ANALYST, JUNE 1986, VOL. 111 601 Linking Low Dispersion Liquid Chromatography with Diode-array Detection for the Sensitive and Selective Determination of Vitamins A, D and E* Mary Mulholland Pye Unicam, York Street, Cambridge CBI ZPX, UK A low dispersion LC system using narrow-bore column technology was linked to a diode-array detector and applied to the determination of vitamins. The LC system had been evaluated for dispersion effects. The technique was robust and reliable and proved to be extremely sensitive and selective for the determination of vitamins A, D and E. Vitamin E was assayed in various vegetable oils and the percentage distribution examined. The high mass sensitivity was demonstrated with an on-column loading of 10 ng of vitamin E, which could be clearly detected and identified.Keywords: Vitamin A, D2, D3 and E determination; narrow-bore column technology; diode-arra y detection; high-performance liquid chromatography; vegetable oils analysis Reducing the size of chromatographic columns results in a decrease in the elution volume required for a given separation. Two major advantages emerge from this. Firstly, as the sample dilution is less, the sensitivity is increased. A compar- able peak size will be obtained for a 1-p1 injection of a given concentration on a 10 cm X 2.1 mm i.d. column as with a 5-1.11 injection volume on a 10 cm x 4.6 mm conventional column, both systems having equivocal flow-rates. This advantage is therefore only relevant when sample volumes are restrictive. Secondly, the amount of solvent required per analysis will be greatly reduced.Worthwhile savings can be achieved when vitamins are assayed in high sample throughput situations. In any chromatographic system the over-all efficiency achieved will be reduced owing to extra-column dispersion. This effect on the measured column effiency, N,, is shown in the following relationship: Nm = (1 - 02) where N is the true column efficiency and 0 represents the relative external contribution to band broadening. As resol- ution is proportional to the square root of the plate number, the relative loss of resolution will be W2.l Accepting, as a guideline, that the maximum acceptable loss in resolution owing to extra-column dispersion is 10%,2 a specification for equipment requirements for a given column size can be drawn UP.Conventional HPLC methods used previously were exam- ined in order to select column types and sizes that would be suitable for the determination of vitamins.3--S After making a suitable choice of column size, the instrumentation could then be optimised to fit the low dispersion requirements. A 1-p1 injection rotor, a specially designed 1-pl flow cell and minimal external tubing were therefore selected. This optimised system was evaluated for compatibility with the selected column geometry.6 The compatibility study consisted of a comparison of the column peak standard deviation with the measured external peak standard deviation to calculate the expected resolution loss? The results show that for capacity factors greater than 1 the instrumentation is fully compatible with the selected column geometry.The determination of fat-soluble vitamins presents many detection problems.8 They have differing UV absorbance maxima and are often accompanied by an excess of com- pounds of similar physical and chemical properties that can interfere with their determination. The choice of diode-array * Presented at the 30th IUPAC Congress, Manchester, UK, September &13th, 1985. detection allows each vitamin to be assayed at its wavelength of maximum adsorption (Amax.) without the loss of selectivity provided by conventional UV detection. Multi-channel UV - visible detection is based on a linear array of light-sensitive elements (diodes) etched on to a silicon chip. These diodes operate in parallel to monitor simul- taneously the absorbance over a given wavelength range.The detector thus acquires three-dimensional data of absorbance versus wavelength versus time, which, in turn, provides both qualitative and quantitative information. The peak purity can also be established by a comparison of spectra taken throughout the peak elution. The instrumen- tation employed for this work can be simply converted for use with conventional HPLC columns. Previous applications of micro-columns with diode-array detection have involved the use of specialised equipment dedicated to this column geometry .9 This paper investigates the applicability of linking narrow- bore column technology to diode-array detection for the determination of vitamins A, D and E. Experimental Materials The vitamins for use as reference materials were obtained from Sigma (St Louis, MO, USA).All solvents were of HPLC grade. (Fisons, Loughborough, UK). Instrumentation The LC system consisted of a PU4015 pump with a PU4021 multi-channel UV - visible detector equipped with a 1-p1 flow cell and a Rheodyne 7520 injection valve with a 1-p1 rotor. The detector output was connected to a PU4850 data station having two additional 128K memory boards. (All components were from Pye Unicam, Cambridge, UK.) Chromatographic Conditions Sample manipulations were performed in the absence of oxygen, direct sunlight or the light of fluorescent tubes in order to avoid degradation of the vitamins. All data were obtained as chromascans from which spectra and chromato- grams could be taken using the PU4850 data station.602 ANALYST, JUNE 1986, VOL.111 ( a ) 325.8 Ib) 110.8 0.2 a.u.f.s I 462.1 ( C) 462.1 0.2 a.u.f.s 1 Wavelength- Fig. 1. 300 nm; and ( c ) , retinol at 330 nm Chromatogram taken at the wavelength of maximum absorption of each vitamin. ( a ) , Ergocalciferol at 270 nm; ( b ) , a-tocopherol at Normal-phase Chromatography An HPLC method was developed, which resolves retinol, a-tocopherol and ergocalciferol (vitamins A, E and D, respectively) using a 3 cm x 2.1 mm i.d. guard and a 10 cm x 2.1 mm i.d. Spheri-5 silica cartridge with a mobile phase of hexane - isopropyl alcohol (99.5 + 0.5 V/V) at a flow-rate of 0.4 ml min-1. This method was altered by changing to a mobile phase of hexane - isopropyl alcohol (99.7 + 0.3 V/V) to resolve a-, p- and y-tocopherol in vegetable oils.All reference materials were dissolved in hexane and the vegetable oils were injected directly. Reversed-phase chromatography To resolve ergocaliferol and cholecalciferol (vitamins D2 and D3, respectively) a method was developed using a 3 cm X 2.1 mm i.d. guard and a 10 cm x 2.1 mm i.d. Spheri-5 RP18 cartridge with a mobile phase of methanol - water (95 + 5 V/V) at a flow-rate of 0.4 ml min-1. All reference materials were dissolved in methanol. The HPLC columns used were supplied by Pye Unicam. Results and Discussion A solution containing 0.5 mg ml-1 each of retinol, cw-toco- pherol and ergocalciferol reference materials was chromato- graphed under the conditions described. Three chromato- grams were taken at the individual wavelength of maximum absorption for each vitamin and these are shown in Fig.1. Each vitamin could therefore be assayed at its maximum sensitivity. The UV spectra illustrated in Fig. 2 were obtained at the peak apex for each vitamin. These spectra are easily distinguished , each having distinct features. The method, therefore, demonstrates good selectiv- ity. Relative standard deviations for the retention time and peak area of each vitamin obtained from ten injections were below 1%. However, care was taken to avoid degradation of the vitamins in solution by using sealed, dark glassware. Vegetable oils contain tocopherols in the unesterified form and can thus be injected on to the column directly without any pre-treatment. Fig. 3 demonstrates the resolution that was achieved between a-, p- and &tocopherol.As each tocopherol has a different degree of biological activity it is necessary to observe their percentage distribution. The results for sun- flower oil and soya oil in Table 1 show sunflower oil containing predominantly a-tocopherol whereas y-tocopherol is the largest component in soya oil. Although there was some 0.2 0.1 0 1.2 t Q) C m e a $ 0.6 0 0.1 0.05 1 I I I I 210 2 50 290 330 370 Wavelengthinm Fig. 2. Fig. 1 at each respective peak apex Spectra of the vitamins taken from the chromatograms of interference with a-tocopherol, owing to co-eluting materials, this did not greatly affect quantitation. The normalisation and subtraction of the spectrum obtained for the co-eluting material from the spectrum at the peak apex for a-tocopherol shows that the resultant spectrum can be clearly identified (Fig.4).ANALYST, JUNE 1986, VOL. 11 1 603 The features in the UV spectrum of vitamin E vary in magnitude. A graph plotting the logarithm of absorbance will therefore enable the smaller features to be examined more clearly. To examine the spectral purity of y-tocopherol in soya oil the logarithm of absorbance plots of spectra taken throughout the peak elution were compared with a plot for the Table 1. Distribution (YO) of vitamin E Total vitamin El Yo Peak 1, Peak 3, Peak 4, Sample a-tocopherol @-tocopherol y-tocopherol Sunfloweroil . . 93.1 2.2 4.6 Soyaoil . . . . 8.9 8.9 81.6 E l z != 0.02 a.u.f.s. X I I Wavelength - spectra of a-tocopherol reference material. Reference material for a-tocopherol is only available as soya oil and, as the spectrum for y-tocopherol is indistinguishable from that for a-tocopherol, it was decided to use the spectrum obtained for a-tocopherol for which the purity was assured. Examina- tion of the graphs shown in Fig.5 reveals a small shoulder at 230-250 nm. This is due to the fats present in soya oil, which leach slowly off the column. As the assay is carried out at 300 nm this will not affect quantitation. A reproducibility study of five injections of each oil gave relative standard deviations for retention behaviour and peak areas of below 2% for each isomer. Very low levels of detection and identification were achieved using this technique. This is demonstrated by Fig. 6, which shows the chromatogram and spectrum for a 10 ng on-column loading of vitamin E acetate.Ergocalciferol and cholecalciferol were resolved using a reversed-phase HPLC method. The chromascan, chromatogram and spectra for a 100-pg on-column loading of these vitamins are illustrated in 210 250 290 330 370 210 250 290 330 370 Wavelengthmm Fig. 3. Chromatograms of vitamin E in (a) sunflower oil and (b) soya oil. Peaks: 1, a-tocopherol; 2, @-tocopherol; and 3, y-tocopherol Fig. 5. Log absorbance plot of (a) a-tocopherol reference material; (6) y-tocopherol in soya oil taken on the peak slope upwards; (c) y-tocopherol in soya oil taken at the peak apex; and ( d ) y-tocopherol in soya oil taken on the peak slope downwards 210 250 290 330 370 Wavelengthlnm Fig. 4. UV spectrum of (a) a-tocopherol in sunflower oil; (6) material co-eluting with a-tocopherol in sunflower oil; and (c) after normalisation and subtraction of ( b ) from ( u ) 0.002 3.u.f.s ._.89.3 ~ 210 250 290 330 370 Wave leng t h/n m Wavelength - Fig. 6 . using a 10-ng on-column loading (a) Chromatogram and ( b ) spectrum of a-tocopherol acetate604 ANALYST, JUNE 1986, VOL. 111 1 I 2 a 9 330 360 390 300 Ti me/s Time I I I I 210 250 290 330 370 210 250 290 330 370 Wavelengthhrn Fig. 7. (a) Chromascan and ( b ) chromatogram of 1, cholecalciferol and 2, ergocalciferol reference materials. (c) Spectrum of ergocalci- ferol and ( d ) spectrum of cholecalciferol reference material Fig. 7. The method demonstrated good selectivity between the two vitamins and relative standard deviations (from five injections) for the retention time and peak area were below 1% for each vitamin.In order to demonstrate the sensitivity, Fig. 8 shows a chromatogram and spectrum obtained for a 10-ng on-column loading of ergocalciferol. Throughout this work the retention behaviour was reproducible and no column deterioration was observed, demonstrating the robustness of the columns selected. Conclusions The technique of linking narrow-bore column technology to diode-array detection was successfully applied to the determi- nation of vitamins. The use of reduced-size columns enabled exceptionally high mass sensitivities to be achieved. These columns were robust and could be used extensively without any deterioration in performance. Chromatographic peak shapes were good, demonstrating the low dispersion effects of the system.Diode-array detection proved to be extremely versatile as data, obtained from a single injection, could be extensively manipulated to give a large amount of information. The risks of misinterpreted data are minimised as chromatographic peaks can be clearly identified and their spectral purity established. These features are pertinent to the analysis of vitamins, which are often found in very complex sample matrices. The combination of narrow-bore LC and diode-array detection proved to be sensitive and selective for both the qualitative and quantitative determination of vitamins A, D and E. 339.8 I Time --t 210 250 290 330 370 Wavelengthhrn - Fig. 8. 10-ng on-column loading (a) Chromatogram and (b) spectrum of ergocalciferol using a 1. 2. 3. 4. 5. 6. 7. 8. 9. References Kucera, P., Editor, “Micro-column High Performance Liquid Chromatography,” Elsevier, Amsterdam, 1984, pp. 1-36. Bristow, P. A , , “Liquid Chromatography in Practice,” HETP, Van Neikerk, P. J., and du Plessis, L. M., J. Chromatogr., 1980, 187, 436. Abe, K., Yuguchi, Y., and Katsui, G., J . Nutr. Sci. Vitaminol., 1979, 25, 67. Van Neikerk, P. J., and Smit, S. C. C., J. Am. Oil Chem. SOC., 1980, 57, 417. Naish, P. J., Goulder, D. P., and Perkins, C. V., Chromato- graphia, 1985, 20, 335. Hupe, K. P . , Jonker, R. J., and Rozing, G., J. Chrornatogr., 1984,285, 253. Macrae, R., Editor, “HPLC in Food Analysis,” Academic Press, London, 1982, Chapter 8, pp. 187-205. Takeuchi, T., and Ishii, D., J. Chromatogr., 1984, 288, 451. UK, 1976, pp. 239-242. Paper A51400 Received November 4th, 1985 Accepted January 8th, 1986