Implementation of spectroscopy as a rapid measurement tool (RMT) to inform risk assessment at petroleum contaminated sites in the Niger Delta, Nigeria.

Abstract

The recent developments and applications of rapid measurement tools (RMT) such as visible near-infrared (vis-NR) spectroscopy can provide ‘fit for purpose’ and cost effective data for informing risk assessment and managing oil-contaminated sites. Infrared spectroscopy discriminates between chemical compounds by detecting the specific vibrational frequencies of molecular bonds, producing a unique infrared ‘spectral signal’ thereby enhancing its identification and quantification applying chemometrics. The aim of the research was therefore to assess the potential of vis-NIR and mid-infrared (MIR) diffuse reflectance spectroscopy (DRS) techniques as RMT to inform risk decision support for remediation of petroleum contaminated sites. The objectives of the study were to: critically review chromatographic and spectroscopic methods for petroleum hydrocarbon analysis in soils; evaluate vis-NIR sensitivity to detect hydrocarbon concentration differences throughout weathering; predict TPH, PAH and alkanes concentrations in contaminated soils using vis-NIR and MIR DRS coupled with regression techniques. The study further evaluated which spectroscopy technique (vis-NIR or MIR); and which modelling method (RF or PLSR) performs best. In this study, a series of 13 soil mesocosms was set up where each soil sample collected was spiked with 10 ml of Alaskan crude oil and allowed to equilibrate at room temperature for 48 h before analysis. The mesocosms were incubated for two years at roomntemperature in the dark. Soils scanning and gas chromatography coupled to mass spectrometry (GC-MS) analysis were carried out at T0, 4, 12, 16, 20 and 24 months. Prior to scanning, soil samples were air-dried at room temperature (21oC) to reduce the effect of moisture. The soil scanning was done simultaneously using an AgroSpec spectrometer with a spectral range of 305 to 2200 nm (tec5 Technology for Spectroscopy, Germany) and Analytical Spectral Devices LabSpec2500 spectrometer (ASD Inc, USA) with a spectral range of 305 to 2500 nm to assess and compare the sensitivity and response of the two instruments to weathering and hydrocarbon composition change overtime against GC-MS data. Partial least squares (PLS) and random forest (RF) regression models showed that ASD LabSpec2500 performed better than tec5 which may be attributed to the shorter wavelength spectra range of the tec5 spectrometer and therefore not detecting all significant hydrocarbon signals (e.g., 2207, 2220, 2240 and 2460 nm). The sensitivity of the two spectral devices to weathering and REWARD K. DOUGLAS Cranfield University PhD Thesis, 2018 hydrocarbon composition change was, however, comparable; and the predicted concentrations were also comparable to the hydrocarbons concentrations determined by GC-MS. The results (coefficient of determination, R²=0.9; ratio of prediction deviation, RPD=3.79 and root mean square error of prediction, RMSEP=108.56 mg/kg) demonstrate that visible-near infrared diffuse reflectance spectroscopy (vis-NIR DRS) is a proven tool for rapid site investigation and monitoring without the need of collecting soil samples and lengthy hydrocarbon extraction for further analysis..To this end, 85 soil samples collected from three crude oil spill sites in the Niger Delta, Nigeria. Prior to spectral measurement, soil physiochemical properties such as pH, total organic carbon and textural analysis were carried out. Soil samples were scanned (field-moist) and assessed using ASD LabSpec2500 (wavelength 350-2500 nm) and MIR data was acquired with Agilent 4300 handheld Fourier transform infrared (FTIR) spectrometer (Agilent Technologies, Santa Clara, CA, United States) with a spectral range of 4000- 650 cmˉ¹. Specifically, detailed analysis of the hydrocarbon content including total petroleum hydrocarbons (TPH), aliphatic and aromatic hydrocarbon fractions were determined and quantified by GC-MS, vis-NIR and MIR DRS. MIR over-performed vis-NIR with RF modelling method performing better than PLSR in predicting TPH, PAH and alkanes. However, PLSR-vis-NIR produced slightly better results than PLSR- MIR in predicting TPH and alkanes. Overall, vis-NIR (wavelength 350-2500 nm) laboratory-scale study yields better TPH prediction than the field-scale study. The minimised moisture content may have improved the results, as laboratory-scale samples were air-dried. Based on the results, MIR spectroscopy coupled with RF is recommended for the analysis of hydrocarbon contaminated soil. Finally, spectroscopy approach was proposed as RMT for contaminated soil investigation and risk prioritisation.