Sie sind auf Seite 1von 7


Sequential Determination of Salicylic and Acetylsalicylic Acids by

Amperometric Multisite Detection Flow Injection Analysis
University of Porto, CEQUP/Department of Physical Chemistry, Faculty of Pharmacy, Rua Anbal Cunha, 164, 4050-047
Porto, Portugal
University of Valladolid, Department of Analytical Chemistry, Faculty of Science Prado de la Magdalena s/n, 47005
Valladolid, Spain
An amperometric multisite detection flow injection
analysis (FIA) system was developed for sequen-
tial determination of 2 analytes with a single sam-
ple injection and single detector. Tubular compos-
ite carbon electrodes with an inner diameter
similar to that of the FIA manifold tubing were con-
structed so that measurements could be made
without impairing the sample plug hydrodynamic
characteristics. The electrochemical behavior of
the tubular voltammetric cell in a low-dispersion
FIA manifold and the behavior of the FIA system
incorporating this type of voltammetric cell in-
tended for multisite detection were evaluated by
performing measurements with potassium
hexacyanoferrate(II). Feasibility of the approach
was demonstrated in the sequential determination
of salicylic and acetylsalicylic acids in pharmaceu-
tical products at a fixed potential of 0.98 V. The
system allows sequential determination of salicylic
acid concentrations ranging from 1.0 10
5.0 10
M and acetylsalicylic acid concentrations
between 1.0 10
and 5.0 10
M with good preci-
sion on both detection sites and with relative stan-
dard deviations (RSDs) 1.5% (n = 10) and 2.1%
(n = 10), respectively. A comparison of these re-
sults with those of the U.S. Pharmacopeia proce-
dure showed RSDs <5.0 and 1.0% for salicylic acid
and acetylsalicylic acid, respectively. The pro-
posed method enables 15 determinations per hour,
which corresponds to the analysis of approxi-
mately 8 samples per hour. The detection limits of
the methodology were approximately 3.5 10
1.1 10
M, respectively, for the first and second
monitoring sites.
ultisite detection is achieved in flowinjection analy-
sis (FIA) when a single detector is placed at different
sites of the manifold (1). Afewpapers have reported
the analytical potential of relocatable detectors in FIA, with
use limited to spectrophotometric (13) and potentiometric
detection (4). Those papers suggest that advantages of the
multisite detection are limited by the physical characteristics
of the detector owing to difficulties in connecting it to the
manifold. Large sample or carrier volumes may be retained in
the interior of the detector and then dragged by the detection
system at commutation time, causing problems related to liq-
uid interfaces. These difficulties probably explain the few an-
alytical applications found in literature.
The electrochemical cells most frequently used in FIAwith
amperometric detection are wall-jet or thin-layer cells. How-
ever, none of these geometries is appropriate for serial moni-
toring. Considering the way in which wall-jet cells operate
(the flow hits the center of the electrode and is dispersed in
3 dimensions), the sample zone is destroyed. On the other
hand, the use of thin-layer cells impairs the hydrodynamic
characteristics of the sample plug throughout the measure-
ment step by causing a sample dispersion that might compro-
mise a second measurement over the same sample plug.
This paper describes the construction of tubular composite
electrodes for incorporation in an electrochemical cell suitable
to be used in amperometric multisite detection FIAmanifolds.
The performance of this electrochemical cell was assessed
both in a low-dispersion FIAsystemand a sequential monitor-
ing FIA system by using the well-characterized ferri-
cyanide/ferrocyanide redox system.
The applicability of amperometric multisite detection by
using this type of detectors is demonstrated by the sequential
determination of salicylic (SA) and acetylsalicylic (ASA) ac-
ids in pharmaceutical products. ASA is an analgesic agent,
widely used both alone and in combination, which is rapidly
hydrolyzed to SA on exposure to moisture. The SA content as
an impurity in ASA formulations is limited to 0.1% by the
U.S. Pharmacopeia (USP; 5).
Simultaneous determination of these compounds has been
performed by colorimetry (6), fluorimetry (7), Raman spec-
troscopy (8), infrared spectroscopy (9), gasliquid chroma-
Received February 13, 2002. Accepted by JM May 2, 2002.
Author to whom correspondence should be addressed; e-mail:
tography (10), thin-layer chromatography (11), and liquid
chromatography (LC; 12, 13). The fewelectrochemical meth-
ods that have been proposed report only the determination of
SA content (1416).
The present method has been used to determine SA and
ASA in pharmaceuticals available on the Portuguese market.
Results are compared with those provided by the chromato-
graphic method recommended by the USP (5).
(a) FIA systems.Solutions were propelled by a Gilson
Miniplus 3 peristaltic pump; intercalation of sample plugs in
the carrying solutions was conducted through an injec-
tor-commutator of circular configuration with functioning
identical to that described by Krug et al. (17). Besides func-
tioning as an injector device for the sample plugs, the injec-
tor-commutator allowed movement of the tubular detector be-
tween the 2 monitoring sites of the manifold.
(b) Teflon tubing.0.8 mm id Omnifit (Cambridge, UK);
used to make all tubes and connections.
(c) Homemade confluences and reference electrode sup-
port (18)
(d) Detector.Used to make amperometric measure-
ments; Metrohm 641 VA (Buckingham, UK) connected to a
Kipp & Zonen recorder.
(e) LC system.LC determinations of SA and ASA were
made according to the procedure described in the USP (5) and
performed in a Merck Hitachi (Darmstadt, Germany) chro-
matographic system comprising Model 7100 pump,
Rheodyne 7725i injector (10 Lloop; Rohnert Park, CA), and
Lichrocart RP 18 column (250 4 mm) packed with
Lichrosorb 5 m beads (Novato, CA). A Merck Hitachi
Model 7000 diode array detector was used; data were pro-
cessed by D-7000 software of the same brand.
Reagents and Solutions
All reagents were of analytical grade, and aqueous solu-
tions were prepared with purified water from a Millipore
Milli-Q system (Bedford, MA), conductivity <0.1 S/cm.
(a) FIA system.The intrinsic working characteristics of
the tubular electrochemical cell and the multisite detection
FIA manifold were evaluated by using a 0.1M KCl (Riedel de
Han, Seelze, Germany) solution as carrier and K
(Riedel de Han) standard solutions within a concentration
range of 1.0 10
to 1.0 10
M prepared in 0.1M KCl.
(b) Phosphate buffer solution.In the optimization of the
system for sequential determination of SA and ASA in phar-
maceutical products, the support electrolyte consisted of a
phosphate buffer solution (H

, pH, 7) prepared
by mixing 524 mL 0.8M K
solution (Merck,
Darmstadt, Germany) with 476 mL 0.2M KH
(c) Aqueous stock solution.An SA (Sigma, St. Louis,
MO) concentration of 1.0 10
M was prepared weekly by
weighing the solid and keeping it away from light. The stan-
dard solutions containing SAand ASAwere prepared daily by
measuring adequate volumes of SAstock solution and weigh-
ing accurate amounts of ASA (Sigma), which were diluted to
100 mL with water in a volumetric flask. Concentration of
these solutions ranged from 1.0 10
to 5.0 10
M for SA
and from 1.0 10
to 5.0 10
M for ASA.
After preparation, aqueous standards and samples were
placed in an ice bath and protected from light until use in the
automated system to minimize probable hydrolysis (19).
Figure 1. Construction and assembly of tubular electrodes: (A) Electric cable is attached with electric solder to
rectangular silver plate; (B) fragment of master pellet is glued to silver plate with conductive silver-based epoxy resin,
and cavity of Perspex holder is filled with nonconductive epoxy resin (ncr); (C) electrode active surface is housed in
Perspex holder; after hardening a channel is drilled and PVC connections (pvc) are applied.
The active surface of the tubular electrodes (working and
auxiliary) was prepared by dissolving a given amount (0.25 g)
of paraffin wax (Fluka, Buchs, Switzerland) in 10 mL warm
n-hexane (Sigma; 40C) in a beaker placed in a water bath and
subsequently adding 4.75 g graphite powder (Merck) with
stirring until the n-hexane completely evaporated. Then,
0.20 g dry graphite powder, now containing 5% (m/m) paraf-
fin wax, was pressed into a pellet with a 10.0 mm diameter
pellet press at 19 000 kg/cm
for 5 min to obtain a disc of
10.0 mm diameter and 1.2 mm thickness.
Electrical contact was made through an electric cable at-
tached with an electric solder to a small rectangular silver
plate (Aldrich, Milwaukee, WI; 1.0 3.0 mm; Figure 1A) to
which a square fragment of the master pellet was glued with a
conductive silver-based epoxy resin (components A and b,
EPO-TECH 410; Epoxy Technology, Billerica, MA).
The fragment of the master pellet was then housed (Fig-
ure 1B) in a small Perspex cylinder (10.0 13.0 mm) with a
parallel-piped cavity (3.0 4.5 7.5 mm) filled with a
nonconductive epoxy resin (1.0 g Araldite M and 0.4 g HR
hardener; Ciba-Geigy, Basel, Switzerland; Figure 1C). Af-
ter hardening, a channel of 0.8 mm diameter was drilled,
perpendicular to the opposite sides of the housing, through
the center of the Perspex cylinder (Figure 1C). Finally, the
connections for the Teflon tubing were made with polyvi-
nyl chloride (PVC) tubes. The reference electrode was an
Orion 90-02-00 double-junction AgCl/Ag (inner filling solu-
tion, Orion 90-00-02; outer filling solution 0.1M KNO
Tubular Electrochemical Cell
The tubular electrochemical cell (Figure 2A) was com-
posed of a working (Figure 2A, E
) and an auxiliary electrode
(Figure 2A, E
), both tubular and prepared as described
above. These electrodes were separated by the contact (Fig-
ure 2A, T) with the reference electrode (Figure 2A, E
). The
reference electrode incorporated in the detection system was
placed in a Perspex support, previously described (18).
The tubular electrochemical cell was connected to the slid-
ing part of the injector-commutator by Teflon tubes (0.8 mm
id) with the shortest possible length. The total inner volume of
the relocatable detector was 105 L, according to the proce-
dure for determination of the injection volume in classic FIA
systems (20).
The injector-commutator switching mode enabled estab-
lishment of a 2-position FIA manifold. In position X (Fig-
ure 2B) the sample plug is inserted in the flowsystemand car-
ried toward the detector (positioned at x) where the the sample
is measured for the first time. By moving the central part of the
injector-commutator to position Y, the detector is moved to y,
where the sample plug, after passing reactor R, is measured
for a second time. At the same time, the loop (S) is filled with a
new sample solution. When the sliding part of the injec-
tor-commutator is moved back to position X, a new sample
plug is injected and the detector returns to position x. This
Figure 2. (A) Tubular electrochemical cell; (B) schematic representation of injector-commutator; and (C) flow
injection system with serial monitoring. Ew = working electrode; Eaux = auxiliary electrode; T = contact with reference
electrode; Eref = reference electrode.
movement of the sliding central part of the injector-commuta-
tor allows the sample loop to be filled and injected in the car-
rying solution, and the detector to be moved between 2 differ-
ent positions of the manifold (Figure 2C).
Reference Procedure
To assess the accuracy of the results obtained by the devel-
oped procedure, we analyzed ASA tablets according to the
USP method (5). The weight of 20 tablets was averaged, the
tablets were ground to a fine powder, and adequate portions of
the ground tablets were dissolved in a mixture of acetonitrile
and formic acid (99 + 1, v/v) for analysis by the USP method
and in water for analysis by the proposed methodology.
Results and Discussion
Performance of the Tubular Electrochemical Cell
The intrinsic response characteristics of the tubular electro-
chemical cell were determined in a low-dispersion FIA mani-
fold presenting the detector in a fixed position and a 30 cmcoil
between the injection and detection site. A 0.1M KCl solution
was used as carrier, and different volumes of K
] so-
lutions with a concentration of 10
Mwere injected. The peak
current was measured by applying a 0.50 V potential.
The detection systemcomprising the tubular electrodes gave
a fast response to the changes of hexacyanoferrate(II) concen-
tration; the peak current was about 90% relative to that of the
steady-state for injection volumes of 250 L and a flow rate of
1.0 mL/min. The return to baseline, which is normally condi-
tioned by washing the sensor unit surface with carrier solution,
accounts for the dependence of sampling rates on sample con-
centration. Therefore, K
] solutions with concentra-
tions of 10
M yielded sampling rates of 100/h, which in-
creased to 130/h when concentrations of 10
M were used.
Testing the electrochemical detector with potassium
hexacyanoferrate(II) within the concentration range of
1 10
and 1 10
Mgave a linear response between analyt-
ical signal and sample concentration (r $ 0.999, n = 6).
Reproducibility of the analytical signal was assessed by per-
forming 12 replicate injections of 3 hexacyanoferrate(II) stan-
dard solutions at concentrations of 1 10
, 1 10
, and
1 10
M. The relative standard deviation (RSD) of the ana-
lytical signals was #0.9%. Reproducibility of several working
electrodes constructed separately was also assessed by using a
1 10
M hexacyanoferrate(II) solution. RSD of the analyti-
cal signals was about 1.5% (n = 4).
FIA Manifold with Amperometric Multisite Detection
The amperometric sequential detection FIA system (Fig-
ure 3A) was evaluated by using the same solutions as those
used to study the intrinsic working characteristics of the tubu-
lar electrodes. The sample plug was injected into a short coil
(30 cm) and flowed through site x, where the analytical signal
was attained. The commutator was switched to the loop fill-up
position when the sample plug flowed inside coil R
, while the
Figure 3. FIA manifolds used: (A) evaluation of the amperometric multisite detection system; (B) sequential
determination of SA and ASA in pharmaceutical products: C1 and C2 = confluence points; D = electrochemical cell; Q,
Q1, Q2, and Q3 = flow rates; R1, R2, R3, and R4 = reactors; S = sample; V = sample volume injected; W = waste; X and Y
= monitoring sites.
detection system was relocated in site y. In this way the sam-
ple plug was again monitored, providing a second analytical
signal. After the analytical signal reached the maximumvalue
(site y), the detector could be displaced to its original position,
which corresponded to the injection of a new sample plug.
Evaluation of the working conditions of the multisite de-
tection manifold required assessment of several parameters,
such as sample insertion volume, flow rate, coil length be-
tween both detection sites, and commutation times, aiming for
effective separation of analytical signals attained at both de-
tection sites, minimization of sample plug dispersion over the
path between both detection sites, and accomplishment of the
highest sampling rates.
The commutation time, which determines the switching
time of the injector-commutator and, hence, relocation of the
detector in sites x and y, is crucial because it conditions clear
separation of analytical signals; restoration of the baseline; mix-
ing or reaction time, which might occur between both measur-
ing sites; and sampling rate accomplished by the system. After
the analytical signal attained at the first detection site has re-
turned to the baseline, the commutator can be switched at will
before the sample reaches site y. The next commutation, corre-
sponding to the setting of the commutator in its original posi-
tion, could be made after the peak maximumwas reached at site
y, thus avoiding the need to complete passage of the whole
sample plug through the detector at that site, with most of the
tailing portion of the sample zone being directly discarded.
As reported earlier, increasing injection volumes gave ana-
lytical signals closer to steady-state values. However, higher
sample volumes produced longer detection times, which had a
negative effect on the sampling rate. Injection volumes of
about 250 L, as in evaluation of detector working character-
istics, significantly compromised the sampling rates. Thus, in-
jection volumes of 100 L were used for further trials, which
allowed a signal with total restoration of the baseline in about
30 s (site x).
The length of coil R
, commutation time, and flow rate de-
termine the time taken by the sample to flow from the first to
the second detection site. If R
is too short, the front edge of
the sample plug may flow through site y before the detector is
relocated in this site. On the other hand, if this coil is too long,
sample dispersion will be high while the sampling rate will be
low. Several lengths ranging from 180 to 500 cm were tested
for coil R
, as well as flow rates between 0.8 and 2.0 mL/min.
A200 cmcoil and 1.2 mL/min flowrate were selected; hence,
the cell remained in the first site for 30 s and 70 s in the second
site. Under these conditions, sampling rates were 45 sam-
ples/h (Figure 4).
Once these parameters were set, potassium
hexacyanoferrate(II) solutions with a concentration range of
1.0 10
to 1.0 10
Mwere intercalated, yielding analytical
signals with good precision at both sites of detection, and
RSDs always #1.3 and 1.8% (n = 12), respectively, for the
first and second detection positions.
Manifold Optimization for Sequential Determination
of SA and ASA in Pharmaceutical Preparations
The usefulness of an FIA manifold with amperometric
multisite detection was evaluated by performing a sequential
determination of SAand ASAin pharmaceutical products (Fig-
ure 3B). SA undergoes an electron transfer-chemical reaction
pathway at the glassy carbon electrode; the final products are
probably a quinone-type structure, as reported for a structurally
related compound containing a phenolic moiety (13). In ASA,
the phenolic group is protected, making this compound
electroinactive. Therefore, to proceed with its quantitation a
preliminary hydrolysis is needed to convert it to SA.
A sample volume of 100 L (which was selected with re-
gard to the sensitivity required for the first detection) was in-
jected in a water stream (Q
= 0.45 mL/min). This was then
mixed with a phosphate buffer solution (Q
= 0.45 mL/min)
over the shortest possible coil (R
= 50 cm) to minimize the
probable hydrolysis of ASA before determination at site X
(where SA content was quantified). However, this coil must
be long enough to allow conditioning of the sample (that was
assessed by baseline stability). Because ASA is much more
stable in water than in phosphate buffer (19), the slightest hy-
drolysis that might occur over the transport step was thus mini-
mized. Afterwards, the sample plug was mixed with NaOH
proceeding from a confluence channel (Q
= 0.85 mL/min).
The NaOH was added to promote hydrolysis of ASA over coil
. While the sample plug was flowing inside R
, the commuta-
tor was switched to allow a new quantitation of the sample at
site Y. The difference between both analytical signals enabled
determination of the ASAcontent of the corresponding sample.
SA measurements (before and after hydrolysis) at both
sites of the manifold were made under different conditions:
100 L SA (2.0 10
M) was inserted in a phosphate buffer
carrier (pH 7) and in a phosphate buffer carrier to which 0.2M
NaOH solution had been added (pH 13). For these 2 experi-
mental conditions a potential variation from 0.40 to 1.30 V
Figure 4. Recorder output relating to sequential
monitoring of the same sample plug (1.0 10
K4[Fe(CN)6] solution) using the system in Figure 3A.
Higher signals correspond to site X; dotted line
corresponds to a blank signal.
was used to determine the more adequate potential for mea-
surements at sites X and Y. The results led to the selection of
0.98 V, which allowed SA determination under both condi-
tions, even though SAconcentration tested at site Xwas about
100 times lower than its concentration at the second monitor-
ing site.
The 0.2M concentration selected for NaOH (after testing
concentrations within the range of 0.11.0M) was added at
confluence C
to increase the pH of the carrier solution and
promote hydrolysis over coil R
. This was lower than that
used before (5) in the determination of SA and ASA as a
whole. The setting of the hydroxide concentration to a moder-
ate level was related to relocation of the detector, as increasing
hydroxide concentrations corresponded to a greater difference
between pHat the first and second detection sites, thus requir-
ing much more time for restoration of the baseline when the
detector was relocated (site X).
After performing trials with hydroxide concentrations of
0.11.0M, contact of the sample with the hydroxide solution
was prolonged to compensate for the low hydroxide concen-
tration of the carrier in which hydrolysis would occur. Be-
cause of this drawback, the length of coil R
was optimized.
Trials with 300 cm coils or longer showed that the 1000 cm
coil allowed about 40%hydrolysis, owing to insertion of solu-
tions containing several SA concentrations, under the same
experimental conditions.
Because the detection system is moved to its original posi-
tion only after the peak maximumhas been attained at site Y(a
sample portion is dragged), return to the baseline must be
achieved before the new sample plug reaches the detector.
Therefore, the adequate length of coil R
was established by
studying lengths of 200350 cm. The 300 cmlength, which al-
lowed not only good restoration of the baseline but also the
best sampling rate, was selected.
Factors contributing to good working conditions of the
proposed manifold were the time required to restore the base-
line and the time required to convert SAinto ASA. Hence, the
time required between the new switching of the commutator
to the other position was assessed. A minimum of 300 s was
selected for the first position and 200 s for the second. The
prolonged positioning of the detector at the first monitoring
site was determined by the length of the preceding coil and the
flow rate.
The proposed system showed a good precision of the mea-
surements at both sites of detection with RSDs always #1.5
and 2.1% (n = 10), respectively. These assays were repeated
several times during the workday and on consecutive days, at
the end of which the systemwas reconditioned by passing wa-
ter along the complete manifold, with no significant alter-
ations found in the amplitude of analytical signals measured.
These results show the absence of possible adsorption of the
tested drugs or their reaction products. The proportionality be-
tween amperometric currents and SA concentrations was
proven from calibration plots obtained at the first and second
sites of detection. In the first monitoring site the plot, con-
structed through injection of SAsolutions with concentrations
ranging from 1.0 10
to 5.0 10
M presented a slope of
4254.3 A/M, an intercept of 5.86 10
A, and a correlation
coefficient of 0.995 (n = 6). For the second monitoring site, in-
jections of SA solutions with concentrations ranging from
4.0 10
to 2.0 10
M yielded a calibration plot with a
slope of 2569.4 A/M, an intercept of 3.78 10
A, and a
correlation coefficient of 0.995 (n = 6). Detection limits were
about 3.5 10
and 1.1 10
M, respectively, for the first and
second monitoring sites.
After optimization, the manifold was used to analyze 6 dif-
ferent pharmaceutical preparations available on the Portu-
guese market. Results were compared with those obtained
with the USP procedure (Table 1). Although the values ob-
tained for the SA were slightly higher than those permitted by
USP (5), agreement between both sets of results was good,
which attests to the inertness of the excipients and absence of
other interferents, as we observed in previous studies. The rel-
ative deviations between both methods were always <1.0%
for ASA, and about 5.0% for SA. The latter was slightly
higher because of ASA hydrolysis before SA was monitored
at the first detection site.
Results of the F-test showed a value of 0.96 for SA and
0.99 for ASA(tabulated value is 7.15, for a confidence level of
95%; 21).
Table 1. Results of SA and ASA determination in commercial pharmaceutical preparations by the proposed method
and reference procedure
Pharmaceutical tablets
USP reference procedure
Error, %
Melhoral (Sterling) 500 mg 2.00 0.04 505 1.06 1.96 0.04 502 1.00 2.04 0.60
AAS (Sterling) 500 mg 2.61 0.06 506 0.30 2.59 0.08 506 0.35 1.16 0.00
Toldex (Bial) 500 mg 3.00 0.02 530 0.30 2.97 0.03 525 0.30 1.01 0.95
Aspirina (Bayer) 500 mg 2.91 0.03 498 1.10 2.77 0.04 500 1.20 5.05 0.40
Melhoral (Sterling) 100 mg 0.67 0.01 102 0.98 0.65 0.01 102 0.98 3.08 0.00
Aspirina (Bayer) 100 mg 0.54 0.01 100 0.99 0.52 0.01 101 1.04 3.85 1.00
Mean and standard deviation of 3 determinations for the same sample. Results expressed as mg per tablet.
The agreement between the values obtained by both proce-
dures was also assessed by the Students paired t-test in which
the t-values calculated for SA (2.47) and for ASA (0.77) were
lower than the tabulated value (2.57), for a confidence level of
95% (n = 6; 21).
The tubular electrodes incorporated in the detector were
used to determine SA and ASA without prior chemical or
physical treatment other than washing the whole system with
water at the end of each workday. Under these conditions and
considering that >50 determinations were made every day, the
electrodes presented good working characteristics for
>2 months, and were replaced only when the analytical signal
decreased by 20%.
The tubular composite carbon electrodes showed good
working characteristics and an appropriate configuration for
sequential measurements because of their small inner volume
and internal diameter similar to that of the tubing. Thus it is
possible to ensure hydrodynamic characteristics of the sample
plug that do not produce further dispersion, which would im-
pair the second monitoring.
The FIA manifold with amperometric multisite detection
allowed sequential determination over the same sample plug
of SA and ASA in pharmaceutical preparations and yielded
results in good agreement with those obtained by the USP pro-
cedure for all preparations analyzed. Moreover, the use of
amperometric multisite detection can be extended to other ap-
plications, requiring only slight changes according to the
analyte and sample characteristics. It must be stressed that the
construction procedure for the developed electrodes can be
also recommended for electrodes made of other materials
such as glassy carbon, gold, or platinum.
R.I.L. Catarino thanks programPRAXIS XXI for the Ph.D
grant. Funds from projects PRAXIS/C/QUI/10035/1998 and
PB98-0367-C02-02 (DGES, Spain) are greatly appreciated.
(1) Zagatto, E.A.G., Bergamin, H.F., Brienza, S.M.B., Arruda,
M.A.Z., Nogueira, A.R.A., & Lima, J.L.F.C. (1992) Anal.
Chim. Acta 261, 5965
(2) Nogueira, A.R.A., Brienza, S.M.B., Zagatto, E.A.G., Lima,
J.L.F.C., & Arajo, A.N. (1993) Anal. Chim. Acta 276,
(3) Arajo, A.N., Catita, J.A.M., & Lima, J.L.F.C. (1998) Anal.
Sci. 14, 809813
(4) Neto, J.A.G., Nogueira, A.R.A., Bergamin, H.F., Zagatto,
E.A.G., Lima, J.L.F.C., & Montenegro, M.C.B.S.M. (1994)
Anal. Chim. Acta 285, 293299
(5) U.S. Pharmacopeia (2000) 24th Rev., U.S. Pharmacopeial
Convention, Rockville, MD, p. 165
(6) Fernndez, J.M.L., Luque de Castro, M.D., & Valcrcel, M.
(1990) J. Autom. Chem. 12, 263266
(7) Villari, A., Micali, N., Fresta, M., & Puglisi, G. (1992) J.
Pharm. Sci. 81, 895898
(8) Wang, C., Vickers, T.J., & Mann, C.K. (1997) J. Pharm.
Biomed. Anal. 16, 8794
(9) Kister, G., Ribes, M., Chanal, J., & Catterini, A. (1976) Ann.
Pharm. Fr. 34, 215222
(10) Galante, R.N., Egoville, J.C., Visalli, A.J., & Patel, D.M.
(1981) J. Pharm. Sci. 70, 167169
(11) Agbaba, D., Lazarevic, R., Zivanov-Stakic, D., &
Vladimirov, S. (1995) J. Planar Chromatogr. 8, 393395
(12) Gracia, L.G., & Luque de Castro, M.D. (1997) J. Liq.
Chromatogr. Rel. Technol. 20, 21232133
(13) Evans, D., Hart, J.P., & Rees, G. (1991) Analyst 116,
(14) Fung, Y.S., & Luk, S.F. (1989) Analyst 114, 943945
(15) Neumayr, M., Friedrich, O., Sontag, G., & Pittner, F. (1993)
Anal. Chim. Acta 273, 469475
(16) Milagres, B.G., Neto, G.O., Kubota, L.T., & Yamanaka, H.
(1997) Anal. Chim. Acta 347, 3541
(17) Krug, F.J., Bergamin, H.F., & Zagatto, E.A.G. (1986) Anal.
Chim. Acta 179, 103118
(18) Alonso, J., Bartroli, J., Lima, J.L.F.C., & Machado, A.A.S.C.
(1986) Anal. Chim. Acta 179, 503508
(19) Bakar, S.K., & Niazi, S. (1983) J. Pharm. Sci. 72, 10241026
(20) Karlberg, B., & Pacey, G.E. (1989) Flow Injection Analysis.
A Practical Guide, Elsevier, Amsterdam, The Netherlands, p.
(21) McCormick, D., & Raoch, A. (1987) Measurements, Statis-
tics and Computation, John Wiley & Sons, Chichester, UK,
pp 202204