April, 19661 PEARSON 247 The Determination of Water in Lubricating Oils by a Near-inf rared Spectrophotometric Method BY B. D. PEARSON (Dmby and District College of Technology, Derby) The water content of clear mineral oils can be quickly and accurately determined by dissolving the wet oil in ethyl acetate and measuring the near-infrared absorbance of the soh tion, relative to a similar reference solution previously dried by molecular sieves. In discoloured or cloudy, wet oils the water can be removed by azeotropic distillation with ethyl acetate, and the water content of the distillate determined spectroscopically. With the latter procedure, errors that arise at low water concentrations may be due to the absorption of atmospheric moisture by the solvent. MANY workers have shown that water can be simply, rapidly and accurately determined in such liquids as glycerol, ethylene and propylene glycols, methanol, isopropanol, butanol and fuming nitric acid, by near-infrared spectroph~tometry.~ v 2 9 3 9 4 9 5 In general, the method has been considered to be applicable to the determination of water in alcohols, aldehydes, amines, esters and ketones.6 Because spectrophotometric measurements can be rapidly made over a single sharp absorption peak with a high degree of accuracy, it would be useful if the technique could be extended to hydrocarbons, particularly lubricating oils.Undiluted oils are difficult to manipulate into absorption cells ; measurements would be greatly simplified by the use of a solvent, not particularly hygroscopic and yet easily dried, for the oil and a reasonable amount of water.Ethyl acetate is a solvent which fulfils these conditions. Water in water - oil emulsions could also be determined using this procedure. Neither typical lubricating-oil additives nor lubricating oils interfere with the near-infrared water absorption band at 5290cm-l. Difficulties arise in the spectroscopic determination of water in used oils. These contain finely divided particles that are suspended in the oil and scatter much of the incident radiation, so introducing intolerably large errors. No simple method of removing the suspended material is available, and so this difficulty is overcome by the rapid azeotropic distillation of the water from the oil with ethyl acetate. The water content of the ester distillate is then determined spectroscopically.A similar procedure has been used to determine the water content of powders.' METHOD AND RESULTS REAGENTS- for at least 48 hours over molecular sieves (with approximately 25 g per 24 litres). sieves, regenerate them in a stream of argon at 250" to 300" C for 10 hours. Ethyl acetate-Dry ethyl acetate (Hopkin and Williams AnalaR grade) was further dried LWolecular sieves, Linde Air Products 4A grade (supplied by B.D. H.)-After using the CALIBRATION-BEER'S LAW MEASUREMENTS- Near-infrared spectrophotometric measurements were made on a Unicam SP700 instru- ment, with 10-mm stoppered cells. Standard solutions of water in ethyl acetate (or in ethyl acetate and oil) were prepared by the weight-fraction method, from solvents (or solutions) previously dried over molecular sieves.The absorbance of these wet solutions relative to the dry solvent (or solution) was measured. The results obtained are shown in Fig. 1. REMOVAL OF WATER FROM ETHYL ACETATE BY MOLECULAR SIEVES- To 57 g of ethyl acetate containing 0424 per cent, of water, w/w were added 5.9 g of molecular sieves. Samples of the solution were withdrawn at intervals ; their absorbance was measured and then the samples were returned to the bulk of the solution.248 PEARSON : DETERMINATION OF WATER IN LUBRICATING OILS [AndySt, VOl. 91 Resdts- Time, in minutes . . .. 10 20 35 55 86 125 230 3G0 480 Percentage of water (w/w) in solution . . . . . . 0.290 0.210 0.155 0.105 0.070 0.040 0.020 0.015 0.007 The above results were obtained when the samples were withdrawn at the times indicated.Identical results for the same water content were obtained for solutions of 40g of ethyl acetate and 17 g of oil, regardless of whether the oil was of the straight or detergent type, In each instance the time required to remove half the water from the solutions was 20 minutes. Water, per cent. (w/w) Fig. 1. Beer’s Law relationships for water in lubricating oil samples. Graph A, in ethyl acetate; graph B, in ethyl acetate and straight oil, 2 to 1, v/v; graph C, in ethyl acetate and detergent oil, 2 t o 1, viv UPTAKE OF ATMOSPHERIC MOISTURE BY ETHYL ACETATE- Ten-ml samples of ethyl acetate were measured into eight similar specimen-tubes, 7-5 cm in length and 1.7 cm in diameter. These tubes were placed in a tank saturated with water vapour, at 24” C.One tube was immediately withdrawn to determine the small “handling correction,” and the other tubes were withdrawn at intervals. RCSW~~S- Time, in minutes . . . . . . . . 30 63 90 110 195 255 305 Percentage of water (w/w) dissolved (corrected) 0.034 0.081 0-102 0.124 0.215 0.278 0.374 The above results were obtained when the samples were withdrawn at the times indicated. AZEOTROPIC DISTILLATION APPARATUS- Attach a round-bottomed, side-arm flask of 500 nil capacity to a lagged column 50 cm long and 3 cm in diameter, which has been loosely packed with glass rings. Connect this, via a distillation head and a 35-cm long condenser, to a calibrated receiver protected by a drying tube. Support the flask-heating mantle by means of a laboratory jack so that it can be lowered for fast cooling.Before use, dry the apparatus by distilling at least 80 ml of dry ethyl acetate. REMOVAL OF WATER BY AZEOTROPIC DISTILLATION- 1.360 per cent. of water, w/w. intervals from the distillation flask was determined. The apparatus described above was used to distil 345 ml of ethyl acetate containing The near-infrared absorbance of 5-ml samples withdrawn at Volume of ethyl acetate distilled, in ml . . 50 100 150 200 250 Percentage of water (w/w) remaining . . 0.565 0.337 0.150 0.072 G.018 Percentage of water removed . . . . 58.5 75.2 89.0 94.7 ’38.7April, 19661 BY A NEAR-INFRARED SPECTROPHOTOMETRIC METHOD 249 DETERMINATION OF WATER IN OIL SAMPLES BY USING AZEOTROPIC DISTILLATION- Wash a 50-ml wet-oil sample into the distillation flask with 100 ml of ethyl acetate.Add two dry anti-bumping granules and distil the mixture until 80 ml of distillate have been collected. Measure the absorbance of the distillate in the region 5450 to 5150cm-l, and determine the percentage of water in the distillate by reference to the calibration graph (Fig. 1 ) . It was found that the results for twelve 50-ml samples, containing between 0.0440 and 0-2550 g of water, determined by this method showed an average error of 0.04 per cent. The maximum error was +2.6 per cent. DIRECT DETERMINATION OF WATER IN NEW OR CLEAR OILS- Dissolve 33 ml of wet oil in ethyl acetate and make the volume up to 100 ml. Measure the near-infrared absorbance of this solution relative to a similar solution that has been dried over molecular sieves. Find the water content of the solution from the ethyl acetate and oil calibration graph.DISCUSSION Meeker, Critchfield and Bishop6 have shown that dry reference solutions for near- infrared spectrophotometry can be quickly obtained by using molecular sieves, type 4A. Their results by this method, for the water determination in ethyl acetate and a number of other organic liquids, were in close agreement with those obtained by the Karl Fischer method. This investigation has shown that this drying technique is as equally effective for ethyl acetate - oil mixtures as for ethyl acetate alone. Water can be accurately determined in new or clear oil samples by measuring the absorbance of such systems as water - ethyl acetate - oil or water - diethyl ether - acetone - oil, in the region 5450 to 5100 cm-l.The former system is the simpler of the two for practical use. It is not particularly hygroscopic, absorbing approximately 0.001 per cent. of its own weight of atmospheric moisture per minute, under the experimental conditions used. The ethyl acetate can be readily recovered, and in the proportion of two parts of ethyl acetate to one part of oil (by volume), approximately 0-5 per cent. of water (i.e., 1-5 per cent. of water with respect to the oil), readily dissolves. A disadvantage of this system is that ethyl acetate precipitates some oil additives, but this factor does not lead to errors in the determination of the water content of the oil. The system, diethyl ether - acetone - oil (5.5 to 2.5 to 2.0, by volume) is less convenient ; solvent recovery is more difficult, and cell-window cleaning before measurement is troublesome due to solvent creepage.This system does, however, take up rather more water (2.6 per cent. with respect to the oil). The experimental procedure indicated has the advantage that a dry-oil solution, for reference purposes, can be readily obtained from a wet-oil sample. The wet oil is dissolved in ethyl acetate, and the solution is divided into two parts, one of which is dried with molecular sieves for about 24 hours. This forms the reference solution from which calibration solutions containing, say, 0.1, 0.2, 0.3 and 0.4 per cent. of water may be prepared. A calibration graph is then constructed and the absorbance of the retained wet-oil solution determined.Such a procedure would be tedious if it had to be carried out for every oil sample, but where the same type of oil is dealt with many times only a few minutes are required for each deter- mination, as the same refercnce solution and calibration graph can be used in each case. Calibration graphs are very similar for straight or detergent oils (Fig. l), and if some accuracy can be sacrificed, an “average” calibration graph can be used. However, if the maximum degree of accuracy is required, solutions should be made up by weight rather than volume, and the base-line accuracy of the spectrophotometer periodically checked against solutions of known water content. Under such conditions the limit of accuracy previously suggested6 for the near-infrared determination of water at 5290 cm-l, i.e., k0-02 per cent.absolute, in the range 0-02 to 1.00 per cent. of water, can probably be improved to +0.004 per cent. absolute above the 0-075 per cent. water level. The absorbance of ethyl acetate solutions of many used and all cloudy oils cannot be measured spectroscopically as the solutions scatter much light. In these instances the water in the oil may be removed by azeotropic distillation with ethyl acetate and the amount of water in the distillate determined spectrophotometrically. The efficiency of water-removal by this type of distillation was illustrated by distilling 250 ml of ethyl acetate from 345 ml250 PEARSON [Analyst, VOl. 91 of the ester containing 1.360 per cent.of water. Examination of the distillation residue at this stage showed that 98.8 per cent. of the water had been removed. So, for an ethyl acetate and oil solution containing only up to about 0-5 per cent. water, all the water will have been removed if the volume is reduced to one-fifth by distillation. This is the procedure that was adopted. To determine the accuracy of this method, known amounts of water were added to mixtures of ethyl acetate and oil which had previously been dried over molecular sieves. The ester was distilled and the water content of the distillate determined. For twelve such samples containing 0.1 to 0-6 per cent. of water with respect to the oil, the average error was 0.04 per cent., with a maximum error of +2.6 per cent. The accuracy of the method is highest for oil samples containing more than 0.3 per cent, of water.At lower water con- centration levels the accuracy decreases, possibly due to the now important effect of atmos- pheric moisture during the transference of solutions. For new or clear oil samples good agreement was obtained between the direct spectrophotometric and azeotropic distillation methods. It is interesting that Beer’s law, for solutions of water in ethyl acetate, is only obeyed up to a concentration of 0-42 per cent. w/w of water. Deviations that occur above this value are possibly due to the association of water molecules. A similar deviation was observed for solutions of water in the esters, ethyl propionate and methyl acetate. I thank Mr. J. Duncalf for help with the preliminary experimental work. REFERENCES 1. 2. 3. 4. 5. 6. 7 . Keyworth, D. A., Talanta, 1961, 8, 461. Cordes, H. F., and Tait, C. W., Analyt. Chem., 1957, 29, 485. Chapman, D., and Nacey, J. F., Analyst, 1958, 83, 377. White, L., and Barrett, W. J., Analyt. Chem., 1956, 28, 1538. Sakai, K. et aZ., J . Chern. SOC. Japan, Ind. Chem. Sect., 1959, 62, 632; AnaZyt. Abstr., 1961, 8, 1832. Meeker, R. L., Critchfield, F. E., and Bishop, E. T., Analyt. Chem., 1962, 34, 1510. Brandenberger, H., and Bader, H., Ibid., 1961, 33, 1947. Received June 2nd, 1965