Discussion/Findings

Method Performance and Data Quality

Overall, the solution-mode ICP-MS method performed well for the majority of analytes, with high calibration linearity (R² > 0.9993 for all isotopes) and low method detection limits across most trace metals. Major cations such as Na, K, and Ca exhibited stable response curves spanning several orders of magnitude. Precision metrics were strong, with Quant RSD values typically <5% for major elements and <10% for trace metals, indicating highly repeatable signal behavior.

Phosphorus was the only analyte that required significant reprocessing during calibration. The initial multi-point calibration exhibited curvature at the low-concentration end (levels L2–L4), which is a well-documented limitation of 31P⁺ analysis in He mode. Phosphorus has comparatively low ionization efficiency in the plasma and is strongly affected by polyatomic background (e.g., NOH⁺, NO⁺, COH⁺), making low-level measurements highly unstable. He collision mode, while effective at suppressing these interferences, further reduces sensitivity and contributes to nonlinear behavior at the low end of the calibration curve. After excluding L2–L4, the remaining calibration levels produced an acceptable linear fit (R² ≈ 0.9998), though precision and sensitivity remained lower than for other analytes. These results reflect known analytical challenges associated with phosphorus in solution-mode ICP-MS rather than instrument instability, and they highlight the importance of optimized reaction-gas methods (e.g., O₂ mass-shift to 47PO⁺) for improved P performance.

A systematic offset in internal standard recoveries was observed between student-prepared solutions and instructor-prepared blanks/QC1. While the instrument and acquisition parameters remained stable, the internal standards responded differently depending on the sample matrix. This pattern suggests a matrix mismatch rather than instrument drift. Despite this offset, the overall analytical performance remained adequate for quantifying the major and trace metals relevant to the project.

Relation of Findings to the Research Question

The method successfully distinguished major compositional differences among samples. Na, K, and Ca varied by more than an order of magnitude across the 16 waters analyzed, reflecting contrasting mineral sources and water-treatment histories. Trace metals such as Mn, Fe, Sr, and Ba were consistently detected above their MDLs, enabling meaningful comparison of geochemical signatures among brands.

These findings show that the method was effective for characterizing bottled-water chemistry, especially for metals with strong ionization efficiency and well-behaved calibration curves. For analytes near their detection limits (As, Se, Pb), the method still provided quantifiable data in some samples, though uncertainty increased. The stability of major ions and reproducibility across replicates supports the method’s utility for environmental and drinking-water geochemistry applications.

In summary, the method worked well for the intended research question: identifying differences in major and trace metal concentrations between bottled-water and tap-water products. Even with the internal standard offset, the relative concentrations and geochemical trends remained robust enough to interpret water sources and treatment differences.

Method Improvements and Missing Elements

While the method was largely successful, several improvements could enhance accuracy and reproducibility:

1. Matrix matching of standards and samples
   Using calibration standards prepared in the same acid matrix as the samples (including ionic strength or TDS adjustments) would minimize internal standard bias.

2. Improved phosphorus method
   Switching P to H₂ mode, high-mass-shift O₂ mode, or using a reaction gas (e.g., O₂ to convert 31P⁺ → 47PO⁺) could substantially improve sensitivity and linearity.

3. Additional QC checks
   More frequent CCVs could document the internal standard drift more precisely and allow correction.

4. Replication and dilution testing
   Running samples at multiple dilution factors helps distinguish true sample chemistry from matrix suppression.

Overall, the method produced usable data, but matrix control and calibration robustness could be strengthened.

Budget Evaluation

Solution-mode ICP-MS analysis costs $20 per analysis. At 37 runs, including blanks, calibration standards, quality controls, unknowns, and continued calibration, this leads to a total solution-mode cost of $740.