36Cl Chronologies and ELA reconstructions from the northern boundary of the South American Arid Diagonal
A thesis submitted to the Graduate School of the University of Cincinnati in partial fulfillment of the requirements of the degree of
Master of Science
in the Department of Geology of the McMicken College of Arts and Sciences by
Rachel Thornton
B.S. (Geology), Kent State University, 2011
Advisory Committee:
Dylan Ward, Ph.D. (Committee Chair) Thomas Lowell, Ph.D. Aaron Diefendorf, Ph.D.
i ABSTRACT
This study focused on glaciation in the South American Arid Diagonal (AD) climatic feature which intersects in the Andes mountain range at approximately 24°S. This NE-SW- trending zone of hyper-aridity separates the Atlantic-sourced Easterly moisture belt from the
Pacific-sourced Westerly precipitation in the central Andes. We present new geomorphological maps of the landforms, particularly of glacial moraines across two mountain ranges within the
Western Cordillera of the central Andes. Additionally, new 36Cl chronologies using surface exposure dating were produced from two sites near the northern boundary of the modern AD.
One site is located in the Cordillera del Tatio at ~21°S (Tatio field site), the other is the southernmost valley in a stratovolcano chain called the Cordón de Puntas Negras at ~24°S (SPN field site). Paleo-ELAs were reconstructed utilizing a two-dimensional numerical model to simulate glaciers over a 90m digital elevation model of the upstream catchment.
We find that four glacial stages were present at both field sites sampled. Glacial stabilizations from the El Tatio field site, once assumed Lateglacial in age are on the order of 10-
20 thousand years older than previously published. Tatio ages ranged from ~25 ka to as early as
MIS 5. We also found that dated glacial stages from El Tatio and the Puntas Negras are not synchronous with lake highstands on the Altiplano, but projected ages of the younger glacial stages indicate both sites were glaciated during the Tauca phase. Regional comparison indicated the possibility of glacial occupation of the Tatio site into the Lateglacial during the Coipasa phase (12 ka) after rapid deglaciation of the SPN field site. Regional ELA comparisons supported the trend of moisture loss from the NE to SW portions of the climatic AD feature.
ii
iii ACKNOWLEDGEMENTS
First and foremost, I thank my advisor Dr. Dylan Ward for this opportunity, his patience, and for pushing me to accomplish things I once considered out of reach. I’d also like to thank my supportive committee Dr. Thomas Lowell and Dr. Aaron Diefendorf for their feedback, guidance, and encouragement. I would not have made it through this project without the help of my lab group, particularly Chris Sheehan and Jason Cesta for their help collecting and processing my samples, for which I also owe thanks to Sarah Hammer. I’d like to acknowledge Dr. Esteban Sagredo for his advice and help during our field season in Chile. For help, support, much-needed laughter, and friendship I thank my friends inside and outside the Department of Geology. Finally, I’d like to thank my family, particularly my husband Robert Naylor, for supporting me and encouraging me the whole way. This project was funded by a National Science Foundation award from Geomorphology & Land-use Dynamics (GLD) (EAR-1226611) to Dr. Dylan Ward. Additional funding was granted by the Geological Society of America and Association for Women Geoscientists.
iv TABLE OF CONTENTS
Abstract…………………………………………………………………………………………….i
Acknowledgements………………………………………………………………...... iv
List of Figures...... ………………………...... ix
List of Tables...... ………………………...... x
List of Appendices...... ……………...... xi
1. Introduction...... 1
2. Geologic Background...... 4
3. Regional Climate...... 6
4. Field Sites...... 10
4.1. Site selection...... 10
4.2. SPN site...... 10
4.3. Tatio site...... 12
5. Literature Review...... 14
5.1. Overview...... 14
5.2. Lake and salar basins...... 14
5.3. Hydrologic budget...... 16
5.4. Glacial history...... 17
6. Methods...... 21
6.1. Cosmogenic surface exposure dating...... 21
6.1.1. Production & scaling...... 22
6.1.2. Limitations of surface exposure dating...... 24
v 6.2. This study...... 26
6.2.1. Sample locations...... 28
6.2.2. Sampling methodology...... 28
6.3. Bulk rock 36Cl sample preparation...... 29
6.3.1. Physical preparation...... 30
6.3.2. Chemical preparation...... 30
6.3.3. Carrier/Isotope dilution...... 31
6.3.4. Dissolution & procedural blank preparation...... 31
6.3.5. Precipitation...... 32
6.3.6. Anion exchange chromatography...... 33
6.3.7. Final sample preparation...... 34
6.3.8. Target preparation for AMS measurement...... 34
6.4. TCN age calculation...... 35
6.5. Geomorphological mapping methods...... 37
6.6. Numerical modeling methods...... 42
7. Results
7.1. El Tatio region mapping...... 43
7.1.1. Tatio site mapping...... 45
7.2. Puntas Negras region...... 48
7.2.1. SPN site mapping...... 50
7.3. Mapping interpretations...... 53
7.4. Cosmogenic results...... 55
7.4.1. SPN ratios...... 55
7.4.2. Tatio ratios...... 57
vi 7.4.3. SPN apparent ages...... 57
7.4.4. Tatio apparent ages...... 58
7.5. Apparent age interpretations: SPN & Tatio...... 60
7.5.1. Scenario 1...... 63
7.5.1.a. SPN...... 63
7.5.1.b. Tatio...... 64
7.5.2. Scenario 2...... 65
7.5.2.a. SPN...... 65
7.5.2.b. Tatio...... 65
7.5.3. Scenario 3...... 66
7.5.3.a. SPN...... 66
7.5.3.b. Tatio...... 67
7.6. Numerical modeling results...... 68
7.7. ELA interpretations...... 70
8. Discussion...... 72
8.1. Regional glacial trends: Northern Altiplano...... 72
8.2. Tatio site chronology: Cosmogenic ages...... 73
8.3. SPN site chronology: Cosmogenic ages...... 75
8.4. Paleo-ELA plots...... 77
9. Conclusions...... 79
References...... 81
Appendix...... 96
vii
viii LIST OF FIGURES
Figure 1: Global atmospheric circulation cells and climatic features affecting South America....2
Figure 2: Moisture loss to the southwest AD between Tauca Phase and the Lateglacial...... 3
Figure 3: Regional DEM and morpho-tectonics of study sites in the Central Andes...... 4
Figure 4: South Puntas Negras field site...... 7
Figure 5: AD modern climatic conditions and glacially-delineated AD boundaries...... 11
Figure 6: How surface processes affect cosmogenic apparent ages...... 25
Figure 7: Field images of boulder sampling...... 29
Figure 8: Examples of landforms in the field...... 41
Figure 9: El Tatio regional glacial map and elevation profile...... 45
Figure 10: Map of glacial stages and sample locations from El Tatio field site ...... 46
Figure 11: SPN regional glacial map and elevation profile...... 49
Figure 12: Map of glacial stages and sample locations from SPN field site...... 50
Figure 13: Map of glacial stages and sample ages from SPN field site...... 58
Figure 14: Map of glacial stages and sample ages from El Tatio field site...... 58
Figure 15: TCN apparent ages and errors plotted with lake curve from Laguna Miscanti...... 61
Figure 16: Probability density functions or camel plots for Tatio and SPN glacial stages...... 62
Figure 17: Locations of each formerly-glaciated site used to calculate paleo-ELAs...... 69
Figure 18: Numerical modeling paleo-ELA distance plots sites...... 70
Figure 19: Regional synthesis of paleo-ELA and projected moraine ages...... 71
ix LIST OF TABLES
Table 1. Boulder sample properties...... 56
Table 2. Sample apparent ages...... 59
x APPENDICES
Appendix A.
Table 3. Raw AMS data SPN site...... 103
Table 4. Raw AMS data Tatio site...... 104
xi 1. Introduction
Studying paleoclimatic transitions is crucial to understanding the impact and extent of modern climate components during periods of environmental change. Glaciers are excellent tracers of climatic conditions (Oerlemans, 1991; Lowell, 2000); thus, preserved palaeoglacial deposits provide an integral archive as a proxy for conditions able to sustain patterns of prior glacial occupation. A more thorough understanding of patterns, extent, and timing of glaciation in Chile are pivotal to understanding climatic events from the Southern
Hemisphere.
The South American Arid Diagonal (AD) is a NW-SE-trending climatic feature in the arid Andes between 18-27° S. This zone of hyper-aridity physically separates the northern
Andes, which receive summer Atlantic moisture, from the southern Andes, which receive winter westerly moisture from the Pacific. Within the AD climatic feature is a zone with no evidence of prior glaciation from 24-27° S, which represents the modern glacially-delineated AD. It is clear from the glacial deposits that the AD has not always been this dry, which begs the question: when was this region last glaciated?
For the last 30 years, the majority of glacial chronologies from South America have been produced in the tropical northern Andes and southern Andes near the Chilean lake district and
Patagonia. In the modern climate, these regions are affected by the Northern and Southern
Trades and Westerlies, respectively. Work in this area has resulted in understanding primary glacial sensitivity (Sagredo and Lowell, 2012) (Figure 1B) as a lens to interpret levels of temperature and precipitation changes supporting observed glacial patterns. It is, however, unclear whether glacial deposits adjacent to the Arid Diagonal (AD) spatially or temporally
1 correlate across the hyper-arid Central Andes, where there is a marked paucity of hydrologic proxies in the literature.
There was little to no age constraint in the
southern Puna plateau portion of the AD until a study
from the Chajnantor Plateau (23°S) added a valuable
new data set which, at the time, was the southernmost
data set from the northern edge of the AD (Ward et al.,
2015). Results from Ward et al., 2015 indicated
widespread glaciation during Tauca phase (15-20 ka),
named after the largest Altiplanic lake highstand from
the Lake Titicaca basin. During the Lateglacial (13-10
ka), however, there are glacial deposits in the
subtropics, but they disappear southwest of ~21° S
latitude (Figure 2). Adjacent field sites exhibited
evidence of glaciation both north and eastward of the
Chajnantor Plateau (Zech et al., 2009; Blard et al.,
2014) during the Lateglacial. Figure 1A. Global atmospheric circulation cells. 1B atmospheric and oceanic circulation cells affecting field Given the primary moisture-limiting sensitivity sites and seasonal climatic features responsible for seasonal gradient in precipitation patterns. Colored and numbered polygons represent primary glacial in this region, further examination of an apparent sensitivities. Number 3 in pink is affected by Atlantic precipitation and is both temperature- and precipitation- sensitive. Group 4 & 5 in green and blue are subtropical spatiotemporal trend of moisture loss southwest of and mid-latitude glacial groups. These groups are precipitation-sensitive. Yellow markers are cosmogenic glacial records from El Tatio (north) and Nevado de 21°S is the motivation for this study. The goal of this Chañi (east). The red marker is the south Puntas Negras site (SPN)
2 project was to develop a
new cosmogenic glacial
chronology near the
northern boundary of the
AD climatic feature and
south of the Chajnantor
plateau where there are no
current glacial records. Figure 2: Apparent loss of moisture delivery to support glaciation between Tauca Phase (left) through the Lateglacial (right) SW of ~20°S. Orange and blue dashed lines indicate modern AD climatic feature. The red polygon represents the boundaries of the glacially-delineated AD with no This new data set, along preserved evidence of prior glaciation. Yellow markers are glacial chronologies published in other studies and red indicates glacial chronology from this study. The Chajnantor plateau chronology is represented with a star. Figure modified from Ward et al., 2015. with modeling efforts, will shed light on the hypothesis that the glacially-delineated AD expanded and/or migrated to the northeast, encompassing the Western Cordillera of the central Andes during the last glacial occupation of the AD.
3
2. Geologic background
The Andes, along the western margin of South America, are composed of semi- continuous mountain ranges and volcanic complexes spanning 7000 km. The Central Andes
(21°- 25°S) are characterized by distinct structurally-controlled physiographic units resulting from subduction of oceanic crust from the Pacific plate under the continental South American plate since the Jurassic (Figure 3).
Figure 3: Regional digital elevation map of study sites in the Central Andes Along the Pacific coast, these morphotectonic units include the Coastal Cordillera and
Central Depression which comprise the Andean forearc. Arc magmatism originated in the
Coastal Cordillera during the Jurassic and migrated inland and eastward to the Western
Cordillera throughout the Miocene-Quaternary from continuous subduction.
4 The Altiplano-Puna plateau is bounded by the Eastern and Western Cordillera of the
Andes. The surface of the plateau is characterized by ignimbrite flows, dacite domes, dacite- andesite composite stratovolcano chains, and basalt-andesite scoria cones. This central portion of the Andes from 14°-28° S, the Central Volcanic Zone, was built from volcanism beginning as early as the Upper Oligocene to recent (de Silva 1989; Isacks 1988; Matteini et al., 2002).
Average elevation is ~4000 m across the arid Altiplano-Puna with volcanic peaks as high as
6000 m.
Crustal shortening trending WNW-WSE and extension trending ENE-WSW created
Basin-and-Range-like surface relief. The intersection of extension and NW-SE transversal lineaments led to a transtensional regime opening pull-apart basins along the Central Puna. The topographic lows associated with both types of deformation are the modern N-S aligned salares, which are characteristic features of the modern landscape. Uplift of the Eastern flank of the
Andes from 14 to 9 Ma is thought to be the mechanism for drainage separation from the forearc and establishment of Altiplano-Puna internal drainage ~15 Ma (Vandervoort et al., 1995;
Hartley, 2003).
5 3. Regional Climate
The Andes range intersects regional and global atmospheric circulation cells (Figure 2A).
The central Andes are located at the sub-tropical zone of high pressure between the south Hadley circulation cell and the south Ferrel Cell; as such, they fall between two different precipitation regimes which exhibit steep east-west and seasonal gradients. Mean annual precipitation (MAP) is highly variable with respect to latitude due to these differing precipitation sources and the topographic modification of moisture distribution (Vuille 1999; Garreaud 1999). As such, the range is separated into broad climatic sections: the tropical Andes, the arid central Andes region containing the Atacama Desert, and the southern Andes.
The central Andes remain dry due to the upwelling Humboldt current off the west coast of Chile, the descending arm of the S Hadley circulation cell, and the orographic effect across the
Andes range. Glaciers cannot be sustained in the arid Andes despite peaks exceeding 6000 m due to precipitation levels under 200 mm/yr (Betancourt et al., 2000; Clapperton 2000; Ammann et al., 2001; Vuille et al., 2012). The climate of the AD is also affected inter-annually by the South
American Seasonal Monsoon and El Niño/Southern Oscillation (ENSO) circulation (Garreaud,
R.D. et al., 1999; Kanner et al., 2012; Baker and Fritz, 2015).
North of the AD, easterly precipitation reaches a maximum of 400 mm/y at 18°S and decreases linearly from 18° to 20°S with a lapse rate of 100 mm/y per degree of latitude. At approximately 23°S, precipitation rapidly plunges to a minimum of 100-200 mm/yr from 23-
26°S (Ammann et al., 2001). Average elevations within the AD are ~4000 m and mean annual temperatures ~4°C (Kull and Grosjean 2000); therefore, the high peaks intersect the 0°C isotherm, a condition previously described as “thermal readiness” (Messerli, 1973). Despite continuous permafrost (Grenon, 2007; Kull and Grosjean, 2000; Amman et al., 2001), however,
6 there is no modern glaciation from 18°S to 27°S (De Martonne, 1934; Houston and Hartley,
Zech et al., 2008) with the exception of small snowfields and rock glaciers (Azócar & Brenning
2010).
Figure 4: Region of the AD affected by the easterly trades during austral summer is shaded in orange beginning at the equator and tapering off around 18°S latitude. Region affected in winter by the westerlies shaded blue starting at ~28°S latitude. The 0° isotherm is plotted on the figure and is continuous across the arid diagonal as it intercepts the topography throughout the region. Precipitation data is plotted in the center across the AD, illustrating the average precipitation around 200 mm/year. This moisture-limitation is why snowline disappears across the AD and boundaries of the AD are marked by orange and blue vertical lines. Field sites are plotted at their latitude within the AD. The yellow markers are El Tatio field site and Nevado de Chañi, which were not visited during the 2016 field season. The SPN site is marked in red as this field site was visited during austral summer in 2016. Figure modified from Ammann et al., 2001. The tropical Andes receive Atlantic-sourced moisture from the easterly trade winds of the
Intertropical convergence zone (ITCZ), which are displaced southward during austral summer
(DJF). During this time, components of the South American Seasonal Monsoon (SASM) (Zhou and Lau, 1998; Vera et al, 2006) direct easterly moisture to the eastern Andes and southern
Altiplano (Garreaud et al., 2009; Lenters and Cook, 1997). A continental low over the Gran
Chaco region in Argentina (~25°S) forms due to transport of Atlantic moisture onto the continent. Convective moisture from the eastern lowlands and Amazon basin is diverted southward via a low-level jet (Seluchi et al., 2003). Upper-level (300 hPa) anticyclonic circulation resulting from latent condensational heat forms the “Bolivian High” (BH), (Lenters
7 and Cook, 1997) which transports monsoonal precipitation southward to the northern Puna and portions of the arid Andes. In summary, strengthened and southward migration of the BH increases moisture flux to the northern AD. The opposite (northward migration and decreased moisture flux) occurs when the BH is weakened. The BH system is modulated by the SASM and continental convection, which is closely tied to both Atlantic and Pacific convergence zones, or
ITCZ.
In addition to the annual anticyclonic system described above, the central Andes are affected by phases of El Niño/Southern Oscillation (ENSO). ENSO drives precipitation and atmospheric circulation heterogeneity on inter-annual timescales (Vuille, 1999; Garreaud and
Aceituno, 2001). Using daily precipitation and reanalysis data, Vuille (1999) demonstrated that
ENSO significantly decreases summer precipitation over the Altiplano.
The Southern Andes receive advective extra-tropical moisture sourced from the Pacific and transported by westerly circulation such as the South American Westerly Wind Belt. The
Altiplano receives very little precipitation from May to October due to the maximum northwardly-displaced position (27°S) of the subtropical westerly jet stream, typically focused at
~50°S (Zech et al., 2007). Upwelling along the western margin of South America associated with the Humboldt current limits evaporation and transport of Pacific moisture onto the continent.
Prevailing dry, cool westerly air flow collides with easterly winds in the upper troposphere, limiting convective precipitation on the eastern flank of the Andes during austral winter months
(JJA). Rare winter and inter-seasonal precipitation in the central Andes have been attributed to extra-tropical polar cold fronts or cut-off low pressure systems from the westerly wind zone.
These produce short-lived but significant precipitation events (Vuille 1999).
8 In summary, hemispheric (meridional) and tropical-subtropical pressure gradients are key regulatory components of the easterly (wet) – westerly (dry) seasonal climate dynamic in this region (Garreaud et al., 2003).
9 4. Field sites
4.1. Site selection
Two sampling sites were chosen to analyze the timing and extent of glaciation across the northern edge of the AD. The northernmost site El Tatio and a valley from the southern Cordón de Puntas Negras (SPN). The Tatio site, near El Tatio geyser field, is located at 23°44’ S, 67°46’
W, approximately 3° of latitude further north from SPN. Although the Tatio moraines had not been dated, they were reported to be late glacial in age by earlier studies which made the assumption that they were concurrent with Altiplanic lake phases (Jenny et al., 1996; Kull and
Grosjean, 2000; Glasser et al., 2009).
4.2. SPN site
The SPN valley is the southernmost formerly-glaciated valley of the Puntas Negras stratovolcano chain (23.8°S, 67.5°W). The Cordon de Puntas Negras, approximately 40 km south of Lascar volcano, are the northern-most chain of stratovolcanoes associated with the Upper
Cenozoic, NW-SE-trending Calama-Olacapato-El Toro (COT) lineament. Here, Quaternary eruptions produced basaltic-andesitic lava flows and scoria cones which are characteristic of this portion of the Puna Plateau.
The location of the SPN site was chosen foremost for the preservation of multiple stages of glaciation identifiable by aerial photography. SPN is also adjacent to El Laco mining facility with infrastructure such as access roads and off-road parking to ease sampling in this remote location. The SPN site is a relatively narrow valley with an impressive expression of nested lateral, frontal, and ground moraines (Figure 5). It is located 30 km from the Quebrada
Nacimiento fault which bounds three endorheic lake basins: Laguna Miñiques, Miscanti, and
10 A B
Figure 5: A. Overview of South Puntas Negras (SPN) field site. B: Lower left photo from Laguna Miscanti on the western end of the SPN site. C: Lower right photo from the formerly-glaciated valley on the southeast end of the SPN site. Preliminary glacial stages are marked on this photo. The red is the relatively older glacial stage I. The blue is the relatively younger glacial stage II.
Pampa Varela (Ramirez and Gardeweg, 1982; Grosjean et al., 2001). The valley is 12 km from
Laguna Miscanti (23°44’ S, 67°46’ W), a saline, alkaline (pH 8 - 8.8) lake 10 m in depth that has been cored and dated using radiocarbon (Grosjean et al., 2001). The presence of Miscanti allowed lake stages and glacial stabilizations to be temporally compared to test the hypothesis that these moraines and lake highstands occurred synchronously.
11 4.3. Tatio site
The Cordillera del Tatio is a stratovolcano chain built predominantly from interbedded
Late Cenozoic-to-modern volcanic peaks, lava domes, and expansive dacitic ignimbrite sheets
(Davidson et al., 1995). Topography of this portion of the arid Andes is affected by the Altiplano
Puna Volcanic Complex (APCV) (de Silva, 1989b; Pritchard and Simons, 2002; Lucchi et al.,
2009a,b). The APCV is highly uplifted and bordered by the thick-skinned fold-and-thrust Eastern
Cordillera to the east and an additional thrust system, the West-Vergent Thrust System to the west (Lucchi et al., 2009). Today there is active magmatism here, responsible for the Tatio geothermal field and relatively young dacitic domes such as Co. La Torta (34 ka). This volcanic arc region is bounded to the west by the Preandean Depression with several N-S striking tectonic basins such as the Salar de Atacama from extensional tectonic activity.
The Tatio site, near El Tatio geyser field, is located at 23°44’ S, 67°46’ W, approximately
3° of latitude further north from SPN. Although the Tatio moraines had not been dated, they were reported to be late glacial in age by earlier studies which made the assumption that they were concurrent with Altiplanic lake phases (Jenny et al., 1996; Kull and Grosjean, 2000;
Glasser et al., 2009).
A recent 10Be chronology from Nevado de Chañi in Argentina at ~ 24-25°S by Martini et al., 2017 was compared with the southern Puntas Negras and Tatio records from this study. The record from Chañi is approximately the same latitude and altitude as the SPN site but located in the Eastern Cordillera. The addition of the Martini 2017 study enabled the evaluation of regional trends in glaciation from the eastern and western flanks of the central Andes. The Tatio, SPN, and Chañi sites envelop the northern boundary of the AD, which is where an apparent threshold of moisture delivery may have existed during the last glacial occupation of the region (Ward et al., 2015).
12
13 5. Literature Review
5.1. Overview
Moisture variability and changes in precipitation gradient near and within the AD have been studied using numerous hydrologic budget proxies. Sediment cores and shoreline deposits from lake and salar basins (Minchin, 1882; Servant and Fontes, 1978; Servant et al., 1995;
Sylvestre et al., 1999; Plackzek et al., 2006a,b; 2009a,b, 2011) are often correlated with glacial records and hydrologic modeling (Hastenrath and Kutzbach, 1985; Blodgett et al., 1997) to identify periods of increased precipitation or aridity. Glacial chronologies are paired with these proxy data to study trends in relative effective moisture variability, as glacial geometry is affected by precipitation and temperature (Ohmura et al., 1992). Lacustrine records from this region yield continuous paleohydrologic data exceeding 100 ka and are used to construct time series analyses to correlate the southern hemisphere expressions of north Atlantic climatic events such as the Younger Dryas and Heinrich events. Reconstructions from the Altiplano reveal that the AD may not have been as arid throughout the Pleistocene as it is today. The AD may have migrated or contracted/expanded through the Pleistocene/Holocene transition to include portions of the Western Andes around the time of deglaciation. Below, I review the key literature and summarize existing knowledge of paleoclimate variability in the AD and adjacent Altiplano.
5.2. Lake and Salar Basins
Between 14° and 22°S on the Altiplano, there are four expansive internally-drained basins: Titicaca (16°S, 69°W), which is separated from the Poópo (18.5°S, 67°W) basin by the
Río Desaguadero, Coipasa (19.5°S, 68°W), separated from the Poópo by the Laka sill, and Uyuni
(20°S, 67°W). Fresh-water Lake Titicaca occupies 8560 km2 of the northern Altiplano with a depth of 285 m (Argollo and Mourguiart, 2000; Placzek et al., 2013). On the southern Altiplano,
14 Lake Poópo is 2530 km2. The Coipasa and Uyuni salar basins on the southern Altiplano are separated by a mere 1m topographic high in the modern climate; therefore, during past lake transgressions, they have overflowed to connect with a total area of ~12,100 km2 (Argollo and
Mourguiart, 2000; Placzek et al., 2013). Two cores were collected from the Salar de Uyuni
(Risacher and Fritz, 2000; Fornari et al., 2001; Baker et al., 2001a; Fritz et al., 2004), both containing alternating salt and mud layers, salt indicating desiccation, mud indicating expansion, respectively. This evidence along with shoreline deposits exposed the current lake levels and salt flats indicate that moisture delivery to the Altiplano/Puna region has fluctuated in geologic time
(Minchin, 1882; Servant and Fontes, 1978; Sylvestre et al., 1995; Sylvestre et al., 1999; Placzek et al., 2006a,b; 2009a,b, 2011). The amalgamation of shoreline deposits, ages from Uranium- series geochronological methods used on sediment cores, natural gamma data, and crosscutting landform observations were used to re-construct lake levels.
Pioneering studies of the Altiplanic lakes focused on shoreline deposits as evidence for the largest lake highstand, the Tauca highstand from 17-14 ka and utilized hydrologic budget modeling to estimate lake surface area and volume (Servant and Fontes, 1978; Hastenrath and
Kutzbach, 1985). Re-evaluation of U-Th and Ar/Ar dates from two cores collected from the
Uyuni-Coipasa basin (Fornari et al., 2001; Risacher and Fritz, 2000; Baker et al., 2001a; Fritz et al., 2004) and coupling with shoreline tufa deposits (Placzek et al., 2006a,b) allowed identification of six lake oscillations from a unified chronology (Placzek et al., 2013). Lake highstands include: Ouki (125-95 ka), Salinas (95-80 ka) which the authors state could have been a separate phase or perhaps was part of the Ouki phase, Inca Huasi (50-40 ka), Sajsi (25-20 ka),
Tauca (17-14), and Coipasa (13-11 ka). The lake stages were corroborated by both natural gamma and salt content data from the core analyzed by Fornari et al., 2001; moreover, among the Late Pleistocene lake phases, only Ouki and Tauca were deep lakes. Lake Tauca is estimated
15 to have reached a maximum depth of 140m during highstand (Bills et al., 1994; Baker et al.,
2001a). Baker et al. (2001a) attributed the wet periods from the Salar de Uyuni to maxima in summer insolation. This group suggested that land-sea temperature gradients were increased compared to the modern climate, enhancing the SASM and bringing more easterly-sourced precipitation to the tropics in South America.
Further south on the Altiplano of northern Chile (23°S, 68°W), a salt core was drilled from the Salar de Atacama from the western end of the central Andes. Bobst et al., 2001 used sedimentary structures and evaporite textures to study net hydrologic balance and identify wet/dry periods. They determined that from 75.7-60.7 ka, 53.4-15.3 ka, and 26.7-16.5 ka were wet periods. Moreover, the largest, wettest lake stage was from 26.7 to 16.5 ka. Brief phases of expansive mudflats rather than lake stages also occurred in the Holocene from 11.4 to 10.2 ka and from 6.2 to 3.5 ka. These wet periods match hydrologic budget studies from the Bolivian
Altiplano with the wettest cycle (26.7 to 16.5) overlapping the Tauca phase. The early Holocene stage (11.4 to 10.2) is synchronous with the Coipasa phase.
5.3. Hydrologic budget
Hydrologic budget and balance modeling have provided estimates of mean annual precipitation (MAP) and temperature changes necessary to support field observations. These models follow the basic premise that in this closed basin, evaporation must not outpace precipitation during a lake transgression.
Hastenrath and Kutzbach (1985) concluded, by using deposits estimating the former surface of paleolake Tauca, that a 300 mm/yr increase in precipitation must have occurred to support the Tauca highstand. Similarly, in his 2001 paper, Grosjean estimated precipitation rates of >500 mm (an increase of ~300 mm relative to modern) occurred at 23°S. These estimates are
16 in agreement with glacier modeling done in the valley sampled at the Tatio site by Kull and
Grosjean (2000) suggesting precipitation rates must have been 515±45mm/yr in the northern
AD. These modeling studies utilized the paleolake deposits to determine P/E relationships that match the paleo-shorelines, estimated depth, and estimated heat balance. These studies, along with the “thermal readiness” (Messerli 1973) concept, illustrated in Figure 4, provided support to accept assumptions of moisture sensitivity to glaciation during the late Pleistocene, similar to modern AD conditions. It has since been an assumption from field- and model-based support that glacial moraines from the northern AD are approximately Lateglacial in age.
Some of the first researchers to use absolute dating methods on deposits from Lake Tauca once thought meltwater filled Lake Tauca with highstands occurring post-deglaciation (Servant and Fontes, 1978). This was later considered implausible based on calculations of the LGM glacial ice volume falling far short of the estimates for total volume of the maximum extent of
Lake Tauca (Hastenrath and Kutzbach, 1985; Blodgett et al., 1997). This exemplifies the value of combining field and hydrologic modeling studies to evaluate existing hypotheses about lake and glacial stages assumed to be contemporaneous.
5.4. Glacial History
As earlier qualitative work gave way to technological advancements and additional fieldwork in this remote area, evidence suggested glaciers may have re-advanced or reached maximum extent at the same time as the Tauca phase (~14-17 ka BP) (Seltzer, 1992; Clayton and Clapperton, 1997; Ammann et al., 2001; Placzek et al., 2013). Early modeling by Hastenrath and Kutzbach, 1985 suggested an increase in summer precipitation may have been the forcing responsible for glacial maxima and lake transgressions occurring synchronously. This hypothesis is supported by correlations of terminal moraine complexes in similar configurations from the
17 Andean Eastern Cordillera in Southern Peru (Seltzer, 1990), Bolivia, as well as the northern boundary of the AD between 18 and 25°S (Jenny et al., 1996). Additionally, Salar de Uyuni and the Salar de Atacama Tauca and Coipasa phases also align with high accumulation and low d18O values from the Sajama ice core (Thompson et al, 1998) from southwest Bolivia (18°S).
A data set describing the chronology of the Tauca phase from erosional and depositional shoreline deposits surrounding the Uyuni-Coipasa basin (Servant and Fontes, 1978) was produced by Sylvestre et al., 1995. This study also noted a terminal moraine complex of three stages at roughly 15, 9, and 6 km, respectively, from modern valley glaciers of the Andean
Eastern Cordillera. Moraines in a similar configuration were also observed in Southern Peru
(Seltzer, 1990) as well as the northern boundary of the AD (18-25°S) (Jenny et al., 1996). These studies qualitatively provided support to accept the assumption of humidity-limitation to glaciation during the late Pleistocene, similar to modern AD conditions.
Additional support for the tentative Lateglacial age assignment (Jenny et al., 1996) was found after stages were further described, correlated via geomorphological descriptions and ELA reconstructions (Clayton and Clapperton, 1997), and paired with minimum radiocarbon ages
(Amman et al., 2001). The oldest and most degraded stage, “N-I” from Amman et al., 2001 relatively pre-dated an intermediate stage “N-II”, characterized by high-relief, sharp-crested lateral moraines and observed as far south as 25°S, linked most probably to “Advance 3” from
Clayton and Clapperton, 1997.
Clayton and Clapperton inferred that their “Advance 3” was synchronous with Tauca phase based initially on fan-deltas of glacifluvial outwash associated with Advance 3 building into the paleolake Tauca basin. The fans lie below the altitude of the highest paleoshoreline, implying a glacial advance followed by a lake level rise. Advance 3 glacifluvial outwash fan- deltas were built into the basin of paleolake Tauca, which imply they respectively advanced and
18 transgressed synchronously. The “N-II”/Advance 3 correlation assigns a radiocarbon minimum age of 13,300 14C yr BP from peat at the base of a terminal moraine, placing the age of “N-
II”/Advance 3, and local LGM at this latitude in the Late Glacial (Clayton and Clapperton, 1997;
Sylvestre et al., 1999; Blard et al., 2009; Placzek et al., 2013).
Clayton and Clapperton re-constructed equilibrium line altitudes (ELA) from the N-
II/Advance 3 moraine using maximum altitude of lateral moraines (Meierding, 1982) from valley glaciers where lateral moraines were well-preserved with clearly-delineated limits. ELA for advance 3 or N-II, or the Tauca phase moraine was estimated at 4435±50 m for Co. Anzanaques and 4525±50m for Co. Tunupa. The authors cite the 90m difference as evidence suggesting
Advance 3 regional ELA exhibited a gradient which parallels the maximum snowline altitudes from Gosse et al., 1995; implicating a NE or Atlantic source of humidity. The modeling efforts of Clayton and Clapperton, 1997 estimated 3 possible precipitation and temperature configurations (relative to modern values) necessary to support the geometry and ELA of
Advance 3. They include: (1) a 4.5-7°C temperature depression with no change in precipitation
(2) a 300 mm/yr increase in precipitation, suggested initially by Hastenrath and Kutzbach, with a temperature depression of 3-5°C, or (3) an increase by 600 mm/yr in precipitation, as suggested by Grosjean (1994). The 600mm/yr with a temperature depression of 1.8-3.6°C match an interpreted 15-25 m rise in paleolake Lejía (23°S, 67°W, 4325m), bringing the surface to 9-11 km2 as opposed to just 2 km2 today. Laguna Lejía is considered representative of the Altiplano area from 21-24°S.
The Altiplano lake phases from the Salar de Uyuni not only overlap wet phases from the
Salar de Atacama sediment core, they also match glacial moraine stabilizations north of ~20°S and south of ~40°S. There is a bias in available data sets, though, as Bobst et al., 2001 suggest.
Most paleoclimate records from the Altiplano are less than 100 ka and the majority of records are
19 from the northern tropical or southern Andes. There is little data from the arid, central Andes where there is no modern glaciation and the majority of lakes are evaporite-encrusted pans. One of the most recent studies from this region was from the Chajnantor plateau, which is located between the Tatio and SPN field sites at ~22.5°S. Samples were collected and dated using cosmogenic 36Cl from the peaks of the Chajnantor Plateau surrounding the Atacama Large millimeter/submillimeter Array (ALMA) observatory. New 36Cl ages from Ward et al. (2015) were reviewed in context with proximal previously published glacial records (Figure 2). Results from this study indicated glacial stabilizations at 25-40 ka, 15-17 ka, and 12-14 ka. The youngest of these moraines, however, were absent south of ~20°S, which implicates a loss of enough moisture delivery to support glaciation south of 20°S within the climatic AD.
Pairing glacial moraine ages with the age constraints on the salar and lake basins is where the research evolved in the tropical portion of the range, but as glacial records become scarce toward the AD, this coupling becomes more challenging.
6. Methods
6.1. Cosmogenic Surface Exposure Dating
Advances in techniques such as terrestrial cosmogenic nuclide (TCN) exposure dating have improved the quantitative study of Earth-surface processes and rates of landscape evolution. We are now able to study a range of spatial and temporal scales which were not possible just decades ago (National Research Council, 2010). Directly dating exposure times of young volcanic rocks, geomorphic surfaces, and soils is a powerful tool for investigating the timing and rates of surface processes. In situ TCN concentration studies have contributed vital chronological constraints for regional and global paleoclimatic reconstructions (Gosse et al.,
20 1995; Ivy-Ochs et al., 1999; Owen et al., 2003; Liccardi et al., 2004; Douglass et al., 2006). This dating method is based on the influx of cosmic radiation, or particles (predominantly protons and alpha particles; 83 and 13 percent, respectively), at energy levels sufficient (~1 GeV < E < ~1010
GeV) to induce reactions with Earth’s upper atmosphere, producing secondary particle showers.
Resultant secondary cosmic rays interact with target atoms in minerals at and within the first ~3-
5 meters of the Earth’s surface. Past this depth, cosmic rays are attenuated with depth due to successive energy loss, described by Beer’s law. The product of spallation reactions of target minerals rare isotopes, in situ cosmogenic nuclides, which are diagnostic of surface exposure.
Six terrestrial cosmogenic nuclides (3He, 10Be, 14C, 21Ne, 26Al, and 36Cl) are commonly used for geologic applications due to: (1) lack of natural occurrence from other processes, making radiogenic origin clear (2) reasonable decay rates relative to durations of surface Earth processes being investigated (3) abundance of target mineral (4) production rate is high enough for them to be measured within detection limits of current analytical methods. In principle, the time since exposure of a sample can be determined by measuring the net concentration of cosmogenic nuclides in a sample (accounting for variation in cosmic radiation flux, decay of the nuclide, and erosion) and dividing it by the rate of production. (Gosse and Phillips, 2001).
6.1.1. Production & Scaling
Cosmogenic isotopes are produced by a few different types of reactions including: spallation, secondary thermal and epithermal neutron capture, and muon-induced reactions (Lal
& Peters 1967, Lal 1988a). Spallation reactions involve collision of secondary particles possessing sufficient energy levels to shatter nuclei of target minerals, which produce rare daughter nuclides with smaller mass relative to the parent element. Muon capture reactions involve negatively charged muon particles occupying the electron shells of target atoms, where
21 they are subsequently captured by the nucleus to produce daughter cosmogenic nuclides.
Relatively less apt to interact with atoms than other particles, muons maintain energy levels suitable for nuclide production (attenuation length approximately 1500 g cm-2; Brown et al.,
1995a); thus, are a more prominent production source at depth (Heisinger et al., 1997; Stone et al., 1998b). Thermal neutron absorption reactions occur due to energy discrepancy with other particles in the radiation flux; therefore, low energy thermal neutrons are effectively suspended between paths of particles with higher energy and follow paths described by Brownian motion until target nuclei uptake them to produce cosmogenic nuclides. Contrary to muonic production, thermal neutron absorption occurs more readily at and proximal to the Earth’s surface (Kurz,
1986b; Fabryka-Martin, 1988). One or more of these production mechanisms may contribute to the production rate of an individual nuclide.
Accuracy of TCN dating is dependent, in part, upon appropriate estimations of regional production rates. Rates of nuclide production are affected by temporal variability in the strength of the geomagnetic field as well as solar modulation of cosmic-ray flux (Lal & Peters, 1967;
Kurz et al., 1990; Lifton et al., 2008). A weaker magnetic field increases the production of cosmogenic nuclides due to greater penetration of cosmic rays into the atmosphere. Increased solar activity decreases the cosmic ray flux to Earth’s atmosphere, particularly at high latitude and altitude. Production rates are additionally affected by site-specific variables including: latitude, atmospheric pressure, and compositional properties of target material (Lal, 1988a; Lal
1991). To calculate an age constraint on an exposure event as a function of time in years (a), an estimated local production rate must be used (reported in units of atoms g-1 a-1) which accounts for the processes and variability described above.
22 Numerous scaling schemes have been developed to incorporate time and spatial variability of TCN production (Lal, 1991; Stone, 2000; Dunai, 2000; Dunai, 2001a; Desilets and
Zreda, 2003; Pigati and Lifton, 2004; Farber et al., 2005; Lifton et al., 2005; Lifton et al., 2014).
The scaling scheme based on observations of Lal & Stone, abbreviated “St” in the review by
Balco, 2008 is most prevalent in the literature as this was the first and only scaling scheme for over a decade. This scheme scaled as a function of latitude and atmospheric pressure (Stone,
2000). Production rates of nuclides were assumed constant over time; thus, temporal variability of the geomagnetic field was not incorporated into this scaling scheme. The scaling schemes of
Dunai 2000 “Du” and Desilets et al., 2006 “De” incorporate magnetic field modulation of production. Due to additional data availability of cosmic ray flux near the surface from additional calibration locations, the elevation-dependence of production was modified to fit modern observation data more closely. The Lifton et al., 2005 “Li” model added temporal solar variability. The “Lm” model was produced to retain altitude scaling from Lal 1991 with the addition of a paleomagnetic correction by Nishiizumi et al., 1989. Fully described in Lifton et al.,
2014, the Lifton-Sato-Dunai scaling model incorporated analytical parameterizations, allowing for calculation of cosmic ray spectra based on first principals of physics rather than empirical fit models. Temporal variability in production rates from solar and geomagnetic modulation is also retained in this new scaling framework. Two versions of the Lifton-Sato-Dunai scaling model were used as part of the CRONUS calculator. The “LSD” denoted “SF” in the calculator is flux- based and the “LSDn” denoted “SA” model contains nuclide-dependent and energy-dependent terms resulting in a nuclide-dependent scaling factor.
Accuracy of empirical measurements used to calibrate scaling schemes as well as increased geographical distribution of calibration data sets will inevitably lead to improvements in scaling methods. Future discrepancies in exposure ages calculated from the same nuclide
23 concentrations are expected due to ongoing advancements of age calculators and scaling schemes; therefore, detailed observations and measurements necessary to recalculate results are reported as legacy data to ensure future applicability of this data set.
6.1.2. Limitations of Surface Exposure Dating
Cosmogenic exposure dating methods are routinely used to constrain minimum or maximum ages of deglaciation by dating landforms deposited during or after ice retreat.
Exposure ages from glacial-geomorphic landforms such as moraine boulders and striated bedrock are assumed to equate to the duration of exposure; therefore, age of moraine or time of deglaciation, respectively (Putkonen and Swanson, 2003; Fabel et al., 2012; Stone and
Ballantyne, 2005; Small et al., 2016). TCN exposure ages are affected by many sources of uncertainty including 2-5% from analytical errors and accelerator mass spectrometry (AMS) and
10-20% from production rate uncertainty associated with the number and distribution of calibration sites and equipment used to measure production variation with latitude and altitude
(Gosse and Phillips, 2001). Finally, there is geological uncertainty, which is difficult if not impossible to quantify. Geological uncertainty of chronological methods was described by Small et al., 2016 as falling into two categories: (1) factors affecting the net concentration before sampling and (2) assumptions involving equivalence, or strength of contemporaneous association between events of interest and the material used for dating. Both types of geological sources of uncertainty are discussed in this section, action to minimize both types of uncertainty are explained in sample methodology section (7.2.2.).
Erosion and weathering processes result in nuclide loss, yielding a younger exposure age than predicted by rates of production. Nuclide loss from partial exposure due to topographic, dust, ash, or snow cover are examples of shielding, an example of the first type of geological
24 uncertainty. Partial exposure of samples due to boulder exhumation or post-depositional toppling are additional ways to calculate erroneously young ages of a target event. The latter examples of partial exposure have the additional component of the second type of geological uncertainty.
Samples from glacially-transported boulders are collected under the assumption that the material had been continuously irradiated since deglaciation with no periods of prior exposure.
Previous events of exposure accumulate TCN concentration that, if not eliminated by erosion or decay before a subsequent exposure event, yields an older exposure age than expected for a duration of irradiation. This excess nuclide inventory from a previous process with respect to the target exposure event is referred to as inheritance; thus, inheritance results in an erroneously old age for the landform or event being dated. Deposition of rockfall- or landslide- sourced boulders from valley walls or topography upstream of the glacial catchment are examples of processes that contribute to inheritance. Inheritance is most common in
allochthonous deposits
with sediment that may
have more complex
exposure histories from
reworking and in
landscapes lacking
sufficient erosion to reset
the sample.
Surface exposure ages
Figure 6: From J Heyman et al., 2011, this figure illustrates how cosmogenic results can vary based of boulders on moraines on surface processes. A: Ideal case with complete shielding, no prior exposure, and no post- depositional modifications which would affect ages. B: prior exposure affecting the exposure ages of boulders making them appear older than age of deglaciation. C: Shielding and erosional and striated bedrock processes make ages appear younger than the age of deglaciation. calculated from a site-
25 scaled production rate, sample properties, and nuclide concentrations theoretically represent a minimum age estimate of deposition/abandonment. For a sample set yielding scatter that is not attributed to inheritance, the maximum apparent exposure age indicates a minimum deglaciation age (Heyman et al., 2011). Geologic uncertainty is incredibly difficult to quantify as the processes described above as issues of equivalence and exposure histories prior to sampling are impossible to confirm; therefore, geologic context must be incorporated into the interpretation of calculated apparent exposure ages (Heyman et al., 2011; Douglass et al., 2006). Figure 6, modified from Heyman et al., 2011, illustrates this concept of how processes including inheritance and partial exposure affect apparent exposure ages.
6.2. This Study
Mafic igneous lithologies from the Cordon de Puntas Negras including andesite, basaltic andesite, and dacite, render 36Cl the most suitable nuclide for this study. Concentrations of in situ cosmogenic 36Cl were measured from moraine boulders to calculate surface exposure ages, or ages of deposition/ice abandonment of glacial landforms. The use of 36Cl accumulation for exposure dating was first suggested over 60 years ago (Davis and Schaeffer, 1955) as the half- life (308,000 years) makes this nuclide of particular interest for studying the Pleistocene epoch.
As such, 36Cl or the stable cosmogenic isotope 3He are often utilized to analyze glacial landforms lacking sufficient crystalline quartz for 10Be dating including mafic volcanics and carbonate-rich lithologies. After advancement in analytical techniques of accelerator mass spectrometry (AMS) measurements, the method was further characterized and reviewed by Phillips et al., 1986 who stressed the hydrophilic nature of chlorine and the use for this mobility in separating in situ from meteoric components.
26 At the surface, 36Cl primarily forms by spallation of 39K and 40Ca and epithermal and thermal-neutron absorption of 35Cl. Spallation production from Ti and Fe is considered insignificant at the surface due to atmospheric attenuation of secondary particles (Zreda et al.,
1991). In the subsurface, muon capture by 40Ca is an important production mechanism due to muon penetration beyond depths of spallation-production attenuation (1-2 meters) (Phillips et al.,
1986; Zreda et al., 1991; Gosse and Phillips, 2001). Whole-rock 36Cl analysis was used for this study, although plagioclase mineral separate analyses may be performed on a subset of samples in the future.
Due to the variety of production pathways, rates of 36Cl production are the most difficult to constrain of commonly measured cosmogenic nuclides (Marrero et al., 2016; Phillips et al.,
2016). One of the most prominent sources of computational error in 36Cl exposure age calculations is the accuracy of the production rate used to convert net concentration measurements into years of irradiation, which is inherently tied to availability of calibration data.
Data from calibration localities exhibit evidence to support the following criteria: minimal erosion, minimal to no shielding or partial exposure, continuous exposure to cosmic rays, and an independent age constraint such as radiocarbon ages. The 36Cl production rates used by the
CRONUS-Earth calculator were calibrated using three sites to constrain Ca and K spallation.
Together the sites yielded a rate of 56.0 ± 4.1 atoms of 36Cl (g Ca)-1 yr-1 and 155 ± 11 atoms of
36Cl (g K)-1 yr-1, scaled using the Lifton-Sato-Dunai “SA” model (Marrero et al., 2016; Lifton et al., 2014). Ages and erosion rates acquired from moraine samples surrounding Quelccaya ice cap in Huancané, Peru (13° S latitude, 4850m elevation) were among the three spallation production calibration sites (Mercer and Palacior, 1977; Kelly et al., 2015, 2012; Phillips et al., 2016;
Marrero et al., 2016). As such, the CRONUS-Earth calculator and the “SA” scaling model were the most appropriate to use for exposure age calculation for this study.
27
6.2.1. Sample locations
Eight glacially-transported andesite boulders were sampled from lateral moraines along the walls of the formerly-glaciated valley at the Tatio site in 2014. During our field season in
2016, twenty-one samples were collected from large glacial erratic boulders at the SPN site. Of the 21 samples from SPN, 2 were collected from the hummocky, degraded terminal moraine, glacial stage 1, GS1). The remaining samples were collected from two additional glacial stages
(GS2b and GS3) up the valley with higher relief relative to GS1.
6.2.2. Sampling Methodology
Several samples were collected from the same moraine ridge, where possible, due to presence of stable boulders acceptable for sampling. Samples were taken via hammer and chisel from the largest boulders in a stable position at the top of moraine ridges, where possible, to minimize the risk of interrupted exposure from till, dust, snow shielding, or toppling. Boulders were chosen for sampling if there was a flat surface with evidence of varnish and ventifaction, indicating little erosion since deglaciation and a stable position since deposition. Field notes were taken for each sample including the dimensions of the boulder, the precise lat/long location, lithology, and sample thickness.
28 Figure 7: Images of our lab group collecting samples from the upper centimeters of Andesite boulders.
6.3. Bulk Rock 36Cl Sample Preparation
Chemical extraction of 36Cl from bulk-rock dissolutions was performed at the University of Cincinnati (UC) Cosmogenic Chlorine Laboratory and purified as AgCl using the procedure given in Stone et al., 1996 with revision per preparatory recommendations from PRIME lab facility at Purdue University (Radler, personal communication). 36Cl was measured with accelerator mass spectrometry (AMS) at Purdue Rare Isotope Measurement (PRIME) Lab,
Purdue University with the addition of two chemical blanks for each unique combination of preparatory reagents and a direct precipitation blank for each carrier solution to correct for reagent contamination. Geochemical analysis of bulk and trace elemental composition of samples was performed at Acme Analytical Laboratories Ltd., now operating as Bureau Veritas
29 Mineral Laboratories in Vancouver, BC, Canada. Exposure ages were calculated using the
Cosmic Ray-Produced Nuclide Systematics on Earth Project (CRONUS-Earth Project) calculator.
6.3.1. Physical Preparation
To acquire the recommended 250 grams of fine grains (<250 μm), samples were first crushed using a Sturtevant 2x6 fixed-plate jaw crusher. Second, each sample was pulverized in a
Bico UD direct drive disc mill, sieved, and sorted by grain sizes (>500 μm, 250-500 μm, and
<250 μm). The jaw crusher, disc mill, and sieves were carefully and thoroughly cleaned of grains and dust between each use. The jaw crusher and disc mill were vacuumed and cleaned with a wire brush and compressed air. Sieves were cleaned using brushes and grain picks. They were also wiped free of dust between each sample. A small hand sample was reserved to measure density of each sample using mass and volumetric water displacement and to archive for future reference.
6.3.2. Chemical Preparation
Two sets of approximately 10 g of <250 μm grains were reserved for whole-rock geochemical analysis. One aliquot was set aside before chemical leaching (described below) and one after. Geochemical analysis of bulk and trace elemental sample composition was performed on the post-leach aliquot at Acme Analytical Laboratories Ltd., now operating as Bureau Veritas
Mineral Laboratories in Vancouver, BC, Canada.
Approximately 100 g of the fine fraction of grains (<250 μm) from each crushed sample were initially leached for over 12 hours using 70mL of 18 MΩ water and 7-8 mL of TraceMetal
Grade (TMG) nitric acid (HNO3) solution in 250 mL nalgene bottles. This process removed
30 potential contamination from atmospheric, or meteoric 36Cl. Leaching solution was decanted, and samples were rinsed with 18 MΩ water to neutralize residual HNO3 and to separate and discard fine particles, then dried overnight.
6.3.3. Carrier/Isotope dilution
After measuring 10 g of post-leached sample and reserving it for geochemical analysis, as mentioned above, ~30g of each sample (exact mass ±0.0001g recorded on lab bench sheets and in sample spreadsheet for archival and working use) was transferred to a 500mL bottle for dissolution and the addition of ~1g of a 35Cl-enriched carrier solution. Adding a spike enriched in isotopically-stable Cl, or the isotope dilution method, has been widely adopted due to the following benefits: (1) ability to measure 36Cl/Cl and total Cl on the same AMS target (2) increase measurement precision for samples with low Cl concentration (3) reduce minimum sample size (4) reduce sensitivity of Cl concentrations to contamination from reagents (Desilets et al., 2006; Sharma et al., 2000; Elmore et al., 1997). Carrier solutions were prepared at the
University of Cincinnati with ICON Isotope Industries 35Cl standard using the approximate ratio of 1.0000 (mg/g) 35Cl. To calculate the amount (mg) of carrier present in each sample, exact concentration of the carrier solution and weight of carrier added to each sample was recorded to ±0.0001 g on bench sheets.
6.3.4. Dissolution & procedural blank preparation
Samples were chemically dissolved in a solution containing 150g of 18 MΩ water,
~150g, or 500% of sample mass, of low chloride 40% hydrofluoric acid (HF) solution, and ~45g, or 150% of sample mass, concentrated TMG 70% HNO3 solution. At this stage, each stock solution and reagent were given a unique identifier. This name was recorded alongside other measurements on the sample bench sheets, which was used to determine the minimum number of
31 procedural blanks to prepare. Each unique combination of HF, HNO3, and carrier spike used on the aliquot of samples was mixed in duplicate pairs and as a procedural blank. Procedural blanks were processed using the same procedure as the bulk-rock samples. Eight procedural blanks were prepared beginning with the addition of ~1 g of carrier spike solution and the solutions for digestion of the rock samples. The blanks serve the purpose of developing a correction factor for chlorine contamination in the reagents used to prepare the sample for AMS measurement.
Samples reacted with the digestion solution to form a gaseous product and fluoride gel; therefore, bottles were vented in a fume hood for 12 hours and occasionally perturbed to ensure the entire sample was exposed to the solution. Samples were subsequently capped tightly and transferred to a heated rolling apparatus for constant agitation at 60°C. The heat ensured complete dissolution and prevented the grains from being coated in fluoride. Time required for dissolution ranged from 2-4 days.
Dissolved samples were transferred from their 500 mL bottle to two 250 mL Nalgene bottles for compatibility with Beckman Coulter Allegra X-14 centrifuge. They were then centrifuged for 30 minutes at 2600 rpm. This step condensed and separated the fluoride solid from solution and enabled sample solution to be decanted into a clean and leached 500 mL bottle in preparation for initial precipitation of silver chloride (AgCl).
6.3.5. Precipitation
AgCl was precipitated from solution by adding 20 drops of 0.15M HNO3 followed by 24-
48 hours of refrigeration while the crystalline solid settled. A direct precipitation blank for the carrier solution used to spike samples was prepared in a 50 mL centrifuge tube. At this stage in the procedure, samples and both varieties of blanks went through the same procedural steps.
Supernatant fluid was carefully decanted into Ag/HF/HNO3 waste without disturbing the solid at the bottom of the bottle. The solid AgCl product was transferred to clean and leached 50
32 mL centrifuge tubes along with a rinse of 18 MΩ water to transfer any residual solids. Samples were centrifuged at 3600 rpm for 10 minutes to compact the solid for rinsing with 10 mL of 18
MΩ using a VWR Analog Vortex Mixer. The sample was then compacted once more via centrifuge and rinse water was decanted into Ag/HF/HNO3 waste.
6.3.6. Anion exchange chromatography
Crude samples were purified for AMS measurement using anion exchange chromatography. One positively charged, stationary phase resin column per sample was suspended over a waste container and conditioned with 10 mL of 4.0 M HNO3. After the HNO3 drained to the resin bed, the pH was neutralized to ~7 using 18 MΩ water (60-80 mL).
To prepare samples for ion exchange columns, 20 drops of ~30% TraceMetal Grade ammonium hydroxide (NH4OH) and enough 18 MΩ water to bring the total volume to 10 mL was added to each centrifuge tube to dissolve the AgCl solid. Samples were re-capped, the pellet of crude AgCl was dislodged from the bottom of the sample and was fully dissolved by mixing with the VWR Analog Vortex. The silver chloride samples, now in solution, were transferred to their corresponding pre-conditioned anion exchange columns. Each centrifuge tube was rinsed with an additional 10 mL of 0.1 M NH4OH solution, vortexed, and decanted into the respective column to transfer residual sample. The columns were drained to the surface of the resin bed.
Next, 10 mL of 0.05 M HNO3 was eluted through each column.
Clean pre-leached 50 mL centrifuge tubes were labeled and prepped with 2 mL of silver nitrate solution in each. After the columns were drained to the resin bed, the new batch of centrifuge tubes replaced the waste bottles below the suspended columns. Chlorine from each column was eluted into the centrifuge tubes by adding 20 mL of 1.5 M HNO3. Isolated AgCl began to precipitate in a white turbidity in the centrifuge tubes. Finally, 20 drops of TraceMetal grade HNO3 was added to each sample. Samples were capped, vortexed for 10 seconds, and
33 placed in a refrigerator for 12-18 hours to allow the AgCl to precipitate from solution and accumulate.
6.3.7. Final Sample Preparation
After precipitation of AgCl was complete, the samples were centrifuged for 10 minutes at
3600 rpm to discard supernatant fluid. Samples were rinsed using 10 mL of 18 MΩ water and vortex. AgCl was re-centrifuged following the rinse in preparation for transfer to 1.7 mL microcentrifuge tubes for final consolidation of the solid sample pellet. Pre-leached microcentrifuge tubes were labeled and weighed, mass recorded in grams. Each microcentrifuge tube received 1 mL of 18 MΩ water as a medium for sample transfer from the 50 mL centrifuge tube to the 1.7 mL microcentrifuge tube using a separate 2 mL disposable pipette per sample.
The 50 mL centrifuge tubes were rinsed, and the rinse water was transferred to the corresponding microcentrifuge tube. The solid sample, now in the microcentrifuge tubes, was compacted into a pellet by centrifuging for 10 minutes at 3600 rpm. Supernatant liquid was drawn up with a new pipette, one for each sample to prevent contamination, and samples were dried in an oven for 12-
24 hours. After the samples were completely dry and had cooled to room temperature, the final weight of the microcentrifuge tube plus sample was weighed and recorded on bench sheets.
6.3.8. Target Preparation for AMS Measurement
Samples were loaded into copper cathodes with a lens of silver bromide (AgBr) for AMS measurement at Purdue Rare Isotope Measurement (PRIME) lab. A loading workspace was cleaned by kimwipe and covered with an 11x7 sheet of cardstock. A 6x6 cm metal plate with a drilled center for holding copper cathodes in place was lined with kimwipes (set on top of one and an additional covered the plate itself) to be replaced between each sample. Hand tools such
34 as microspatulas, forceps, and one 11/64” diameter high speed steel drill blank for each sample were ultrasonically cleaned, sanded, and sterilized with acetone to prevent contamination. The
Cu cathodes were heated in glass vials via hot plate for 15 minutes. Heating the cathodes before and after they have been loaded removes any water moisture, makes the AgBr less brittle, and reduces 36S. After the cathode was initially heated, it was placed into the slot of the metal plate standing freely with the AgBr lens facing upward. A drill blank pin was used to press a small dimple in the center of the AgCl target to focus the sample in the center of the cathode. Each sample microcentrifuge tube was opened and passed through an antistatic machine to prevent the sample pellet from clinging to the sides. The sample was carefully placed in the center of the cathode and pressed into the AgBr using the drill blank pin and a ball hammer to secure the loose material. The work surface was cleaned between each sample to reduce the risk of cross- contamination. After the cathodes were loaded, they were placed back onto a hot plate for an additional 15 minutes. They were then capped, and sample information was recorded. Sample
ID, Cu cathode holder number, and PRIME lab sample ID were recorded on a sample loading spreadsheet.
6.4. TCN Age Calculation
PRIME Lab facility directly measures 36Cl/37Cl and 35Cl/37Cl, which are used to calculate the total atoms of rock chlorine using equation (1):