A Review of NIST Primary Activity Standards for (18)F: 1982 to 2013

Abstract

The new NIST activity standardization for (18)F, described in 2014 in Applied Radiation and Isotopes (v. 85, p. 77), differs from results obtained between 1998 and 2008 by 4 %. The new results are considered to be very reliable; they are based on a battery of robust primary measurement techniques and bring the NIST standard into accord with other national metrology institutes. This paper reviews all ten (18)F activity standardizations performed at NIST from 1982 to 2013, with a focus on experimental variables that might account for discrepancies. We have identified many possible sources of measurement bias and eliminated most of them, but we have not adequately accounted for the 1998-2008 results.

Keywords: F-18; anticoincidence; discrepancy; ionization chamber; liquid scintillation counting; primary standard.

Fig. 1

Decay scheme for ¹⁸F, with LS efficiencies for each branch as calculated by MICELLE2. The calculations were for a Hionic Fluor cocktail with 10 % added water. The CIEMAT/NIST output (H3X.TAB) at ε_H-3 = 0.50 is shown.

Fig. 2

A comparison of the results of NIST LSC-based standardizations of ¹⁸F, normalized to the 2012 LTAC-based standard. See text (Sec. 4.1 and Eq. (5)) for the derivation of Δ°. The bar colors identify the specific ionization chambers used to establish links between the experiments. Blue – VIC; Red – Chamber “A”; Green – Capintec dose calibrator, DS = 439; Purple – Capintec dose calibrator, DS = 472 to 477. The 1982-I link was calculated from the average of the NIH CRC-16 and CRC-30 measurements, with the uncertainty bars corresponding to the standard deviation on those measurements. The 2006 Capintec links are based on measurements with a CRC-15R dose calibrator. All other Capintec links are based on measurements with the NIST CRC-12. Uncertainty bars are estimated standard uncertainties (k = 1); unless more detailed uncertainty information was available, the uncertainties are simply taken from the standard uncertainty on the corresponding primary activity standard. Chamber “A”-based links do not include chamber-height corrections. *The 2008 links are given in outline to emphasize that no LSC measurements were performed in 2008; the 2008 activity was based on the VIC measurements, using the K_VIC determined from the 2001 LSC measurements. Thus, the blue VIC bars from 2001 and 2008 are identical.

Fig. 3

Calculated by attributing the difference in the ⁴⁸V impurity determinations by liquid scintillation counting and HPGe to ³H (see text, Sec. 3.4, for details), this plot qualitatively reproduces the observed rate-dependence of the massic activity from Fig. 2 of Ref. [38]. The calculations assume efficiency values consistent with the 1998 experiments: ε_F-18 = 0.9686, ε_V-48 = 0.6, and ε_H-3 = 0.41. The bands surrounding the dark line represent the bounds calculated by propagating the 1998 estimated uncertainty on the activity of the ⁴⁸V impurity. Since only the first few cycles were used for the standard activity determination, the average and maximum biases stemming from the impurity misattribution are 0.25 % and 0.49 %, respectively.

Fig. 4

Comparison of NIST ¹⁸F standardization results from 2012 to 2013 [41]. All results are normalized (via K_VIC links) to the 2012 LTAC determination. The different symbols correspond to the three different experiments: open diamonds – 2012-I; open squares – 2012-II; open triangles – 2013. The uncertainty bars represent combined standard uncertainties (k = 1). The dotted lines represent the KCRV from the CCRI(II)-K3.F-18 comparison [25] and its standard uncertainty. TDCR data from 2012-I is not represented in this figure (see Sec. 3.8 for details).