Quantitative analysis of the Uncompahgre National Forest soil and its extracts by ¹³C DP- and CP-MAS solid state NMR methods

Abstract

We have carried out the first absolutely quantitative solid-state 13C NMR study on a soil and its separated components. The Uncompahgre National Forest soil was collected and separated into a humin fraction, humic acid, and fulvic acid using a classic soil separation method. The soil separation was performed in a quantitative manner such that all the organic carbon in the soil and its separated fractions was accounted for by elemental analysis. Compared to the carbon detected by elemental analysis, 1H-to-13C CP-MAS and 13C DP-MAS spin-counting accounted for substantially less than 100 percent of the carbon, from a low of 48 percent of the carbon, to a high of about 86 percent of the carbon, depending on the method and particular soil fraction. The peak areas for spin counting spectra generated by DP- and CP-MAS were corrected to account for relaxation and cross-polarization processes to yield NMR peak integrals that are directly proportional to the amount of carbon in a particular functional group that the peak represents. The DP-MAS generated spectra, following the aforementioned peak area corrections, were found to have greater relative intensity in the aromatic region compared to the corresponding CP-MAS spectra for the same soil components. The quantitative 13C MAS NMR analysis was accompanied by an extensive error analysis of the regressions, calibrations, and deconvolution procedures used in this thesis. The percent of the carbon detected by NMR compared to that detected by elemental analysis for the whole soil increased dramatically following a de-ashing treatment that employed the use of hydrofluoric acid. This treatment was found to greatly reduce the iron content, as detected by elemental analysis, while preserving the integrity of the organic content of the soil. Vibrating sample magnetometer (VSM) measurements were taken on all the soil and soil extracts, along with the HF-treated soil, to better understand the nature of the iron in these samples. These VSM measurements qualitatively confirmed the iron elemental analysis results for these samples and indicated that ferrimagnetic particles were retained in the whole soil, HF-treated whole soil, and humin fraction. Calculations using the Curie law of paramagnetism confirmed the susceptibility measurements on these samples and showed that paramagnetic iron was the dominate form of iron in the soil and soil fractions. A second method for estimating the carbon content by NMR, the empirical relationship method, utilizes peak areas and their organic carbon functionality assignments. This method more closely predicted some of the carbon contents, but is based on many subjective functionality assignments and the more error prone individual peak areas. For this reason this second method, while gaining popularity in some soil research areas, was evaluated but not a major focus in this thesis. Large systematic errors associated with the external reference calibration for DP- and CP-MAS spin counting spurred investigation into BI RF-field homogeneity and its relationship to quantitative spin counting in NMR. The BI field for the spinner and coil configuration was measured and found to be in rough agreement with BI fields simulated by using the law of Biot-Savart. A CP- and DP-MAS peak area study for a mixture of three different 13C labeled L-alanine compounds, as a function of sample position, showed greater deviation in signal intensities for CP-MAS generated peak areas. The methodology used to create coil shapes and predict their BI field profile was employed in creating two types of coil designs, of which the simulations predict that the variable pitch and radius coil produced a nearly homogeneous BI field throughout the sample volume.