A number of additional instruments are directly related to TIMS analysis. These include but are not limited to making single and double filaments by spot welding ultra-pure metal wire (e.g. Rhenium) onto filament holders followed by filament outgassing under high vacuum. Samples are loaded onto filaments in an ISO 14644-1 Class 4 workstation. Compressed air is required to operate the beam-valve that separates the ion source from the analyzer. Similarily a water chiller supplies 20°C water to “cool” the magnet and turbo pump. Oil free scroll pumps generate a fore vacuum of ~2x 10^-2 hPa to operate the turbo pumps. A drying cabinet is used for drying liquid nitrogen traps after use and filaments and shield plates after cleaning them in an ultrasonic bath.

Prior to TIMS analysis the element of interest has to be separated and purified from the sample matrix by chromatographic methods in a clean room chemistry lab. Samples are then loaded onto rhenium filaments that are mounted on a sample wheel which is then inserted into the source housing of the TIMS. The latter is evacuated to 1x10^-7 to 10^-8 hPa over 6 to 12 hours before a filament can be heated up by applying electrical current. Depending on the ionization potential of elements, variable currents respectively temperatures are needed for ionization. Ions of e.g. Pb+, Sr+ and Nd+ typically require temperatures of 1180-1240°C, 1350-1450°C and 1600-1800°C, respectively. They are accelerated by a 10kV ionization potential in the ion source. The ion source not only accelerates ions to a common "speed" but also focuses the ion cloud by a series of lenses into an ion beam that is introduced into the high vacuum analyzer (~3x 10^-9 hPa) of the MS. Here, the ion beam enters a stable magnetic field that separates ions by mass as a function of mass and charge. In other words, after the ions have passed the magnet the ion beam is split up into a fan of mass specific ion beams. For example, an ion beam of Sr is separated into masses 84, 86, 87 and 88; the four naturally occurring Sr isotopes. The mass specific ion beams are then "collected" by faraday cups. They house graphite shells where the ions are electrically neutralized, causing a very small flow of electrons in the range of 0.001 up to 50x 10^-11 A. The latter is then wired to high resistance amplifiers (usually 10^11 Ω) and the resulting voltage (V = Ω x A) processed by the software. The ion beam intensities are thus a reflection of isotope abundance of an element. Since the ion beam intensities of the various isotopes are measured simultaneously in the faraday multi-collector, a measured isotope ratio of -for example- mass 87/86 is "simply" the division of voltages measured for mass 87 and 86. Although this may sound trivial the ion currents are measured at very high precision so that differences in isotope abundance can be measured to ≤0.0008% 2SE precision for ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd and to ≤0.005% 2SE precision for Pb isotope ratios.

Ionization of an element causes fractionation of its isotopes leading to depletion of lighter isotopes relative to heavier isotopes. In case of a TIMS ion source the same sample load is usually ionized over a period of 1-2 hours causing ever increasing mass-bias with time. In isotope systems with at least two stable isotopes the mass bias (%/amu) is corrected by normalization of the measured isotope ratios to an internationally agreed stable isotope ratio such as ⁸⁶Sr/⁸⁸Sr = 0.1194 or ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. In contrast ²⁰⁴Pb is the only stable Pb isotope and thus within run mass bias correction using naturally occurring Pb isotopes is not possible. For many years Pb mass bias correction in our lab successfully applied an external correction based on repeated measurements of a Pb standard (e.g. NBS981) and correction to reference values (e.g., Todt et al 1996). The mass bias correction values obtained for the Pb standard were then applied to the measured Pb isotope ratios of samples. This type of mass bias correction, however, limits 2SD of Pb isotope ratio analysis to about 0.03%/amu or 300ppm/amu based on ~800, unfiltered NBS981 measurements from 1997 to 2007 on the MAT262 using 100 ng loads of Pb and an emitter with high silica content. To overcome these analytical limitations and to resolve small-scale Pb isotopic variations in sample sets, we established a Pb double spike (DS) technique in 2007 (Hoernle et al. 2011, EPSL). Pb-DS basically involves a second, independent measurement of the sample Pb admixed with a ²⁰⁴Pb-²⁰⁷Pb spike. The latter contains artificially enriched ²⁰⁴Pb and ²⁰⁷Pb and its composition is calibrated against a fixed ²⁰⁸Pb/²⁰⁶Pb ratio of 1.00016 for NBS982. The mass bias correction is then obtained by deconvolution of the two Pb measurements (IC + DS) and the spike composition. Pb-DS allows for stringent mass bias correction of individual samples and thus overcomes matrix effects that influence the fractionation behavior of individual samples. The use of Pb-DS has improved the external reproducibility (2SD) of NBS981 to ≤ 0.005%/amu (50 ppm/amu) on the MAT262 and TRITON Plus which represents at least a six-fold enhancement of analytical precision compared to external mass bias correction. Pb specialists will have noticed that 2SD for the NBS981 IC measurements increased to 500 ppm/amu compared to the 1997-2007 measurements. This reflects ~10 times smaller Pb loads (8-16 ng vs 100 ng) in conjunction with a high efficiency silica gel emitter (Gerstenberger & Haase 1997). Still the joint DS measurement is capable to intercept the larger mass bias variations of the IC determinations and provide correction to fairly uniform Pb ratios.

The analytical quality of TIMS mesurements is monitored by repeated measurements of single element reference materials NBS987 (Sr), La Jolla and SPEX CertiPrep© (Nd) and NBS981 (Pb) along with the samples on each wheel. Ions accumulating over time can lead to alteration of the faraday cups and cause long-term drifts of Sr and Nd isotope ratios. Therefore our lab routinely normalizes the averages of standard measurements (n=4 to 5) obtained for each wheel to the accepted / preferred standard values and applies the offset value to the measured sample data. For instance, all measured NBS987 on the TRITON Plus averages ⁸⁷Sr/⁸⁶Sr = 0.710268 ± 0.000027 (2SD, n=860, since 2014) while the "wheel normalized" NBS987 gives ⁸⁷Sr/⁸⁶Sr = 0.710250 ± 0.000009 for the same set of measurements. Similarly, all measured La Jolla Nd on the TRITON Plus averages ¹⁴³Nd/¹⁴⁴Nd = 0.511847 ± 0.000011 (2SD, n=638, since 2014) and the wheel normalized La Jolla data set gives ¹⁴³Nd/¹⁴⁴Nd = 0.511850 ± 0.000006. The normalization per wheel ensures unhampered comparability of sample data generated over longer periods and ensures best possible analytical output. Despite multiple replacements of H1 and H3 faraday cup shells of the TRITON Plus, ⁸⁷Sr/⁸⁶Sr of NBS987 repeatedly underwent systematic long-term drifts in static collection mode. This is presumably related to alteration of H3 reflecting the 82.58% natural abundance of ⁸⁸Sr measured in this cup. Similar observations but less distinct were made for ¹⁴³Nd/¹⁴⁴Nd. To overcome the long-term drift problem, in 2020 Sr & Nd measurements were switched from static -where each isotope is measured in a dedicated faraday cup- to multidynamic collection where critical isotopes (^{86, 87, 88}Sr and ^{143, 144, 146}Nd) are measured twice in different cups by rotation of the magnetic field. This measure leads to equalization of differences in cup efficiency but requires stable ion beams and increases analytical time per position by 50%. Long-term Pb-DS corrected NBS981 values on the TRITON Plus are ²⁰⁸Pb/²⁰⁴Pb = 36.7206 ± 0.0050, ²⁰⁷Pb/²⁰⁴Pb = 15.4975 ± 0.0019, ²⁰⁶Pb/²⁰⁴Pb = 16.9408 ± 0.0019, ²⁰⁸Pb/²⁰⁶Pb = 2.167585 ± 0.000097 and ²⁰⁷Pb/²⁰⁶Pb = 0.914801 ± 0.000048 (2SD, n=236 since 2014). These values compare very well with published double and triple spike Pb data of NBS981 and our lab is listed in Taylor et al. 2014.