756 Analyst, August, 1979, Vol. 104, $9. 756-765 Automatic Emission Spectrometer for the Deter m i nation of N i trog en -1 5 J. D. S. Goulden and D. N. Salter* National Institute for Research in Dairying, Shinfield, Reading, RG2 9A T An automatic nitrogen-15 analyser, employing a novel use of a rhodium- platinum catalyst for the generation of nitrogen and capable of analysing 60 samples per hour, is described. Nitrogen compounds of biological origin are first converted into ammonium chloride by conventional Kj eldahl digestion and distillation methods. The ammonium chloride sample (about 5 pl containing about 10 pg of nitrogen) is injected into a soda-lime reactor at 590 "C through which flows a stream of purified helium. Ammonia that is released passes directly into the catalyst tube and the generated nitrogen and hydrogen are separated by passage through a gas-chromatographic column, which also retains the water.After passing through a pressure restrictor the nitrogen flows in the helium stream through a Spectrosil discharge tube located in a microwave cavity. The emitted radiation is analysed by means of a specially con- structed dual-wavelength monochromator and the intensities of the 14N14N (297.7 nm) and 14N15N (298.3 nm) bands are measured simultaneously by two photomultipliers. Amplified signals, proportional to the peak intensities, are fed through phase-sensitive detectors into a ratiometer, the output from which is fed to a digital voltmeter and printed out in terms of nitrogen-15 abundance. A peak detector indicates the total nitrogen content of each sample and actuates the nitrogen-15 print-out. The response of the instrument is slightly curvilinear but may be regarded as linear over limited ranges.Calibration can therefore be achieved by running suitably chosen standards to fix upper and lower set points. Carry- over between samples is very small and is eliminated by running duplicates. Standard deviations of replicate measurements of natural abundance and enriched standards are less than 0.01 atom-%, while determinations of nitrogen-15 in biological samples were shown to be accurate to fO.01 atom-% by comparison with a Statron NOI-4 nitrogen-15 analyser. Keywords ; Nitrogen- 15 determination ; catalytic nitrogen generation ; auto- mated emission spectrometer The analysis of nitrogen-15 (15N) by the classical method of Rittenbergl using a mass spectrometer is a tedious and demanding operation, which for many years severely limited the use of nitrogen-15 as a tracer in biological research.Subsequently, the technique of emission spectrometry was applied to the analysis of nitrogen isotopes by Broida and Chapman,2 and the developments of Meier and Muller3 and Faust4 led to the introduction of reliable and relatively cheap commercial emission spectrometers. Further developments by Lloyd- Jones and co-workersk7 have resulted in improved accuracy and speed of analysis. The methods for the preparation of nitrogen for isotopic analysis have, however, continued to be time consuming and proved to be a limiting factor in nutritional studies.8 The application of optical and mass-spectrometric techniques has been discussed in a recent review.9 .In this paper an automatic nitrogen-15 analyser, capable of analysing 60 samples per hour, is described. The generation of pure nitrogen gas from a solution of ammonium chloride prepared from the test material is carried out automatically and the result printed out as nitrogen-15 atom per cent. Total nitrogen content of the sample can also be indicated simultaneously and displayed as a peak on a chart recording. The method is used routinely for samples containing 10 pg of nitrogen, and with some loss in accuracy may be used for samples containing as little as 2 pg of nitrogen. It is therefore ideally suited to the deter- mination of nitrogen-15 in compounds available only in very small amounts, e.g., amino acids prepared by ion-exchange chromatography of protein hydrolysates or physiological fluids.A preliminary report of this apparatus has been published.10 * To whom all correspondence should be addressed.GOULDEN AND SALTER 757 Principle of the Method With the manual nitrogen-15 analyser (e.g., the Statron NOI-4 or NOI-5), part of the emission spectrum of nitrogen in the 300-nm region is recorded and the peak heights for 14N14N and 14N15N in a series of replicate scans are measured. Nitrogen-15 isotopic abundance is then calculated from the mean ratio of these peak heights, after correction for background emission. In the National Institute for Research in Dairying (NIRD) automatic nitrogen-15 analyser, the two peaks at 297.7 and 298.3 nm are measured simultaneously, thus eliminating errors in the ratio due to variations of source intensity during the time needed for a wavelength scan.For both the manual and the automatic methods the nitrogen-15 labelled biological material is first subjected to a Kjeldahl digestion; alkali is added to the digest and the ammonia released is steam distilled into dilute hydrochloric acid, from which ammonium chloride is recovered by evaporation. From this stage onwards analysis with the NIRD analyser is fully automatic apart from the manual injection of samples by syringe, avoiding the troublesome vacuum technique involved in the preparation of emission tubes containing nitrogen. Ammonium chloride solution (approximately 10 pg of nitrogen) is injected into a stream of helium and passes immediately through a heated reactor tube, the front portion of which contains soda-lime.Ammonia is generated and is dissociated by a catalyst in the rear portion of the tube. The generated gases are separated by a gas-chromatographic column, which also retains the water. They are then eluted sequentially through a flow restrictor into a discharge tube located in a microwave cavity. The emitted radiation is analysed by means of a dual-wavelength monochromator. Signals proportional to the intensities of the bands at 297.7 and 298.3 nm, measured simultaneously, are amplified and fed through a pair of phase-sensitive detectors into a ratiometer whose output is fed t o a digital voltmeter and printed out as nitrogen-15 abundance in atom per cent.As the nitrogen flows through the discharge tube a peak detector actuates the printer and gives an indication of the total mass of nitrogen present in the sample. Materials and Methods The Statron NOI-4 emission spectrometer, the preparation of reagents and samples for analysis and the operation of the equipment have been described previously.6 Standard nitrogen-15 labelled ammonium chloride samples (measured in a mass spectrometer) were obtained from C.Z. Scientific Ltd. For analysis in the NIRD automatic nitrogen-15 analyser, samples of biological material (normally containing 200-1 000 pg of nitrogen, but for amino acid fractions from a chromatography column sometimes as little as 20-50 pg of nitrogen) are digested with 3 ml of concentrated sulphuric acid (low nitrogen, Aristar grade, BDH, Poole, Dorset) and one tablet of mercury(I1) oxide catalyst (Kjeltabs F.M., Thompson & Capper Ltd.), heating being continued for at least 1.5 times the clearing time.The ammonia released from the digest, by addition of an excess of 10 M sodium hydroxide solution containing 25 g 1-1 of sodium thiosulphate, is distilled in a Markham's apparatus and collected in 10 ml of 0.025 M hydrochloric acid. Between distillations of different nitrogen samples, 20-25 ml of ethanol are distilled to remove trace amounts of adsorbed ammonia and prevent cross-contamination. The resulting ammonium chloride solution is evaporated to dryness at 70 "C and the solid re-dissolved in glass-distilled water to give a solution containing approximately 2 pg pl-l of nitrogen.Alternatively, the microdiffusion method of Conwayll can be used for samples of low nitrogen content Ammonia released from portions of the diluted digest with 10 M potassium hydroxide solution containing 100 g 1-1 of sodium thio- sulphate is collected in 1 ml of 0.05 N hydrochloric acid in the centre well of a 60-mm Conway unit. The nitrogen-15 labelled ammonium chloride solution is evaporated to dryness and reconstituted to approximately 2 pg pl-1 of nitrogen. Components of the NIRD Nitrogen-15 Analyser in Fig. 1. Helium supply passing it through a BOC Rare Gas Purifier. A block diagram of the main components of the automatic nitrogen-15 analyser is shown Cylinder helium, first dried by passing through a molecular sieve (grade 5A), is purified byAmp1 ifier B .Phase- sen sit ive Peak - Chart detector - detector recorder B Wide slit mirror I - 1 ' EHT power unit - Microwave generator Ratiometer Phase- 1 sensitive detector A I I Vacuum pump I Photo- mu1 ti pl ier / I ' Discharge tube Dual wavelength monochromator 11 Furnace Soda- I i me Printer 'z FH-1 Calibration voltmeter Digital n Helium Helium purifier I 1 Oveh < I Nitrogen Pressure regulators controller I I I Valve Sample injection Fig. 1. Block diagram of main components of automatic nitrogen-15 analyser. bAugust, 1979 SPECTROMETER FOR THE DETERMINATION OF NITROGEN-15 769 Reactor tube In an earlier version of the NIRD nitrogen-15 analyser,12 nitrogen was generated from the ammonium chloride solution by reaction with copper and copper(I1) oxide, but a “memory effect” became apparent and produced serious cross-contamination of successive nitrogen-15 samples.Tests showed that the memory effect was most. probably caused by the formation of oxides of nitrogen as by-products of the Dumas reaction. These reacted with water already adsorbed on the chromatographic column to produce nitric acid, which could take part in exchange reactions with nitrogen. With the catalytic method of nitrogen generation, the front portion of a fused silica reactor tube (about 300 mm long, 14 mm 0.d.) is packed with soda-lime (BDH, 10-16 mesh). The rear 100-mm portion (7mm 0.d.) of the tube is packed with about 1.4g of the catalyst (Engelhard Industries Ltd., Chemical Group, Cinderford, Gloucestershire, Code No.99966, 0.25% rhodium - 0.25% platinum coated on alumina spheres, diameter approximately 0.2 mm) and the whole reactor tube is enclosed in a furnace at 590 “C. As the alumina spheres on which the catalyst is deposited are known to sinter at temperatures above 600 “C, the furnace temperature is reduced to about 500 “C during overnight regeneration. The sample of ammonium chloride solution (about 10 pg of nitrogen for normal operation) is injected manually through a silicone-rubber septum into the soda-lime tube and the liberated ammonia is dissociated. Because the dissociation reaction is endothermic, it is promoted by high temperatures. Ammonia dissociation is also accompanied by an increase in volume so that decomposition is favoured by high dilution in the helium carrier gas and small sample size.Exhaustion of the soda-lime (indicated by an erratic decrease in the intensity of the nitrogen peak on the chart record) leads to poisoning of the catalyst with the consequent presence of undissociated ammonia and the possibility of a memory effect between successive samples. Several thousand samples can be analysed before replacement is necessary. Gas-chromatographic column The column removes water and readily separates the nitrogen from attendant hydrogen. The nitrogen peak emerges 36 s after the hydrogen peak and has a pulse time of 2.4 s for half-peak maximum. The molecular sieve (13X, 30-60 mesh, Chromatography Services Ltd.) is packed into a stainless-steel tube 1.3 m long x 4 mm i.d., which has a water capacity equivalent to several hundred samples before regeneration.As water is absorbed the effective column length is reduced so that the elution time of about 1 min for nitrogen on a freshly regenerated column is progressively reduced during the analysis of a series of samples. The column is regenerated overnight by reversing the helium flow and setting the oven temperature surrounding the column to about 220 “C. Sintered stainless-steel filters are fitted at both column and reactor outlet connectors as a precaution against blockage of the flow restrict or. Discharge tube As the discharge operates at 10-15 Torr and the remainder of the preparation unit is at a positive helium pressure (approximately 30 p.s.i., which gave an elution time of just under 1 min for a 1.3-m column), it is necessary to introduce a throttle between the column exit and the discharge tube.This consists of a short stainless-steel tube flattened sufficiently to give the desired pressure drop. To avoid fluorescence effects the discharge tube was made from Spectrosil, its dimensions being 250 mm over-all length x 3 mm i.d. Light output is affected only slightly by changes in the discharge tube internal diameter, operating temperature, microwave power and pressure. It is convenient to operate the discharge at room temperature at about 22 W. Optical system A Rank-Hilger D331 double monochromator fitted with a 2400 lines per millimetre grating blazed for 300 nm was converted into a dual- wavelength monochromator. At 300 nm, the stated aperture and reciprocal dispersion were j75.5 and 1.3 nm mm-1, respectively.The transmitted beam enters a second compart- ment through a relatively wide slit that enables a narrow range of wavelengths to fall on the An optical diagram is shown in Fig. 1.760 GOULDEN AND SALTER: AUTOMATIC EMISSION Analyst, Vol. 104 wavelength change mirror. This mirror can be rotated through a small angle by a long lever working between a pair of adjustable stops. Choice of optimum slit widths and electronic parameters proved to be particularly difficult as all of these factors are interrelated, and a compromise had to be made between optical resolution and electronic noise in relation to sample size. Satisfactory performance was obtained with 0.10-mm entrance and exit slits, giving a spectral band width of 0.13 nm. The intermediate slit was sufficiently wide to allow both the 14N14N and 14N15N bands to be transmitted.The nitrogen supply was used only for setting up the optical system. Electronics system The electronics system was assembled from commercially available units with a minihum amount of modification. A Microtron 200 Mk I11 [Electro-Medical Supplies (Greenham) Ltd.] microwave generator is used with about 80% depth of modulation at approximately 1 kHz to excite the discharge. To avoid troublesome earth-loop problems, the screen on the modulation lead from the case is disconnected, and the 3A/4 cavity (Electro-Medical Supplies, Type 215L) insulated from the monochromator chassis. A pair of EM1 9789 QB photomultipliers (EMI, Hayes, Middlesex) are fitted into Rank-Hilger mounts and the electronics altered for ax.operation with the cathode earthed, supplied by an EM1 Type PM 25A EHT unit. The measuring system consists of a pair of precision amplifiers (Brook- deal Electronics, Model 9452), and a pair of phase-sensitive detectors (Brookdeal Electronics, Model 941 1) supplying the two input channels of a ratiometer (Brookdeal Electronics, Model 9547). The A lead (signal beam, see Fig. 1) is connected directly to the ratiometer whilst the B lead (reference beam) is connected through a potentiometer circuit (not shown), which serves as a fine control for balancing the two signals. An adjustable signal is taken off this circuit to operate a 1-mV chart recorder by means of which the approximate nitrogen concentration of the sample is monitored. A specially designed electro-mechanical peak detector built into the recorder actuates the printer at the peak maximum.13 The output of the ratiometer supplies the input to the printer (Anadex DP 501) via a calibration box (providing variable scale expansion and back-off) and a digital voltmeter (Fenlow Type 701-BCD).This allows the output to be printed as either millivolts or atom per cent. nitrogen-15 with appropriate adjustments for the slope and intercept of the straight- line calibration relationship. Tem$erature control For satisfactory reproducibility it was found necessary to house the nitrogen-15 analyser in a temperature-controlled room. A number of different components contribute to the temperature sensitivity of the instrument, e.g., wavelength shifts at the monochromator, photometric imbalance due to the differing temperature coefficients of the photomultipliers and temperature drifts of the electronic units. Tests have shown that provided the mono- chromator temperature is held to 30.2 O C , the errors in recorded nitrogen-15 are less than hO.01 atom-%. Regular temperature checks and occasional adjustments of the setting of the room thermostat enabled the temperature to be held within these limits without difficulty. Vacwm and @essure system The vacuum pump is not required to have a high performance and even a small rotary pump, such as the Edwards ISP 30, needs several feet of 2-mm bore tubing to restrict pumping capacity. Calibration and Correction for the Background Effect Even with no nitrogen present, the discharge shows some emission in the 300-nm region.Increasing the helium flow-rate, however, reduces the background emission , possibly owing to a change in pressure in the discharge zone. When the total amount of nitrogen in the sample decreases, it is necessary to increase the amplifier gain or photomultiplier EHT. This leads to an increase in the relative error due to background emission. The signal at 298.3nm (14N15N band) was backed off by a voltage corresponding to the backgroundAugust, 1979 SPECTROMETER FOR THE DETERMINATION OF NITROGEN-15 761 emission. This enables a two-wavelength measurement to provide results that are substantially independent of the nitrogen content. In the diagram the heights of the 14N14N and 14N15N bands relative to the signal with an opaque shutter are represented by x and y, respectively.The corresponding background levels are a and b. The true intensity ratio (R) is given by R = ( x - a)/(y - b) and hence x = Ry - Rb + a. The signal ratio (R’) observed during normal signal measurement is given by The theoretical basis for this technique is illustrated in Fig. 2. X Rb a (Rb - a) Y Y Y Y R ‘ = - = R - - + - = R - and for R’ to approach the true value R, the condition to make uncorrected background errors negligible is therefore that y > (Rb - a). This does not necessarily hold at low nitrogen-15 enrichments when R is fairly large, hence the need to make a background correc- tion by subtracting a constant signal (c, the back-off) from each channel.The measured ratio then becomes R’ = ( x - c)/(y - c). By making the back-off equal to the background emission (b) at the wavelength of the 14N15N peak, which is the smaller signal at low nitrogen- 15 enrichments, we minimise errors, and R’ = ( x - b)/(y - b). At low nitrogen-15 enrich- ments b < x and b = a, so that R’ = ( x - a)/(y - b) = R. In a practical test it was found that signals a and b were 0.6 and o.5y0, respectively, of the signal x for a natural abundance sample. At very high nitrogen-15 abundance, i.e., if y > x , it may be more accurate to put c = a, but this is seldom necessary in practice. 1 4 ~ 1 4 ~ 297.7 nm - f- :/ .1 r------ ~a /Opaque shutter zero Fig. 2. Theoretical basis of correction for bacl Wave1 ength/nm ground effect.For normal operation, a 5-pl sample of an ammonium chloride solution containing 10 pug of nitrogen is injected into the helium stream and the EHT adjusted so that the nitrogen peak on the recorder reaches about 70% of full-scale deflection. Tests have shown that using the zero off-set procedure, errors in nitrogen-15 atom per cent. for the low enrichment samples are within hO.01 when the peak height is within the range 60-9070 deflection. Adjustment of the volume injected from the 10-pl syringe enabled sample concentrations within the range 1 4 pg pl-l to be accommodated. Samples containing as little as 2 pg of nitrogen may be measured if the photomultiplier EHT is increased and the instrument calibration is set up at this level, but the error then increases to about 0.1 atom-%.The calibration procedure is carried out using a known low nitrogen-16 abundance sample (usually natural abundance) and one other of known enrichment chosen in relation to the enrichment of the samples of interest. For normal use, the instrument is first conditioned762 GOULDEN AND SALTER: AUTOMATIC EMISSION Analyst, Vol. 104 by injecting 10 samples of ammonium chloride solution each containing 10 pg of nitrogen in 5 p1. After checking EHT, zeros and balance, the low nitrogen-15 sample is injected and the intercept (back-off) potentiometer adjusted to show the correct reading. The high nitrogen-15 sample is then injected and the slope (variable scale expansion) potentiometer adjusted to give a correct reading. These two potentiometers are housed in the calibration box (Fig.1). When the upper set point is fixed (usually at 1.88 atom-%) and the lower at natural abundance (0.365 atom-%), readings between 0.365 and 2 atom-% are correct to hO.01 atom-% nitrogen-15. As the true relationship between millivolts and nitrogen-15 atom per cent. is curvilinear, print-out values for samples above 2 atom-% can be taken as approximate, and a first-order correction made using an error curve. More accurate results for such samples can be obtained by running the samples with an appropriate series of standards and carrying out a regression analysis on the printed nitrogen-15 readings. Alternatively, if a number of samples to be run are expected to have high enrichments, the instrument may be calibrated with two standard nitrogen-15 samples with values on either side of the mean expected, using the lower enrichment to set the intercept.The results can then be taken directly from the print-out. Re+eatability and accuracy chloride solutions are shown in Table I. The results of measurements of isotopic enrichments in three standard [15N] ammonium Six samples from each solution were measured in TABLE I REPEATABILITY SAMPLES OF MEASUREMENTS OF STANDARD [15N]AMMONIUM CHLORIDE WITH THE NIRD AUTOMATIC NITROGEN-15 ANALYSER Measured nitrogen-15, atom-% I A \ Enrichment of standard NIRD automatic used for calibration.* r A \ atom- % nitrogen- 15 Values 0.37 0.364 0.373 0.386 0.379 0.378 0.361 1.44 2.54 1.430 1.441 1.427 1.432 1.424 1.424 2.536 2.541 2.545 2.527 2.548 2.554 * Supplied by C.Z.Scientific Ltd. t S.D. = Standard deviation. Mean S.D.? 0.374 0.009 J -l Manual emission spectrometer Values 0.398 0.393 0.352 0.395 0.411 0.360 1.468 1.433 1.430 1.400 1.389 1.404 2.556 2.566 2.525 2.523 2.618 2.530 Mean 0.385 J 1.421 1 J i 2.553 S.D..f 0.023 0.029 0.036 the Statron NOI-4 and a further six samples from each of the same solutions were measured in the NIRD automatic nitrogen-15 analyser. These results show that the automatic analyser gave acceptable values for the standards (within 0.01 atom-%) and also indicate a higher reproducibility using the automatic method. The accuracy of the NIRD auto- matic nitrogen-15 analyser is illustrated in Table 11, which shows the results of measurements of nitrogen-15 in samples of the duodenal contents of a steer that had been given a feed containing [15N]urea.In each instance two samples of the biological material were digestedAagust, 1979 SPECTROMETER FOR THE DETERMINATION OF NITROGEN-15 TABLE I1 ACCURACY OF MEASUREMENT OF NITROGEN-15 IN BIOLOGICAL SAMPLES The nitrogen-15 enrichment of the same ammonium chloride solutions prepared from various samples of duodenal contents and rumen bacteria from a steer were measured both in the NIRD automatic analyser and in the Statron NOI-4 emission spectrometer. Sample Sample No. Duodenal contents . . .. 1 2 3 4 5 6 Rumen bacteria, C375/1 . . 1 2 3 4 5 6 and a sample of the ammonium chloride NIRD automatic Aliquot 1 Aliquot 2 0.59 0.58 0.79 0.81 1.11 1.12 1.40 1.38 1.67 1.67 1.51 1.52 w- 0.37 - 0.93 - 1.81 - 2.06 I 1.93 - 1.69 - Manual emission spectrometer Aliquot 1 Aliquot 2 0.59 0.55 0.79 0.76 1.10 1.10 1.36 1.37 1.73 1.76 1.52 1.55 7 0.39 - 0.92 - 1.83 - 2.06 - 1.94 - 1.68 - 763 prepared from each was measured, and also measured again using the manual emission spectiometer. Results from the two instruments agreed well, the automatic analyser showing less variation between duplicates than the manual analyser.The standard deviation obtained when 10 measurements were made with the NIRD analyser on a sample with a mean nitrogen-15 enrichment of 0.608 atom-% was 0.006 atom-%, similar to that obtained with standard ammonium chloride solutions. Memory efect The memory effect with the NIRD automatic analyser is very small and for practical purposes can be ignored.To test the effect four natural abundance samples, injected at l-min intervals, were followed by four samples of exceptionally high nitrogen-15 abundance (19.5 atom-%) and these were followed by five samples of natural abundance. The results (Table 111) show that even under these extreme conditions an increase in the natural abundance sample immediately following the highly enriched sample was only just detected. TABLE I11 TEST OF THE MEMORY EFFECT IN NIRD AUTOMATIC NITROGEN-15 ANALYSER Four natural abundance samples of ammonium chloride were injected at 1-min intervals and immediately followed by four samples containing 19.5 atom-% nitrogen-16 and a further five samples of natural abundance, at the same rate. Each sample contained 10 pg of nitrogen. Sample No.1 2 3 4 5 6 7 8 9 10 11 12 13 Nominal atom-% 15N 0.365 0.365 0.365 0.365 19.5 19.5 19.5 19.5 0.365 0.365 0.365 0.365 0.365 Print-out value atom-% lfiN 0.365 0.359 0.364 0.360 19.519 19.513 19.413 19.514 0.407 0.372 0.362 0.362 0.364764 GOULDEN AND SALTER : AUTOMATIC EMISSION Analyst, Vol. 104 Discussion The most important feature of the NIRD automatic nitrogen-15 analyser is the high rate of analysis that is possible (one sample per minute) compared with manually operated optical emission spectrometers. This can be achieved without sacrificing the high chemical sensi- tivity obtainable manually, and with at least as high accuracy (hO.01 atom-% in the normal operating range). It is also simple to operate, completely dispensing with time- consuming and tedious vacuum techniques.The essential development that has enabled this high throughout is the use of a rhodium - platinum catalyst for the generation of pure nitrogen. At the time when this work was started the use of such a catalyst in nitrogen-15 analysis had not been reported. Subse- quently, however, a report of the use of a rhenium-filament catalyst for the batchwise conversion of ammonia to nitrogen for the mass-spectrometric determination of 14N to l5N ratios has appeared.14 Under the conditions maintained in the analyser this method does not suffer significantly from interfering side-reactions that hinder the use of traditional methods such as that of Dumas, using copper - copper(I1) oxide, or that of Rittenberg, using alkaline hypobromite solution. The latter method has been used in an East German auto- matic nitrogen-15 analyser (Statron Isonitromat) , which was announced after the work reported here was started, and the procedures required to correct for the memory effect in this instrument result in a slower rate of analysis (18 samples per hour).I t is quoted to have a similar reproducibility (hO.01 atom-%) to that of the NIRD analyser. However, in the Statron system, measurements are made at three different wavelengths, enabling an automatic correction to be made for background emission. The use of triple wavelength measurement in conjunction with the NIRD nitrogen preparation system might reduce background drift. This drift is probably caused by changes in the background level due to emission from residual water molecules.As the effective length of the molecular sieve column becomes reduced, owing to saturation with water molecules, the drift tends to increase towards the end of a run. Similarly, a further improvement in stability with a consequent extension of the limits of accuracy below kO.01 atom-% might be achieved by replacement of the double photomultiplier measurement system with a single photo- multiplier and suitably modified electronics to allow simultaneous measurements at three wavelengths. Within the limits used, however, the gradual downward drift observed in calibration is not a serious problem, as it shows up as a change in intercept and, to a first approximation, affects all readings equally. Correction can be readily made either by adjustment of the intercept potentiometer or by analysing a natural abundance sample a t intervals throughout a run and applying the correction to the printed results.Although mass spectrometry remains the most accurate means of measuring nitrogen-15, with a reproducibility of &0.001 atom-% possible in modern apparatus, automatic optical emission spectrometry offers a number of advantages. Not least among these are the ease of operation and the much higher rate of analysis possible over extended periods. The very small amount of nitrogen required (10 pg), compared with the mass spectrometer (200- 2000 pg for highest accuracy), permits measurement of nitrogen-15 in nitrogen metabolites obtainable only in small amounts. We are grateful to Dr. E. W. Evans of the Physics Department, NIRD, for helpful criticism of the manuscript, and to the NIRD Instrumentation Section Workshop for assistance in the construction of the instrument. References 1. 2. 3. 4. 5. 6. 7. 8. 9. Rittenberg, D., in Wilson, D. W., Nier, A. 0. C., and Reimann, S. P., Editors, “Preparation and Broida, H. P., and Chapman, M. W., Analyt. Chew., 1958, 30, 2049. Meier, G., and Miiller, G., Isotopenpraxis, 1965, 1, 53. Faust. H., Isotopenpraxis, 1965, 1. 62. Lloyd-Jones, C. P., Hudd, G. A., and Hill-Cottingham, D. G., Analyst, 1974, 99, 580. Lloyd-Jones, C . P., Adam, J.. and Salter, D. N., Analyst, 1975, 100, 891. Lloyd-Jones, C. P., Adam, J., Hudd, G. A., and Hill-Cottingham, D. G., Analyst, 1977, 102, 473. Salter, D. N., and Smith, R. H., Br. J . Nutr., 1977, 38, 207. Hauck, R. D., and Bremner, J. M., Adv. Agron., 1976, 28, 219. Measurement of Isotopic Tracers,” Ann Arbor, Mich., 1947, pp. 31-42.Aztgast, 1979 SPECTROMETER FOR THE DETERMINATION OF NITROGEN-15 765 10. 11. 12. 13. 14. Goulden, J. D. S., and Salter, D. N., Proc. Nutr. Soc., 1977, 36, 132A. Conway, E. G., “Microdiffusion Analysis and Volumetric Error,” Fourth Edition, Crosby Lockwood, Goulden, J. D. S., and Salter, D. N., UV Spectrom. Gr$ Bull., 1975, 3, 74. Goulden, J. D. S., and Salter, D. N., J . Chromat., submitted for publication. Walker, R. L., Walton, J. R., Carter, J. A., and Matthews, D. R., in “Proceedings of the I.A.E.A. Symposium on Isotope Ratios as Pollutant Source and Behaviour Indicators, 18-22 November, 1974, Vienna,” I.A.E.A., ‘Vienna, 1975, p. 429. Received November 17th. 1978 Accepted February 16th, 1979 London, 1957, p. 138.