Refinement and evaluation of an automated mass spectrometer for nitrogen isotope analysis by the Rittenberg technique

An apparatus designed to automatically perform hypobromite oxidations of ammonium salt samples for nitrogen isotope analyses with a mass spectrometer was modified to improve performance and reduce analysis time. As modified, reference N2 is admitted to the mass spectrometer between samples from a dedicated inlet manifold, for calibration at the same pressure as that of the preceding sample. Analyses can be performed on samples containing 10 μg to 1 mg of N (or more), at a rate of up to 350 samples/day. When operated with a double-collector mass spectrometer, the standard deviation at the natural abundance level (10 analyses, 50-150 μg N) was <0.0001 atom % 15N. Very little memory was observed when natural abundance samples (0.366 atom % 15N) were analysed. following samples containing 40 atom % 15N. Analyses in the range, 0.2 to 1 atom % 15N (50-150 μg N), were in good agreement with manual Rittenberg analyses (1 mg N) using a dual-inlet system, and precision was comparable. For enrichments of 2 to 20 atom % 15N, automated analyses were slightly lower than manual analyses, which was attributed to outgassing of N2 from the plastic microplate used to contain samples.


Introduction
Use of nitrogen-15 as an isotopic tracer has been stimulated by the automation of mass spectrometers for nitrogen isotope analysis. Automation may be accomplished by interfacing an automatic N/C analyser (ANCA) to an isotope ratio mass spectrometer [1][2][3][4]; the combination is commonly referred to as ANCA-MS.
During analyses, N in the sample is converted to N and N oxides by flash combustion of a Sn sample container ("-1700C) in the presence of CuO and a catalyst (usually CrO3). Reduction of the N oxides to N occurs as the combustion products are swept over Cu at 600 C. The N is purified by gas chromatography, and a small fraction (-1%) of the effluent is admitted to the mass spectrometer for measurement of the ion currents at rn/z 28, 29 and 30, from which both total N and 15N are determined.
Nitrogen isotope analyses can also be performed by the Rittenberg technique, which utilizes alkaline hypobromite to oxidize ammonium (NH4+) to N in the absence of air. This technique, named after its originator, has been the method of choice for manual N isotope analyses for more than 50 years. Conversion of N in the sample to NH4 + is commonly carried out by the Kjeldahl method [5], which involves digestion with concentrated H:SO4 to convert organic forms of N to NHa+-N, followed by steam distillation of the digest with alkali.
To test the automated mass spectrometer developed at the Los Alamos National Laboratory, thousands of analyses were performed for scientists engaged in N isotope research. Eventually, a private business was established to continue this service (Isotope Services, Los Alamos, New Mexico), using the same type of automated mass spectrometer. The second instrument incorporated several refinements, including the use of disposable microplates to contain samples and the capability for multiple loading of plates [7].
A commercial system based on the design of McInteer and Montoya [6] was recently developed for automation of a mass spectrometer in the authors' laboratory [8].
Experience gained in the operation of this system for analyses of more than 30 000 samples has led to several modifications which improve performance. The major modifications are described in this article together with an evaluation of analytical performance.  PT, pressure transducer (MAAS Model TDI01); VG, vacuum gauge (thermocouple); R, regulator. The ARA and a dual-inlet system connect to the mass spectrometer via manually operated bellows valves (Nupro Model SS-4H). operation with a double-collector mass spectrometer (Nuclide/MAAS Model 3-60-RMS) [8]. The mass spectrometer is also equipped with a conventional dual-inlet system. This system consists of a vacuum manifold with two variable volume metal bellows (10--300 ml), matched viscous leaks and a set of four changeover valves (Nupro Model SS-4BK-1 C) to sequentially admit sample and reference gases to the mass spectrometer and to a waste pump (an 11 1/s Perkin-Elmer Ultek ion pump). Connection of the ARA and dual-inlet system to the mass spectrometer is made via a stainless-steel tee with three bellows valves (Nupro Model SS-4H), which allows either dual-inlet or automated operation to be conveniently selected.
The ARA operates under computer control to liberate N2 from a preselected number of NH4 + salt samples, regulate the pressure of N2 liberated, admit the N2 to the As originally designed [8], the ARA was equipped with a single inlet manifold, from which either sample or reference N2 was admitted to the mass spectrometer. To increase throughput capacity, multiple samples could be analysed between calibrations; however, higher precision was achieved by analysing reference N after every sample. Ideally, the latter approach is carried out using separate inlet manifolds for sample and reference N, so that calibration can be accomplished as soon as possible following isotope-ratio analysis of the sample, without prolonging the operational routine. This is achieved with the design illustrated by figure 1 An important consideration in the design of the ARA is the internal volume of the inlet manifold, which must be minimized to obtain sufficient pressure from microgram quantities of N2 for analyses with the mass spectrometer [6,7]. To meet this requirement, the original design utilized 1/16th in o.d. tubing and tube fittings and miniature solenoid valves [8]. Subsequent work showed that a further increase in pressure could be attained by reducing the internal volume of the sample drawback valve (valve 8 in figure ), used to introduce Freon during purging of the sample well and admit N2 to the sample manifold following hypobromite oxidation of the NH4 + salt sample. The reduction in volume was achieved by replacing the body and plunger of the original three-way valve (Honeywell-Skinner Model B13ADK1150), which had 1/8th in NPT fittings, with those from a two-way valve (Honeywell-Skinner Model B2DA9400) having 1/16th in PTF fittings. The new plunger, like the original, was crossdrilled to ensure complete purging of air or N2   + + Purging was performed for the period specified following an initial purge for 10 and a 4-min period during which the sample well was pressurized with Freon (3-4 psig).
from a previous sample. The transducer used to measure pressure in the sample manifold (PT1 in figure 1), originally mounted to the body of the sample drawback valve, was relocated on the other side of the cold trap. In this position, the transducer is protected against corrosion in the event that hypobromite is admitted to the manifold during drawback (i.e. due to malfunction).
As specified previously [8], LiOBr is used to oxidize NHe+-N to N. The LiOBr is purged with He to exclude any air. It is stored in a sealed bulb under a He atmosphere during use. Experience revealed the need to maintain a constant pressure of He in the bulb as the volume of LiOBr decreased during operation, so He is now supplied at 1-4 psig via the stopcock used as a vent in purging (figure ). To reduce contamination of the LiOBr by air during delivery to the reaction head, the Tygon INLET PRESSURE (torr) Figure 2. Ratio data for different inlet pressures of reference N2.
The software supports sensing and control functions via the PDISO-8 board, and enhances the control capabilities of the IEEE-488 interface utilized for I/O connection to the integrating ratiometer and valve interface [8]. For example, ion accelerating voltage is automatically controlled such that the measured ratio is 29N2/(2aN 2 + 3N2) for samples containing <5 atom % 15N, and 3N2/ (aN + gN) for higher enrichments. In all cases, ratio measurements are performed only after voltage data have been collected to check for numerator overflow. If overflow occurs, inlet pressure is reduced until the condition is no longer detected, and the ratio analysis is then performed.

Operation
Analyses with the modified ARA follow the routine listed below (numbering of valves follows figure 1)" (1) In standby status, valves and 5 are open to evacuate the sample and reference inlet manifolds.
All other valves are closed.
(2) Upon receiving keyboard input to begin a run, the x-y table loads a sample tray (loading can be done from a stack of up to eight trays) and moves it to position the starting well (96 wells/tray) beneath the reaction head.
(3) The reaction head drops, establishing a gas-tight seal with the well.
(4) The sample well is purged with Freon by opening valves 9 and 11. Initial purging is for 10 s, during which the flow rate is measured, and valve 10 is opened briefly to eliminate air from the line that connects this valve to the exhaust line. Valves 9 and 11 are then closed, leaving the sample well under a positive pressure of Freon to promote exchange with air adsorbed on the plastic surface. After approximately 4 min, valves 9 and 11 are opened for 10-60 s to remove residual air. (9) The reaction head is raised after momentarily opening valve 9 to pressurize the drawback line (this eliminates a partial vacuum from drawback that would otherwise lead to the entry of air into the drawback line and enhance jarring of the tray as the head is raised) and then operating the peristaltic pump for a few seconds to eliminate any voids from the reagent line. The head is cleaned by positioning a polypropylene box with the x-y table, lowering the head onto paper towels in the box, opening valves 9 and 10 for a few seconds and then raising the head.
(10) The sample tray is moved with the x-y table to position the next well beneath the reaction head.
(11) The head is lowered, and purging is performed for 10 s as described in step 4. Valves 9 and 11 are then closed to maintain a Freon atmosphere in the sample well.
(12) Filling of liquid N is enabled.
(13) Pressure in the sample manifold is measured with the pressure transducer (PT1). If the measured pressure (Pro) is < torr, the analysis is aborted, and clean-up of the cold trap occurs. If Pm exceeds a specified inlet regulation pressure (Pr 1-10 torr), pumping is accomplished by opening valve 7. Initial pumping may be continuous (Pro > 1'25 Pr), in which case the pressure is also monitored continuously; final pumping (Pr < Pm 1"25P) is intermittent, with valve 7 being opened momentarily after a 3-s delay for pressure measurement. The final sample pressure (Ps), obtained with (Ps P) or without (Ps <-P,.) regulation, is stored in memory.
(14) Valve 6 is opened to admit sample N2 to the mass spectrometer.
(15) A delay of 30-60 s is provided to allow equilibration of the ion source. Immediately after beginning this delay, valve is closed to isolate the reference inlet manifold, and valve 2 is opened for a few milliseconds to admit reference N2 from a 1-1 bulb.
Pressure in the reference manifold is then measured with the pressure transducer (PT). If the measured pressure (Pro') is <Ps, drawback via valve 2 is repeated. If Prn' is > l'25Ps, valve.3 is opened, and Prn' is continuously updated. Valve 3 is closed when P,,' is <-Ps, or upon completing the delay period. (16) Ratio data are collected for sample N2 after checking for numerator overflow. If preliminary measurements ofgN/(aN + N) indicate a 15N content >5 atom %, the ion accelerating voltage is decreased to measure 3N/(SN2 + 29N). (17) Valve 5 is opened to evacuate residual N from the sample manifold. A delay of 15-60 s ensures complete evacuation. During this period, final pumping of reference N is carried out by intermittently opening valve 3 until Pro' is --<P.
(18) Valve 6 is closed to isolate the sample inlet manifold from the mass spectrometer, and valve 4 is opened to admit reference N. (20) Ratio data are collected for reference N2 after checking for numerator overflow. As data collection begins, valves 9 and 11 are opened to complete purging of the next sample well. Purging is terminated (i.e by closing valves 9 and 11) after 10-60 s, which may occur during or after data collection. Heating of the cold trap is concluded during data collection.
(21) Upon completing the analysis of reference N, atom % 15N is calculated for the sample (see table   1), and a report is printed. Valve is opened to evacuate the reference manifold. Evacuation of the sample manifold is completed by closing valve 7 and opening valve 5. The routine continues via step 5.

Evaluation
The automated mass spectrometer is designed to process large numbers of samples without operator intervention. When purified to remove N2 [7,8], Freon-12 (CC12F) is an excellent purge gas for the ARA. However, international concern over depletion of stratospheric ozone by chlorofluorocarbon (CFC) refrigerants has led to recent legislation that progressively restricts their consumption during the 1990s, and bans all further production as of 2000 [9]. This ban does not apply to hydrochlorofluorocarbon (HCFC) refrigerants, which have much shorter atmospheric lifetimes than CFCs, and, hence, lower potential for depletion of ozone [9]. The latter group includes Freon-22 (CHC1F), which is commonly used in air-conditioning systems. Data in table 2 show that Freon-22 is a satisfactory purge gas for the ARA. As with Freon-12, purification is necessary to remove N. This can be accomplished using the apparatus described previously [8]; however, a higher pressure is generated with Freon-22 than with Freon-12 (the pressure during purification of Freon-22 will reach approximately 200 psig, as opposed to 130-140 psig for Freon-12), and the bubbler valve (used to exhaust Freon contaminated with N) must be heated to maintain a stable flow rate.
Attempts to use CO2 as a purge gas were unsuccessful. due to formation of Br upon reaction of LiOBr with H2CO.
In a previous evaluation of analytical precision with the ARA [8], analyses of 15N-enriched (NH4)2804 were found to decrease, and analyses of 15N-depleted (NH4)SO4 to increase, with decrease in sample size from 100 to 20 btg of N. These findings were attributed to isotopic contamination from atmospheric NHa [8]. An important consideration in the use of mass spectrometry to measure isotope ratios is that ratio measurements depend directly upon inlet pressure. This is illustrated by figure 2, which shows data obtained for reference N when the ratio, 29N2/(28N + N), was measured using inlet pressures of to 8 torr. In every case, the measured ratio was higher than the theoretical value (0"00735 for the natural abundance level of 15N).
Analyses of sample N2 are subject to the same error, which necessitates the use of reference N2 so that a ratio difference can be measured. An accurate value of atom % 15N can then be obtained for the sample, ,provided both analyses were performed at the same pressure of N.
Calibrations with the ARA were originally carried out by regulating the same inlet pressure for sample and reference N2 [8], but this was found to give a larger ion current at m/z 28 + 30 (which is proportional to the pressure of N2) for the reference N2 than for the sample N2, owing to a lower content of non-liquid N2 condensable impurities, and the result was overestimation of atom % 15N. To avoid such difficulty, regression equations were developed that relate the inlet pressure of sample or reference N2 to the ion current generated at m/z 28 + 30. These equations are utilized to obtain the same ion current with sample and reference N. Their effect on analytical performance is apparent from table 4, which summarizes data obtained when analyses were performed at different inlet pressures with and without the use of regression for pressure regulation of reference N2. Without regression, atom % 15N was consistently overestimated, and serious error occurred at low inlet pressures. With regression, overestimation was largely eliminated, although precision was limited at torr by the tolerance in pressure regulation (+5%), due to the marked effect of low inlet pressures on ratio measurements (see figure 2).
As noted previously, the drawback valve utilized in the ARA is equipped with a crossdrilled plunger to ensure that all traces of N2 from a previous sample are removed during purging with Freon. Moreover, strong pumping is provided by a 175 1/s diffusion pump connected to the inlet manifolds, and by a 50 1/s turbomolecular pump connected to the ion source of the mass spectrometer [8].
The combination is so effective that memory is virtually nonexistent, even when a natural abundance sample is analysed following a sample that contains 40 atom % 15N (see table 5).
To evaluate the accuracy and precision of analyses with the ARA, a comparison was made to manual Rittenberg analyses with the dual-inlet system, using (NH4)2SO4 at eight different 15N concentrations between 0"2 and 20 atom %. Sample size for the ARA usually ranged from 10 to 150 gg of N, but in one case the range was from 10 to 1000 big of N. Manual analyses were performed on 1000 tg of NH4+-N, which was oxidized to N2 in disposable glass vials [10]. Table 6 summarizes the data obtained.
Examination of table 6 reveals that the accuracy and precision of analyses with the ARA depend upon the amount of N in the sample and the 15N content. Best results were achieved with 50-150 btg of N, and an 15N concentration between 0"2 and atom %. Accuracy and precision were sacrificed with smaller samples and/or higher 15N enrichments, presumably due to isotopic dilution by trace amounts of natural abundance N2 outgassed from the plastic sample tray. No difficulty was encountered in the analysis of samples containing 250-1000 btg of N; however, the values obtained for 15Nenriched (NH4)2SO4 (0"5 atom % 15N) were somewhat lower than with 50-150 tg of N, which can be attributed to isotopic fractionation from incomplete oxidation of NHe+-N to N2 (the 0" ml of LiOBr added is sufficient to oxidize approximately 200 tg of N).