As noted in the introduction to Section 3, in the early- to mid-1980s I not only took my first steps away from solely studying lunar samples to becoming a meteoriticist, but also developed, more or less in parallel, an interest in noble gases (and radionuclides) in terrestrial samples. For a physicist with a meagre (to say the least) background in geology, this naturally required extensive collaboration with trained Earth scientists. The first of my two main terrestrial research areas were cosmogenic noble gases – largely in combination with cosmogenic radionuclides – as a tool in geomorphology. The second was the study of noble gases in water samples from lakes, groundwaters, and fluid inclusions in speleothems as a tool in limnology and climatology. These activities will be the topics of Sections 4 and 5. But first, in Section 4.1, I review the first two studies on noble gases in terrestrial samples in which I was involved.

4.1 My First Steps in Geology: Fissiogenic and Nucleogenic Noble Gases

For his doctoral thesis published in 1988, Jost Eikenberg proposed to Peter Signer a study of noble gases in uranium-rich minerals. The two long lived uranium isotopes 235U and 238U decay mostly along the well known α and β- decay chains to 207Pb and 206Pb, respectively, but they can also split into two roughly equal sized nuclides, including several xenon isotopes which are of interest here. Some of these fission reactions are triggered by fast or slow neutrons, but most importantly 238U can also spontaneously fission. The half-life for spontaneous fission (sf) is much longer than the cumulative half-life of 4.47 Ga along the decay chain to 206Pb. Nonetheless the spontaneous fission process can also be used to date U-rich minerals. For example, in zircon and monazite, Xe from spontaneous fission is often retained better than radiogenic 206,207Pb. However, one problem has been that the fission half-life of 238U has been controversial. The relevant quantity is λsf × 136Ysf, i.e. the decay constant λ for spontaneous fission of 238U times the fractional yield per fission event of 136Xe, the main fission Xe isotope. Values for this product determined with U oxides such as pitchblende were typically about 20 % lower than those measured with accessory U minerals such as zircon or monazite. Jost Eikenberg analysed a large series of carefully characterised uraninite and pitchblende samples with known, undisturbed geological histories and concordant U-Pb ages. In selecting the samples, Jost profited from the expertise of his thesis co-supervisor Victor Köppel. Jost’s average λsf × 136Ysf of (5.7 ± 0.4) × 10-18/a was more robust than the U oxide-based values obtained by previous researchers, but essentially confirmed them (Eikenberg et al., 1993). Hence, the discrepancy with the values based on U-rich accessory minerals remained. Therefore, in her doctoral thesis, Riccarda Ragettli did an equally careful study by combining U-Pb and U-Xe analyses of zircons and monazites, again with Victor as co-supervisor (Ragettli et al., 1994; Fig. 4.1). She obtained λsf × 136Ysf = (6.83 ± 0.18) × 10-18/a, again essentially confirming previous values obtained on accessory U minerals, but with improved precision. Ragettli et al. (1994) considered it very unlikely that diverse U-bearing oxides would lose more or less constant fractions of their fission Xe by diffusion, but noted that these minerals can readily regenerate their crystal structure after radiation induced lattice damage. We suggested that either this regeneration or changes in oxidation state could lead to a systematic loss of Xe. In a later paper, Jost Eikenberg and I investigated another possibility, namely that zircons and monazites might contain sizeable fractions of Xe from the spontaneous fission of 232Th (Wieler and Eikenberg, 1999). However, with a series of very Th-rich monazites from China, we were able to establish an upper limit on λsf (232Th) that ruled out this possibility.

Jost Eikenberg also measured He, Ne, Ar, and Kr in many of the samples studied in his 1993 paper. One U-rich fluorite yielded essentially pure 22Ne from the reaction 19F(α,n)22Ne. To the best of our knowledge, the ratio 20Ne/22Ne = 0.042 is the lowest value ever measured in a terrestrial sample. While this is not an Earth-shaking finding, it was quite remarkable for a meteorite noble gas person like me to learn that almost pure 22Ne not only occurs in presolar grains in meteorites – remember the exotic Ne-E component in meteorites mentioned in Section 3.2.1 – but is also produced in natural samples on our own planet. Jost found in his samples also several noble gas components that were formed by so called Wetherill reactions, e.g., 18O(α,n)21Ne. He estimated that about 2.4 % of the 21Ne in the Earth’s atmosphere could be produced by such reactions. This work inspired Ingo Leya and me to study the production of Ne by alpha particles from U and Th decay in the terrestrial crust and upper mantle (Leya and Wieler, 1999). These calculations relied on the same tools Ingo used to calculate cosmogenic nuclide production in meteorites and on planetary surfaces, as discussed above. Calculated production rates of Ne isotopes as functions of the concentrations of the major taget elements oxygen and fluorine and the number of alpha particles agree well with experimental data.

4.2 Cosmogenic Noble Gases and Radionuclides in Geomorphology

About a decade after Davis and Schaeffer (1955) had suggested that cosmogenic nuclides produced in rock samples near the Earth’s surface through interactions with galactic cosmic rays could be used to study geologic problems (Section 3.1.1), Devendra Lal and Bernard Peters (Lal and Peters, 1967) provided a theoretical framework for such studies. However, analyses of in situ cosmogenic nuclides in terrestrial samples only became routine in the 1980s (e.g. Klein et al., 1986; Nishiizumi et al., 1986; Kurz, 1986; see Gosse and Phillips, 2001; Niedermann, 2002; Granger et al., 2013), although already twenty years earlier cosmogenic nuclides had become a widely used tool for studying the exposure history of meteorites and lunar samples, as mentioned in Section 3.1. The reason for this delay is the strong shielding by the Earth’s atmosphere, which causes production rates at the Earth’s surface to be several orders of magnitude lower than those on the lunar surface or in meteorites. The crucial break through in the early-1980s was the development of accelerator mass spectrometry (AMS), which had already revolutionised 14C dating. AMS also allowed routine analyses of in situ cosmogenic radionuclides such as 10Be, 26Al, and 36Cl in small samples (Section 3.1.2 and Fig. 3.2). Within a few years its use transformed quantitative geomorphology and quaternary geology (Gosse and Phillips, 2001; Dunai, 2010; Granger et al., 2013). Landforms can now be dated and erosion rates measured over timescales relevant to soil-forming processes. Current radionuclide detection limits are low enough to allow the determination of surface exposure ages in some cases as low as <1000 years.

The first report of cosmogenic noble gases in terrestrial rocks was by Srinivasan (1976), who detected excesses of the lightest (and rarest) Xe isotopes in barites from South Africa and Australia. However, this remained the only report of cosmogenic heavy noble gases in terrestrial samples until Dunai et al. (2022) detected cosmogenic Kr in zircons from a suite of surface rocks. Much more convenient to analyse in terrestrial samples are the cosmogenic contributions to the two lighest noble gases He and Ne. This was pioneered by Mark Kurz at the Woods Hole Oceanographic Institution and by Harmon Craig, Kurt Marti, and co-workers at the University of California in San Diego, who reported cosmogenic 3He and 21Ne in high altitude samples from Hawaii (Kurz, 1986; Marti and Craig, 1987). Since then, analytical developments in noble gas mass spectrometry have enabled routine analysis of cosmogenic noble gases in a variety of rocks on Earth (Niedermann, 2002; Blard, 2021). However, the detection limits for cosmogenic noble gases in terrestrial rocks have never reached the very low levels that are now routine for several radionuclides. Therefore, we decided to focus on the study of very old landscapes where the stable noble gases could have advantages over radionuclides or at least provide complementary information, as discussed in the following.

4.2.1 Antarctica

In 1991 we were contacted by Christian Schlüchter, then at ETH before he moved to the University of Bern two years later. Christian is a Quaternary Geologist who works almost everywhere on the planet and has a reputation as a globetrotter. Among his favourite terrains are the Dry Valleys in Victoria Land in East Antarctica. These ice-free valleys are among the oldest landscapes on Earth, with very little precipitation and very little erosion. For a geomorphologist like Christian, the prospect that all of a sudden age and erosion rates of such landscapes could be quantified with nuclides produced by cosmic rays must have been no less exciting than was the realisation by mid-20th century geologists that they could absolutely and reliably date the formation of rocks with radiogenic isotopes. Christian thus sought cooperation with the AMS group at the ETH Physics Department as well as with us at the Earth Science Department. As it just happened, we were looking for a dissertation project for Laura Bruno. So, because noble gases are ideal for investigating very old landscapes, as they do not decay like radionuclides, Laura set out to study the glacial history of Antarctica, or more exactly the history of Antarctica as reflected in the Dry Valleys. Laura’s doctoral thesis was the first chapter of a very fruitful collaboration with Christian that led to several jointly supervised dissertations and many other projects. Laura’s work also marked the beginning of our collaboration with the AMS group at ETH on terrestrial problems, extending our previous joint activities in cosmochemistry (Section 3.1). Our main partners in the AMS team were Susan Ivy-Ochs, Peter Kubik and later Lukas Wacker. For our cosmogenic nuclide studies in Antarctica, Carlo Baroni of the University of Pisa was also an important colleague. I already had met Carlo during my meteorite search at Frontier Mountain in the 1990/91 season at the Italian Antarctic Station at Terra Nova Bay (see Box 3.2).

The development and subsequent history of the Antarctic ice masses are major issues in palaeoclimatology. The most important question was the extent to which the Antarctic ice masses had been affected by the Pliocene climate oscillation. This was a global warming interval from 4.8 to 3.2 Ma ago with an increase in sea surface temperature of about 3 ºC. Some workers argued that such warm conditions caused the ice sheet to collapse. Others proposed that the East Antarctic Ice sheet has remained stable since its formation perhaps 15 or 20 Ma ago and that the Antarctic climate had been cold and hyperarid ever since (Denton et al., 1993). For Christian Schlüchter and colleagues, the Dry Valleys were a key area (Fig. 4.2). Had there been major erosional events indicating the existence of running water in summer since the Dry Valley landscape formed? Cosmogenic nuclides – here primarily the stable noble gases – were the ideal (then new) tool to tackle this question.

Because the production rate of cosmogenic nuclides near a rock surface decreases with increasing depth and because even in the hyperarid Antarctic climate rocks slowly erode, concentrations of cosmogenic nuclides basically only provide minimum values for the time a sample was exposed to cosmic rays at the Earth’s surface. The nominal age would equal the true exposure age only if no erosion had occurred (Fig. 4.3a). The analysis of two or more nuclides with different half-lives in the same sample allows, in principle, one to determine both the erosion rate and the exposure age (Fig. 4.3b), but for the moment we are concerned only with stable noble gases. More about cosmogenic nuclide systematics can be found in Section 4.3.1 about the CRONUS programmes.

Laura Bruno analysed 3He and 21Ne in pyroxene and quartz separates from boulders at two locations (Mt. Fleming and Table Mountain) in glacial deposits of the so called Sirius group. Pyroxenes were important because they retain cosmogenic helium much better than quartz and hence allow estimates of noble gas losses by diffusion. Christian Schlüchter had taken the samples with the eye of an experienced field geologist, allowing him to recognise boulders least disturbed since deposition. Laura’s oldest minimum values of ~5 and 5.5 Ma for the two sites were among the oldest exposure ages of terrestrial samples ever measured until then (and also higher than the maximum values of about 4 Ma to be deduced from radioactive 10Be). Laura’s work (Bruno et al., 1997) was thus the first unequivocal evidence based on cosmogenic nuclides that the East Antarctic Ice sheet had remained stable for at least the past 5 Ma, but likely even longer, presumably at least for 6-6.5 Ma. Bruno and co-workers also concluded that the maximum possible uplift rate of the Transantarctic Mountains was only ~170 m/Ma, much slower than necessary to explain the microfossils found in Sirius group sediments if these sediments were of much younger age, as argued by the proponents of an unstable ice sheet.

In his doctoral thesis, Jörg Schäfer extended the work of Laura Bruno to other locations in the Dry Valleys. His noble gas data, supplemented by 10Be data provided by Susan Ivy-Ochs, gave minimum ages of up to 10 Ma (Schäfer et al., 1999). As in other studies discussed here, samples were taken from erratic boulders (examples shown in Fig. 4.4) and/or bedrock surfaces. Jörg’s data set confirmed Laura Bruno’s conclusion as to the Pre-Pliocene age of the Sirius group sediments, which he estimated to be older than 20 million years. Maximum long term erosion rates were below 15 cm/Ma even at altitudes below 1000 m above sea level. Jörg concluded that this implies a permanently cold and hyperarid climate and thus a decoupling of the Antarctic climate from that of lower southern latitudes.

Jörg Schäfer provided additional evidence for a long term continuously arid climate in the Dry Valleys by measuring the age of a spectacular glacial remnant. This ice body in Beacon Valley is covered by a till layer with an 40Ar/39Ar age of ~8 Ma. As this age likely represents a lower limit on the age of the underlying ice, this implies exceedingly low ice sublimation rates, probably due to saturation of the till with moisture. This would also imply a constantly cold climate in Beacon Valley for at least 8 million years. Jörg measured cosmogenic 3He and 21Ne in erratic boulders from the till surface and from within the ice, respectively (Fig. 4.5). He derived a conservative lower limit of 2.3 Ma for the cosmic ray exposure age of the surface boulders, while the difference between the noble gas concentrations of the surface samples and one from the interior allowed him to derive an ice sublimation rate of no more than a few metres per million years (Schäfer et al., 2000). This and the high age (likely considerably older than the 2.3 Ma minimum age) support the conclusion from the Ar-Ar data that the ice in Beacon Valley was deposited many millions of years ago and has never been thinner than it is today.

The work of three further doctoral students confirmed the long term stability of a cold and hyperarid climate in East Antarctica. Peter Oberholzer and Stefan Strasky analysed noble gases in our laboratory. Luigia di Nicola from the Universities of Siena and Bern and later at the SUERC in East Kilbride, Scotland, measured the radionuclides 26Al and 10Be at the ETH AMS facility. These projects were carried out together with Carlo Baroni and Christian Schlüchter. The study area a few hundred km north of the Dry Valleys in North Victoria Land was chosen to address the objection of Van der Wateren et al. (1999) that landscape stability inferred from the Dry Valleys should not be generalised to the entire Transantarctic Mountains. The noble gas cosmic ray exposure ages showed that the long term stability and the very limited dynamics of the East Antarctic Ice Shield is also evidenced in northern Victoria Land (Oberholzer et al., 2003; 2008). Long term maximum erosion rates there are similarly low as those in the Dry Valleys, testifying to a constantly cold and hyperarid climate also further north. Luigia and Stefan also found some very old exposure, again confirming long term hyperarid climates, but their data also showed that many samples had undergone a complex exposure history, reflecting, e.g., ice level fluctuations (Strasky et al., 2009a; Di Nicola et al., 2012).

I should add here that I myself was not involved in the details of the project definitions of all these studies or the geological interpretation of the data. My contributions focused on the noble gas analyses and the more technical aspects of data interpretation.

4.2.2 Tibet and from there to the Indian Ocean

The Himalayas and adjacent high altitude regions in central Asia are sometimes referred to as Earth’s third pole, and understanding the role of this vast region in shaping the global climate system is of paramount importance. The history of glaciations is of particular importance, since glacial cycles reflect broader climate changes. Himalayan glaciations may affect the global radiation budget and influence the Asian monsoon. Christian Schlüchter recognised the potential of cosmogenic nuclides for studying Tibet’s glacial history and he invited three of the doctoral students who had worked with him on projects in Antarctica to join him for field work in Tibet. The acknowledgments in the dissertations of Jörg Schäfer, Peter Oberholzer, and Stefan Strasky not only testify to how much they benefitted from Christian’s experience as a field geologist but also to how deeply impressed they were by his empathy for foreign cultures. They studied noble gases and 10Be in glacial moraine samples from different key areas in Tibet (Fig. 4.4). Beryllium-10 was again analysed in cooperation with Susan Ivy-Ochs and Peter Kubik from the AMS group at ETH. In short, these studies showed that glaciations in Tibet had had rather limited influence on global climate over the past ~170,000 years (Schäfer et al., 2002; Schaefer et al., 2008; Oberholzer, 2004; Strasky et al., 2009b). A putative ice dome that once covered the entire plateau, as proposed by Kuhle (1998), was previously viewed with great scepticism and could now be rejected based on the cosmogenic nuclide data. The data also showed that increased moisture supply by an intensified summer monsoon had only limited influence on the Tibetan glaciations. Rather, these were influenced by the North Atlantic climate system (Oberholzer, 2004; Schaefer et al., 2008). The latter paper by Joerg Schaefer (note the slight name change to avoid the pain with German Umlaut characters in citation records!) was published when he had already moved to the Lamont-Doherty Earth Observatory near New York, where he still works today.

Interactions between tectonics and climate make the Himalayas the world’s largest sediment source to the oceans. The Tsangpo-Brahmaputra river system drains the Himalayan range and the Tibetan plateau, and this large catchment is ideal for studying how denudation processes are reflected in the sediment load. Maarten Lupker therefore decided to study the interplay between tectonic uplift, surface denudation processes, and sediment transport along the Tsango-Brahmaputra catchment (Lupker et al., 2017). Maarten had been a Master’s student at our institute under the supervision of Bernard Bourdon and Sarah Aciego. For his Master’s thesis on the isotopic composition of Sr, Nd, and Hf in dust and ice from an ice core in Greenland, he received the inaugural Prix de Quervain of the Swiss Committee on Polar and High Altitude Research (in honour of Alfred de Quervain, the Swiss Geophysicist who crossed Greenland in 1912, the second traverse ever after Fridtjof Nansen’s in 1888). I had been a member of the selection committee for the award and gladly agreed when Maarten expressed his wish to join our group as a postdoc after completing his PhD at CRPG in Nancy. He later moved to the Institute of Geology in our Department, but continued to work with us thanks to his interest in cosmogenic nuclides. These are an excellent tool for determining catchment-wide denudation rates over a wide range of spatial scales. Basically, the nuclide concentration in a sample of loose sediments from a river reflects how fast, on average, the river catchment is eroding upstream to the position of the sample (Brown et al., 1995). Lupker et al. (2017) showed that denudation rates (measured with 10Be in Zürich and at CEREGE in Aix-en-Provence) vary by two orders of magnitude between individual sub-catchments along the Tsangpo-Brahmaputra system. The sediment flux at the outlet in Bangladesh corresponds to a denudation rate on the order of one millimetre per year. Most importantly, Maarten Lupkers’ study is an excellent example of the complications that can arise when attempting to derive catchment-wide denudation rates “from a bag of sand”.

4.2.3 Northern Chile

The Western Central Andes constitute another area where the interplay of erosion, climate, and tectonics can be studied in an exemplary way. Fritz Schlunegger (now at the University of Bern) and our joint doctoral student Florian Kober combined analyses of 10Be, 26Al, and 21Ne to determine erosion rates and cosmic ray exposure ages across a west-east transect in northern Chile. The transect spanned an elevation range from close to sea level to >4000 m above sea level and a climate range from hyperarid to semi-arid (Kober et al., 2007). I had never joined my colleagues in Tibet, but this time I did not want to miss the opportunity to get at least a vague idea of how trained geologists like Fritz and Florian can assess a terrain to extract the most information from a minimum number of samples. To me, this remains both an art and a science, but there is definitely more to it than just walking on an Antarctic ice field or a hot desert surface in search of (dark) stones that might turn out to be meteorites. It was the first time I had been back to South America since my stay in Guayaquil and the subsequent long trip across the entire continent in 1974–75. I had been informed that in some parts of the Atacama desert not a single rainfall had been documented since the arrival of the Spanish conquistadores. Yet, during my two weeks in the port city of Arica we enjoyed half an hour of pleasant evening rain. On an unforgettable day trip we drove up to the Bolivian border at an altitude of over 4000 m, crossing the hot Atacama desert before finally passing Vicuna herds grazing in the freshly fallen snow. On several other trips along the Panamericana, we had seen a lone hiker heading north, and one day we met him at one of the small makeshift restaurants along the road. Yoshi had a good position in Japan that allowed him to take a break every few years to hike the world. He had crossed Asia in two legs from Western China to Istanbul, and once he participated in a deca-ultratriathlon in Mexico (38 km swim, 1800 km bike, and 421.95 km run, the latter on a 400 m track!). The present trip would eventually take him from Buenos Aires to Cuzco. He relied on the smallest backpack one can possibly imagine, with two half-litre bottles of water, a thin blanket for the night, and some stuff to periodically repair his sneakers. He often ate what others dumped out of their cars and when in need of water, he would simply signal to the truck drivers, who often politely stopped. Apparently this is one way to survive in one of the driest places on Earth. But when we met him, he was suffering quite a bit, perhaps from an upset stomach. So he accepted our offer of a ride to Arica, albeit with a bit of guilty conscience. After having recovered a few days later, he insisted that we take him back to where we had picked him up some 50 km south of Arica. A few months later I received a letter from him posted in Cuzco. No doubt he finished the journey without leaving a gap. I wonder if he has now also fulfilled his lifelong dream of hiking from Alaska to Tierra del Fuego.

As expected, Florian Kober showed that erosion rates and exposure ages along the transect correlate with altitude and present day rainfall. In the hyperarid Coastal Cordillera, erosion rates are extremely low, in the range of 10-100 cm/Ma, resulting in nominal minimum exposure ages between 1-6 Ma. In the semi-arid Western Cordillera, erosion rates are much higher, up to 46 m/Ma, corresponding to nominal minimum exposure ages of 20,000–100,000 years. The erosion rates in the hyperarid zones are as extremely low as those in Antarctica and therefore can also be studied particularly well by stable cosmogenic noble gas isotopes. Dunai et al. (2005) reported exposure ages of up to 37 Ma measured by cosmogenic 21Ne in quartz clasts from an area south of Arica. These authors concluded that their data require predominantly hyperarid conditions for several tens of million years, consistent with the hypothesis that the onset of aridity in the Atacama Desert may be the cause – rather than the consequence – of the uplift of the high Andes. The combination of cosmogenic noble gases and radionuclides allowed Kober et al. (2007) to study erosion in more detail than would have been possible with noble gases alone. When only one nuclide is analysed, one usually assumes nuclide saturation in a constant erosion regime, but this is not the rule in the study area. The data require more complicated exposure scenarios, such as episodic erosion by spalling of larger slabs. Kober et al. (2009) further showed for a river system in northern Chile that catchment-wide denudation rates determined by only one nuclide would not have been correct for most sub-catchments because sediments sometimes are buried for long periods of time. This study again demonstrated the usefulness of combined cosmogenic nuclide analyses.

4.3 Method Development on Terrestrial Cosmogenic Nuclides

As much as I enjoy working with geologists on field studies, I have never come close to being an expert in the field. Since the beginning of my cooperation on terrestrial cosmogenic nuclides, I have therefore instead tried to contribute to method developments in terrestrial cosmogenic nuclide research with my students, postdocs, and other colleagues. On the one hand, we have attempted to improve values for nuclide production rates and scaling factors, including exploring the use of less common minerals. On the other hand, we developed one of the first successful extraction lines for in situ produced cosmogenic 14C, a very important cosmogenic nuclide, albeit one rather difficult to analyse. Much of this work was carried out in the framework of the CRONUS collaboration. So, before we go into the details of our work, let me introduce the CRONUS programmes.

4.3.1 CRONUS

We have seen in Section 3.1 that the production rate of a cosmogenic nuclide in a given meteorite is a fundamental quantity whose uncertainty often dominates the total error of a stated exposure age. Production rates are an even greater problem in terrestrial cosmogenic nuclide research. For a meteorite we can assume that the production rate at a given pre-atmospheric depth depends essentially only on the target element abundances (neglecting, for example, the small radial gradient of the galactic cosmic ray flux in the inner solar system). The main uncertainties are therefore the pre-atmospheric size of the meteorite and the sample depth. In contrast, the main factors that govern production rates in terrestrial samples – apart from major target element abundances, i.e. mineral type – are the geographic location of a sample and its exhumation/erosion history. Location is important for two reasons (e.g., Gosse and Phillips, 2001). First, the Earth’s magnetic field allows only primary cosmic ray protons above a certain energy to reach the upper atmosphere, where they will induce nuclear reactions that produce the secondary cosmic ray neutrons that eventually reach the surface. This threshold energy depends on the geomagnetic latitude. Fortunately, the average position of the magnetic pole coincides with the geographic pole over timescales of about 10,000 years or more. Therefore, in most cases, the latitude correction can simply be based on the geographic latitude of a sample. Second, and even more importantly, the Earth’s atmosphere efficiently shields the surface of our planet from cosmic rays. To illustrate this, the atmospheric load of ~1 kg/cm2 at sea level is equivalent to about 4-5 m of concrete. As an example, nuclide production rates at 3000 m altitude and latitudes higher than about 60º are almost 20 times larger than values near sea level in regions close to the equator. A reliable “scaling” of production rates for altitude and latitude is therefore required. Devendra Lal (Lal, 1991) provided an early method based on cosmic ray flux data obtained by exposing photographic emulsions at different altitudes in the atmosphere. The “Lal scaling” became very popular and widely used for many years (Fig. 4.6a). However, scaling methods needed further testing and improvements. For example, there was an urgent need for a much larger database of directly measured production rates at different locations and for different mineral types. Production anomalies in certain regions of the world needed further investigation, the most striking being an approximately 20 % lower than expected value in Antarctica (Stone, 2000). In addition, possible fluctuations of production rates over time needed to be considered.

A large community effort was therefore launched in 2004–2005 under the name CRONUS (Cosmic Ray Produced Nuclide Systematics on Earth). The North American branch (CRONUS-Earth) was coordinated by Fred Phillips and colleagues (Phillips et al., 2016a) and the European branch (CRONUS-EU) by Tibor Dunai, with regular advice (more than once over a beer) provided by Finlay Stuart (Stuart and Dunai, 2009). Tibor set up CRONUS-EU while working in Amsterdam, but soon afterwards moved to Edinburgh and later to the University of Cologne. I had known him since his time as doctoral student at ETH. We were very impressed when – within a few days after Peter Signer had offered him the position – without previous experience in noble gas geochemistry Tibor presented a proposal to study noble gas signatures in the sub-continental mantle, a region that hitherto had received much less attention than the mid-ocean ridge and ocean island basalt sources (Dunai and Baur, 1995).

The main goals of CRONUS included (i) testing scaling procedures by experiments with artificial targets, (ii) geological calibration of production rates using samples with independently known exposure ages, (iii) development of improved numerical nuclide production models, and (iv) improvement of sample processing and analysis programmes, including inter-laboratory standard comparisons.

Both the AMS group and the noble gas group at ETH participated in CRONUS-EU, which was funded by the European Union as a “Marie Curie Research Training Network”. As the name implies, the funding agency placed great emphasis on training young scientists in multinational cooperations. All students and postdocs funded by CRONUS-EU thus worked in institutions abroad. The programme started with a meeting in Amsterdam, where all the group leaders met all candidates for doctoral or postdoctoral positions. The group leaders presented their projects, the candidates their previous research, and then both sides expressed their preferences. My favourite postdoc candidate was Pieter Vermeesch, who had just completed his PhD on fission track dating at Stanford. Pieter was also the favourite of several of my colleagues. So imagine how happy I was when Pieter actually expressed his wish to join our team in Zürich, to work on an experiment with artifical quartz targets in vacuum containers exposed at different altitudes in the Swiss Alps (Vermeesch et al., 2009; see below). At Stanford, Pieter had developed a deep interest in scientific statistics and today, as a service to the community, he is contributing several widely used tools. One of these is the Excel add-in “Cosmocalc”, allowing cosmogenic nuclide calculations, which Pieter developed in Zürich. Later at Birkbeck and the University College, London, he provided a tool to visualise ages of detrital zircon populations (“Density Plotter”; Vermeesch, 2012), and a programme facilitating many different geochronological applications (“IsoplotR”; Vermeesch, 2018). After Pieter left for London, I was able to hire Kristina Hippe as a doctoral student with CRONUS-EU funds. Kristina’s work will be discussed in the section on in situ14C below.

CRONUS-EU has undoubtedly fulfilled its goals, in particular as a starting point to motivate young scientists to become interested in cosmogenic nuclide research. One of several training weeks, during which students and postdocs met many of the senior scientists, took place on Mt. Etna in Sicily. As a particular highlight, I remember a lecture in the field by the late Pete Burnard, in which he vividly explained good sampling strategies. In his honour, the Pete Burnard award is now presented to a young scientist every second year at the small but highly successful DINGUE noble gas meetings that Pete initiated. DINGUE here stands for “Developments In Noble Gas Understanding and Expertise”, but if you speak French, you will understand its other meaning as well!

CRONUS was a success story but work continues on cosmogenic nuclide production rates and their scaling factors. In their “Reflections on future directions for cosmogenic nuclide research” Fred Phillips and co-workers (Phillips et al., 2016b) make several first order recommendations, including continuing efforts to understand the sources of the remaining discrepancies between models and data, and to devise consistent ways of calculating and reporting exposure ages benchmarked against reference ages. They further recommend efforts to facilitate the use of intercomparison materials and integrate the work of ongoing calibration campaigns. Many of these efforts are currently underway.

4.3.2 Work on Cosmogenic Nuclide Production Rates – in Zürich and Elsewhere

As noted above, production rates of cosmogenic nuclides on Earth depend critically on the altitude and latitude of the sampling site and thus are “scaled” almost universally to sea level and high latitudes. After the pioneering work by Lal (1991), scaling procedures were provided, e.g., by Argento et al., (2013) and Lifton, Sato, and Dunai (2014), the latter work also known as “LSD” (Fig. 4.6b). Artificial targets serve as input data and ground truth tests of scaling procedures. Ideally, a series of such targets are exposed simultaneously at different altitudes and, if possible, also at substantially differing latitudes. Given that production rates are very low and exposure durations limited by, for example, the duration of a PhD thesis, targets must be very large, in the range of kilograms or tens of kilograms, rather than on the order of tens of milligrams as is typical for natural samples. A first successful experiment was performed by Kuni Nishiizumi at the UC Berkeley Space Sciences Laboratory and co-workers (Nishiizumi et al., 1996). They exposed several hundred kilograms of water (stored in waterbeds) for five years at one high and one low altitude site in California and then quantitatively extracted the cosmogenic 10Be accumulated in this large amount of water. Brown et al. (2000) repeated the 10Be experiment with 20 litres of water per sample exposed at three different elevations in the French Alps. These workers also determined cosmogenic 3He in a second set of ten litre water samples.

These experiments motivated us to perform our own irradiations, in which several generations of graduate students became engaged. Some of the experiments were very successful, but I will also have to admit failures. In a first round, we used (artificial) quartz as the target, enclosed in ultrahigh vacuum stainless steel containers. In quartz, spallation of Si produces 21Ne, unlike water, for which 3He is the only detectable stable cosmogenic noble gas nuclide. This was deemed important as it should help to decide whether the production of 3He, 21Ne, and radionuclides such as 10Be all follow the same altitude dependence (Fig. 4.6b). On heating, quartz releases 21Ne at fairly low temperatures of about 600 ºC, which we thought should be easy to achieve even for multi-kg samples in steel containers. Although quartz may partly lose 3He over short periods of time, this could be controlled by analysing the headspace of the containers before heating.

Our initial plans were very ambitious and included at least two altitude profiles, one in the Alps, the other in North America (if possible at a high latitude). A complete latitude transect from 60 ºN to close to the equator was also envisaged. In the end, the published project only covered one altitude profile in the Switzerland, with five targets exposed at elevations ranging from 550 metres (in Zürich) to 4550 metres at Monte Rosa, the second highest mountain in the Alps, on the Swiss-Italian border (Vermeesch et al., 2009). Even this down-scaled project kept busy several generations of doctoral students, postdocs, and their supervisors, as documented in four different dissertations by Jörg Schäfer, Florian Kober, Peter Oberholzer and Stefan Strasky. We first had to learn how to make suitable quartz target samples (artificial quartz slabs ground to small grains), and build containers and furnaces to ensure quantitative degassing of one kg of quartz at a temperature of 800 ºC, which was less easy than initially imagined (Strasky 2008; Fig. 4.7). We also had to make sure that Ne blanks were low enough, and we had to evaluate ultrahigh vacuum valves that remain tight even at the low winter temperatures at high altitudes. Finally, the five targets in Switzerland were exposed for a year and analysed after recovery for cosmogenic 3He and 21Ne in the ultrahigh sensitivity Tom Dooley mass spectrometer. Two other samples exposed at different altitudes in Antarctica, deployed and recovered by Christian Schlüchter on two of his trips, were also analysed but unfortunately the results were not precise enough to be useful (Strasky, 2008). Two additional samples, also deployed by Christian in Tibet, unfortunately never made it back to Switzerland. I wonder whether they still languish on a rooftop awaiting recovery.

Pieter Vermeesch performed the data analysis of the targets from the Alps (Vermeesch et al., 2009; Fig. 4.7). Attenuation lengths for 3He and 21Ne agreed within uncertainties. Basically as expected, production rates of both nuclides were proportional to neutron monitor count rates and followed the same scaling relationship as radionuclides. However, the uncertainties of our data and the limited altitude range covered did not allow us to test the slight differences in the modelled altitude scaling for different nuclides as predicted by Lifton et al. (2014) shown in Figure 4.6. The 3He and 21Ne production rates derived from the artificial targets agree – albeit with considerable uncertainties – with values obtained from natural samples. Hence, this first experiment found a mostly successful end more than 10 years after its start. Unfortunately, the same cannot be said for a second experiment, an altitude profile again in the Swiss Alps, but this time with water as the target. We aimed to improve the accuracy by increasing the exposure durations by an order of magnitude to about ten years, and chose water because one of the main goals was to measure the branching ratio of directly produced 3He and 3He that “grew in” by decay of its radioactive precursor 3H. Targets were deployed around 2005 and retrieved and analysed in 2015, but unfortunately the data did not allow a consistent interpretation. The reasons for this failure are unclear, but I suspect that such a long term project would have required better monitoring by the ultimately responsible person – me!

Determining the production rate of cosmogenic noble gases in various minerals has been a side issue in several of our papers, and two of them have been devoted specifically to this topic. Kober et al. (2005) measured cosmogenic 10Be and noble gases in sanidine and Fe oxide minerals in silicic volcanic rocks from Chile that also contained quartz. The well known 10Be and 21Ne production rates in the latter mineral allowed us to determine P(10Be) and P(21Ne) in sanidine and P(3He) in Fe oxides. Such minerals are useful in exposure age studies in volcanic regions where the rocks lack quartz and pyroxene. With Vasily Alfimov and co-workers, Florian Kober also re-determined P(21Ne) in quartz using a large data set from samples with known 21Ne and 10Be concentrations, assuming that P(10Be) in quartz is known (Kober et al., 2011). The approach we used minimised the influence of samples with a complex exposure history.

Martin Frank, now at GEOMAR in Kiel, is an isotope geochemist mainly interested in palaeooceanography. He joined Alex Halliday’s team at ETH in the late 1990s. In an in house seminar we realised that Martin’s expertise matched Jörg Schäfer’s and my interests in the systematics of cosmogenic nuclide production. As mentioned earlier, production rates on Earth are influenced by the modulation of the cosmic ray intensity by the Earth’s magnetic field. It is well known that the geomagnetic field intensity varies with time, but this is usually ignored in exposure age calculations, i.e. production rates are assumed to be constant at a given location regardless of when and how long a sample was exposed to cosmic rays. Shortly before his arrival in Zürich, Martin Frank had produced a field intensity curve over the last 200,000 years based on 10Be data in a global series of deep-sea sediments (Frank et al., 1997). To progress further we contacted Jozef Masarik at Comenius University in Bratislava. Jozef is not only an expert in physical models describing cosmogenic nuclide production in meteorites (as we saw above) but also simulates particle fluxes and cosmogenic nuclide production in the Earth’s atmosphere and solid surface. Martin supplied Jozef with the then best palaeomagnetic field reconstructions over the past 800,000 years and – during several visits to Zürich – Jozef modelled the effects of variable field intensity on the nuclide concentrations of samples taken today (Masarik et al., 2001a). As I recall, we were quite relieved when it turned out that corrections are small (Fig. 4.10). At latitudes >40º, field intensity variations hardly influence production rates, and even at the equator integrated production rates for exposure ages between 40,000 to 800,000 years are only 10 to 12 % higher than the present day values, and for ages below 40,000 years the difference is even smaller. Perhaps somewhat counter intuitively, correction factors for stable nuclides and radionuclides with half-lives longer than a few hundred thousand years are almost identical. An early excursion of the magnetic field has less of an impact on the present day radionuclide concentration than a more recent field excursion of the same magnitude and duration, but for practical purposes this is negligible. For 14C, with its comparably very short half-live of 5730 a, corrections are always smaller than ~2 % because the magnetic field intensity has remained fairly constant during the past ~10 ka, when most of the 14C extant today was produced.

Later Jozef and I showed that effective production rates also depend to some extent on the shape and size of a sampled boulder, because cosmic ray neutrons are more easily lost back to the atmosphere from a non-flat sample than from a flat surface (Masarik and Wieler, 2003).

4.4 In situ14C in Terrestrial Samples

To a scientifically interested public, carbon-14 is without doubt the best known radioactive nuclide used for dating purposes. When I tell someone that we measure ages of rocks with radioactive nuclides, the response is often: “Oh I know, carbon-14”. This refers to the classical 14C dating of organic material, in which the cosmogenic radionuclide 14C is produced in the atmosphere and subsequently incorporated into trees and other plants. In fact, organic 14C dating is by far the most widely used method for isotopic age determination, today overwhelmingly by using Accelerator Mass Spectrometry. However, 14C is also produced as a cosmogenic nuclide in solid matter, mostly by spallation of oxygen atoms by fast neutrons. But this in situ produced 14C is rather rarely used in terrestrial applications. A seemingly good reason for this could be viewed in the short half-life of 14C, which limits the time frame available for study to about 20,000 years at best, much less than the up to several million years covered by other radionuclides such as 10Be. However, it is precisely the much shorter half-life of 14C which makes in situ14C an attractive addition to the cosmogenic radionuclide toolkit. Not only is 14C well suited for dating very young surfaces, but in combination with a long lived nuclide, it also allows the detection of complex exposure histories caused by interruptions in surface exposure over timescales ranging from the latest Pleistocene to the Holocene. Examples are interruptions caused by glacier re-advances, sediment transport, changes in surface erosion rates, and mass removal events (Hippe, 2017).

In situ14C in terrestrial samples so far is mostly studied only in niche applications because the analyses are difficult and time consuming. The difficulties include the needs for quantitative extraction of very low amounts, separation from atmospherically-produced 14C, and ultralow blanks. Pioneering work was done at the University of Arizona by Tim Jull and Nathaniel (Nat) Lifton (Lal and Jull, 1994; Lifton et al. 2001). For his PhD thesis, Nat built a 14C extraction line for quartz samples whose main features were a lower blank than obtained in previous systems, and – most importantly – the ability to extract 14C as CO2 only (preventing potential losses of CO) and to quantitatively separate the in situ14C from atmospheric or organic contamination. We therefore decided to build our own 14C extraction line by essentially following the procedure developed in Tucson. Nat Lifton welcomed Florian Kober to visit his lab and kindly introduced him to his philosophy of 14C extraction and its intricacies. Nat remained a constant source of support and inspiration during the construction of our facility in Zürich.

We copied much of the procedure developed by Nat Lifton, but added some important modifications. First, our extraction line is made of stainless steel tubing, not glass. All of our noble gas extraction and purification lines are stainless steel, and our workshop has extensive expertise in building metal-based ultrahigh vacuum systems. We therefore relied on this experience also for 14C extraction, although common wisdom at the time seemed to be that only a glass line would guarantee a sufficiently low carbon blank. This turned out not to be true. Second, we heated the quartz samples to about 1600 ºC instead of the 1200 ºC used in Tucson. This allowed us to omit the need for lithium metaborate (LiBO2) as fluxing agent, eliminating one source of blank carbon. Third, the MICADAS 200 kV AMS system at ETH/PSI in Zürich (Fig. 3.2b) can be fed with CO2, avoiding the sample graphitisation step, another potential source of blank carbon.

Our system as of 2009 (Fig 4.11) is described by Hippe et al. (2009). One day, Tibor Dunai offered to finance a doctoral student for two years with remaining funds of the CRONUS-EU programme if I could guarantee follow up support. Christian Schlüchter gave this guarantee, and so Florian Kober and I could happily welcome Kristina Hippe in our team in 2008. Kristina had completed her Master’s thesis at the Freie Universität Berlin and arrived in Zürich when Florian and our workshop were working on the completion of the 14C line. Kristina’s first task was to set up analytical protocols and carry out an extensive test programme, consisting of blank, standard, and sample reproducibility tests and much more, all with great support from the AMS team at ETH, especially Lukas Wacker. When the line became operational around 2010 it was one of the very few systems that allowed routine analyses of in situ terrestrial 14C. Kristina and her colleagues used 14C in conjunction with other cosmogenic radionuclides to study denudation rates and sediment storage in the Bolivian Altiplano and the deglacation history in the Gotthard region in the Swiss Alps after the last ice age (Hippe et al., 2012; 2014). An overview of scientific problems to be addressed with 14C is given in her review paper (Hippe, 2017). A remarkable example is the 14C depth profile presented by Lupker et al. (2015) along a quartzite core in Spain from the surface down to about 15 m. These data allowed Maarten Lupker to constrain the fraction of 14C produced by secondary cosmic ray muons. Muons have a considerably larger depth range than neutrons and thus become the dominant producer of 14C below a few meters depth, but their contribution should already be taken into account also in near-surface samples.

While 14C is by now well integrated into the cosmogenic nuclide tool box, its measurement remains very time consuming, which results in a very small number of analyses overall, e.g.,  compared to the workhorse 10Be. Fortunately, sometimes even very few 14C data provide important additional information. For example, Fogwill et al. (2014) complemented 10Be analyses and ice sheet modelling with two 14C analyses to reconstruct the evolution of two major ice streams entering the Weddell Sea in West Antarctica over the past 20,000 years. Nevertheless, more efficient 14C analysis protocols and further reductions of blank values are highly desirable. An important step in this direction is the new 14C extraction line that Maarten Lupker, Kristina Hippe, Lukas Wacker, and colleagues set up at the ETH AMS facility (Lupker et al., 2019). Paul Muzikar discusses recent progress in combining 14C and 10Be analyses and gives an overview of operating in situ14C facilities (Muzikar, 2020).