53Mn and 60Fe in iron meteorites—New data, model calculations

We measured specific activities of the long‐lived cosmogenic radionuclides 60Fe in 28 iron meteorites and 53Mn in 41 iron meteorites. Accelerator mass spectrometry was applied at the 14 MV Heavy Ion Accelerator Facility at ANU Canberra for all samples except for two which were measured at the Maier‐Leibnitz Laboratory, Munich. For the large iron meteorite Twannberg (IIG), we measured six samples for 53Mn. This work doubles the number of existing individual 60Fe data and quadruples the number of iron meteorites studied for 60Fe. We also significantly extended the entire 53Mn database for iron meteorites. The 53Mn data for the iron meteorite Twannberg vary by more than a factor of 30, indicating a significant shielding dependency. In addition, we performed new model calculations for the production of 60Fe and 53Mn in iron meteorites. While the new model is based on the same particle spectra as the earlier model, we no longer use experimental cross sections but instead use cross sections that were calculated using the latest version of the nuclear model code INCL. The new model predictions differ substantially from results obtained with the previous model. Predictions for the 60Fe activity concentrations are about a factor of 2 higher, for 53Mn, they are ~30% lower, compared to the earlier model, which gives now a better agreement with the experimental data.


INTRODUCTION
Most meteorites are routinely measured by accelerator mass spectrometry (AMS) for the radionuclides 10 Be, 26 Al, 36 Cl, and (more rarely) 41 Ca, which, if combined with concentrations for cosmogenic noble gases, provide information on cosmic ray exposure (CRE) histories, that is, CRE ages, terrestrial ages, pre-atmospheric sizes, and shielding depths. In contrast, the data for two other cosmogenic radionuclides, 53 Mn (T 1/2 = 3.7 AE 0.4 Ma; Honda and Imamura 1971) and 60 Fe (T 1/2 = 2.61 AE 0.04 Ma), are scarce because for their measurements dedicated AMS systems are needed to generate ions of 150-200 MeV energy, which is required to reduce the otherwise strongly interfering isobaric background and to achieve a sufficiently high sensitivity. Only two AMS facilities report routine 53 Mn and 60 Fe measurements. Before AMS, 53 Mn has been measured via radiochemical neutron activation techniques (e.g., Imamura et al. 1980), but there are currently only very few suitable nuclear reactors available for such studies.
The 60 Fe half-life has been under debate for the last few decades. The value of T 1/2 = 2.61 AE 0.04 Ma determined by Rugel et al. (2009) was confirmed recently by additional independent measurements (Wallner et al. 2015;Ostdiek et al. 2017). The new value is one order of magnitude higher than the first estimate of T 1/2 = 0.3 AE 0.9 Ma (Roy and Kohman 1957) and~75% higher than the previously adopted value of T 1/2 = 1.49 AE 0.27 Ma .
Another challenge in 60 Fe measurements is due to the fact that 60 Fe in iron meteorites is only produced from 62 Ni and 64 Ni. Since both Ni isotopes have a low abundance ( 62 Ni = 3.63%, 64 Ni = 0.93%), the 60 Fe production rates are relatively low, that is, in the range of disintegration per minute per kg (dpm kg À1 ) or less (e.g., Knie et al. 1999b;Nishiizumi and Honda 2007). Because the concentration of stable 56 Fe atoms is high in meteoritic material, the resulting 60 Fe/ 56 Fe ratio is low, that is, in the range 10 À14 . Such low isotope ratios cannot at present be measured with AMS systems commonly used for 10 Be, 26 Al, 36 Cl,or 41 Ca. For 53 Mn, it is necessary to remove or suppress the ubiquitous, interfering isobar 53 Cr. It is not widely appreciated that this can make an AMS measurement of 53 Mn as challenging or even more so than AMS measurements for 60 Fe. As a consequence, 53 Mn measurements of meteorites require similarly large AMS facilities (~14 MV tandem accelerators) as needed for 60 Fe.
Against those odds, 53 Mn and 60 Fe both have the potential to constrain CRE histories of iron meteorites, potentially even better than the classical lighter radionuclides. First, 60 Fe is only produced from a single target element, that is, Ni, and 53 Mn is only produced from Fe and Ni. Consequently, neither nuclide is affected by problems caused by variable contributions from the lighter elements sulfur and phosphorous (with the exception of sample dilution, which typically is irrelevant compared to overall uncertainties of the data). Very often microinclusions of troilite (FeS) and schreibersite ([Fe,Ni]P 3 ) compromise detailed studies of the mainstream cosmogenic nuclides 10 Be, 26 Al, and 21 Ne and therefore limit their use for accurate determination of CRE histories of iron meteorites. Second, due to its relatively long half-life compared to 10 Be, 26 Al, 36 Cl,and 41 Ca,53 Mn is less affected by decay during terrestrial residence, which makes the interpretation easier, because most meteorites are finds and not falls. Consequently, although the measurements of 53 Mn and 60 Fe are more challenging than the measurements of the classical radionuclides, they can provide valuable additional constraints on CRE histories of iron meteorites.
Recently, 60 Fe (and in some cases 53 Mn) have been analyzed in deep sea materials (crusts, nodules, sediments), Antarctic snow, and the lunar regolith as a tracer for nearby supernova explosions (e.g., Knie et al. 1999aKnie et al. , 2004Fitoussi et al. 2008;Wallner et al. 2016;Fimiani et al. 2016;Ludwig et al. 2016;Koll et al. 2019aKoll et al. , 2019bWallner, personal communication). Considering meteorites, there are only very few studies that include 60 Fe. The first 60 Fe measurement in a meteorite was performed by Goel and Honda (1965) by decay counting of the radioactive daughter 60 Co in 2.5 kg of the chemically processed iron meteorite Odessa. Kutschera (1984) described the first successful 60 Fe detection by AMS of a meteorite sample (Treysa). Knie et al. (1999b) presented 60 Fe activities in the iron meteorites Dermbach and Tlacotepec and in the metal fractions of the mesosiderite Emery and the LL chondrite Saint-S everin. More recently, Berger et al. (2007) studied the 60 Fe shielding dependence in the three iron meteorites Canyon Diablo, Grant, and Dorofeevka, which cover a large range of different shielding conditions. They found that 60 Fe production rates in large iron meteorites decrease with increasing shielding but that the trend in smaller iron meteorites is not that clear; the production rates might also slightly decrease with increasing shielding. In addition, Nishiizumi and Honda (2007) measured 60 Fe activities in six iron meteorites using low-level counting and found a good correlation between 60 Fe/kg(Ni) and measured 53 Mn activities. The Odessa sample studied by Nishiizumi and Honda (2007) was measured before by Goel and Honda (1965). The large discrepancy of a factor of six in specific 60 Fe activities was explained by a change of the 60 Co half-life in-between the two studies (Nishiizumi and Honda 2007). The latest study that includes both 53 Mn and 60 Fe was for the very large iron meteorite find Gebel Kamil (Ott et al. 2014). The Gebel Kamil data were used together with data from the aforementioned AMS studies to constrain the activity ratio of 60 Fe to 53 Mn of (2.68 AE 0.35)910 À3 (dpm kg À1 [Ni]/dpm kg À1 [Fe]). This ratio allowed workers to disentangle the cosmogenically produced 60 Fe from interstellar 60 Fe in lunar and terrestrial material Koll et al. 2019aKoll et al. , 2019b. To summarize, for the last 12 years, only one meteorite study was published that analyzed both heavy radionuclides.
The data presented here are part of a larger project determining CRE histories of iron meteorites and thereby studying the constancy of galactic cosmic rays (Smith et al. 2019). In the course of this project, we measured cosmogenic noble gas and radionuclide concentrations in~60 iron meteorites, mainly from group IIIAB. We chemically separated 53 Mn and 60 Fe and prepared AMS targets for all of them; so far 28 were measured for 60 Fe and 41 were measured for 53 Mn (plus six additional samples from the large Twannberg iron meteorite). Although the database is not yet complete, it may nevertheless help to better understand the production systematics of 53 Mn and 60 Fe in iron meteorites and to validate the improved model calculations for cosmogenic nuclide production in meteorites (Ammon et al. 2009 (Cook et al. 2018). In addition, we studied six samples from the large Twannberg iron meteorite (for the corresponding lighter radionuclides, see Smith et al. 2017).

Sample Preparation
In a related project, all samples have been studied for 10 Be, 26 Al, 36 Cl, and 41 Ca as well as for the light cosmogenic noble gases He, Ne, and Ar (Smith et al. 2019). The chemical preparation was performed at the Dresden Accelerator Mass Spectrometry (DREAMS) facility of the Helmholtz-Zentrum Dresden-Rossendorf (HZDR) and was adapted from the procedure described earlier by Merchel and Herpers (1999). A full description of the chemical separation procedure is given by Smith et al. (2017Smith et al. ( , 2019. Here, we give some details, especially for the 53 Mn and 60 Fe analysis. The solution from the anion exchange (7.1 M HCl fraction, height 20 cm, diameter 1 cm, DOWEX 1X8, 100-200 mesh) containing mainly Mn was further purified from the interfering isobar 53 Cr using the following procedure: First, the solution was evaporated to dryness on a hot plate. The residue was dissolved in a mixture of 5 ml H 2 O, 5 ml HNO 3 , and 0.5 ml H 2 O 2 to reduce Mn 4+ to Mn 2+ . The solution was then heated for~1 h to fully destroy H 2 O 2 . Subsequently, KClO 3 was added to oxidize Mn 2+ back to Mn 4+ and finally precipitate it as MnO(OH) 2 by heating for~1 h. This precipitate was rinsed three times with water, subsequently transferred into microreaction vessels (Eppendorf tubes), and finally dried at 80°C in an oven. In an attempt to speed up the chemical processing, we slightly changed the protocol for one sample batch, which resulted in chemical yields larger than 100% (e.g., Turtle River, Elyria, Casas Grandes). An SEM-EDX (scanning electron microscope/energy-dispersive spectroscopy) measurement of these samples revealed that they were contaminated with AgCl from the earlier Ag 36 Cl AMS target preparation. Hence, we "rescued" the MnO 2 samples by washing them three times with~12.5% NH 3aq solution to dissolve the AgCl. A second SEM-EDX scan of these samples after the cleaning showed that the precipitate was pure MnO 2 . We applied the original longer protocol for all remaining samples. The MnO 2 powder was mixed with Ag powder, with a mass ratio MnO 2 :Ag = 1:4 and was then pressed into Cu sample holders.
For the 60 Fe analysis, the isobar 60 Ni introduces an interfering background in AMS, a background that needs to be suppressed during sample preparation. We applied the following procedure: During the anion exchange step, Fe 3+ is present as a chloro-complex [FeCl 4 ]and is absorbed on the DOWEX 1x8. Nickel is eluted with 10.2 M HCl, while Fe can be stripped after all other elements by elution with H 2 O (27 ml). The iron was then precipitated as FeO(OH) by adding~9 ml of 25% NH 3aq . The precipitate was rinsed three times with dilute ammonia solution (i.e., two drops of 25% NH 3aq in 250 ml H 2 O), then dried as iron oxide in an oven at 90°C, and was later ignited at 800°C for~2 h. The iron oxide powder was mixed with Ag powder, with a mass ratio Fe 2 O 3 :Ag~1:2, and was pressed into Cu sample holders for subsequent AMS measurements.

AMS Measurements
The nuclide 60 Fe was measured at the ANU Canberra relative to the PSI-12 standard material, which was produced at the Paul Scherrer Institute (PSI) from a dilution series that is based on material extracted from a beam dump. The 60 Fe/Fe ratio of PSI-12 is 1.234(7) 9 10 À12 ). The original material was used for the half-life measurements of 60 Fe (e.g., Rugel et al. 2009). The Munich group used a primary standard with a concentration of 60 Fe/Fe = (9 AE 1) 9 10 À12 , which is described in Knie et al. (1999aKnie et al. ( , 1999b. All 53 Mn measurements were performed at the ANU Canberra relative to a piece of the Grant iron meteorite that was provided by Greg Herzog (personal communication); the nominal 53 Mn/ 55 Mn ratio is 2.59 9 10 À10 (Gladkis 2006). This value has been obtained measuring the 53 Mn activity in 200 g of the iron meteorite Grant via the 53 Cr K a -line. The ratio used at ANU is 10% lower than the value used by others.
Note that a change in AMS standards and/or halflives has a direct influence on the meteorite data. If we consider, as an example, the radionuclide 60 Fe we measure the concentration of 60 Fe atoms in a sample, whereas the given production rates are saturation activities calculated via decay constant times the nuclide concentration. The recent change in the half-life of 75%, therefore, reduces the production rates by the same 75%. Consequently, all data that are based on the old half-life must be reduced by 75% to be comparable to the new measurements.
ANU Canberra: Samples were loaded into an MC SNICS ion source that is equipped with a sample wheel holding up to 32 positions. Either MnO -(for 53 Mn) or FeO -(for 60 Fe) was extracted and injected into the 14UD tandem accelerator at the Heavy Ion Accelerator Facility (HIAF; Fifield et al. 2013;Wallner et al. 2015). For these measurements, the 14UD accelerator was operated at terminal voltages between 13.8 and 14.3 MV. By selecting charge states of 11+ for Fe and 12+ or 13 + for Mn, we obtained particle energies between 165 MeV for 60 Fe and up to 200 MeV for 53 Mn. Typical currents were several µA of FeO À and several 100 nA for MnO -. Beam intensities of the stable isotopes 54 Fe, 56 Fe, and 55 Mn, respectively, were measured with Faraday cups at the low-and highenergy side of the spectrometer. The rare isotopes 53 Mn and 60 Fe were directed into a gas-filled magnet (ENGE spectrometer converted into gas-filled mode) and then counted atom by atom in a multi-anode ionization chamber (Fifield et al. 2013;Martschini et al. 2019). The gas-filled magnet allows for a spatial separation of the stable isobar from the radionuclide due to their different mean charge states and therefore different deflection angles caused by the interaction with the gas. This separation blocks the majority of the stable isobars from entering the particle detector and consequently reduces the beam intensity of the background isobars in the ionization chamber to acceptable levels (typically less than 100 60 Ni and up to a few 1000 53 Cr events per second). The background for 60 Fe was measured as low as 60 Fe/Fe = 3910 À17 (typical blanks for our study 60 Fe/Fe~10 À16 ; Wallner, personal communication) and the 53 Mn-background was in the range 53 Mn/ 55 Mn <10 À12 . The reproducibility of the AMS measurements, based on repeated measurements of identical samples, is 3-5% for 60 Fe and~5-10% for 53 Mn, respectively.
TUM Munich: The AMS facility in Munich at the Maier-Leibnitz-Laboratory in Garching is also based on a 14 MV tandem accelerator combined with a gas-filled analyzing magnet system. The isobaric background for the 60 Fe measurements was reduced in the same way as described above for the ANU Canberra setup Koll et al., 2019aKoll et al., , 2019b.

NEW MODEL CALCULATIONS
Our previous calculations modeling the production rates of 53 Mn and 60 Fe were based on the nuclear reaction cross sections that were available at that time (Merchel et al. [2000] and references therein), on adjusted cross sections for the neutron-induced reactions (Ammon et al. 2009;Leya and Michel 2011), and on the depth-and size-dependent spectra for primary and secondary particles calculated using Monte Carlo methods. For the production of 53 Mn from nat Ni and 60 Fe from nat Ni, the cross section database was limited and/or the data scattered far outside the range of the given uncertainties (e.g., Merchel et al. [2000] and references therein; Ammon et al. 2009). The earlier model predictions for 53 Mn overestimated measuredspecific activities for the meteorite Grant by up to 50%, which Ammon et al. (2009) argued could be due to AMS normalization problems for the proton-induced cross sections and/or the thick target production rates used to determine the neutron-induced cross sections. Note that such problems would only partially cancel out during the adjustment procedure used to determine the neutron-induced cross sections.
As already stated by Ammon et al. (2009), the model predictions for the production of 60 Fe were much lower than most of the experimental data. The calculated upper limit for the 60 Fe-specific activity was 0.9 dpm kg À1 (Ni) and was at the center of a 25 cm iron meteoroid. Since then, the recommended half-life value for 60 Fe has seen an increase of~75%. Consequently, it is possible that the discrepancies between the earlier model predictions and the experimental data were (at least partly) caused by problems related to the experimental input data used for modeling, that is, proton-induced cross sections and/or thick target production rates used to determine neutron-induced cross sections. Here, we try to overcome the problem by using only calculated cross sections for modeling. In doing so, we rely on the latest version of the INCL (Li ege Intra Nuclear Cascade) code, which has recently been improved for higher energies, that is, in the range above 1 GeV, and for the emission of light complex particles (e.g., David et al. 2013;Mancusi et al. 2015;Pedoux and Cugnon 2011). We consider the current version of the code to be for the first time reliable enough for calculating sufficiently accurate production rates. The depth-and sizedependent energy spectra of primary and secondary particles used for modeling have been calculated using Monte Carlo methods, and they are the same as used by Ammon et al. (2009), that is, we use a solar modulation parameter M = 550 and a particle flux in the meteoroid orbits of 4.47 cm À2 s À1 .
Manganese-53: The proton-(solid line) and neutron-(dashed line) induced cross sections calculated by INCL for the production of 53 Mn from nat Fe are shown in Fig. 1 (upper panel). Also shown are the experimental cross sections for the proton-induced production given by Furukawa (1973), Gensho et al. (1972Gensho et al. ( , 1979, Kumabe et al. (1963), Lavrukina et al. (1964), Shore et al. (1961), Perron (1976), and Merchel et al. (2000). In addition, we show the data for the reaction 56 Fe(p, X) 53 Mn obtained in inverse kinematics experiments (Villagrasa-Canton et al. 2007). With an 56 Fe abundance of 92%, the inverse kinematics data for 56 Fe should be comparable to the other data obtained by irradiating Fe with a natural isotopic composition. It can be seen that the experimental data give a consistent excitation function, at least up to~60 MeV incident proton energy. In the energy range 400-1600 MeV, the data scatter significantly. Conversely, the INCL model produces a smoother excitation function. For energies below 40 MeV, the experimental cross sections are significantly higher than the model predictions and also the threshold energies differ. Whereas the model predicts a threshold energy of~14 MeV, the experimental data show the production of 53 Mn already at 12 MeV; the difference is important because reactions with lower reaction thresholds very often produce higher meteorite production rates. Above 40 MeV, there is reasonable agreement between measured and calculated cross sections. Also shown are the modeled results for the neutron-induced production of 53 Mn from nat Fe. Below~40 MeV, the neutron-induced cross sections are higher than the proton-induced data by up to a factor of two (the average is 40%); at the local minimum close to 40 MeV, they are lower than the proton-induced cross sections by up to 50%, and they are again~20% larger than the proton data for energies between 40 MeV and 65 MeV. For energies in the range 65 MeV-1 GeV, the proton-induced cross sections are on average 20% larger than the neutroninduced cross sections. Above 1 GeV, the cross sections for both projectile types are similar.
The proton-(solid line) and neutron-induced (dashed line) cross sections for the production of 53 Mn from nat Ni are shown in the middle panel of Fig. 1. Also shown are the experimental data from Merchel et al. (2000). While there is a reasonable agreement between experimental and calculated data, the experimental data are too scarce for establishing an excitation function consistent enough for model calculations. Again, the INCL calculations produce a smooth excitation function.
The new model predictions for cosmogenic 53 Mn production in iron meteoroids with pre-atmospheric radii of 5, 10, 15, 25, 30, 32, 40, 50, 60, 65, 85, 100, 120 cm, and for the outermost 200 cm of a 10 m object calculated using the new proton-and neutron-induced cross sections are shown in Fig. 2. The 53 Mn production is for almost all radii and all shielding depths dominated by neutrons. For example, already at the center of a 5 cm iron meteoroid more than 50% of 53 Mn is produced by secondary neutrons. This value increases to~80% at the center of a 25 cm iron meteoroid and reaches more than~90% at the center of a 50 cm iron meteoroid.
The specific 53 Mn activities for iron meteorites found in literature are between 23 AE 1 dpm kg À1 and 583 AE 25 dpm kg À1 (e.g., Nishiizumi et al. 1991), which is in good agreement with the range 33-567 dpm kg À1 predicted by the model. Furthermore, and experimental data. The experimental data for the protoninduced production are from Furukawa (1973), Gensho et al. (1972Gensho et al. ( , 1979, Kumabe et al. (1963), Lavrukina et al. (1964), Shore et al. (1961), Perron (1976), Merchel et al. (2000), and Villagrasa-Canton et al. (2007) and are not corrected for the new half-life value. Honda et al. (1961) measured 53 Mn in the four iron meteorites Grant, Williamstown, Odessa, and Canyon Diablo and found production rates in the range 92 AE 12-299 AE 11 dpm kg À1 , again in the range of the model predictions. However, there is a slight discrepancy for the iron meteorite Grant. While the model predicts 53 Mn production rates in the range 342-500 dpm kg À1 for an iron meteoroid with a radius of 40 cm, Honda et al. (1961) measured a production rate of 299 AE 11 dpm kg À1 . In a later measurement of the iron meteorite Grant, Imamura et al. (1980) measured specific 53 Mn activities between 304 AE 15 and 374 AE 16 dpm kg À1 , still somewhat low but slightly closer to the model predictions. The 53 Mn AMS value by Merchel (1998) of 435 AE 65 dpm kg À1 fits perfectly with the new model.
Iron-60: Our new purely theoretical data support the statement that the earlier experiment-based model significantly underestimated the 60 Fe production rates in iron meteorites as 21 of 28 of the obtained data are higher than the (former) modeled upper limit. Figure 1 (lower panel) depicts the INCL results for the proton-(black solid line) and neutron-induced (dashed line) production of 60 Fe from nat Ni. Also shown are the experimental data from Merchel et al. (2000) for the proton-induced production. Note that changing the halflife from 1.49 to 2.61 Ma will not change the cross sections, because the change in the measured activity cancels with the factor accounting for saturation. The two major reactions for the proton-induced production of 60 Fe are 62 Ni(p,3p) 60 Fe and 64 Ni(p,3p2n) 60 Fe or 64 Ni (p,ap) 60 Fe (i.e., an alpha particle instead of two protons and two neutrons in the exit channel). Most importantly and very surprisingly, the cross sections for the neutron-induced production of 60 Fe from Ni are up to five times higher than the cross sections for the proton-induced production. The most relevant reactions for the neutron-induced production of 60 Fe are 62 Ni(n,2pn) 60 Fe and 64 Ni(n,2p3n) 60 Fe or 64 Ni(n, an) 60 Fe. The results of the new model calculations are shown in Fig. 3 as depth-dependent production rates for iron meteorites with pre-atmospheric radii of 5, 10,15,25,30,32,40,50,60,65,85, 100, 120 cm, and for the outermost 200 cm of a 10 m object.

Mn Activities in Iron Meteorites
The 53 Mn data for the 46 samples from 41 meteorites are given in Table 1. For the large iron meteorite Twannberg, we studied six samples. The terrestrial ages used below are from the study by Smith et al. (2019). Owing to the long half-life of 53 Mn, decay corrections due to the terrestrial residence are insignificant, that is, the measured activity concentrations can just be converted to production rates. Note that even the longest terrestrial age of 285 ka for Puentel del Zacate reduces the 53 Mn concentration by less than 5%, which is below typical uncertainties for the AMS measurements (note that no 53 Mn has been measured for Puentel del Zacate). The 53 Mn activities based on the AMS measurements range from 4.3 to 658 dpm kg À1 . Considering only Twannberg samples, the 53 Mn activities range from 4.3 to 165 dpm kg À1 , that is, the spread is almost a factor of 40. Such a large spread for data from one meteorite confirms the large pre-atmospheric size of Twannberg (see also below). For the iron meteorite Grant, we measured a 53 Mn production rate of 441 AE 45 dpm kg À1 , which is significantly higher than the 53 Mn production rates given by Imamura et al. (1980), which range between 304 AE 15 and 374 AE 16 dpm kg À1 .
The new model predictions for 53 Mn are betweeñ 20% (surface of small meteoroids) and~30% (center of large meteoroids) lower than the previous model predictions ( Ammon et al. 2009). Averaged over all radii and all shielding depths, the difference is~30%. With the new 53 Mn data for Grant and the new model calculations, there is now a good agreement between experimental data and model predictions. For example, the model predicts 53 Mn-specific activities in the range 331-478 dpm kg À1 , in very good agreement with the measured specific activity of 441 AE 45 dpm kg À1 .
Nevertheless, there are still some discrepancies between model predictions and experimental data. For example, the model predicts maximum 53 Mn production rates of 547 dpm kg À1 in the center of a 25 cm meteoroid (Fig. 2). In contrast, the 53 Mn production rate for 8 of the 41 studied meteorites (Avoca, Bristol, Calico Rock, Chulafinne, Dalton, Fort Pierre, Greenbrier County, Zerhamra) is higher than the upper limit given by the model. For seven of the eight meteorites, however, there is agreement within the 1r standard deviation; the only exception is Chulafinne, for which the agreement is only within the 2r standard deviation.
We now discuss the shielding dependence of the 53 Mn data for Twannberg. The model predicts that for an iron meteoroid of radius 120 cm, the production rate varies only by a factor of~10 with shielding, compared to the measured factor of 40. Consequently, Twannberg must have been larger than 120 cm in radius. Moreover, 53 Mn activities as low as 4.26 dpm kg À1 as measured for one of the Twannberg samples are only possible at a depth of 120 cm in an object with a radius of 10 m (e.g., Smith et al. 2017). Larger depths in smaller objects (1.2 m <radius <10 m) are also possible; however, we have no model predictions for such objects.

Fe Activities in Iron Meteorites
The 60 Fe concentrations normalized to the Ni content of the 28 iron meteorites are compiled in Table 1. The calculated specific activities range from 0.38 dpm kg À1 (Ni) for Casas Grandes to 2.02 dpm kg À1 (Ni) for Gan Gan. The new 60 Fe data are in the range of values found in the literature for other iron meteorites (recalculated for the new half-life). For example, Casas Grandes and Lombard have low 60 Fe activities of 0.38 dpm kg À1 (Ni) and 0.46 dpm kg À1 (Ni), respectively. Such low concentrations have also been determined in the large meteorites Gebel Kamil (Ott et al. 2014) and Canyon Diablo (Berger et al. 2007).
For further discussion, we calculate production rates, that is, we correct the 60 Fe activities for radioactive decay during terrestrial residence. The terrestrial ages have been determined using 41 Ca/ 36 Cl atom ratios (Smith et al. 2019). The changes are minor; the maximum is 7% for Casas Grandes (terrestrial age of 247 AE 98 ka) and the average change is 1%. The calculated production rates are given in Table 1 as 60 Fe (0; dpm kg À1 [Ni]). Figure 4 depicts a histogram of all existing (to our knowledge) 60 Fe data for iron meteorites. In addition to our new data (Table 1), literature data from Knie et al. (1999b), Nishiizumi and Honda (2007), Berger et al. (2007), and Ott et al. (2014) are included. The data from Knie et al. (1999b), Berger et al. (2007, and Nishiizumi and Honda (2007) have been recalculated for the new 60 Fe half-life.
The new model calculations for 60 Fe production rates are significantly higher than the earlier model by Ammon et al. (2009). While the maximum is still at the center of a 25 cm meteoroid, it is now slightly above 2 dpm kg À1 (Ni), that is, more than a factor of 2 higher (see Fig. 3). Since for most of the studied meteorites neither the pre-atmospheric radius nor the preatmospheric shielding depth of the studied sample is known, a comparison of the experimental data with the model predictions is only possible for production rate averages and ranges. The improved model is in accord with measured activities. For example, the predicted 60 Fe production rates for all shielding depths in iron meteoroids with radii between 5 and 120 cm range c Cannot be given. 53 Mn and 60 Fe in meteorites between 0.2 and 2 dpm kg À1 (Ni), which covers all of the measured 60 Fe activities. Again, there are still some discrepancies between model predictions and experimental data. The modeled 60 Fe production rates for an iron meteorite with a radius of 40 cm, that is, very close to the preatmospheric radius of Grant (Ammon et al. 2008), vary between 1.2 dpm kg À1 (Ni) and 1.8 dpm kg À1 (Ni). This is significantly higher than the measured data for Grant from Berger et al. (2007), which are-after recalculating them using the new 60 Fe half-life-0.57 dpm kg À1 (Ni) and 0.69 dpm kg À1 (Ni). According to the model, such low 60 Fe activities are only reached in objects at least 100 cm in radius, which is unreasonable for Grant (e.g., Ammon et al. 2008).

Nuclide Correlations
To search for cosmogenic nuclide correlations that might help decipher cosmic ray exposure histories in iron meteorites, we plot in Fig. 5 the 53 Mn production rates as a function of 36 Cl production rates for the studied samples. Also shown are the results from the new model predictions for all shielding depths in meteorites with pre-atmospheric radii between 5 and 120 cm and the outermost 2 m of a 10 m radius object (dashed lines). The two solid black lines connect the results for the centers and the surfaces, respectively. The model calculations define an area of allowed 36 Cl-53 Mn production rates for meteorites that fall within the above-mentioned size range and that experienced singlestage exposure histories. The solid black symbols are experimental data; the 53 Mn data are from Table 1; and the 36 Cl production rates, which were determined in the same aliquots, are from Smith et al. (2017Smith et al. ( , 2019. In total, we have 36 Cl and 53 Mn data for 35 iron meteorites; of these, 25 plot within and 10 plot outside the allowed data field. The three meteorites North Chile, Turtle River, and Benedict plot below the allowed data field. The 36 Cl production rates for all three meteorites are in a range typical for iron meteorites, that is, they range between~6 dpm kg À1 for North Chile and~24.4 dpm kg À1 for Benedict. In contrast, the 53 Mn production rates are very low. For example, with a 36 Cl production rate of 23.9 AE 0.9 dpm kg À1 for Turtle River, the model predicts 53 Mn production rates in the range 360-530 dpm kg À1 , that is, far higher than the 56.4 dpm kg À1 measured by us. According to the model calculations, 53 Mn production rates as low as 56.4 dpm kg À1 are only possible in iron meteorites with pre-atmospheric radii larger thañ 100 cm. This is in contrast to the relatively low 4 He/ 21 Ne ratio of~220 and the activity ratios of the light cosmogenic radionuclides (e.g., Smith et al. 2019). The reason for the apparently too low 53 Mn production rates is not clear. It might be due to (1) a complex Fig. 5. Production rates of 53 Mn as a function of 36 Cl production rates for all shielding depths in iron meteorites with pre-atmospheric radii between 5 cm and 120 cm and the outermost 2 m of a 10 m object. The thin dotted lines connect the model calculations for an individual meteorite (from the surface toward the center). The thick black lines connect the results for all surfaces and centers, respectively. The model predictions define an area of allowed 36 Cl-53 Mn production rate combinations. Also shown are experimental data. Meteorites plotting outside the allowed field are labeled. Fig. 4. Histogram of 60 Fe activities (dpm kg À1 [Ni]) of the 28 studied meteorites. Also shown are data from the literature (Ott et al. 2014;Berger et al. 2007;Nishiizumi and Honda 2007;Knie et al. 1999b). The data from Knie et al. (1999b), Berger et al. (2007), and Nishiizumi and Honda (2007) were recalculated for the now accepted 60 Fe half-life of T 1/2 = 2.61 AE 0.04 Ma (Rugel et al. 2009). exposure history, (2) unrecognized problems during sample preparation and/or AMS measurements, and/or (3) an unusually high concentration of natural 55 Mn in the studied iron meteorite. For the last point, we discuss as an example the data for Turtle River. The studied sample had a mass of 102 mg, and during chemical extraction, we added~4 mg of Mn carrier. For calculating specific 53 Mn activities, we used the measured 53 Mn/ 55 Mn ratio, the amount of 55 Mn carrier added, and we assumed that the concentration of native 55 Mn in the sample is negligible. Consequently, calculating an 53 Mn activity~10% too low requires in addition to the 4 mg of 55 Mn carrier added~0.4 mg of native 55 Mn in the sample. This value is unreasonably high considering that the Mn/Fe ratio in iron meteorites is in the range 10 À7 (e.g., Sugiura and Hoshino 2003), which corresponds to~10 ng of native 55 Mn in the Turtle River sample, which is lower by more than four orders of magnitude. In addition, Herpers et al. (1969) measured native 55 Mn concentrations in the range <5.5-199 ppm, again too low to compromise 53 Mn-specific activity measurements by AMS. Next, we consider the second possibility, a complex exposure history. Doing so, we assume a recent break-up of an originally much larger Turtle River iron meteorite. The recovered mass of Turtle River is~23 kg, which corresponds to a minimum pre-atmospheric radius of~9 cm (after the hypothetical recent break-up). The 36 Cl production rate in such an object is~25 dpm kg À1 , that is, very close to the measured value. To reach this value, the meteorite must have been irradiated for at least 1 Ma. During the same time, 80 dpm kg À1 53 Mn would have been produced in such a meteorite, that is, much more than measured by us for Turtle River. Therefore, a complex exposure with a very recent break-up leaving some of the radionuclides under-saturated cannot explain the measured data, leaving unrecognized problems during sample preparation and/or AMS measurements the most likely explanation. Indeed, Turtle River belongs to the batch for which some samples had "virtual" chemical yields larger than 100% for Mn and needed reprocessing (note that the excess was not MnO 2 but AgCl, see above). For Benedict, we cannot completely exclude that the sample (or at least parts of it) got lost during chemical processing. However, there are no indications for any such problems for the North Chile sample.
There are also five meteorites (Casas Grandes, Sikhote-Alin, Calico Rock, Schwetz, Trenton) for which the 36 Cl and the 53 Mn data individually fall into the allowed range, but for which the combination of 36 Cl and 53 Mn data are outside the range predicted by the model. For the five meteorites in question, the 53 Mn production rates are higher than expected based on the model calculations. Note, however, that our data for Casas Grandes of 237 AE 29 agree well with the 210 AE 6 dpm kg À1 measured by Herpers et al. (1969). For Sikhote-Alin, we measured 358 AE 35 dpm kg À1 and Herpers et al. (1969) measured 335 AE 8 dpm kg À1 , again a good agreement. Finally, Herpers et al. (1969) measured 590 AE 14 dpm kg À1 compared to the 516 AE 52 dpm kg À1 measured by us. However, to be more precise, the 53 Mn data for Calico Rock, Schwetz, and Trenton with greater than 500 dpm kg À1 are, according to the model calculations, only possible for iron meteorites with pre-atmospheric radii in the range 20-40 cm. For meteorites in this size range, however, the 36 Cl production rates are~20 dpm kg À1 (e.g., Smith et al. 2019), that is, far higher than the values of less than~14 dpm kg À1 measured for two of the three meteorites in question. The found masses of 7.28 kg, 21.5 kg, and 505 kg for Calico Rock, Schwetz, and Trenton are in accord with preatmospheric radii in the range 20-40 cm. While we cannot fully exclude that the 36 Cl data are too low, we infer that the 53 Mn data are too high, which might be caused by unrecognized problems during sample preparation and/or AMS measurements, or it could be due to a complex exposure history. The latter might be as follows: The 53 Mn production rate decreases more slowly with depth than the 36 Cl production rate. For example, the 53 Mn production rate in a 10 m object decreases from the surface toward a shielding depth of~2 m by about three orders of magnitude. In contrast, the 36 Cl production rates decrease by about four orders of magnitude in the same range of shielding depths. Consequently, there are regions in a large iron meteoroid with measurable amounts of 53 Mn but without any 36 Cl. If, after further break-up, those regions get close to the (new) pre-atmospheric surface, they might have inherited some excess 53 Mn from the first irradiation stage leading to high 53 Mn/ 36 Cl activity ratios. For the break-up to have a measurable effect on the 53 Mn/ 36 Cl ratio, a requirement would be that it occurred during the last few half-lives of 53 Mn, that is, within the last 10 Ma or so. While we consider it as unlikely that such a recent break-up has remained unnoticed so far, considering the variety of measured cosmogenic nuclides (e.g., Smith et al. 2019), such a scenario is not impossible. However, such a scenario cannot explain the data for the five meteorites in question. We discuss here as an example the data for the meteorite Calico Rock. Assuming after the hypothetical recent break-up, a relatively large object with a 36 Cl production rate in the range 4 dpm kg À1 (close to the value measured by us), the 53 Mn production rate would be~200 dpm kg À1 , that is, far lower than the 500 dpm kg À1 measured by us. Consequently, in such a scenario, 300 dpm kg À1 of 53 Mn must have been inherited from the earlier irradiation stage. From Fig. 5, we can conclude that at least in the range of studied pre-atmospheric radii, there is no region in an iron meteorite in which 300 dpm kg À1 53 Mn is produced without any collateral production of 36 Cl, making such a scenario impossible. A special case is the data for Greenbrier County; the 36 Cl data of 29 dpm kg À1 are unexpectedly high, whereas according to the model calculations (Smith et al. 2017), the upper limit for the 36 Cl production rate in meteoritic metal is $ 25 dpm kg À1 . Currently, however, we have no reason to consider the 36 Cl data for Greenbrier County as unreliable. We might speculate that neutron-capture reactions on natural chlorine might be the reason for the too high 36 Cl concentrations. Such a mineral could be lawrencite, which occurs in some iron meteorites (e.g., Goldschmidt 1954;Honda et al. 1961). For a discussion, see also Smith et al. (2019).
The new model predicts 60 Fe/ 53 Mn production rate ratios in the relatively narrow range 0.0021-0.0031 (dpm kg À1 [Ni]/dpm kg À1 ); the average for all radii and all shielding depths is (2.8 AE 0.1) 9 10 À3 . Considering now that the model calculations are for metal consisting of 90% Fe and 10% Ni, the production rate ratio 60 Fe (kgNi) À1 to 53 Mn (kgFe) À1 changes to (2.5 AE 0.1) 910 À3 . This is in excellent agreement with the activity ratio of 60 Fe (kgNi) À1 to 53 Mn (kgFe) À1 of (2.68 AE 0.35)910 À3 deduced by Fimiani et al. (2016) for meteorite data, which has been used by these authors to disentangle measured 60 Fe data for lunar Apollo 12, 15, and 16 samples into cosmogenic and interstellar components. Both activity ratios are slightly higher than the ratios expected for extraterrestrial dust ( 60 Fe/ 53 Mñ 10 À4 dpm kg À1 [Ni]/dpm kg À1 ; Knie et al. 1999a). Figure 6 depicts the 60 Fe production rates (dpm kg À1 [Ni]) as a function of the 53 Mn production rates (dpm kg À1 ). The experimental data are shown by the solid black symbols. Also shown is the linear correlation predicted by the model calculations (gray band). Twenty of the available 23 experimental data follow the predicted linear trend; three irons (Benedict, Gan Gan, Turtle River) plot well above the correlation line. The meteorites Benedict and Turtle River have already been discussed before for their low 53 Mn data (see Fig. 5). Gan Gan has a very high 60 Fe production rate of 2.02 AE 0.37 dpm kg À1 (Ni), which indicates a rather small pre-atmospheric radius. Values that high are only possible close to the center of an iron meteorite with a pre-atmospheric radius in the range 20-30 cm (see Fig. 3). In such objects, however, the 53 Mn activities are in the range 550 dpm kg À1 , that is, far higher than the measured value of 323 AE 14 dpm kg À1 . The calculated pre-atmospheric radius in the range 20-30 cm is in reasonable agreement with the recovered mass of 83 kg, which corresponds to a postatmospheric radius of~13 cm. From these arguments, we speculate that the measured 53 Mn activity for Gan Gan is slightly too low.

CONCLUSIONS
We measured 53 Mn and 60 Fe activities in 41 and 28 iron meteorites, respectively, including six samples from the large iron meteorite Twannberg for 53 Mn. Measurements of 60 Fe and 53 Mn by accelerator mass spectrometry are both experimentally challenging. Consequently, prior to this study, the database for cosmogenic 53 Mn and for 60 Fe was limited to a few measurements only. In addition, we performed new model calculations for the production of 60 Fe and 53 Mn in iron meteorites. The model is based on the same particle spectra as a function of size and depth as used by Ammon et al. (2009), but our model uses only theoretical cross sections for proton-and neutron-induced reactions obtained from the INCL nuclear model code.
The new model predictions for 60 Fe are significantly higher than earlier ones and, with one exception (Grant; Berger et al. 2007), are in generally good agreement with the older and newer measurements for iron meteorites. There is still a discrepancy between measured and modeled 60 Fe data for the iron meteorite Fig. 6. Iron-60 production rates (dpm kg À1 [Ni]) as a function of 53 Mn production rates (dpm kg À1 ). The experimental data are shown by solid black symbols. The gray shaded area indicates the linear correlation predicted by the new model calculations. Samples that deviate significantly from the predicted linear correlation are labeled.
Grant, which could well be due to inconsistent Grant data in a previous publication (Berger et al. 2007).
For 53 Mn, the new model predictions are on average 30% lower than the earlier model and are therefore in better agreement with experimental data. The new 53 Mn data for Grant are higher than the earlier data from Imamura et al. (1980) but are consistent with early AMS data (Merchel 1998) and are now in good agreement with the range predicted by the model for iron meteorites with a 40 cm pre-atmospheric radius. We found large variations among the 53 Mn activity concentrations of the six studied Twannberg samples, clearly confirming its exceptionally large preatmospheric size (Smith et al. 2017).
There are, however, still some discrepancies. Some of the measured 53 Mn/ 36 Cl and 60 Fe/ 53 Mn ratios do not fit into the allowed range in 53 Mn/ 36 Cl and do not follow the linear correlation between 60 Fe and 53 Mn predicted by our model, respectively. The discrepant data cannot be explained by complex exposure histories but might indicate some unrecognized problems either during sample preparation and/or during AMS measurements. The grand average of all measured data, however, agrees well with the average production rate ratio of 53 Mn/ 36 Cl~31 and 60 Fe/ 53 Mn~3.6 9 10 À3 calculated with our model by considering all shielding depths in iron meteorites with pre-atmospheric radii between 5 and 120 cm.
In iron meteorites, 10 Be and 26 Al are often not very reliable due to possible contributions from traces of sulfur and/or phosphorous. For 26 Al, such contributions can be in the range of tens of percent, even in samples that were visually expected to be devoid of any inclusions (see Smith et al. 2019). In addition, 41 Ca is often very difficult to measure and is strongly affected by decay during terrestrial residence. Furthermore, 36 Cl is difficult to study because of the need for a dedicated chemistry laboratory for the chemical extraction and high risk of cross contamination, especially in the AMS ion source. Consequently, it might well be that 53 Mn and 60 Fe are more reliable cosmogenic radionuclides for studying cosmic ray exposure histories of iron meteorites. However, for 53 Mn to become reliable, there is a need to establish a consistent and well-documented standard, which ideally should not be from a meteorite.
The quality of the model calculations has improved considerably by using calculated data, instead of using experimental cross sections that are based on a few measurements only and that are sometimes inconsistent. While this clearly demonstrates the good quality currently achieved by nuclear model codes to calculate nuclear cross sections (at least for some target-product combinations), it also indicates that there is still a need for more and more reliable cross section measurements.