MS and NMR Analysis of Isotopically Labeled Chloramination Disinfection Byproducts: Hyperlinks and Chemical Reactions

Abstract

FT-ICR MS and NMR analysis of an isotopically labeled complex mixture of water disinfection byproducts formed by chloramine disinfection of model phenolic acids is described. A new molecular formula assignment procedure using the CoreMS Python library able to assign isotopically enriched formulas is proposed. Statistical analysis of the assigned formulas showed that the number of compounds, the diversity of the mixture, and the chlorine count increase during the chloramination reaction. The complex reaction mixture was investigated as a network of reactions using PageRank and Reverse PageRank algorithms. Independent of the MS signal intensities, the PageRank algorithm calculates the formulas with the highest probability at convergence of the reaction; these were chlorinated and nitrated derivatives of the starting materials. The Reverse PageRank revealed that the most probable chemical transformations in the complex mixture were chlorination and decarboxylation. These agree with the data obtained from INADEQUATE NMR spectra and literature data, indicating that this approach could be applied to gain insight into reactions pathways taking place in complex mixtures without any prior knowledge.

Scheme 1. Starting Materials Used to Investigate Chloramine Disinfection Byproducts

Figure 1

Negative mode ESI mass spectrum of 20 ¹⁵N-labeled amino acids. The amino acid peaks are highlighted in blue and labeled.

Figure 2

Number of peaks detected and formulas assigned for technical triplicates of three samples for compounds 1 and 2. Contaminants refers to formulas present in control samples. The % relative cumulative intensity of the assigned peaks is shown by the scatter plot on the secondary axis.

Figure 3

Reconstructed mass spectra of chloramine DBPs of (a) 1 and (b) 2 after 1, 3, and 6 days of reaction time. Only formulas present in all 3 technical replicates are plotted, and their intensities are averaged (mean). Spectra are normalized to the highest intensity peak in each spectrum corresponding to the shown structural formulas.

Figure 4

UpSet plots of chloramination DBPs for days 1, 3, and 6 samples of (a) 1, colored by N and/or Cl content and (b) 2, colored by number of F atoms.

Figure 5

Correlation for chloramination of 2 of (a) average number of formulas, (b) Shannon diversity index; (c) mean m/z, (d) double bond equivalent, (e) hydrogen-to-carbon ratio, (f) oxygen-to-carbon ratio, and (g–i) count of Cl, N, and F against time. Correlations were determined using Spearman’s rho (ρ), and statistically significant correlations are highlighted in red. Each point represents a technical replicate.

Figure 6

(a) Reaction network of chloramination DBPs of 1 after 6 days of reaction time. The nodes are scaled to their PageRank score and colored according to their MS intensity (log scale). The nodes are linked by edges, which represent different reaction types. The starting material (1) and formulas with highest PageRank score are labeled. (b) The Reverse PageRank reaction probabilities are shown for each reaction type. S_NAr–nucleophilic aromatic substitution.

Figure 7

(a) Structural information provided by INADEQUATE and ADEQUATE NMR experiments. (b) The structures of the major products of chloramination of 3 as identified by NMR spectroscopy.