714 



Fishery Bulletin 101 (3) 



were calculated from normalized mole fractions of 

 Sr and Ca. Limit of detection (3a: Potts, 1987) was 

 calculated for all points in both high and low Sr/Ca 

 regions. 



Effect of beam diameter and beam current 



We conducted five repeat measurements at each of 

 five points within the high and low Sr/Ca regions 

 using a 1-. 7-, 15-, 20-. and 25-pm-diameter beam at 

 15 kV and 50 nA. This resulted in beam power densi- 

 ties of 961, 19.5, 4.2, 2.39, and 1.52pW/pm^ respec- 

 tively. We conducted five repeated measurements at 

 each of four locations within the high and low Sr/Ca 

 regions using beam currents of 5, 10, 20, and 30 nA 

 with a 10-m-diameter beam and accelerating voltage 

 of 15 kV, with resulting beam power densities of 1.0, 

 1.9, 3.8, and 5.7pW/pm-, respectively. The coefficient 

 of variation (CV) of Sr/Ca ratios was calculated as the 

 ^^Sr/Ca ^ Mean Sr/Ca"' for each beam power density. 

 If beam damage affects precision and accuracy of 

 Sr/Ca ratios, subsequent measurements at the same 

 spot should be increasingly divergent from the first 

 and such divergence should be evident in high coeffi- 

 cients of variation. The limit of detection was used as 

 a measure of precision. Limit of detection (3a; Potts, 

 1987 ) for Sr was calculated for the first measurement 

 taken at each beam power density in each region. 



Results 



Effect of spectrometer (crystal) choice 



Spectrometer (crystal) choice for the measurement of Sr 

 had an apparent systematic effect on Sr/Ca ratios at high 

 Sr/Ca levels but no effect at low Sr/Ca levels (Fig. lA). The 

 mean Sr/Ca level in the high Sr/Ca region was significantly 

 lower (-15%) when Sr was measured on the PET crystal 

 «=7.189; P<0.001; df=38). Crystal choice had an effect on 

 measurement of both Ca and Sr in the high Sr/Ca region. 

 Ca did not differ significantly between the high and low Sr/ 

 Ca regions (P>0.05) but was approximately 2% lower when 

 Sr was measured with the PET crystal (Fig. IB). This dif- 

 ference was attributable to beam damage, which occurred 

 as Sr was measured. The mean Ca was 197.400 ppm when 

 Sr was measured on the PET crystal and 202,100 ppm 

 when Sr was measured on the TAP crystal. 



In the high Sr/Ca region, mean Sr was significantly 

 lower when Sr was measured on the PET crystal (^=11. 58; 

 P<0.001;df=38)(Fig. lBi.Themean(±SD)was7270±4ppm 

 when Sr was measured with the PET crystal and 8870 

 ±3 ppm when Sr was measured with the TAP crystal. This 

 difference is also reflected in the higher minimum limit of 

 determination for Sr for PET in both the high Sr/Ca (695 

 ppm) and low Sr/Ca regions (126 ppm). Using the TAP 

 crystal to measure Sr, we found that the minimum limit of 

 determination for Sr was 103 ppm and 65 ppm in the high 

 Sr/Ca and low Sr/Ca regions, respectively. To achieve simi- 

 lar counting statistics for Sr with the PET crystal, count 



times would need to be increased to 200 seconds on both 

 the peak and background. 



Effect of beam diameter and beam current 



In repeated measurements with different beam diameters 

 (same beam current) at the same locations, Sr/Ca ratios 

 did not vary greatly with the exception of measurements 

 made with the 1-pm beam (Fig. 2A). The CV of the Sr/Ca 

 ratios for the 1-m beam was high and led to significant 

 variation of Sr/Ca ratios in subsequent measurements at 

 the same point (Table 2). The Sr/Ca ratio was least vari- 

 able for the 7-pm beam in the high Sr/Ca region (Table 2). 

 Limit of detection (3a) for Sr ranged from 80 ppm to 172 

 ppm in the low Sr/Ca region and from 299 ppm to 315 ppm 

 in the high Sr/Ca region under the various beam diameters 

 (Fig. 3A). 



Beam current had a significant effect on variation of 

 Sr/Ca and limit of detection for Sr. The greatest variation 

 in Sr/Ca ratios was observed at beam currents 5nA and 10 

 nA (Fig. 2B). The CV of Sr/Ca ratios was negatively related 

 to beam current (r=-0.93; P>0.05) (Table 3). The CV of Sr 

 was high in all treatments, ranging from 0.23 to 1.17 in the 

 high Sr/Ca region and from 0.11 to 0.46 in the low Sr/Ca 

 region. The CV of Ca was 0.04 for all beam configurations. 

 The limit of detection of Sr as measured at the first sample 



