based on the experiments conducted for the factorial design. The
variables’ domains of the Doehlert experimental design were
moved in relation to the full factorial design because better
recoveries were observed at the maximum of these two variables.
The planning used to build the Doehlert matrix is presented in
Table 2, including the three central point repetitions to evaluate
the experimental error, resulting in nine experiments in total.
Because the acid concentration affects the sample introduction
system and the transport of aerosol (Todolı´ and Mermet, 1999),
this factor was studied at five levels, while the sonication time was
studied at three levels. The centrifugation time was fixed at its
lowest level, 5 min, because, as observed in Fig. 1, it had very little
influence on extraction. The centrifugation rotation speed was also
fixed at its lowest level, 2000 rpm, because it had only a small
negative effect on extraction.
The results obtained from the Doehlert matrix in the
optimisation of the ultrasound-assisted extraction procedure are
shown in Table 2. The extraction percentages were in the following
ranges: 91–127% for Ca, 73–93% for Cu, 85–98% for Fe, 91–104% for