide range of environments in South Australia were scanned under laboratory conditions. These were combined with reference laboratory data in partial least squares regression (PLSR) calibration models. The calibrations were validated by leave-one-out cross validation, with a further test set available for validation. Predictions with coefficient of determination (R2) > 0.75, were obtained for TOC (0.95), TAA (0.88), RIS (0.86), LC (0.76) and ANC (0.76), but models for ESS (0.66) and RA (0.41) were less satisfactory. The handheld spectrometer performed similarly to the benchtop spectrometer in terms of PLSR prediction accuracies with the potential for in-field sampling. Results thus confirmed the possibility of using MIR spectroscopy for the rapid and cost-effective characterisation of ASS.

Rapid prediction of total petroleum hydrocarbons in soil using a hand-held mid-infrared field instrument.

This manuscript reports on the performance of a hand-held diffuse reflectance (mid)-infrared Fourier transform (DRIFT) spectrometer for the prediction of total petroleum hydrocarbons (TPH) in three different diesel-contaminated soils. These soils include: a carbonate dominated clay, a kaolinite dominated clay and a loam from Padova Italy, north Western Australia and southern Nigeria, respectively. Soils were analysed for TPH concentration using a standard laboratory methods and scanned in DRIFT mode with the hand-held spectrometer to determine TPH calibration models. Successful partial least square regression (PLSR) predictions, with coefficient of determination (R(2)) ~0.99 and root mean square error (RMSE) <200mg/kg, were obtained for the low range TPH concentrations of 0 to ~3,000mg/kg. These predictions were carried out using a set of independent samples for each soil type. Prediction models were also tested for the full concentration range (0-60,000mg/kg) for each soil type model with R(2) and RMSE values of ~0.99 and <1,255mg/kg, respectively. Furthermore, a number of intermediate concentration range models were also generated for each soil type with similar R(2) values of ~0.99 and RMSE values <800mg/kg. This study shows the capability of using a portable mid-infrared (MIR) DRIFT spectrometer for predicting TPH in a variety of soil types and the potential for being a rapid in-field screening method for TPH concentration levels at common regulatory thresholds. A novel hand-held mid-infrared instrument can accurately detect TPH across different soil types and concentrations, which paves the way for a variety of applications in the field.

GEMAS: prediction of solid-solution phase partitioning coefficients (Kd) for oxoanions and boric acid in soils using mid-infrared diffuse reflectance spectroscopy.

The authors' aim was to develop rapid and inexpensive regression models for the prediction of partitioning coefficients (Kd), defined as the ratio of the total or surface-bound metal/metalloid concentration of the solid phase to the total concentration in the solution phase. Values of Kd were measured for boric acid (B[OH]3(0)) and selected added soluble oxoanions: molybdate (MoO4(2-)), antimonate (Sb[OH](6-)), selenate (SeO4(2-)), tellurate (TeO4(2-)) and vanadate (VO4(3-)). Models were developed using approximately 500 spectrally representative soils of the Geochemical Mapping of Agricultural Soils of Europe (GEMAS) program. These calibration soils represented the major properties of the entire 4813 soils of the GEMAS project. Multiple linear regression (MLR) from soil properties, partial least-squares regression (PLSR) using mid-infrared diffuse reflectance Fourier-transformed (DRIFT) spectra, and models using DRIFT spectra plus analytical pH values (DRIFT + pH), were compared with predicted log K(d + 1) values. Apart from selenate (R(2) = 0.43), the DRIFT + pH calibrations resulted in marginally better models to predict log K(d + 1) values (R(2) = 0.62-0.79), compared with those from PSLR-DRIFT (R(2) = 0.61-0.72) and MLR (R(2) = 0.54-0.79). The DRIFT + pH calibrations were applied to the prediction of log K(d + 1) values in the remaining 4313 soils. An example map of predicted log K(d + 1) values for added soluble MoO4(2-) in soils across Europe is presented. The DRIFT + pH PLSR models provided a rapid and inexpensive tool to assess the risk of mobility and potential availability of boric acid and selected oxoanions in European soils. For these models to be used in the prediction of log K(d + 1) values in soils globally, additional research will be needed to determine if soil variability is accounted on the calibration.

GEMAS: prediction of solid-solution partitioning coefficients (Kd) for cationic metals in soils using mid-infrared diffuse reflectance spectroscopy.

Partial least squares regression (PLSR) models, using mid-infrared (MIR) diffuse reflectance Fourier-transformed (DRIFT) spectra, were used to predict distribution coefficient (Kd) values for selected added soluble metal cations (Ag(+), Co(2+), Cu(2+), Mn(2+), Ni(2+), Pb(2+), Sn(4+), and Zn(2+)) in 4813 soils of the Geochemical Mapping of Agricultural Soils (GEMAS) program. For the development of the PLSR models, approximately 500 representative soils were selected based on the spectra, and Kd values were determined using a single-point soluble metal or radioactive isotope spike. The optimum models, using a combination of MIR-DRIFT spectra and soil pH, resulted in good predictions for log Kd+1 for Co, Mn, Ni, Pb, and Zn (R(2) ≥ 0.83) but poor predictions for Ag, Cu, and Sn (R(2) < 0.50). These models were applied to the prediction of log Kd+1 values in the remaining 4313 unknown soils. The PLSR models provide a rapid and inexpensive tool to assess the mobility and potential availability of selected metallic cations in European soils. Further model development and validation will be needed to enable the prediction of log K(d+1) values in soils worldwide with different soil types and properties not covered in the existing model.

Influence of the nature of soil organic matter on the sorption behaviour of pentadecane as determined by PLS analysis of mid-infrared DRIFT and solid-state 13C NMR spectra.

The nature of soil organic matter (SOM) functional groups associated with sorption processes was determined by correlating partitioning coefficients with solid-state (13)C nuclear magnetic resonance (NMR) and diffuse reflectance mid-infrared (DRIFT) spectral features using partial least squares (PLS) regression analysis. Partitioning sorption coefficients for n-pentadecane (n-C(15)) were determined for three alternative models: the Langmuir model, the dual distributed reactive domain model (DRDM) and the Freundlich model, where the latter was found to be the most appropriate. NMR-derived constitutional descriptors did not correlate with Freundlich model parameters. By contrast, PLS analysis revealed the most likely nature of the functional groups in SOM associated with n-C(15) sorption coefficients (K(F)) to be aromatic, possibly porous soil char, rather than aliphatic organic components for the presently investigated soils. High PLS cross-validation correlation suggested that the model was robust for the purpose of characterising the functional group chemistry important for n-C(15) sorption.

Direct comparison between visible near- and mid-infrared spectroscopy for describing diuron sorption in soils.

Both visible near-infrared (VNIR) and mid-infrared (MIR) spectroscopy have been claimed to better predict pesticide sorption in soils than other methods. We compared the performances of VNIR and MIR spectroscopy for predicting both organic carbon content (foc) and the sorption affinity (Kd) of diuron in 112 surface soils from South Australia. Separate calibration models were developed between VNIR and MIR spectra, and foc and Kd using partial least-squares (PLS) regression. MIR clearly outperformed VNIR for predictions of both foc and Kd in soils. Correlation (R2) and accuracy (RPD) indices were 0.4 and 1.3 for the VNIR-PLS model versus 0.8 and 2.3 for the MIR-PLS model, respectively, for Kd prediction. PLS loadings for sorption prediction were compared in terms of the soil information they contained. While VNIR loading did not include any direct spectral information regarding soil minerals, MIR loading included peaks associated with sand, clays, and carbonates. Perhaps by better predicting foc and integrating the effects of OC as well as minerals, the MIR-PLS model provided a better prediction for diuron Kd values in our calibration set.

Midinfrared spectroscopy and chemometrics to predict diuron sorption coefficients in soils.

The potential of mid-infrared (MIR) spectroscopy in combination with partial least-squares (PLS) regression was investigated to predict the soil sorption (distribution) coefficient (K(d)) of a nonionic pesticide (diuron). A calibration set of 101 surface soils collected from South Australia was utilized for reference sorption data and MIR spectra. Principal component analysis (PCA) was performed on the spectra to detect spectral outliers. The MIR-PLS model was developed and validated by dividing the initial data set into four validation sets. The model resulted in a coefficient of determination (R2) of 0.69, a standard error (SE) of 5.57, and a residual predictive deviation (RPD) of 1.63. The normalized sorption coefficient for the organic compound (K(oc)) approach, on the other hand, resulted in R2, SE, and RPD values of 0.42, 7.26, and 1.25, respectively. However, the significant statistical difference between the two models was mainly due to two outliers detected via PCA. Apart from spectral outliers, the performance of the two models was essentially similar for the rest of the calibration set. Outlier detection by the MIR-PLS model may gainfully be employed as a tool for improving prediction of K(d). The MIR-based model can provide a direct estimation of K(d) values based on the integrated properties of organic and mineral matter reflected in the infrared spectra.

Prediction of atrazine sorption coefficients in soils using mid-infrared spectroscopy and partial least-squares analysis.

This study explored the potential of mid-infrared spectroscopy (MIR) with partial least-squares (PLS) analysis to predict sorption coefficients (Kd) of pesticides in soil. The MIR technique has the advantage of being sensitive to both the content and the chemistry of soil organic matter and mineralogy, the important factors in the sorption of nonionic pesticides. MIR spectra and batch Kd values of atrazine were determined on a set of 31 soil samples as reference data for PLS calibration. The samples, with high variability in soil organic carbon content (SOC), were chosen from 10 southern Australian soil profiles (A1, A2, B, and C in one case). PLS calibrations, developed for the prediction of Kd from the MIR spectra and reference Kd data, were compared with predictions from Koc-based indirect estimation using SOC content. The reference Kd data for the 31 samples ranged from 0.31 to 5.48 L/kg, whereas Koc ranged from 30 to 680 L/kg. Both coefficients generally increased with total SOC content but showed a relatively poor coefficient of determination (R2 = 0.53; P > 0.0001) and a high standard error of prediction (SEP =1.22) for the prediction of Kd from Koc. This poor prediction suggested that total SOC content alone could explain only half of the variation in Kd. In contrast, the regression plot of PLS predicted versus measured Kd resulted in an improved correlation, with R2 = 0.72 ( P > 0.0001) and standard error of cross-validation (SECV) = 0.63 for three PLS factors. With the advantages of MIR-PLS in mind, (i) more accurate prediction of Kd, (ii) an ability to reflect the nature and content of SOC as well as mineralogy, and (iii) high repeatability and throughput, it is proposed that MIR-PLS has the potential for an improved and rapid assessment of pesticide sorption in soils.