Fast and Environmentally Friendly Quantitative Analysis of Active Agents in Anti-Diabetic Tablets by an Alternative Laser-Induced Breakdown Spectroscopy (LIBS)

Method and Comparison to a Validated Reversed-Phase High-Performance Liquid Chromatography (RP-HPLC) Method
Victor Ulises Contreras,a,* Marco A. Meneses-Nava,a Nancy Ornelas-Soto,a Oracio Barbosa pez-de-Alba,b Jose L. Maldonado,a Gabriel Ramos-Ortiz,a a,a Pedro L. Lo Garc b pez-Mart nezb Francisco J. Acevedo-Aguilar, Leticia Lo
a b

ptica, A.P. 1-948, 37150 Leo Centro de Investigaciones en O n, Gto., Me xico Departamento de Qu mica, Universidad de Guanajuato, Cerro de la Venada s/n, C.P. 36040Gto., Me xico

Laser-induced breakdown spectroscopy (LIBS) is evaluated as a potential analytic technique for rapid screening and quality control of anti-diabetic tablets. This paper proposes a simple LIBS-based method for the quantitative analysis of two active pharmaceutical ingredients (APIs): metformin (Met) and glybenclamide (Gly). In order to quantify both APIs, chlorine (Cl) concentration was estimated by employing the Cl/Br optical emission ratio, where Br was introduced as internal standard. Calibration curves were prepared, achieving linearity higher than 99%. On the other hand, for comparison to the proposed method, an isocratic reversed-phase high-performance liquid chromatography (RP-HPLC) method was also developed for quantitative determination of the same analytes by ultraviolet (UV) detection. The chromatographic separation was achieved on a Phenomenex Hypersil C18, 250 mm 6 4.6 mm, 5 lm column. The mobile phase was K2HPO4/H3PO4CH3OH and ow rate was 1.0 mL min1. The method is linear over a range of 1060 lg mL1 for Gly and 530 lg mL1 for Met and the correlation coefcients were !0.99. Recoveries were found to be in the range of 95101%. Furthermore, four different commercial brands of each active agent were evaluated by both proposed LIBS and chromatographic methods and results were compared with each other. The comparison was satisfactorily validated by analysis of variance (ANOVA). Index Headings: Pharmaceuticals; API; Metformin; Glybenclamide; Laser-induced breakdown spectroscopy; LIBS; Spectroscopy.

INTRODUCTION
Diabetes is a chronic disease that not only affects peoples health but also imposes a large economic burden on individuals, families, national healthcare systems, and economies worldwide. Type 2 diabetes is characterized by high glucose concentration in the blood, which is caused by both low pancreas secretion and reduced sensitivity to insulin. Currently, there is a wide variety of oral agents clinically available to help regulate glucose levels in blood such as sulfonylureas, meglitinides, biguanides, and thiazolidinediones.1,2 Sulfonylureas, such as glybenclamide (Gly), interact with sulfonylurea receptors on pancreatic b-cells to enhance insulin secretion and decrease blood-glucose levels. On the other hand, metformin (Met), the most common biguanide clinically available, decreases hepatic glucose production and
Received 14 May 2012; accepted 9 July 2012. * Author to whom correspondence should be sent. E-mail: ulises@cio.mx. DOI: 10.1366/12-06724

increases the sensitivity of peripheral tissues to insulin.3 At present, the blend of Gly/Met is the oral drug combination of choice in clinical practice.4 Nowadays more than 220 million people worldwide have diabetes and the World Health Organization (WHO) projects that diabetes deaths will double between 2005 and 2030.5 Besides the increase of diabetes, the production of both counterfeit and low-quality drugs is increasing and this fact also affects human health and results in loss of prot to pharmaceutical industries.6 Control of the active agent content in commercial products is important to guarantee the quality of pharmaceuticals worldwide. Laser-induced breakdown spectroscopy (LIBS) represents a powerful atomic analytical technique because of its intrinsic characteristics such as simple pretreatment or none at all, speed, and availability of multi-elemental analysis. LIBS requires no additional chemical compounds or solvents at all, and consequently, the analysis is environmentally friendly. At present, LIBS-based analysis of food, agricultural, geological, environmental, and biomedical solid samples for macro and micro elemental determination has emerged as an important analytical eld because of its viability.713 LIBS is a technique based on the analysis of the optical emission of a plasma plume containing individual neutral and ionic species from elements present in a sample. In general, a pulsed laser vaporizes and excites the surface of the sample in order to produce timedependent emission lines. Emission line intensities indicate the relative population of each species, allowing a qualitative and quantitative analysis of solid, liquid, and gaseous samples. In the pharmaceutical eld, some works have shown the effectiveness of applying the LIBS technique mostly for qualitative determination of some active ingredients in solid and liquid formulations.1416 Quantitative analysis of pharmaceutical products has been reported using noble gas environments to produce plasma that enhances the sensitivity to halogen elements present in active pharmaceutical ingredients (APIs).17 Lubricants, coating characterization, thickness, and uniformity of the pharmaceutical tablets have also been analyzed using LIBS.1820 On the other hand, chromatography is the most frequently used technique for pharmaceutical analysis, and for antidiabetic drugs, several high-performance liquid chromatography (HPLC) methods have been reported for Met and a bit less for Gly analysis but all of them are restricted for their

1294

Volume 66, Number 11, 2012

0003-7028/12/6611-1294$2.00/0 2012 Society for Applied Spectroscopy

APPLIED SPECTROSCOPY

determination in biological uids, which involve exhaustive and complicated solvent extraction, pre-concentration, or dilution processes.2130 The reported methods are strongly dependent on the sample preparation, which is a critical and the most time-consuming step, and thus, they are not economically feasible for routine analysis in pharmaceutical industries; in addition, the adverse environmental impact of these analytical methods can be reduced by minimizing the amount of solvents employed in sample pretreatment. However, the downscaling is not straightforward because this enhances the risk of errors such as incomplete analyte extraction, which manifests in poorer analytical performance. The aim of this study was to develop new rapid, accurate, precise, economic, and environmentally friendly methods based on spectroscopic and chromatographic methods for industrial analysis of anti-diabetic active pharmaceutical ingredients, i.e., Met and Gly present in tablet formulations, by the addition of internal standards, to improve the analytical performance and thus control undetermined changes in API concentration and instrument response uctuations, and also to reduce the problem of the many-fold dilution required in the classical batch procedures. All experiments were performed at ambient conditions. For both methods, calibration curves were obtained by plotting the ratio of analyte signal to internal standard signal as a function of analyte concentration, and for the quantitative analysis of eight different commercial antidiabetic brands the results were compared with each other. Furthermore, excipients such as calcium carbonate (CaCO3), lactose (C12H22O11H2O), talc (3MgO4SiO2H2O), and starch were studied by LIBS and Raman spectroscopy to characterize them and to select an adequate excipient for synthetic sample preparation.

EXPERIMENTAL
Reagents. All experiments were performed with analytical reagent grade chemicals. Metformin hydrochloride (C4H11N5HCl) and Glybenclamide (C23H28ClN3O5S) were obtained from Sigma Aldrich, USA. Excipients were purchased from Qu mica Meyer Mex. The internal standards, KBr and benzophenone, were obtained from PIKE Technologies, USA and Sigma-Aldrich, USA, respectively; K2HPO4 and H3PO4, were also obtained from PIKE Technologies. High purity ethanol (99.5%) was used in sample preparation. HPLC grade methanol (Merck, USA) and ultrapure water were used for preparing mobile-phase solutions. Procedures. LIBS Procedure and Apparatus. Synthetic samples were prepared by grinding analyte, excipient, and internal standard together in an agate mortar. Samples were prepared starting with quantity A (mg) of analytical standard and mixing with quantity B (mg) of excipient. The quantity A was selected according to the desired Cl concentration for each sample and the total sum of A and B was 400 mg for all synthetic samples. As a reference, 100 mg of internal standard (KBr) were added to improve quantitative measurements. So as to procure an optimal mixture, the grinding step in the agate mortar was performed in ethanolic suspension; this promotes smaller particle size and better homogenization. The mixing process was repeated twice until the sample was dried. After the mixing process, the sample was compacted under controlled pressure. The obtained samples were 8 mm diameter circular tablets. In addition, four commercial samples for each active agent

were prepared as follows: ten commercial tablets from the same box were ground and mixed in order to obtain an appropriate sampling process. Then, 400 mg of ground pharmaceutical product and 100 mg of internal standard were homogenized in ethanolic suspension, as described in the previous paragraph. Three replicates were made for all different samples (both synthetic and commercial). Pulses at 532 nm from an Nd : YAG laser (10 ns pulse duration at 10 Hz) were used for the experiment. Pulse energy was in the range of 10 to 20 mJ. The laser pulse was focused perpendicularly on the sample surface using a bi-convex lens (50-mm focal length). Samples were held in a custom-made aluminum holder that kept a constant lens-to-sample distance over all tablets. The sample holder was mounted on an XY stage allowing several shots at different places on the tablet surface. Light was collected and focused into a CzernyTurner spectrograph (Spectra Pro 500i Action Research Corp.). Then light was spectrally dispersed by a 1200 grooves/mm grating and detected by an intensied charge-coupled device (ICCD) camera (Princeton Instruments, Inc.). The 10243256 pixel array covered a spectral window of 35 nm with central wavelength at 830 nm. The laser pulse and detection system were synchronized in order to avoid bremsstrahlung and early plasma emission. Typical delay time td and gate time tg were on the microsecond scale. The possible non-uniformities of the laser energy and heterogeneities on the sample surface were minimized by averaging ten spectra acquisitions for 50 different places on each tablet surface. For a given sample the average of 500 spectra took less than one minute, giving a single spectrum for each analyzed tablet. Repeatability was observed by analyzing each sample three times. The background continuum emission was subtracted from the line integrated intensity. Spectral correlation was performed using homemade software on the Matlab platform via the atomic spectral database from NIST.31 Data processing and calculations were made in OriginPro software. Lorentzian t was performed before the intensity ratio analysis. HPLC Procedure, Conditions, and Apparatus. Stock internal standard solution (IS), containing 100 lg mL1 of benzophenone, was prepared by diluting the reagent in a mixture of chloroformmethanol (3 : 1). Benzophenone was chosen as internal standard due to the similar of its physicochemical properties to those of the analyte Gly. Stock standard solutions of Met and Gly were prepared separately by dissolving 10 mg of each drug in 10 mL distilled water and 10 mL stock IS solution, respectively. Appropriate Met and Gly working standard solutions were prepared individually by diluting the stock solutions with mobile phase and IS solution, respectively. Triplicate 20 lL injections were made for each working standard solution. The peak area for each Met concentration was recorded and then plotted against the corresponding concentration to obtain the calibration curve. In the case of Gly, the calibration curve was obtained by the ratio Gly peak area to IS peak area as a function of analyte concentration. Four different brands labeled to contain 5 mg of Gly and another four containing 500 mg of Met, all them with excipients, were used. For each brand, 10 tablets were weighed and nely powdered. In the case of Met, an accurate weight of the powder equivalent to one tablets content was transferred into a 100 mL volumetric ask, diluted with ultrapure water, and then sonicated for 10 min. Then, a 1 mL aliquot was taken

APPLIED SPECTROSCOPY

1295

TABLE I. Gly.

Statistical parameters for LIBS determination of Met and

Met Equation r2 Dynamic range (% Cl) Linearity (%) Slope standard error Intercept standard error LOD (% Cl) LOQ (% Cl) (Cl/Br)r = 0.036(%Cl) 0.034 0.9903 5.7311.43 99.5 12.8 3 103 1.46 3 103 3.50 3 102 11.2 3 102

Gly (Cl/Br)r = 0.042(%Cl) 0.010 0.9993 0.0990.723 99.59 2.97 3 104 3.27 3 102 1.10 3 102 3.65 3 102

from the previous solution and centrifuged to 10 000 rpm for 10 min. Afterward, 30 lL of supernatant was used for desired dilutions with the mobile phase and triplicate 20 lL injections were made for each sample. For Gly, an accurate weight of the powder equivalent to one tablets content was transferred into a 10 mL volumetric ask, diluted with IS solution, and then sonicated for 5 min. Later, a 1 mL aliquot was taken and centrifuged to 13 000 rpm for 5 min. Then, 300 lL of supernatant was dried by a stream of nitrogen and re-dissolved with 600 lL of methanol, and nally, this solution was sonicated for 5 min and centrifuged to 13 000 for 5 min and triplicate 20 lL injections from supernatant were made for each sample. The experiments were carried out using a PerkinElmer series 200 liquid chromatograph equipped with an autosampler, a pump, a diode-array detector, and an oven attached to the

FIG. 2. Emitted spectra from (a) metformin and (b) glybenclamide commercial samples mixed with an internal standard (KBr). The Cl and Br lines used for the analysis are indicated by arrows on both graphs. In the case of glybenclamide spectra, an enlarged zone illustrates the Cl emission at 837.24 nm.

PerkinElmer TotalChrom Workstation V 6.2.1 (PerkinElmer Instruments LLC, Shelton, Connecticut, USA) and connected to an ultraviolet-visible (UV-Vis) detector operating at 230 nm. The chromatographic separation was performed at 40 8C using a Phenomenex Hypersil C18, 250 mm 3 4.6 mm, 5 lm analytical column (Waters Corp., Milford, MA). The mobile phase was K2HPO4/H3PO4 10 mM pH 3.5-CH3OH (40 60, v/v) at a ow rate of 1.0 mL min1.

RESULTS AND DISCUSSION

LIBS Analysis. An optical ngerprint of biguanides and sulfonylureas can be obtained by measuring the spectral emission of certain elements such as Cl. To determine the Cl concentration, the peak area is plotted versus Cl concentration. For quantitative analysis the measured signal can be improved by using an internal standard as specic reference, e.g., Br. After the addition of the reference, peak areas from both elements (Cl and Br) will be affected equally by shot-to-shot pulse variation, minimizing error caused by possible energy variation during the experiment and any other variable instrumental response. In the following text ClBr ratio means the area ratio of analyte (Cl(I)) peak to internal standard (Br(I)) peak. The symbol (I) is used for emission of neutral species and (II) for rst ionized species. Synthetic samples may be divided in two sets depending on the target analyte used, that is: Met and Gly sets. For each set, an average calibration curve was obtained by using the ClBr ratio and the Cl concentration. The dynamic range of calibration curves was chosen to cover the Cl concentration from commercial samples. Statistical parameters from calibration curves are summarized in Table I. In this work, typical ranges of concentration from commercial samples lie between 6 to 11% and 0.1 to 0.2% for Met and Gly, respectively.

FIG. 1. Typical excipients used in anti-diabetic tablets show spectral differences between each other over the visibleIR region. As an example, two spectral windows are shown at (a) 732769 nm and (b) 833867 nm. Spectral differences can be used as particular excipient optical ngerprints.

1296

Volume 66, Number 11, 2012

FIG. 3. (a) Metformin-containing commercial samples (14) are well predicted according to calibration curve analysis. Seven synthetic samples at different concentrations of Cl were tted achieving linearity higher than 99%. (b) According to statistical analysis two pharmaceutical brands (2 and 3) present wider spread distributions. The outlier (in brand 1) is present in the analysis.

Qualitative Analysis. Frequently, matrix effects are the source of error in the quantitative determination of sample concentration in LIBS spectroscopy. Therefore, the use of the same excipient for the synthetic samples as for the commercial formulations is of special importance. Typical excipients used in the pharmaceutical industry such as calcium carbonate (CaCO3), lactose (C12H22O11H2O), talc (3MgO4SiO2H2O), and starch were characterized using LIBS and Raman spectroscopy. As can be seen in Fig. 3, LIBS spectra from two different wavelength ranges were evaluated from the above-mentioned excipients. Particularly, Mg(I) (Fig. 1a) and Ca(II) (Fig. 1b) emissions (talc and calcium carbonate, respectively) are present in those selected spectral windows. In the case of lactose and starch it is more difcult to identify its optical ngerprint in air due to their compositional elements (C, O, and H). Nevertheless, it is possible to identify excipients by Raman spectroscopy by using the same LIBS experimental setup, but adding a notch lter and changing some experimental parameters in order to avoid laser ablation (lens-to-sample distance, laser energy, td, and tg). In the case of Met, it is not possible to identify the excipient because all signicant Raman emission bands correspond to the Metformin molecule due to its high concentration contents in all commercial samples (around 90%). For the Gly set, Raman spectroscopy identies lactose as the excipient for all the commercial formulations. Therefore, lactose was used as excipient for both Met and Gly sets of synthetic samples for LIBS experiments.

Quantitative Analysis. Internal Standard. The fundamental principle of internal standard in chemical analysis is based on the addition of a component that has similar physicochemical properties to the analyte (depending on the pretreatment procedure). The internal standard should present chemical stability under experimental conditions applied in the procedure; there should be no interactions between the internal standard and the analyte, nor with other sample components. It should be soluble in the assay medium and should not be a natural component of the sample. In this work, potassium bromide (KBr) was chosen as internal standard because Br and Cl have similar spectral and chemical characteristics, e.g., both elements have high and comparable rst potential ionization values (13.01 and 11.84 eV for Cl and Br, respectively) remaining mostly neutral through laser energy uctuations throughout the experiment. Spectral databases show that the strongest single line emitted by Br(I) and Cl(I) (827.44 and 837.59 nm, respectively) are sufciently close to be detected simultaneously in the same spectral window. Furthermore, there were not interference signals from K in the spectral window of interest.31 In addition to the spectral characteristics, KBr is highly soluble in ethanol, so the homogenization in the sample preparation process was enhanced. Figure 2a shows the spectral window between 812 and 847 nm, where the emission of both a commercial sample of Met with a 10% Cl concentration and internal standard are present. The spectrum shows the emission of neutral and ionic species from Met such as N(I), N(II), and Cl(I) and the emission of Br(I) from the internal standard. It should be pointed out that the emission of N(I), N(II), and O(I) from the API could not be discriminated from emissions of the same species present in atmospheric air. Nevertheless, air signals do not interfere with the Cl and Br lines of interest. On the other hand, Fig. 2b shows a spectrum of commercial Gly with a Cl concentration of 0.19%, where internal standard has also been added. With the purpose to resolve the Cl(I) emission at 837.59 nm, higher energy was employed than in Met experiments. Thus, the emission of N(I), N(II), and Cl(I) was enhanced as shown in Fig. 2b. Notice that the O(I) signal is not spectrally resolved because of the broadening of the N(II) line at 845 nm due to the higher pulse energy used. Individual resolved signals from Br(I) and Cl(I) at 827.24 and 837.59 nm, respectively, were used for the ClBr ratio in the quantitative analysis. Therefore, the same spectral range but different energy and time parameters were used for the analysis of each active agent. Calibration Curves and Analytical Parameters. The Cl/Br emission ratio was measured under optimum physicochemical conditions. The calibration curves were obtained for different standard samples containing between 5.73 and 11.43% for Met and between 0.09 and 0.72% for Gly. For each standard sample, three replicates were made. Statistical parameters and analytical characteristics for the determination of both Met and Gly are summarized in Table I. The calibration curves were linear for all concentrations tested. A linearity !99.50% was obtained (Table I). It is worth noting that a high linearity was achieved despite the fact that halogen elements, such as Cl, are spectrally difcult to detect due to their high excitation energies (.10 eV) and neither saturation nor linear deviations were observed. With the proposed method, quantitative information for Cl can be easily obtained within the tested range. For this experiment, the optimal temporal acquisition parameters were

APPLIED SPECTROSCOPY

1297

TABLE II. Recoveries obtained with the proposed LIBS method in the analysis of the pharmaceutical formulations of Met and Gly. Met Samples 1 2 3 4 Labeled weight (mg/tablet) 500 500 500 500 Found weight (mg/tablet) 524.39 462.88 513.46 471.11 % R (RSD) 104.88 92.58 102.69 94.22 (6.29) (7.02) (9.24) (6.37) Labeled weight (mg/tablet) 5 5 5 5 Gly Found weight (mg/tablet) 4.47 5.33 5.39 4.79 % R (RSD) 89.78 106.53 107.77 95.91 (4.62) (9.77) (8.77) (6.97)

0.20 ls (td) and 1.00 ls (tg), producing the optimal signal-tonoise ratio (SNR). It must be taken into account that besides the high energies necessary for Cl excitation, the Gly samples have relatively low Cl concentration. It has been shown by other authors that the sensitivity in the case of halogen elements and SNR can both be enhanced by producing the induced breakdown in a noble gas atmosphere.14,32,33 St-Onge et al. reported an improved factor of 7.3 for signal detection when they analyzed the emitted Cl line at 837.59 nm using a helium ow over pharmaceutical samples. Nevertheless, in this work, the Cl concentration in pharmaceutical samples was sufciently high to be detected under atmospheric conditions (around ;0.035% using the same line at 837.59 nm).33,34 In order to quantify the lowest content of Cl in Gly samples (;0.1 %), experimental parameters were optimized (laser pulse energy between 16 and 20 mJ per pulse, 2.0 ls tg, and 0.25 ls td) to maximize the signal-to-background ratio. With the parameters mentioned above, the achieved linearity for Gly was 99.59%. Application to Pharmaceutical Formulations. The LIBS method has been applied to the determination of the active ingredients Met and Gly by using their Cl content. Figure 3a shows the Met calibration curve obtained from synthetic samples (see previous section) and four different commercial samples containing 500 mg of Met, labeled from 1 to 4. As can be seen in this gure, commercial samples are well tted on the calibration curve taking into account the API concentration labeled on the commercial boxes. According to the above results, matrix effects are negligible or not evident because Met represents around 90% of the total weight in commercial tablets. Vertical error bars represent the standard deviation of the three used replicates. One outlier was observed according to a box and whisker analysis as shown in Fig. 3b (small triangle). This outlier from sample 1 was not considered in the analysis of the Cl content. In the same gure it can be noticed that the Cl concentration shows a Gaussian distribution in brands 1, 2, and 3. Furthermore, brands 2 and 3 present considerable spread of data according to their box size. Nevertheless, the simple grinding process proposed in this method gives satisfactory predictions for Met in commercial samples, as is shown in Table II. For the case of samples containing Gly as the active agent, typical values of Cl concentration are about 0.18% of the total mass, which corresponds to 5 mg of API over 160 mg of the total tablet mass. Figure 4a shows four commercial samples labeled from 1 to 4 and the respective calibration curve. In the case of the Gly LIBS results, no matrix effect is expected because both synthetic and commercial samples contain lactose as excipient. The four Gly brands show normal distribution and no outliers are present, as can be seen in Fig. 4b. Boxes of brands 1 and 4 are smaller because those brands present smaller

standard deviations than brands 2 and 3 (Table II). Results based on LIBS and the simple preparation steps suggest the use of this method for a rapid on-line/in-situ monitoring of APIs due to their elemental composition in the pharmaceutical industry with an acceptable level of accuracy. HPLC Analysis. In order to separate Met and Gly analytes from matrix, mixtures of K2HPO4/H3PO4 pH 3.5 and methanol in different combinations at various ow rates were tested. The optimum wavelength for detection was 230 nm for both drugs. The mobile-phase K2HPO4/H3PO4 10 mM pH 3.5CH3OH (40 60, v/v) at a ow rate of 1.0 mL min1 proved to be the best as the chromatographic peaks were well dened, resolved, and free from tailing. Increasing the methanol concentration to more than 70% led to inadequate separation. At a lower methanol concentration (,45%), separation occurred, but with excessive tailing and increased retention time. As shown in Figs. 5 and 6, the retention times were 4 min for Met, 9.4 min

FIG. 4. (a) Glybenclamide-containing commercial samples (14) are well predicted according to their respective calibration curve analysis. For the sake of clarity, only a fraction of the curve containing commercial samples is shown (;0.10.5%). Achieved linearity is higher than 99%. (b) According to statistical analysis the two pharmaceutical brands (1 and 2) present wider spread distributions.

1298

Volume 66, Number 11, 2012

TABLE III. Statistical parameters for RP-HPLC determination of Met and Gly. Met Equation r2 Dynamic range (mg L1) Slope standard error Intercept standard error LOD (lg L1)b LOQ (lg L1)
a b

Gly AGly /ABena = 0.193[Gly] 0.0013 0.9998 10.060.0 0.02 1.166 6.3 21.1

A = 82682 [Met] 51778 0.9986 5.030.0 2101.1 42860 12.1 40.3

FIG. 5. Typical chromatograms obtained from standard solutions of Met at (a) 15 mg L1, (b) 20 mg L1, and (c) 25 mg L1.

A = Area. LOD was calculated as three standard deviations of the baseline noise versus calibration slope.

for Ben, and 11.3 for Gly. The variation in retention times among three replicate injections of Met, Ben, and Gly standard solutions was ,0.07 min. Peak areas from Met and from Gly/ benzophenone were plotted against corresponding concentrations in the range of 5 to 30 lg mL1 for Met and 10 to 60 lg mL1 for Gly. Linear regression parameters of Met and Gly determination are presented in Table III. The results showed highly reproducible calibration graphs with correlation coefcients of .0.99. Limit of detection (LOD) and limit of quantication (LOQ) values are also shown in Table III. In order to demonstrate the validity and applicability of the proposed HPLC method, recovery tests were carried out by analyzing three laboratory-prepared samples of Met, Gly, and benzophenone (Fig. 7). Analysis of Tablets. The results with laboratory-prepared samples suggested the successful use of the proposed HPLC method for the assay of Met and Gly in commercial tablet dosage form. Eight different lm-coated tablets were analyzed; four contained 500 mg Met and another four contained 5 mg Gly each; all with excipients. The results show that the proposed HPLC method can be used for the determination of Met and Gly without prior separation of the excipients. Figures 8 and 9 show the typical chromatograms obtained from the analysis of Met and Gly in tablets. The drugs formed wellshaped, symmetrical, single peaks, well separated from the solvent front. No interfering peaks were obtained in the chromatogram due to tablet excipients. The utility of all of the proposed methods was veried by means of replicate estimations of pharmaceutical preparations, and the results obtained were evaluated statistically. Table IV shows the

FIG. 7. Chromatogram obtained from laboratory-prepared mixture containing 500 mg L1 Met, 5.25 mg L1 Gly, and 100 mg L1 benzophenone.

FIG. 8. Typical HPLC chromatogram obtained from two different pharmaceutical tablets of Met: (a) Metixor-Et; and (b) Glucophaget.

FIG. 6. Typical chromatograms obtained from standard solutions of (a) Gly 20 mg L1 IS (benzophenone 100 mg L1); (b) Gly 40 mg L1 IS; and (c) Gly 60 mg L1 IS.

FIG. 9. Typical HPLC chromatogram obtained from two different pharmaceutical tablets of Gly: (a) Euglucont and (b) Kenert.

APPLIED SPECTROSCOPY

1299

TABLE IV. Recoveries obtained with the proposed HPLC method in the analysis of the pharmaceutical formulations of Met and Gly. Met Samples 1 2 3 4 Labeled weight (mg/tablet) 500 500 500 500 Found weight (mg/tablet) 480.98 478.57 475.64 503 % R (RSD) 96.20 95.71 95.13 100.60 (1.21) (7.38) (7.05) (1.47) Labeled weight (mg/tablet) 5 5 5 5 Gly Estimated weight (mg/tablet) 4.96 4.98 5.04 5.05 % R (RSD) 99.19 99.57 100.72 101.03 (1.79) (0.61) (1.10) (0.78)

results obtained from the analysis of tablets by the HPLC method. Results obtained from the proposed methods (Tables II and IV) of the analysis of both drugs in tablets indicated that they can be used for direct quantization and routine QC analysis of these pharmaceuticals. Comparison of Methods. The statistical parameters and the results obtained in the analysis of the test set of synthetic samples are summarized in Tables I and III. The results obtained for the two analytes are satisfactory for all the methods employed and a comparison between LIBSbased and HPLC-based methods was carried out by using the same eight commercial samples (Tables II and IV). Results from both analyses were compared using singlefactor analysis of variance (ANOVA). From the ANOVA result, it can be observed that there are no signicant differences between results of both analyses; i.e., differences can be attributed to random variations in the experiment. An analysis of the recoveries leads to the conclusion that for the assayed concentrations the two methods are adequate. In the case of Met determination, satisfactory recoveries were obtained, between 93 % and 105 % by using both methods, while recoveries closer to 100 % were obtained by using the HPLC method (between 95 % and 101 %). On the other hand, the Gly recoveries obtained by LIBS present the lowest and highest values for samples 1 and 4, respectively (Table II); meanwhile, recoveries very close to 100 % were also obtained by HPLC-determination (between 99 % and 101 %). It is convenient to indicate that the most reproducible method is the HPLC method (very low relative standard deviation (RSD) values); for this reason, it would be the most indicated method for trace analysis of Gly. Overall, adequate recoveries for both analytes and no signicant differences in the recovery values for the two methods were found. On the other hand, for Gly analysis, successful recoveries were obtained when the HPLC method was used and recoveries with high deviations were found with LIBS. In this particular case, the intrinsic detection limits of LIBS limit the improvement of analytical recoveries and therefore Gly results from HPLC analysis are signicantly more accurate and precise than results based on LIBS, according to the manufacturer information (Table II and IV); however, it must be taken into account that ANOVA analysis showed that both methods are statistically the same, despite the differences in accuracy and RSD. In addition, the LIBS-based method offers unbeatable advantages such as collection of a lot of information in a rather fast and cost-effective way, easy sample pretreatment or none at all, availability of multielemental analysis, and no requirement of additional chemical compounds or solvents at all, making the analysis environmentally friendly, which makes a more

appropriate approach for screening industrial data. In general, the results obtained in this work represent a good alternative for the pharmaceutical industry primarily because demands are made to increase production efciency, while at the same time maintaining or improving product quality by monitoring production in a fast, easy, and environmentally friendly way.

CONCLUSIONS
Time- and spectral-resolved optical emissions of a pharmaceutical sample can be used to characterize its atomic composition and quantify an analyte in a faster and more practical way than conventional methods. Also, no additional chemicals are used in the proposed LIBS method, making it safe and environmentally friendly. Results obtained in this work show that the proposed LIBS method is able to quantify pharmaceutical ingredients in solid dosages by monitoring specic elements such as Cl. Results can be obtained in just a few minutes with simple sample preparation using small amounts of pharmaceutical dosage (grams). Therefore, the proposed methodology can be implemented to monitor the pharmaceutical production process in situ in real time and qualitative and quantitative analysis could be used for inspection and recognition of authenticity. On the other hand, the reported HPLC method enables the quantication of Met and Gly with superior accuracy and precision compared to the LIBS method, but it is not suitable for use as a screening method. Although the data acquisition time could be improved, avoidance of the use of expensive and environmentally harmful solvents in the mobile phase was preferred. On the other hand the extraction of Met is done only with ultrapure water. All of these HPLC procedures are relatively rapid and work without solving equations or performing separation steps. The proposed HPLC method provides good separation of Met and Gly from matrix within a relatively short analysis time. It is simple and accurate for the determination of Met and Gly in raw material and pharmaceutical formulations. The proposed methods are suitable for QC laboratories. High recovery shows that the methods are free from the interference of commonly used excipients and additives in the formulations of drugs.
ACKNOWLEDGMENTS The authors acknowledge the nancial support from Council of Science and Technology of the State of Guanajuato, CONCyTEG, Mexico (Project 09-04K662-072-A02).

1. S.E. Inzzuchi. Oral antihyperglycemic theraphy for type 2 diabetes: scientic review. J. Am. Med. Assoc. 2002. 287(3): 360-372. 2. A.J. Krentz, C.J. Beiley. Oral antidiabetic agents: current role in type 2 diabetes mellitus. Drugs. 2005. 65(3): 385-411. 3. A. Nourparvar, A. Bulotta, U. Di Mario. R. Perfetti. Novel strategies for

1300

Volume 66, Number 11, 2012

4. 5. 6. 7. 8.

9.

10.

11.

12. 13. 14. 15.

16. 17. 18.

19.

the pharmacological management of type 2 diabetes. Trends Pharmacol. Sci. 2004. 25(2): 86-91. G.E. Dailey. Glybenclamide/metformin tablets: A new therapeutic option for the management of type 2 diabetes. Exp. Opin. Pharmacother. 2003. 4(8): 1417-1430. World Health Organization (WHO). Media centre. Diabetes. Fact sheet No. 312. 2011. [Available at http://www.who.int/mediacentre/factsheets/ fs312/en/]. A. Patrignani, M. Musumeci, F. Bosco, M. Pane. Counterfeiting: a global spread, a global threat. Trends Organ. Crim. 2009. 12(1): 59-77. C.F. Ferreira, E.A. Menezes, W.O. Matos, D. Milori, A. Noriega, L. Martin-Neto. Determination of Ca in breakfast cereals by laser induced breakdown spectroscopy. Food Control. 2010. 21(10): 1327-1330. L.C. Trevizan, D. Santos, R. Samad, N. Vieira, L. Nunes, I. Runa, F. Krug. Evaluation of laser induced breakdown spectroscopy for the determination of micronutrients in plants. Spectrochim. Acta, Part B. 2009. 64(5): 369-377. M.R. Martelli, F. Brygo, A. Sadoudi, P. Delaporte, C. Barron. LaserInduced Breakdown Spectroscopy and Chemometrics: A Novel Potential Method to Analyze Wheat Grains. J. Agric. Food Chem. 2010. 58(12): 7126-7134. D.C. Alvey, K. Morton, R.S. Harmon, J.L. Gottfried, J.J. Remus, L.M. Collins, M.A. Wise. Laser-induced breakdown spectroscopy-based geochemical ngerprint for the rapid analysis and discrimination of minerals: the example of garnet. Appl. Opt. 2009. 49(13): 168-180. R.S. Harmon, F.C. De Lucia, N.J. McMillan, J.F. Jenkins, M.E. Walsh, A.W. Miziolek. Laser-induced breakdown spectroscopyAn emerging chemical sensor technology for real-time eld portable, geochemical, mineralogical and environmental applications. Appl. Geochem. 2006. 21(5): 730-747. S. Pandhija, N.K. Rai, N.K. Thakur. Contaminant concentration in environmental samples using LIBS and CF-LIBS. Appl. Phys. B. 2010. 98(1): 231. V.K. Singh, A.K. Rai, P.K. Rai, P.K. Jindal. Cross-sectional study of kidney stones by laser-induced breakdown spectroscopy. Lasers Med. Sci. 2009. 24(5): 749-759. J. Anzano, B. Bonilla, B. Montul-Ibor, J. Casas-Gonzalez. Rapid characterization of analgesic tablets by laser-induced breakdown spectroscopy (LIBS). Med. Chem. Res. 2009. 18(8): 656-664. G.G. Arantes de carvalho, L.C. Nunes, P. Florencio de Souza, F.J. Krug, T.C. Alegre, D. Santos. Evaluation of laser induced breakdown spectroscopy for the determination of macro and micronutrients inpharmaceutical tablets. J. Anal. At. Spectrom. 2010. 25: 803-809. L. St-Onge, E. Kwong, M. Sabsabi, E.B. Vadas. Rapid analysis of liquid formulations containing sodium chloride using laser-induced breakdown spectroscopy. J. Pharm. Biomed. Anal. 2004. 36(2): 277-284. L. St-Onge, E. Kwong, M. Sabsabi, E.B. Vadas. Quantitative analysis of pharmaceutical products by laser-induced breakdown spectroscopy. Spectrochim. Acta, Part B. 2002. 57(7): 1131-1140. J.A. Good, M.D. Mowery, R.A. Reed. Rapid at-line determination of lubricant distribution in powder blends using laser induced breakdown spectroscopy (LIBS). AAPS Pharm. Sci. Suppl. 3. 2001. Abstract No. 2344. L. St-Onge, J.F. Archambault, E. Kwong, M. Sabsabi. E.B. Vadas. Rapid

20.

21.

22.

23.

24.

25.

26. 27. 28. 29.

30.

31.

32.

33.

34.

quantitative analysis of magnesium stearate in tablets using laserinduced breakdown spectroscopy. J. Pharm. Pharmaceut. Sci. 2005. 8(2): 272288. M.D. Mowery, R. Sing, J. Kirsch, A. Razaghi, S. Be chard, R.A. Reed. Rapid at-line analysis of coating thickness and uniformity in solid dosage forms using laser induced breakdown spectroscopy. J. Pharm. Biomed. Anal. 2002. 28: 935-943. E.P. Lai, S.Y. Feng. Solid phase extractionnon-aqueous capillary electrophoresis for determination of metformin, phenformin and glyburide in human plasma. J. Chromatogr. B, Anal. Technol. Biomed. Life Sci. 2006. 843(1): 94-99. C. Hess, F. Musshoff, B. Madea. Simultaneous identication and validated quantication of 11 oral hypoglycaemic drugs in plasma by electrospray ionisation liquid chromatography-mass spectrometry. Anal. Bioanal. Chem. 2011. 400(1): 33-41. Y. Wang, Y. Tang, J. Gu, J. Fawcett, X. Bai. Rapid and sensitive liquid chromatography-tandem mass spectrometric method for the quantitation of metformin in human plasma. J. Chromatogr. B. 2004. 808(2): 215-219. F. Collin, H. Khoury, D. Bonnefont-Rousselot, P. Therond, A. Legrand, D. Jore, M. Gardes-Albert. Liquid chromatographic/electrospray ionization mass spectrometric identication of the oxidation end-products of metformin in aqueous solutions. J. Mass Spectrom. 2004. 39(8): 890-902. N. Koseki, H. Kawashita, M. Niina, Y. Nagae, N. Masuda. Development and validation for high selective quantitative determination of metformin in human plasma by cation exchanging with normal-phase LC/MS/MS. J. Pharm. Biomed. Anal. 2005. 36(5): 1063-1072. V. David, A. Medvedovici, F.J. Albu. Liq. Chromatogr. Relat. Technol. 2005. 28: 81-95. S. Aburuz, J. Millership, J.J. McElnay. Chromatogr. B. 2006. 832: 202207. W. Zhang, B. Xiang, Y. Wu, E. Shang. Anal. Chim. Acta. 2005. 550: 7781. S. Naraharisetti, J. Kirby, F. Hebert, R. Easterling, D. Unadkat. Validation of a sensitive LC-MS assay for quantication of glyburide and its metabolite 4-trashydroxy glyburide in plasma and urine: an OPRU Network study. J. Chromatogr. B. 2007. 860(1): 34-41. C. Gedeon, B. Kapur, K. Aleksa, G. Koren. A simple and rapid HPLC method for the detection of glyburide in plasma original research communication (analytical). Clin. Biochem. 2008. 41(3): 167-173. Y. Ralchenko, A.E. Kramida, J. Reader, NIST ASD Team NIST Atomic Spectra Database (ver.4.0.1). [Online]. Gaithersburg, MD: National Institute of Standards and Technology, 2010. http://physics.nist.gov/asd [2010, May 9]. D.A. Cremers, L. J. Radziemski. Detection of chlorine and uorine in air by laser-induced breakdown spectroscopy. Anal. Chem. 1983. 55(8): 1252-1256. M. Tran, Q. Sun, B.W. Smith, J.D. Winefordner. Determination of F, Cl, and Br in solid organic compounds by laser-induced plasma spectroscopy. Appl. Spectrosc. 2001. 55(6): 739-744. L.D. Ph, A.J. Amourox, Time-Resolved Laser-Induced Breakdown Spectroscopy: Application for Qualitative and Quantitative Detection of Fluorine, Chlorine, Sulfur and Carbon in Air. Appl. Spectrosc. 1998. 52(10): 1321-27.

APPLIED SPECTROSCOPY

1301