Technical note: apatite and zircon (U-Th)/He analysis using quadrupole and magnetic sector mass spectrometry

Apatite and zircon (U-Th)/He thermochronological data are obtained through a combination of crystal selection, He 10 content measurement by extraction from crystal and analysis using noble gas mass spectrometry, and measurement of U, Th and Sm contents by dissolution and solution analysis using inductively coupled plasma mass spectrometry (ICP-MS). In this contribution, we detail the complete protocols developed for over more than a decade that allow apatite and zircon (U-Th)/He data to be obtained with precision. More specifically, we show that the He content can be determined with a high precision using a calibration of the He sensibility based on the Durango apatite and its use also appears crucial to check for He, U-Th15 Sm analytical problems. The Durango apatite used as a standard is therefore a suitable mineral to perform precise He calibration and yield (U-Th)/He ages of 31.1±1.4 Ma with an analytical error of less than 5%. The (U-Th)/He ages for the FCT zircon standard yields a dispersion of about 9%, with mean age of 27.0±2.6 Ma comparable to other laboratories. For the long-term quality control of the (U-Th)/He data, attention has been paid to evaluate the drift of He sensibility, blanks through time and those of (U-Th)/He ages and Th/U ratios (with Sm/Th when possible), all associated with the use of Durango apatite and Fish 20 Canyon Tuff zircon as standards.

The geological implications of these data rely on the precision of measurements of He, U, Th and Sm contents of apatite and 25 zircon crystals, by: (i) crystal picking; (ii) non-destructive He degassing and content determination by mass spectrometry; (iii) U, Th and Sm analysis after crystal dissolution and solution analysis by ICP-MS . Different contributions have already presented parts of the analytical protocols, for example, He degassing using laser beam (e.g. Foeken et al., 2006;House et al., 2000), dissolution and analysis of U, Th, Sm, Ca or Zr (e.g. Evans et al., 2005;Guenthner et al., 2016;Reiners and Nicolescu, https://doi.org/10.5194/gchron-2021-1 Preprint.  packaging is strictly related to the acid attack protocol. During U, Th and Sm analysis by means of ICP-MS instruments, the presence of Pt + ions at high concentration (>320 μg/ml) in the sample solution may lead to the formation of complex, platinum 55 argides 194 Pt 40 Ar + , 195 Pt 40 Ar + , 198 Pt 40 Ar + , that cause isobaric interferences with the measured U isotopes at masses 234, 235 and 238 (Evans et al., 2005;Reiners and Nicolescu, 2007). Apatite chemical digestion can be achieved using acid digestion at low concentration (HNO3 5N ultra-pure) and low temperature (65°C), is not able to dissolve the capsule, therefore Pt tubes can be used. However, an acid digestion method using concentrated acids is instead adopted for zircon (HCl and pure HF 27N) at high temperatures (220°C), which leads to a total dissolution of the Pt capsule, therefore, Nb tubes were adopted instead. 60 Although niobium-argon complexes do not create isobaric interferences with the analyzed U masses, the solution, highly concentrated in Nb may, however, cause a partial precipitation of the uranium and thorium (Evans et al., 2005).

Helium analysis protocol
The helium content analyses were performed at GEOPS laboratory, Paris Saclay University (Orsay, France). Each capsule containing a crystal, fragments, or grain(s) was degassed using either a homemade He extraction line coupled with a quadrupole 65 mass spectrometer (Prisma QMG 100 Pfeiffer©), further referred to as the He line, or another homemade line connected to a rehabilitated VG5400 magnetic sector mass spectrometer, further referred to as the VG line. The He and VG lines are fully automated using LabView Std, from the heating phase to the helium analysis. Each portion of the line is divided into sections (extraction, purification, analysis) by pneumatic Swagelok© valves coupled to electro-valves (E.V 3/2 NF Direct Flasque.D2,4 ALU BUNA, TH France) and activated by pressurized air. Ultra-high vacuum conditions (<10 -9 mbar) are guaranteed by using 70 a system of turbomolecular (HighCube -Pfeiffer SAS©) and ionic pumps (StarCell -Varian©). Figure 1 presents the schematic geometry of the two homemade He and VG lines, with the different parts that are controlled using LabView.  Platinum and niobium tubes adopted for sample packaging are suitable because of the low hydrogen they release at high vacuum, their malleability and their U, Th, Sm and REE purity. Being metallic materials, they ensure a homogeneous heat transfer during laser shooting. For the He line, the Pt/Nb tubes are deposited on a copper planchette containing 25 or 49 positions and are placed into a cell that moves in front of the laser beam using a X-Y motor system (SMC100CC -Newport 80 ©) controlled by LabView. An Ytterbium doped infra-red diode laser coupled with an optic (wavelength 1064 nm -1080 nm, 10W ManLight; Laser2000), placed at a focal distance of 4 cm from the sample, allows heating up the capsules with a beam of 70 µm diameter. For the VG line, the Pt/Nb tubes are placed on an inox or copper planchette containing 12 or 49 positions and the heating is ensured using an infrared diode fusion laser (Teledyne, USA) moving in front of the cell. After each sample loading, the line (He and VG line) is heated overnight at low temperature (<50°C) to remove any gas absorbed on the inner 85 walls of the lines. In addition, ultra-high vacuum can be quickly obtained by heating an empty capsule placed in the planchette in order to remove air absorbed on the cupper/inox planchette. Cupper or inox were selected for planchette material due to their good thermal conductivity and their inertia in vacuum conditions. The cell is sealed with a CF63 sapphire window (Caburn) allowing a good transmission of the whole IR laser beam. Each capsule is heated using the heating protocol summarized on Table 1. 90  (2001) The heating scheduled procedure was repeated on each sample until all 4 He was degassed giving a signal back to the background level. The sample temperature achieved using the laser of the He line is recorded by means of a LabView camera 95 and a homemade algorithm that converts the total red, green and blue visible light into a temperature. We used the light emission in visible light associated with black body light emission during heating. For this aim, a Pt capsule has been heated with increasing values of the laser intensity and pictures have been taken at different temperature settings, as presented in Figure 2A. At the same time, the temperature of the heated capsule has been measured using an external filament extinction pyrometer. This type of pyrometer is currently used to calibrate TIMS filament temperatures. For each recorded picture, a 100 simple image treatment has been realized using LabView in order to retrieve the red, blue and green value on the RGB colorimetric coding system that ranges from 0 to 255 (Fig. 2B). As the red signal is already saturated when the capsule is emitting visible light, we chose to sum the RBG signal of the three colors, as shown on Figure 2C. The obtained RBG is correlated with the temperature of the heated capsule. The simple image treatment procedure has been calibrated from 950 to 1150°C for a fixed value of the exposure time of 500 ms. The same heating calibration has been automatically applied to each 105 heated capsule, with an estimated error of ± 20°C on temperature calculation.  The analysis protocols differ as a function of the type of mass spectrometer used and the type of analyzed minerals.

Helium line
The diffused 4 He gas is mixed with a known amount of 3 He in the purification line, used as a spike, in concentration of 115 about 100 to 1000 times higher than the 4 He to be determined. A ~4000 cc (cubic centimeter) cylinder (V1), filled by 3 He gas, is connected to a pipette made by two welded valves with a small, 5 mm diameter, stainless steel cylinder placed inside to reduce the volume of the pipette (Fig. 1). The approximate volume of the pipette is ~0.5 cc (V2) and allows an 3 He amount of 10 -9 to 10 -10 ccSTP (cubic-centimeters-at-standard-temperature-and-pressure) to be introduced into the line. The amount of 3 He decreases in the cylinder by a factor of ~0.9999 (V1/(V1+V2)) for every shot of gas extracted, therefore the number of pipettes 120 taken is automatically recorded to take the decrease of 3 He in the cylinder into account. These statistics allow the data for He age computation to be calculated with the right amount of 3 He spike introduced into the line. According to the spike conditions reported above ( 3 He in concentration of about 100 to 1000 times higher than 4 He), it has been observed that the He line can perform a total of about 6000 analyzes.
The sample gas is purified from most of the H2O, CO2, H2, Ar gases using two liquid nitrogen-cooled traps of activated 125 charcoal, a ST707 and a ST701 SAES getter, according to different purification protocols adapted for various minerals ( Table   1). The use of a hot (>850°C) titanium sponge getter is dedicated to minerals with high CO2 or H2O contents. The access to the entire system of traps, individually connected to the line by means of ultra-high vacuum valves (Fig. 1), allows the analysis of a large variety of minerals containing variable abundances of CO2 or H2O such as calcite or goethite (Allard et al., 2018;Cros et al., 2014). Beside helium isotopes ( 3 He and 4 He), H2O, CO2, H2 and Ar gases are additionally measured on the electron 130 multiplier of the Prisma QMG 100 Pfeiffer quadrupole mass spectrometer to check for the effective purification of the analyzed gas. Such measurements of the gas are repeated 16 times and a linear regression of the data for the 4 He/ 3 He ratio is then calculated and includes a correction of the HD + isobaric contribution on the 3 He signal, even if this contribution is insignificant compared to the 3 He spike signal. In addition, we also observe that the signal at mass 4 slightly increases when the H2 signal is higher, which we interpret as either a double H2 molecule having an isobaric impact on mass 4 or the tail of the H2 peak 135 having an influence on the shape of the mass 4 peak. This effect is not negligible for the low 4 He signals of typical samples, but an adapted H2 purification protocol allows to remove this effect: a 707 SAES getter unit is positioned in the quadrupole mass spectrometer (Fig. 1).
The gas purification protocol ensures to get a close to constant (although slightly decreasing) total pressure in the line and in the quadrupole mass spectrometer. The 4 He concentration is calculated using the 3 He content such as: 140 with ( 4 He/ 3 He)s and ( 4 He/ 3 He)b the ratios measured for the sample and blank respectively. The 3 He content is determined using the following equation: V1 is the volume of the 3 He cylinder (~4000 cc), V2 the pipette volume (~0,5 cc) and N the pipette number (i.e. N is the 145 number of introductions of the pipette volume). 3 Hec is the 3 He content value adapted for each calibration and D is the 'drift', an additional parameter introduced allowing to take into account the evolution of sensitivity of the quadrupole mass spectrometer along with external parameters such as temperature or even power failures. D acts as if the pipette volume V2 could vary to mime the variations of the quadrupole sensitivity. D is determined manually and changes according to the Durango standard results, especially every time the source is tuned again. The product 3 "# 2 × & 34 34536 × + 8 thus decreases 150 regularly and homogeneously.

VG line
The diffused 4 He is purified from the H2O, CO2, H2, Ar gases using two ST707 SAES-getters and a Ti sponge getter ( Fig.   1B). A cryogenic trap from Advanced Research Systems (ARS)©, installed more recently, has the capacity to cool activated charcoal down to 8 K. At this temperature He is efficiently trapped, then further released into a smaller volume at about 50 K. 155 Again, according to the nature of the sample to be analyzed, different purification protocols are adopted (Table 1). Every protocol is fully automatized and the 4 He gas is introduced into the VG5400 magnetic sector mass spectrometer. The filament amperage is fixed at a compromising value ranging from 300 to 400 µamps, lower than the recommended value for He analysis but ensures a longer life of the filament. Isotope 4 He is analyzed using a Pfeiffer electron multiplier (17 dynodes) which is connected to an Ortec discriminator and a LabView counting card. 20 analyzes of the 4 He signal integrated over 1 sec are 160 performed and the mean 4 He signal is recorded. The linear regression over the 20 measurements allows us to get a mean signal and the associated standard deviation. The dead time of the electronic chain is close to the width of the signal delivered by the electron multiplier which is a few ns. The maximal recorded counting rate being about 3´10 5 c/s, the dead time correction is always lower than 1% and it is neglected. The system sensitivity is determined using an internal 4 He standard from a ~4000 cc cylinder calibrated over the Durango apatite analyses. The 4 He concentration is computed using 165 with 4 Hes and 4 Heb are the signal for the sample and blank respectively and s the sensitivity (He/cps; cps: counts per second).
Durango apatite fragments and/or Fish Canyon Tuff zircon crystals are analyzed regularly (1 Durango/Fish Canyon standard analyzed every 7 unknown samples) to check the (U-Th)/He analysis reproducibility.
The degree of cleanness after the series of baths is checked by analysis of 238 U, 232 Th natural isotopes intensities and spike isotopes 230 Th and 235 U. The tests were carried out by filling 200 µl MilliQ water in cleaned vials, refluxed for 2h at 100°C, and transferred to PP tubes with the addition of 800 µl MilliQ water.

Samples digestion protocol 185
After degassing, each Pt/Nb-conditioned sample is transferred from the planchette into a vial for grain dissolution by acid digestion. The sample digestion is carried out by adding a volume of 50 µl of a spike solution in each vial (in HNO3 5N and containing a known amount of 235 U, 230 Th -plus addition of 149 Sm and 42 Ca for apatite). According to the nature of the sample, a specific dissolution protocol may be followed (acids and heating temperature) ( Table 2). Step 2: Evaporation at 100

100°C
Step 5: Add 200 µL HNO3 1N + some drops HF 0.1N; 1h reflux at 100°C Step 6: Transfer to PP vial, add 800 µL HNO3 1N Step 7: dilution 1/10 with HNO3 1N For apatite, the dissolution protocol has been adapted from Farley (2002). The dissolution requires a soft acid digestion (HNO3 5N -bidistilled from HNO3 65-Normapur -VWR) performed in a 4 mL single-use polypropylene tube (VWR) by adding 50 µL of spike (~4 ppb of 235 U, 230 Th, 149 Sm and ~800 ppb 42 Ca) and 50 µl of HNO3 5N. The tube is then placed on a 195 hot plate at 65°C during 3 h for digestion. After digestion, samples are diluted with 1.9 ml of HNO3 1N and stored at 4°C before ICP-MS analysis ( Table 2). Due to the digestion conditions (using diluted acids), the Pt capsule does not dissolve and does not interfere during ICP-MS analysis. Sample digestion is always made with freshly diluted nitric acid by addition of MilliQ water. To minimize possible source contaminations from the environment, storing tubes, evaporation, ICP-MS analysis is programmed within a few days of sample digestion. After analysis, the Pt capsules are promptly collected, cleaned and sent 200 back to the factory company in a recycling/reselling loop contract.
For zircon, the dissolution protocol was slightly adapted from Reiners (2005) and Reiners and Nicolescu (2007). The dissolution is performed in 350 µl PFA parrish style vials (Savillex SAS). Zircons are first spiked with 100 µL of 235 U and 230 Th (~45 to ~55 ppb of 235 U, ~15 to ~20 ppb of 230 Th; Table 2). The vials are then placed into a digestion vessel hermetically sealed with a metallic gasket (IN/PFA OUT/Stainless steel PA4748, Parr Instrument Company) to hold high pressures. The 205 digestion follows several steps summarized in Table 2. Step 1: inside the vials: addition of 200 µL HF 27N (Suprapur® -VWR) and few drops of HNO3 7N (Suprapur® -VWR); inside the digestion vessel: addition of 10 ml HF 27N and 1 mL of HNO3 7N. Once sealed, the vessel is heated up at 220°C and held at high pressure for 96 hours.
Step 2: the acid solution is evaporated to dryness by placing the vials on a hot plate at 100°C.
Step 3: 300 µL HCl 6N (Suprapur® -VWR) are added to each vial, the Parr vessel is filled with 12 ml HCl 6N, sealed and heated back at 220°C, under pressure for 24 hours in an oven. 210 Step 4: the vials are evaporated to dryness on a hot plate at 100°C.
Step 5: a reflux is carried out with a combination of HNO3 5N (200 µL) and HF 0.1 N (few drops) at 100°C for 1 hour.
Step 6: the solutions are transferred to 4 ml polypropylene tubes where 800 µL HNO3 1N are added.
Step 7: a final dilution (1/10) is done with freshly prepared HNO3 1N in a second 4 mL PP single-use tube. The solutions are stored at 4°C before isotopic analysis. To avoid pollution released from the storage tubes or changes in concentration by evaporation of the solutions, the ICP-MS session is always scheduled within a few days after 215 In addition to the apatite and zircon samples, the following solutions are also prepared as summarized in Table 2; spiked sample (Sp), including Durango and FCT standards samples; acid blank (Blk): to check acid purity and potential contaminations of tubes; chemistry acid blank (Blk-ch): to check the enrichment contamination in acid caused by the chemistry protocol; spiked blank (BSP): a weighted volume of spike is added to a volume of acid in order to check variations in 220 concentration of the spike and to take into account the contribution of natural isotopes contained in the spike; spiked blank chemistry (BSP-ch): a volume of spike in acid undergoes the same dissolution protocol than the samples, which allows to quantify contamination coming from contingencies during the dissolution protocol (vessel, user, acid); Durango solution (DUR): a single fragment of Durango is dissolved in a volume of acid, no spike is added, which allows the natural isotopic ratio the uranium to be measured for checking, as it has to be in isotopic equilibrium with 235 U/ 238 U=0.00725. 225

Spike solution composition and calibration
As reported above, for every sample to be analyzed and according to the isotopic dilution method, a volume of 50 µl of the spike solution MR2 is introduced into the vial before the dissolution protocol. The spike solution is prepared (60 ml) every 6 to 12 months from elemental mother solutions MR1 and MR (Table 3)

U, Th (Sm and Ca) analysis by ICP-MS 250
The solutions obtained by chemical dissolution of the samples are then analyzed with an ICP-MS in order to determine the U, Th and Sm (Ca for apatite) signal intensities. We now mainly use a High Resolution Inductively Coupled Plasma Mass Spectrometer -HR-ICP-MS -ELEMENT XR from Thermo Scientific) at GEOPS laboratory since 2016, that allows the U, Th and Sm isotopes to be measured at low resolution (300) while Ca is measured at high resolution (10 000). In addition, we also use a quadrupole ICP-QMS seriesII CCT Thermo-Electron at LSCE (Gif/Yvette; France) and an Agilent 7900 quadrupole 255 ICP-MS at IPGP (France) in order to measure the U, Th and Sm contents. (4) 260 . We obtain abundances in nanogram (10 -9 g) and spike 265 concentrations in ppb. In addition, the same equations can be reversed to determine the concentration of the spike isotopes 235 U, 230 Th, 149 Sm and 42 Ca.
We obtain the weight of the apatite grain(s) from the measurement of 43 Ca, for which we use the composition of a pure fluoro-apatite (Ca5(PO4)3F) containing 40 wt. % Ca in one apatite crystal. Thus, the apatite weight is given by equation (8) The factor 0.135 refers to the natural isotope abundance of the 43 Ca isotope.
The measurement of the Ca content and the deduced apatite crystal weight combined to the measurements of U, Th, Sm abundances then allow the U, Th and Sm concentrations of the Durango fragments to be determined. Determination of crystal weight is also useful to ensure that the criteria imposed for grain selection, i.e. (i) crystal size (L,W,T > 60 µm), (ii) geometry (well-shaped prisms), and (iii) purity (inclusion-free crystals) have been respected. Guenthner et al. (2016) have already 275 performed a complete work on determining apatite and zircon weights by measuring the Ca and Zr contents.

(U-Th)/He age reduction
The (U-Th)/He age (in Ma) is calculated assuming a linear production of 4 He with time, using the determined U, Th and Sm concentrations such as: where P* is the instantaneous production of 4 He, i.e. the 4 He concentration produced in one year in ccSTP/g, and 4 He is the measured 4 He concentration in ccSTP/g. P* is calculated using the following equation: * %  For an ICP-MS session, the (U-Th)/He data reduction Excel WorkBook is able to calculate the U-Th-Sm contents and their associated standard deviations for the different minerals analyzed, according to the chemical dissolution protocol followed. The Excel WorkBook also calculates the (U-Th)/He data, which include U, Th, Sm contents and the effective uranium content (eU) both in ppm, the Th/U and Sm/Th ratios and the (U-Th)/He age.

Helium quadrupole analysis
One aspect of quadrupole mass spectrometry is the variable response in terms of ionization and signal analysis of such instruments. This behavior has also been observed on the quadrupole adopted here for analyses of rare gases and is illustrated on Figure 3A. The signal of the 3 He spike is reported as a function of the number of the pipettes extracted, over a period of 5 months of analysis. The 3 He signal fluctuates significantly although the 3 He amount in each pipette should decrease smoothly 300 following a law that depends on the volume of the cylinder and the pipette volume (here ~4000 and ~0.5 cc, respectively; see equation (1)). The advantage of using a 3 He spike for isotopic dilution is to thwart the impact of the nonlinear answer of the quadrupole mass spectrometer with time (Farley, 2002;House et al., 2000), but it also allows the total gas pressure to be buffered in the mass spectrometer if the introduced 3 He signal is large enough compared to the other signals. Figure 3B presents the various signals measured with the quadrupole mass spectrometer (H2, 3 He, 4 He, 40 Ar, CO2, and 305 masse 5 that represents the background noise) during the same 5 months of analysis (~10 to 30 analyses per days, 5 days a week). The 3 He clearly controls the total pressure inside the mass spectrometer during the analysis, independently from the pressure of 4 He gas released from the sample. The use of spike rich in 3 He allows to get a stable and uniform total pressure in the mass spectrometer, for any degassed sample analyzed. However, the quadrupole signals are variable and sometimes erratic within few percent of the signal over weeks to months 315 periods of time (Fig. 3A). To correct for the quadrupole drift, we introduce a correcting factor, noted D in the calculation of the 3 Hes pipette content (equation 1). This parameter acts by modifying the value of the pipette volume, V2 in equation (1), to counterbalance the variations of the quadrupole sensitivity. Figures 4C and D presents the Durango apatite (U-Th)/He ages obtained from measurements over a period of two months using two coupled values of the drift parameter D and 3 Hec calibrated content. One can observe that the (U-Th)/He ages obtained for the Durango apatite remain constant and are more reproducible 320 using a D value of 99.937x10 -2 as compared to 99.987x10 -2 . In the present case, it just depletes the tank in 3 He more quickly.  In addition to the 3 He pipette number associated to each measurement, we use a specific code name. As an example, for the Durango apatite, the code name D19P11A can be read as Durango, year 2019, planchette n°11, aliquot A. This designation using the name of sample and its analysis date allows to better organize the He analyses database and data backup from a 335 chronological point of view.

Helium magnetic sector analysis
In comparison to quadrupole mass spectrometers, magnetic sector mass spectrometers have a more stable and linear response of ionization and thus allow for a better analysis. To test the response of the modified VG5400 mass spectrometer and better calibrate the 4 He cylinder, we performed multiple analyzes of fragments of Durango apatites having different sizes 340 (dozens to hundreds of micrometers long fragments). After degassing, the U, Th and Sm contents of the fragments are determined and, assuming an age of 31.02±1.01 (McDowell et al., 2005), the amount of 4 He for each fragment is calculated.
This operation was performed twice at different times, using the VG5400 tuned with different source parameters and using different He line conditions, one with the use of a cryogenic trap allowing to concentrate the gas in a smaller volume, the other without. Using the correlation between the calculated 4 He in ccSTP and the 4 He signal in count per second (cps), the sensitivity 345 was determined (Fig. 5).  For the two different conditions, the measured 4 He signal and the 4 He content calculated from measured U-Th-Sm display a very good linear correlation (r 2 >0.99, Fig. 5). Sensitivity values of 1.1´10 -12 ccSTP He/cps and 2.9´10 -14 ccSTP He/cps were thus determined, additionally showing that the use of the cryogenic trap increases the sensitivity by a factor ~40. Such analyzes 355 of Durango apatite fragments are therefore useful to test the electron multiplier and counting system responses, which turned out to be linear from thousands to hundreds of thousands of cps for He, without any impact of the dead counting time. The use of both a 4 He cylinder and the Durango apatite fragments is thus important to follow the evolution of the filament and analyzer conditions through time.

U, Th, Sm chemistry and blanks 360
Acid blanks are regularly analyzed and allow the acid quality to be controlled. Low intensities are measured for 235 U and 230 Th, i.e. <20 cps, meanwhile the intensities for 238 U and 232 Th are hundred times higher. Higher signal intensities for all the isotopes are measured in Blk-ch, indicating that the chemical dissolution method adds some contamination to the sample solutions. Such contribution is nevertheless very negligible compared to the intensities of the signals observed for the apatite and zircon samples (100 000 cps). 365 For apatite, U, Th and Sm in blanks are low in comparison with the U, Th, Sm contents of the apatite, as already stated by Reiners and Nicolescu (2007). Figure 6 presents the evolution of the measured 235 U/ 238 U, 230 Th/ 232 Th and 149 Sm/ 147 Sm isotopic ratios for spiked blank (BSP), spiked blank chemistry (BSP-ch) and for Durango apatite sample (Sp) solutions. For a spiked sample (Sp), the 235 U/ 238 U, 230 Th/ 232 Th and 149 Sm/ 147 Sm ratios range between the BSP-ch value and the natural value (i.e. 235 U/ 238 U=0.00725; 230 Th/ 232 Th=0 (no natural 230 Th atoms) and 149 Sm/ 147 Sm=0.9), as can be observed in Figure 6. The isotopic 370 ratio values for the BSP and BSP-ch blanks do not vary by more than few percent through the different analyzes and are orders of magnitude higher than for the sample (Sp). The BSP-ch display lower 235 U/ 238 U, 230 Th/ 232 Th and 149 Sm/ 147 Sm values compared to the BSP (Fig. 6), showing that the chemistry protocol has an impact on the U, Th and Sm isotopes in solution.
However, this effect remains insignificant compared to the Durango apatite U, Th, Sm contents, and does not influence the (U-Th)/He age. Nevertheless, blanks are always well characterized, as for natural apatite crystals, the U, Th and Sm contents 375 are usually lower than for Durango. more specifically of Th when HF+HNO3 acids are used has already been well described (e.g. Révillon and Hureau-Mazaudier, 2009;Yokoyama et al., 1999), and the effect can be seen on Figure 7, where the 235 U/ 238 U and 230 Th/ 232 Th ratios are lower than for the BSP. In addition, the use of Nb capsules (BSP-ch-Nb) also impacts the U and Th budgets and leads to a massive 390 reduction of the ratios (maxima 235 U/ 238 U=16.2±1.0; 230 Th/ 232 Th=4.1±0.9 and recorded minima down to 235 U/ 238U= 12.0; 230 Th/ 232 Th=1.7; Fig. 7). The impact of the niobium capsule on U and Th signals has already been noted by Reiners and Nicolescu (2007), and reported to be more significant for the Th content. The differences in 235 U/ 238 U and 230 Th/ 232 Th ratios between the different blanks (BSP, BSP-ch, BSP-Ch-Nb) are associated with some isotope fractionation, with a calculated decrease of respectively 4 to 6% for the 235 U/ 238 U ratio and 10 to 20% for the 230 Th/ 232 Th ratio, compared to the expected ratios given by the spiked 400 solutions without Nb. The shift for these ratios is systematic but variable from one solution to another, particularly for the 230

Durango apatite and FCT zircon (U-Th)/He age reproducibility 405
The Durango apatite is constantly analyzed in the laboratory to check for the He mass spectrometer sensitivity evolution though time as well as the evolution of (U-Th)/He age and U, Th and Sm contents. As the dissolution protocol for apatite has a very low impact on the U, Th and Sm content determination by ICP-MS analysis, the measurement of Durango apatite acts as a sensor and regularly allows to detect any analytical problem. As an example, Figure 8   Histogram representation of the He ages, Th/U and Sm/Th ratios has been constructed using Radial Plotter (Vermeesch, 2009).
A typical mean error of <5% (1s) is obtained on each AHe age by using either quadrupole or magnetic sector mass spectrometers, without any evident difference over a large period of time. This error can be interpreted as the quadratic sum of the errors on the coupled analyzes of the U-Th-Sm and the He contents and are associated with the calibration. Our (U-420 Th/He) ages on unknown apatite compare well within error with other laboratories data (e.g Ketcham et al., 2018). In addition, our strategy developed to determine the Ca concentration allows us to obtain the weight of the Durango fragment(s) and thus to calculate the U, Th, Sm concentrations in ppm (Table S1). Mean values of U=19±4 ppm, Th=412±68 ppm and Sm=38±7 ppm have been obtained and the U content is similar to that obtained by Schneider et al. (2015) and Yanga et al. (2014).
Fish Canyon Tuff zircon crystals have been analyzed at the GEOPS laboratory as standards for the ZHe method. The U 425 and Th losses during dissolution, due to the niobium impact, are corrected for the determination of the U and Th concentrations in zircons. (U-Th)/He ages were obtained on 57 crystals of FCT zircons analyzed using the He and VG lines and are reported in Figure 9 as a function of the Th/U ratio, and in Table S2. Plotter (Vermeesch, 2009). An U/Pb age of 28.5±0.06 Ma has been published by Schmitz and Bowring (2001). We obtained a mean age of 27.0±2.6 Ma (1s) and a mean Th/U ratio of 0.4 on zircon crystal from C.W. Naeser collection (K/Ar age of 27.9±0.7 Ma; Naeser et al., 1991) (Fig. 9). The standard dispersion of the ZHe ages is ~9% and is 435 comparable to the natural dispersion observed in the ZHe values given in the literature (e.g. Ault et al., 2018;Guenthner et al., 2014;Reiners, 2005). The Th/U dispersion of 37% also corresponds to the natural dispersion observed in the Th/U ratio of the Fish Canyon zircon standard (e.g. Reiners et al., 2002). The (U-Th)/He age results are comparable with (U-Th)/He literature data that range from 27.3±1.0 to 29.8±2.7 Ma (Dobson et al., 2008;Gleadow et al., 2015;Reiners et al., 2002;Tagami et al., 2003;Tibari et al., 2016). Th/U ratios vary between 0.42±0.15 Ma (Tagami et al., 2003) and 0.63±0.14 Ma (Tibari et al., 2016) 440 in the literature. The mean (U-Th)/He age obtained in this study is slightly younger by 5% than the U/Pb age of 28.5±0.06 Ma obtained by Schmitz and Bowring (2001), but still in good agreement within error bars (Fig. 9). The slight ZHe age difference could also be explained by the measured variable ages of the Fish Canyon Tuff zircon as a function of the sampling site (Gleadow et al., 2015). An second option is that since similar ages are obtained by degassing either on the He or VG line, that the slight shift in the (U-Th)/He age may be associated with the He content determination for few percent, but also associated 445 to the U and Th content determination and finally to the impact of the niobium precipitation during zircon dissolution. The loss of U and more specifically of Th associated to the use of HF+HNO3 (Révillon and Hureau-Mazaudier, 2009;Yokoyama et al., 1999), and from the use of Nb capsule (Reiners and Nicolescu, 2007) have already been taken into account on the blank correction. However, the slight lower ZHe ages obtain in this study could also be associated with an under correction of the impact of the zirconium brought into solution that could cause a slight loss in Th that is not taken into account in the blank 450 correction. To estimate such an impact, additional work should be carried out to fully understand the U and Th isotope fractionation during this chemical protocol.

Conclusion
This contribution presents the (U-Th)/He analysis protocols developed for over more than ten years and shares all the empirical and analytical aspects observed during the different steps of the protocol: sample preparation, mineral hand picking, 455 -a simple method to determine the temperature of the heated metallic (Pt and Nb) capsules that contain the apatite or zircon crystals during laser firing in the range 900-1200°C (visible light emission wavelength); -a method to calibrate He sensitivity using quadrupole and magnetic sector mass spectrometers; -the protocols to dissolve apatite and zircon crystals and to clean laboratory vessels after chemical digestion; 460 -the protocol to calibrate the U, Th and Sm spikes; -the method used to track the U, Th and Sm blank evolution and determine U, Th and Sm contents; We adopted the Durango apatite as a standard to perform the He calibration and check for He, U-Th-Sm analytical problems and are able to determine (U-Th)/He ages with an error of less than 5%. Our choice is also related to the fact that Durango is an easy-to-use mineral due to its high purity, its rapid dissolution protocol and the strong reproducibility during 465 analyses. For the long-term quality control of the (U-Th)/He data, attention needs to be paid to evaluate precisely the drift of in the data acquisition. LT developed the Monte Carlo simulation QTLFT software and AD the Excel WorkBook automatization software. CG prepared the manuscript with contributions from all co-authors.

Competing interests:
The authors declare that they have no conflict of interest.