Arsenic in Rice Flour: A 172% CRM Recovery
An ICP-MS laboratory determines arsenic in rice flour. A 0.500 g portion is microwave-digested in 5 mL concentrated HNO3 and 2 mL H2O2 and diluted to 50.0 mL. The method monitors m/z 75 for As and uses 72Ge as internal standard; no collision gas is used. A rice flour certified reference material certified at 0.285 mg/kg As gives 0.490 mg/kg. A sample gives 0.42 mg/kg, and a 0.50 mg/kg spike gives 155% recovery. Duplicate sample results agree within 4% RSD. The 72Ge internal-standard response in the samples is 70% of that in the calibration standards. The laboratory's reporting limit is 0.05 mg/kg, and the food limit is 0.50 mg/kg. The analyst suggests correcting the result using the internal-standard response and reporting the sample as 0.42 mg/kg, below the limit.
1. Diagnose the most likely cause of the high CRM and spike results. Why does the internal-standard correction not fix it?
Read the worked answer
The pattern is an additive interference, not simple signal drift. At m/z 75, arsenic is monoisotopic, and the common polyatomic interference is 40Ar35Cl+, which appears at the same nominal mass. Rice contains chloride, and the digest can also contribute chloride. With no collision gas, the interference is transmitted with the analyte. A CRM recovery of 172% and a spike recovery of 155% are too large for ordinary drift, and the 4% duplicate agreement shows the error is reproducible.
The 72Ge internal standard corrects multiplicative changes in sensitivity, not an additive isobaric or polyatomic signal. The 72Ge response at 70% indicates matrix suppression, but normalizing to it would raise the reported arsenic further, not lower it. The internal-standard correction is therefore inappropriate as a fix. The laboratory should check chloride in the digest, monitor 35Cl or 37Cl, and test a chloride-only blank. The likely remedy is a collision/reaction cell, such as helium with kinetic energy discrimination, or ICP-MS/MS with oxygen mass shift to AsO+ at m/z 91.
2. Design confirmation experiments to prove the interference and select a corrected mode.
Read the worked answer
First, analyze a chloride-only solution and a method blank at m/z 75; if a signal appears at the arsenic retention mass, the interference is demonstrated. Add NaCl to a low-arsenic rice sample at several chloride levels and show that the apparent arsenic increases with chloride. Monitor 35Cl or 37Cl in the same runs. Then compare three modes on the same digest: no gas, helium collision with kinetic energy discrimination, and ICP-MS/MS with oxygen mass shift. In each mode, run the rice CRM, a spike, a duplicate, and a method blank.
Select the mode that gives CRM recovery within 85–115%, spike recovery within 85–115%, and duplicate RSD ≤10%, while keeping the internal standard within the laboratory's acceptance range. Verify that the m/z 75 signal in the chloride-only solution is reduced to the blank level. If collision gas is insufficient, use a different interference-free mass or a confirmatory technique such as hydride-generation AAS or ICP-OES after separation. Document the chosen mode and the chloride check in the SOP.
3. What can be reported for the sample, and what validation/QA criteria should be set before re-reporting?
Read the worked answer
Do not report 0.42 mg/kg. The CRM and spike failures invalidate the run, and the result cannot be rescued by applying the internal-standard response. Re-digest and re-analyze the sample with the interference-controlled mode, then report the new result with its uncertainty.
Before re-reporting, require: CRM recovery 85–115%; spike recovery 85–115%; duplicate RSD ≤10%; internal standard response within 60–125% of the calibration mean; method blank below the reporting limit; and a chloride check or interference check at m/z 75. If the sample is near the 0.50 mg/kg limit, confirm with an independent method or a second interference-free mass. Update the SOP to require collision/reaction gas for arsenic in rice and to monitor chloride in every digestion batch. Retain the failed run data and the investigation record.