Flow analysis�/hydride generation�/Fourier transform infraredspectrometric determination of antimony in pharmaceuticals

M. Gallignani *, C. Ayala, M.R. Brunetto, M. Burguera, J.L. Burguera

Venezuelan Andean Institute for Chemical Research (IVAIQUIM), Faculty of Science, University of Los Andes, P.O. Box 440, Merida

5101-A, Venezuela

Received 8 July 2002; received in revised form 27 November 2002; accepted 27 November 2002

Abstract

In this work, a flow analysis system with hydride generation and Fourier transform infrared (FTIR) spectrometric

detection has been developed for the determination of antimony in pharmaceuticals. The method is based on the on-line

mineralization/oxidation of the organic antimonials present in the sample and pre-reduction of Sb(V) to Sb(III) with

K2S2O8 and KI, respectively; prior to the stibine generation. The gaseous SbH3 is separated from the solution in a gas

phase separator, and transported by means of a nitrogen carrier into a short pathway (10 cm) IR gas cell, where the

corresponding FTIR spectrum is acquired by accumulating 3 scans in a continuous mode. The 1893 cm�1 band was

used for the quantification of the antimony. The procedure is carried out in a closed system, which reduces sample

handling and makes possible the complete automation of the antimony determination. The figures of merit of the

proposed method (linear range: 0�/600 mg l�1, limit of detection (3s )�/0.9 mg l�1, limit of quantification (10s )�/3

mg Sb l�1, precision (R.S.D.) less than 1% and sample frequency�/28 h�1), are appropriate for the designed

application. Furthermore, precise and accurate results were found for the analysis of different antimonial

pharmaceutical samples, indicating that the methodology developed represents a valid alternative for the determination

of antimony in pharmaceuticals, which could be suitable for the routine control analysis.

# 2003 Elsevier Science B.V. All rights reserved.

Keywords: Antimony determination; Stibine; SbH3; Glucantime; Flow analysis; Hydride generation; Fourier transform infrared

spectrometry

1. Introduction

Although antimony is a potentially significant

element for plants, it does not have known

essential function in animals; on the contrary its

toxicity has been demonstrated [1,2]. However,

systemic antimonial therapy is still recommended

for multiple lesions caused by human leishmaniasis

disease [3,4]. The toxicological and physiological

behavior of this element is dependent on its

oxidation state. Elemental antimony is more toxic

than its salts, and trivalent antimony salts are ten

times more toxic than pentavalent salts [5,6]. At

present, almost all forms of leishmaniasis are

* Corresponding author. Tel.: �/58-274-2401375; fax: �/58-

274-2401286.

E-mail address: maximo@ula.ve (M. Gallignani).

Talanta 59 (2003) 923�/934

www.elsevier.com/locate/talanta

0039-9140/03/$ - see front matter # 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0039-9140(02)00648-3

initially treated with tri- and penta-valent anti-monials in the form of sodium stibogluconate

(pentostan ), N -methyl glucamine antimoniate or

meglubine antimoniate (glucantime ). Other anti-

monial compounds such as sodium and potassium

antimonyl-tartrate (emetic tartar) has been used in

the treatment of schistosomiasis, protozoal infec-

tions and an as emetic [7,8].

Molecular absorption spectrophotometry [9,10],fluorimetry [11], atomic fluorescence spectrometry

[12], anodic stripping voltammetry [13] and gra-

phite furnace atomic absorption spectrometry

(AAS) [14] have been reported to be suitable for

antimony determination in different matrixes.

However, the most commonly employed analytical

approach involved hydride generation (HG) inter-

faced with some form of atomic spectroscopy(AS), popularly AAS [15]. HG methods have the

advantage of being simple, rapid and relatively

interferences free.

Fourier transform infrared (FTIR) spectrome-

try is a fast analytical technique that provides very

interesting quantitative information from solid,

liquid and gaseous samples [16]. However, in the

last years, developments in gas phase generation-FTIR spectrometry, based on the on-line genera-

tion of vapor phases from liquid and solid samples

have improved direct determination by FTIR due

to the high transparency of gases, the low back-

ground values achieved and the possibilities of-

fered by using multiple-pass cells to increase the

analytical sensitivity [17,18]. Additionally, and in

the same way that in HG, vapor-phase generationallows an efficient matrix removal, thus reducing

spectral interferences. Various flow injection (FI)-

vapor generation-FTIR methodologies have been

developed for the determination of alcohols (eth-

anol and methanol) in different matrixes [19�/22],

carbonates in liquid and solid samples [18,23�/25];

as well as benzene, toluene and methyl tert -butyl

ether in gasolines [26] and butyl acetate, tolueneand methyl ethyl ether in paint solvents [27].

Most recently Gallignani et al. developed the

flow analysis (FA) with HG and FTIR (FA�/HG�/

FTIR) coupling for the individual and simulta-

neous determination of antimony, arsenic and tin,

via the generation of stibine (SbH3), arsine (AsH3)

and stannane (SnH4) [28]. This hybridized system

enhanced the analytical potentialities of the FTIRinstrumentation, allowing the determination of Sb

by using the medium intense band at 1893 cm�1.

Quality control of pharmaceutical products

demands a continuous development and improve-

ment of new analytical chemical methods. Most

on-line continuous flow and FI manifolds coupled

to different AS detection techniques have been

used for sample pretreatment for the determina-tion of antimony in various types of samples [15].

However, to the best knowledge of the authors, no

work has been reported for the Sb determination

by HG�/FTIR spectrometry with on-line sample

treatment.

The main aim of this work was to evaluate the

possibilities of the FA�/HG�/FTIR spectrometry

for the analysis of antimony in real samples, and inthis way, to develop an alternative methodology

for the determination of antimony in pharmaceu-

ticals which could be suitable for the routine

control analysis.

2. Experimental

2.1. Instrumentation

Fig. 1 depicts the schematic diagram of the FA�/

HG�/FTIR manifold used in this work. It consists

of an assembly of three Ismatec (Glattbrugg,

Switzerland) IPC peristaltic pumps (P1�/P3) of

four channels, furnished with Tygon tubes, six

channels for the transport of samples and reagents

(C1�/C6), three PTFE reaction coils (R1�/R3), aVarian (Springvale, Australia) gas�/liquid separa-

tor (GPS), a gas IR cell, a FTIR detector, two

entry of N2 with independent flow rate Cole

Palmer (Illinois, USA) regulators and a home-

made gas trap (GT). The experimental design

includes a simple time-based injection system for

the sequential introduction of sample and carrier.

It was designed using two Ismatec IPC fourchannel programmable peristaltic pumps (P1 and

P2) synchronized in the time, which propelled the

sample and the carrier (water), respectively. De-

tails of the construction and functioning of the

time-based injector were previously described by

Gallignani et al. [28,29].

M. Gallignani et al. / Talanta 59 (2003) 923�/934924

A Perkin�/Elmer (Norwalk, CT), Spectrum 2000

series FTIR spectrometer equipped with a tem-

perature stabilized DTGS detector, a Perkin�/

Elmer MIR-IR source and a KBr beamsplitterwas employed to carry out the spectral measure-

ments, accumulating 3 scans at a nominal resolu-

tion of 2 cm�1; a Wilmad (New Jersey, USA) 10

cm path-way gas cell with a 60 ml internal volume,

equipped with 32�/2 mm circular SeZn windows.

The SPECTRUM 2000 software was used to control

the instrument, for data acquisition and also for

processing the results.

2.2. Reagents

All chemicals used were of analytical-reagent

grade and Milli-Q double de-ionized water of 18

MV cm�1 identified in the text as DI water, wasused throughout. Stock standard solution (2000

and 1000 mg l�1) of antimony(III) was prepared

by dissolving K(SbO)C4H4O6 �/0.5H2O in 10% (v/v)

HCl. This standard solution can be stored for at

least 6 month.

A 500 mg l�1 standard solution of antimony(V)

was prepared from a 2500 mg l�1 antimony(III)

solution: 50 ml of the antimony(III) solution weretreated with 50 ml of KMnO4 2.5% (w/v) potas-

sium permanganate and 50 ml of 2.5 M sulfuric

acid. This mixture was heated during 30 min at

80 8C and after cooling, the excess of KMnO4 and

the produced MnO2 were removed by adding a few

drops of 15% (v/v) hydrogen peroxide solution

until a clear solution was obtained. This solution

was heated again for 10 min, in order to eliminate

any remaining excess of H2O2 and then, after

cooling, diluted to 250 ml with DI water [30,31].

The Sb(V) solution was prepared just before use. A

1% (w/v) of sodium tetrahydroborate(III) solution

stabilized with 0.5% sodium hydroxide was pre-

pared daily by dissolving 2.5510 g of NaBH4 in

0.5% (w/v) NaOH and diluting quantitatively to

250 ml with the same solution. At 4 8C, this

solution remained stable for about a week. Potas-

sium persulfate (K2S2O8), the oxidizing reagent,

was freshly prepared daily by dissolving 7.5 g in

250 ml of 10% HCl to give a final concentration of

3% (w/v). A 10% (w/v) potassium iodide (KI) pre-

reducing agent was daily prepared by dissolving 25

g in 30% (v/v) hydrochloric acid, and diluting to

250 ml with the same solution.

Less concentrated solutions were prepared daily

just before use, by appropriate dilutions with the

desired medium, as indicated in the text. Solutions

of different HCl concentrations were used for the

optimization studies; a 10% (v/v) HCl solution was

used as a carrier. The nitrogen used in this study

was from AGA, which certify a purity of 99%. Its

flow was controlled with a needle valve with a flow

meter from Cole Palmer.

All glassware were soaked overnight in HNO3

1:1, and rinsed with NaBH4 solution in order to

remove any Sb by forming the hydride, and then

washed thoroughly with distilled and DI water just

before use [32].

Fig. 1. Schematic diagram of the experimental (FA�/HG�/FTIR) system. P1�/P3, peristaltic pumps; C1, carrier; C2, sample (standard);

C3, oxidizing agent; C4, pre-reducing agent; C5, hydrochloric acid; C6, reductant; R1�/R3, reaction coils of 30, 30 (0.8 mm i.d.) and 60

cm (1.5 mm i.d.), respectively; GPS, gas phase separator; GR, gas flow rate regulator; GT, gas trapping. For further details and

operating procedure, see Table 1 and text.

M. Gallignani et al. / Talanta 59 (2003) 923�/934 925

2.3. Samples

Two pharmaceuticals containing N -methyl glu-

camine antimoniate, ‘Glucantime ’, (from Aventis

Pharma SA, Brazil and Rhone-Poulenc Rorer,

Paris, France), in presentation of 5.0 ml ampoules,

with a nominal content of 0.405 g Sb ampoule�1,

were purchased in Venezuelan pharmacies. Ula-

mine-I , (glucamine antimoniate) an experimentalformulation (not commercialized ) for the leishma-

niasis treatment (from Laboratorio de Investiga-

ciones Parasitologicas ‘Jose Witremumolo

Torrealba’), Department of Biology, University

of los Andes, Merida, Venezuela) in presentation

of 10.0 ml ampoules, with a nominal content of

0.135 g Sb ampoule�1, was provided by the

producers. Samples of Stibogluconate of anti-mony, Pentostan , (from Glaxo Smith Klane,

England) in presentation of 6 ml ampoules with

an antimony content of 100 mg Sb ml�1, were also

furnished by the mentioned laboratory. Addition-

ally, 0.5 and 0.005% (w/v) solutions of sodium

antimonyl-tartrate (C8H4Na2O12Sb2) and potas-

sium antimonyl-tartrate (emetic tartar,

C8H4Sb2O12 �/3H2O) [7,8], were provided by a localchemist drugstore. For the 0.5% solutions, aliquots

of 5 ml; with a nominal Sb content of 10.47 and

9.12 mg of antimony, respectively, were analyzed.

The antimony content in pharmaceuticals was

evaluated by the proposed method and by AAS.

2.4. Procedure

The samples and reagents were fed through theirrespective lines at room temperature, as indicated

in Fig. 1 under the operating conditions indicated

in Table 1. P3 was on during the analysis to propel

the oxidizing reagent (C3), the pre-reducing agent

(C4), the hydrochloric acid solution (C5) and the

reducing agent (C6), while P1 and P2 were syn-

chronized in the time, in order to introduce the

carrier or samples (C1 and C2). Initially, in orderto establish the baseline of the continuous system

(step I-a), P2 was stopped while P1 and P3 were

switched on to propel the carrier (HCl 10% v/v)

and the reagent solutions, respectively, via R3. The

first 45 s were used to fill the system with the

carrier and reagents. Thus, after 37 s, the gaseous

phase background was obtained by accumulating3 scans (step I-b), and then, the reference spectrum

of the blank was registered (37 s), also by

accumulating 3 scans (step I-c). The background

and the reference spectrum of the blank obtained

were used during the analysis. The initialization of

the system took around 120 s, and at this point the

system is ready for the analysis.

The procedure runs through a cycle of twosequences, i.e. (1) sample introduction, and (2)

gaseous SbH3 generation (sequence 2a) and FTIR

spectrum acquisition (sequence 2b).

In sequence (1), P1 was stopped, while P2 was

turned on to introduce the sample in the system

(around 45 s). The sample solution mixes down-

stream with the oxidizing agent in R1, where the

organic pentavalent antimony is completely con-verted in free Sb(V). Then, the resulting solution

flows to R2 where mixes with the pre-reducing

agent propelled by means of P3, providing the

appropriate conditions for the quantitative pre-

reduction of Sb(V) to Sb(III). Thereafter, this

solution mixes downstream with the NaBH4 (C6)

in R3 where Sb(III) is converted into its hydride

form (sequence 2a). Once the mixture reaches thegas phase (GP) separator, the liquid phase flows to

waste, and the antimony hydride is carried to the

IR gas cell with a stream of nitrogen. The

introduction of two N2 flows of 25 and 50

ml min�1, respectively (Fig. 1) helps to improve

(i) the stripping of the SbH3 vapor and (ii) its

transport to the measurement cell. At this point

(sequence 2b), the SbH3 FTIR spectrum wasacquired in a continuous mode by accumulating

3 scans. The whole sequence 2 took about 37 s.

Finally, the toxic SbH3 is trapped in a GT,

containing a AgNO3 0.25 M solution, according

to the following reactions:

SbH3(g)�3AgNO30 Ag3Sb�3HNO3

2Ag3Sb�6AgNO30 12Ag�SbO3�6HNO3

�Sb

The previously described sequences (1 and 2)

were repeated when the same standard or sample

solution was under evaluation. However, prior to

the evaluation of a different standard or sample

M. Gallignani et al. / Talanta 59 (2003) 923�/934926

solution, the carrier was introduced by means of

P1 (step I-a), in order to clean the system, and

then, to avoid the memory effect of the previously

introduced solution.

The first derivative spectra were obtained from

the FTIR spectra registered, and then, the peak-to-

valley (1891�/1895 cm�1) measurement of the 1893

cm�1 band were evaluated. Data obtained for

samples were interpolated on the corresponding

calibration graph established for aqueous standard

solutions of Sb(III), introduced in the system

identical to samples.

For comparison analysis, the liquid effluents of

R2 (Fig. 1) adequately diluted with DI water, were

Table 1

Operating conditions FA�/HG�/FTIR

Instrumental parameters Value

FTIR Radiation source Perkin�/Elmer MID-IR source

Detector DGTS

Beamsplitter KBr

Spectral range 4500�/500 cm�1

Analytical band 1893 cm�1

Measurement criterion Peak-to-valley measurement (1895�/1891 cm�1) of the first order derivative

spectra at the 1893 cm�1 SbH3 band

Nominal resolution 2 cm�1

Number of scans accumulated for the

background definition

5

Number of scans accumulated for each

spectrum

3

Gas cell Short path gas cell*/Wilmad

Path-way: 10 cm

Internal volume: $/60 ml

Windows: ZnSe 32�/2 mm (circular)

FA-

HG

Carrier composition (C1) HCl 10% (v/v)

Carrier flow rate (C1) 2 ml min�1

Sample solutions (C2) Sample (glucantime ) dilution in HCl 10% (v/v): M1 (500 ml�/100 ml) and M2

(250 ml�/100 ml)

Antimony standard solutions (C2) 0�/600 mg Sb(III) l�1

Sample (standard) flow rate (C2) 2 ml min�1

K2S2O8 solution concentration (C3) 2% (w/v) in HCl 10% (v/v)

K2S2O8 flow rate (C3) 2 ml min�1

Reaction coil (R1) PTFE (30 cm, 0.8 mm i.d.)

Pre-reducing agent (KI) composition (C4) 5% (w/v) in HCl 30% (v/v)

KI flow rate (C4) 4 ml min�1

Reaction coil (R2) PTFE (30 cm, 0.8 mm i.d.)

HCl solution concentration (C5) 20% (v/v)

HCl flow rate (C5) 1 ml min�1

NaBH4 concentration (C6) 0.5% (w/v) in NaOH 0.1 M

NaBH4 flow rate (C6) 1 ml min�1

HG reaction coil (R3) PTFE (60 cm, 1.5 mm i.d.)

N2 flow rate (stripping ) 50 ml min�1

Gas carrier flow rate (N2 carrier) 25 ml min�1

Gas trapping solution AgNO3 0.25 M

M. Gallignani et al. / Talanta 59 (2003) 923�/934 927

analyzed by conventional flame AAS, at the 217.6nm characteristic absorption line of antimony

[15,30,31].

3. Results and discussion

3.1. Preliminary studies. Gaseous SbH3 FTIR

spectra obtained from standards and samples

The preliminary studies and the optimization of

the proposed method were carried out for the

determination of antimony in glucantime samples.

For this purpose, two glucantime sample dilutions

were prepared: 500 and 250 ml were diluted to a

final volume of 100 ml with HCl 10% (v/v). These

sample dilutions are identified in the text as M1

and M2.Fig. 2A presents the gaseous SbH3 FTIR

spectra, obtained in the proposed system, from:

(a) a standard solution of 500 mg Sb l�1, and (b) a

sample solution (M1). SbH3 shows a medium

intense band at 1893 cm�1, and two less intense

bands localized at 830 and 781 cm�1. The insert in

the figure shows an amplification of the 1893

cm�1 band. At this point, it is important to pointout, that the FTIR spectra of standards and

samples treated in the proposed system are iden-

tical, and provide a way for the determination of

antimony in this kind of pharmaceuticals.

Any of the mentioned bands could be employed

for the determination of antimony. However, in

the present study, the band at 1893 cm�1 was

selected for the quantitative determination of Sb,via the generation of the stibine, because this band:

(i) is strong, (ii) is well resolved for standards and

samples and (iii) appears in a really transparent

zone of the blank. This can be seen in Fig. 2B,

where the behavior of the analytical band obtained

from: (a) a standard solution (250 mg Sb l�1), (b)

a sample solution (M2), (c) a sample solution

fortified with 250 mg Sb l�1 (50 ml of M1�/50 mlof 500 mg Sb l�1), and (d) the blank of the system

are shown.

Different baseline correction can be established

to carry out the quantitative determination at the

1893 cm�1 band. However, in this study the peak-

to-valley (1895�/1891 cm�1) measurement of the

first order derivative spectra was selected as the

analytical signal, in order to automatically correct

the displacement of the spectra with respect to the

base line, and to avoid the associated problems in

the selection of the baseline correction [28]. This

Fig. 2. SbH3(g) FTIR spectra from standard and sample

solutions. (A) Spectra obtained from: (a) standard solution of

500 mg Sb l�1, and (b) sample solution (M1). Insert: amplifica-

tion of the 1893 cm�1 analytical band. (B) FTIR spectra (1893

cm�1 band) from: (a) standard solution (500 mg l�1), (b)

sample solution (M2) and (c) sample solution fortified with 500

mg Sb l�1, (d) blank of the system. (C) Measurement criteria

(a) 1893 cm�1 absorption band of SbH3(g), (b) First derivative

order of the same band. Standard solution [Sb]�/100 mg l�1.

Other experimental conditions as specified in Table 1.

M. Gallignani et al. / Talanta 59 (2003) 923�/934928

behavior can be observed in Fig. 2C, where the1893 cm�1 band, and the corresponding first

order derivate obtained from a 100 mg Sb l�1

solution are shown.

The analytical sensitivity of the system (the

slope of the calibration graph) obtained by using

the derivative measurements is approximately one

half of those obtained by using the absorbance at

1893 cm�1 establishing the baseline correctionbetween 1750 and 2050 cm�1. However, the

related loss in the sensitivity, does not cause

problems, due to the high antimony content of

samples.

3.2. Optimization of the operating conditions

In the proposed procedure, three antimony

species (Sb(V), Sb(III) and organic-Sb (Org-Sb))were involved. For this reason, the optimization of

the parameters related to the chemical reactions

were monitored in all cases from 250 mg l�1

standard solutions of Sb(III) and Sb(V), and

from a sample solution (M2).

It is well established that the stibine is preferably

generated from Sb(III) [15]. However, the anti-

mony in glucantime is present as Org-Sb(V) [8]. Inpreliminary experiments, in absence of K2S2O8

and KI, a very little band was obtained, probably

due to the SbH3 generated from a free pentavalent

antimonial fraction present in the sample solution.

When the oxidizing reagent was added, the 1893

cm�1 band increased, possibly owed to an increase

in the Sb(V) concentration, product of the miner-

alization of the sample solution, followed by thegeneration of SbH3 from Sb(V). Lastly, in pre-

sence of the oxidizing and the pre-reducing agents,

the analytical band strongly increased, due the

quantitative conversion of Org-Sb(V) to Sb(III),

prior to the SbH3 generation reaction.

The FA�/HG�/FTIR method for determination

of antimony proposed in this work, was then

carried out in six steps: (i) oxidation/mineraliza-tion of the sample solution, (ii) pre-reduction of

the Sb(V) to Sb(III), (iii) generation of the gaseous

SbH3, (iv) transport of the gaseous phase to the

measurement cell, (v) acquisition of the FTIR

spectrum, and (vi) treatment of the spectral data.

Instrumental spectroscopic parameters, reagent

concentration and carrier gas flow rate wereoptimized using the univariate method, while other

experimental conditions such as reagent flow rates,

the transfer line length and the reaction coils

length, were previously fixed based on previous

experiences in our laboratory [28,29,33]. However,

before each optimization, the validity of the

previously selected parameters were checked. In

our system, under the reagent flow rates indicatedin Table 1, and by using the Varian GPS, it was

not possible the variation of the N2 stripping flow

rate (Fig. 1), which was fixed at 50 ml min�1.

The sample solutions were prepared in HCl 10%

(v/v), and the same medium was used as a carrier

in order to maintain similar acidic media in the

initialization of the system and during the analysis

(Section 2.4).

3.3. Effect of the instrumental parameters

The effect of the instrumental parameters such

as nominal resolution, number of scans and back-

ground conditions on the quality of the analytical

signal were evaluated using a solution of 250 mg

Sb l�1, unless otherwise stated.

Initially, the influence of the number of scansemployed to establish the background and to

obtain each spectrum was tested from 1 to 25

scans. Referring to the background definition, 5

scans are sufficient to establish a stable and clean

background. On the other hand, the accumulation

of scans in the proposed system does not have a

significant effect on the S/N ratio, probably due to

the very clean background obtained. These resultsare in good agreement with previously obtained

results in our laboratory in HG�/FTIR systems

[28]. Thus, 3 scans were selected for further

studies. Higher values does not improve the

quality of the analytical signal, and deteriorate

the sample throughput.

The nominal resolution (R ), significantly affects

the shape of the FTIR absorption bands. Table 2presents the effect of this parameter, varied from 2

to 32 cm�1, on the 1893 cm�1 absorption band.

An increase in this parameter causes the depres-

sion and the broadening of the analytical band.

This table includes the time required for the

acquisition of 3 scans, which greatly decreases

M. Gallignani et al. / Talanta 59 (2003) 923�/934 929

with increasing R (from 37 s for R�/2 cm�1 to 5 s

for R�/32 cm�1). In this work a nominal resolu-

tion of 2 cm�1 was used in further studies.

3.4. Optimization of the chemical reaction

parameters

The first chemical reaction, involved in the

proposed procedure, was the on-line mineraliza-

tion of the sample solution. In the literature a

bewildering variety of combinations of strong

acids (e.g. HNO3, HCl, H2SO4), oxidants (e.g.

KMnO4, K2Cr2O7, Br�/BrO/�3 ; K2S2O8)

[15,29,34], an UV radiation [35] have been used

for this purpose. However, in our case, thereaction with K2S2O8 was found to be suitable to

the on-line system presented in this work. Fig. 3A

shows the effect of the oxidizing agent on the

analytical signal generated from sample and stan-

dard solutions. For the Sb(III) and the Sb(V)

solutions (Fig. 3A, a and b), the analytical signal

was independent of the K2S2O8 concentration,

thus indicating that the presence of the oxidizingagent in the range from 0 to 3% (w/v) does not

affect the SbH3 generation in the experimental

design used in this work. The minimum K2S2O8

concentration required for the complete minerali-

zation of the sample solution (Fig. 3A, c) was

found to be 1% (w/v). Thus, in order to guarantee

a quantitative conversion of the organic antimonypresent in all the sample solutions, a 2% (w/v)

potassium persulfate solution was used as the

oxidant reagent. The analytical signal found in

the absence of K2S2O8, is probably due to the

presence of a pentavalent free antimony fraction in

the sample solution.

The second chemical reaction required in the

proposed system was the pre-reduction of theSb(V) produced in the first step, to Sb(III). For

this purpose, different reducing agents such as: L-

cysteine, ascorbic acid, tiourea, KI and mixtures of

them has been used [15]. Based on previous

experiences in SbH3 generation [30,31], the reac-

tion with KI was found to be adequate for the on-

line pre-reduction proposed in this work. The

influence of the potassium iodide concentrationon the SbH3 generation from sample and stan-

dards solutions was studied over the range 0�/10%

(w/v). The presence of the pre-reducing agent at

high concentrations as 10% (w/v) does not affect

the estibine generation from the Sb(III) solution,

as can be seen in Fig. 3B, a.

The behavior of the pentavalent antimony

standard and the sample solutions was similar.As can be seen in Fig. 3B, b and c, the analytical

response increased with the KI concentration up

to 2.0 and 2.5% (w/v), respectively. Thereafter, a

plateau were reached, indicating the completeness

of the pre-reduction of Sb(V) to Sb(III). The

difference between both curves is probably due

to the presence of a residual concentration of

KMnO4 or H2O2 in the pentavalent antimonystock (Section 2.2). Newly, in the absence of the

pre-reducing agent a little signal was obtained

from the Sb(V) and the sample solutions. This

behavior is probably due to the generation of

SbH3 directly from the pentavalent antimony [15].

A concentration of 5% (w/v) was chosen for

further determinations.

It is worthwhile to point out that the pre-reduction of the Sb(V) to Sb(III) required an

acid medium. However, the generation of the

stibine also required this medium.

Fig. 3C presents the effect of the HCl concen-

tration in the KI solution on the pre-reduction

process. For the Sb(III) solution (Fig. 3C, a), a

very little effect was observed, achieving a constant

Table 2

Effect of the nominal resolution

Nominal resolution

(cm�1)

x9/s a R.S.D.

(%)b

Time

(s)c

R�/2 0.10659/

0.0003

0.28 37

R�/4 0.04199/

0.0002

0.48 21

R�/8 0.01309/

0.0002

1.5 12

R�/16 0.00359/

0.0001

2.9 7

R�/32 0.00089/

0.0001

12.5 5

a Means9/S.D. of five independent measurements of an

antimony standard solution of 250 mg Sb l�1.b Relative standard deviation (n�/5).c Time required for the acquisition of 3 scans.

M. Gallignani et al. / Talanta 59 (2003) 923�/934930

response for HCl concentrations higher than 10%

(v/v). On the other hand, for the Sb(V) and the

sample (M2) solutions (Fig. 3C, b and c), the

analytical signal increase with increasing the HCl

concentration up to 20% (v/v). At higher concen-

trations, a plateau was observed, thus indicating

the quantitative conversion of Sb(V) to Sb(III).

Based on these results, the pre-reducing agent was

prepared in HCl 30% (v/v). Concerning the HCl

concentration in the channel C5, a little effect was

observed, due to the acidic media present in all

reagents and samples solutions. Then, an HCl

concentration of 20% (v/v) in the C5 was chosen

for further measurements.

The last chemical reaction related in our system

is the SbH3 formation. To determine the optimum

NaBH4 concentration, different concentrations in

the range 0�/1% (w/v) were tested (Fig. 3D). For all

solutions, the analytical signal continuously in-

crease with the NaBH4 concentration up to 0.4%

(w/v). A wide maximum was observed between 0.4

and 0.7% (w/v). Then, the analytical signal de-

creased at concentrations higher than 0.8% (w/v),

probably, because of the generation of hydrogen

Fig. 3. Effect of the chemical variables. (A) Effect of the K2S2O8 concentration on the analytical signal obtained from standard and

sample solutions. (B) Effect of the KI concentration on the analytical signal obtained from standard and sample solutions. (C) Effect of

the HCl concentration on the analytical signal obtained from standard and sample solutions. (D) Effect of the NaBH4 concentration

on the analytical signal obtained from standard and sample solutions. ((a) 250 mg Sb(III) l�1 (m); (b) 250 mg Sb(V) l�1 (^); (c)

sample solution (M2) (j)). Other experimental conditions as indicated in Table 1.

M. Gallignani et al. / Talanta 59 (2003) 923�/934 931

and its transfer to the gas cell exceeds the rate offormation of the stibine considerably [32]. A

concentration of 0.5% (w/v) was then used for

further studies.

3.5. Effect of the N2 carrier flow rate

The carrier gas flow rate is a critical parameter

in vapor-phase FTIR spectrometric analysis,

which affects the volatility of the gaseous analytes

and controls the speed of vapor introduction into

the measurement cell [17,28]. Therefore, the effect

of this parameter from 25 to 200 ml min�1, on theanalytical signal of a 250 mg Sb l�1 solution, was

studied. The results obtained: 0.10809/0.0003,

0.07909/0.0002, 0.06119/0.0002, 0.02489/0.0004,

and 0.01369/0.0004 for R�/2, 4, 8, 16 and 32

cm�1, respectively, show that the increase of this

parameter decrease the analytical signal, due to a

hardly dilution effect. In the configuration used in

this work, the use of values lower than 25ml min�1 was not possible. Consequently, a N2

flow rate of 25 ml min�1 was chosen for all further

work. As had been mentioned before, the N2

stripping flow rate was fixed in 50 ml min�1.

3.6. Analytical figures of merit

Under the experimental conditions indicated in

Table 1, the analytical signals obtained in the

proposed system, increased linearly with increas-

ing the antimony concentration from 0 to 600

mg l�1. The equation which describes the simplecalibration line is jdA /dn j�/2�/10�59/0.000426

[Sb] with r�/0.99989, where jdA /dn j is the peak-

to-valley (1895�/1891 cm�1) measurement of the

first order derivative of the SbH3 absorption

spectrum at the 1893 cm�1 band, and [Sb] is

antimony concentration mentioned above.

The precision of the procedure was estimated by

measuring ten replicates of a sample dilution (M2,5:2000) and standard solutions of 250 and 25 mg

Sb(III) l�1 by the recommended procedure, under

the optimum conditions given in Table 1. The

corresponding R.S.D. were 0.3 (0.08509/0.0002),

0.28 (0.10659/0.0003) and 1.8% (0.01069/0.0002),

respectively.

The detection and quantification limits for Sb,defined as three and ten times the S.D. of the blank

(3s , 10s) were 0.9 and 3 mg Sb l�1, respectively;

while the sample frequency of the proposed

method was 28 h�1.

3.7. Interference effects

In order to study the possible matrix effect on

the antimony analytical signal, standard additiongraphs were prepared by adding various amounts

of Sb (from 0 to 250 mg l�1) to 20 and 50 ml of a

sample solution (M1), respectively, and diluting to

a final volume of 100 ml. The equations obtained

for the simple calibration and standard additions

were: jdA /dn j�/2�/10�59/0.000426 [Sb], jdA /

dn j�/0.04309/0.000425 [Sb] and jdA /dn j�/

0.08569/0.000430 [Sb], respectively. These curvesdid not show a significant difference in their slope

(P B/0.0005), which denoted by the link of any

physical or chemical interference matrix effects.

On the other hand, the antimony content obtained

from the standard addition calibration graphs

were 4009/3 and 3989/3 mg ampoule�1, respec-

tively; while the Sb content found from different

sample dilutions: M1, M1(50�/50) and M1(20�/

80), by means of the simple calibration graph were

3989/3, 4019/3 and 3999/4 mg ampoule�1, re-

spectively. Newly, this values did not show sig-

nificant differences and were nearly closed to the

value reported by the producer (405 mg) which

denoted the absence of any kind of interferences.

The accuracy of the proposed procedure was

initially evaluated by spiking two different sampledilutions of glucantime. For this purpose 20 and

50 ml of M1 solution of glucantime were fortified

with known amounts of antimony between 0 and

400 mg Sb l�1, and diluted to a final volume of

100 ml with HCl 10% (v/v). In all cases the

recoveries ranged between 98.2 and 102%, which

again demonstrates the general reliability of the

method. Furthermore, pharmaceutical sampleswere analyzed by the proposed method and by

AAS, and the obtained results show a good

agreement (Table 3). Similar results were obtained

in the standard addition and recovery experiences

carried out with the other samples under evalua-

tion.

M. Gallignani et al. / Talanta 59 (2003) 923�/934932

3.8. Analysis of real samples

Samples described in Section 2 were analyzed by

the proposed method, by using the simple calibra-

tion line. Samples of Glucantime , Pentostan and

Ulamine were diluted off-line 0.5:100, 0.25:100,

and 1:25, respectively, prior the analysis. On the

other hand, sodium and potassium antimonyl-

tartrate samples were previously diluted 1:5, whilethe 0.005% (p/v) solutions were directly analyzed.

The obtained results are summarized in Table 3,

and as can be seen the values found agree with

those reported by the producer, and with those

obtained by AAS, thus indicating that the pro-

posed FA�/HG�/FTIR determination of antimony

provides a fast means of control analysis of this

kind of pharmaceutical formulations.

4. Conclusions

In this work, a novel analytical method based on

the coupling of FA�/HG�/FTIR spectrometry was

developed for the determination of antimony. The

method has been applied to the determination ofthe analyte in pharmaceuticals. The on-line sys-

tem, which included the complete mineralization

and pre-reduction of the sample, prior to the SbH3

generation reaction, reduces sample handling and

makes possible the complete automation of the

antimony determination. On the other hand, the

use of 2% (w/v) K2S2O8 and 5% (w/v) KI solutions,

provide an appropriate medium for the completeoxidation/mineralization and sample pre-reduc-

tion, under soft experimental conditions.

The figures of merit of the proposed method

(limit of detection�/0.9 mg l�1, linear range: 5�/

600 mg l�1, and sample throughput (28

samples h�1) are adequate for the designed appli-

cation. Furthermore, precise and accurate results

were found in the analysis of antimony pharma-ceutical samples, indicating that the methodology

developed is appropriate to solve real analytical

problems. Moreover, it represents an alternative

methodology for the determination of antimony in

pharmaceuticals, which could be suitable for the

routine control analysis.

Table 3

Analysis of real samples

Samplea FA�/HG�/FTIRb Reported AASb

1a 0.4009/0.005

g ampoule�1

0.405

g ampoule�1

0.3989/

0.008

1b 0.3999/0.004

g ampoule�1

0.405

g ampoule�1

0.4029/

0.006

1c 0.3989/0.005

g ampoule�1

0.405

g ampoule�1

0.4039/

0.009

1d 0.3979/0.004

g ampoule�1

0.405

g ampoule�1

0.3959/

0.007

1e 0.3999/0.001

g ampoule�1

0.405

g ampoule�1

0.4009/

0.010

2a 0.4019/0.005

g ampoule�1

0.405

g ampoule�1

�/

2b 0.3969/0.003

g ampoule�1

0.405

g ampoule�1

�/

2c 0.3999/0.003

g ampoule�1

0.405

g ampoule�1

�/

2d 0.3989/0.002

g ampoule�1

0.405

g ampoule�1

�/

2e 0.3989/0.002

g ampoule�1

0.405

g ampoule�1

�/

3a 0.1389/0.001

g ampoule�1

0.135

g ampoule�1

0.1399/

0.005

3b 0.13509/0.0007

g ampoule�1

0.135

g ampoule�1

0.1339/

0.004

3c 0.13709/0.0008

g ampoule�1

0.135

g ampoule�1

0.1389/

0.005

4a 0.5959/0.003

g ampoule�1

0.600

g ampoule�1

0.6039/

0.009

4b 0.6019/0.004

g ampoule�1

0.600

g ampoule�1

0.6089/

0.012

4c 0.5969/0.005

g ampoule�1

0.600

g ampoule�1

0.5909/

0.009

5 10.59/0.01

mg aliquot�1

10.47

mg aliquot�1

�/

6 21.09/0.02

mg aliquot�1

20.94

mg aliquot�1

�/

7 9.109/0.03 mg ml�1 9.12 mg ml�1 �/

8 18.39/0.02 mg ml�1 18.24 mg ml�1 �/

a samples: 1a�/e, Glucantime (Aventis Pharma SA); 2a�/e,

Glucantime (Rhone-Poulenc Rorer); 3a�/c, Ulamine-I ; 4a�/c,

Pentostan ; 5, Sodium antimonyl tartrate-I: 0.5% (w/v) solution

(aliquots of 5 ml); 6, Sodium antimonyl tartrate-II: solution

0.005% (w/v); 7, Potassium antimonyl tartrate-I: 0.5% (w/v)

solution (aliquots of 5 ml); 8, Potassium antimonyl tartrate-II:

solution 0.005% (w/v). Results were obtained under the

experimental conditions specified in Table 1. More details of

sample preparation in Sections 2.4 and 3.8b Means9/S.D. of five measurements.

M. Gallignani et al. / Talanta 59 (2003) 923�/934 933

New studies are under way concerning thedetermination of antimony in other samples and

matrixes.

Acknowledgements

The authors acknowledge the financial support

from the Consejo de Desarrollo Cientıfico, Huma-nıstico y Tecnologico (CDCHT-ULA) of the

University of los Andes, the Consejo Nacional de

Investigaciones Cientıficas y Tecnologicas (CON-

ICIT, Project S1-2000000815) and the BID-CON-

ICIT (Project QF-46).

References

[1] E. Underwood, Trace Elements in Human and Animal

Health Nutrition, fourth ed., Academic Press, New York,

1977.

[2] J.R. Castillo, C. Martinez, P. Chamorro, J.M. Mits,

Mikrochim. Acta 3 (1986) 95.

[3] J.D. Berman, in: K.P. Chang, R.S. Bray (Eds.), Experi-

mental Chemotherapy of Leishmaniasis. A Critical Re-

view, Leishmaniasis, Elsevier, Amsterdam, 1985, p. 111.

[4] WHO, The Leishmaniasis. Report of the Word Health

Organization Expert Committee, WHO Tech. Rep. Ser.

701 (1984) 99.

[5] T.D. Luchey, V. Venugopal, Metal Toxicity in Mammals.

Physiologic and Chemical Basis for Metal Toxicity,

Plenum Press, New York, 1977.

[6] P.D. Marsden, P.D. Jones, in: C.P. Chang, R.S. Bray

(Eds.), Clinical Manifestations, Diagnosis and Treatment

of Leishmaniasis, Leishmaniasis, Elsevier, Amasterdan,

1985, p. 193.

[7] J.D. Berman, in: K.P. Chang, R.S. Bray (Eds.), Experi-

mental Chemotherapy of Leishmaniasis*/a Critical Re-

view, Leishmaniasis, Elsevier, Amsterdam, 1985.

[8] W. Peters, R. Killick-Kendrick, The Leishmaniasis in

Biology and Medicine. Clinical Aspects and Control,

Academic Press, London, 1987.

[9] S. Sato, S. Uchikawa, E. Iwamoto, Y. Yamamoto, Anal.

Lett. 16 (1983) 827.

[10] S. Sato, Talanta 32 (1985) 341.

[11] I.M. Kolthoff, P.J. Elving, E.B. Sandell, Treatise on

Analytical Chemistry, Part II, vol. 7, Wiley Interscience,

New York, 1961.

[12] J.D. Norris, T.S. West, Anal. Chim. Acta 71 (1974) 458.

[13] S. Constantini, R. Giordano, M. Rizzica, F. Benedetti,

Analyst 110 (1985) 1355.

[14] I. Iwamoto, Y. Inoike, Y. Yamamoto, Y. Yahashi,

Analyst 111 (1986) 295.

[15] J. Dedina, D.L. Tsalev, Chemical analysis, in: J.D. Wine-

rfordner (Ed.), Hydride Generation Atomic Absorption

Spectrometry, vol. 130, Wiley, Chichester, 1995.

[16] J.R. Ferraro, L.J. Basile (Eds.), Fourier Transform Infra-

red Spectroscopy. Application to Chemical Systems, Aca-

demic Press, New York, USA, 1978.

[17] E. Lopez-Anreus, S. Garrigues, M. de la Guardia, Anal.

Chim. Acta 308 (1995) 28.

[18] A. Perez-Ponce, S. Garrigues, J.M. Garrigues, M. de la

Guardia, Analyst 123 (1998) 1817.

[19] A. Perez-Ponce, S. Garrigues, M. de la Guardia, Analyst

121 (1996) 923.

[20] A. Perez-Ponce, S. Garrigues, M. de la Guardia, Anal.

Chim. Acta 336 (1996) 123.

[21] J.M. Garrigues, A. Perez-Ponce, S. Garrigues, M. de la

Guardia, Vibr. Spectrosc. 15 (1997) 219.

[22] A. Perez-Ponce, R.F. Rambla, J.M. Garrigues, S. Garri-

gues, M. de la Guardia, Analyst 123 (1998) 1253.

[23] A.R. Casella, R.C. de Campos, S. Garrigues, M. de la

Guardia, A. Rossi, Frezenius J. Anal. Chem. 367 (2000)

556.

[24] A. Perez-Ponce, S. Garrigues, M. de la Guardia, Anal.

Chim. Acta 358 (1998) 235.

[25] A. Perez-Ponce, S. Garrigues, M. de la Guardia, Vibr.

Spectrosc. 16 (1998) 61.

[26] E. Lopez-Anreus, S. Garrigues, M. de la Guardia, Anal.

Chim. Acta 333 (1996) 157.

[27] E. Lopez-Anreus, S. Garrigues, M. de la Guardia, Analyst

123 (1998) 1247.

[28] M. Gallignani, C. Ayala, M.R. Brunetto, J.L. Burguera,

M. Burguera, Analyst 27 (2002) 1705.

[29] M. Gallignani, H. Bahsas, M.R. Brunetto, J.L. Burguera,

M. Burguera, Y. Petit de Pena, Anal. Chim. Acta 369

(1998) 57.

[30] C. Rondon, J.L. Burguera, M. Burguera, M.R. Brunetto,

M. Gallignani, Y. Petit de Pena, Frezenius J. Anal. Chem.

353 (1995) 133.

[31] Y. Petit de Pena, M. Gallignani, M. Burguera, J.L.

Burguera, N. Anez, A. Lugo, J. Braz. Chem. Soc. 1

(1990) 72.

[32] M. Burguera, J.L. Burguera, C. Rivas, P. Carrero, M.R.

Brunetto, M. Gallignani, Anal. Chim. Acta 308 (1995) 339.

[33] M. Gallignani, M. Valero, M.R. Brunetto, J.L. Burguera,

M. Burguera, Y. Petit de Pena, Talanta 52 (2000) 1015.

[34] J.L. Burguera, M. Burguera, C. Rivas, Quımica Analıtica

16 (1997) 165.

[35] M. Goto, T. Shibakawa, T. Arita, D. Ishii, Anal. Chim.

Acta 140 (1982) 179.

M. Gallignani et al. / Talanta 59 (2003) 923�/934934