Biomedical Engineering Reference
In-Depth Information
0
−
50
Zn 1 ppm Cu 0.3 ppm
Intermetallic compound
−
100
Zn 0.5 ppm Cu 0.3 ppm
Zn 0.3 ppm Cu 0.3 ppm
−
150
Cu 0.3 ppm
−
200
−
250
−
1.2
−
0.9
−
0.6
−
0.3
0
0.3
0.6
0.9
1.2
1.5
Bias potential (V)
FIGURE 9.2
Examples of curves obtained with our PSA analyzer with synthetic samples containing lead, copper, cadmium,
and zinc. The figure shows the linear behavior between peaks area and concentration for all the analytes we
tested. The curves are obtained under the experimental conditions specified in Adami et al. (2005).
tools, which are user-friendly and easy to use, allow the detection of the analytical signals
and extraction of the information data. With synthetic sample we obtained enough sensitiv-
ity to propose a first-screening apparatus but, when the sample is a real matrix, it can con-
tain some different components, in particular some different metal ions, each one acting as
a possible interferant. To overcome this problem, we tried to integrate in the proposed sys-
tem a neural network algorithm, the multilayer perceptron (MLP), which appears to be able
to reveal the real concentrations of the different metal ions, after a severe training (Figure
9.3). The PSA analyzer is based on the schematic block diagram proposed in Figure 9.1(a). A
computer drives the apparatus and a suitable software interface allows even nontechnical
personnel, without any specific hardware knowledge, to perform the experiments. The ana-
lyzer drives a glassy-carbon working electrode, a calomel reference electrode, and a plat-
inum counter electrode; each user can utilize its proper electrochemical cell and its proper
electrodes. The system is designed to perform the two-step process described in an article by
Adami, Sartore, and Nicolini (61): during the first stage the working electrode polarization
causes the deposition of the reduced metal ions on it. A potentiostat (62,63) drives the elec-
trochemical cell through the counter electrode ensuring the proper biasing voltage
V
ref/wk
. In
the second step (Figure 9.1(b)), a constant oxidizing current (1-100
A) is fed into the work-
ing electrode and the metal is forced to move again into the solution, producing a potential
drop between reference and working electrodes. The system becomes a galvanostat and the
potential drop
V
ref/wk
is now recorded as a function of time. The instrument software consists
of two distinct parts: a low-level driver, which manages the data transfer to and from the
interface electronics, and a high-level application, which provides a user-friendly interface
with data acquisition and analysis options. The software tools are entirely written in C under
the LabWindows CVI 5.5 environment (National Instruments). It operates under Microsoft
Windows 95/98/NT/2000. The data acquisition software implements an algorithm to drive
the experimental procedure, by setting the proper parameters on the experimental windows
(61). The data analysis software transforms and plots the acquired data
E
(
t
) in the inverse
derivative form d
t
/d
E
(64,65) as a function of the recorded potential (Figure 9.3): those por-
tions of the stripping potentiogram exhibiting a plateau are evidentiated, in the derivative
