Introduction

Capsaicinoids are a group of pungent compounds found mostlyin capsicum fruits, the structures of which are acid amides ofvanillylamine and C9 C11 branched-chain fatty acids. Thereare five naturally occurring capsaicinoids which have beenreported, namely, capsaicin, nordihydrocapsaicin,dihydrocapsaicin, homocapsaicin and homodihydrocapsaicin.1,2Of these, capsaicin and dihydrocapsaicin are the majorcomponents of most capsicum species. Capsaicinoidcompounds have been considered as a major indicator of thechili product qualities. Besides their pungent properties, thecapsaicinoid compounds have also been studied and used formedical and military purposes, such as analgesic creams anddefensive spray.3,4 The first reliable reported measurement ofchili pungency was the Scoville Organoleptic Test.5 Anaccurate determination of the levels of various capsaicinoids hasbecome important because of the increasing demand bycomsumers for foods, and the increasing use inpharmaceuticals.6,7 Similarly, scientists in the area of genetics,biogenesis, food chemistry and physiology, also need reliable,safe, and reproducible standard analytical procedures and therapid methods for the separation and quantitation of thesecapsaicinoid compounds that are useful for comparingpungency levels among different samples. Therefore, theScoville Organoleptic Test has since been replaced byinstrumental methods. The analysis of capsaicinoids has beenconducted by using spectrophotometric,811 gaschromatographic,1216 micellar electrokinetic capillarychromatographic,17 high-performance liquid chromatographicprocedures,1825 and liquid chromatographic-mass spectrometricprocedures.26 Techniques using high-performance liquid

chromatography provide accurate and efficient analysis ofcontent and type of capsaicinoids present in a chili sample.However, it is still necessary to optimize the method for eachchromatographic system in order to identify each of theremaining closely related capsaicinoids in the extract. Theliterature regarding the optimization of this techniques for thedetermination of capsaicinoid compounds of Thai capsicumfruits is still inadequate. Therefore, this study was conducted tooptimize the sample preparation, separation, detection andidentification for Thai capsicum fruits and to achieve aconvenient, rapid and efficient analysis of capsaicinoidcompounds.

Experimental

Plant materialThe matured chili pods with stems removed were dried in a

hot-air oven under 55C for 24 30 h, ground with seeds to passthrough a 60 mesh sieve and stored in sealed plastic bags at 5Cuntil examined. Some of the chili pods used for this study hadbeen grown under farming practices at Lampang AgriculturalResearch and Training Center (LARTC) and some werepurchased from local markets.

ReagentsAcetonitrile (HPLC-grade, J. T. Baker, USA), methanol

(HPLC-grade, Carlo Erba, Italy) and acetone (ACS-grade,Fluka, Switzerland) were used. Deionized, double distilledwater was used throughout. All solvents for thechromatographic system were filtered and degassed using a 0.45m pore size polyamide (nylon) filter.

Capsaicinoid standardsStandards of capsaicin (CAPS, 8-methyl-N-vanillyl-6-

nonenamide), and dihydrocapsaicin (DHC, 8-methyl-N-vanillyl-

661ANALYTICAL SCIENCES JUNE 2002, VOL. 182002 The Japan Society for Analytical Chemistry

Optimization of High-Performance Liquid ChromatographicParameters for the Determination of Capsaicinoid Compounds Using the Simplex Method

Rachaneewan KARNKA,*,** Mongkon RAYANAKORN,* Surasak WATANESK,*and Yuthsak VANEESORN*

*Department of Chemistry, Faculty of Science, Chiang Mai University, Chiang Mai, 50200, Thailand**Lampang Agricultural Research and Training Center, Lampang 52000, Thailand

A high-performance liquid chromatographic method was developed for the analysis of capsaicinoid compounds, thepungent principles of capsicum fruits. A sequential simplex method was applied to optimize the chromatographicresponse function used to assess the quality of separation by varying the chromatographic parameters. The separationwas achieved in 11 min using a C-8 column of 15-cm length and 4.6 mm diameter using a UV detector. A flow rate of1.15 ml min1 at a column temperature of 43.5C using 63.7% methanol in water gave the most efficient separation. Themethod was found to be suitable for the determination of the major capsaicinoid compounds in the capsicum samples.

(Received November 21, 2001; Accepted March 22, 2002)

To whom correspondence should be addressed.E-mail: mongkon@chiangmai.ac.th

nonanamide) were purchased from Sigma (Sigma-Aldrich, St.Louis, MO, USA). The solutions of all standards were preparedin acetonitrile. Appropriate dilutions of the initial solution wereprepared in order to obtain a calibration curve.

ApparatusThe HPLC system consisted of an HPLC Shimadzu pump

10AD, an SPD-M10AV variable wavelength UV detector andCLASS-LC10 Software for data processing.

An Inertsil RP-8 column was used with a controlled-temperature oven.

Octadecyl (C18) 40 m Prep LC packing for solid phaseextraction (SPE) was obtained from J. T. Baker (Phillipsburg,USA).

Optimization of capsaicinoids extraction by ultrasonicationThe extraction efficiencies for the capsaicinoids from chili

peppers in various organic solvents (acetone, acetonitrile andmethanol) were compared, and it was found that acetonitrilegave the highest extraction rate with the fewest impurities. Theoptimum volume of the solvent and the time of sonication werethen determined using only acetonitrile. A 0.3 g sample of chiliwas sonicated for 60 min with 10, 15, 20, 30 and 40 ml ofacetonitrile and another 0.3 g of chili was sonicated for 10, 30,45, 60, 75 and 105 min with 10 ml acetonitrile, respectively.The amounts of the individual extracted capsaicinoids weredetermined and the peak areas were calculated.

Clean-upA C18 Sep-pak (200 mg) was washed with 0.5 ml of double-

distilled water and 0.5 ml of methanol, and then conditionedwith about 0.5 ml of acetonitrile. A 0.5 ml portion of extractwas injected into the conditioned Sep-pak. After thecapsaicinoids were eluted with 0.5 ml of acetonitrile, thecartridge was washed three times with 0.5 ml of acetonitrile andall washed solutions were collected in the same tube. Thesolution was filtered through a 0.45 m filter membrane with asyringe filter into a small glass vial. The obtained filtrate waslater used for an HPLC analysis with each injection volume of 5l.

Chromatographic optimizationAn Inertsil RP-8 column was used with a controlled-

temperature oven. The overall temperature control wasmaintained within 0.5C with a variation of from 26.0C to50.0C. The flow rate used varied from 0.7 to 1.2 ml min1.The mobile phase was a mixture of methanolwater withvarying percentages of methanol from 55% to 70% for the RP-8

column. Percentages below 55% were not used because of theexcessively high column pressure obtained with a flow rate of1.2 ml min1. The detector wavelength was set at 280 nm.

Chromatographic response functionThe separation quality of capsaicinoid compounds for

achieving the maximum resolution with the minimum assaytime was assessed at the end of the chromatogram bycalculating the value of a chromatographic response function(CRF). The CRF is a flexible function that allows desirabletime and resolution criteria to be specified. The correspondingterms in the chromatogram are then compared to these criteriaand the function is maximized by changing the experimentalvariables. It is represented by the following equation:27,28

CRF = Ri + La b|TM TL| c(T0 T1) (1)where Ri is the resolution between adjacent pairs of peaks. Inpractice it is limited to a maximum value of 2.00 so that all pairsof well-resolved peaks make no further contribution to the CRF.L is the total number of peaks detected, TM is an acceptableanalysis time, TL is the retention time of the last eluted peak, T1is the elution time of the first peak, T0 is a specified minimumretention time, and a, b, c are the arbitrary weighting factors (avalue of 1 was used in the this work).

Results and Discussion

Ultrasonic extractionIn this work ultrasonic solvent extraction was used as a simple

and inexpensive method applicable to capsicum samples. Thegoal of the optimization procedure was to improve theextraction efficiency with minimum solvent consumption andminimum time needed for the extraction procedure. Theefficiency of the extraction procedure was verified by the peakarea of the same samples. From the results of the extractionefficiencies of various organic solvents (acetone, acetonitrileand methanol), acetonitrile was used, because it gave areasonably high extraction rate and fewer impurities for thecapsicum studied.

The best extract of capsaicinoids from capsicum samples wasobtained with 10 ml of acetonitrile in one extraction step for 60min with a controlled column temperature of 43.0 0.5C. Theresults are shown in Figs. 1 and 2. The reproducibility of theultrasonic extraction of 0.3 g-capsicum samples with 10 ml ofacetonitrile for 60 min sonication was evaluated using 10consecutive analyses. The reproducibility of the peak areas ofcapsaicin and dihydrocapsaicin was found to be satisfactory

662 ANALYTICAL SCIENCES JUNE 2002, VOL. 18

Fig. 1 Peak areas of capsaicinoids as a function of the volume ofacetonitrile.

Fig. 2 Peak areas of capsaicinoids as a function of the time ofsonication.

with 1.05% RSD (n = 10) and 1.08% RSD (n = 10),respectively.

The recovery of clean-up columnThere was almost a complete elimination of interferents when

SPE-C18 was used with acetonitrile as the eluent. The worstinterferents, especially pigments, were reduced. The percentrecovery for capsaicin was found to be 98.91 and fordihydrocapsaicin it was 99.23 with standard deviations of 0.09and 0.19, respectively.

Chromatographic optimizationThe chromatographic parameters were optimized using a

chemometric approach based on the use of the simplexmethod.27,28 When using a simplex, each vertex corresponds to

a set of experimental conditions. In this work, the factors whichwere varied to improve the separation of capsaicinoidcompounds included the mobile-phase composition, the flowrate and the column temperature. From the resultingchromatogram under each set of conditions the chromatographicresponse function (CRF) was calculated and the relativeresponses were ranked. The advantage of the CRF for thesimplex method is that it allows a weighting of importantchromatographic features (Ri and TM) for simplex movementsthat result in a higher CRF value with increased resolution,reasonably short analysis time (TM TL) and good retention (T1> T0). The modified simplex was started after introducing upperand lower boundary conditions for the above three variables. Inthis process, the CRF value is calculated for m sets of startingconditions, where m is the number of factors to be optimizedplus 1. In this case m is 4. The corresponding initialexperimental conditions and CRF values are given in Table 1.

The point corresponding to the lowest value of CRF was thenreflected about the surface (hyperface) defined by the remainingthree points to give a fifth set of conditions to evaluate and rank.Expansion and contraction for the simplex was allowed, basedon the usual rules.2729 To select whether expansion, contractionor keeping the reflection steps, by using the differences in theresponses at the vertices to estimate the vertex with a betterresponse when the vertex violated a boundary condition on oneof the experimental factors.30 The next vertex and process wererepeated sequentially until an apparent optimum had beenobtained.

The results of the sequential simplex progress are given inTable 2. The simplex was halted after 30 experiments, sincethere was no further significant improvement towards themaximization of the CRF value after vertex 21. Figure 3 showsthe variation in CRF with the experiment number; it can be seenthat an optimum response was achieved rapidly. Althoughexperiment number 7 had the highest CRF value, because nearlythe same set of conditions reappeared near vertices 22 30, theearly high CRF value was considered to be fortuitous. Thisreturn to the optimum value increases ones confidence that themethod is rugged and efficient. Chromatograms obtained underoptimum conditions selected as vertex number 7 are presentedin Fig. 4, showing excellent resolution among the three expectedpeaks. The data indicate that the variation of the columntemperature has a more significant effect on the CRF value thando the other two parameters.

663ANALYTICAL SCIENCES JUNE 2002, VOL. 18

55.0 70.0 70.0 60.0 60.0 55.0

0.70 1.20 1.00 1.00 0.70 1.2026.0 50.0 31.0 26.0 43.0 50.0

Table 1 Boundary conditions of the experimental parameters and the initial conditions used for simplex optimization with the C8 column

Boundary condition Minimum MaximumExperiment No.

1 2 3 4Mobile phase

composition/% methanol in water

Flow rate/ml min1Temperature/C

1 B 70.0 1.00 31.0 3.9812 W 60.0 1.00 26.0 3.7093 N 60.0 0.70 43.0 0.7024 N 55.0 1.20 50.0 0.9715 CR 62.5 0.95 48.9 4.5186 CW 60.9 1.10 33.7 0.9377 CR 63.7 1.15 43.5 5.9708 CW 61.3 0.88 43.1 2.2399 CW 60.2 1.12 45.5 3.92210 CW 66.1 1.03 38.5 4.17011 R 68.0 0.98 41.8 3.84512 CW 62.2 1.07 44.6 4.76213 CR 61.2 1.07 49.3 4.56214 CW 64.5 1.05 42.1 4.78015 CR 64.0 1.20 40.7 5.52716 CW 63.0 1.00 46.2 5.16817 CR 65.0 1.19 40.9 5.77718 CW 63.1 1.11 43.4 5.88719 CW 64.0 1.11 42.3 5.82120 R 63.3 1.08 45.4 5.48721 CW 63.8 1.17 41.9 5.71622 CR 63.3 1.18 43.2 5.90623 CW 63.8 1.13 42.6 5.80524 R 63.0 1.16 44.8 5.85925 CR 63.2 1.16 44.1 5.89526 CW 63.3 1.15 43.5 5.89427 R 63.7 1.20 42.7 5.96528 CW 63.4 1.17 43.3 5.90729 R 63.9 1.20 43.2 5.93430 CW 63.6 1.18 43.2 5.960

Table 2 Relationship between chromatographic response function and vertex number during simplex optimization with the C8 column

VertexNo.a % MeOH

Flow rate/ml min1

Column temperature/C CRF

a. B, best; N, next to the worst; W, worst; R, reflection; CR, contraction on the R side; CW, contraction on the W side.

Fig. 3 Relationship between the chromatographic responsefunction (CRF) and the experiment number during simplexoptimization with the C8 column.

CalibrationThe determination of the two capsaicinoids in the fruit

extracts was performed using the external standard method.The calibration graphs were expressed as chromatographic peakareas of standard capsaicinoids versus correspondingconcentrations of the standards in the concentration range of 1 100 mg l1. The definition of the limit of detection31,32 used hereis the concentration corresponding to a signal of the blank(calculated from the extrapolation of the regression line of thedata rather than from separate measurements) plus threestandard deviations of the noise, assuming a normallydistributed variation around the regression line derived from theactual data over the concentration range studied. Although thesignals for 1 mg l1 samples were measured in this work, thelimits of detection for capsaicin and dihydrocapsaicin werefound to be 1.65 and 1.87 mg l1, respectively, using the abovedefinition. The regression lines, correlation coefficients, limitsof detection and limits of quantitation are summarized in Table3.

ReproducibilityThe reproducibility was evaluated by 8 consecutive analyses

with both capsaicin and dihydrocapsaicin in a standard solution.The relative standard deviations of the retention time forcapsaicin and dihydrocapsaicin were excellent with a meanretention time of 7.76 min, 1.26% RSD for capsaicin and 1.32%RSD with a mean retention time of 10.46 min fordihydrocapsaicin. The instrument repeatability data for thecorrected area calculation for a standard solution of aconcentration of 50 mg l1 for capsaicin and 50 mg l1 fordihydrocapsaicin were satisfactory with the mean of thecalculated concentration being 48.79 mg l1 with 1.58% RSD forcapsaicin and 3.49% RSD with a mean calculated concentrationof 51.60 mg l1 for dihydrocapsaicin.

Determination of capsaicinoids in capsicum extractionsThe optimal conditions were determined for the standard

solution. A peak just before capsaicin and afterdihydrocapsaicin appeared to be impurities contained incapsaicin and dihydrocapsaicin, and probably represented minorcapsaicinoids, such as nordihydrocapsaicin, homocapsaicin andhomodihydrocapsaicin. The peak just before capsaicin is likelyto be nordihydrocapsaicin, as reported in another study.33

Using the optimum condition, the results were obtained forindividual capsaicinoid peaks of capsicum species tested,

including capsicum samples collected from farming practices atLampang Agricultural Research and Training Centre (LARTC)and some from local markets, as shown in Table 4. As for therecovery rate of the column clean-up, it was found to be 93.3%0.6 (n = 3) and 89.6% 0.2 (n = 3) for capsaicin anddihydrocapsaicin, respectively.

Conclusion

The determination of the optimum conditions for samplepreparation and capsaicinoid extraction was relativelystraightforward. Optimization of the column temperature, theflow rate and the mobile phase composition to achieve goodanalytical separation was achieved rapidly using the simplexmethod. In this study, it was found that the effect of changingthe column temperature was more important than changes in themobile phase or flow rate. The separation of compounds with asimilar structure appears to be particularly sensitive totemperature changes. In summary, the optimumchromatographic separation of capsaicinoid compounds withgood resolution in a short time was accomplished using thesimplex method. It has proved to be a useful tool fordeveloping the analysis method.

664 ANALYTICAL SCIENCES JUNE 2002, VOL. 18

Fig. 4 Chromatogram of capsaicinoids obtained with the C8column using the optimum conditions (1, nordihydrocapsaicin; 2,capsaicin; 3, dihydrocapsaicin).

Table 3 Calibration range, regression lines, correlation coefficients, limits of detection and limits of quantitation for capsaicinoids studied

Parameter Capsaicin Dihydrocapsaicin1 100 1 100

Y = 2374.6X + 953.82 Y = 1969.3X + 1066.90.9998 0.9998

1.65 1.87

5.49 6.23

Calibration range/mg l1

Regression lineCorrelation

coefficientLimit of

detection/mg l1aLimit of

quantitation/mg1b

a. yL = yB + 3sB.31,32b. yL = yB + 10sB.31,32

Table 4 Various capsicum samples and their contents of capsaicin and dihydrocapsaicin

Capsicumsamplea

Capsaicin content Dihydrocapsaicin content

%g g1 SDb %g g1 SDb

C1 0.0209 0.0002 0.0170 0.0010C2 0.0362 0.0010 0.0437 0.0003C3 0.0146 0.0001 0.0125 0.0008C4 0.0515 0.0012 0.0510 0.0012C5 0.0214 0.0004 0.0228 0.0007C6 0.0244 0.0005 0.0285 0.0007C7 0.3289 0.0082 0.2439 0.0057C8 0.5038 0.0020 0.4641 0.0008C9 0.6985 0.0156 0.4075 0.0072C10 0.5398 0.0007 0.3251 0.0013

a. C1 C5, from LARTC farming practices; C6 C10, from local markets.b. SD: standard deviation, n = 3.

Acknowledgements

The authors gratefully acknowledge Professors S. N. Demingand R. L. Deming for their initial suggestion concerning anapproach to the simplex method of optimization and thePostgraduate Education and Research Program in Chemistry(PERCH) of Thailand and the Graduate School of Chiang MaiUniversity for their partial support.

References

1. K. Lee, T. Suzuki, M. Kobashi, K. Hasegawa, and K. Iwai,J. Chromatogr., 1976, 123, 119.

2. K. Iwai, T. Suzuki, and H. Fujiwake, J. Chromatogr., 1979,172, 303.

3. M. W. Law, J. Assoc. Off. Anal. Chem., 1983, 63, 1304.4. A. Szallasi and P. M. Blumberg, Pain, 1996, 68, 195.5. V. Govindarajan, S. Narasimhan, and S. Dhanaraj, J. Food

Sci. Technol., 1977, 14, 28.6. Y. J. Surh and S. S. Lee, Life Sci., 1995, 56, 1845.7. Y. J. Surh, S. H. Ahn, K. C. Kim, J. B. Park, Y. W. Sohn,

and S. S. Lee, Life Sci., 1995, 56, 305.8. A. S. L. Tirimanna, Analyst, 1972, 97, 372.9. A. Trejo-Gonzalez and C. Wild-Altamirano, J. Food Sci.,

1973, 38, 342.10. K. L. Bajaj, J. Assoc. Off. Anal. Chem., 1980, 63, 1314.11. T. Anan, H. Ito, H. Matsunaga, and S. Monma, Caps.

Eggplant. Newslett., 1996, 15, 51.12. J. J. DiCecco, J. Assoc. Off. Anal. Chem., 1976, 59, 1.13. B. V. Thomas, A. A. Schreiber, and C. P. Weisskopf, J.

Agric. Food Chem., 1998, 46, 2655.14. J. Jurenitsch and R. Leinmuller, J. Chromatogr., 1980, 189,

389.15. W. S. Hawer, J. Ha, J. Hwang, and Y. Nam, Food Chem.,

1994, 49, 99.16. A. M. Krajewska and J. Powers, J. Assoc. Off. Anal. Chem.,

1987, 70, 926.17. A. Laskaridou-Monnerville, J. Chromatogr. A, 1999, 838,

293.18. J. E. Woodbury, J. Assoc. Off. Anal. Chem., 1980, 63, 556.19. P. G. Hoffman, M. C. Lego, and W. G. Galetto, J. Agric.

Food Chem., 1983, 31, 1326.20. K. M. Weaver, R. G. Luker, and M. E. Neale, J.

Chromatogr., 1984, 301, 288.21. T. H. Cooper, J. A. Guzinski, and C. Fisher, J. Agric. Food

Chem., 1991, 39, 2253.22. T. S. Johnson, G. A. Ravishankar, and L. V. Venkataraman,

J. Agric. Food Chem., 1992, 40, 2461.23. M. D. Collins, L. M. Wasmund, and P. W. Bosland, Hort.

Science., 1995, 30, 137.24. M. Parish, J. Assoc. Off. Anal. Chem., 1996, 79, 738.25. J. Lu and M. Cwik, J. Chromatogr. B, 1997, 701, 135.26. C. A. Reilly, D. J. Crouch, G. S. Yost, and A. A. Fatah, J.

Chromatogr. A, 1984, 316, 69.27. J. C. Berridge and E. G. Morrissey, J. Chromatogr., 1984,

316, 69.28. J. C. Berridge, Analyst, 1984, 109, 291.29. F. H. Walters, L. K. Parker, Jr., S. L. Morgan, and S. N.

Deming, Sequential Simplex Optimization: A Techniquefor Improving Quality and Productivity in Research,Development and Manufacturing, 1991, CRC Press, BocaRaton.

30. S. M Sultan and A. H. El-Mubararak, Talanta, 1996, 43,569.

31. J. C. Miller and J. N. Miller, Statistics for AnalyticalChemistry, 1993, Ellis Horwood, New York.

32. J. N. Miller, Analyst, 1991, 116, 3.33. V. K. Attuquayefio and K. A. Buckle, J. Agric. Food

Chem., 1987, 35, 777.

665ANALYTICAL SCIENCES JUNE 2002, VOL. 18