ANALYST SEPTEMBER 1986 VOL. 111 1017 Automatic Nitrogen4 5 Analyser for Use in Biological Research Joha J. Therion Animal and Dairy Science Research Institute Private Bag X2 Irene 1675 Republic of South Africa Hendrik G. C. Human and Cornelius Claase National Institute for Materials Research South African Council for Scientific Research Pretoria 000 1, Republic of South Africa Roderick I. Mackie Animal and Dairy Science Research Institute Private Bag X2 Irene 1675 Republic of South Africa and Albrecht Kistner laboratory for Molecular and Cell Biolog y South African Council for Scientific Research Pretoria 000 I , Republic of South Africa An automatic nitrogen-I5 analyser capable of analysing 12-20 samples per hour has been developed. The generation of pure nitrogen gas from the sample is carried out automatically the nitrogen isotope ratio is determined by emission spectrometry and the results are calculated as atom-% nitrogen-I 5.The total nitrogen content of the sample is also determined. Sample masses can vary between 0.5 and 100 mg and only 2-10 pg amounts of N are necessary for accurate determinations. Keywords Automatic nitrogen analysis; emission spectroscopy; nitrogen- 15 isotope ratio; automated data handling Nitrogen is one of the most important and familiar elements in the biological sciences as it is one of the main constituents of amino acids peptides proteins and nucleic acids. It is conspicuous however that little work has been carried out on the kinetics of the assimilation of nitrogenous compounds into cells their quantitative partitioning between pools and their flow to different metabolites.An important reason for this apparent lack of interest is that all the radioactive isotopes of nitrogen have short half-lives; the longest is only 10 min for nitrogen-13. The use of radioactive isotopes is therefore both difficult and expensive although nitrogen-13 has been utilised to study ammonia assimilation pathways by different bac-teria 1-5 The stable isotope nitrogen-15 has been used for studying the kinetics of ammonia assimilation by Derxia qumrnosa6 and for the dynamics of nitrogen metabolism in ruminants.7 Nitrogen-15 has also been used extensively in plant and soil studies .8,9 In most of these studies isotope-ratio mass spectrometers have been used to measure nitrogen-15; this is a relatively expensive and slow procedure although a high accuracy can be obtained.Broida and Chapmanlo introduced a photoelec-tric method to measure nitrogen-15 abundance and it became evident from this that emission spectrometry could be very useful for the isotopic analysis of microgram amounts of nitrogen. It has several advantages over mass spectrometry in that an emission spectrometer is less expensive than a mass spectrometer less demanding of the operator and the minimum amount of nitrogen required for determination is much smaller.11.12 Since 1958 emission spectrometry has been developed and applied to agricultural studies and many other research areas.12-16 However the methods for the pre-paration of samples to isolate nitrogen for isotopic analysis have so far been time consuming and have proved to be a limiting factor in extensive surveys such as nutritional studies.17 Goulden and Salter16 devised an automatic emission spectrometer capable of analysing 60 samples per hour. However biological samples had to be subjected to Kjeldahl digestion and ammonium chloride ultimately recovered, which was then injected into a reactor tube where nitrogen was generated and passed into the discharge tube of the emission spectrometer. This is a tedious process that has to be carried out manually and on average one operator can only process and measure 70 samples per week.12 This paper describes a fully automated system that measures total nitrogen and nitrogen-15 in biological materials without prior sample preparation.Experimental The system consists of two major components an automatic nitrogen analyser for determining total nitrogen and an emission spectrometer for measuring nitrogen-15. Automatic Nitrogen Analyser The automatic nitrogen analyser (Model NA1500 Carlo Erba Strumentazione Milan Italy) was used with a minor modifi-cation to the outlet tube for coupling to the emission spectrometer. A schematic diagram is shown in Fig. 1. This instrument is designed for micro and macro determinations of total nitrogen present in a wide range of organic and inorganic substances in both solid and liquid form. The analyser can operate completely automatically and is capable of perform-ing 50 analyses sequentially using the automatic sampler.It has been used with great success over a wide range of nitrogen concentrations from 10 pg to 2 mg. The instrument operates on the principle of the Dumas combustion procedure. The sample is placed in a tin container and introduced into the automatic sampler where it is purged with helium for 20 s before it drops into the combustion column which is maintained at 1000 "C. The container melts and the heat of the reaction of the tin with a small amount of pure oxygen, introduced simultaneously primes the flash combustion of the sample at ca. 1700 "C. Under these conditions even thermally resistant substances are completely oxidised. The combustion products are carried by a constant flow of helium through chromium oxide granules maintained at 1000 "C.The oxi 1018 ANALYST SEPTEMBER 1986 VOL. 111 m source 1/4m m o n oc h ro m at o r Printer - Computer Fig. 1. Schematic diagram of the automatic nitrogen-15 analyser dation is completed in a 5-cm layer of silver-coated cobalt oxide at the bottom of the combustion column. This also retains interfering substances produced during the combustion of halogenated compounds. The combustion products a mixture of C02 various oxides of nitrogen and H20 pass through a reduction reactor filled with reduced copper maintained at 650°C. At this temperat-ure the nitrogen oxides are reduced to elemental nitrogen, which together with the C 0 2 and H20 pass through two absorbent filters the first containing Mg( C104)2 to absorb water and the second containing NaOH absorbed on asbestos particles to retain CO2.The elemental nitrogen enters the chromatographic column and together with the carrier gas, flows through the thermal conductivity detector which generates an electrical signal proportional to the concen-tration of nitrogen present. The signal is amplified the area integrated and the result printed out by the data processor (DP110 Carlo Erba). By calibrating the instrument with a standard it is possible to calculate the percentage of nitrogen in the samples. Emission Spectrometer The analysis of nitrogen-15 by emission spectrometry depends OR the property of nitrogen at reduced pressure (2-10 Torr), to emit light of characteristic wavelengths when energised by radio- or microwaves. The presence of an extra neutron in the 15N nucleus causes shifts resulting in a readily measurable displacement of the bandhead that usually appears in the ultraviolet region from 297.7 nm for the l4N'4N molecule to 298.3 nm for 14N15N and 298.8 nm for 15N15N.These bands can be readily separated by a small monochromator with ca. 1 A resolution. The intensities of the bands are proportional to their nitrogen content so that in an equilibrium mixture of the three molecular forms of nitrogen the proportion of 15N can be calculated from the ratio of the bands of mass 28 and 29 in a manner analogous to that used in mass spectrometry.12 The spectrometer features a microwave source for the excitation of the nitrogen gas in a stream of He carrier gas a monochromator that scans rapidly and repeatedly over the relevant spectral region and a computer for the rapid acquisition of spectral information and data processing.Scanning Monochromator A Jarrell-Ash 0.25 m Ebert monochromator (Jarrell-Ash Division Fischer Scientific Waltham MA USA) with a 2360 lines mm-1 grating blazed for 300 nm and a dispersion of 1.65 nm mm-1 was used. Wavelength scans obtained with the monochromator using slits of 25,50 and 150 ym respectively, with a helium and natural nitrogen mixture excited by a microwave discharge at 20 W showed that the 50 pm slits were the best set to use as with these there was only a marginal loss in the definition of the spectrum compared with the 25 pm slits. With the 150 pm slits the resolution was definitely impaired. The rapid scanning over the spectral region was accom-plished with a rotating quartz plate QP (1.9 mm thick) behind the entrance slit (Fig.1). The refraction of the light beam (at approximately 300 mm) is such that a 40" rotation of the plate shifts the image at the exit slit by 0.6 nm the difference in wavelength between the two peaks to be measured. A synchronous motor is used for driving the refractor plate in order to ensure the constant speed of rotation that is necessary as the computer measures intensity at regular intervals (700 ps between measurements) so that the two peaks are separated on a time basis rather than on a wavelength basis. Constant rotational speed also eliminates the effect of a 50 Hz ripple on the emission from the microwave source and on the output of the amplifier.The speed of rotation is 1 rev s-1 resulting in a time lapse of 0.11 s between the measurement of the two peaks. The dispersed signal by photomultiplier tube 1 (PMT1) is constantly related to a signal measured by photomultiplier tube 2 (PMT2) as shown in Fig. 1. The latter measures the intensity of the 337.1 nm bandhead of nitrogen belonging to the same band system through a narrow-band interference filter that excludes any extraneous light. This serves as an efficient reference for the analytical channel to eliminate the effect of the evolutionary nature of the nitrogen content of the gas mixture given off by the source. Excitation Source A Microtron 200 Mark 111 (Electro-Medical Suppliers Green-ham UK) microwave generator (2450 MHz) was used for the excitation of the gas mixture.The instrument is equipped with a power meter a reflected power meter and a magnetron protection cut-out device that becomes operative when the reflected power exceeds 75 W. A 214L type resonant cavity accommodating discharge tubes of up to 13 mm diameter (RC) was used as the termination on the coaxial cable (Fig. 1). The source can supply power to a maximum of 200 W but the minimum of 20 W is normally used ANALYST SEPTEMBER 1986 VOL. 111 1019 Pumping System The Carlo Erba nitrogen analyser supplies the nitrogen -helium gas mixture at a flow-rate of approximately 80 ml min-l at atmospheric pressure. Inside the discharge tube the pressure must be ca. 5 Torr and the reduction in pressure is accomplished using a vacuum pump (Fig.1). The pumping speed of the rotatory vacuum pump (Alcatel 2004 A CIT Alcatel Paris France) is reduced by a fixed diaphragm D (ca. 1 mm diameter) in the line. The discharge cavity is located between this diaphragm and the needle valve V which is adjusted so that the combined effect of the pump D and V is such that the pressure on the supply side is near to atmospheric pressure in order not to interfere with the operation of the nitrogen analyser. A Pirani gauge type pressure meter (Alcatel API 101 T CIT Alcatel Paris France) is available for measuring the pressure in the discharge section whereas a diaphragm type meter measures the pressure on the inlet side. A second needle valve N is available for bleeding air into the system. As this is a constant source of nitrogen for the discharge rather than one changing with time as supplied by the nitrogen analyser it can be conveniently used for preliminary measurements and for checking the operation of the system.Electronics A single stabilised high-voltage supply (Tennelec C 952 with <3 mV regulation and less than 0.001% drift per hour 0-10 mA current) is used for the Hamamatsu 1P28 photomultiplier on the monochromator and the Hamamatsu R166 solar blind photomultiplier tube on the reference channel. The light flux is sufficiently high so that the photomultiplier tubes can be used with a high voltage supply of approximately 360 V. Each photomultiplier signal is amplified and converted by a 12-bit analog to digital converter. The parallel digital infor-mation is loaded into a Zenith microcomputer (Zenith Data Systems Illinois USA) via two parallel interface ports.An optical switch on the rotating quartz plate drive supplies a reference pulse to the computer that initiates a measurement cycle. In this way the measurements are synchronised with the quartz plate position and therefore with wavelength. The computer triggers the A/D converters for each conversion. Measurement by Computer Analysis of a sample begins when the NA1500 cycle is started at time to. After the time interval necessary for the gases to pass through the instrument the first detectable nitrogen appears at the microwave excitation source at time tl and is present until time t2. Programmed measurements begin at ta and end at tb [Fig.2(a)]. Ideally this interval should include the region of maximum intensity. The programme runs through 25 cycles of measurements in the interval ta-tb. Each cycle duration is 1 s (the period of revolution of the rotating quartz plate) but measurements are made only during a fraction of this 1 s period viz. over 80” or an 80/360 s period. During this period the spectral region of interest i. e. 297.4-298.6 nm is scanned. Only three values of intensity are required from the spectrum viz. the intensity at the strong I4N14N peak at 297.7 nm the intensity at the weak 14N15N peak at 298.3 nm and a background value. The first value is easily selected by the computer from the 320 elements stored in the memory of channel 2 by looking for the maximum of the l4NI4N peak [Ip, Fig.2(b)]. Next the intensity at the 14N15N peak is located 56 measurements after the first peak ( I p +56) and this separation remains fixed. Similarly the 93rd measurement after the first peak (Ip +93) represents a reliable position (wavelength) at which to measure the background intensity [Fig. 2(b)]. These three values from channel 2 are normalised with respect to Measure and display I I 1 1 1 I I + 56 I + 93 Fig. 2. spectrum showing the positions at which measurements are made (a) Sequence of events during an analytical cycle; and (b) N channel 1 the reference channel by dividing by the corre-sponding values of the averaged set of 25 reference intensities. Results and Discussion Evaluation of Data Although the background on the long wavelength side of the 14N15N peak appears flat and reproducible it does not represent the real background at the wavelength of this peak, as using this value for calculating the I4N15N to 14N14N peak ratio yields a value of ca.1 50 instead of the value of 1 137 for natural (0.365 atom-%) 15N abundance. This is due to the fact that the emission peak of the 14N2 molecule at 297.7 nm produces a “skirt” of spectral emission towards the long wavelength side that interferes at 298.3 nm the wavelength of the 14N15N molecule maximum. This is a real feature of the spectrum and not due to inadequate resolution by the monochromator. A sample of known 15N abundance is therefore used to establish the background value. For this purpose a sample at natural abundance has proved to be adequate.Measurements are made as follows the intensity of radiation is measured at the 14N14N peak the 14N15N peak and at a minimum in the spectrum on the long wavelength side. These values are called P1 P2 and B respectively. The concentration of 15N C is given by c= - loo (inatom-%) . . . . . . 2R + 1 where P i - A P2 - A R E - . . . . . . . . For a sample of natural abundance R = 137 and the value of the true background A is given by . . . . . . . RP2 - Pi R - 1 A = - (3 1020 ANALYST SEPTEMBER 1986 VOL. 111 Comparing this value with B the background measured away from the peak it was found that the ratio k = A/B was both positive and constant regardless of the signal sizes. The signal size was varied in a 1 8 range by varying the nitrogen concentration and also by varying the high voltage on the photomultiplier tube but no correlation of this ratio with the signal size was found.After establishing the value of k accurately by running three or more samples of natural abundance samples of unknown 15N content can be run and the concentration calculated from equation (1) by substituting kB for A the true background in equation (2) so that R can be defined in terms of the background measured away from the peak ( B ) and the concentration of 15N (C) calculated without needing to know the true absorbance. Hence, 100 2R + 1 c= -where Pi - kB P2 - kB R=-Following this procedure and running a number of natural abundance samples as unknowns a standard deviation of better than 0.02 atom-% of 15N is usually found.Repeated analyses were carried out on defined inorganic and organic substances to test the accuracy and reproducibility of the results obtained with the nitrogen analyser (ANA). These are shown in Table 1. Excellent reproducibility was obtained with all three substances with coefficients of variation between 0.3 and 0.85%. The agreement between the known nitrogen content of the compounds and the values obtained was very good. The difference was 0.05 for (NH4)2S04 whereas it was 0.08 for haemin an organic compound from which it is difficult to release nitrogen. The difference for albumin an organic macromolecule was only 0.01. Table 2 shows a comparison of nitrogen analyses of biological samples performed using the Kjeldahl method and the ANA respectively.Coefficients of variation varied between 1.5 and 2.7% which is good considering the heterogeneity of the samples and with the maize cell walls, the extremely low nitrogen content. It is important to note that the accuracy with which the sample is weighed into the tin cup will influence the performance of the ANA. The use of large samples would therefore increase the accuracy of the method although the size of the tin cup dictates the sample size to a great extent especially when using biological samples of low density. The sample masses shown in Table 2 were the maximum that could be fitted into the sample cups except for the duodenal contents and sheep faeces where smaller samples were also used.From the results of the latter two substances it seems that the problem of heterogeneity of the samples can be overcome by carrying out repeat analyses on the same batch of material. The linearity of the measurements performed on the emission spectrometer was determined using 15N-labelled (NH4)2S04 standards (obtained from Isotope Services Los Alamos National Laboratory Los Alamos NM USA). Table 3 shows the excellent agreement obtained between the nominal values of the standards and those measured with the automatic emission spectrometer. A correlation co-efficient of 0.9996 was found with a y-intercept of -0.061. The slight deviation from a straight line with a y-intercept = 0 could be attributed to the fact that a higher degree of accuracy can be obtained with a mass spectrometer.The memory effect of the automatic 15N analyser is not Table 1. Accuracy and reproducibility of nitrogen analyses performed using the automatic nitrogen analyser Sample (NH4)2S04 . . . . Haemin . . . . Mass/ mg 5.07 5.04 5.05 5.05 5.00 5.00 10.07 10.05 10.05 10.00 10.05 10.06 Atom-% of '5N Measured Mean Expected SD cv 21.29 21.15 21.10 0.094 0.44 21.20 21.15 21.17 21.08 21.02 8.49 8.51 8.59 0.025 0.29 8.53 8.53 8.48 8.53 8.54 Albumin . . . . 14.20 14.96 15.09 15.10 0.121 14.19 14.96 14.23 15.19 14.22 15.20 14.19 15.04 14.24 15.21 0.79 Table 2. Determination of atom-% 15N in biological samples using Kjeldahl procedures and the automatic nitrogen analyser Mean mass/ Sample mg Bacteria .. . . 2.04 Maizecellwalls . . 6.51 Duodenal contents 2.04 10.02 Sheepfaeces . . . . 2.02 10.04 Atom-% N in sample Kjeldahl ANA SD cv 6.66 6.66 0.136 2.0 0.14 0.18 0.004 2.27 3.30 3.85 0.105 2.74 3.30 4.09 0.064 1.57 2.51 2.41 0.038 1.59 2.51 2.51 0.056 2.2 ANALYST SEPTEMBER 1986 VOL. 111 0.357 0.373 0.391 0.342 0.356 , 1021 ' b 0.364 5 0.019 Table 3. Measured 15N atom-% in a range of 'SN standards. Standards obtained from Isotope Services Los Alamos National Laboratory, Los Alamos NM USA Measured atom-% 1sN Nominal atom-% 15N x (n = 6) SD cv 0.37 0.358 0.006 1.68 0.5 0.481 0.018 3.74 1.0 0.898 0.022 2.45 2.0 1.764 0.025 1.42 5.0 4.528 0.081 1.79 10.0 9.521 0.118 1.24 Table 4. Test of the memory effect in the discharge tube of the ADSRI nitrogen-15 analyser Atom-% 1SN Sample No.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Nominal 0.365 0.365 0.365 0.365 0.365 16.603 16.603 16.603 16.603 16.603 0.365 0.365 0.365 0.365 0.365 significant and can be ignored. The results in Table 4 show that even under extreme conditions where the highly enriched samples were immediately followed by a sample at natural abundance no significant increase could be detected. Goulden and Salterl6 also found no memory effect when they subjected the NIRD automatic analyser to a highly enriched sample followed by one at natural abundance. The most important feature of the automatic nitrogen-15 analyser is the fast rate of analysis possible (12-20 samples per hour compared with 6-8 samples per week with manually operated emission spectrometers1*).It is also more advanced than the automatic system described by Goulden and Salter,16 as biological samples can be analysed directly without prior Kjeldahl digestion and subsequent recovery of ammonium chloride. Further the system is fully automated with a direct printout of atom-% 15N for each sample. Although mass spectrometry remains the most accurate means of analysing for nitrogen-15 with a reproducibility of 0.001 atom-% possible in modern apparatus automatic emission spectrometry offers a number of advantages. The most important are the much faster rate of analysis possible and the very small amount of nitrogen (10 pg) required for accurate determinations compared with the 200-2000 pg required for determination by mass spectrometer.The auto-matic nitrogen-15 analyser described in this paper is therefore ideally suited for biological research where metabolites are often only available in small amounts. The authors thank the Protein Research Fund Protein Advisory Committee Republic of South Africa for providing funds for the construction of the instrument. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References Kenealy W. R. Thompson T. E. Schubert K. R. and Zeikus J. G. J. Bacteriol. 1982 150 1357. Kim H-C. and Hollocher C. H. J. Bacteriol. 1982,151,358. Meeks J . C. Wolk C. P. Thomas J. Lockau W. Shaffer, P. W. Austin S. M. and Galonsky A. J. Biol. Chem. 1977, 252 7894. Meeks J. C. Wolk C. P. Lockau W. Schilling N. Shaffer, P. W. and Chien W.-S. J. Bacteriol. 1978 134 125. Wolk C . P. Thomas J. Shaffer P. W. Austin S. M. and Galonsky A. J . Biol. Chem. 1976 251 5027. Wang R. and Nicholas D. J. D. Phytochemistry 1985 24, 1133. Leng R. A. and Nolan J. V. J. Dairy Sci. 1984 67 1072. Bremner J. M. J . Assoc. Off. Anal. Chem. 1985 68 155. Yamamuro S. Soil. Sci. Plant Nutr. 1981 27 405. Broida H. P. and Chapman N. W. Anal. Chem. 1958,30, 2049. Ohmori M. Iizumi H. and Hattori A. Anal. Biochem., 1981 111 83. Salter D. N. Proc. Nutr. SOC. 1981 40 355. Ito O. Yoneyama T. Akiyarna Y. and Kumazawa K., Radioistope (Jpn.) 1976 25 448. Fiedler R. and Proksch G . Plant Soil 1972,36,371. Lloyd-Jones C. P. Adam J. Judd G. A. and Hill-Cottingham D. G. Analyst 1977 102 473. Goulden J . D. S . and Salter D. N. Analyst 1979 104 756. Salter D. N. and Smith R. H. Br. J . Nutr. 1977 38 207. Paper A6157 Received February 20th 1986 Accepted April 7th 198