Ultrasound and microwave techniques were used to extract tobacco alkaloids, and response surface methodology was used to optimize extraction con- ditions.
JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001 309
AGRICULTURAL MATERIALS
Comparison of Methods for Extraction of Tobacco Alkaloids NÁDIA M. JONES Centro de Investigação de Ciências Agrárias Tropicais, DAIAT, Instituto Superior de Agronomia, Tapada da Ajuda, 1349-017 Lisbon, Portugal M. GABRIELA BERNARDO-GIL Centro de Engenharia Biológica e Química, DEQ, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisbon, Portugal M. GRAÇA LOURENÇO1 Centro de Investigação de Ciências Agrárias Tropicais, DAIAT, Instituto Superior de Agronomia, Tapada da Ajuda, 1349-017 Lisbon, Portugal
Ultrasound and microwave techniques were used to extract tobacco alkaloids, and response surface methodology was used to optimize extraction conditions. Ultrasonic technique factors were temperature, 30–85°C; time, 3–45 min; solvent volume, 8–80 mL. Microwave extraction factors were pressure, 15–75 psi; time, 3–40 min; power, 30–90% of the maximum magnetron power of 650 W. Soxhlet and solvent AOAC-modified extraction methods were also applied after some improvements. Nicotine, nornicotine, anabasine, and anatabine were quantified by gas chromatography. A steam distillation International Standards Organization method for total alkaloid evaluation was used as reference. The results obtained by the different methods were compared using a least squares deviation test. The ultrasonic and the proposed modified-AOAC extraction method were the more convenient with regard to practicability and precision. The relative deviations (n = 5) were as follows: For the ultrasonic method in low-level alkaloid tobaccos, 0.7% nicotine and 1.4–14% minor alkaloids; in high-level alkaloid tobaccos, 2.4% nicotine and 4.5–5.1% minor alkaloids. For the modified AOAC method in low-level alkaloid tobaccos, 0.9% nicotine and 2.4–11.6% minor alkaloids; and in high-level alkaloid tobaccos, 1.7% nicotine and 2.0–2.4% minor alkaloids.
he correct evaluation of alkaloids in tobacco is of great importance because of the role of alkaloids in tobacco flavor and in public health (1, 2). Extraction is perhaps the most essential step in the evaluation process. The determination of individual tobacco alkaloids is usually performed by gas chromatography (GC). Independently of the method of quantification, however, samples must be submitted to time-consuming preparation to extract the alkaloids from the
T
Received March 20, 2000. Accepted by WEB June 27, 2000. 1 Author to whom correspondence should be addressed.
complex tobacco matrix. Many of the conventional extraction procedures, such as Soxhlet, involve lengthy operations, large consumption of reagents, and eventually alkaloid degradation. Most laboratories determine total alkaloids in tobacco as nicotine, applying an International Standards Organization (ISO) method (3) based on steam distillation and spectrophotometric quantification. Shaking extraction procedures using different solvents have also been used, and it has been verified that the presence of water increases the extractive power of the solvents (4). Methods based on the selective solubility of alkaloids in organic solvents and in water have been applied, for example in the AOAC method (5), to quantify the total alkaloids (as nicotine), tertiary alkaloids (as nicotine), and secondary alkaloids (as nornicotine) by colorimetry. The extraction of alkaloids by Soxhlet has been used with several solvents, e.g., methanol as proposed by Burns and Collin (6), which quantifies nicotine and minor alkaloids by gas chromatography in different tobacco types. Ultrasonic extraction uses high frequency sound to disrupt or detach the target analyte from the matrix. Severson et al. (7) proposed a method based on ultrasonic extraction, for individual alkaloid quantification with methanolic 0.05N KOH as extracting solvent. In a previous study comparing 3 extraction methods (Soxhlet, AOAC-modified, and an ultrasonic method), the best results were obtained by the AOAC method (8). Microwave energy has been used to digest different matrixes. Improvements in design and software of microwave equipment have permitted safe control of pressure and temperature, making it possible to use microwave energy for the solvent extraction of several organic compounds (9). Microwave-assisted extraction (MAE) enhances the effectiveness of solvents (polar solvents) in contact with solid samples, and has been applied to different matrixes such as biological materials, food, and soil (10–13). Alternatively, the microwave-assisted process (MAP) uses microwave energy-transparent solvents (apolar solvents), and consists of a partitioning mechanism in which the sample acts as a good dielectric in the presence of a low-dielectric, poorly heated solvent (9, 14). MAE was used by Bichi et al. (15) for extraction of
310 JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001 Table 2. Codified variable levels for microwave extraction optimization
Table 1. Codified variable levels for ultrasonic extraction optimization Codified variables
Temperature, °C
Time, min
Preliminary RSM –1
36.0
7.5
16.5
0
45.0
14.0
29.0
1
54.0
20.5
41.5
Pressure, psi
Time, min
Power, %a
–1
27.0
14.0
42.0
0
45.0
21.5
60.0
1
63.0
29.0
78.0
Codified variables
Solvent volume, mL
a
Percent of maximum magnetron power (650 W).
Final RSM –1
41.0
17.0
32.0
0
57.5
27.5
50.0
1
74.0
38.0
68.0
pyrrolizidine alkaloids from Senecio spp. dried plants. The recent progress in ultrasonic and microwave extraction has increased interest in evaluating and optimizing the applicability of such methodologies to the extraction of tobacco alkaloids. In the present study, an ultrasonic and a new microwave-assisted extraction method were optimized using response surface methodology RSM (16). These techniques were based on mathematical and statistical methods used for
modeling and analyzing a response that is influenced by a collection of variables. The optimum response was found by using the fitted model. To simultaneously optimize 3 controllable parameters for ultrasonic and microwave-assisted extractions of tobacco alkaloids, RSM was applied through a central composite rotatable second-order design (CCD). Nicotine concentration in tobacco was the response evaluated. A second-order polynomial equation was used to approximate the response: 3
3
i=1
i=1
3
3
Y = β 0 + ∑ β i xi + ∑ β ii xi2 + ∑ ∑ β ij xi x j i=1 j = i+1
where Y is the response function; β0 is the center point of the system; βi, βii and βij represent the coefficients of the linear,
Table 3. Experimental design and results of nicotine (%) for ultrasonic extraction
Observation No.
Temperature, °C
Time, min
Solvent volume, mL
X1
X2
X3
x1
x2
x3
Nicotine, %
–1
3.279
Coded variablesa
1
41
17
32
–1
–1
2
41
17
68
–1
–1
1
3.401
3
41
38
32
–1
1
–1
3.750
4
41
38
68
–1
1
1
3.745
5
74
17
32
1
–1
–1
3.973
6
74
17
68
1
–1
1
4.319
7
74
38
32
1
1
–1
4.498
8
74
38
68
1
1
1
4.483
9
30
27.5
50
–α
0
0
3.145
10
85
27.5
50
α
0
0
4.595
11
57.5
10
50
0
–α
0
3.945
12
57.5
45
50
0
α
0
4.243
13
57.5
27.5
20
0
0
–α
4.217
14
57.5
27.5
80
0
0
α
4.164
15
57.5
27.5
50
0
0
0
4.393
16
57.5
27.5
50
0
0
0
4.023
17
57.5
27.5
50
0
0
0
4.174
18
57.5
27.5
50
0
0
0
4.126
a
α = Axial spacing (2k)1/4; k = number of factors. In this case, α = 1.68179.
JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001 311 Table 4. Full model ANOVA results for ultrasonic extraction Coded variablesa
p-Valueb
Coefficient estimate
Standard error
t-Test
Intercept
4.184
0.074
56.417
0.0000 ***
x1
0.405
0.040
10.085
0.0000 ***
x2
0.147
0.040
3.652
0.0065 *** 0.5317
0.026
0.040
0.654
x1x1
–0.134
0.042
–3.200
0.0126 **
x1x2
–0.016
0.053
–0.300
0.7719
x1x3
0.027
0.053
0.509
0.6243
x2x2
–0.054
0.042
–1.304
0.2285
x2x3
–0.061
0.053
–1.161
0.2790
x3
x3x3
–0.020
0.042
–0.487
0.6392
Sum of aquares
Degrees of freedom
Mean squares
f0
Model
2.8232
9
0.3137
14.19 ***
Residual
0.1766
8
0.0221
Total
2.9998
17
Source of variation
a b
x1 = Coded temperature; x2 = coded time; x3 = coded solvent volume. * **Significant at 5%; *** significant at 1%.
quadratic, and interactive effects, respectively; and xi, xii, and xixj represent the linear and interactive effects of the independent variables (16).
Experimental
The optimized conventional methods were compared after some modifications to make them more practical. Differences between the methods were investigated by the comparison of means using the least squares deviation (LSD) test (16). The precision and practicability were analyzed and are discussed below.
A flue-cured (low-level alkaloids) tobacco from France and a Burley (high-level alkaloids) tobacco from Guatemala were obtained directly from Tabaqueira, Portugal. The cured tobaccos were dried overnight (15–18 h) in an oven at 50°C, and ground to 0.4 mm in a Wiley No. 4 mill. All reagents used were of analytical grade. Nicotine, nornicotine, and anabasine were obtained from Sigma Chemi-
Materials and Chemicals
Table 5. Second model ANOVA results for ultrasonic extraction Coded variablesa
Coefficient estimate
Standard error
t-Test
p-Valueb
Intercept
4.163
0.055
76.271
0.0000 ***
x1
0.405
0.037
10.900
0.0000 ***
0.147
0.037
3.948
0.0019 ***
x1x1
–0.129
0.038
–3.424
0.0050 ***
x1x2
–0.016
0.049
–0.324
0.7514
x2x2
–0.050
0.038
–1.329
0.2087
Source of variation
Sum of squares
Degrees of freedom
Mean squares
f0
Model
27731
5
0.5546
29.36 ***
0.0189
x2
Residual
0.2267
12
Total
2.9998
17
a b
x1 = Coded temperature; x2 = coded time; x3 = coded solvent volume. **** Significant at 1%.
312 JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001
cal Co., St. Louis, MO, quinaldine from Fluka Chemie AG (Buchs, Switzerland) and 2,4′-bipyridyl from Aldrich Chemical Co. (Milwaukee, WI). METHODS
Extraction Procedures Steam distillation.—A normalized ISO method (3) was followed. Approximately 1 g ground tobacco was subjected to steam distillation under strong alkaline conditions, followed by spectrometric measurement of the distillate absorption to evaluate the total alkaloid concentration, expressed as percentage of nicotine. Solvent extraction.—An AOAC method (5), based on the work of Cundiff and Markunas (17) was applied. Some modifications such as sample weight were introduced; however, the proportion of different reagent volumes was maintained. Centrifugation was used to improve phase separation. An alternative high resolution GC quantification method was used. For the modified method, 1 g ground tobacco was accurately weighed into a 100 mL Erlenmeyer flask; 1 mL of each of 2 internal standards (quinaldine and 2,4′-bipyridyl) and 6 mL 5% acetic acid were added and swirled until the tobacco was thoroughly wetted;
Figure 1. Nicotine concentration (% in tobacco) as a function of temperature and time variations for ultrasonic extraction.
Table 6. Experimental design and results of nicotine (%) for microwave extraction Coded variablesa
Pressure, psi
Time, min
Power, %
X1
X2
X3
x1
x2
x3
Nicotine, %
1
27
14
42
–1
–1
–1
4.462
2
27
14
78
–1
–1
1
4.319
3
27
29
42
–1
1
–1
4.686
4
27
29
78
–1
1
1
4.495
5
63
14
42
1
–1
–1
4.953
6
63
14
78
1
–1
1
4.849
7
63
29
42
1
1
–1
4.933
8
63
29
78
1
1
1
4.858
Observation No.
9
15
21.5
60
–α
0
0
4.660
10
75
21.5
60
α
0
0
5.075
11
45
3
60
0
–α
0
4.256
12
45
40
60
0
α
0
4.996
13
45
21.5
30
0
0
–α
4.659
14
45
21.5
90
0
0
α
4.497
15
45
21.5
60
0
0
0
4.934
16
45
21.5
60
0
0
0
4.538
17
45
21.5
60
0
0
0
4.826
18
45
21.5
60
0
0
0
4.937
a
α = Axial spacing (2k)1/4; k = number of factors. In this case, α = 1.68179.
JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001 313
KOH per liter of methanol were added into a round flask with a vertical condenser. The extraction was performed at 82°C for 40 min. Microwave extraction.—To avoid damaging the tobacco alkaloid extraction, MAP was considered because of its analyte cooling extracting capacity. Preliminary experiments, using a STAR System 2, a microwave technology (CEM) with open vessel microwave digestion, were performed with benzene and hexane. Because moisture is an essential parameter in MAP, leading to cell rupture and diffusion of its content into the cooler solvent, water was added to the tobacco. Experiments were developed with samples containing 1 and 2 g ground tobacco with 20–60% of water by weight, 25 mL solvent, and extracted by microwave digestion with different time/temperature programs. In spite of the minimized risk of analyte degradation, only 20% of the expected value was obtained. MAE was also applied using a CEM MDS-2000 (900 W magnetron delivers at maximum 650 W) microwave digestion system equipped with closed CEM PFA, all Teflon containers and pressure control, which allow higher temperatures without solvent boiling. This equipment has no temperature control and no vapors exhaustion. To be consistent with other methods, a tobacco + solvent proportion of 1 g tobacco to 40 mL extracting solvent volume was used. A solution containing 0.05M KOH per liter methanol was used, as in other reported tobacco alkaloids extractions (7, 15). Considering that pressure, time, and power were the main influencing factors, the RSM was used to optimize these extraction conditions (Table 2). The optimized microwave method was performed at 75 psi for 40 min at 60% total power.
and 40 mL benzene–chloroform solution, and 4 mL 36% sodium hydroxide were then added. The flask was tightly stopped and shaken for 20 min using a wrist-action shaker. Contents of the flask were centrifuged and then filtered through Whatman No. 4 paper, and the benzene–chloroform extract was subsequently analyzed by GC. 6000 rpm during 15 min. Soxhlet extraction.—Ground tobacco (2 g) was extracted with methanol for 2 h in a Soxhlet system according to a method proposed by Burns and Collin (6) for GC alkaloid quantification. Two mL of 2 internal standards, quinaldine and 2,4′-bipyridyl, was added to the tobacco sample. Ultrasonic extraction.—In previous studies, some solvents were tested by ultrasonic extraction. It was verified that water and temperature control improved the extractive capacity of solvents (18). However, taking in account that water is not recommended in subsequent GC analysis, an ultrasonic extraction based on Severson’s method (7) with methanolic potassium hydroxide was implemented. The influence of extracting volume, temperature, and time in ultrasonic extraction was tested and optimized by RSM. Considering previous studies and keeping temperature below methanol boiling point, a preliminary CCD was established, varying the temperature between 30 and 60°C, time from 3 to 25 min, and solvent volume between 8 and 50 mL. From the results, a second CCD was implemented (Table 1). Extractions were performed in an ultrasonic cleaning bath (Elma Transsonic T660/H) with a thermostatic heater (Clifton, NE 4-T, Nichel-Electro, North Somenrset, UK. The optimized ultrasonic extraction method was performed as follows: To 1 g ground tobacco, accurately weighed, 1 mL of each of 2 internal standards (quinaldine and 2,4′-bipyridyl) and 40 mL of a solution containing 0.05M
Table 7. Full model ANOVA results for microwave extraction Coded variablesa
Coefficient estimate
Standard error
t-Test
p-Valueb
Intercept
4.808
0.084
57.234
0.0000 ***
x1
0.171
0.046
3.745
0.0057 *** 0.0303 **
x2
0.120
0.046
2.627
x3
–0.058
0.046
–1.263
0.2421
x1x1
0.023
0.047
0.490
0.6376
x1x2
–0.051
0.059
–0.864
0.4130
x1x3
0.019
0.059
0.326
0.7530
x2x2
–0.062
0.047
–1.315
0.2249
x2x3
–0.002
0.059
–0.040
0.9691
x3x3
–0.079
0.047
–1.674
0.1327
Sum of squares
Degrees of freedom
Mean squares
f0
Model
0.80
9
0.089
3.14 *
Residual
0.23
8
0.028
Total
1.03
17
Source of variation
a b
x1 = Coded pressure; x2 = coded time; x3 = coded power. * Significant at 10%; ** significant at 5%; *** significant at 1%.
314 JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001
GC Analysis The alkaloids were analyzed by a GC method Lourenço et. al. (19) with a 5890 Model series II Hewlett Packard (HP) gas chromatograph equipped with an HP GC automatic liquid sampler and an HP 3396 integrator. A capillary poly 5% diphenyl–95% dimethyl siloxane base modified, bonded PTA5 column (Supelco, Bellefonte, PA; 30 m, 0.25 mm, 0.50 µm) was used with the following oven programs: for nicotine an initial temperature of 160°C for 9 min, 160–280°C at 30°C/min, and finally 280°C for 5 min; for the other alkaloids an initial temperature of 150°C for 20 min, 150–280°C at 25°C/min and finally 280°C for 5 min. Linear flow velocity of helium was adjusted to 30 cm/s at column temperature of 160°C; injector temperature to 250°C, and flame ionization detector to 300°C; 1–3 µL extract volumes were injected in split mode (1:20). The different alkaloids were quantified with calibration curves. Because it was not possible to purchase anatabine, the response factor of this alkaloid was taken to be the same as for anabasine. Two internal standards, quinaldine (10 mg/mL methanol) for nicotine analysis and 2,4′-bipyridyl (0.2 mg/mL methanol) for minor alkaloids evaluation were used. Low level alkaloid extracts were concentrated at 40°C in a rotorvapor before GC analysis. Results and Discussion
Optimization Ultrasonic extraction.—Optimization of the ultrasonic extraction was evaluated by quantification of nicotine with a second-order CCD using the limits shown in Table 1. The first RSM analysis showed that higher temperature, longer time, and larger solvent volume increased nicotine extraction. A second design was implemented using more suitable limits (Table 1) made possible by adaptation of a condenser to the extraction vessel. Table 3 presents the experimental design and the results of nicotine concentration. Four replicates at the central point were performed to allow estimation of the pure error sum of squares. The ordinary least squares regression method was applied to estimate the function. The regression and its analysis were made with Statistica Version 5.0 software (StatSoft). Residual homogeneity and normality were tested, and in both cases, the null hypothesis was accepted. A second order model was fitted and the effects were evaluated by analysis of variance (ANOVA) as a test for significance (Table 4). The full model was significant at 99% confidence with R2 = 0.94 and RAdj2 = 0.87. Temperature and time linear effects were significant at 99% confidence. Quadratic effect of temperature was also significant with 95% confidence. As solvent volume was not significant, a second model was fitted, suppressing that factor (Table 5). This model was also significant at 99% significance with R2 = 0.92 and RAdj2 = 0.89. Temperature and time linear effects and quadratic effect of temperature were also significant at 99% confidence. Figure 1 shows nicotine extraction with time and temperature variations. Optimum point for this model was
Figure 2. Nicotine concentration (% in tobacco) as function of % of magnetron maximum power, pressure, and time variations for microwave extraction.
JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001 315 Table 8. Individual tobacco alkaloids extracted by different methodsa Extraction methodsb
Nicotine
Nornicotine
Anabasine
Anatabine
High-level alkaloid Burley tobacco SE
3.96 ± 0.07c
0.460 ± 0.009c
0.028 ± 0.001c
0.199 ± 0.005c
SO
d
3.55 ± 0.36
c
0.429 ± 0.056
c
0.029 ± 0.004
0.191 ± 0.026c
US
4.00 ± 0.10c
0.481 ± 0.022c
0.032 ± 0.001c
0.203 ± 0.010c
c
d
d
0.019 ± 0.006
0.117 ± 0.066d
MW
3.86 ± 0.10
0.187 ± 0.077
Low-level alkaloid flue-cured tobacco SE
0.75 ± 0.01
0.030 ± 0.003c
0.008 ± 0.000c
0.051 ± 0.001c
SO
0.74 ± 0.02c
—
—
0.032 ± 0.011d
US
c
0.74 ± 0.01
0.030 ± 0.004
0.008 ± 0.001
0.050 ± 0.001c
MW
0.76 ± 0.04c
0.032 ± 0.008c
—
0.042 ± 0.008e
a b c,d,e
c
c
c
The results are expressed as % in tobacco ± ts/√n. Abbreviations: SE = solvent extraction (modified AOAC method); SO = Soxhlet; US = ultrasonic; MW = microwave. Different characters indicate significantly different values (95% LSD test).
temperature = 82ºC and time = 40 min. A volume of 40 mL was adopted. Microwave-assisted extraction.—A set of 18 experiments using the limits of Table 2 was used for optimization. A second order equation was fitted to nicotine concentration (Table 6). The goodness of fit was assessed through ANOVA after normality and homogeneity tests (Table 7). The full model was significant at 90% confidence with R2 = 0.78 and RAdj2 = 0.53. These results reveal a rather low adjustment of the model which may be due to the inclusion of some insignificant terms. However, the model can be considered as acceptable for a preliminary study, having R2 > 60% (20). Pressure and time linear effects were significant at 99 and 95% confidence, respectively. None of the quadratic effects were significant, which allows us to conclude that neither a maximum nor a minimum of the response function was reached. Little curvature may signify that operating conditions are far from the optimum. Analyzing response surfaces (Figure 2), we can state that more pressure and time were needed to reach the optimum; however, the lack of temperature control and practical constraints made the safe use of other limits impossible. In spite of the low significance of power input, the graphs suggest that an intermediate value should give the best result. Consequently, the method was developed with maximum equipment security at pressure of 75 psi for 40 min and with 60% maximum magnetron power.
For low level alkaloid tobaccos, all the methods extracted similarly precise amounts of the different alkaloids except anatabine. For high level alkaloid tobaccos, the Soxhlet method extracted significantly less nicotine with lower precision, and the microwave method was less efficient in the extraction of secondary alkaloids. Considering the practicability of the studied methods, the solvent purification extraction method (AOAC modified method) is perhaps more laborious; however, several experiments can be performed simultaneously. In spite of using benzene, this method has the advantage of producing cleaner extracts, which generated
Comparison of Methods To assess method adequacy for the analysis of high and low alkaloid levels, 2 tobacco samples were chosen. Five replicate experiments by each method were simultaneously developed. Significance of differences between alkaloid mean concentration values and their deviations (ts/ n ) were evaluated using an LSD test with 95% significance (Table 8).
Figure 3. Comparison of total alkaloids extracted by different methods. SD = steam distillation (ISO method); SE = solvent extraction (modified AOAC method); SO = Soxhlet; US = ultrasonic; MW = microwave; B = Burley tobacco; FC = flue-cured tobacco.
316 JONES ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 2, 2001
chromatograms with fewer peaks. The Soxhlet method used long extraction times with large solvent volumes, which required concentration of the extracts and led to eventual alkaloid degradation, possibly one explanation for the lack of precision. Ultrasonic extraction is a practical method, and the use of a higher temperature and a higher extraction time increased extraction efficiency. Microwave-assisted extraction is a simple and rapid method which leads to satisfactory extractions in spite of the less precise minor alkaloid results. Eventually, results could be improved with higher pressure (Figure 2) if a temperature control device is used. Thus, in terms of practicability and precision, the proposed modified AOAC and the ultrasonic methods were the more convenient. The best relative ts / n were obtained with these methdeviations Y × 100% ods, ranging from 0.9 to 11.6% with the modified AOAC method and from 0.7 to 14% with the ultrasonic method. Comparison of total alkaloids obtained by the different methods with the standard ISO method showed that higher contents (especially for high-level tobacco alkaloids) were achieved by either method. The total alkaloids by ISO method agrees with only nicotine quantified by the other methods (Figure 3). Acknowledgments The authors thank Technical University of Lisbon for the Nadiá Jones scholarship; Instituto Superior de Agronomia, Prof. Madeira, Pedology Sec., DCA, for microwave equipment facilities, and Prof. Manuela Neves for statistical advice. References (1) Lourenço, M.G. (1999) Rev. Cienc. Agrar. XXII, 63–75 (2) Gorrod, J.W., & Wahren, J. (1993) Nicotine and Related Alkaloids, Chapman & Hall, London, UK
(3) International Standards Organization (1992) International Standard ISO 2881, Geneva, Switzerland (4) Saunders, J.A., & Blume D.E. (1981) J. Chromatogr. 205, 147–154 (5) Official Methods of Analysis (1997) 16th Ed., AOAC INTERNATIONAL, Gaithersburg, MD, Method 960.08 (6) Burns, D.T., & Collin E.J. (1977) J. Chromatogr. 133, 378–381 (7) Severson, R.F., McDuffie, K.L., Arrendale, R.F., Gwynn, G.R., Chaplin, J.F., & Jonhson, A.W. (1981) J. Chromatogr. 211, 111–121 (8) Lourenço, M.G. (1986) Bioquim. Apl. III, 11–21 (9) Jassie, L., Revesz, R., Kierstead, T., Hasty, E., &. Matz, S. (1997) in Microwave-Enhanced Chemistry: Fundamentals, Sample Preparation, and Applications, H.M. Kingston & S.J. Haswell (Eds), American Chemical Society, Washington, DC, pp 569–609 (10) Ganzler, K., Szinai, I., & Salgó, A. (1990) J. Chromatogr. 520, 257–262 (11) Pare, J.R.J., Sigouin, M., & Lapointe, J. (1991) U.S. Patent No. 5002784 (12) Ganzler, K., Salgó, A., & Valkó, K. (1986) J. Chromatogr. 371, 299–306 (13) Lopez-Avila, V., Young, R., & Teplitsky, N. (1996) J. AOAC Int. 79, 142–156 (14) Pare, J.R.J., Matini, G., Bélanger, J.M.R., Li, K., Rule, C., Thibert, B., Yaylayan, V., Liu, Z., Mathé, D., & Jacquault, P. (1997) J. AOAC Int. 80, 928–933 (15) Bichi, C., Beliarab, F.F., & Rubiolo, P. (1992) Lab 2000 6, 36–38 (16) Montgomery, D.C. (1997) Design and Analysis of Experiments, 4th Ed., John Wiley & Sons, New York, NY (17) Cundiff, R.H., & Markunas, P.C. (1960) J. AOAC Int. 43, 519–524 (18) Lourenço, M.G. (1985) Ph.D Thesis, Instituto Superior de Agronomia, Lisbon, Portugal (19) Lourenço, M. G., Matos, A., Oliveira, M.C. (2000) J. Chromatogr. 898, 235–243 (20) Capanzana, M.V., &. Buckle, K.A. (1997) Lebensm. Wiss. Technol. 30, 155–163