6. AOACSPIFANMethods-2018Awards

158

Pacquette & Thompson: J ournal of AOAC I nternational V ol. 98, N o. 6, 2015  1703

were equipped with modern collision/reaction cells that are thought to be necessary to avoid the Ar/C spectral interferences on the major Cr and Se isotopes. Before actual MLT study samples were analyzed, each participating laboratory was asked to set up the method and evaluate the linearity and the method LOQ with their given instrument model. This exercise is identical to what is done to transfer a mineral method to another site in the authors’ internal laboratory network, as it quickly identifies problems in procuring or preparing suitable standards and standardblanks, or inotherwise setting up the instrument parameters. To check the linearity, standards were analyzed and the calibration curve prepared on each of 3 separate days. On each day, working standards at the lowest concentration level (WS1) and at ½WS1 were analyzed as samples, and then their calculated concentrations were compared to their nominal concentrations. The mean recovery of each standard versus its nominal concentration (i.e., the calibration residual) had to be within 5%. All laboratories passed this test except Laboratories 9 and 11, both of which failed only at the lowest standard level for Se (Table 1). For these laboratories, the practical LOQ (PLOQ) for Se was therefore equal to WS1, whereas the other laboratories could analyze as low as ½ WS1 in concentration. The second setup test was to analyze the sample blank on 5 separate days (done at same time as the linearity study, plus two more days), calculating the SD of these results, multiplying that by 10, and then adding that result to the blank mean. This LOQ was multiplied by the method’s dilution factor of 50 to arrive at the approximate LOQ in terms of sample weight. The SMPRs state an LOQ of 20 ng/g Cr and Mo and 10 ng/g Se on a ready-to-feed (RTF) basis. Table 1 shows the prework results from the participating laboratories (Laboratories 6 and 7 dropped out about this time). Note that it is desirable to have low, consistent blanks for good sensitivity, as well as the linearity, to avoid excessive calibration bias. These trials immediately pointed to Laboratories 1 and 10 as having potential problems; they were allowed to proceed with the MLT, but indeed Laboratory 1’s data were eventually rejected in total. The prework results for Laboratory 10 may not have been so ominous because it did not submit all the data, and the PLOQs could not be calculated.

The final prework for the participating laboratories was to analyze the NIST Standard Reference Material (SRM) 1849a sample. All laboratories passed this test by producing Cr, Mo, and Se results within 5% of the certified means (data not shown, but similar to data collected during the MLT, which is tabulated later). The fact that Laboratories 1 and 10 produced good results on the SRM might be attributed to the relatively high concentration of these elements in the SRM. It should be noted that six of the nine laboratories determined Na, K, P, Mg, Ca, Fe, Cu, Zn, and Mn concurrently with the Cr, Mo, and Se with good precision and accuracy. This work was done under the direction of the Study Director. Results on SPIFAN samples are published in this issue of J. AOAC Int. [AOAC First Action Method 2015.06 by Thompson, J.J., Pacquette, L., & Brunelle, S.L. (2015) J. AOAC Int. 98 , 1711–1720]. Method 2011.19 appears to be viable as a 12-element method, not just as a method for ultratrace elements. Each participating laboratory received blind duplicates of seven of the SPIFAN matrixes (this study used the original SPIFAN set) for a total of 14 samples to test. NIST SRM 1849a was not included as a blind sample, but rather the participants were instructed to analyze it concurrently with the other samples as if it were a control sample. The seven matrixes tested were an infant formula partially hydrolyzed milk-based powder, an adult nutritional low-fat powder, an adult nutritional milk protein- based powder, a child formula powder, an infant elemental powder, an adult high protein nutritional RTF liquid, and an adult high-fat nutritional RTF liquid. Only two infant formula types were chosen (there were four more in the SPIFAN set) because they were known to be unfortified in Cr and Mo and would not yield useful information. Participants were asked to reconstitute all powders prior to analysis with the exception of SRM 1849a, which was unblinded but rather easily identified by its sachet anyway. Participants used a direct weight of 0.2 g SRM powder, which has proven to be homogeneous for minerals at this weight through extensive use in the authors’ laboratories. All other powders were reconstituted by either dissolving 20 g powder in enough laboratory water to make 200 g solution, i.e., a 10% (w/w) reconstitution, or by following the official method with the SPIFAN-recommended 25 g sample + 200 g water (11.1%, w/w). Some laboratories asked to work with the 10% reconstitution rates, as this is certainly an easier

Table 1. Set-up tests for participating laboratories a

LOQ Cr (20 ng/g required), ng/g

LOQ Mo (20 ng/g required), ng/g

LOQ Se (10 ng/g required), ng/g

Lab

PLOQ Cr, µg/L

PLOQ Mo, µg/L

PLOQ Se, µg/L

1 2 3 4 5 6 7 8 9

45

0.4 0.4 0.4 0.4 0.4

30

0.4 0.4 0.4 0.4 0.4

46

0.2 0.2 0.2 0.2 0.2

7 9

5 9 2 1

4 3 1 6

12

9

4

0.4 0.4

1

0.4 0.4

1

0.2 0.4

16 44

13 14

18 66 13

10 11

?

?

?

8 0.4 a  Laboratories 6 and 7 dropped out at this time and Laboratory 10’s data were incomplete. A PLOQ of 0.4 µg/L Cr/Mo and 0.2 µg/L Se along with an LOQ below 20 ng/g (10 ng/g for Se), was desired. See  text for details. 0.4 4 0.4