A New Mapping Protocol for Laser Ablation (with Fast-Funnel) Coupled to a Time-of-Flight Mass Spectrometer (LA-FF-ICP-ToF-MS) for the Rapid, Simultaneous Quantification of Multiple Minerals

Although in situ analysis by LA-ICP-MS is considered a rapid technique with minimal sample preparation and data reduction, mapping areas of millimetres in size using a small beam ( < 15 μ m) can be time consuming (several hours) when a quadrupole ICP-MS is used. In addition, fully quantitative imaging using internal standardisation by LA-ICP-MS is challenging in samples with more than one mineral phase present due to varying ablation rates. A new protocol for the quantification of multiple coexisting phases, mapped at a rate of about 12 mm 2 h -1 and a resolution of 12 μ m × 12 μ m per pixel, is presented. The protocol allows mapping of most atomic masses, ranging from 23 Na to 238 U, using a time-of-flight mass spectrometer (ICP-ToF-MS, TOFWERK) connected to a 193 nm excimer laser. A fast-funnel device was successfully used to increase the aerosol transport speed, reducing the time usually required for mapping by a factor of about ten compared with a quadrupole ICP-MS. The lower limits of detection for mid and heavy masses are in the range 0.1 – 10 μ g g -1 , allowing determination of trace to ultra-trace elements. The presented protocol is intended to be a routine analytical tool that can provide greater access to the spatial distribution of major and trace elements in geological materials.

particular their spatial distribution in minerals, is widely used to trace the source and evolution of ore-forming systems in order to refine petrogenetic models and to develop indicator minerals as tools for exploration of ore deposits (Smythe et al. 2008, Dupuis and Beaudoin 2011, Dare et al. 2014, Mao et al. 2015, Duparc et al. 2016, Aulbach et al. 2021. In the field of geology, electron probe microanalysis (EPMA) and micro-XRF are non-destructive methods extensively used for high resolution (1-3 μm beam) mapping of geological samples but both have higher limits of detection compared with LA-ICP-MS (Nadoll and Koenig 2011) and are thus limited to major and minor elemental mapping. LA-ICP-MS is the method of choice to reach sub μg g -1 mass fractions of a wide number of elements, and spot or line scan analyses are commonly used for this purpose. Spot or line scan analyses requires only seconds to minutes to acquire data, thus many grains can be sampled in a single specimen. However, this produces individual, spatially isolated data or one-dimensional transects, covering a limited portion of the studied minerals. To better reveal spatial distribution, such as zoning in minerals, exsolution and the distribution of elements among different minerals, 2D chemical mapping by LA-ICP-MS is typically carried out by analysing a series of subsequent line scans (Zhou et al. 2017, Duran et al. 2020 or a grid of spots (Kelly et al. 2017. However, with current instrumentation (discussed below), mapping using LA-ICP-MS is time-consuming and is often restricted to small areas (< 1 mm 2 ) of the studied polished section.
The duty cycle of Q-ICP-MS limits the data acquisition rate to about 1 to 6 cycles per second (1-6 pixels s -1 ) under common analytical conditions for mapping, thus limiting the pixel production to this rate. Using a 10-μm beam, an area of 1 mm 2 area (10,000 pixels) can take from 30 min up to 4 h (Lear et al. 2012, Lawley et al. 2020. Moreover, some ablation cells have long wash-out times, which may result in pixels representing mixed ablation sites. To avoid this phenomenon, a reduced production rate is necessary (Douglas et al. 2015). Despite the low pixel production rate, chemical mapping is popular and of growing demand in geoscience as it can reveal incomparable information (Gundlach-Graham and Günther 2016, Sylvester andJackson 2016).
Achieving fast quantitative chemical images at the microscale for the routine determination of both major and trace elements for a wide range of mineral matrices (silicates, sulfides, oxides, phosphates) requires new LA-ICP-MS instrumentation and protocols that fulfil essential requirements. These include: (1) a beam size smaller than the target resolution, (2) a high repetition pulsing rate of the laser (3) a rapid transfer of the aerosol to the ICP-MS to minimise pulseto-pulse mixing by using a two-volume or a low dispersion ablation cell (Müller et al. 2009, Fricker et al. 2011 or high efficiency aerosol dispersion cells (Pisonero et al. 2006, Wang et al. 2013, Gundlach-Graham and Günther 2016, Van Malderen et al. 2016, and (4) a rapid analysis of the generated aerosol. The nature of ICP-ToF-MS allows for fast acquisition (33,000 Hz) (Hendriks et al. 2017) allowing for the possibility of producing larger maps compared with those generated with a quadrupole ICP-MS (Bussweiler et al. 2017, Chew et al. 2021, Van Acker et al. 2021. The (quasi-) simultaneous extraction of multiple ion packages in the ICP-ToF-MS instrument allows nearly the full suite of the periodic table to be determined without pre-selecting isotope packages. Extracted ions are accelerated with the same kinetic energy in a flight tube and the travel time of the ions vary as a function of their mass to charge ratio (m/q), with lighter elements hitting the detector first (Myers et al. 1994, Borovinskaya 2014). The microchannel plate detector allows about 40,000 times frame detection of each extraction (33,000 Hz) in the ICP-ToF-MS (TOFWERK, TOF-R). Based on its speed only, ICP-ToF-MS has the capacity to produce up to five orders of magnitude more pixels per second compared with ICP-Q-MS. However, because each extraction involves very few atoms, a number of ToF extractions must be summed into each mass spectrum to reach a measurable working signal, which also improves baseline characterisation and subtraction (Hendriks et al. 2017). Cumulative extraction also reduces the amount of data to handle. From our experience, a minimum usable bundle of data can be produced in 2 ms (500 Hz) by the summation of sixty-six extractions.
Because large maps are likely to cover multiple phases, a protocol to achieve fully quantitative determination of major and trace elements of multi-phase samples is necessary. Paul et al. (2014) proposed an approach based on mineral identification using different strategies, after which, an internal standardisation (IS) method is applied (Longerich et al. 1996). A selected element at a fixed mass fraction, determined by microprobe analysis or by assuming stoichiometry values, is fixed for each identified phase. Then, a calibration material is assigned for each phase, again relying on user selection. However, there are several limitations using the internal standardisation method for mapping: a risk of false mineral identification or mineral zonation affecting the internal reference value. Lastly, the mass fraction of the element can be significantly different from stoichiometry since solid solution is common in nature and widespread in composition. Deriving chemistry from the theoretical stoichiometry for these cases may thus lead to false results, loss of information and could render false zonation from homogenously distributed elements. The alternative, acquiring microprobe analyses for multiple phases, is time consuming and expensive. For those reasons, mineral identification from LA-ICP-MS maps can be a useful informative tool but undoubtedly presents risks when used to achieve quantitative mass fractions. An alternative method, which we use in this study, is based on the normalisation of the sum of elements in their oxide form (Liu et al. 2008, Selih andvan Elteren 2011), except for minerals with no oxygen such as sulfide, to minimise mistakes in mineral identification and to avoid determining a fixed value for an internal standard for each mineral.
Taking into consideration the requirements above, we have developed a new analytical protocol for fast chemical mapping of all atomic masses, ranging between 23 Na to 238 U, using a time-of-flight mass spectrometer (ICP-ToF-MS, TOFWERK) attached to an 193 nm Excimer laser, equipped with a fast response funnel, that allows quantitative measurement of multiple mineral phases over large areas. The high-resolution maps (12 μm per pixel) are produced at rate of 5 min mm -2 . Since mapping by LA-ICP-ToF-MS is quite recent and under development (Burger et al. 2017, 2 4 4 Bussweiler et al. 2017, Ubide et al. 2019, Rubatto et al. 2020, Sugiono et al. 2022, we detail the optimal analytical parameters required for high-resolution mapping of large areas of thin sections as well as demonstrate that its performance (limits of detection, quality control) is suitable for laser ablation work for various geological applications.

Instruments and settings
Analytical parameters are summarised in Table 1 and are as follows. A RESOlution 193 nm ArF-Excimer laser ablation system (Applied Spectra, California), used in this study, was attached to an ICP-ToF-MS (Model TOF-R, TOFWERK, Thun, Switzerland) built on a ThermoElemental iCap ICP-Q-MS and recently installed at LabMaTer (Université du Québecà Chicoutimi, UQAC). The laser system was equipped with a large S-155 Laurin Technic double-volume ablation cell and controlled by GeoStar software (Norris Scientific). A rotating slit device was used to shape the laser beam into a rectangular of 12 × 6 μm (length × width) to produce high-resolution chemical maps in this study (see Figure S1 in supplementary material for more details about ablation mode and beam shape). Data were acquired in line-scan mode (overlapping mode) with a fluence of 6 J cm -2 , a pulsing rate of 120 Hz, and a beam displacement speed set to 360 μm s -1 . In addition, a small-volume fast-response funnel (FRF) device (Norris Scientific and Laurin Technic) was attached to the laser to ensure rapid transfer of the ablated aerosol out of the ablation cell and through the transfer tubing to the ICP-ToF-MS. The mapping production rate was 30 pixels s -1 , allowing an area of 10 mm 2 to be mapped in 45 min. The fast-response funnel works under positive pressure (~25 kPa) and the S-155 ablation cell was flooded with 1000 ml min -1 of helium mixed with 400 ml min -1 of argon in the FRF cup. An additional 300 ml min -1 auxiliary argon flux was injected into the gas mixture at the torch of the ICP-ToF-MS. To avoid detector saturation and reduce background noise (H 2 O + , 16 O + , 18 O + , 14 N + , 12 C + ), a low-mass cut-off was applied using ThermoElemental QCell technology by setting the collision/reaction cell (CRC) mass and bias voltages to +230.5 V and +0.97 V respectively. Although, Burger et al. (2019), demonstrated that the injection of H 2 and He into the reaction/collision cell can noticeably reduce 40 Ar + and 80 ArAr + interferences and improve selenium detection, no reaction/collision gas nor any N 2 was added to the gas mixture in our experiment. After testing different optional gas mixtures (N 2 /H 2 /He) in our system, it was concluded that the benefits of reducing polyatomic interferences were less than the formation of polyatomic MH + and oxides that formed when adding any gas. Four notch filters are available on the instrument and are RF-only quadrupole to which four auxiliary RF frequencies can be applied to selectively attenuate ions of specific m/z in ICP-ToF-MS (Hendriks et al. 2017). The RF filters were set to reduce signal on masses 40 Ar, 80 ArAr + , 56 Fe and 63 Cu and avoid detector saturation while analysing Fe-sulfides, Fe-oxides, chalcopyrite, and the reference material (RM) MASS-1 (Fe-Cu-Zn sulfide).
The instrument tuning was achieved with NIST SRM 610 glass and GSE-1G (USGS basaltic glass), and the formation of interferences was verified on the mean ion detection of the mapped sample (HN99-46) using the TofDAQ Viewer software (TOFWERK). Oxide and hydroxide interferences were monitored from 248 ThO + / 232 Th + and 249 ThOH + / 232 Th + and were < 1%. Double-charged species were monitored from 141 Pr 2+ / 141 Pr + and 181 Ta 2+ / 181 Ta + and were < 0.5%. Background on the ), 56 (ArO + ) and 54 (ArN + ). Using the 12 × 6 μm beam, total cps were about 1.5 Mcps and 3.0 Mcps on NIST SRM 612 and NIST SRM 610 respectively with a resolution from~2500 (m/ Δm) on 238 U to~2000 (m/Δm) on 23 Na. Isotope precision was assessed from 153 Eu/ 151 Eu on NIST SRM 610 over a 10-s integration period and was 4.4% RSD. Sensitivity on heavy and mid masses ( 238 U and 165 Ho) was similar, about 160 cps per μg g -1 and 150 cps per μg g -1 respectively, and much lower for light masses being at 0.7 cps per μg g -1 ( 28 Si). A total of 107 isotopes from sixty-six elements were treated. Multiple isotopes were acquired for many elements, thus noticeably increasing the quality control of potential interferences. In the interests of brevity, concentrations from a single selected isotope per element (sixty-six in total, Table 1) are presented here but complete data sets are available upon request. To help evaluate the quality control of minerals from the studied sample (HN99-46), SEM analyses were conducted at IOS Services Géoscientifiques (Canada) using a Zeiss Sigma 300 VP field emission scanning electron microscope (FE-SEM) equipped with two Oxford Instruments 170 mm 2 Ultim-Max EDS-SDD. The acceleration voltage was set to 20 kV and an aperture of 30 μm at high current was used to obtain sufficient counts in the EDS detectors with a dead time lower than 40%.

Pixel production rate
The effective wash-out time using specific analytical parameters and instrument set-up was evaluated to optimise the pixel production rate. Both transport speed and efficiency of the aerosol from the ablated site to the ICP-MS are essential to achieve fast mapping at the micrometre level (Van Malderen et al. 2015. The wash-out time, which is defined as the time required for the ablated aerosol to be swept out of the ablation cell and through the transfer tubing (Gundlach-Graham and Günther 2016), is commonly expressed as the time required to reach 1% of the remaining signal (99% wash-out) and noted as full-width at 1% maximum (FW0.01M), compared with its reference maximum peak intensity after a single laser shot (see Figure 1a). As a rule of thumb, the faster the wash-out, the higher the pixel production rate can be achieved when coupled to a fast measurement instrument such as ICP-ToF-MS. A longer wash-out time than the pixel production rate may result in undesirable tailing (memory) effects on the images when running continuous line scan analyses to produce maps. On the other hand, a lower pixel production rate would not take full advantage of the high speed of 2 4 6 et al. 2016), and are referred to as "fast-response-funnels" (FRF). The performance of the Norris Scientific/Laurin Technic FRF used on our laser system was evaluated to optimise the pulsing rate and the beam displacement. The wash-out efficiency greatly depends on the design of the funnel (Gundlach-Graham and Günther 2016, Van Malderen et al. 2016. The effective wash-out time also varies according to: (1) the total amount of aerosol generated by the single-pulse laser, (2) the nature of the elements (volatile versus non-volatile), (3) the isotopic abundance, (4) the mass fraction of the element and (5) the sensitivity of the instrument at the measured mass. For instance, Burger et al. (2017) reported a FW0.01M achieved after 100 ms when using a 44 μm beam while it was reduced down to < 10 ms when using a 5 μm beam. Comparing wash-out times without using identical analytical parameters, materials and instrument settings is thus meaningless and only serves as a crude indicator of the cell potential.
Without the use of the fast-response funnel, the wash-out time of the S-155 cell (Laurin Technic) used in this work was measured at about 420 ms to reach FW0.01M, while it was about ten times faster when using the fast-response funnel (Norris Scientific and Laurin Technic). Using the analytical conditions of the present work (Table 1), we find wash-out times of about 38 ms at FW0.01M and 56 ms at FW0.001M based on an average of the signal from 238 U, 232 Th, 208 Pb, 165 Ho and 28 Si from thirty individual pulses on NIST SRM 610 ( Figure 1a). As shown on Figure 1a, about 6 ms was necessary before the signal reaches the maximum peak. This 6 ms "ramp" should also be considered; thus, the whole event is 44 ms and the ideal acquisition frequency would be 1/0.044 s (= 23 pixels s -1 ). Our results show that a production rate of about 30 pixels per second would be realistic considering that the ablation cell could be washed out to FW0.03M at about that frequency. To verify the efficiency of the fast response funnel, a sequence of 30 Hz pulsing was carried out on GSE-1G. The wash-out between each pulse was about 97% for GSE-1G (FW0.03M, Figure 1b), confirming that a pixel production rate of 30 Hz would be suited to minimise the risk of producing artefact tailing from the wash-out and optimising the production rate.
To produce pixels of square shape, both the stage translation and data production rate must be synchronised with the pixel target production rate, here established at 30 Hz. The data acquisition was thus set to 30 Hz by the integration of 1100 extractions (33,000 Hz instrument frequency extractions / 30 Hz target frequency pixel production) for a total of 33.33 ms of cumulative integration time for each point of data. The pixel width was determined by the beam vertical length, here 12 μm, thus the stage displacement speed was fixed to 360 μm s -1 (30 Hz × 12 μm = 360 μm s -1 ).
While a pulsing effect from the combined oscillating aerosol flow and scan period of sequential quadrupole ICP-MS measurement may lead to aliasing (pulsed / frequency) error , Norris et al. 2021, this undesirable effect is partially resolved by the simultaneous acquisition nature of ICP-ToF-MS. However, signal pulsing could occur from the oscillating aerosol flow, thus a quasistationary mass flow to the ICP-MS should minimise this effect , Norris et al. 2021. The constancy of the material flow was optimised by using a pulse frequency of 120 Hz on the laser. At this rate, a pulse occurs each 8.3 ms leading to a wash-out of about 16% (FW0.84M) between pulses. Consequently, aerosol clouds from the pulses are mixed and no obvious pulsing effect was observed. A simulation was carried out using a 120 Hz pulse rate (Figure 1c) of the laser based on the measured wash-out of a true single test run on NIST SRM 610 ( Figure 1a). It shows that an oscillation of the signal is present when averaging the integrated signal summed over four pulses of the laser; however, the theoretical relative deviation on the mean is negligible at 1.0% (2s). The analysed material generated from the equivalent of four laser pulses is thus accumulated over the 33.3 ms integration period of the ICP-ToF-MS, producing individual pixels with minimum potential for the pulsing effect. The cumulative pulses also improve the signal to noise ratio and reduce the measurement uncertainty compared with pulse-to-pulse mode .

Calibration material
Best practice in LA-ICP-MS recommends the use of matrix-matched reference materials (RM) for the primary calibration (Longerich et al. 1996, Sanborn and Telmer 2003, Danyushevsky et al. 2011, Sylvester and Jackson 2016 or as secondary calibrants to account for matrix effects (Gilbert et al. 2013, Thompson et al. 2021a. A total of six reference materials were used for the new multi-phase calibration method to allow quantification of a wide range of minerals (silicates, oxides, sulfides, phosphates and carbonates). The main block of calibration consisted of three glasses (NIST SRM 610, GSE-1G, GSD-1G) and two sulfides (MASS-1 and UQAC-FeS-1). UQAC-FeS-1 is a nanopowdered pressed pellet developed at LabMaTer ) of a natural Fe(Ni-Cu)-sulfide from Ragland nickel mine (Katinniq, Quebec) and doped with chalcophile elements, including PGE, fused in a carbon crucible and then crushed to a nano-powder and pressed into pellets of 13 mm 2 4 7 diameter using a technique described in Garbe-Schönberg and Müller (2014). A natural apatite from Durango (Lyons 1988, Chew et al. 2016 was also used for the specific calibration of P because the glasses used for calibration only contain a few μg g -1 of P. As underlined by Cook et al. (2016), NIST SRM 610 is best suited for silicates, while GSE-1G and GSD-1G were used with success as calibrants (primary/ secondary) and for quality control of Fe-oxides, (Savard et al. 2012, Dare et al. 2014, Duran et al. 2020, chromite (Arguin et al. 2016), ilmenite (Thompson et al. 2021b), olivine (Batanova et al. 2019), garnet (Guilmette et al. 2018) and sulfides (Duran et al. 2019, Lawley et al. 2020. NIST SRM 610 can also be used for carbonates, such as otoliths (Lazartigues et al. 2014, Jochum et al. 2019, Thompson et al. 2021a).
Finally, MASS-1 and UQAC-FeS-1 are both suited for LA-ICP-MS analysis of sulfide minerals (Wilson et al. 2002, Danyushevsky et al. 2011, Duran et al. 2019, Lawley et al. 2020, Andronikov et al. 2021. Calibration using a variety of matrix RMs has the advantage of producing an extended coverage of the elements because no single reference material can cover all (see online supporting information Table S1). For example, S, Te, Se, Pt, Pd, Os, Ir, Au are noticeably present in higher mass fractions and are better characterised by the sulfide RM compared with the glass RM.

Quality control reference material (QCRM)
In addition to the six RMs used for calibration, sixteen QCRMs were analysed to extend the evaluation of the accuracy of the new calibration protocol for a wide variety of matrices. These include three glasses (G-Probe-15, NIST SRM 612 and NIST SRM 614), a synthetic sulfide doped with PGE and Au (PO-727, Memorial University), a series of natural minerals (see Table S1 for details) and a series of nano-powder pressed pellets prepared at LabMaTer using a Planetary Mill and following the method of Garbe-Schönberg and Müller (2014). These comprise powders of a high purity concentrate of natural apatite (UQAC-APA-1) from Lac-à-Paul Ti-P deposit (Arianne Phosphate, Quebec), an ilmenite-rich sand (AMIS-0454), a natural otolith (in-house RM under development) and three sulfides (PTC-1b and CCu-1e from CCRMP; UQAC-FeS-5, an in-house RM under development at LabMaTer ). A full description of reference materials is provided in Table S1 and the measurement results for the quality control RMs are given in Table S2.

Data reduction
Post-acquisition data processing of ToF data required two steps. The raw data (h5 format) were first treated in Tofware software (TOFWERK) to select isotopes, refine the peak (shape, position, width) and proceed to baseline subtraction (Hendriks et al. 2017). The modified h5 files were then loaded in IOLITE v4 software (Paton et al. 2011) where a new data reduction scheme, the "3D Trace Elements" (3D-TE-DRS) was, in part, developed in cooperation with the IOLITE team during this study in order to convert the raw data to quantitative multi-mineral maps. The 3D-TE-DRS can combine multiple RMs and produce blocks of calibration over time. The normalisation to 100% weight method was used, which has been proven to work well for RM synthetic glasses (Gagnon et al. 2008, Liu et al. 2008. For the calibration, the median of common isotopes of each RM were normalised to GSE-1G and the linear sensitivity were all forced to the origin. In the new 3D-TE-DRS, each individually acquired datum (pixel) in the map can be attributed to a mineral phase based on the signal detection in the instrument from userselected elements and using threshold or gap limits (< or > larger or less than) from as many elements as necessary. For example, sulfide and apatite can be easily identified using the detection threshold signal (> × cps) of 34 S and 31 P respectively, whereas differentiating silicate phases, such as amphibole and pyroxene, will require different threshold or gaps from the same elements. Minor elements can also be of help to refine the identification. An appropriate element oxide form is then applied based on the phase attribution. When a sulfide is recognised, oxygen is excluded from the normalisation. Also, as Fe occurs in various oxidation states, such as Fe 2+ in ilmenite, Fe 3+ in haematite and a combination of Fe 2+ and Fe 3+ in magnetite (equivalent of Fe 3 O 4 ), the appropriate oxide conversion is applied according to mineral identification based on the chemical criteria set by the users. Finally, an appropriate sum for the normalisation is applied to each identified phase and can differ from 100% to compensate for any major component that cannot be detected by LA-ICP-MS, such as CO 2 , OH, H 2 O and the halogens (F, Cl, Br, I). For most anhydrous or volatile-free minerals, the normalisation will be forced to 100%, while calcite, which contains 44% CO 2 , normalisation was forced to 56%. Having the flexibility to adjust the normalisation protocol for each identified pixel is only now possible using the new IOLITE 3D-TE-DRS.

Analysed sample
To demonstrate the versatility of the new mapping protocol, a sample (HN99-46, Figure 2) of magmatic Fe-Ti-P mineralisation from the Critical Zone of Sept-Iles layered intrusion, Quebec, Canada (Nabil 2003, Tollari et al. 2008 containing a diversity of mineral phases (oxides, phosphate, sulfide, carbonate and silicates) was used to produce a 2 mm × 5 mm map. The mineralised sample is dominated by igneous apatite, coarse-grained ilmenite, and titanomagnetite (type-A) containing very fine exsolution of ilmenite ( Figure 2e). In order to demonstrate the high-resolution of the new mapping protocol, the 2 mm × 5 mm mapped area was chosen to include fine-scaled features, such as a latestage vein of serpentine with calcite and pyrite (Figure 2a). A close-up image of pyrite in the mapped area is provided in Figure 2b. Thinner veinlets of serpentine and magnetite (Figure 2c, d) also cross-cut titanomagnetite and apatite. Titanomagnetite, near this late-stage vein, is partially altered (type-B) and contains small inclusions (< 15 μm) of serpentine and chlorite as identified by SEM. The H 2 O content (20%) of serpentine in the vein was calculated using SEM analysis (at IOS Services Geoscientifiques, Chicoutimi) and the normalisation was therefore forced to 80%.

Results and discussion
Since this is one of the first applications of laser ablation mapping of geological samples using ICP-ToF-MS, emphasis is placed on evaluating the method performance and figures of merit regarding limits of detection and quality of the data. In addition to using "external" reference materials to evaluate the quality of the data, fully quantitative data extracted from the mapped area of the different phases is directly compared with data obtained by traditional means from previous work (by LA-ICP-MS, EPMA) and this study

4 9
(SEM analysis). The partitioning behaviour of trace elements for the minerals also supports the evaluation. The advantages and limits of the method are highlighted by the analytical capabilities in the geochemical context of the project presented here.

Limits of detection
Both a lower limit of detection (LLoD) and an upper limit of detection (ULoD) specific to the method must be calculated because of the limited linearity of ICP-ToF-MS. The peak tailing effect on adjacent masses, combined with overlapping peak shapes and stray ions, produce a massspectral baseline that requires post-acquisition peak fitting and baseline subtraction (Figure 3). This ICP-ToF-MS baseline is known to have negative effects on the sensitivity and the stability of the signal for important elements in geology, such as B, Li and K (Hendriks et al. 2017. In addition, approaching detector saturation affects the linearity of the instrument (Hendriks et al. 2017) and can occur in two different ways: (1) the signal on a single mass exceeds 10 Mcps (> 300 ions detection per extraction) and (2) the signal of the entire mass spectrum exceeds 30 Mcps (> 900 ions detection per extraction). Reducing the sensitivity of the light-mass elements ( 23 Na to about 75 As) significantly increases their lower limits of detection (LLoD), but since many minerals of interest are composed of light mass atoms as major components (calcite, magnetite, pyrite, apatite, etc.), working with higher LLoD in this mass range has only a moderate negative impact on the results.
The LLoD were calculated on the baseline instead of the gas blank, since this represents more realistic values for ICP-ToF-spectrometry, as demonstrated by Thompson et al. (2020). The ULoD were estimated from the sensitivity (ions per extraction per μg g -1 ) of each element, assuming saturation occurs on individual peaks at 300 ions detection per extraction. For light masses ranging between 23 Na and 75 As, detector saturation should not be possible under the analytical conditions used here. However, for heavy masses, noticeably 238 U, 232 Th, 205 Tl, 181 Ta, 209 Bi, 175 Lu and 169 Tm, saturation could occur between 5-10% m/m. The broad detector saturation, which can occur from the whole spectrum at 30 Mcps, was not considered in the estimation of the ULoD. Both LLoD and ULoD, using analytical parameters used for mapping (12 µm × 6 μm beam), are provided (Table S3) and are summarised in Figure 4. The LLoD for elements < 85 amu ( 23 Na to 85 Rb) generally range from 15-350 μg g -1 , with the highest LLoD at 2500 μg g -1 ( 34 S). LLoD for elements > 85(Rb) masses generally range between 0.15 to 5 μg g -1 , with the lower values ranging between 0.11 and 0.15 μg g -1 for U, Th, Tl and Ta. This significant difference in LLoD between the light masses and mid-heavy masses is due, in part, to the reduced sensitivity applied by the low-mass cut-off setting on the Qcell. The negative impact of the higher LLoD for light masses is minimal for many minerals, whereas elements of mid-heavy masses (REE and most transition metals) commonly occur as minor to ultra-trace mass fractions and thus require much lower LLoD. Low levels of U and Th (< 0.12 μg g -1 ) can potentially be measured in a very small volume of sample. We evaluated that each laser pulse was removing  ULoD LLoD 2 5 0 about 0.1 μm of material, and the integration of four laser pulses (12 × 6 μm beam) used to produce each pixel of data would thus remove a volume of about 28.8 μm 3 of material. Assuming a density of a mineral being about 3.5 g cm -3 , the integrated mass sample per pixel equivalent is 1 × 10 -10 g (100 pg) and the total amount of U and Th potentially detected in this sample is 1.2 × 10 -17 g (0.012 fg), which is about two orders of magnitude lower that the limits of detection reported elsewhere based on a test portion of 4 μg of material (Macholdt et al. 2014).
The effective linear dynamic range for light masses for the method are as low as three to four orders of magnitude, whereas it is generally six orders of magnitude for the mid and heavy masses. These observations are in good agreement with previously reported performance of the ICP-ToF-MS (Bleiner et al. 2000, Tanner and Günther 2008, Hendriks et al. 2017. As a consequence, the determination of trace elements (< 10 μg g -1 ) in the light mass range (< 85 Rb, with the exception of Ga) is not possible with the instrument settings used. Also, the detector saturation can be, in some cases, a limiting factor in terms of flexibility in mapping analysis of multiple mineral phases. Minerals with major elements having multiple isotopes of high abundance, such as Pb (208, 207 and 206), would make it impossible to 'notch' all three main isotopes and thus would make the analysis of minerals, such as galena (PbS), impossible as saturation is expected to occur at about 11% m/m, comparable to an experiment conducted by Gundlach-Graham et al. (2018) using a 5 μm beam. The same issue is expected to occur for platinum-group minerals and alloys, and REE-rich phases. On the other hand, the analysis of minerals for which major elements are mono-isotopic or are dominated by a single isotope, such as uraninite (UO 2 ) with 238 U (natural abundance of 99.3%), pyrochlore [(Na, Ca) 2 Nb 2 O 6 (OH,F)] with mono-isotopic 93 Nb and arsenopyrite (FeAsS), might be possible if the notch on 63 Cu is switched to 238 U, 93 Nb or 75 As respectively. This implies that these minerals could be analysed only under specific notch filter tuning.

LLoD comparison with LA-ICP-Q-MS
Compared with single-quadrupole ICP-MS, time-of-flight spectrometers have lower sensitivity in solution mode (Hendriks et al. 2017). To compare the LLoD with ICP-Q-MS (see results in Figure 5), the laser was attached to an Agilent 7900 ICP-Q-MS using the same laser settings. However, the fast-funnel was not used since the wash-out time is too fast for the acquisition time of the ICP-Q-MS. A total of fifty-five isotopes were pre-selected and a dwell time of 5 ms was set to each mass for a total sweep time of 0.377s per cycle. 44 Ca was used as the internal standard element and the LLoD as obtained from the USGS basaltic glass GSE-1G was used for the comparison, thus data are missing for 34 S, 77 Se and the PGE. Except for 28 Si, which has comparable LLoD in both instruments, the LA-ICP-ToF-MS LLoD are typically higher for the light masses by a factor of about 10 to 1000 (  39K  47Ti  55Mn  60Ni  71Ga  77Se  89Y  95Mo  105Pd  115In  125Te  139La  146Nd  157Gd  165Ho  172Yb  181Ta  189Os  197Au 208Pb 238U 2 5 1 compared with quadrupole and for masses 205 Tl to 238 U the LLoD are 5× higher by ToF-MS. The detailed data are provided in Table S3.

LoD (ToF-glasses only) LoD(Q-MS) LLoD ToF-MS / LLoD Q-MS
It should be noted however that acquiring data in such conditions by ICP-Q-MS would not be possible for mapping. The sweep time of ICP-Q-MS, being 0.377 s, would limit the pixel production rate to less than 3 pixels s -1 . With a displacement speed of 360 μm s -1 , each pixel would have been about 120 μm wide. To produce a 12-μm wide pixel, the beam displacement speed should be set to 36 μm s -1 instead of 360 μm s -1 . In doing so, the cumulated pulses would have been forty instead of four to produce one pixel. Assuming that the ablation rate is about 0.1 to 0.3 μm per pulse, cumulating forty pulses would have produced a laser trench of about 4 to 12 μm depth. At such a depth, the angle of ablation would be very steep and the material ejection from the plume at the impact site would be oblique (Conde et al. 2004). Therefore, a true effective comparison between ICP-Q-MS and ICP-ToF-MS is not possible as the trade-off of the apparent lower sensitivity of the ICP-ToF-MS is counterbalanced by the effective, fast data acquisition.

Quality control
Reference materials: Comparisons of results for reference materials with working values are shown in Figures 6, 7 and 8, and the statistics are provided in Table 2. Mass fractions below LLoD were removed and the data set comprised 605 measurement results, averaged from two measurements from a total of twenty-two reference materials. Detailed results are provided in Table S2. Analytical conditions were the same for the RMs as for the mapping experiment (Table 1). It should be noted that some analyses were as short as 0.5 to 2.5 s because the speed of beam displacement (360 μm s -1 ) covered rapidly the small grains (< 1 mm) of natural minerals, such as the Madagascar apatite, Laflamme PO-727 sulfide, chromite CHR-M1a, R-10 rutile, and the Mongolian Olivine (sh11-2). Furthermore, the natural minerals may have heterogeneities, thus resulting in large uncertainties in the working values (Eggins and Shelley 2002, Danyushevsky et al. 2011, Gilbert et al. 2013, Batanova et al. 2019, Thompson et al. 2021a) and/or there could be low confidence in the working values resulting from challenging elements and matrices. Therefore, we propose a simplified approach in order to track systematic problems that could increase the risk of false measurement on the unknown samples.
In Figure 6, the results from the analysed RMs are compared with their working values according to the following three categories: glass RMs, natural RMs and sulfide RMs. For each datum (i), an absolute relative difference (ARD%) was calculated by comparing measured values (m i ) with their working value (W V i ) from the following equation: (1) For clarity and simplification, each datum is subdivided into three classes based on their ARD(%): data with less than 25% ARD are considered as acceptable. This threshold is based on the relative uncertainties (U) of both the reference   2 5 2 material and the analytical uncertainties which were trivially estimated at 15% and 20% (RSD%) respectively for all RMs, and cumulated by the equation U Budget 2 = (U RM 2 + U Measurement 2 ). A second level was set for data ranging between 25% and 50% of ARD to the working values. The data mostly plot close to the accepted values ( Figure 6), however, users should be aware that the occurrence of lowquality results is possible. Finally, data with > 50% ARD of the working values are flagged as "potentially problematic" and were investigated to find if any bias or systematic problems occurred during data acquisition.
From the overall data set (n = 605), 76% are within < 25% ARD, 14% are between 25% to 50% ARD and only 10% are > 50% ARD (Table 2), which is encouraging and suggests that the new mapping protocol produce nearquantitative measurements that are fit for purpose for most geological contexts. From the overall data quality assessment displayed in Figures 6, 7 and 8, no systematic problems were found regarding matrices, atomic mass or mass fraction level. The atomic masses with the most problems (> 50% ARD) are typically between 185 (Re) and 208 (Pb) which include PGE, Au, W, while masses ranging close to REE are for most < 50% ARD. Both major and minor elements (> 1000 μg g -1 ) for the most part had < 25% ARD (Figure 8).
The complex interactions in LA-ICP-MS (volatility, condensation, transport, ionisation, etc.) in addition to the level of characterisation and inhomogeneous distribution of the elements in the RM may provide some explanation. Our data suggest that fractionation could differ in the mafic versus felsic nature of the glasses from this experiment, and thus more investigation would be needed to better understand the complexity of multiple matrix-induced fractionation. Extensive testing of the new fast-funnel would also be required to evaluate its impact independently on the presented results. A recent study demonstrated that the fine tuning of laser focus on the surface could be another factor contributing to fractionation (Huang et al. 2021), and using a small beam, such as in the present work, is likely to be affected by this parameter and could contribute to the quality of some data. Improving the quality of the data would be profitable, but while homogeneous and wellcharacterised RMs are mostly limited to glasses, it will remain challenging to achieve. Users should be warned that despite the overall good quality, the data should still be interpreted with caution. This highlights a problem commonly raised about the lack of good quality matrix-matched reference materials for the calibration and quality control of LA-ICP-MS analysis, which is noticeable for sulfide minerals (Günther and Hattendorf 2005, Fernández et al. 2007, Danyushevsky et al. 2011, Yuan et al. 2012, Feng et al. 2018, Sio et al. 2019, Phyo et al. 2020, Steenstra et al. 2020.
Discussing each occurrence of potential problematic measurement is beyond the scope of this work and more detailed assessment of the quality control of the results for the glass, natural and sulfide reference materials is provided in online supporting information Appendix S1.
New mapping protocol: An example  Table 3. Some features of the sample are not readily visible on the reflected light photomicrograph (Figure 2), such as calcite in the serpentine vein, the fine ilmenite exsolution lamellae in titanomagnetite and the difficult distinction between titanomagnetite and ilmenite. However, the phase identification ( Figure 9) highlights their presence and thus can help to support petrographic observation. Figure 10 presents RGB images produced in IOLITE v4 software of the distribution of major elements, highlighting fine-scale details of a serpentine veinlet (20-50 μm wide) using Si-Al ( Figure 10a) and Mg (Figure 10b), as well as some ilmenite exsolution lamellae (5-10 μm wide) in titanomagnetite using Ti (Figure 10b). The ilmenite exsolution lamellae cannot be considered as 100% resolved because the beam used was 12 μm. They were however detected, which is another good example of the fine spatial resolution of the method using the fast-response-funnel (Norris Scientific and Laurin Technic). Figures 10a, b highlight a chemical change of Mg and Si in the micro-veinlets of serpentine when hosted by different minerals. Iron oxide is present in the serpentine vein where titanomagnetite is the host mineral, whereas serpentine mostly fills the vein when apatite is the See Table S1 for the list of reference materials.
2 5 4 host mineral. The chemical changes recorded by the new mapping protocol agree with the petrographic observations ( Figure 2) and is another demonstration of the resolving power of the new protocol. Figure 10c highlights the relation between S, Mn and Ca. Calcium is present in both apatite and calcite, while Mn is enriched in calcite and present as well in ilmenite and magnetite respectively.
Quality control of major elements in the sample: The IOLITE "Inspector" (Petrus et al. 2017, Chew et al. 2021 is an elegant tool for the integration of a selected area and was used on the mineral phases to extract major and trace element mass fractions. A comparison of the extracted mineral data from the map with values from EPMA measurement (Nabil 2003, Tollari et al. 2008, Méric 2011, SEM analysis (IOS Service Géoscientifiques) and stoichiometry is provided in are good examples to demonstrate that fixing the mass fraction of an internal standard based on stoichiometry could lead to bias of all measured elements. The measured Fe 3 O 4 (87%) in the magnetite from sample HN99-46 is lower than the stoichiometry (100% Fe 3 O 4 ) because TiO 2 (8-12%), MgO (0.9-1.2%) and Al 2 O 3 (1.8-2.8%) are also present in the natural oxide, as determined in this work and confirmed by EPMA and SEM (Table 4)   and P (blue); (c) S (red), Mn (green) and Ca (blue).
2 5 5 major elements match closely to the SEM data, except that SiO 2 is lower by 20% relative and balanced by 18% relative higher FeO. A matrix effect could explain this difference, such as the presence of water (assumed at 20% by EPMA) that could affect the ablation and ionisation processes relative to the free-H 2 O reference material. Using a matrix-matched serpentine as secondary calibrant would be advised in the future. The good overall agreement for the major elements supports the data accuracy of the new mapping protocol (i.e., it is nearly fully quantitative).
Quantification of trace elements: Trace elements cannot be measured by EPMA or SEM or compared with stoichiometry, thus, a series of selected trace element maps are provided and discussed in more details in online supporting information for each main mineral phase (see Figures S5 to S9). In summary, trace element data that were extracted from the map using the inspector tool compared well to LA-ICP-Q-MS values from previous work on this thin section (Tollari et al. 2008, Méric 2011, Kieffer et al. 2022. The distribution of the measured trace elements are also supported by their partitioning behaviour for the minerals as reported in the literature (Bédard 2001, 2014, Milani et al. 2017, Arguin et al. 2018, Duran et al. 2020. The minimum mass fraction on the scale for each elemental map was fixed to their respective LLoD, thus removing any pixel < LLoD.
Ablation yield and total effective yield (TEY): As previously mentioned, internal standardisation or normalisation is necessary to correct for differences in ablation yield to achieve fully-quantitative measurement (Jackson et al. 1992, Jochum and Stoll 2008, Liu et al. 2008. The overall signal response in certain phases such as iron oxide is lower compared with more easily ablated phases, such as sulfide and hydrated silicate (see Table 5 and discussions below). Semi-quantitative or qualitative approaches in LA-ICP-MS mapping are often based on the raw signal and reported in terms of raw counts per second (McCarron et al. 2014, Ubide et al. 2015, Gourcerol et al. 2018. Although it can be very useful, this approach can still represent a risk that can potentially lead to erroneous results and interpretation, not only for the mass fraction (semi-quantitative) of the elements in each phase, but also in the perception of their spatial distribution. It is thus essential to underline the importance of tracking and apply corrections regarding the ablation yield when no fixed internal standard element can be used. This work supports the useful normalisation approach and highlights the risk of potential error associated with mapping using semiquantitative analysis or raw counts to study elemental distribution in the analysed samples.
The laser ablation yield is described as the amount of material ablated (Jeffries et al. 1995) and can be affected by the depth of ablation, the laser conditions (wavelength, energy, beam size) in addition to the physical and chemical properties of the ablated material (Koch et al. 2008, Van Malderen et al. 2016. The rate of ablated material can significantly differ amongst different minerals and will have an effect on the sensitivity, but a multitude of complex interacting processes, such as transport and ionisation in the plasma, are also involved in the detection of atomic components (Van Malderen et al. 2016). Consequently, the term "ablation yield" is more appropriate to designate the amount of ablated material, while we propose to use the "total effective yield" (TEY) to include all accumulated causes that influence the variation of sensitivity between the ablated phases. To our knowledge, no detailed information is available about the relative TEY between phases, although this information can be easily extracted from a quantitative analysis (map and reference materials). Using the IOLITE calculator tool (Petrus et al. 2017), the TEY of mineral phases using six major elements (Fe, Mg, Si, Al, Ca and S) were obtained by dividing the mass fraction (μg g -1 ) with their net count per seconds (ToF ions/extraction). The six major elements provided similar distributions of the TEY, with the example of Fe plotted in Figure 11, illustrating that sulfide clearly has a higher TEY than the other phases. Results extracted from the ToF map using the inspector tool, along with the six minerals from the sample are provided in Table 5. The unaltered titanomagnetite (Type-A) in the mapped sample has the lowest TEY and was thus used as the reference for relative comparison with other materials ( Table 5). The TEY of other oxides (chromite, ilmenite and rutile) and apatite are comparable (< 1.18), while pyrite is the mineral with the highest TEY (4.54) relative to titanomagnetite Type-A. An extended relative TEY table comprising a total of thirty-one samples, analysed in the same condition as of the present work, and including reference materials and other natural minerals is available in Table S5. As underlined by Cook et al. (2016), mass fraction cannot be inferred across different phases from the raw signal. An example of such an effect on the Fe and Ti distribution is provided in Figure 12, where the normalised mass fraction maps determined in this study are compared with the raw signal maps. The raw signal map of Fe ( Figure 12a) suggests at first that Fe in the pyrite is much higher than any other phases, whereas Fe is actually higher in the magnetite phase, as shown on the quantitative map ( Figure 12b). On the other hand, the raw signal map of titanium ( Figure 12c) is almost similar to the quantitative Ti map (Figure 12d) because Ti is mainly present in ilmenite and magnetite, and these two phases have comparable TEYs (Table 5).  Figure 11. Distribution of the calculated "total effective yield" (TEY) for Fe in the sample HN99-46; (srp) serpentine, (apa) apatite, (mag) magnetite, (ilm) ilmenite, (py) pyrite, (cal) calcite.

Conclusions
A new LA-ICP-MS protocol for the simultaneous determination of trace and major elements in multiple minerals combining a 193 nm laser equipped with a fastfunnel device and coupled to a time-of-flight mass spectrometer is presented. The technique combined multiple reference materials of varied matrices (sulfide, mafic-glasses, felsic-glasses, natural apatite) to produce blocks of calibration. The data were quantified using a recently developed data reduction scheme in IOLITE v4 (3D-TE-DRS). The protocol thus provides an extended coverage of the elements and points toward best practice in LA-ICP-MS by using (multiple) matrix-matched calibration materials. Mineral recognition is possible using the new IOLITE 3D-TE-DRS, and appropriate conversion of oxide forms is applied and correction for undetected components such as CO 2 , F, Cl, H 2 O is possible to achieve quantitative results by personalised normalisation for contrasting minerals such as sulfides, silicates and carbonates. The normalisation to 100% oxide approach avoids having to use external determination of major elements, by for example electron microprobe analysis, to use as internal standards. In addition, we introduce the concept of "total effective yield" (TEY) and show that there can be significant difference between some minerals, which emphasises the importance of the normalisation approach to achieve quantitative measurement. The technique has the advantage of producing chemical images at a rate of 30 pixels s -1 , which is about ten times faster compared with methods for quadrupole-MS. The analysis of multiple quality control reference materials and a comparison of the analysed minerals with other external sources (LA-ICP-MS, SEM, EPMA) demonstrated that the quality of the data fit the purpose. However, like any methods under development, there is room for improvement. The proximity of limits of detection, confidence on the working mass fractions of the reference material used, possible fast-funnel induced fractionation, or problems with the tuning of the ICP-ToF-MS itself are to be improved but are beyond the scope of this work. Consequently, users must be aware that uncertainties can affect the measurement results. In summary, the protocol offers efficiency and high analytical speed for fine resolution chemical mapping of trace and major elements over much larger areas than previously available. quantitative Ti (μg g -1 ). The Fe distribution differs significantly while Ti had quite similar distribution.
2 5 9 suggestions. There are no conflicts of interests of author and co-author.

Data availability statement
Data sharing is not applicable to this article as no new data were created or analyzed in this study. materials (NIST SRM 610, GSD-1G, GSE-1G, MASS-1 and UQAC-FeS-1). Figure S3. Trace element associated with apatite from sample HN99-46. Figure S4. Mantle-normalised multi-element diagram for apatite trace elements extracted from the new mapping protocol (this work) and compared with previous laser work on the same thin section (Tollari et al. 2008, Kieffer et al. 2022.     Appendix S1. Discussion of (1) ablation mode and beam shape and (2) quality control. (3) Quantification of trace elements in sample HN-99-46. Table S1. List of reference materials used to calibrate and monitor the new multi-phase mapping protocol. Table S2. Measurement results for analysed reference materials. Table S3. Details results of lower limits of detection (LLoD), upper limits of detection (ULoD), and LLoD comparison with LA-Q-ICP-MS. Table S4. Comparison of the measured trace elements in ilmenite and titanomagnetite in sample HN99-46 with previously published data (Méric 2011).