Normal phase liquid chromatographyThe most promising approach to substitute the referencemethod by an automated analysis deviates as little aspossible from the original principle. Aitzetmüller pro-posed frontal elution liquid chromatography  using amoving wire detector which is no longer commerciallyavailable. He not only reported one of the earliest at-tempts to automate liquid chromatography (in 1973) butalso gave a still valid discussion about the unique char-acter of the normal phase liquid chromatography on sili-ca gels.Chromatography on silica is based on both partition-ing and adsorption. A water-free silica surface containsactive sites and primarily separates by adsorption .If the activity of silica is reduced by water or ethanol,chromatography occurs by partitioning. As a conse-quence, the control of the water content of silica is es-sential. Silica adsorbs water from the eluent or desorbswater into a dry eluent until a steady state is reached.These changes cause drifting retention times . If agradient is employed, a steady-state situation mightnever be reached. The concept of isohydric solvents wasintroduced [15, 16], permitting polarity gradients with-out affecting the water concentration of the stationaryphase. However, isohydric eluents require the careful ad- justment of the water concentration of the two eluents.Mostly a dry and a saturated solvent are mixed.Detection techniqueHow can gravimetric quantification be replaced by achromatographic detection? UV response cannot be usedas it strongly differs for the compounds of interest. RIdetectors approach mass-sensitive measurement moreclosely, but can only be employed for isocratic runs.Hence the peak of the polar compounds (requiring gradi-ent elution) cannot be quantified. ELSD represents an-other mass-sensitive type of detector. Its main featuresare excellent gradient capability, ease of operation, andruggedness. However, because of the different underly-ing light scattering mechanisms, ELSDs do not producelinear response .This paper presents an HPLC method which mimicsthe reference method for the determination of the polarcompounds  as closely as possible. The main advanta-ges are speed, low labor-requirement, low solvent con-sumption, and improved repeatability. The validation of the method as well as the comparison with the referencemethod for polar compounds and the method measuringthe dielectric constant (FOS) will be presented in asecond paper .
Materials and methods
. The HPLC instrument consisted of a gradient pump(Gynkotek P 580A NDG/Agilent HP1100 quaternary), with anautosampler (Gynkotek Gina 50), a column thermostat (Gynkotek STH 585), and a light scattering detector (Polymer LaboratoriesPL-ELS 1000).The column flow rate was 2.0ml/min, while the column temper-ature was held at 10°C. Injection volume was 20µl. The gradientprogram consisted of 0.0–0.5min 0% B, 0.5–1.9min 0–100%B,1.9–2.4min 100% B, 2.4–2.9min 100–0%B, 2.9-x min 0% B. Theconditioning time (x) required depends on the dead volume of theHPLC pump and might vary between 4 and 9min. The dead vol-ume was determined by injecting the reference oil while maintain-ing 100% eluent B. The conditioning time was increased stepwiseuntil the first peak of the reference oil (non-polar compounds) elut-ed between k
0.6 and 1.0. Three consecutive injections were madefor any evaluated conditioning time in order to ensure that stableperformance was reached. A longer conditioning prolongs the k
value of the first eluting peak while the k
value of the second peak (polar compounds) remains constant.The ELSD spray temperature was set to 40°C and the evapo-ration temperature to 90°C. A nitrogen gas flow of 1l/min wasused.
Chemicals and materials
. Hexane 96% (multisolvent, HPLCgrade, Scharlau, Barcelona), ethanol, abs., HPLC quality, MächlerAG, Reinach, Switzerland), bidistilled water, diethyl ether (Ph.H.V.Siegfried, Zofingen, Switzerland). A reference oil (25–30% polarcompounds) was used as a standard.A 30
4.6mm i.d., 10µm particle size cyano Nucleosil HPLCcolumn, (Macherey Nagel) was used for the separation. A peek tubing (3.0m
0.5mm i.d.) was installed between the column andthe detector.
. The extraction solvent consisted of hexane:diethylether, 10:3 (by volume), the mobile phase A of hexane, and mo-bile phase B of hexane:ethanol:water, 50:50:1 (by volume). Botheluents were stored in brown bottles.
. Stock solution: 0.625±0.02g reference oil warmed to40–50°C was weighed in a graduated flask. The latter was filledwith extraction solvent to 50ml. Reference oil was produced byprolonged heating of a frying fat in an open, air-exposed vesseluntil FOS measurement indicated values corresponding to 25–30%polar compounds. The exact concentration of polar compoundswas determined by the reference method (six independent determi-nations, each with an individually deactivated silica batch per-formed by three different operators).Six standards solutions (A, B, C, D, E, F) were prepared bytransferring 1, 3, 6, 8, 10, and 12ml stock solution into 50-mlgraduated flasks and filling up to the mark with hexane. In a re-frigerator solutions are stable for more than three months.
Preparation of samples
. Warmed oil (15 drops, 40–50°C) weredissolved in 20ml extraction solvent. Then 0.4ml of this solutionwas diluted in the HPLC sample vial with 1.2ml hexane and thevial closed with a PTFE only or a Viton seal.
. Peak areas of the polar and non-polar compounds arecalibrated by a quadratic, non-zero offset calibration curve. Thecalibration included the concentration of the polar and non polarcompounds. For example, if 0.625g reference oil of 30.4% polarcompound is used, standard A contains 0.174g/l non-polar and0.076g/l polar compounds. The percentage of polar compounds ina sample was calculated as polar compounds (%)=100
amountpolar/(amount non-polar+amount polar).
Figure1 shows a chromatogram of a heavily used (top)and a fresh frying oil. Two well separated peaks wereobtained in less than 2.5min.