30 / 65
FOR EXPERT REVIEW PANEL USE ONLY
NOT FOR DISTRIBUTION
2
•
ICP-AES is more common in labs (Availability / instrument affordability).
Cons/Weaknesses
•
Performance at low concentration levels for some elements (Mn, Fe, Cu)
Supporting Data
•
General Comment:
Well written and comprehensive report on the MLT. The report provides details in the study design and
results analysis. It contains collaborators, study designs, equipment used, data results, comments from the
participating labs, and final conclusion. The conclusion is fairly reasonable.
-
Method Optimization:
•
Performance Characteristics:
Analytical Range:
LOQ: Meet the SMPR requirement on Ca, Mg, P, K, Na, and Zn.
Accuracy/Recovery: Good performance on the NIST Reference material (1849a)
Precision (RSD
r
) and Reproducibility (RSD
R
):
Good repeatability and reproducibility for all elements, except the three elements, Mn, Fe, and Cu
at low concentrations.
For manganese, iron, and copper, the RSDs at or near the LOQ (up to 50 times of the LOQ) are
larger than the SMPR requirement.
1) Copper, at about 0.05 mg/100g RTF, the RSDs are about 10% (which is the requirement limit).
The SMPR LOQ is 0.001 mg/100g RTF.
2) Manganese, at 0.008 the RSD
R
still above 16% (the limit is 10%), the SMPR LOQ is 0.001.
The report has pointed out that if look at these RSDs in terms of HorRat value, they are not too bad,
all are less than 2.0.
•
System suitability:
Recommendation:
The MLT results of this method showed excellent accuracy, precision and reproducibility results except that
for Cu, Mn, and Fe at low concentration levels the RSD
R
values do not meet the SMPR requirements.
I would reluctant to recommend this method as the final method against the existing SMPR. On the other
hand, based on the MLT results, I am confident that this method will allow accurate and precise
determination of these nine elements at the majority of the SMPR analytical ranges. Only for the two or
three elements (Copper, Manganese, and Iron) at near their SMPR required LOQ levers, they show large
determination uncertainties, or RSD. I think this method is worth to be recommended for slightly modified
analytical ranges.