Page:NIOSH Manual of Analytical Methods - 5524.pdf/9

METALWORKING FLUIDS (MWF) ALL CATEGORIES: METHOD 5524, Issue 2, dated 29 December 2014 - Page 9 of 10

Cleanup Procedure Evaluation This cleanup procedure has been incorporated into NMAM Method 5524 in order to deal with reported spurious weight gains and losses before and after analysis of the PTFE filters used with this method. It is believed that trace levels of dust or extractable material are entrained in these filters during the manufacturing process. This cleanup procedure has been evaluated using 60 filters from three different batches of PTFE filters (20 filters/batch) [1]. Prior to cleaning, the filters were weighed (untreated filters), then washed with the ternary blend according to the procedure in Appendix 1, dried, and then reweighed (treated filters). Results: The differences in each of the 3 batches were compared by subtracting the treated filter weights from their untreated weights. For all 60 samples, the overall mean difference in weights (+/-the standard error of the mean) was 0.2 µg (± 1.4 µg). For each of the 3 batches, the differences in weight were reported as mean (+/-standard error): 5 µg (± 2.5 µg) , -2 µg (± 2.0 µg), and -3 µg(± 2.7 µg) for batches 1, 2 and 3, respectively. The weight differences were not statistically significant from zero overall or by batch using a paired t-test (p = 0.05), and allowing for multiple comparisons. To determine if the washing procedure affected the filter’s performance for analysis of metalworking fluids, each of the filters was analyzed according to the procedure of NMAM Method 5524, which includes extraction with the binary and ternary solvent blends. The following differences were computed: post-analysis weights of the filters minus their treated or untreated weights. The average difference in the weights of the 60 untreated filters and their post-analysis weights reported as mean (± standard error of the mean) was 34 µg (± 1.9 µg). For each of the 3 batches, the differences in the untreated and post-analysis weights were: 30 µg (± 3.9 µg), 38 µg (± 2.0 µg) and 34 µg (± 3.3 µg) for batches 1, 2 and 3, respectively. These differences are statistically significantly different from zero, both overall and individually by batch, using a paired t-test (p = 0.05), and allowing for multiple comparisons. The average differences in the weights of the 60 treated filters and their post-analysis weights were: 34 µg (± 1.4 µg). Again the differences in weight are reported as mean (± standard error). For each of the 3 batches, the differences, (post-analysis minus treated) weights, were: 35 µg (± 3.4 µg), 36 µg (± 2.0 µg), and 31 µg (± 1.8 µg) for batches 1, 2 and 3, respectively. These differences are statistically significantly different from zero, both overall and individually by batch, using a paired t-test (p = 0.05), and allowing for multiple comparisons. In summary, the weight differences were statistically different from zero by batch and overall using a paired t-test for both the post-analysis minus treated weights and for the post analysis-untreated weights (p = 0.05). However, the untreated/treated differences were not statistically different using the same tests. These experiments indicate that cleaning the filters lowered the overall LOQ of the analytical method. The LOQ determined from the differences in weights between the untreated and analyzed filters was 140 µg and was greater than the LOQ of 110 µg determined from the differences in weights between the treated and analyzed filters. The extraction procedure produced a more consistent blank and therefore a lower average standard deviation from which lower LOD s and LOQs were determined. [1] NIOSH [2014]. CEMB Analytical Services Report: Sequence 11600-CA. Novi, MI: Bureau Veritas North America, unpublished.

NIOSH Manual of Analytical Methods (NMAM), Fifth Edition