520 YOUNG: THE DETERMINATION OF DIHYDROXYBENZENES [Vol. 86 The Determination Liquid of Dihydroxybenzenes by Par ti tion Chromatography Liquid - BY JOHN H. YOUNG [ T h Midland Tar Distillers Limited, Research and Development Department, Four Ashes, nr. Wolverhamnpton) Liquid - liquid partition chromatography is slow, largely because of the time involved in identifying and determining components in the eluate. With the apparatus described, most of these time-consuming operations are avoided by continuously recording the percentage of ultra-violet light of fixed wavelength transmitted by the eluate. By this means, the emergence from the column of separated components is indicated by the appearance of peaks on the recorder chart. Some components can be identified and deter- mined directly from the trace by the relative positions of and areas under the peaks.For others, an automatic fraction collector is used, and the trace serves as a guide for blending fractions for subsequent ultra-violet analysis. The method has been applied mainly, but not exclusively, to the determination of dihydric phenols in aqueous and organic solutions. THE phenolic fraction extracted from coal-tar distillates boiling over the range 230" to 270" C is essentially a complex mixture of monohydroxy aromatics containing minor amounts of dihydroxybenzenes. As the latter may have some bearing on the properties and utilisation of such a fraction, a method was required for determining individual dihydroxybenzenes in the presence of much larger amounts of monohydric phenols. Such a method would also be of value in classifying aqueous phenolic effluents and in evaluating procedures for treating them; for this purpose, the method should be applicable to a dilute aqueous solution of the sample.Chromatographic procedures were clearly suggested, and liquid - liquid partition was preferred to paper chromatography because it could more easily be made quantitative. Such a method had already been successfully devisedl and appeared to be suitable for our purpose. The chief disadvantages of the column procedure were the slowness of operation and the time required for identifying and determining the eluted components. In view of this, a semi-automatic apparatus was designed to work without attention overnight and to record a trace showing the eluted components.The amounts of individual dihydric phenols could then be determined after blending fractions (the trace being used as a guide) or some- times directly from the trace. Although the apparatus has been constructed for determining dihydric phenols, it can also be used for other chromatographic determinations, with a suitable choice of phases, provided that a wavelength can be found in the visible or ultra-violet region at which the solvent transmits light appreciably more than do the components to be measured. DESCRIFTION OF APPARATUS SOLVENT HANDLING- The eluting agent (cyclohexane) is supplied from a 1-litre flask, A, equipped with a bottom off-take leading to a tap and a male spherical joint (14 mm diameter). When gradient elution is used, the more polar solution (20 per cent.v/v of n-butyl alcohol in cyclohexane) is contained in a second 1-litre flask, B, and is continuously supplied to flask A via siphon tube C (2mm i.d.). A stirrer in flask A ensures mixing, and a side-tube is provided to prime the siphon by slight air pressure. I t has been found that the siphon gives a more linear change in concentration with volume of eluting solution than does the more conventional arrangement of a separating funnel mounted on the flask so that the flask remains full. When analysing phenolic mixtures it is often desirable to use only pure cyclohexane as eluting agent during the early stages of elution. A delay before the siphon begins to operate can be achieved by fitting a non-return valve of the type shown at D in Fig.1. A B10 air leak is cut off close to the joint, and a 3/16-inch ball-bearing is ground in with emery powder. Final grinding is done with very fine emery, a second identical ball being used, The device for supplying the eluting agent is shown in Fig. 1.August, 19611 BY LIQUID - LIQUID PARTITION CHROMATOGRAPHY 521 4 f Fig. 1. Details of supply of eluting solution - Spherical joint ( I 4 mm) U Teflon tube (2.5 mm i.d.) Fig. 2. Details of pre-column522 YOUNG : THE DETERMINATION OF DIHXDROXYBENZENES [Vol. 86 and this ball is used in the completed valve. The valve opens only under a positive hydrostatic head of about 40mm of the butyl alcohol-cyclohexane solution, so that the delay before the gradient is applied can be adjusted by varying the relative heights of the flasks and the liquids in them.It is important that eluting agent passing to the column is saturated with the stationary phase (water) at the temperature of the column, otherwise each component may give rise to two partly resolved peaks. When an eluting solution of changing composition is used, water is added to the more polar solution to the extent of about two-thirds of the amount required for saturation. Final adjustments to the water content are then made by passing the eluting agent through a small pre-column (see Fig. 2) packed with the same stationary phase as the analytical column. The eluting solution attains equilibrium by accepting or losing water in this pre-column. INTRODUCTION OF SAMPLES- Aqueous sam$Zes-Portions (0.5 to Eiml) are acidified and then mixed with twice their The resulting free-flowing powder is put into an weight of dry 60- to 80-mesh silica gel.adapter (see Fig. 3) that fits on to the top of the column. Spherical joint (14 mrn)--\ 4 024 B24 NO. I s i n t e r - d Spherical ,'joint (14 mm) Fig. 3. Adapter for aqueous samples Fig. 4. Sample injector Non-aqueous sampEes-Portions (0.5 ml) of samples dissolved in a cyclohexane - butyl alcohol mixture are introduced into the side-limb of the injection device shown in Fig. 4. A three-way tap is set to by-pass the sample limb until the flow has been adjusted. The tap is then turned, and the flow of eluting agent is diverted to wash the sample on to the column. PREPARATION OF COLUMN- The chromatographic tube (see Fig.5) has an internal diameter of about 14mm and is 640mm long. It is fitted at the top with a ground-glass B24 socket and at the bottom with a B10 cone having a No. 1 sinter below it. This position for the sinter decreases the space in which fractions of eluate can diffuse. For the analysis of dihydric phenols, the tube is packed to a depth of 570 mm with 55 per cent. w/w of water on acid-washed 60- to SO-mesh silica gel. This system is based on the original work of Blackburn, Barker, Catchpole and Hollingworthl and is similar to that used more recently by Barker and Hollingworth2ps concurrent with the developments described here.August, 19611 BY LIQUID - LIQUID PARTITION CHROMATOGRAPHY 523 The most satisfactory method of packing the column is to add the correct proportion of water to a stirred slurry of dry silica gel in a large volume of cyclohexane and to continue the stirring for a short time.The slurry obtained can be packed to form a column showing no “bands” by the technique described below. The chromatographic tube is filled with cyclohexane and connected via a siphon tube (see Fig. 6) to a 250-ml flask containing about 60g of the prepared gel in about 150ml of cyclohexane. A micro pump (obtainable from the Distillers Company Ltd.) is then used to pump cyclohexane into the flask at 1500ml per hour so as to force the #24 i.d. mm sinter Fig. 5. Details of column Teflon tube (3 mm i.d.) I Cyclohexane in at 1500 ml per hour Fig. 6. Column-packing device slurry through the siphon tube and into the column; the surplus cyclohexane running from the bottom of the column can be re-cycled.It is sometimes necessary to tap the 250-ml flask lightly to ensure an even flow of slurry. When packing has been completed, the top surface of the column is protected from disturbance by a disc of filter-paper, and the level of cyclohexane is never allowed to fall below this disc. COLLECTION AND MEASUREMENT OF ELUATE- The eluate from the column passes through a Teflon tube (16mm i.d.) to a cell con- structed as shown in Fig. 7 from silica plates secured with Araldite adhesive to a brass plate 4mm thick. This assembly fits in the cell-well of a Unicam SP500 spectrophotometer modified for automatic scanning as described by Shrewsbury.* In this work, the instrument is set at a fixed wavelength (2800 A), and the percentage transmission is recorded on a chart moving at 1 inch per hour.(The cyclohexane used must have a low optical density at all wavelengths accessible to the instrument ; commercial grades must generally be purified before use by passage through dry silica gel.) From the cell, the eluate passes into a Teflon tube (1-5 mm i.d.) coupled by a B10 joint to a flow controller (see Fig. 8). Accurate control is obtained by adjusting the height of a stainless-steel wire (17 s.w.g.) and so varying the restriction to flow imposed by it. It is an advantage to file the wire so that its cross-sectional area is decreased towards its lower tip. The rate of flow is measured by a simple calibrated flowmeter of the type shown in Fig. 9, which indicates the hydrostatic head of eluting solution required for a corresponding rate of flow through a fixed capillary restriction.The flow is normally in the range 50 to 80ml per hour.524 YOUNG : THE DETERMINATION OF DIHYDKOXYBENZEMES Teflon tube [Vol. 86 Brass lid 111 11 Adjustable bracket mm X , 4 mm) L / I Brass pl& 4 mm thick Fig. 7. Details of cell The eluate from the flowmeter is usually collected with a Shandon fraction collector equipped with fifty 15-ml tubes. This type of fraction collector has a scoop that will take the eluate to waste if the mains electricity supply is interrupted, and use is made of this by controlling the mains input in two ways. A time switch stops and starts the collection, and a Simmerstat working on a time cycle of just over 1 minute will direct a fixed proportion of the eluate to waste.This means that a known proportion of each component can be r Loose-fi t t i ng,l ad j ustab I e stainless-steel wire Wide- bore capillary \ Fig. 8. Details of flow controller Overflow tube Fig. 9. Details of flowmeterAugust, 19611 BY LIQUID - LIQUID PARTITION CHROMATOGRAPHY 526 collected during a long run without exceeding the volumetric capacity of the fraction collector. Used in conjunction, the two controls permit complete collection of part of the eluate and partial collection of the remainder. Each change of fraction is marked on the trace by a small “spike.” This is achieved by connecting a micro switch to a pawl that keeps contact with the notched rim of the turn-table. During each change of fraction, the signal to the recorder can then be partly shorted by a 9000-ohm resistor. OPERATION As the spectrophotometer is used for other purposes during the day, it has been con- venient to operate the column overnight as described below.During the day, the pre-column is re-packed, and butyl alcohol is washed from the analytical column with about 500 ml of purified cyclohexane; the rate of flow may be as high as 200 to 300 ml per hour and no air pressure is required. The flow is then adjusted to about 70ml per hour, and, during the last half-hour of the afternoon, the recorder is started, set to zero for dark current and then set at about 90 per cent. transmission at 2800 A with a slit width of 0.46mrn. The sample is introduced, the fraction collector is switched on, and the gradient siphon is primed (both flasks being full).The apparatus is then left without attention until the following morning. RESULTS In order to illustrate the procedure, the trace for a prepared mixture is shown in Fig. 10, the amounts of individual phenols present being in the range 0.2 to 2 mg. It has been found that the butyl alcohol is not eluted evenly, but builds up on the column and, during the run, suddenly “breaks through” at a concentration of just over 1 per cent. At this point on the trace a spurious peak occurs, the nature of which is not known. It may be produced by traces of impurities adsorbed on the silica gel and then displaced as a sharp band by the butyl alcohol. QUANTITATIVE ANALYSIS- The amount of a component present in a peak is given by the expression- kAtq Weight present, mg = - El in which k is a constant, A is the maximum height of the peak expressed as an optical density, t is the time in hours for the elution of the peak (the intercept along the base-line between the tangents to the sides of the Deak), q is the rate of flow in millilitres per hour, E is the extinct& coefficient (Ei&) at <he centimetres.wavelength used and 1 is the thickness of the cell in I 4- Time, hours ( I division = 2 hours) Fig. 10. Trace recorded for a prepared mixture526 YOUNG THE DETERMINATION OF DIHYDROXYBENZENES [Vol. 86 By calibration with prepared mixtures, the value of k was found to be about 5.7. For a true Gaussian peak: k should be equal to 6-27; the lower value is a purely empirical factor, which takes account of the flow conditions in the cell.In practice, it is convenient to tabulate values of k/El for the series of compounds to be determined, and the values used at 2800 A for some common phenols are shown in Table I. Phenol Phenol .. . . o-Cresol . . . . m-Cresol . . .. p-Cresol . . .. 2,3-Xylenol . . .. 2,4-Xylenol , . . . 2,5-Xylenol . . . . 2,g-Xylenol . . .. 3,4-Xylenol . . . . 3,5-Xylenol . . . . 2-Ethylphenol . . 3-Ethylphenol . . 4-Ethylphenol . . TABLE I VALUES OF k/El FOR VARIOUS PHENOLS Value of k/El, mg per ml Phenol . . 0.142 Catechol . . . . . . 0.108 3-Methylcatechol . . . . 0.088 4-Methylcatechol . . . . 0.075 3,4-Dimethylcatechol . . 0-114 3,6-Dimethylcatechol .. 0.087 4,5-Dimethylcatechol . . 0.111 3-Ethylcatechol .. .. 0.163 4-Ethylcatechol . . . . 0.098 Resorcinol . . . . . . 0.135 2-Methylresorcinol . . . . 0.105 4-Methylresorcinol . . . . 0.097 5-Methylresorcinol . . . . 0.103 2,4-Dimethylresorcinol 4-Ethylresorcinol . . Quinol .. . . Value of k/El, mg per ml .. 0.070 . . 0.113 . . 0.077 . . 0.060 a . 0-1 12 . . 0.097 . . 0.118 . . 0.085 . . 0.089 . . 0-186 . . 0-072 , . 0.127 .. 0.257 . . 0,079 . . 0.095 As an alternative quantitative method, ultra-violet analysis of collected fractions is slower, but slightly more accurate; it is essential when the identity of a peak is in doubt. In applying methods of quantitative analysis (especially by measurement of areas) it is important to allow for the polarity of the solvent used. The presence of butyl alcohol is likely to produce a considerable change in the spectrum of a phenol in cyclohexane and TABLE I1 RESULTS FOUND FOR PREPARED MIXTURES Mixture No.1 is that for which the trace is shown in Fig. 10. Average values for extinction coefficient were used in the two instances when the components were not resolved Total amount of collected Amount of dihydroxybenzenes- Amount component found by- A > Component o-Cresol , . .. m- and p-Cresols . . Phenol .. .. 3-Methylcatechol . . 4-Methylcatechol . . Catechol . . .. 4-Methylresorcinol . . 2- and 5-Methyl- resorcinols .. Resorcinol . . .. Total . . .. 3-Methylcatechol . . 4-Methylcatechol . . Catechol . . .. 4-Methylresorcinol . . 5-Methylresorcinol . . Resorcinol . . .. Mixture No. 1- Quinol .. . . Mixture No. 2- Quinol .. .. of --+-A, component ultra-violet measurement present, mg 0.34 0.65 1-62 0.30 0.32 1.86 0.3 1 0.62 1.67 1.68 9.27 0.57 0.15 1.60 0.46 0.26 1.79 1-68 analysis, mg -* -* -* -* 0.34 2-00 0.28 0.71 1.76 1.64 - 0.61 0.19 1.60 0.49 0-42 1.78 1-54 of area, mg 0.33 0.64 1-26 0.31 0.33 0.67 1.85 1-57 9.41 0.57 0.16 0-27 1.80 1.59 found by found by ultra-violet measurement present, analysis, of area, mg mg mg 6.46 6-51 6.73 6-63 6-87 6.57 * Fraction not collected.August, 19611 BY LIQUID - LIQUID PARTITION CHROMATOGRAPHY 527 will therefore affect the extinction coefficient at the wavelength used during the run. The extinction coefficient of the component in the correct solvent must always be used. Quantitative results for two prepared mixtures are shown in Table I1 to indicate the accuracy to be expected from the method. I thank the Directors of the Midland Tar Distillers Ltd. for permission to publish this paper and Mr. D. D. Shrewsbury for designing and constructing the cell and carrying out the necessary ultra-violet analyses. REFERENCES 1. 2. 3. 4. 5. Blackburn, W. H., Barker, L., Catchpole, J. R., and Hollingworth, N. W., Gus Councd Res. Comm. Barker, L., and Hollingworth, N. W., Gas Council Res. Comm. No. 52, 1958. -,-- , J . Appl. Chem., 1959, 9, 16. Shrewsbury, D. D., Analyst, 1959, 84, 68. Burrows, G., Trans. Inst. Chem. Engrs., 1957, 35, 245. No. 24, 1955. Received Januavy 24th, 1961