782 ABaZyst, December, 1968, Vol. 93, pp. 782-787 Recent Methods for Determining Traces of Nitrogen in (Mineral Oils BY P. GOUVERNEUR AND F. VAN DE CWATS (KoninklijkelSheEl-Laboratorium, A msterdam (Shell Research N . V.)) Three methods for the determination of traces of nitrogen in oil are discussed, viz., those based on extractive percolation, oxy-hydrogen combus- tion and hydrogenation - coulometry. Their scopes are compared with respect to their lower detection limit, the range of products to which they can be applied and their speed. ALTHOUGH organic nitrogen compounds are only minor constituents of mineral oils, they citn play a significant rdle, both during the manufacture and in the performance of oil products. The nitrogen concentrations involved are usually small, most often in the parts per million range, and sometimes below that level. Fortunately, there is a choice of analytical methods available today, thus enabling successful results to be achieved in this concentration range.Three of them are discussed in some detail and their scopes compared with respect to their lower detection limit, the range of products to which they can be applied and their speed. In each of the following methods ammonia is the final reaction product, but the des- tructive treatment and the method of ammonia formation are different. The first method involves extractive percolation of the oil sample through sulphuric acid on a carrier, followed by Kjeldahl digestion of the nitrogen-containing concentrate; the second, oxy-hydrogen com- bustion, whereby the nitrogen compounds of the oil are initially converted into nitrogen oxides, and later into ammonia by Devarda reduction; and the third, catalytic hydrogenation, followed by microcoulometric titration of the ammonia formed. Dumas’ combustion method is not included in the present discussion because of the interference encountered in this method from dissolved molecular nitrogen, the amount of which often exceeds the combined nitrogen fraction by several times. EXTRACTIVE PERCOLATION THROUGH SULPHURIC ACID ON A CARRIER- This is essentially a pre-concentration method and has been found a successful tech- nique for concentrating all of the nitrogen compounds (basic and non-basic) contained in oil fractions.It is based on percolation of the sample through concentrated sulphuric acid distributed on an inert carrier, e.g., fine-grade pumice, in a small column.After passage of the oil, the contents of the column are subjected to Kjeldahl treatment and the ammonia is determined titrimetricallyl or, with small samples, spectrophotometricaUy.2 Since its introduction in 1962, the technique has proved useful for a wide variety of petroleum distillate fractions ranging from low-boiling fractions to lubricating oils with nitrogen levels from 0 to 500 p.p.m. TABLE I EXTRACTIVE PERCOLATION METHOD : NITROGEN CONTENT OF REFINED KEROSENES 2-litre samples; percolation rate 500 ml per hour Nitrogen content, parts per thousand million Sample Natural Added Total present Total found A 19* 1137 132 122; 131 B - 8; 14 C - 31; 28 D - 194; 199 A I \ * Average from multiple determinations by same method.t Nitrogen-containing kerosene. Paper presented at the Joint Symposium on Limits of Detection in Analysis, April 17th and 18th. 0 SAC and the authors. 1968, Enschede.GOUVERNEUR AND VAN DE CRAATS 783 It has been found more recently that this technique can be easily extended to the parts per thousand million range for highly refined oil fractions. Table I shows parts per thousand million amounts of nitrogen in kerosene samples, detected by percolation of 2-litre samples at a rate of 500ml per hour. 0 /El I00 90 Y c 8 80- 5 L Q) Q 70 5 6 0 - f! 2 $ 50- M E 40- Z 30 - - - 0 I I I 20' Qo 70 80 90 IC Sulphuric acid strength, per cent. Fig. 1. Strength of sulphuric acid distributed on pumice versus nitrogen recovery for the extractive percolation method: 0 straight-run heavy gas oil, 200 p.p.m. of nitrogen; 0 catalytically cracked light gas oil, 122 p.p.m.of nitrogen A = Reducing valves B = Flame arrester C = Precision regulating valves D = Control valve and rotameter for primary E = Control valve and rotameter for hydrogen F = Control valve and rotameter for secondary G = Metal pressure safety valves H = Flame arrester J = Burner oxygen oxygen K = Sample feed unit L =Condenser M = Adaptor N = Flow indicator 0 = Scrubber P = Condenser Q = Absorber R = Splash bulb S = Control valve and rotameter for suction T = Water flow meter Fig. 2. Diagram of apparatus for nitrogen determination by oxy-hydrogen combustion784 [Analyst, Vol. 93 The recommended concentration of the sulphuric acid is 98 per cent., although in later work this proved to be less critical than originally believed.Fig. 1 indicates that results are not greatly affected unless the concentration of the sulphuric acid drops below 94 per cent. OXY-HYDROGEN COMBUSTION- Oxy-hydrogen combustion is a well known method for decomposing organic material quickly and effectively, and the Wickbold apparatus developed for this purpose is widely used nowadays for determining traces of sulphur and halogen.5 In this method the sample is fed into an oxy-hydrogen pilot flame and burned with a hot flame in an excess of oxygen. The combustion products are then drawn through a scrubber containing a suitable absorbent for trapping the component to be determined.Combustion rates can be high, e.g., from 1 to 5 ml per minute, depending on the nature of the sample. The determination of nitrogen in this way is a more recent development4 and com- prises the following steps: conversion of the organic nitrogen compounds in the hot flame, producing nitrogen oxides (mainly nitric oxide) ; trapping of the oxides formed on sodium chlorite on an alumina carrier, thereby producing a nitrite - nitrate mixture; and wet reduction of this mixture with Devarda alloy producing ammonia, which can be determined in the usual way. The combustion apparatus used is shown in Fig. 2. A satisfactory sample feed-rate is about 2 ml per minute, and a useful upper limit for the amount of nitrogen to be convertedin one run is 5 mg, preferably contained in 5 to 50 ml of sample.In this way any overloading of the sodium chlorite reagent is prevented. The reduction of the nitrite - nitrate mixture to ammonia and the isolation of the latter is a one-step operation that proceeds simply and conveniently, so that the whole method is relatively quick in comparison with traditional methods for the detection of trace nitrogen. Fifteen determinations can be made in a single working day. GOUVERNEUR AND VAN DE CRAATS: RECENT METHODS FOR TABLE I1 OXY-HYDROGEN METHOD : CONVERSION OF NITROGEN COMPOUNDS All compounds dissolved in benzene - kerosene Compound Nitrobenzene . . .. Aniline . . .. .. Azobenzene . . .. Pyridine . . a . .. Octylamine . . .. Heptyl cyanide . . .. Quinoline .. .. .. Indole . . .. .. Nitrogen theory, p.p.m.23 1 1016 296 246 204 171 203 215 Nitrogen average found, p.p.m. 222 1028 281 99 204 169 200 207 Recovered as ammonia, per cent. 96 101 96 40 100 99 99 96 In model experiments with blends of pure nitrogen compounds in nitrogen-free fuel, the conversions into ammonia were between 95 and 101 per cent., as Table I1 shows. Azobenzene, however, is an exception. This compound evidently splits off an appreciable fraction of its nitrogen as elemental nitrogen, which is only sparingly oxidised in the flame. However, this is of little consequence for the present purpose because tke occurrence of compounds of this type in petroleum has not been reported. TABLE I11 OXY-HYDROGEN METHOD : NITROGEN CONTENT OF OIL FRACTIONS Nitrogen by Nitrogen by extractive percolation, oxy-hydrogen combustion, Sample p.p.m.p.p.m. Naphtha .. .. .. .. .. 97 94; 95 Light gas oil . . .. .. .. 15 16; 13; 16 Heavy gas oil* . . .. .. .. 267 261; 254 Gas oil concentrate* . . .. .. 577 542; 540 Luboil* . . .. .. .. .. 616 618; 613 Dark steam cylinder lubricant* .. 478 470; 496 Shale oil* .. .. .. . . 1.59 per cent. 1.57; 1.59 per cent. * Diluted.December, 19681 DETERMINING TRACES OF NITROGEN IN MINERAL OILS 785 Some results obtained on actual oil samples are presented in Table 111, the oils having nitrogen contents from 15 p.p.m. upwards. In general, satisfactory agreement was found with results obtained by the extractive percolation method, both for low-boiling and for residual oil fractions, and also in the exceptional case of the shale oil sample.It may be concluded, therefore, that the nitrogen compounds of oils are generally converted into nitrogen oxides during the oxy-hydrogen combustion treatment. The conversion of molecular nitrogen, on the other hand, is poor and amounts to only a few per cent. under the flame conditions. In consequence, the effect of the portion of dissolved nitrogen present in the oils is not sig- nificant. Unfortunately the situation is quite different for the gases used for the combustion. Their molecular nitrogen impurity level often ranges from 100 to 1000 p.p.m. v/v and, as the gas consumption is relatively high, the blank contribution is significant. For this reason the method is no longer attractive for oils containing less than 15 p.p.m. of nitrogen. CATALYTIC HYDROGENATION - COULOMETRIC METHOD- This method was published by Martin6 in 1966 and comprises catalytic hydrogenation of the oil sample with nickel catalyst, followed by microcoulometric acid - base titration, in a special cell, of the ammonia formed.An outstanding feature of this method is that it is capable of detecting nanogram amounts of nitrogen in milligram amounts of oil in a short time. Moreover, the use of such small samples and a relatively large excess of nickel catalyst appreciably reduces the effect of nickel de-activation by poisoning and coke deposits (a tradi- tional drawback in Ter Meulen type hydrogenations). Hydrogen r Insert scrubber t Hydrogenation furnace injection Fig. 3. Hydrogenation - coulometric nitrogen method A schematic arrangement is given in Fig. 3.Fig. 4 shows the equipment, which is available from Dohrmann Instruments Company (Mountain View, California). The pro- cedure is as follows: the 1 to 10-pl oil sample is injected into a 400" C zone and subjected to hydrogenation in a current of ultra-pure hydrogen (24 litres per hour) through nickel catalyst. The catalyst supplied by the manufacturer, viz., granular nickel of high purity, is used at 900°C in contrast with Martin's catalyst, which was nickel on magnesium oxide used at 440" C. An insert scrubber tube containing, e.g., magnesium oxide serves to remove inter- fering acidic products that originate from accompanying sulphur and halogens. The am- monia leaving the tube is fed into a titration cell designed to operate at a pre-determined constant hydrogen-ion concentration in the pH range 5 to 6.The cell contains 0.04 per cent. sodium sulphate solution as the electrolyte and has four electrodes, a sensing pair (hydrogen electrode v e m w lead - lead sulphate reference electrode) and a generating pair (two platinum electrodes). Any change in concentration (incoming ammonia) is detected by the sensing electrode pair as a potential difference, which leads through the coulometer amplifier to the generation of hydrogen ions at the generator electrode. The required current is recorded via a precision series resistance on a potentiometric recorder, the peak area representing the current integral with the chart speed as the time basis. The number of coulombs required for neutralisation is then known. Initially, the application of this method to oil samples failed, although ammonia produced by heating weighed sub-micro amounts of ammonium chloride was completely recovered.This suggested that the trouble was in the catalytic hydrogenation step, and the experiencesFig. 4. Hydrogenation - coulometric apparatus To fuce page 7851786 GOUVERNEUR AND VAN DE CRAATS: RECENT METHODS FOR [Arta&St, VOl. 93 of colleagues (E. M. Fredericks and E. D. Peters in a private communication) have confinned it. They suggested a pre-treatment of the nickel catalyst and also ultra-purification of the hydrogen used. Fresh catalyst should be subjected in sit% to an oxidative treatment (pure oxygen at 900" C) and subsequently to a reducing treatment (pure hydrogen at 400" C), with pure helium at the intermediate stages for safety reasons.This treatment was repeated as soon as the catalyst began to show signs of de-activation. The hydrogen used was purified by palladium diffusion with an A.5 Diffusion Unit manufactured by Johnson, Matthey and Co. Ltd. HYDROGENATION - Nitrogen theory, p.p.m. 1000 500 250 125 50 TABLE IV COULOMETRY: OCTYLAMINE I N KEROSENE Nitrogen average Injected, found, Recovery] P1 p.p.m. per cent. 0-5 to 1 667 67 0-5 t o 2 374 75 1 237 95 1 130 104 2 48 96 Table IV illustrates the conversion of octylamine when blended with nitrogen-free kerosene. Best results were obtained for the more dilute blends, and for practical reasons the 50 p.p.m. nitrogen level was selected for further work. Results relating to the conversion of other nitrogen compounds at this level are shown in Table V.Satisfactory results were obtained for the compounds tested, except for azobenzene. Here again, the tendency of TABLE V HYDROGENATION - COULOMETRY : CONVERSION OF NITROGEN COMPOUNDS All compounds dissolved in kerosene; 2 4 samples injected, inlet temperature 400" C Compound Nitrobenzene . . .. Aniline . . .. .. Azobenzene . . .. Pyridine . . .. .. Octylamine . . .. Heptyl cyanide . . .. Quinoline .. .. .. Indole . . .. .. Nitrogen theory, p.p.m. 65 50 50 50 49 55 60 60 Nitrogen average found, p.p.m. Gl 48 50 40 47 55 53 53 Recovered as ammonia, per cent. 94 96 100 80 96 100 106 106 this compound to produce elemental nitrogen when decomposed obviously prevents complete conversion, as in the oxy-hydrogen method.Results for oil samples are shown in Table VI. In general they are in reasonably good agreement with known values, even those for the shale oil sample. The high-boiling dark steam cylinder lubricant gave a low value, however. TABLE VI Inlet temperature 400" C, magnesium oxide scrubber 450" C HYDROGENATION - COULOMETRY: NITROGEN CONTENT OF OIL FRACTIONS Nitrogen by extractive percolation, Sample p.p.m. Naphtha .. .. .. .. 97 Kerosene . . .. .. .. 3.2 Light gas oil* . . .. .. .. 187 Heavy gas oil* .. .. .. 267 Gas oil concentrate* . . .. .. 577 Dark steam cylinder lubricant* . . 478 Shale oil* - . .. .. .. 1.59 per cent. * Diluted in kerosene to a 50 p.p.m. level. Nitrogen by hydrogenation- coulometry, p.p.m. 94; 95 3; 4 165; 180 251; 283 588; 623 334; 392 1.43; 1.69 per cent.December, 19681 DETERMINING TRACES OF NITROGEN IN MINERAL OILS 787 These experiments led us to the conclusion that the hydrogenation - coulometric method is most effective for products that are sufficiently volatile below 400” C, and that it is rapid, the signal being available almost instantly and the calculated result well within 15 minutes, including the necessary dilution.The lower detection limit at present appears to be I p.p.m. Further work will be necessary to extend the method to products boiling well above 400” C. CONCLUSION In Table VII the salient features of the methods discussed have been summarised. TABLE VII SUMMARY OF CHARACTERISTICS OF THREE METHODS FOR TRACE NITROGEN DETERMINATION IN OILS Extractive percolation Useful product range : low boiling ..* . .. + highboiling .. .. .. + - residual . . .. .. .. Useful nitrogen range, p.p.m. . . 0.01 to 600 Sample requirements . . .. . . 10 ml to 2 litres Repeatability, usual . . .. . . 6 per cent. amount Repeatability, best . . .. . . 0.01 p.p.m. Analysis time: Number of analyses per person per present 1 test elapsed . . .. .. 6 hours 8-hour day .. .. .. 3 Oxy-h ydrogen combustion + + + 16 to 1000 6 to 40 ml 6 per cent. amount present 6 p.p.m. 46 minutes 16 Hydrogenation - coulometry + (+I 1 to 60 6 per cent. amount present 1 p.p.m. 16 minutes I 2 to 10 pl 30 The extractive percolation method, although of limited speed, is capable of determining parts per thousand million concentrations of nitrogen in certain refined oil fractions. The oxy-hydrogen combustion method is more rapid, and is characterised by its general applicability to oil products with nitrogen contents of over 15 p.p.m. The “Dohrmann” technique (hydrogenation - coulometry) appears to be the most rapid, and is useful for products that are sufficiently volatile below 400” C and have nitrogen contents of over 1 p.p.m. REFERENCES 1. 2. 3. Gouverneur, P., Analytica Chim. Acta, 1962, 26, 212. Smith, A. J., Cooper, F. F., jun., Rice, J. O., and Shaner, W. C., jun., Ibid., 1968, 40, 341. “Proceedings of the Colloquium on Experiences with the Wickbold Combustion Apparatus for Organic Elementary Analysis, Hanau, February, 1966,’’ Heraeus-Schott Quarzschmelze GmbH, Hanau, Germany. Gouverneur, P., Snoek, 0. I., and Heeringa-Kommer, M., Artalytica Chim. Acla, 1967, 39, 413. Martin, R. L., Analyt. Chem., 1966, 38, 1209. 4. 6. Received May 7th, 1968