Analytica Chimica Acta 539 (2005) 257–261 Direct determination of paracetamol in powdered pharmaceutical samples by fluorescence spectroscopy Altair B. Moreira a , Hueder P.M. Oliveira a , Teresa D.Z. Atvars a , Iara L.T. Dias b , Graciliano O. Neto a , Elias A.G. Zagatto c , Lauro T. Kubota a,∗ a Instituto de Quı́mica, Universidade Estadual de Campinas, Unicamp P.O. Box 6154, CEP 13083-970, Campinas SP, Brazil b Universidade São Francisco, Bragança Paulista SP, Brazil c CENA, Universidade de São Paulo, Piracicaba SP, Brazil Received 8 November 2004; received in revised form 22 February 2005; accepted 7 March 2005 Available online 31 March 2005 Abstract The native fluorescence of paracetamol (PA) in the solid state is demonstrated, allowing the development of a rapid, simple and rugged method for direct analysis of pharmaceutical formulations. It is easily adaptable to any spectrofluorimeter, and no chemical treatment of the sample is needed. The fluorescence measurements (λex = 333 nm; λem = 382 nm) are performed directly on the powdered sample, the active substance being diluted in lactose, maize starch, poly(vinylpyrrolidone), talc and stearic acid. The influence of the ingredients of PA formulations is discussed. Fluorescence intensity is linearly dependent on PA concentration within the 100–400 mg g−1 range. The analytical frequency is 200 h−1 . Detection and quantification limits were estimated within the 13.0–16.7 and 43.1–55.7 mg g−1 ranges for samples with different ingredient proportions. The method was applied to pharmaceutical formulations and the relative standard deviation of results was <2.7% (n = 20) for all tested ingredient proportions. Results were compared with those obtained by a method recommended by the British Pharmacopoeia and no statistical difference between methods was found at the 95% confidence level. © 2005 Elsevier B.V. All rights reserved. Keywords: Solid-phase analysis; Fluorescence spectroscopy; Paracetamol; Acetaminophen 1. Introduction Paracetamol (acetaminophen or N-acetyl-4-aminophenol), herein referred to as PA (Fig. 1), is a popular antipyretic and analgesic agent. In several countries, it is one of the most used medicines as an alternative to aspirin (acetylsalicylic acid). For PA determination in drug formulations, different methods exploiting, e.g. titrimetric [1,2], electrochemical [3–5], spectrofluorimetric [6–8], chromatographic [9–11] and spectrophotometric [12–19] techniques among others have been proposed. Most of the reported procedures require a previous oxidation of PA [7,9,12,13,16,17,19], which may increase the reagent consumption, time required for ∗ Corresponding author. Tel.: +55 37883091; fax: +55 37882987. E-mail address: kubota@iqm.unicamp.br (L.T. Kubota). 0003-2670/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2005.03.012 analysis and possibility of sample contamination. A sample preparation step prior to the determination makes the above-mentioned procedures somewhat laborious and less attractive for muti-sample analyses. This is often a limiting factor in the chemical and pharmaceutical industries, where a large number of samples must be assayed for quality control purposes. Simpler procedures performed in a shorter time are preferred and analytical procedures that do not involve use of chemicals and/or sample pre-treatment steps are then a good alternative. In this context, near-infrared (NIR) [20,21] and FT-Raman [22] spectroscopy have been used for PA determination in the solid phase. However, quantitative determinations with these techniques are only feasible by applying chemometric algorithms, especially in relation to NIR, where the spectra do not have high resolution, and complex-calibrating models must be constructed. 258 A.B. Moreira et al. / Analytica Chimica Acta 539 (2005) 257–261 Fig. 1. Molecular formula of paracetamol. Spectrofluorimetry in the UV–vis region can be used to perform the measurements directly in the solid matrix [23,24], leading to favorable characteristics of simplicity, sensitivity, ruggedness, selectivity, rapidity, etc. Nondestructive analyses are carried out that follow the present tendency towards Clean Chemistry. Moreover, the possibility of using optical-fiber accessories for on-line and/or in situ analysis [25] becomes feasible. To date, this strategy has scarcely been exploited in relation to solid samples of pharmaceutical interest. Regarding PA determination, a literature survey reveals that it usually requires the use of derivative reactions [6–8] because PA is not intrinsically fluorescent in aqueous solutions. In this regard, preliminary experiments demonstrated that PA is fluorescent in the solid phase. The aim of this work was the development of a simple, rugged and reliable procedure for PA determination in pharmaceutical preparations exploiting its inherent fluorescence in the solid phase. The measurements are performed directly on the solid sample without any prior chemical treatment. 2. Experimental 2.1. Chemicals and samples PA was of analytical reagent grade (Synth) and lactose, maize starch, poly(vinylpyrrolidone), talc and stearic acid used for the tablet preparation were of pharmaceutical grade (Galena). These ingredients were chosen because they are usually employed in pharmaceutical formulations. The proportions used were those indicated in a pharmaceutical handbook [26] and used to prepare dilutions of the active ingredient in accordance with the desired concentrations. The samples were prepared according to the recommendations of the British Pharmacopoeia [27]. The sample dilutions to reach the required concentrations were carried out after powdering the samples until a homogeneous particle size was attained; the powder was further mixed with the solid ingredients. The average particle sizes were smaller than 100 m and in this size range no significant changes in the fluorescence signal were observed. 2.2. Apparatus Steady-state fluorescence measurements were carried out with a model LS-55 Perkin-Elmer spectrofluorimeter, fur- nished with a xenon discharge lamp (20 kW, 8 s), two MonkGillieson monochromators, a Hamamatsu photomultiplier, an optical-fiber accessory and an ELISA plate, where the powdered samples were placed. Slits for the excitation and emission beams were set at 13 and 15 nm, the photomultiplier voltage was adjusted to 800 mV and the monochromator scan rates were 700 nm min−1 . For data acquisition and treatment, a PC microcomputer running FT Winlab software was used. Fluorescence decays were obtained with a pulsed FL 900 spectrofluorometer (Edinburgh Analytical Instruments, Edinburgh, UK) using the time correlated single photon counting (TCSPC) method. Excitation of the samples was carried out by a pulsed hydrogen flash-lamp controlled by a Thyratron tube operating with a repetition frequency rate of 40 kHz. Small disks prepared by compressing solid PA (41 MPa) were placed inside 1-cm quartz cuvettes. Measurements were performed at room temperature (λex = 333 nm; λem = 382 nm). The observed decays of fluorescence intensity R(t) are given by the convolution integral [28]: t R(t) = F (t − t )L(t )dt (1) 0 where L(t ) = time distribution of the lamp pulse; F(t ) = (theoretical) response function corresponding to an infinitely short excitation lamp pulse; t = sample decay time and t = standard decay time. L(t ) was measured with a solution of LUDOX (DuPont) used to determine instrument response function closely spaced in time to specific fluorescence decay experiments. Data were analyzed by the exponential series method (ESM) that assumes that the fluorescence decay can be analyzed as a multiple exponential function where the best fit uses the “Marquardt algorithm” to minimize the χ2 (summation of the square of the differences between experimental data and modeling function multiplied by the weighting factor) [28]. A good fit is obtained when χ2 approaches 1.0 and a random residual distribution is observed. 2.3. Procedure Synthetic samples containing lactose, maize starch, talc, poly(vinylpyrrolidone) and stearic acid in the 70:14: 9:4:3 w/w (#1), 75:12:8:3:2 w/w (#2) or 80:10:7:2:1 w/w (#3) proportions were prepared in order to match the paracetamol tablets that generally present these proportions [26]. The ingredients were mixed with PA and powdered in an agata mortar for 0.5 min. The tablets underwent dilutions in order to get PA concentrations within the 100–400 mg g−1 range. Twenty-five milligram samples were placed into the 96-well plate, where the fluorescence measurements were carried out directly on the powdered samples. Spectra were obtained at 25 ± 1 ◦ C. For wavelength selection, excitation A.B. Moreira et al. / Analytica Chimica Acta 539 (2005) 257–261 259 and emission radiation were scanned from 305 to 355 nm and from 360 to 475 nm, respectively. 3. Results and discussion 3.1. Solid-phase PA fluorescence For the fluorimetric determination of PA in aqueous medium, the analyte is usually oxidized under slightly alkaline conditions by hexacyanoferrate(III), yielding a fluorescent substance [29]. Fluorescence emission of native PA in aqueous medium seems to be not yet observed and, in agreement with the literature, the authors did not observe emission in aqueous phase. To the best of our knowledge, PA native fluorescent in the solid phase has not yet been reported. Surprisingly, some emission was observed in the solid phase, probably due to the rigidity of the PA molecule in the solid phase. In fact, it is well known that increasing structural rigidity can improve the fluorescent quantum efficiency [30]. In view of this pioneering aspect, it is important to demonstrate experimentally the PA fluorescence. PA fluorescence in the solid phase was investigated under different excitation wavelengths, and excitation with lower absorption radiation produces emission spectrum with lower intensity (Fig. 2, inset). In addition, no spectral shifts in the emission spectra were observed by varying the excitation wavelength, which is often observed for light scattering phenomena [31]. In order to get some additional insights into this emission, the emission decay (Fig. 3) of PA in the solid state was studied using the single photon counting time resolved (SPCTR) fluorescence technique. A mono-exponential decay was observed, with lifetime τ F = 1.77 ± 0.01 ns, χ2 = 1.093. Both the steady-state emission and the dynamic spectra indicate that solid PA does exhibit fluorescence emission. This emission Fig. 2. Fluorescence spectra (2) of pure PA related to excitation wavelength (1) between 325 and 349 nm with increments of 1 nm. Inset: dependence of the emission intensity with the excitation wavelength. Fig. 3. Fluorescence decays for PA collected at λem = 382 nm and λex = 333 nm; spectrum 1 is the instrument response function of the LUDOX solution and spectrum 2 is the sample decay, both obtained at 298 K in the solid state. signal can thus be exploited for quantitative PA determination. 3.2. Analytical studies The method developed in the present work involved fluorescence measurements directly on the solid sample. Wavelengths for maximum absorption and emission were selected as 333 and 382 nm, respectively (Fig. 4). 3.2.1. Distance between optical fiber and sample The angle of incident light for total internal reflectance, as well as the acceptable angle, is an important aspect to be considered when using optical fibers. As the capacity of an optical fiber to catch light is linked to this angle, it is necessary to optimize the distance between the optical fiber Fig. 4. Excitation (1) and emission (2) spectra of PA and their corresponding blank spectra (3 and 4) in the solid phase. Ingredient proportion = 70:14:9:4:3 w/w; scan rate = 700 nm min−1 ; PA concentration = 250 mg g−1 . 260 A.B. Moreira et al. / Analytica Chimica Acta 539 (2005) 257–261 Table 1 Effect of some ingredients Ingredient C I Lactose Maize starch Talc Poly(vinylpyrrolidone) Stearic acid 133 147 189 186 196 166 62 149 35 353 I/C ± ± ± ± ± 4 2 3 1 8 1.3 0.4 0.8 0.2 1.8 I = fluorescence intensity with estimated standard deviation for n = 3, in arbitrary units; C = paracetamol concentration, in mg g−1 . in the formulations have fluorescence and any of them was observed. Fig. 5. Effects of the sample amount (A, dashed line) and the distance between optical fiber and sample (B, solid line). Figure refers to 250 mg g−1 PA. and the sample. In the present procedure, best sensitivity was attained with a distance of 0.7–0.9 cm (Fig. 5). 3.2.2. Sample amount The sample amount added to each well of the 96-well plate was also investigated. Depending on this amount, a variation in the fluorescence intensity might occur due to the eventual radiation scattering provoked by the radiation crossing the sample. Moreover, the number of molecules might not be enough for a proper interaction with the incident radiation. Thus, the diameter of the well is an important parameter to be controlled as well as the diameter of the radiation light beam. Increasing the sample amount increased the analytical signal, and maximum analytical signal was attained for an amount of 17-mg sample. Increasing the sample amount beyond this value did not improve the fluorescence intensity. 3.2.3. Presence of ingredients Influence of ingredients such as lactose, maize starch, talc, poly(vinylpyrrolidone), and stearic acid on the PA fluorescence intensity was evaluated individually. For this task, different mixtures of each ingredient with PA were prepared. It was found that lower signal/concentration (I/C) relationship was observed for the PA plus poly(vinylpyrrolidone) mixtures (Table 1). These data reveal that the proposed method is dependent on the quantity and kind of the ingredients used as constituents of the medicine. It is important to have previous knowledge of the sample bulk composition before applying the method, in order to prepare standards matching the sample matrices. Hopefully, the ingredients used in the production process are always known; therefore, their effect can be easily circumvented. It is also very important to know if any of the ingredients used 3.2.4. Analytical curve For a 25-mg sample and the proportion #1 (70:14: 9:4:3 w/w), a linear response was noted within 100 and 400 mg g−1 PA. The analytical equation was: y = 31(±7) + 0.97(±0.03)x (r = 0. 9990, n = 7) where x = PA concentration, expressed as mg g−1 ; y = fluorescence intensity, in arbitrary units. For the other investigated proportions (#2 and #3, 75:12:8:3:2 and 80:10:7:2:1 w/w, respectively) the analytical equations were: #2 : y = 46(±9) + 0.90(±0.03)x #3 : y = 41(±10) + 0.97(±0.04)x (r = 0.9981, n = 7) (r = 0.9985, n = 7) The different equations related to the different ingredient proportions suggest that it is necessary to calibrate the method for each production process, using the proportion actually involved. This is a limitation of the method that the previous knowledge of the ingredients and proportions is needed. Even though the PA contents in these samples were higher than the optimized linear range, the proposed method can be applied without diluting the sample. In fact, modifying the distance between optical fiber and sample, reducing the intensity of the radiation source, and/or the excitation slit width can be performed in order to match the dynamic instrument range with the assayed samples. Moreover, smaller amounts of sample can be used. This potentiality of the proposed method minimizes sample handling, allowing the direct determination in the production line. However, for much more concentrated samples, self-absorption eventually decreasing the quantic yield cannot be disregarded. 3.2.5. Figures of merit Repeatability of the analytical results was evaluated by performing 10 determinations on different samples (ca. 250 mg g−1 PA). The R.S.D. was lower than 2.7% for all ingredient proportions, highlighting the good repeatability inherent in the proposed method. Detection and quantification limits were estimated in accordance with the 3σ and the 10σ criteria [32], respectively, where σ is the standard deviation of three measurements of A.B. Moreira et al. / Analytica Chimica Acta 539 (2005) 257–261 Table 2 Paracetamol contents in tablets (mg per tablet) as determined by the proposed and reference [27] (British Pharmacopoeia) methods Sample Nominal Proposed method 1 2 3 4 5 6 750a 750a 750b 750b 750c 750c 763 759 761 745 768 742 ± ± ± ± ± ± 14 13 10 11 10 8 British Pharmacopoeia 759 746 753 756 755 754 ± ± ± ± ± ± 6 7 6 4 5 7 Experimental data (with uncertainties) based on three replicates. a Proportion 1. b Proportion 2. c Proportion 3. the blank. The values for the three ingredient proportions studied were 13.0 and 43.6 mg g−1 (proportion #1), 14.2 and 47.4 mg g−1 (proportion #2), and 16.7 and 55.7 mg g−1 (proportion #3). Even though the detection and quantification limits obtained for this method are somewhat high, a great advance is reached as it is not necessary to use chemometric methods, normally employed in NIR and Ramam techniques for PA determination [20–22], and in which there is the need to build a model for each sample. The fluorescence spectrum was recorded in 18 s, leading to a possible analytical frequency of 200 samples per hour, considering an automated system or that the samples were ready to be assayed. 3.2.6. Applications The applicability of this new method was demonstrated by applying it to six samples containing lactose, maize starch, talc, poly(vinylpyrrolidone) and stearic acid as ingredients in three different proportions, and the results are shown in Table 2 together with results obtained with the UV spectrophotometric method recommended by the British Pharmacopoeia [27]. After applying the Student’s t-test, no difference between these methods was found at 95% of confidence level. 4. Conclusions The results presented in this work demonstrate the potentiality of the proposed method for PA determination in solid matrices. Precise results, in agreement with those obtained by the British Pharmacopoeia reference method, were obtained. The procedure is simple, rapid and non-destructive. No chemicals are required; therefore, hazardous residues are not generated, thus contributing to Clean Chemistry. The method, however, is dependent on the ingredients and powdering, being necessary a previous knowledge of the ingredients of the sample, and a good powdering system. Circumvented these points, a simple reading of the fluorescence intensity in the solid sample makes an easy and fast determination possible. 261 Acknowledgments Partial support from the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and the Conselho Nacional de Desenvolvimento Cientı́fico e Tecnológico (CNPq) is greatly appreciated. References [1] K.G. Kumar, R. Letha, J. Pharm. Biomed. Anal. 15 (1997) 1725. [2] G. Burgot, F. Auffret, J.L. Burgot, Anal. Chim. Acta 343 (1997) 125. [3] J.M. Zen, Y.S. Ting, Anal. Chim. Acta 342 (1997) 175. [4] C.Y. Wang, X.Y. Hu, Z.Z. Leng, G.J. Yang, G.D. Jin, Anal. Lett. 34 (2001) 2747. [5] M.S.M. Quintino, K. Araki, H.E. Toma, L. Angnes, Electroanalysis 14 (2002) 1629. [6] J.M. Calatayud, C.G. Benito, Anal. Chim. Acta 231 (1990) 259. [7] J.A.M. Pulgarin, L.F.G. Bermejo, Anal. Chim. Acta 333 (1996) 59. [8] J.A. Murillo, L.F. Garcia, Anal. Lett. 29 (1996) 423. [9] J. Meyer, U. Karst, Chromatographia 54 (2001) 163. [10] M.L. Qi, P. Wang, Y.X. Leng, J.L. Gu, R.N. Fu, Chromatographia 56 (2002) 295. [11] A. Marin, E. Garcia, A. Garcia, C. Barbas, J. Pharm. Biomed. Anal. 29 (2002) 701. [12] S.M. Sultan, I.Z. Alzamil, A.M.A. Alrahman, S.A. Altamrah, Y. Asha, Analyst 111 (1986) 919. [13] S.M. Sultan, Talanta 34 (1987) 605. [14] K.K. Verma, A. Jain, K.K. Stewart, Anal. Chim. Acta 261 (1992) 261. [15] Z. Bouhsain, S. Garrigues, A.M. Rubio, M. Guardia, Anal. Chim. Acta 330 (1996) 59. [16] M.L. Ramos, J.F. Tyson, D.J. Curran, Anal. Chim. Acta 364 (1998) 107. [17] A. Criado, S. Cardenas, M. Gallego, M. Valcarcel, Talanta 53 (2000) 417. [18] M.J.A. Canada, M.I.P. Reguera, A.R. Medina, M.L.F. Cordova, A.M. Diaz, J. Pharm. Biomed. Anal. 22 (2000) 59. [19] J.F.V. Staden, M. Tsanwani, Talanta 58 (2002) 1095. [20] A. Eustaquio, P. Graham, R.D. Jee, A.C. Moffatt, Analyst 123 (1998) 2303. [21] A. Eustaquio, M. Blanco, R.D. Jee, A.C. Moffat, Anal. Chim. Acta 383 (1999) 283. [22] R. Szostak, S. Mazurek, Analyst 127 (2002) 144. [23] A. Villari, N. Micali, M. Fresta, G. Puglisi, Analyst 119 (1994) 1561. [24] A.B. Moreira, I.L.T. Dias, G.O. Neto, E.A.G. Zagatto, L.T. Kubota, Anal. Chim. Acta 523 (2004) 49. [25] U. Utzinger, R.R. Richards-Kortum, J. Biomed. Opt. 8 (2003) 121. [26] A. Wade, P.J. Weller, Handbook of Pharmaceutical Excipients, second ed., The Pharmaceutical Press, London, 1994. [27] British Pharmacopoeia, vol. II, HMSO, London, 2002. [28] Y.S. Liu, W.R.J. Ware, J. Phys. Chem. 97 (1993) 5980. [29] J. Shibasaki, R. Konishi, K. Yamada, Chem. Pharm. Bull. 28 (1980) 669. [30] A.C. Franville, B. Dunn, J.I. Zink, J. Phys. Chem. B. 105 (2001) 10335. [31] J.B. Birks, The Theory and Practice of Scintillation Counting, Pergamon, Oxford, 1964. [32] A.C.S. Committee on Environmental Improvement, Sub-committee on Environmental Analytical Chemistry, Anal. Chem. 52 (1980) 2242.