MIRA: a new isotope ratio mass spectrometer for clumped isotope ...
1
MIRA: a new isotope ratio mass spectrometer for clumped isotope studies of CO2 Paul F Dennis∗ , Stuart Vinen, Alina Marca-Bell, Peter J Rowe Stable Isotope Laboratory, School of Environmental Sciences, University of East Anglia, Norwich, Norfolk, NR4 7TJ, UK ∗ E-mail: [email protected]
Abstract Accurate measurement of multiply substituted isotopologues using an isotope ratio mass spectrometer (IRMS) imposes stringent requirements on instrument performance in terms of overall sensitivity, abundance sensitivity and linearity. To meet these requirements we have designed, constructed and evaluated the performance of a new IRMS, optimising its performance for clumped isotope ratio measurements of CO2 . We determined an instrument and magnet geometry that combined excellent ion optics and transmissivity with relatively straightforward engineering such that the instrument could be self-built. MIRA (Multi-Isotopologue Ratio Analyser) is a 120o magnetic sector isotope ratio mass spectrometer with a 25cm physical radius and extended geometry for an effective 50cm dispersion. It is equipped with a Nier type high sensitivity electron impact ion source and six faraday cup collectors for measurements of all the cardinal isotopologue masses of the CO2 molecule. The mass spectrometer is constructed using UHV techniques, compatible materials and components. Modular high voltage electronics modules are used for the source power supplies and instrument control and data acquisition electronics are based on National Instruments compactRIO hardware and LabVIEW software. In dual-inlet mode the MIRA mass spectrometer has enhanced sensitivity (approximately 600 CO2 molecules ion−1 ) combined with high abundance sensitivity and a linear response in terms of measured ∆47 over a greater than 100‰ range in measured bulk isotope composition δ 47 with respect to the working reference gas. Precision of measured ∆47 is at the shot-noise limit of 0.008‰ for standard measurement cycles. When compared with commercially available instruments the MIRA mass spectrometer offers enhanced performance for clumped isotope ratio measurements of CO2 . The high sensitivity combined with the measurement linearity allows for increased overall sample throughput whilst minimising the number of calibration samples that need to be run in order to correct for non-linearity in ∆47 measurements.
2
Introduction Multiply substituted isotopologues are molecules, formula units and other moities in which there are two or more isotopic substitutions by rare heavy isotopes. In geochemistry they are more commonly and colloquially referred to as clumped isotopes and over the past ten years there has been a growing interest in measuring the variability of the relative abundance of clumped isotopes in naturally occuring materials [1, 2]. The most widely developed application has been isotope thermometry of carbonate minerals, particularly those that are characteristic of conditions at the surface and in the upper crust of Earth and even Mars [3–7]. However clumped isotope studies in other systems, including tropospheric and stratospheric CO2 [8–11] and stratospheric O3 [12], point to potentially important applications across a wide range of problems in the atmosphere, biosphere and solid earth. To date all published measurements, bar one, have been made using commercially available ThermoFinnigan 253 gas source isotope ratio mass spectrometers that have been modified to include a six faraday cup collector array with amplifier gains optimised for measurement of isotopologues in the CO2 and O2 systems [13]. The exception is a single study describing the performance of a Thermo-Finnegan Delta XP IRMS for clumped isotope measurements [14]. These studies demonstrate that it is possible to achieve high precision measurements on multiply substituted isotopologues that are present in low abundance. For example the excess abundance of the m/z = 47 isotopologue in CO2 (predominantly
18
O13 C16 O)
which occurs with a nominal stochastic abundance of just 44ppm in CO2 that is characteristic of the VPDB standard can be measured to a relative precision of better than 0.01‰ ( [13] and references therein). The excess abundance of an isotopologue is expressed as [1]:
∆i =
!
" Ri − 1 × 1000 Ri∗
where Ri represents the measured abundance ratio of isotopologue i relative to that of the non isotopically substituted isotopologue, and R∗i represents the expected abundance ratio for the isotopologue when all the isotopes are stochastically distributed. The stochastic abundance ratio is calculated from the measured bulk isotopic composition of the sample. Not with standing the fact that high precision measurements of ∆47 in CO2 can be made the measurement procedure requires careful calibration in order to correct for analytical artefacts that are instrument
3 dependent. The two most important of these are: (i) so called non-linearity of the measured ∆47 composition as a function of the bulk isotopic composition of the sample, and (ii) scrambling of the clumped isotope signal. The non-linearity refers to an increasing divergence between the measured and true ∆47 composition with difference in bulk isotopic composition between the sample and instrument working reference gas. 47 This is characterised as a linear trend in plots of ∆47 versus δ(sam−wrg) for gas samples that have been
heated at 1000oC in order to randomise the isotope distribution (the so-called heated gas line) [13]. To correct for this effect measured ∆47 compositions are extrapolated parallel to the heated gas line 47 towards a zero δ(sam−wrg) value. The gradient of the heated gas lines are subject to differences between
individual instruments and to secular variations with both long term drift and sudden step changes occurring. These can sometimes be related to changes in instrument tuning, filament changes, or even vacuum pump maintenance. Whatever the cause it is necessary to monitor carefully the heated gas line by regular measurements of heated gases interspersed with samples to ensure the integrity of results. The origin of the measured dispersion in ∆47 compositions is not fully understood. Careful observations have suggested that there is a pressure dependent negative, peak baseline shift at all the minor isotopologue peaks when a sample gas is introduced to the mass spectrometer [15, 16]. This may be the result of one or more factors including: (i) insufficient screening of secondary electrons and negative ions generated in the flight tube post magnetic separation that perhaps result from the interaction between a broad, non-gaussian tail of the major ion beam with the flight tube and/or faraday cup assembly; (ii) loss of secondary ions from the faraday assembly, and (iii) electronic issues associated with the high gain (1012 ) amplifiers used for registering the minor isotopologue beams. Whilst further work is required to fully understand the phenomenon a number of different correction schemes for the baseline shift have been proposed to correct for the ∆47 dispersion [13, 15, 16]. Scrambling in the ion source results from ion-molecule interactions and recombination of CO2 ion fragments resulting in a degree of randomisation of the isotope distributions and a scale compression of the clumped isotope signal. The degree of scale compression appears to be a function of source pressure, ion residence time [13] and possibly electron current and energy with values ranging from factors of 0.9 to 0.7. Since one is wanting to measure to high precision small excess values on rare clumped isotopes it is advantageous to maximise the signal by using high source pressures and electron currents thus whilst the full details of the electron-molecule/ion interactions in the source are not fully understood it is important
4 that they are subject to detailed investigation. In the absence of a phenomenological or constitutive description of the scrambling effects a protocol for normalising analytical data to a theoretical reference frame scaled according to statistical mechanical calculations of the expected isotopologue excess based on spectroscopic data for CO2 has been proposed and implemented by most laboratories [17]. The method is based on measuring gases that have a varying degree of isotope ordering as a result of equilibration at different temperatures and comparing the measured excess to theoretically calculated values. When results from different laboratories are scaled appropriately the data are in good agreement. Given the growing interest in clumped isotope measurements and the known instrumental effects that affect current instruments we have designed a new isotope ratio mass spectrometer that optimises performance especially in respect of linearity and sensitivity. The MIRA (Multi Isotopologue Ratio Analyser) employs a novel symmetric 120o extended geometry that optimises transmission and beam quality. In this paper we describe the ion-optic design, basic construction details and analytical performance of MIRA. In particular we demonstrate that the MIRA design is linear with no dispersion of the measured ∆47 value with bulk isotopic composition. Currently MIRA uses a standard Nier type electron impact source. We have conducted experiments to help understand the electron-molecule interactions in this source and the processes that lead to scrambling of the clumped isotope signal and will use this information in the design of a new high sensitivity source for clumped isotope studies.
Instrument design Instrument geometry and ion optics The geometry and ion optical properties of the MIRA instrument are illustrated in Figure 1. The on-axis trajectory has a physical radius, ρo , of 250mm and deflection angle, Θ, in the magnetic field of 120o . The ion beam has positive oblique entrance and exit angles, #’ and #”, of 40.89o. The distance from the source slit to the magnet entrance and from the magnet exit to the resolving slit is 288.7mm (1.1547 x ρo ). The oblique magnet entrance and exit angles of the ion beam result in an extended dispersion of 500mm and stigmatic focussing for optimum transmission. The symmetric geometry has unit magnification in both the x-y and x-z planes. In terms of effective dispersion and image aberrations to the second order 90o and 120o extended geometry instruments have a similar performance. However, we chose the 120o geometry because it offers
5 a shorter overall path length between ion source and resolving slits. Reducing the path length results in less ion-molecule collisions and beam broadening ultimately resulting in improved beam profile quality, and reduced ion scattering and secondary electron generation from interactions between the beam and ion optic elements (flight tube, lenses, resolving slits etc). These considerations may be an important contribution to eliminating the effect of non-linearity in clumped isotope measurements [15, 16]. Trajectories for the on- and off-axis beams were determined using a spreadsheet program, TRAJCALC. This program calculates the first- and second-order field expansion coefficients for ion trajectories in the mid-plane of uniform wedge field magnets of different geometries [18, 19]. Combined with trajectories computed for the pre- and post-magnet field free drift regions it is possible to determine the focal plane geometry and second-order aberrations for both on-axis and off axis beams. Figure 2 shows the detailed trajectories for the inner, central and outer rays of the m/z = 44-49 beams plotted with respect to the ion optic, or x-axis. These cover the 6 cardinal masses of the CO2 isotopologues. The ray trajectories were calculated for a maximum beam divergence of ±0.033 radians exiting the source slit. This is characteristic of the Nier type ion source used in MIRA. The focal plane is oriented at an acute angle of approximately 28o with respect to the optic axis. Unit mass beam separations are approximately 10mm for the CO2 isotopologues. At the focal point beam images are a maximum 1mm wide. These widths are calculated assuming a line source of negligible width at the source slit. In practice the ionisation region in the source is projected by a 3-element immersion lens on to the source slit and has a width much less than that of the source slit width of 0.6mm. Thus the width of the source image at the resolving slits is dominated by the second-order aberrations. Contributions from chromatic aberrations due to the energy spread of ions exiting the source have been neglected in these calculations. This is reasonable given the very low ∆E of electron impact ion sources. Using a resolving slit width of 2mm will theoretically produce well resolved peaks with flat tops and a mass resolution of 250 (dispersion/resolving slit width). On-axis trajectories have also been computed for the x-z plane and are plotted in Figure 1b. The schematic diagram shows the strong focussing action of the magnet fringing fields. Parallel trajectories originating from the source slit (5mm long) are brought to a focus within the magnetic field. The acceptance envelope for trajectories oriented obliquely to the optic axis is shown by the dash-dot line in Figure 1b. Maximum deviations from the optic axis are + and - 0.017 radians for beams originating from the bottom and top of the source slit respectively. This is restricted by placing a 5mm aperture located
6 close to the magnet entrance within the flight tube. The envelope for these beams has a maximum width of 12mm at the magnet exit. To minimise ion scattering from the upper and lower surfaces of the flight tube we have designed this with an internal height of 12mm. This is wider than typical flight tubes in commercial IRMS instruments which tend to have widths of close to the source slit height.
Ion source The ion source is based on a standard Nier type [20,21] with transverse beam electron impact ionisation. The detailed geometry was optimised using the SIMION 3D ion trajectory modelling package (INEEL95/0403). The source consists of a small, enclosed ionisation cavity with an ion exit slit (5mm x0.6mm). This slit forms the first active component of a three element immersion lens consisting of the ion exit slit, the half- or focus plates and the source slit. The focus plates are separated by a gap of 2mm, and the source slit is sized 5mm x 0.5mm. Following the source slit is a further α slit located 50mm beyond the source slit and used to limit the overall beam divergence in the x-y plane. The final active component of the ion source is a small repeller plate located towards the rear of the ionisation chamber. The repeller provides fine control of the field gradient immediately behind the ion-exit slit and in conjunction with the focus potential affects overall beam quality as well as the residence time of ions in the source. The transverse ionising electron beam is collimated by a cross-axis magnetic field generated by two small ALNICO permanent magnets. Modelling of ion trajectories using SIMION shows that appropriate selection of focus and ion repeller potentials for a source potential of +8kV produces a strongly demagnified image of the ionisation region projected onto the plane of the source slit and forming a beam with a narrow beam divergence (< ± 0.03 rads) in the x-y plane. In the x-z plane the beam is extracted as a ribbon of parallel trajectories 5mm high and oriented parallel to the x axis. Ions extracted close to the bottom and top of the slits are deflected by fringing fields and follow trajectories oriented at both positive and negative angles with respect to the x-axis. The majority of these are intercepted by the α slit and don’t pass into the flight tube.
Faraday cup detectors The faraday detector array consists of six individually screened, deep faraday cups with integral resolving slits and secondary electron supression, Figure 4. Each cup is 35mm deep x 5mm wide x 15mm high
7 and is machined from a single block of stainless steel with side plates of 0.2mm thick stainless steel spot welded on to form the cup sides. The insides of the cups are coated in graphite. The slit array has a front resolving aperture (2 x 5mm), followed by an electron suppression plate with an aperture of 2.5 x 5.5mm and a final screening aperture (3 x 6mm). The faraday cups and integral slit array are mounted onto individual supports using ceramic rods with 1mm spacers between all the elements. The front faraday support has a deep aperture that is 7mm wide x 20mm high x 8mm depth. This provides an effective screen between adjacent cups that collects scattered secondary ions and electrons originating from interaction between the resolving slit edges and ion beam. Each faraday cup assembly is covered by a stainless steel wrap around shield. The individual faraday assemblies are mounted onto a base plate that has been photo-etched with locating slots that allow for fine adjustment of position in both x- and y-directions. This ensures exact registration of the faraday cups with the six CO2 beams and that the resolving slits are located on the focal plane of the instrument.
Electromagnet The electromagnet is constructed from 10cm thick milled Maximag high purity iron plates (Tennant Metallurgical Group Ltd). Three sections are used for the top, bottom and back of the yoke. The pole faces are fabricated from two further 10cm thick sections with care taken to ensure accurate machining of the magnet entrance and exit angles. The whole assembly is accurately aligned using precision dowels and bolted together. The coils consist of 2x600 turn assemblies made from 2mm diameter polyester resin coated copper wire wound onto armatures that fit around the pole faces. The total coil resistance is approximately 10Ω. The total power dissipation for a 0.35T field is approximately 125W and it is not necessary to cool the magnet or coils.
Dual inlet MIRA uses a standard VG micro dual-inlet system from a SIRA series II instrument. The inlet has a sample cold finger, reference and sample gas variable volume bellows, matched 50cm long capillaries and change-over valve for rapid switching between sample and reference gas. The working volumes for the sample gas cold finger, sample and reference bellows are respectively 2 x 10−4 L, 2-40 x 10−3 L and 5-100
8 x 10−3 L. The cold finger volume in the sample block is carefully matched by a dummy volume on the reference gas valve block. For measurement an aliquot of gas at 160mb pressure is isolated in these volumes and subsequently allowed to deplete during the measurement cycle. A 20 minute data acquisition results in a 20% depletion of gas. Using matched volumes and carefully balanced capillaries ensures identical sample and reference gas depletion rates and no drift in measured isotope ratios during the run. The sample and reference gas variable volumes are used to balance pressures and signals prior to the measurement run. A typical sample measurement is averaged over 5 acquisitions, with each acquisition consisting of 20 reference-sample cycles. Each reference and sample cycle has a 10s deadtime following switching of the changeover valve and a 20s integration period. Total integration periods are 2000s each for the sample and reference gas.
Vacuum system and construction MIRA is constructed using UHV techniques with all metal sealed conflat (CF) type flanges. The source and detector housings are made from single blocks of 316LN stainless steel with ports machined directly into the blocks using standard CF knife flange geometries. The top and bottom sections of the flight tube are milled from 316LN plate and welded together with full penetration welds to eliminate any potential trapped volumes. The flight tube is attached to the source and detector housing using 316LN flanges. All the internal surfaces of the machined blocks and flight tube are electropolished and heat treated under vacuum to reduce outgassing rates. To ensure a clean, oil free vacuum the mass spectrometer is differentially pumped using three Leybold TW70H 70L.s−1 drag stage turbomolecular pumps (Oerlikon Leybold Vacuum GmbH). One pump is dedicated to pumping the source gas load, whilst the analyser is pumped by two pumps located at either end of the flight tube at the source and detector housings. The high vacuum pumps are backed by two Leybold DiVac 0.8T diaphragm pumps running in series. In this mode backing pressures are reduced to ca. 0.1mb and a clean vacuum with ultimate pressure of 1 x 10−9 mb is readily achieved. The source and analyser vacuum are monitored by two cold cathode ion gauges located either side of the differential pumping pressure step in the source housing. The dual-inlet change-over valve is pumped by a dedicated 70L.s−1 turbomolecular and diaphragm pump, whilst the sample and reference gas inlet blocks and valves are pumped via a further 70L.s−1 pump
9 backed by two diaphragm pumps operated in series. Roughing down the inlet from atmospheric pressure is provided for by a valved connection to the intermediate stage between the two diaphragm pumps.
System electronics, instrument control and data acquisition Source high voltage power supplies The source high voltage power supplies are of a modular design and built using Applied Kilovolts HV units for the source and focus plate potentials, and an Applied Kilovolts floating filament power supply unit for the filament (ITT Exelis Ltd). Sitting on top of the source and focus potentials are UltraVolt programmable floating power supply units to provide bias voltages for the filament, ion repeller, trap voltage and ∆ focus for beam steering (UltraVolt Inc). Twelve volt DC power to drive the UltraVolt units and bi-directional analogue channels for programmable voltage control and read-back are provided by UltraVolt TNFL1 15kV isolated units. The trap current is monitored using a current sensing resistor and a high common mode voltage op-amp (AD629). The output of the op-amp is used as input to a PID circuit for control of the filament current (see below).
Electromagnet power supply The electromagnet is powered by an Amcron Macro-Tech 601 600W audio amplifier with a DC input stage. The coils have a resistance of 10Ω at ambient temperatures and a current of approximately 3.5A is required to produce a field strength of 0.35T. The voltage gain of the amplifier is 20x such that an input voltage of 1.75V is needed for a 35V output. Even at a relatively low mass resolution of 250, and with good laboratory temperature control (±1o C) it is still necessary to use current or field feedback in order to stabilise the magnetic field. We use the voltage drop across a low value current sensing resistor (0.1Ω) with a very low temperature coefficent of resistance as the process input into a PID control to maintain field stability to better than 1 in 5000.
10
Head amplifiers and signal integration The ion currents at m/z = 44, 45 and 46 are amplified by low noise transimpedance amplifiers (AD549L) with gains of 108 , 1010 and 1010 respectively and time constants of 0.1 second. Ion currents at m/z = 47, 48 and 49 are amplified by ultra low noise transimpedance amplifiers (Femto Messtechnik GmbH model LCA-2-10T) operated with a gain of 1012 and a time constant of 1 second. All the amplifiers are individually shielded and mounted directly onto the signal feedthroughs of the faraday array flange. The output voltages are buffered and processed via synchronous voltage to frequency converters (AD652). To optimise linearity these are operated at a clock frequency of 1MHz giving a maximum count rates of 5x105 Hz for a 10V signal. High precision ∆47 , ∆48 and ∆49 measurements requires low amplifier input bias currents with minimal drift. Typical bias currents for the LCA-2-10T amplifiers are 10fA with temperature induced drift of ca. 1fA per degree C. To keep the absolute drift to a minimum value of less than 0.5fA the temperature of the amplifier and VFC circuits is regulated to better than 0.5 degree C. An offset of 0.5fA results in an error of 0.03 per mille in measured δ 47 compositions when the sample and reference gas differ by 100 per mille. This is at the level of precision expected for a single δ 47 acquisition of 20 minutes and is deemed acceptable given that at smaller differences in bulk isotope composition between the mass spectrometer working reference gas and sample gas the error is proportionately smaller.
Instrument control and data acquisition System control and data acquisition electronics are based on National Instruments compactRIO technology with an NI cRIO-9004 Real Time Controller interfaced to a cRIO-9104 unit consisting of an FPGA back plane and eight ports for analogue and digital signal conditioning and input-output modules (National Instruments Inc). Analogue input signals (vacuum gauges, source read back parameters, and trap current feedback) are processed by two NI 9205 32 channel, 16-bit analogue to digital converters. Two NI 9263 4 channel, 16-bit digital-analogue converters are used for all the system analogue control voltages (0-10 volts). An NI 9476 32 channel 24V sourcing digital output module is used to control pneumatic interfaces for low and high vacuum isolation valves, and valves in the micro dual-inlet. Fast digital signals, both input and output, are managed by two NI 9401 8 channel high speed digital
11 input/output modules. These modules provide the TTL clock frequency (1MHz) for the VFC’s and act as the interface for the VFC outputs. FPGA (Field Point Gate Array) and Real-Time controller technology allows rapid development and deployment of system control and data acquisition functions that would conventionally require the design and construction of dedicated electronic circuits. For example we have configured the Real-Time controller to carry out the feedback system control functions for the ion source (filament current) and the electromagnet. Using proprietary NI PID modules implemented in the LabVIEW Real-Time programming module the electron beam trap current is regulated to better than 0.1% and the field to 1 part in 5000. The signals from the six faraday cups are counted using a six channel leading edge counter implemented on the FPGA. Timing for the counter and the clock pulses for the external VFC circuits is provided by the 40MHz clock on the FPGA thus enabling a fully synchronous VFC-counter circuit. Counter outputs are transferred via an FPGA enabled FIFO buffer to the Real-Time controller and from there to the desktop PC for display and data storage.
Performance In this section we characterise the performance of the MIRA IRMS for clumped isotope analysis concentrating on (i) beam quality as defined by mass resolution, abundance sensitivity and flatness of the peak top; (ii) overall sensitivity in terms of sample molecules per ion detected, (iii) system linearity with respect to measured ∆47 and ∆48 versus the sample bulk isotope composition as represented by the δ 47 and δ 48 composition. Finally we note that MIRA exhibits a greater degree of scale compression for clumped isotope measurements than other instruments reporting that this effect appears to be constant and readily calibrated. In a separate paper we present the results of preliminary experiments aimed at understanding this effect (unpublished results) and anticipate that the results of these studies will be used to inform the design of a new high sensitivity source.
Beam quality and peak shape Examples of the peak shape and quality are shown in Figure 4. The peaks were recorded at a major beam intensity of 4.2x10−8A and an analyser vacuum of 2x10−8mbar. In panel A the output of all six faraday collectors are recorded as the source HV voltage is scanned over a span of 60V at a nominal 8kV
12 acceleration voltage. The six cardinal masses for the CO2 isotopologues are in excellent registration with the beam centres within four volts of each other, and the peaks having flat base lines, steep sides and broad flat tops. This is characteristic of the fact that the resolving slits are accurately located on the focal plane of the instrument. The mass resolution, defined as M/∆M = V/∆V where M is the mass of interest, ∆M is the peak width in amu, V is the accelerating voltage at which the peak appears and ∆V is the peak width in volts at half the peak height (fwhm definition), is 235. This is in excellent agreement with the theoretical mass resolution of 250 for an instrument with 500mm dispersion and resolving apertures 2mm wide. An estimate of the beam width at the resolving slit of the faraday detectors is given by the distance over which the beam intensity rises from 10% to 90% of its maximum value as it crosses the edge of the aperture. The 10-90 beam width is 0.34mm, somewhat less than the beam width of 1mm calculated assuming a maximum beam divergence of ±0.033 rads shown in Figure 2. This latter width defines the outer envelope of the beam and for flat topped peaks must be less than the resolving slit width. To test the beam quality and flat topped nature of the peaks we have plotted the beam ratios 45/44, 46/44 and 47/44 over a 10V span in Figure 4a. The y-axis of these plots is scaled in per mille and covers a range of -40 to +40 per mille deviation in the measured ratio. All three ratios are flat to within 2‰ over the whole of the 10V range. Whilst the overall beam profile and peak quality of MIRA is excellent a major concern for clumped isotope ratio measurements are the corrections required for dispersion of the measured ∆47 and ∆48 compositions as a function of the bulk isotopic composition of a sample. This is the so-called non-linearity effect observed in existing commercial IRMS instruments. Whilst there is a good phenomenological description of the effect it’s cause is not fully understood though recent work strongly supports the idea that secondary electrons generated by the major beam and which are not fully suppressed are a major contributing factor [15, 16]. These electrons cause negative shifts in the baseline signal for the weak clumped isotope peaks at masses 47 to 49. The location for generation of the secondary electrons has not been identified but is likely to result from a broad, markedly non-gaussian distribution of ions in the tails of the major peak. These ions are not collected by the faraday detectors and may interact with the flight tube and ion-optic elements to generate the secondary electrons. Another effect resulting from broad tails associated with the major beams is to lower the abundance sensitivity of the instrument, defined as the signal intensity in the tail normalised with respect to the peak height at distances of one, two or
13 more amu displacement from the major peak. Whilst for routine isotope ratio measurements abundance sensitivity is not usually an issue one needs to be aware of potential problems when considering clumped isotopes. The isotopologues at masses 47, 48 and 49 are nominally 40, 4ppm and 40ppb respectively of the mass 44 peak intensity. As with scattered secondary electrons a low abundance sensitivity can also lead to non-linearity in clumped isotope measurements. Accurate measurement of these species over a wide compositional range of 100 per mille requires abundance sensitivities of better than 1 in 108 and 1 in 109 respectively for the major beam at 3 and 4 amu distance. It is not easy to measure such high abundance sensitivities. In Figure 4b we have plotted the peaks at masses 48 and 49 together with the lower portion of all the peaks as all six peaks are simultaneously scanned and collected. In doing this test we have not compensated for the input bias current of the amplifiers. All six peaks show the same basic characteristics. Towards the high mass side (lower voltage) the peaks demonstrate a markedly sharp rise. On the low mass side (higher voltage) all the peaks show a more marked decay towards the baseline. Such a shape is characteristic of ion beams in mass spectrometers where ion-molecule collisions tends to modify the ion energy distribution with a tail towards lower energy. This is reflected in the tail on the low energy (low mass) side of the mass spectrum. Not-with-standing this tail the baseline for all the peaks is flat and at the level of the amplifier input bias currents as measured when there is no beam in the instrument (typically 5 to 10 fA). Thus we can exclude any contribution of a secondary electron current to the measured beam signals. Similarly, the flat baselines are strongly suggestive of a very high abundance sensitivity. Overall MIRA has excellent ion optics resulting in high beam quality and peak shapes that are characterised by a sharp rise and decay, with broad, flat tops across which the measured isotope ratio is stable. Abundance sensitivity is high and there is no contribution of secondary electrons to the minor ion beam signals.
Sensitivity We determined the absolute sensitivity of the MIRA mass spectrometer under typical analysis conditions following the method of Prosser [22]. The reference gas side of the mass spectrometer was filled with CO2 to a pressure of 160mb. This is the inlet pressure required for a major beam signal of 4.2x10−8A. The dummy cold finger volume (0.2mL) was then isolated and the signal monitored whilst the gas depletes. The change in the ion signal as a function of time is given by:
14
I = Io .e(−c/v).t where I is the ion signal at time t, I0 is the ion signal at time t = 0, c is the capillary conductance in units of m3 s−1 and v is the volume (m3 ) from which the analyte gas is depleting. The capillary conductance for a 20% depletion from the cold finger volume over a 20 minute run cycle is 3.8x10−11m3 s−1 (Figure 5). Assuming that the ion signal is directly proportional to the gas pressure in the cold finger volume then at t = 0 the gas flow rate into the ion source is given by: dµ −c = .µ0 dt v where µ0 is the initial amount of CO2 in the volume v in molecules, c is the capillary conductance, v is the volume of the inlet dummy cold finger. µ0 is calculated as 8.6x1017 molecules and the molecular flux into the source at t = 0 as 1.63x1014 molecules.s−1 . The ion signal at t = 0 is 4.24x10−8A, or 2.65x1011 ions s−1 . Thus the sensitivity is 615 molecules.ion−1 Considering the measured sensitivity in terms of overall efficiency of ion extraction from the ion source and transmission through the analyser we can describe the measured ion current from a Nier type electron impact ion source as [23]:
I + = β.Qi .Se .I − .p where I+ is the measured ion current (A), Qi is the differential ionisation coefficient (m−1 ).Pa−1 , Se is the effective path length of the ionising electron beam (m), I− is the ionisation (trap) current (A), p is the source pressure (Pa) and β is the combined source and transmission efficiency. For the MIRA source I+ is 4.2x10−8A, I− is 750x10−6A, Se is taken as the source slit length of 5x10−3m. The source pressure is estimated using kinetic theory and the equation for gas transport in the molecular flow regime through a thin aperture [24]:
µ = (P1 − P2 ).A.
!
R.T 2.π.M
"0.5
where µ is the gas flux (N.m.s−1 ), P1 and P2 are the pressures inside and outside the source box
15 respectively (Pa), A is the cross sectional area of the source slit aperture (m2 ), R is the gas constant (J.K−1 .mol−1 ), T absolute temperature (K) and M the mean molecular mass (kg.mol−1 ). Since the source pressure P1 is very much greater than the pressure P2 outside the source box we can make the approximation P1 = (P1 - P2 ). The measured gas flux of 1.63x1014 molecules.s−1 determined above for intitial gas flow through the capillary converts to a gas flow rate of 1.06x10−6 N.m.s−1 at an assumed source temperature of 200o C. Using this value yields a source pressure of 2.25x10−3Pa. We can now estimate the overall source efficiency as:
β=
I+ Qi .Se .I − .p
The differential ionisation coefficient Qi is estimated as 7.6m−1 .P a−1 using the reported cross sections for electron impact ionisation of CO2 by 70eV electrons [25,26]. Together with the derived source pressure and source design parameters we determine an overall efficiency of 67% i.e. 2 out of every 3 ions produced within the active region of the ion source are extracted, transmitted and then detected by the faraday cups. The overall sensitivity and efficiency of MIRA is commensurate with, but certainly towards the high end of the performance range for IRMS instruments fitted with electron impact ion sources. For example the Europa 20-20 IRMS (now Sercon 20-22 IRMS) equipped with a similar ion source but operating at 2.7kV has a sensitivity of 780 molecules.ion−1 [22]. From the data supplied by Sercon we have calculated an efficiency of 49% for the 20-20 and 20-22 instruments. The improvement in sensitivity and efficiency of MIRA over other similar analysers probably reflects a combination of several factors including: (i) enhanced ion extraction efficiency resulting from the higher source potential (8kV compared to 2.7kV) and higher field gradients in the ion extraction region of the source, and; (ii) an improved overall instrument transmission from source to detector.
Linearity As previously discussed a key performance parameter for clumped isotope measurements is the overall linearity of the measured ∆47 composition as a function of the difference in bulk isotope composition (δ 47 and δ 48 ) between the mass spectrometer working reference gas and sample gas [13, 15, 16]. Linearity is
16 evaluated using sample gases with different bulk isotopic compositions and known ∆47 and ∆48 values that have been prepared by either heating CO2 at high temperature (1000oC) in order to produce a close to stochastic distribution of the isotopes, or equilibrating CO2 with water at lower temperatures (e.g. [17]). In Figure 6 we plot the results of ∆47 and ∆48 measurements on heated and water equilibrated gases covering a range of δ 47 and δ 48 compositions. The gases were prepared using BOC cylinder gas (δ 13 C = -34.34‰V P DB ; δ 18 O = 10.29‰V SMOW ) and CO2 produced by reaction of BDH Carrara marble chips, BDHCM, with phosphoric acid at 25o C (δ 13 Cvpdb = 1.98‰V P DB , δ 18 Ovsmow = +39.30‰V SMOW ) as two end members, and a 50:50 mixture of the two gases. In order to: (i) cover a wider compositional range, and; (ii) check that there are no systematic offsets associated with differences between the sample and reference sides of the dual-inlet and changeover valve we used the two different end member gases (BOC and BDHCM) as working reference gas in the mass spectrometer. The BOC CO2 was used as received whilst we equilibrated the BDHCM CO2 with water at 25o C in order to bring it’s ∆47 composition closer to equilibrium at ambient temperatures. The total range in δ 47 and δ 48 compositions measured with respect to the working reference gas are on the order of 120‰ from -60 to +60‰. As the reference gases have different ∆47 and ∆48 compositions to facilitate comparison we have corrected our results to the absolute reference frame [17]. In making this correction we have adjusted for scale compression but do not make any correction for non-linearity. As shown in Figure 6a and b the measured ∆47 and ∆48 compositions for both the 1000o and 25o C gas samples are invariant with respect to the bulk isotope composition of the gas. These results were obtained over a 9 month period using different reference gases and show no secular variations, or sudden step change. During this time we have changed the reference gas several times, the ion source filament and modified several of the electronic circuits. Thus we are confident that the linear behaviour of MIRA is an inherent feature of its design. Also plotted in Figure 6 is the region of non-linear behaviour for 1000oC heated gases reported for measurements made on a Thermo-Finnigan 253 IRMS [13]. These authors report both secular drift and step changes in the dispersion. Similar non-linear behaviour has been observed by other groups including laboratories at Harvard, Yale, Johns Hopkins, Chicago and Zurich [15–17]. The dispersion in ∆47 ranges from 0.0064 to 0.0128‰ ∆47 per ‰ δ 47 and from 0.125 to 0.35‰ ∆48 per ‰ δ 48 [13]. At large differences in bulk isotopic composition of 70‰ the absolute magnitude of the dispersion for ∆47 is on the order of the total range in expected ∆47 composition for naturally occurring samples. For ∆48 measurements the
17 dispersion is several orders of magnitude greater than the expected range of natural samples. Though we observe that the MIRA mass spectrometer is linear with respect to ∆48 the external measurement precision is such that it is not possible to resolve systematic changes in ∆48 composition as a function of equilibration temperature i.e. the 1000o and 25o C equilibration lines in Figure 6b are indistinguishable. ∆48 is very sensitive to sample contamination (e.g. [13]) and we use positive values greater than 1 as an indication that the sample may be subject to contamination and reject the ∆47 measurement. Internal measurement precisions for ∆48 on single samples are on the order of 0.05 to 0.1‰ and we conclude that present preparation and gas purification methods are not yet sufficient to produce sample gases that will yield a meaningful ∆48 measurement.
Scrambling and scale compression During ionisation of the analyte gas there is potential for reordering of the isotopes amongst the ensemble of molecules in the ion source moving compositions closer towards a stochastic distribution. This scrambling process leads to scale compression in clumped isotope measurements [13, 17]. We are publishing a phenomenological description of the scrambling effect in a separate paper. It is likely that the key processes that cause scrambling are ion-molecule and dissociative ionisation-recombination reactions with threshold energies of 10-20eV within the high pressure region of the source. The degree of scrambling is dependent on both source pressure and the ion residence time and is described by the function (Dennis, unpublished results):
ln
∆∗47 = k.t.P ∆47
where ∆∗47 and ∆47 are the true (non-scrambled) and scrambled compositions respectively, k is a constant, t is the residence time and P the source pressure. i.e. the degree of scrambling is dependent on the probability of an ion-molecule interaction. The effect of scrambling is shown in Figure 7 in which the theoretical ∆47 compositions of sample gases equilibrated with water at temperatures of 25o , 50o and 80o C and heated to 1000oC are plotted versus the measured ∆47 compositions determined with respect to the mass spectrometer working reference gas. The theoretical compositions were estimated using the data of Wang et al. (2004) [1] [17]. The empirical lines plotted in Figure 7 correspond to the different reference gas compositions used
18 over the past year: BOC cylinder gas and a reference gas prepared by reacting BDH Cararra Marble chips with phosphoric acid and then equilibrating the gas with water to draw the ∆47 composition closer to that of an ambient temperature CO2 (see discussion on linearity above). There is a linear relationship between ∆47 expressed in the absolute reference frame and the measured ∆47 composition. The two lines are parallel to each other with a gradient of 1.3233. This represents an overall scale compression of 24.5%. The different line intercepts result from the fact that the mass spectrometer working reference gases have different states of isotope ordering depending on their source, or how they were prepared. Note that neither the BOC, or BDHCM gases are in clumped isotope equilibrium at 25o C. Scale compressions reported for other instruments tend towards a lower value, ranging from 1 to 12.5% [13, 14, 17]. The ion source in MIRA and that in the Thermo-Finnigan mass spectrometers are of different designs. Given the high efficiency of the MIRA source (see above) the operating pressure can’t be significantly greater than that of the Thermo-Finnegan design. Thus we conclude that the greater scale compression that we observe in MIRA results from a longer ion residence time in the source. We are currently re-designing the source in order to minimise the residence time and reduce scale compression. It has also been reported that scrambling and scale compression is subject to secular changes [13]. These are possibly related to subtle variations in source tuning affecting ion residence time and source pressure required to sustain the measurement signal. We are unaware of any other reported studies of the long term stability of scale compression during clumped isotope measurements but expect subtle secular variations to be the norm. However, our experience is that provided the source is operated under identical tuning and with a constant source pressure, then the degree of scale compression is stable and not subject to secular drift. Over the past year we have not had to vary source tuning and overall source and instrument sensitivity has remained constant.
Conclusions We have described the design and construction of a new IRMS, MIRA, that has been specifically optimised for clumped isotope ratio measurements. The main aim of the design was to produce an instrument with a completely neutral response such that the measured ∆47 and ∆48 compositions are independent of the bulk isotopic composition of the sample being measured. This would remove the need for careful calibration of any non-linear response that would be needed to correct for the dispersion in ∆47 and ∆48
19 that is evident in commercially available instruments. Thus special care has been paid to ensure that ion-optic aberrations are small particularly with reference to components of the beam that contribute to a significant, broad non-gaussian component. This combined with careful design of the faraday cup detectors ensures that there are no contributions from secondary ions and electrons to the measured beams, particularly for the minor, multiply substituted isotopologues. With a standard design Nier type electron impact ion source the instrument performance is exceptional. Overall sensitivity is very high at 615 CO2 molecules per detected ion. This represents a combined source and transmission efficiency of 69%. Beam and peak quality in terms of overall shape, flatness and baseline stability are all very high. The ion beam has a narrow 10-90 width of 0.34mm, significantly smaller than the resolving aperture width of 2mm, with a negligible non-gaussian component in the beam tails. This contributes to a sharp rise and fall of the peaks with broad flat tops whilst retaining an overall mass resolution of 235 at an operating HV of 8kV. The peak tops are flat to within 2‰ over a greater than 10V window. There is no measurable secondary ion, or secondary electron component to the measured ion beams, particularly for the multiply substituted isotopologues at masses 47 to 49. This is reflected in the flat response of the instrument when comparing measured ∆47 and ∆48 compositions across a greater than 100‰ bulk composition range. We have observed no change in this response over more than nine months of running. We do observe a higher degree of scale compression at 24.5% than is seen with other instruments. However this is stable and not subject to secular variation. Calibration of the local ∆47 scale for different working reference gases to the absolute reference frame show that the compression is independent of working reference gas composition. MIRA allows routine measurement of clumped isotope compositions at the shot noise limit of precision for ∆47 (< ±0.01‰) for samples as small as 5mg with 2000s integration of sample and reference beams and overall measurement times of 100 minutes. The limit to minimum sample size is currently fixed by the minimum bellows and cold finger micro-volume on the dual inlet. Sample consumption during a measurement is 3.2x10−5 bar.L. This suggests that with optimisation of the inlet volumes samples as small as 1.6x10−4 bar.L can be readily run with a 20% depletion. This is equivalent to approximately 750 microgrammes for carbonate mineral samples.
20
Acknowledgments PFD gratefully acknowledges the support of the University of East Anglia who provided the funding for design and development of the MIRA mass spectrometer. SJV acknowledges the support of NERC and Sercon via a CASE studentship. ADM and PFD acknowledges NERC support and funding via grants NE/H012311/1 and NE/E010105/1. Sercon Ltd continue to provide financial support for the ongoing development of MIRA and the authors acknowledge the constructive discussions and long term support that Sam Barker of Sercon has given the project. PFD is very grateful to Dave Harbour of Isotope Precision Ltd who provided guidance, support and all the engineering expertise required to see through the project from its original concept to completion. Chell Instruments constructed the instrument.
21
References 1. Wang Z, Schauble EA, Eiler JM (2004) Equilibrium thermodynamics of multiply substituted isotopologues of molecular gases. Geochimica et Cosmochimica Acta 68: 4779–4797. 2. Eiler J (2007) “Clumped-isotope” geochemistry—The study of naturally-occurring, multiplysubstituted isotopologues. Earth and Planetary Science Letters 262: 309–327. 3. Came R, Eiler J, Veizer J, Azmy K, Brand U, et al. (2007) Coupling of surface temperatures and atmospheric CO2 concentrations during the Palaeozoic era. Nature 449: 198–2007. 4. Affek HP, Bar-Matthews M, Ayalon A, Matthews A, Eiler JM (2008) Glacial/interglacial temperature variations in Soreq cave speleothems as recorded by ‘clumped isotope’ thermometry. Geochimica et Cosmochimica Acta 72: 5351–5360. 5. Da¨eron M, Guo W, Eiler J, Genty D, Blamart D, et al. (2011)
13
C18 O clumping in speleothems:
Observations from natural caves and precipitation experiments. Geochimica et Cosmochimica Acta 75: 3303–3317. 6. Swanson EM, Wernicke BP, Eiler JM, Losh S (2012) Temperatures and fluids on faults based on carbonate clumped-isotope thermometry. American Journal of Science 312: 1–21. 7. Halevy I, Fischer WW, Eiler JM (2011) Carbonates in the Martian meteorite Allan Hills 84001 formed at 18 ± 4 ◦ C in a near-surface aqueous environment. Proceedings of the National Academy of Sciences 108: 16895–16899. 8. Eiler JM, Schauble E (2004) 18 O13 C16 O in Earth’s atmosphere. Geochimica et Cosmochimica Acta 68: 4767–4777. 9. Affek HP, Eiler JM (2006) Abundance of mass 47 CO2 in urban air, car exhaust, and human breath. Geochimica et Cosmochimica Acta 70: 1–12. 10. Affek HP, Xu X, Eiler JM (2007) Seasonal and diurnal variations of
13
C18 O16 O in air: Initial
observations from Pasadena, CA. Geochimica et Cosmochimica Acta 71: 5033–5043.
22 11. Yeung LY, Affek HP, Hoag KJ, Guo W, Wiegel AA, et al. (2009) Large and unexpected enrichment in stratospheric
16
O13 C18 O and its meridional variation. Proceedings of the National Academy of
Sciences 106: 11496–11501. 12. Yeung LY, Young ED, Schauble EA (2012) Measurements of 18 O18 O and 17 O18 O in the atmosphere and the role of isotope-exchange reactions. Journal of Geophysical Research 117: 1–20. 13. Huntington K, Eiler J, Affek H (2009) Methods and limitations of ’clumped’ CO2 isotope (∆47 ) analysis by gas-source isotope ratio mass spectrometry. Journal of Mass Spectrometry 44: 1318– 1329. 14. Yoshida N, Vasilev M, Ghosh P, Abe O, Yamada K, et al. (2013) Precision and long-term stability of clumped-isotope analysis of CO2 using a small-sector isotope ratio mass spectrometer. Rapid Communications in Mass Spectrometry 27: 207–215. 15. He B, Olack GA, Colman AS (2012) Pressure baseline correction and high-precision CO2 clumpedisotope (∆47 ) measurements in bellows and micro-volume modes. Rapid Communications in Mass Spectrometry 26: 2837–2853. 16. Bernasconi SM, Hu B, Wacker U, Fiebig J, Breitenbach SFM, et al. (2013) Background effects on Faraday collectors in gas-source mass spectrometry and implications for clumped isotope measurements. Rapid Communications in Mass Spectrometry 27: 603–612. 17. Dennis KJ, Affek HP, Passey BH, Schrag DP, Eiler JM (2011) Defining an absolute reference frame for ’clumped’ isotope studies of CO2 . Geochimica et Cosmochimica Acta 75: 7117–7131. 18. Brown KL, Belbeoch R, Bounin P (1964) First- and Second-Order Magnetic Optics Matrix Equations for the Midplane of Uniform-Field Wedge Magnets. Review of Scientific Instruments 35: 481–485. 19. Wollnik H (1987) Optics of Charged Particles. Academic Press. 20. Nier AOC (1940) A mass spectrometer for routine isotope abundance measurements. Review of Scientific Instruments 11: 212–218. 21. Nier AOC (1947) A mass spectrometer for isotopic and gas analysis. Review of Scientific Instruments 18: 398–406.
23 22. Prosser SJ (1996) A novel magnetic sector mass spectrometer for isotope ratio determination of light gases. International Journal of Mass Spectrometry and Ion Processes 125: 241–266. 23. Elliot RM (1963) Ion Sources. In: McDowell CA, editor, Mass Spectrometry, McGraw-Hill. 24. Berman A (1992) Vacuum Engineering, Calculations, Formulas and Solved Exercises. Academic Press. 25. Rapp D, Englander-Golden P (1965) Total Cross Sections for Ionization and Attachment in Gases by Electron Impact. I. Positive Ionization. The Journal of Chemical Physics 43: 1464–1479. 26. Itikawa Y (2002) Cross Sections for Electron Collisions With Carbon Dioxide. Journal of Physical and Chemical Reference Data 31: 749–767.
24
Figure Legends Figure 1.
The geometry of the MIRA mass spectrometer. The MIRA instrument uses a
120o , symmetric extended geometry with non-normal magnet entrance and exit trajectories for enhanced dispersion and stigmatic focussing. The physical radius is 250mm with source slit to magnet entrance and magnet exit to resolving slit distances of 288.7mm. The boundaries of the magnetic field are rotated by 40.89o with respect to the normal of the beam trajectories. In the x-z plane the flight tube height of 12mm allows full transmission of all ion trajectories exiting the source thus minimising scattered elements of the beam that may result from interaction between the ion beam and flight tube. Magnification in both the x-y and x-z planes is unity. Figure 2. Focal plane trajectories for the six CO2 cardinal masses. Focal plane solution for the six CO2 beams between m/z =44-49. The inner, outer and central rays of beams are plotted in the x-y plane of the instrument for an initial beam divergence from the source of +/-0.033 radians. The focal points are separated by a distance of approximately 10mm normal to the beam trajectories, with the focal plane oriented at approximately 28o to the x-axis. Figure 3. Faraday cup assembly schematic. The faraday cups (C) are machined from a block of stainless steel with the sides spot welded on. The internal surfaces of the cups are coated in graphite. The lens element consists of a front resolving slit (0V), a central electron suppression plate (-36V), and to ensure minimal field gradients across the aperture of the faraday cup a second screening plate (0V). The cups are supported in individual screened cases using ceramic rods and spacers that are tensioned with a nut assembly. Figure 4. Peak shapes and beam quality. Characteristic peak shapes for all six cardinal masses of the CO2 molecule are plotted in panel A. Note the excellent peak quality with a steep rise and fall on the peak sides and a broad flat top. The peaks centres are coincident to within four volts. Beam ratios over a 10V span, indicated by the outline box superposed on the peaks, are stable to better than 2‰, Figure 4a top. Overall beam quality is excellent with a very high abundance sensitivity and absence of any secondary electron currents introducing baseline offsets, panel B. The peak base lines coincide with the amplifier input bias currents on both the high and low mass side. Figure 5. Source and instrument sensitivity. A plot of the major beam intensity as a function of time recorded during a typical analysis (open circles: reference, filled circles: sample). The sample and
25 reference gases are isolated in identical, matched volumes (0.2mL) with an initial pressure of 160hPa. As the gas depletes the signal strength reduces. The capillaries are crimped to ensure identical conductance. The run consists of 20 cycles of reference-sample pairs with a 10 second dead-time following switching of the changeover valve followed by a 20 second integration period. The total run time is 1200 seconds during which time there is an approximate 20% depletion of the sample and reference gas corresponding to an average gas utilisation rate of approximately 1.2x10−9 mols.s−1 over the 20 minute measurement period. The overall sensitivity is 620 CO2 molecules per ion detected (see text). Figure 6. Linearity. Plotted in panel A are the ∆47 compositions expressed in the absolute reference frame [17] of CO2 gas with differing bulk isotopic compositions that have been equilibrated at either 1000o or 25o C. The data have been corrected for scale compression and have had no linearity correction applied (see text). The compositional range measured with respect to the mass spectrometer working reference gas is greater than 120‰ from approximately -60 to +60‰. The horizontal lines are the theoretical ∆47 compositions for 1000o and 25o C [1]. The MIRA mass spectrometer has a completely neutral response. For comparison the range of linearity reported for the Thermo-Finnigan IRMS is bounded by the two 48 sloping lines plotted in panel A [13]. Panel B is a plot of the raw measured ∆48 versus δsam−wrg for
the same gases. Again the MIRA response is neutral over a wide compositional range however it is not possible to resolve the 25o C (squares) from the 1000oC (circles) data. Non-linearity ranges for the Thermo-Finnigan 253 instrument lie between the two sloping lines [13]. Figure 7. Scrambling and scale compression. Measured compositions of 25o , 50o, 80o and 1000oC equilibrated CO2 gases plotted versus theoretical ∆47 compositions [1]. Data are plotted for two different reference gases: (i) BOC cylinder gas, and; (ii) CO2 produced by reacting BDH Carrara Marble chips with phosphoric acid and partially equilibrating the gas with water at 25o C. Note that both reference gases are not in equilibrium at ambient temperatures. The conversion between scales has a correction factor of 1.3233 for scrambling/scale compression that we have observed is invariant with time over a nine month period. Scale compression results from scrambling of the isotope distribution in the ion source and is dependent on source pressure and ion residence time within the high pressure region of the source.
2
1
= 250mm = 120° L1 = 288.7mm
L2 = 288.7mm
X 1
Y
2
= 40.89° focal plane
source slit
resolving slit
Z 12mm X
focal plane
A
A
B C
10mm
+40‰
45/44
A
0 -40‰ +40‰
46/44
0 -40‰ +40‰
47/44
0 -40‰
25x104
45 44
Intensity [cps]
20x104
15x104 47
10x104
46 5x104 48 0
49
7910
7920
7940
7930
7950
7960
7970
Accelerating potential [volts]
47 45 48 44 46 49
B
x10-11A x10 A x10-13A -10
Intensity [A]
2.0 48
1.5
1.0
0.5 49 0 7910
7920
7930
7940
7950
7960
Accelerating potential [volts]
7970
7980
7990
5×10−8 4
I (A)
3
I = Io exp(-c/v).t
2
Io = 4.24x10-8A, c = 3.8x10-11m3.s-1
1 0
v = 2x10-7m3 0
500
1000 t (s)
Δ47(URF)
A
1.0
0.5
0 −50
0
δ47 sam-wrg
50
B
Δ48
20
0 −50
0 δ48 sam-wrg
50
100
1.0
47
ΔURF
47 BDH: Δ47 URF = 1.3233*Δ(EG-WG) + 0.87
0.5
47 BOC: Δ47 URF = 1.3233*Δ(EG-WG) + 0.75961
0 −0.6
−0.4
−0.2 47 ΔEG-WG
0
0.2
MIRA: a new isotope ratio mass spectrometer for clumped isotope studies of CO2 Paul F Dennis∗ , Stuart Vinen, Alina Marca-Bell, Peter J Rowe Stable Isotope Laboratory, School of Environmental Sciences, University of East Anglia, Norwich, Norfolk, NR4 7TJ, UK ∗ E-mail: [email protected]
Abstract Accurate measurement of multiply substituted isotopologues using an isotope ratio mass spectrometer (IRMS) imposes stringent requirements on instrument performance in terms of overall sensitivity, abundance sensitivity and linearity. To meet these requirements we have designed, constructed and evaluated the performance of a new IRMS, optimising its performance for clumped isotope ratio measurements of CO2 . We determined an instrument and magnet geometry that combined excellent ion optics and transmissivity with relatively straightforward engineering such that the instrument could be self-built. MIRA (Multi-Isotopologue Ratio Analyser) is a 120o magnetic sector isotope ratio mass spectrometer with a 25cm physical radius and extended geometry for an effective 50cm dispersion. It is equipped with a Nier type high sensitivity electron impact ion source and six faraday cup collectors for measurements of all the cardinal isotopologue masses of the CO2 molecule. The mass spectrometer is constructed using UHV techniques, compatible materials and components. Modular high voltage electronics modules are used for the source power supplies and instrument control and data acquisition electronics are based on National Instruments compactRIO hardware and LabVIEW software. In dual-inlet mode the MIRA mass spectrometer has enhanced sensitivity (approximately 600 CO2 molecules ion−1 ) combined with high abundance sensitivity and a linear response in terms of measured ∆47 over a greater than 100‰ range in measured bulk isotope composition δ 47 with respect to the working reference gas. Precision of measured ∆47 is at the shot-noise limit of 0.008‰ for standard measurement cycles. When compared with commercially available instruments the MIRA mass spectrometer offers enhanced performance for clumped isotope ratio measurements of CO2 . The high sensitivity combined with the measurement linearity allows for increased overall sample throughput whilst minimising the number of calibration samples that need to be run in order to correct for non-linearity in ∆47 measurements.
2
Introduction Multiply substituted isotopologues are molecules, formula units and other moities in which there are two or more isotopic substitutions by rare heavy isotopes. In geochemistry they are more commonly and colloquially referred to as clumped isotopes and over the past ten years there has been a growing interest in measuring the variability of the relative abundance of clumped isotopes in naturally occuring materials [1, 2]. The most widely developed application has been isotope thermometry of carbonate minerals, particularly those that are characteristic of conditions at the surface and in the upper crust of Earth and even Mars [3–7]. However clumped isotope studies in other systems, including tropospheric and stratospheric CO2 [8–11] and stratospheric O3 [12], point to potentially important applications across a wide range of problems in the atmosphere, biosphere and solid earth. To date all published measurements, bar one, have been made using commercially available ThermoFinnigan 253 gas source isotope ratio mass spectrometers that have been modified to include a six faraday cup collector array with amplifier gains optimised for measurement of isotopologues in the CO2 and O2 systems [13]. The exception is a single study describing the performance of a Thermo-Finnegan Delta XP IRMS for clumped isotope measurements [14]. These studies demonstrate that it is possible to achieve high precision measurements on multiply substituted isotopologues that are present in low abundance. For example the excess abundance of the m/z = 47 isotopologue in CO2 (predominantly
18
O13 C16 O)
which occurs with a nominal stochastic abundance of just 44ppm in CO2 that is characteristic of the VPDB standard can be measured to a relative precision of better than 0.01‰ ( [13] and references therein). The excess abundance of an isotopologue is expressed as [1]:
∆i =
!
" Ri − 1 × 1000 Ri∗
where Ri represents the measured abundance ratio of isotopologue i relative to that of the non isotopically substituted isotopologue, and R∗i represents the expected abundance ratio for the isotopologue when all the isotopes are stochastically distributed. The stochastic abundance ratio is calculated from the measured bulk isotopic composition of the sample. Not with standing the fact that high precision measurements of ∆47 in CO2 can be made the measurement procedure requires careful calibration in order to correct for analytical artefacts that are instrument
3 dependent. The two most important of these are: (i) so called non-linearity of the measured ∆47 composition as a function of the bulk isotopic composition of the sample, and (ii) scrambling of the clumped isotope signal. The non-linearity refers to an increasing divergence between the measured and true ∆47 composition with difference in bulk isotopic composition between the sample and instrument working reference gas. 47 This is characterised as a linear trend in plots of ∆47 versus δ(sam−wrg) for gas samples that have been
heated at 1000oC in order to randomise the isotope distribution (the so-called heated gas line) [13]. To correct for this effect measured ∆47 compositions are extrapolated parallel to the heated gas line 47 towards a zero δ(sam−wrg) value. The gradient of the heated gas lines are subject to differences between
individual instruments and to secular variations with both long term drift and sudden step changes occurring. These can sometimes be related to changes in instrument tuning, filament changes, or even vacuum pump maintenance. Whatever the cause it is necessary to monitor carefully the heated gas line by regular measurements of heated gases interspersed with samples to ensure the integrity of results. The origin of the measured dispersion in ∆47 compositions is not fully understood. Careful observations have suggested that there is a pressure dependent negative, peak baseline shift at all the minor isotopologue peaks when a sample gas is introduced to the mass spectrometer [15, 16]. This may be the result of one or more factors including: (i) insufficient screening of secondary electrons and negative ions generated in the flight tube post magnetic separation that perhaps result from the interaction between a broad, non-gaussian tail of the major ion beam with the flight tube and/or faraday cup assembly; (ii) loss of secondary ions from the faraday assembly, and (iii) electronic issues associated with the high gain (1012 ) amplifiers used for registering the minor isotopologue beams. Whilst further work is required to fully understand the phenomenon a number of different correction schemes for the baseline shift have been proposed to correct for the ∆47 dispersion [13, 15, 16]. Scrambling in the ion source results from ion-molecule interactions and recombination of CO2 ion fragments resulting in a degree of randomisation of the isotope distributions and a scale compression of the clumped isotope signal. The degree of scale compression appears to be a function of source pressure, ion residence time [13] and possibly electron current and energy with values ranging from factors of 0.9 to 0.7. Since one is wanting to measure to high precision small excess values on rare clumped isotopes it is advantageous to maximise the signal by using high source pressures and electron currents thus whilst the full details of the electron-molecule/ion interactions in the source are not fully understood it is important
4 that they are subject to detailed investigation. In the absence of a phenomenological or constitutive description of the scrambling effects a protocol for normalising analytical data to a theoretical reference frame scaled according to statistical mechanical calculations of the expected isotopologue excess based on spectroscopic data for CO2 has been proposed and implemented by most laboratories [17]. The method is based on measuring gases that have a varying degree of isotope ordering as a result of equilibration at different temperatures and comparing the measured excess to theoretically calculated values. When results from different laboratories are scaled appropriately the data are in good agreement. Given the growing interest in clumped isotope measurements and the known instrumental effects that affect current instruments we have designed a new isotope ratio mass spectrometer that optimises performance especially in respect of linearity and sensitivity. The MIRA (Multi Isotopologue Ratio Analyser) employs a novel symmetric 120o extended geometry that optimises transmission and beam quality. In this paper we describe the ion-optic design, basic construction details and analytical performance of MIRA. In particular we demonstrate that the MIRA design is linear with no dispersion of the measured ∆47 value with bulk isotopic composition. Currently MIRA uses a standard Nier type electron impact source. We have conducted experiments to help understand the electron-molecule interactions in this source and the processes that lead to scrambling of the clumped isotope signal and will use this information in the design of a new high sensitivity source for clumped isotope studies.
Instrument design Instrument geometry and ion optics The geometry and ion optical properties of the MIRA instrument are illustrated in Figure 1. The on-axis trajectory has a physical radius, ρo , of 250mm and deflection angle, Θ, in the magnetic field of 120o . The ion beam has positive oblique entrance and exit angles, #’ and #”, of 40.89o. The distance from the source slit to the magnet entrance and from the magnet exit to the resolving slit is 288.7mm (1.1547 x ρo ). The oblique magnet entrance and exit angles of the ion beam result in an extended dispersion of 500mm and stigmatic focussing for optimum transmission. The symmetric geometry has unit magnification in both the x-y and x-z planes. In terms of effective dispersion and image aberrations to the second order 90o and 120o extended geometry instruments have a similar performance. However, we chose the 120o geometry because it offers
5 a shorter overall path length between ion source and resolving slits. Reducing the path length results in less ion-molecule collisions and beam broadening ultimately resulting in improved beam profile quality, and reduced ion scattering and secondary electron generation from interactions between the beam and ion optic elements (flight tube, lenses, resolving slits etc). These considerations may be an important contribution to eliminating the effect of non-linearity in clumped isotope measurements [15, 16]. Trajectories for the on- and off-axis beams were determined using a spreadsheet program, TRAJCALC. This program calculates the first- and second-order field expansion coefficients for ion trajectories in the mid-plane of uniform wedge field magnets of different geometries [18, 19]. Combined with trajectories computed for the pre- and post-magnet field free drift regions it is possible to determine the focal plane geometry and second-order aberrations for both on-axis and off axis beams. Figure 2 shows the detailed trajectories for the inner, central and outer rays of the m/z = 44-49 beams plotted with respect to the ion optic, or x-axis. These cover the 6 cardinal masses of the CO2 isotopologues. The ray trajectories were calculated for a maximum beam divergence of ±0.033 radians exiting the source slit. This is characteristic of the Nier type ion source used in MIRA. The focal plane is oriented at an acute angle of approximately 28o with respect to the optic axis. Unit mass beam separations are approximately 10mm for the CO2 isotopologues. At the focal point beam images are a maximum 1mm wide. These widths are calculated assuming a line source of negligible width at the source slit. In practice the ionisation region in the source is projected by a 3-element immersion lens on to the source slit and has a width much less than that of the source slit width of 0.6mm. Thus the width of the source image at the resolving slits is dominated by the second-order aberrations. Contributions from chromatic aberrations due to the energy spread of ions exiting the source have been neglected in these calculations. This is reasonable given the very low ∆E of electron impact ion sources. Using a resolving slit width of 2mm will theoretically produce well resolved peaks with flat tops and a mass resolution of 250 (dispersion/resolving slit width). On-axis trajectories have also been computed for the x-z plane and are plotted in Figure 1b. The schematic diagram shows the strong focussing action of the magnet fringing fields. Parallel trajectories originating from the source slit (5mm long) are brought to a focus within the magnetic field. The acceptance envelope for trajectories oriented obliquely to the optic axis is shown by the dash-dot line in Figure 1b. Maximum deviations from the optic axis are + and - 0.017 radians for beams originating from the bottom and top of the source slit respectively. This is restricted by placing a 5mm aperture located
6 close to the magnet entrance within the flight tube. The envelope for these beams has a maximum width of 12mm at the magnet exit. To minimise ion scattering from the upper and lower surfaces of the flight tube we have designed this with an internal height of 12mm. This is wider than typical flight tubes in commercial IRMS instruments which tend to have widths of close to the source slit height.
Ion source The ion source is based on a standard Nier type [20,21] with transverse beam electron impact ionisation. The detailed geometry was optimised using the SIMION 3D ion trajectory modelling package (INEEL95/0403). The source consists of a small, enclosed ionisation cavity with an ion exit slit (5mm x0.6mm). This slit forms the first active component of a three element immersion lens consisting of the ion exit slit, the half- or focus plates and the source slit. The focus plates are separated by a gap of 2mm, and the source slit is sized 5mm x 0.5mm. Following the source slit is a further α slit located 50mm beyond the source slit and used to limit the overall beam divergence in the x-y plane. The final active component of the ion source is a small repeller plate located towards the rear of the ionisation chamber. The repeller provides fine control of the field gradient immediately behind the ion-exit slit and in conjunction with the focus potential affects overall beam quality as well as the residence time of ions in the source. The transverse ionising electron beam is collimated by a cross-axis magnetic field generated by two small ALNICO permanent magnets. Modelling of ion trajectories using SIMION shows that appropriate selection of focus and ion repeller potentials for a source potential of +8kV produces a strongly demagnified image of the ionisation region projected onto the plane of the source slit and forming a beam with a narrow beam divergence (< ± 0.03 rads) in the x-y plane. In the x-z plane the beam is extracted as a ribbon of parallel trajectories 5mm high and oriented parallel to the x axis. Ions extracted close to the bottom and top of the slits are deflected by fringing fields and follow trajectories oriented at both positive and negative angles with respect to the x-axis. The majority of these are intercepted by the α slit and don’t pass into the flight tube.
Faraday cup detectors The faraday detector array consists of six individually screened, deep faraday cups with integral resolving slits and secondary electron supression, Figure 4. Each cup is 35mm deep x 5mm wide x 15mm high
7 and is machined from a single block of stainless steel with side plates of 0.2mm thick stainless steel spot welded on to form the cup sides. The insides of the cups are coated in graphite. The slit array has a front resolving aperture (2 x 5mm), followed by an electron suppression plate with an aperture of 2.5 x 5.5mm and a final screening aperture (3 x 6mm). The faraday cups and integral slit array are mounted onto individual supports using ceramic rods with 1mm spacers between all the elements. The front faraday support has a deep aperture that is 7mm wide x 20mm high x 8mm depth. This provides an effective screen between adjacent cups that collects scattered secondary ions and electrons originating from interaction between the resolving slit edges and ion beam. Each faraday cup assembly is covered by a stainless steel wrap around shield. The individual faraday assemblies are mounted onto a base plate that has been photo-etched with locating slots that allow for fine adjustment of position in both x- and y-directions. This ensures exact registration of the faraday cups with the six CO2 beams and that the resolving slits are located on the focal plane of the instrument.
Electromagnet The electromagnet is constructed from 10cm thick milled Maximag high purity iron plates (Tennant Metallurgical Group Ltd). Three sections are used for the top, bottom and back of the yoke. The pole faces are fabricated from two further 10cm thick sections with care taken to ensure accurate machining of the magnet entrance and exit angles. The whole assembly is accurately aligned using precision dowels and bolted together. The coils consist of 2x600 turn assemblies made from 2mm diameter polyester resin coated copper wire wound onto armatures that fit around the pole faces. The total coil resistance is approximately 10Ω. The total power dissipation for a 0.35T field is approximately 125W and it is not necessary to cool the magnet or coils.
Dual inlet MIRA uses a standard VG micro dual-inlet system from a SIRA series II instrument. The inlet has a sample cold finger, reference and sample gas variable volume bellows, matched 50cm long capillaries and change-over valve for rapid switching between sample and reference gas. The working volumes for the sample gas cold finger, sample and reference bellows are respectively 2 x 10−4 L, 2-40 x 10−3 L and 5-100
8 x 10−3 L. The cold finger volume in the sample block is carefully matched by a dummy volume on the reference gas valve block. For measurement an aliquot of gas at 160mb pressure is isolated in these volumes and subsequently allowed to deplete during the measurement cycle. A 20 minute data acquisition results in a 20% depletion of gas. Using matched volumes and carefully balanced capillaries ensures identical sample and reference gas depletion rates and no drift in measured isotope ratios during the run. The sample and reference gas variable volumes are used to balance pressures and signals prior to the measurement run. A typical sample measurement is averaged over 5 acquisitions, with each acquisition consisting of 20 reference-sample cycles. Each reference and sample cycle has a 10s deadtime following switching of the changeover valve and a 20s integration period. Total integration periods are 2000s each for the sample and reference gas.
Vacuum system and construction MIRA is constructed using UHV techniques with all metal sealed conflat (CF) type flanges. The source and detector housings are made from single blocks of 316LN stainless steel with ports machined directly into the blocks using standard CF knife flange geometries. The top and bottom sections of the flight tube are milled from 316LN plate and welded together with full penetration welds to eliminate any potential trapped volumes. The flight tube is attached to the source and detector housing using 316LN flanges. All the internal surfaces of the machined blocks and flight tube are electropolished and heat treated under vacuum to reduce outgassing rates. To ensure a clean, oil free vacuum the mass spectrometer is differentially pumped using three Leybold TW70H 70L.s−1 drag stage turbomolecular pumps (Oerlikon Leybold Vacuum GmbH). One pump is dedicated to pumping the source gas load, whilst the analyser is pumped by two pumps located at either end of the flight tube at the source and detector housings. The high vacuum pumps are backed by two Leybold DiVac 0.8T diaphragm pumps running in series. In this mode backing pressures are reduced to ca. 0.1mb and a clean vacuum with ultimate pressure of 1 x 10−9 mb is readily achieved. The source and analyser vacuum are monitored by two cold cathode ion gauges located either side of the differential pumping pressure step in the source housing. The dual-inlet change-over valve is pumped by a dedicated 70L.s−1 turbomolecular and diaphragm pump, whilst the sample and reference gas inlet blocks and valves are pumped via a further 70L.s−1 pump
9 backed by two diaphragm pumps operated in series. Roughing down the inlet from atmospheric pressure is provided for by a valved connection to the intermediate stage between the two diaphragm pumps.
System electronics, instrument control and data acquisition Source high voltage power supplies The source high voltage power supplies are of a modular design and built using Applied Kilovolts HV units for the source and focus plate potentials, and an Applied Kilovolts floating filament power supply unit for the filament (ITT Exelis Ltd). Sitting on top of the source and focus potentials are UltraVolt programmable floating power supply units to provide bias voltages for the filament, ion repeller, trap voltage and ∆ focus for beam steering (UltraVolt Inc). Twelve volt DC power to drive the UltraVolt units and bi-directional analogue channels for programmable voltage control and read-back are provided by UltraVolt TNFL1 15kV isolated units. The trap current is monitored using a current sensing resistor and a high common mode voltage op-amp (AD629). The output of the op-amp is used as input to a PID circuit for control of the filament current (see below).
Electromagnet power supply The electromagnet is powered by an Amcron Macro-Tech 601 600W audio amplifier with a DC input stage. The coils have a resistance of 10Ω at ambient temperatures and a current of approximately 3.5A is required to produce a field strength of 0.35T. The voltage gain of the amplifier is 20x such that an input voltage of 1.75V is needed for a 35V output. Even at a relatively low mass resolution of 250, and with good laboratory temperature control (±1o C) it is still necessary to use current or field feedback in order to stabilise the magnetic field. We use the voltage drop across a low value current sensing resistor (0.1Ω) with a very low temperature coefficent of resistance as the process input into a PID control to maintain field stability to better than 1 in 5000.
10
Head amplifiers and signal integration The ion currents at m/z = 44, 45 and 46 are amplified by low noise transimpedance amplifiers (AD549L) with gains of 108 , 1010 and 1010 respectively and time constants of 0.1 second. Ion currents at m/z = 47, 48 and 49 are amplified by ultra low noise transimpedance amplifiers (Femto Messtechnik GmbH model LCA-2-10T) operated with a gain of 1012 and a time constant of 1 second. All the amplifiers are individually shielded and mounted directly onto the signal feedthroughs of the faraday array flange. The output voltages are buffered and processed via synchronous voltage to frequency converters (AD652). To optimise linearity these are operated at a clock frequency of 1MHz giving a maximum count rates of 5x105 Hz for a 10V signal. High precision ∆47 , ∆48 and ∆49 measurements requires low amplifier input bias currents with minimal drift. Typical bias currents for the LCA-2-10T amplifiers are 10fA with temperature induced drift of ca. 1fA per degree C. To keep the absolute drift to a minimum value of less than 0.5fA the temperature of the amplifier and VFC circuits is regulated to better than 0.5 degree C. An offset of 0.5fA results in an error of 0.03 per mille in measured δ 47 compositions when the sample and reference gas differ by 100 per mille. This is at the level of precision expected for a single δ 47 acquisition of 20 minutes and is deemed acceptable given that at smaller differences in bulk isotope composition between the mass spectrometer working reference gas and sample gas the error is proportionately smaller.
Instrument control and data acquisition System control and data acquisition electronics are based on National Instruments compactRIO technology with an NI cRIO-9004 Real Time Controller interfaced to a cRIO-9104 unit consisting of an FPGA back plane and eight ports for analogue and digital signal conditioning and input-output modules (National Instruments Inc). Analogue input signals (vacuum gauges, source read back parameters, and trap current feedback) are processed by two NI 9205 32 channel, 16-bit analogue to digital converters. Two NI 9263 4 channel, 16-bit digital-analogue converters are used for all the system analogue control voltages (0-10 volts). An NI 9476 32 channel 24V sourcing digital output module is used to control pneumatic interfaces for low and high vacuum isolation valves, and valves in the micro dual-inlet. Fast digital signals, both input and output, are managed by two NI 9401 8 channel high speed digital
11 input/output modules. These modules provide the TTL clock frequency (1MHz) for the VFC’s and act as the interface for the VFC outputs. FPGA (Field Point Gate Array) and Real-Time controller technology allows rapid development and deployment of system control and data acquisition functions that would conventionally require the design and construction of dedicated electronic circuits. For example we have configured the Real-Time controller to carry out the feedback system control functions for the ion source (filament current) and the electromagnet. Using proprietary NI PID modules implemented in the LabVIEW Real-Time programming module the electron beam trap current is regulated to better than 0.1% and the field to 1 part in 5000. The signals from the six faraday cups are counted using a six channel leading edge counter implemented on the FPGA. Timing for the counter and the clock pulses for the external VFC circuits is provided by the 40MHz clock on the FPGA thus enabling a fully synchronous VFC-counter circuit. Counter outputs are transferred via an FPGA enabled FIFO buffer to the Real-Time controller and from there to the desktop PC for display and data storage.
Performance In this section we characterise the performance of the MIRA IRMS for clumped isotope analysis concentrating on (i) beam quality as defined by mass resolution, abundance sensitivity and flatness of the peak top; (ii) overall sensitivity in terms of sample molecules per ion detected, (iii) system linearity with respect to measured ∆47 and ∆48 versus the sample bulk isotope composition as represented by the δ 47 and δ 48 composition. Finally we note that MIRA exhibits a greater degree of scale compression for clumped isotope measurements than other instruments reporting that this effect appears to be constant and readily calibrated. In a separate paper we present the results of preliminary experiments aimed at understanding this effect (unpublished results) and anticipate that the results of these studies will be used to inform the design of a new high sensitivity source.
Beam quality and peak shape Examples of the peak shape and quality are shown in Figure 4. The peaks were recorded at a major beam intensity of 4.2x10−8A and an analyser vacuum of 2x10−8mbar. In panel A the output of all six faraday collectors are recorded as the source HV voltage is scanned over a span of 60V at a nominal 8kV
12 acceleration voltage. The six cardinal masses for the CO2 isotopologues are in excellent registration with the beam centres within four volts of each other, and the peaks having flat base lines, steep sides and broad flat tops. This is characteristic of the fact that the resolving slits are accurately located on the focal plane of the instrument. The mass resolution, defined as M/∆M = V/∆V where M is the mass of interest, ∆M is the peak width in amu, V is the accelerating voltage at which the peak appears and ∆V is the peak width in volts at half the peak height (fwhm definition), is 235. This is in excellent agreement with the theoretical mass resolution of 250 for an instrument with 500mm dispersion and resolving apertures 2mm wide. An estimate of the beam width at the resolving slit of the faraday detectors is given by the distance over which the beam intensity rises from 10% to 90% of its maximum value as it crosses the edge of the aperture. The 10-90 beam width is 0.34mm, somewhat less than the beam width of 1mm calculated assuming a maximum beam divergence of ±0.033 rads shown in Figure 2. This latter width defines the outer envelope of the beam and for flat topped peaks must be less than the resolving slit width. To test the beam quality and flat topped nature of the peaks we have plotted the beam ratios 45/44, 46/44 and 47/44 over a 10V span in Figure 4a. The y-axis of these plots is scaled in per mille and covers a range of -40 to +40 per mille deviation in the measured ratio. All three ratios are flat to within 2‰ over the whole of the 10V range. Whilst the overall beam profile and peak quality of MIRA is excellent a major concern for clumped isotope ratio measurements are the corrections required for dispersion of the measured ∆47 and ∆48 compositions as a function of the bulk isotopic composition of a sample. This is the so-called non-linearity effect observed in existing commercial IRMS instruments. Whilst there is a good phenomenological description of the effect it’s cause is not fully understood though recent work strongly supports the idea that secondary electrons generated by the major beam and which are not fully suppressed are a major contributing factor [15, 16]. These electrons cause negative shifts in the baseline signal for the weak clumped isotope peaks at masses 47 to 49. The location for generation of the secondary electrons has not been identified but is likely to result from a broad, markedly non-gaussian distribution of ions in the tails of the major peak. These ions are not collected by the faraday detectors and may interact with the flight tube and ion-optic elements to generate the secondary electrons. Another effect resulting from broad tails associated with the major beams is to lower the abundance sensitivity of the instrument, defined as the signal intensity in the tail normalised with respect to the peak height at distances of one, two or
13 more amu displacement from the major peak. Whilst for routine isotope ratio measurements abundance sensitivity is not usually an issue one needs to be aware of potential problems when considering clumped isotopes. The isotopologues at masses 47, 48 and 49 are nominally 40, 4ppm and 40ppb respectively of the mass 44 peak intensity. As with scattered secondary electrons a low abundance sensitivity can also lead to non-linearity in clumped isotope measurements. Accurate measurement of these species over a wide compositional range of 100 per mille requires abundance sensitivities of better than 1 in 108 and 1 in 109 respectively for the major beam at 3 and 4 amu distance. It is not easy to measure such high abundance sensitivities. In Figure 4b we have plotted the peaks at masses 48 and 49 together with the lower portion of all the peaks as all six peaks are simultaneously scanned and collected. In doing this test we have not compensated for the input bias current of the amplifiers. All six peaks show the same basic characteristics. Towards the high mass side (lower voltage) the peaks demonstrate a markedly sharp rise. On the low mass side (higher voltage) all the peaks show a more marked decay towards the baseline. Such a shape is characteristic of ion beams in mass spectrometers where ion-molecule collisions tends to modify the ion energy distribution with a tail towards lower energy. This is reflected in the tail on the low energy (low mass) side of the mass spectrum. Not-with-standing this tail the baseline for all the peaks is flat and at the level of the amplifier input bias currents as measured when there is no beam in the instrument (typically 5 to 10 fA). Thus we can exclude any contribution of a secondary electron current to the measured beam signals. Similarly, the flat baselines are strongly suggestive of a very high abundance sensitivity. Overall MIRA has excellent ion optics resulting in high beam quality and peak shapes that are characterised by a sharp rise and decay, with broad, flat tops across which the measured isotope ratio is stable. Abundance sensitivity is high and there is no contribution of secondary electrons to the minor ion beam signals.
Sensitivity We determined the absolute sensitivity of the MIRA mass spectrometer under typical analysis conditions following the method of Prosser [22]. The reference gas side of the mass spectrometer was filled with CO2 to a pressure of 160mb. This is the inlet pressure required for a major beam signal of 4.2x10−8A. The dummy cold finger volume (0.2mL) was then isolated and the signal monitored whilst the gas depletes. The change in the ion signal as a function of time is given by:
14
I = Io .e(−c/v).t where I is the ion signal at time t, I0 is the ion signal at time t = 0, c is the capillary conductance in units of m3 s−1 and v is the volume (m3 ) from which the analyte gas is depleting. The capillary conductance for a 20% depletion from the cold finger volume over a 20 minute run cycle is 3.8x10−11m3 s−1 (Figure 5). Assuming that the ion signal is directly proportional to the gas pressure in the cold finger volume then at t = 0 the gas flow rate into the ion source is given by: dµ −c = .µ0 dt v where µ0 is the initial amount of CO2 in the volume v in molecules, c is the capillary conductance, v is the volume of the inlet dummy cold finger. µ0 is calculated as 8.6x1017 molecules and the molecular flux into the source at t = 0 as 1.63x1014 molecules.s−1 . The ion signal at t = 0 is 4.24x10−8A, or 2.65x1011 ions s−1 . Thus the sensitivity is 615 molecules.ion−1 Considering the measured sensitivity in terms of overall efficiency of ion extraction from the ion source and transmission through the analyser we can describe the measured ion current from a Nier type electron impact ion source as [23]:
I + = β.Qi .Se .I − .p where I+ is the measured ion current (A), Qi is the differential ionisation coefficient (m−1 ).Pa−1 , Se is the effective path length of the ionising electron beam (m), I− is the ionisation (trap) current (A), p is the source pressure (Pa) and β is the combined source and transmission efficiency. For the MIRA source I+ is 4.2x10−8A, I− is 750x10−6A, Se is taken as the source slit length of 5x10−3m. The source pressure is estimated using kinetic theory and the equation for gas transport in the molecular flow regime through a thin aperture [24]:
µ = (P1 − P2 ).A.
!
R.T 2.π.M
"0.5
where µ is the gas flux (N.m.s−1 ), P1 and P2 are the pressures inside and outside the source box
15 respectively (Pa), A is the cross sectional area of the source slit aperture (m2 ), R is the gas constant (J.K−1 .mol−1 ), T absolute temperature (K) and M the mean molecular mass (kg.mol−1 ). Since the source pressure P1 is very much greater than the pressure P2 outside the source box we can make the approximation P1 = (P1 - P2 ). The measured gas flux of 1.63x1014 molecules.s−1 determined above for intitial gas flow through the capillary converts to a gas flow rate of 1.06x10−6 N.m.s−1 at an assumed source temperature of 200o C. Using this value yields a source pressure of 2.25x10−3Pa. We can now estimate the overall source efficiency as:
β=
I+ Qi .Se .I − .p
The differential ionisation coefficient Qi is estimated as 7.6m−1 .P a−1 using the reported cross sections for electron impact ionisation of CO2 by 70eV electrons [25,26]. Together with the derived source pressure and source design parameters we determine an overall efficiency of 67% i.e. 2 out of every 3 ions produced within the active region of the ion source are extracted, transmitted and then detected by the faraday cups. The overall sensitivity and efficiency of MIRA is commensurate with, but certainly towards the high end of the performance range for IRMS instruments fitted with electron impact ion sources. For example the Europa 20-20 IRMS (now Sercon 20-22 IRMS) equipped with a similar ion source but operating at 2.7kV has a sensitivity of 780 molecules.ion−1 [22]. From the data supplied by Sercon we have calculated an efficiency of 49% for the 20-20 and 20-22 instruments. The improvement in sensitivity and efficiency of MIRA over other similar analysers probably reflects a combination of several factors including: (i) enhanced ion extraction efficiency resulting from the higher source potential (8kV compared to 2.7kV) and higher field gradients in the ion extraction region of the source, and; (ii) an improved overall instrument transmission from source to detector.
Linearity As previously discussed a key performance parameter for clumped isotope measurements is the overall linearity of the measured ∆47 composition as a function of the difference in bulk isotope composition (δ 47 and δ 48 ) between the mass spectrometer working reference gas and sample gas [13, 15, 16]. Linearity is
16 evaluated using sample gases with different bulk isotopic compositions and known ∆47 and ∆48 values that have been prepared by either heating CO2 at high temperature (1000oC) in order to produce a close to stochastic distribution of the isotopes, or equilibrating CO2 with water at lower temperatures (e.g. [17]). In Figure 6 we plot the results of ∆47 and ∆48 measurements on heated and water equilibrated gases covering a range of δ 47 and δ 48 compositions. The gases were prepared using BOC cylinder gas (δ 13 C = -34.34‰V P DB ; δ 18 O = 10.29‰V SMOW ) and CO2 produced by reaction of BDH Carrara marble chips, BDHCM, with phosphoric acid at 25o C (δ 13 Cvpdb = 1.98‰V P DB , δ 18 Ovsmow = +39.30‰V SMOW ) as two end members, and a 50:50 mixture of the two gases. In order to: (i) cover a wider compositional range, and; (ii) check that there are no systematic offsets associated with differences between the sample and reference sides of the dual-inlet and changeover valve we used the two different end member gases (BOC and BDHCM) as working reference gas in the mass spectrometer. The BOC CO2 was used as received whilst we equilibrated the BDHCM CO2 with water at 25o C in order to bring it’s ∆47 composition closer to equilibrium at ambient temperatures. The total range in δ 47 and δ 48 compositions measured with respect to the working reference gas are on the order of 120‰ from -60 to +60‰. As the reference gases have different ∆47 and ∆48 compositions to facilitate comparison we have corrected our results to the absolute reference frame [17]. In making this correction we have adjusted for scale compression but do not make any correction for non-linearity. As shown in Figure 6a and b the measured ∆47 and ∆48 compositions for both the 1000o and 25o C gas samples are invariant with respect to the bulk isotope composition of the gas. These results were obtained over a 9 month period using different reference gases and show no secular variations, or sudden step change. During this time we have changed the reference gas several times, the ion source filament and modified several of the electronic circuits. Thus we are confident that the linear behaviour of MIRA is an inherent feature of its design. Also plotted in Figure 6 is the region of non-linear behaviour for 1000oC heated gases reported for measurements made on a Thermo-Finnigan 253 IRMS [13]. These authors report both secular drift and step changes in the dispersion. Similar non-linear behaviour has been observed by other groups including laboratories at Harvard, Yale, Johns Hopkins, Chicago and Zurich [15–17]. The dispersion in ∆47 ranges from 0.0064 to 0.0128‰ ∆47 per ‰ δ 47 and from 0.125 to 0.35‰ ∆48 per ‰ δ 48 [13]. At large differences in bulk isotopic composition of 70‰ the absolute magnitude of the dispersion for ∆47 is on the order of the total range in expected ∆47 composition for naturally occurring samples. For ∆48 measurements the
17 dispersion is several orders of magnitude greater than the expected range of natural samples. Though we observe that the MIRA mass spectrometer is linear with respect to ∆48 the external measurement precision is such that it is not possible to resolve systematic changes in ∆48 composition as a function of equilibration temperature i.e. the 1000o and 25o C equilibration lines in Figure 6b are indistinguishable. ∆48 is very sensitive to sample contamination (e.g. [13]) and we use positive values greater than 1 as an indication that the sample may be subject to contamination and reject the ∆47 measurement. Internal measurement precisions for ∆48 on single samples are on the order of 0.05 to 0.1‰ and we conclude that present preparation and gas purification methods are not yet sufficient to produce sample gases that will yield a meaningful ∆48 measurement.
Scrambling and scale compression During ionisation of the analyte gas there is potential for reordering of the isotopes amongst the ensemble of molecules in the ion source moving compositions closer towards a stochastic distribution. This scrambling process leads to scale compression in clumped isotope measurements [13, 17]. We are publishing a phenomenological description of the scrambling effect in a separate paper. It is likely that the key processes that cause scrambling are ion-molecule and dissociative ionisation-recombination reactions with threshold energies of 10-20eV within the high pressure region of the source. The degree of scrambling is dependent on both source pressure and the ion residence time and is described by the function (Dennis, unpublished results):
ln
∆∗47 = k.t.P ∆47
where ∆∗47 and ∆47 are the true (non-scrambled) and scrambled compositions respectively, k is a constant, t is the residence time and P the source pressure. i.e. the degree of scrambling is dependent on the probability of an ion-molecule interaction. The effect of scrambling is shown in Figure 7 in which the theoretical ∆47 compositions of sample gases equilibrated with water at temperatures of 25o , 50o and 80o C and heated to 1000oC are plotted versus the measured ∆47 compositions determined with respect to the mass spectrometer working reference gas. The theoretical compositions were estimated using the data of Wang et al. (2004) [1] [17]. The empirical lines plotted in Figure 7 correspond to the different reference gas compositions used
18 over the past year: BOC cylinder gas and a reference gas prepared by reacting BDH Cararra Marble chips with phosphoric acid and then equilibrating the gas with water to draw the ∆47 composition closer to that of an ambient temperature CO2 (see discussion on linearity above). There is a linear relationship between ∆47 expressed in the absolute reference frame and the measured ∆47 composition. The two lines are parallel to each other with a gradient of 1.3233. This represents an overall scale compression of 24.5%. The different line intercepts result from the fact that the mass spectrometer working reference gases have different states of isotope ordering depending on their source, or how they were prepared. Note that neither the BOC, or BDHCM gases are in clumped isotope equilibrium at 25o C. Scale compressions reported for other instruments tend towards a lower value, ranging from 1 to 12.5% [13, 14, 17]. The ion source in MIRA and that in the Thermo-Finnigan mass spectrometers are of different designs. Given the high efficiency of the MIRA source (see above) the operating pressure can’t be significantly greater than that of the Thermo-Finnegan design. Thus we conclude that the greater scale compression that we observe in MIRA results from a longer ion residence time in the source. We are currently re-designing the source in order to minimise the residence time and reduce scale compression. It has also been reported that scrambling and scale compression is subject to secular changes [13]. These are possibly related to subtle variations in source tuning affecting ion residence time and source pressure required to sustain the measurement signal. We are unaware of any other reported studies of the long term stability of scale compression during clumped isotope measurements but expect subtle secular variations to be the norm. However, our experience is that provided the source is operated under identical tuning and with a constant source pressure, then the degree of scale compression is stable and not subject to secular drift. Over the past year we have not had to vary source tuning and overall source and instrument sensitivity has remained constant.
Conclusions We have described the design and construction of a new IRMS, MIRA, that has been specifically optimised for clumped isotope ratio measurements. The main aim of the design was to produce an instrument with a completely neutral response such that the measured ∆47 and ∆48 compositions are independent of the bulk isotopic composition of the sample being measured. This would remove the need for careful calibration of any non-linear response that would be needed to correct for the dispersion in ∆47 and ∆48
19 that is evident in commercially available instruments. Thus special care has been paid to ensure that ion-optic aberrations are small particularly with reference to components of the beam that contribute to a significant, broad non-gaussian component. This combined with careful design of the faraday cup detectors ensures that there are no contributions from secondary ions and electrons to the measured beams, particularly for the minor, multiply substituted isotopologues. With a standard design Nier type electron impact ion source the instrument performance is exceptional. Overall sensitivity is very high at 615 CO2 molecules per detected ion. This represents a combined source and transmission efficiency of 69%. Beam and peak quality in terms of overall shape, flatness and baseline stability are all very high. The ion beam has a narrow 10-90 width of 0.34mm, significantly smaller than the resolving aperture width of 2mm, with a negligible non-gaussian component in the beam tails. This contributes to a sharp rise and fall of the peaks with broad flat tops whilst retaining an overall mass resolution of 235 at an operating HV of 8kV. The peak tops are flat to within 2‰ over a greater than 10V window. There is no measurable secondary ion, or secondary electron component to the measured ion beams, particularly for the multiply substituted isotopologues at masses 47 to 49. This is reflected in the flat response of the instrument when comparing measured ∆47 and ∆48 compositions across a greater than 100‰ bulk composition range. We have observed no change in this response over more than nine months of running. We do observe a higher degree of scale compression at 24.5% than is seen with other instruments. However this is stable and not subject to secular variation. Calibration of the local ∆47 scale for different working reference gases to the absolute reference frame show that the compression is independent of working reference gas composition. MIRA allows routine measurement of clumped isotope compositions at the shot noise limit of precision for ∆47 (< ±0.01‰) for samples as small as 5mg with 2000s integration of sample and reference beams and overall measurement times of 100 minutes. The limit to minimum sample size is currently fixed by the minimum bellows and cold finger micro-volume on the dual inlet. Sample consumption during a measurement is 3.2x10−5 bar.L. This suggests that with optimisation of the inlet volumes samples as small as 1.6x10−4 bar.L can be readily run with a 20% depletion. This is equivalent to approximately 750 microgrammes for carbonate mineral samples.
20
Acknowledgments PFD gratefully acknowledges the support of the University of East Anglia who provided the funding for design and development of the MIRA mass spectrometer. SJV acknowledges the support of NERC and Sercon via a CASE studentship. ADM and PFD acknowledges NERC support and funding via grants NE/H012311/1 and NE/E010105/1. Sercon Ltd continue to provide financial support for the ongoing development of MIRA and the authors acknowledge the constructive discussions and long term support that Sam Barker of Sercon has given the project. PFD is very grateful to Dave Harbour of Isotope Precision Ltd who provided guidance, support and all the engineering expertise required to see through the project from its original concept to completion. Chell Instruments constructed the instrument.
21
References 1. Wang Z, Schauble EA, Eiler JM (2004) Equilibrium thermodynamics of multiply substituted isotopologues of molecular gases. Geochimica et Cosmochimica Acta 68: 4779–4797. 2. Eiler J (2007) “Clumped-isotope” geochemistry—The study of naturally-occurring, multiplysubstituted isotopologues. Earth and Planetary Science Letters 262: 309–327. 3. Came R, Eiler J, Veizer J, Azmy K, Brand U, et al. (2007) Coupling of surface temperatures and atmospheric CO2 concentrations during the Palaeozoic era. Nature 449: 198–2007. 4. Affek HP, Bar-Matthews M, Ayalon A, Matthews A, Eiler JM (2008) Glacial/interglacial temperature variations in Soreq cave speleothems as recorded by ‘clumped isotope’ thermometry. Geochimica et Cosmochimica Acta 72: 5351–5360. 5. Da¨eron M, Guo W, Eiler J, Genty D, Blamart D, et al. (2011)
13
C18 O clumping in speleothems:
Observations from natural caves and precipitation experiments. Geochimica et Cosmochimica Acta 75: 3303–3317. 6. Swanson EM, Wernicke BP, Eiler JM, Losh S (2012) Temperatures and fluids on faults based on carbonate clumped-isotope thermometry. American Journal of Science 312: 1–21. 7. Halevy I, Fischer WW, Eiler JM (2011) Carbonates in the Martian meteorite Allan Hills 84001 formed at 18 ± 4 ◦ C in a near-surface aqueous environment. Proceedings of the National Academy of Sciences 108: 16895–16899. 8. Eiler JM, Schauble E (2004) 18 O13 C16 O in Earth’s atmosphere. Geochimica et Cosmochimica Acta 68: 4767–4777. 9. Affek HP, Eiler JM (2006) Abundance of mass 47 CO2 in urban air, car exhaust, and human breath. Geochimica et Cosmochimica Acta 70: 1–12. 10. Affek HP, Xu X, Eiler JM (2007) Seasonal and diurnal variations of
13
C18 O16 O in air: Initial
observations from Pasadena, CA. Geochimica et Cosmochimica Acta 71: 5033–5043.
22 11. Yeung LY, Affek HP, Hoag KJ, Guo W, Wiegel AA, et al. (2009) Large and unexpected enrichment in stratospheric
16
O13 C18 O and its meridional variation. Proceedings of the National Academy of
Sciences 106: 11496–11501. 12. Yeung LY, Young ED, Schauble EA (2012) Measurements of 18 O18 O and 17 O18 O in the atmosphere and the role of isotope-exchange reactions. Journal of Geophysical Research 117: 1–20. 13. Huntington K, Eiler J, Affek H (2009) Methods and limitations of ’clumped’ CO2 isotope (∆47 ) analysis by gas-source isotope ratio mass spectrometry. Journal of Mass Spectrometry 44: 1318– 1329. 14. Yoshida N, Vasilev M, Ghosh P, Abe O, Yamada K, et al. (2013) Precision and long-term stability of clumped-isotope analysis of CO2 using a small-sector isotope ratio mass spectrometer. Rapid Communications in Mass Spectrometry 27: 207–215. 15. He B, Olack GA, Colman AS (2012) Pressure baseline correction and high-precision CO2 clumpedisotope (∆47 ) measurements in bellows and micro-volume modes. Rapid Communications in Mass Spectrometry 26: 2837–2853. 16. Bernasconi SM, Hu B, Wacker U, Fiebig J, Breitenbach SFM, et al. (2013) Background effects on Faraday collectors in gas-source mass spectrometry and implications for clumped isotope measurements. Rapid Communications in Mass Spectrometry 27: 603–612. 17. Dennis KJ, Affek HP, Passey BH, Schrag DP, Eiler JM (2011) Defining an absolute reference frame for ’clumped’ isotope studies of CO2 . Geochimica et Cosmochimica Acta 75: 7117–7131. 18. Brown KL, Belbeoch R, Bounin P (1964) First- and Second-Order Magnetic Optics Matrix Equations for the Midplane of Uniform-Field Wedge Magnets. Review of Scientific Instruments 35: 481–485. 19. Wollnik H (1987) Optics of Charged Particles. Academic Press. 20. Nier AOC (1940) A mass spectrometer for routine isotope abundance measurements. Review of Scientific Instruments 11: 212–218. 21. Nier AOC (1947) A mass spectrometer for isotopic and gas analysis. Review of Scientific Instruments 18: 398–406.
23 22. Prosser SJ (1996) A novel magnetic sector mass spectrometer for isotope ratio determination of light gases. International Journal of Mass Spectrometry and Ion Processes 125: 241–266. 23. Elliot RM (1963) Ion Sources. In: McDowell CA, editor, Mass Spectrometry, McGraw-Hill. 24. Berman A (1992) Vacuum Engineering, Calculations, Formulas and Solved Exercises. Academic Press. 25. Rapp D, Englander-Golden P (1965) Total Cross Sections for Ionization and Attachment in Gases by Electron Impact. I. Positive Ionization. The Journal of Chemical Physics 43: 1464–1479. 26. Itikawa Y (2002) Cross Sections for Electron Collisions With Carbon Dioxide. Journal of Physical and Chemical Reference Data 31: 749–767.
24
Figure Legends Figure 1.
The geometry of the MIRA mass spectrometer. The MIRA instrument uses a
120o , symmetric extended geometry with non-normal magnet entrance and exit trajectories for enhanced dispersion and stigmatic focussing. The physical radius is 250mm with source slit to magnet entrance and magnet exit to resolving slit distances of 288.7mm. The boundaries of the magnetic field are rotated by 40.89o with respect to the normal of the beam trajectories. In the x-z plane the flight tube height of 12mm allows full transmission of all ion trajectories exiting the source thus minimising scattered elements of the beam that may result from interaction between the ion beam and flight tube. Magnification in both the x-y and x-z planes is unity. Figure 2. Focal plane trajectories for the six CO2 cardinal masses. Focal plane solution for the six CO2 beams between m/z =44-49. The inner, outer and central rays of beams are plotted in the x-y plane of the instrument for an initial beam divergence from the source of +/-0.033 radians. The focal points are separated by a distance of approximately 10mm normal to the beam trajectories, with the focal plane oriented at approximately 28o to the x-axis. Figure 3. Faraday cup assembly schematic. The faraday cups (C) are machined from a block of stainless steel with the sides spot welded on. The internal surfaces of the cups are coated in graphite. The lens element consists of a front resolving slit (0V), a central electron suppression plate (-36V), and to ensure minimal field gradients across the aperture of the faraday cup a second screening plate (0V). The cups are supported in individual screened cases using ceramic rods and spacers that are tensioned with a nut assembly. Figure 4. Peak shapes and beam quality. Characteristic peak shapes for all six cardinal masses of the CO2 molecule are plotted in panel A. Note the excellent peak quality with a steep rise and fall on the peak sides and a broad flat top. The peaks centres are coincident to within four volts. Beam ratios over a 10V span, indicated by the outline box superposed on the peaks, are stable to better than 2‰, Figure 4a top. Overall beam quality is excellent with a very high abundance sensitivity and absence of any secondary electron currents introducing baseline offsets, panel B. The peak base lines coincide with the amplifier input bias currents on both the high and low mass side. Figure 5. Source and instrument sensitivity. A plot of the major beam intensity as a function of time recorded during a typical analysis (open circles: reference, filled circles: sample). The sample and
25 reference gases are isolated in identical, matched volumes (0.2mL) with an initial pressure of 160hPa. As the gas depletes the signal strength reduces. The capillaries are crimped to ensure identical conductance. The run consists of 20 cycles of reference-sample pairs with a 10 second dead-time following switching of the changeover valve followed by a 20 second integration period. The total run time is 1200 seconds during which time there is an approximate 20% depletion of the sample and reference gas corresponding to an average gas utilisation rate of approximately 1.2x10−9 mols.s−1 over the 20 minute measurement period. The overall sensitivity is 620 CO2 molecules per ion detected (see text). Figure 6. Linearity. Plotted in panel A are the ∆47 compositions expressed in the absolute reference frame [17] of CO2 gas with differing bulk isotopic compositions that have been equilibrated at either 1000o or 25o C. The data have been corrected for scale compression and have had no linearity correction applied (see text). The compositional range measured with respect to the mass spectrometer working reference gas is greater than 120‰ from approximately -60 to +60‰. The horizontal lines are the theoretical ∆47 compositions for 1000o and 25o C [1]. The MIRA mass spectrometer has a completely neutral response. For comparison the range of linearity reported for the Thermo-Finnigan IRMS is bounded by the two 48 sloping lines plotted in panel A [13]. Panel B is a plot of the raw measured ∆48 versus δsam−wrg for
the same gases. Again the MIRA response is neutral over a wide compositional range however it is not possible to resolve the 25o C (squares) from the 1000oC (circles) data. Non-linearity ranges for the Thermo-Finnigan 253 instrument lie between the two sloping lines [13]. Figure 7. Scrambling and scale compression. Measured compositions of 25o , 50o, 80o and 1000oC equilibrated CO2 gases plotted versus theoretical ∆47 compositions [1]. Data are plotted for two different reference gases: (i) BOC cylinder gas, and; (ii) CO2 produced by reacting BDH Carrara Marble chips with phosphoric acid and partially equilibrating the gas with water at 25o C. Note that both reference gases are not in equilibrium at ambient temperatures. The conversion between scales has a correction factor of 1.3233 for scrambling/scale compression that we have observed is invariant with time over a nine month period. Scale compression results from scrambling of the isotope distribution in the ion source and is dependent on source pressure and ion residence time within the high pressure region of the source.
2
1
= 250mm = 120° L1 = 288.7mm
L2 = 288.7mm
X 1
Y
2
= 40.89° focal plane
source slit
resolving slit
Z 12mm X
focal plane
A
A
B C
10mm
+40‰
45/44
A
0 -40‰ +40‰
46/44
0 -40‰ +40‰
47/44
0 -40‰
25x104
45 44
Intensity [cps]
20x104
15x104 47
10x104
46 5x104 48 0
49
7910
7920
7940
7930
7950
7960
7970
Accelerating potential [volts]
47 45 48 44 46 49
B
x10-11A x10 A x10-13A -10
Intensity [A]
2.0 48
1.5
1.0
0.5 49 0 7910
7920
7930
7940
7950
7960
Accelerating potential [volts]
7970
7980
7990
5×10−8 4
I (A)
3
I = Io exp(-c/v).t
2
Io = 4.24x10-8A, c = 3.8x10-11m3.s-1
1 0
v = 2x10-7m3 0
500
1000 t (s)
Δ47(URF)
A
1.0
0.5
0 −50
0
δ47 sam-wrg
50
B
Δ48
20
0 −50
0 δ48 sam-wrg
50
100
1.0
47
ΔURF
47 BDH: Δ47 URF = 1.3233*Δ(EG-WG) + 0.87
0.5
47 BOC: Δ47 URF = 1.3233*Δ(EG-WG) + 0.75961
0 −0.6
−0.4
−0.2 47 ΔEG-WG
0
0.2
