Methodology

Methodology: ICP-MS Elemental Concentrations

Using an Agilent 7500ce ICP-MS at the University of Texas at Austin, trace Zn and As concentrations were measured. Spiked samples of known concentrations based on previous sample ranges were used. Oxide production was reduced to less than 2% based on cerium oxides in no-gas mode as the instrument was tuned for optimal sensitivity over the AMU range. The quantitative method utilized an octopole reaction system (ORS) in helium mode (collision-mode) to mitigate polyatomic interferences and in no gas mode to use as a reference for potential interferences. Internal standards were introduced during analysis to correct for instrumental drift, with internal standard sensitivity variations maintained within QA tolerances (±25%).

Limits of detection were calculated from replicate blank (2% HNO3) analyses conducted throughout the sequence, and ranged from 0.07 to 18.13 ppb (median = 0.18 ppb) for all analytes measured in helium mode or no-gas mode. 

Replicates of QCBL4, showed analyte recoveries within 6% of certified values excluding Zn-70 which will be discussed at a later point. Replicates of NIST 1643 f ranged from within 7% of certified values to within 230% of certified values, excluding Zn-70. We recognize that some of these values are large, which indicates that some modes may not be optimal. Relative precision for QCBL4 replicates (n = 4) was generally between 1 and 10% of replicate averages, excluding Zn-70. Relative precision for NIST 1643 replicates (n = 4) was generally between 1 and 8% of replicate averages, excluding Zn-70. A summary of the ICP-MS analytical parameters is provided below.

ICP-MS operating conditions and instrument parameters

Analyte detection limits and recoveries on quality control (QC) standards

Data Processing

Raw data was exported to excel for further analysis. Internal standard variation was then calculated for each run based on counts per second of the internal standard in the calibration blank. Next, samples were separated based on type (calibration, blank, QCs, and spikes). Statistics were calculated for all of the sample types, and the limit of detection was calculated based on the standard deviation of the blank concentration and a student’s t-test for a 95% confidence interval.

MDL = t_{(n-1, 0.95)}S

Where t_{(n-1, 0.95)} is the Student’s t statistic value based on the number of replicates and S is the standard deviation of the replicate measurements (40 CFR 136). The %QRSD, which represents the precision of in-analysis replicates, was averaged for replicate runs and reported as well.

The first two blanks on the run were excluded to ensure no carryover from previous runs. No other blanks were excluded due to this project not using any sticky analytes.