This ICP-MS method was developed to enable high-resolution comparison of inorganic water chemistry between municipal tap water and commercially bottled drinking water sources consumed in the West Campus area of Austin, Texas. West Campus has a uniquely dense residential population with high bottled-water usage, despite access to treated municipal water from the City of Austin system (TX2270001). Comparing trace and major-element concentrations between these sources allows assessment of whether bottled water provides measurable chemical advantages over tap water, or whether differences are primarily driven by marketing and consumer perception rather than composition. The method was therefore designed to achieve low detection limits, robust interference control, and high precision across a wide concentration range, enabling direct comparison of regulatory-relevant constituents and naturally occurring metals in both water types.
Methodology
Sample Preparation
Water samples, field duplicates, and quality-control standards were prepared for solution-mode ICP-MS analysis following EPA Method 200.8–style procedures. All samples were transferred into acid-washed polypropylene vials. Each aliquot was acidified to pH < 2 using ultrapure HNO₃ (Optima grade) to preserve dissolved metals and prevent adsorption to container walls. Calibration standards, continuing calibration verification standards (CCVs), laboratory fortified blanks (LFBs), and matrix spikes (MS/MSD) were prepared from NIST-traceable multi-element stock solutions using 18.2 MΩ∙cm DI water.
Dilutions of 10x were performed to minimize matrix suppression and reduce total dissolved solids below 0.2%. Internal standards (Sc, Ge, Rh, In, Tb, Bi) were added online to all samples at 10 µg/L.
Instrumental Method (Agilent, MassHunter Workstation)
Analytes and Collision/Reaction Cell (ORS) Modes
The method quantified major and trace elements including Na, Mg, Al, P, K, Ca, V, Cr, Mn, Fe, Cu, Zn, As, Se, Sr, Ba, and Pb. Each isotope was assigned to an ORS mode depending on expected polyatomic interferences:
| Isotope | Mode | Rationale |
|---|---|---|
| 23Na, 27Al, 31P, 39K | No Gas | Minimal interferences |
| 40Ca, 56Fe, 78Se | H₂ mode | Mitigates ArO⁺, ArN⁺, Ar₂⁺ |
| 51V, 52Cr | He mode | Reduces ClO⁺, ClOH⁺ |
| 75As | He mode | Removes ArCl⁺ |
| 88Sr, 137Ba, 206Pb | No Gas | High sensitivity isotopes |
Dwell times ranged between 0.1–0.3 s per mass, with 3–5 replicate readings per integration and 100 sweeps per sample.
Instrument Optimization
Before acquisition, the ICP-MS was tuned using a multi-element tuning solution (1–10 µg/L Li, Y, Ce, Tl):
-
Oxide ratio (CeO⁺/Ce⁺) adjusted to < 2.0%.
-
Doubly-charged ratio (Ce²⁺/Ce⁺) tuned to < 3.0%.
-
Sensitivity maximized on 7Li⁺, 89Y⁺, and 205Tl⁺.
-
Background noise minimized across m/z 5–220.
-
Lens voltages, nebulizer gas flow, and plasma RF power optimized for stability and < 2% RSD on tuning solution replicates.
Cell gas flows (He or H₂) were finalized by maximizing analyte counts while minimizing known polyatomic interferences.
Analytical Sequence
1. Plasma Warm-Up and Auto-Tune Verification
Before sample analysis, the instrument completed an automated performance tune using a multi-element tuning solution. Oxide, doubly charged ion ratios, sensitivity checks, and mass calibration were verified to be within manufacturer specifications.
2. Initial Calibration Blank (ICB)
A 2% HNO₃ reagent blank was analyzed to establish baseline background intensities and ensure that no residual analyte signal persisted from previous runs.
3. Multi-Point Calibration (L2–L6)
Five calibration standards spanning 1–4 orders of magnitude were analyzed in increasing concentration:
-
L2
-
L3
-
L4
-
L5
-
L6
For phosphorus, calibration levels L2–L4 were later excluded due to curvature in the response; the calibration was reprocessed using L5 and L6 only.
Each standard included online internal standards (Sc, Ge, Rh, In, Tb, Bi), and Quant RSDs were monitored to assess short-term precision.
4. Initial Calibration Verification (ICV / QC2)
A second-source or mid-level calibration check (equivalent to L4) was analyzed immediately after calibration to confirm accuracy before unknowns were run. Acceptable recovery was defined as within ±10% of the expected concentration.
5. Procedural Blank (LRB)
A reagent blank was analyzed to quantify background contributions from sample preparation, acid source, or autosampler memory effects.
6. NIST 1643f (10×) Quality-Control Sample (QC1)
A diluted certified reference material (10× NIST 1643f) was measured to evaluate external accuracy and to confirm that the instrument reproduced known concentrations for major and trace metals.
7. Unknown Samples (AP1–AP16 and Others)
All bottled-water samples, student-prepared dilutions, and additional controls were analyzed in randomized order to minimize drift-related bias. Each sample was measured using the best analyte/mode combination established during method development (No Gas, He, or H₂ modes depending on the isotope). Ca was quantified using 40Ca in H₂ mode to suppress Ar₂⁺, and supported by No Gas measurement of 43Ca or 44Ca for comparison.
8. Continuing Calibration Verification (CCV)
After every 10 samples, a mid-level calibration check was analyzed to evaluate instrumental drift. Recoveries outside ±10% triggered reinstrument calibration or data review.
9. Intermittent Blanks
Instrument rinse blanks (2% HNO₃) were inserted periodically to monitor carryover. These blanks also helped identify the matrix differences later observed in internal standard recoveries.
10. End-of-Run QC Checks
The sequence concluded with:
-
CCV (final)
-
CCB (continuing calibration blank)
These ensured that calibration remained valid throughout the full analytical sequence and that no drift or contamination accumulated by the end of the run.
Data Quality Evaluation
Calibration Linearity
Each analyte was calibrated using a six-point external curve (0, L2–L6). Calibration data was based on the EPA and TCEQ data for the CITY OF AUSTIN WATER & WASTEWATER water area (TX2270001). Back-calculated concentrations for the standards increased monotonically over 3–4 orders of magnitude (for example, Na from 0 to 18,600 µg/L; K from 0 to 4,600 µg/L; Pb isotopes from 0 to ~100 µg/L). All calibration functions in MassHunter were strongly linear, with coefficients of determination R² ≥ 0.9993 for every isotope, and most analytes exhibiting R² ≥ 0.9997.
Limits of Detection (LOD)
Method detection limits were estimated from five instrument blanks for each isotope using the Student-t approach (LOD = t₀.₉₉,ₙ₋₁ · σ_blank). The resulting LODs ranged from approximately 0.001 to 3.4 µg/L. Most trace metals (V, Cr, Mn, Fe, Cu, Zn, As, Sr, Ba, Pb) had LODs below 0.1 µg/L, whereas some major ions and difficult analytes (K, P, Se in He mode) showed higher LODs (≈1–3 µg/L).
Precision
Within-sample precision (MassHunter “Quant RSD”) for calibration standards and unknowns was generally <5 % for major cations (Na, K, Ca, Sr, Ba) and <5–10 % for most trace metals over the calibrated range. Higher RSD values occurred mainly for low-level P, As, and Se near the detection limit.
Accuracy and QC Stability
Accuracy and instrumental stability were monitored using NIST 1643f (10× dilution; QC1) and an independent mid-range check solution at the L4 level (QC2). Across the run, QC1 and QC2 recoveries varied by only a few percent for Na, K, Ca, Sr, Ba, and most trace metals, indicating stable sensitivity and negligible drift over the analytical sequence.
Detection Relative to Blanks
Concentrations in bottled-water and tap-water samples were typically one to three orders of magnitude higher than blank-equivalent concentrations. For example, Na in samples ranged from ~3 to ~1900 µg/L (LOD ≈ 0.07 µg/L), and K from ~9 to ~545 µg/L (LOD ≈ 3.4 µg/L). This confirms that, for most analytes, the reported values are well above the method detection limits.