Quaternary Geochronology 28 (2015) 29e39
Contents lists available at ScienceDirect
Quaternary Geochronology
journal homepage: www.elsevier.com/locate/quageo
Research paper
Geochemical analyses of air from an ancient debris-covered glacier,
Antarctica
Audrey M. Yau a, *, Michael L. Bender a, David R. Marchant b, Sean L. Mackay b
a
b
Department of Geosciences, Princeton University, Princeton, NJ 08540, USA
Department of Earth & Environment, Boston University, Boston, MA 02215, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 26 April 2014
Received in revised form
13 March 2015
Accepted 14 March 2015
Available online 17 March 2015
We examined air trapped in ancient ice from three shallow cores (<35 m deep) recovered from stagnant
portions of the Mullins glacier, an 8 km long debris-covered alpine glacier in the McMurdo Dry Valleys
that is overlain by several in-situ volcanic ash-fall deposits. Previously reported 40Ar/39Ar dates on ashfall in the vicinity of the core sites average 4.0 Ma, and underlying ice is presumably as old in some areas.
We analyzed the elemental and isotopic composition of O2, N2, and Ar and total air content of the glacial
ice. We also dated the trapped air directly to an uncertainty of ±220 kyr (1s) by measuring its 40Ar/36Ar
and 38Ar/36Ar ratios. Our results suggest that the air analyzed is likely a mixture of ancient atmosphere
trapped at the time of ice formation and more recent air introduced via cracks in the ice that penetrate to
at least 33 m. The isotopic signatures of gases have been complicated by gas loss, as well as a mixture of
thermal and gravitational fractionation. The oldest age estimated for the trapped air dates to 1.6 Ma,
indicating that the original air is at least as old as 1.6 ± 0.2 Ma. A convergence to older ice ages with
increasing depth in the deepest core analyzed (33 m) hints at the possibility that pristine air might be
recovered at greater depths. Minor interstitial debris present in the glacial ice (<1%), along with
geochemical evidence for in-situ microbial respiration, prohibit direct analysis of CO2. We measured the
triple isotopic composition of O2 as a proxy for CO2 and infer that, in the air represented in our ice
samples, CO2 concentrations are within the range observed over the last 800 ka.
© 2015 Elsevier B.V. All rights reserved.
Keywords:
Ice cores
Ar-dating
Stable isotopes
Antarctica
Dry Valleys
1. Introduction
The debris-covered glacier in Mullins Valley (hereafter informally termed Mullins glacier) is located in the coldest and driest
region of the McMurdo Dry Valleys, Transantarctic Mountains
(Fig. 1). Minimum ages for stagnant portions of Mullins glacier have
been constrained by morphological analysis and 40Ar/39Ar dating of
overlying volcanic ash fall, with tephra ages increasing from
~4.0 Ma in the central part of the glacier up to ~8.0 Ma at its distal
end (Marchant et al., 2007; Kowalewski et al., 2011). Given its old
age, sublimation rates must be exceedingly slow (melting is limited
given the low atmospheric temperatures, see below), and calculated rates of ice loss via sublimation inferred from cosmogenic
nuclide exposure-age studies have been used to both support
(Schafer et al., 2000; Marchant et al., 2002) and refute (Ng et al.,
2005) old ice ages. More recently, however, results from simple
* Corresponding author.
E-mail address: [email protected] (A.M. Yau).
http://dx.doi.org/10.1016/j.quageo.2015.03.008
1871-1014/© 2015 Elsevier B.V. All rights reserved.
1-D vapor diffusion models, using local meteorological data as
input, suggest that the buried surface of Mullins glacier near the
study site is in near-equilibrium condition, with the annual rate of
ice-surface lowering via sublimation being <0.16 mm a1
(Schorghofer, 2005; Kowalewski et al., 2011). Robust 2D models,
which more faithfully incorporate changes in topography, debris
texture, and surface conditions suggest that ice loss may be as low
as 0.022 mm a1, and imply that ice could survive indefinitely with
very modest changes in atmospheric conditions, e.g., a ~1.9 C
decrease in mean annual air temperature or an increase in relative
humidity of ~12% (Kowalewski et al., 2012). The objectives of this
study are to determine whether air bubbles in Mullins glacier have
preserved pristine, ancient atmosphere or were contaminated with
younger air, and more generally, to examine the geochemistry of
the trapped air in order to understand the limits of the paleoclimate
information it contains.
We focus on the elemental and isotopic composition of O2, N2,
and Ar, as well as the total air content (TAC) of the trapped gases to
determine the likely history of the preserved atmosphere. Several
properties of the air are of particular interest: d15N, d38Ar/36Ar, TAC,
30
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
Fig. 1. Mullins glacier field site showing locations of shallow ice cores: Core 1 (160.53720 E, 77.87680 S), Core 2 (160.53786, 77.87679), Core 3 (160.53572, 77.87799). Inset map
shows the location of Mullins Valley (arrowed) within the larger context of the McMurdo Dry Valleys. Basemap is a hillshade generated from a high resolution airborne LiDAR digital
elevation model (DEM), collected as a joint effort by US National Science Foundation (NSF)/NASA/US Geological Survey (USGS) (Schenk et al., 2004).
dAr/N2, dO2/N2, dO2/Ar, and d18O of O2. These particular geochemical signatures provide information as to whether the trapped air
has been subject to fractionation by gravitational and thermal
processes (d15N and d38Ar/36Ar) (Craig et al., 1988; Severinghaus
et al., 1996), whether the ice has undergone partial melting or gas
loss (dAr/N2), whether air is trapped by typical firn densification
processes (TAC) (Herron and Langway, 1987; Martinerie et al.,
1992), and whether the trapped air has been subject to microbial
respiration (dO2/N2, dO2/Ar, and d18O of O2). With the context
provided by these geochemical analyses, we can better understand
the processes preserving ancient ice and air in Mullins glacier.
We also date the trapped air by measuring the paleoatmospheric 40Ar/36Ar ratio (40Ar/36Aratm), which has been increasing in
the atmosphere monotonically with time (Bender et al., 2008).
Concentrations of 36Ar and 38Ar have been essentially constant
throughout geologic time, but 40Ar has been slowly and steadily
increasing through Earth's history as a result of the decay of 40K.
The ratio of 40Ar/36Ar has increased in the last million years at a rate
of 0.066 ± 0.007‰/Ma (Bender et al., 2008). We precisely measure
the ratio of 40Ar/36Ar of the trapped air and determine the age of the
air based on this rate of 40Ar/36Ar increase. Samples with intact
trapped air would give an Ar isotope age consistent with independent dating constraints (~4 Ma).
1.1. Geographic context
Mullins glacier is a cold-based alpine glacier occupying the floor
of Mullins Valley and the upper portion of the adjacent Beacon
Valley (Fig. 1). Its ice is derived from a combination of direct precipitation at the valley headwall and from snow previously
deposited on the surface of the East Antarctic Ice Sheet and blown
to Mullins Valley by strong katabatic winds. Mean annual air
temperature and relative humidity at the study site
average ~22 C and ~49%, respectively, and annual precipitation is
limited to a few centimeters per year (Marchant and Head, 2007;
Fountain et al., 2010). Under these conditions, ice loss is achieved
by sublimation, rather than by melting (Kowalewski et al., 2011). As
noted in Fig. 1, the vast majority of Mullins glacier is covered with a
thin layer of debris, <70 cm thick, sourced from rock fall off cliffs of
Ferrar Dolerite and Beacon sandstone at the valley head
(Kowalewski et al., 2011; Mackay et al., 2014). Both the glacier and
overlying debris are incised by thermal-contraction cracks
(Kowalewski et al., 2011), which in plan-view create a series of
interlocking polygons that are up to 20 m in diameter and show a
vertical relief of 1e2 m between the polygon troughs and polygon
centers (Levy et al., 2006). The thermal-contraction cracks extend
an unknown distance down into the glacier. Typically, they fill with
sand, forming veins up to several cm wide that taper at depth to just
over a few mm wide (Kowalewski et al., 2011). Relict sand veins,
e.g., those no longer associated with active thermal cracking and
polygon formation, appear irregularly across the ice surface and at
depth (Kowalewski et al., 2011).
Shallow cores from three sites on the Mullins glacier, collected
during an expedition over the 2009 austral summer, are the focus of
this study (e.g., Mackay et al., 2014) (Fig. 1). Cores 1 and 2 were
collected in the middle portion of the Mullins glacier. They are
situated about 20 m apart, at the center of adjacent polygons, and
reach depths of 16.9 m and 13.8 m, respectively. Radioisotopic dates
on ash fall deposits overlying Mullins glacier near these core sites (6
deposits within 0.5 km) average 4.0 Ma, giving a minimum age for
the underlying ice at these sites (Marchant et al., 2007). Core site 3,
also drilled at the center of a polygon, is situated ~500 m up-glacier
from Core sites 1 and 2 and reached a depth of 33.6 m. Ash fall does
not occur near Core site 3, but its up-glacier location would suggest
an ice age somewhat younger than that at Core sites 1 and 2. Ice
recovery in all three cores was in excess of 95%. Visually, all three
cores show evidence for brittle fracture and sand veins (Fig. 2). The
sand veins are most common in near-surface ice, but occur down to
the base of Cores 1 and 2, and to at least 25 m depth in Core 3. Clean
fractures, absent of sand, also truncate the cores; these fractures
may be related to brittle deformation from uneven ice flow, thermal
contraction, or some combination of the two (Fig. 2). Finally, it is
important to emphasize that Mullins glacier is not an ice sheet, but
an alpine glacier, continuously eroded from the surface by slow
sublimation. Its age should generally increase with depth and distance from the headwall, but there may be age reversals associated
with folding during flow.
1.2. Fractionation effects
The elemental and isotopic composition of gases in ice is biased
from the atmospheric composition by three different processes.
The first is the equilibrium effect of gravitational fractionation, the
progressive enrichment of heavier gases and isotopes with depth in
the firn (Craig et al., 1988; Schwander, 1989). The enrichment of the
heavy species relative to the light species is derived from the
barometric equation:
d ¼ ½expðDmgz=RTÞ 1 * 1000‰
(1)
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
31
Fig. 2. Core characteristics for Core 2 (AeC) and Core 3 (DeF). Panel A shows sand-filled cracks at ~1.5 m depth in Core 2. Scattered rock fragments and sand grains (dolerite) are
seen in core sections in B and D; relatively cleaner ice occurs at greater depths (C, E, F). Cores are ~3 inches in diameter and ~1 foot long.
d ¼ fractional deviation of the sample to reference ratio
Dm ¼ mass difference
g ¼ constant for gravitational acceleration
z ¼ depth below the surface
R ¼ gas constant
T ¼ temperature
The second is the equilibrium effect of thermal fractionation, in
which heavy elements and isotopes are enriched in the cold region
of a porous medium (such as firn) in the presence of a temperature
gradient (Grew and Ibbs, 1952). The magnitude of thermal fractionation is described by Chapman and Cowling (1970) as:
d ¼ ð½To =Ta 1Þ * 1000‰
(2)
gases and/or their varying molecular sizes. The effect of gas loss has
been qualitatively observed, though poorly quantified. We do not
use a gas loss correction term for 15N or 14N, as there does not
appear to be a fractionation of these isotopes associated with gas
loss (Severinghaus et al., 2003; Huber et al., 2006). And we do not
correct 18O, 17O, and 16O, for gas loss because the magnitude of this
fractionation has not been quantified, but is likely to be small. We
do make a correction, based on dAr/N2, for the effect of Ar loss on
d40Ar/36Ar and d38Ar/36Ar as per Kobashi et al. (2010). This correction, though small, is marginally significant for Ar isotope data.
The contribution of these three processes (gravitational, thermal, and gas loss fractionation) to the isotopic and elemental
composition of the trapped air provides a history for the trapping of
this ancient air. These processes are accounted for in the following
discussion of the geochemical data.
d ¼ fractional deviation of the sample to reference isotopic ratio
To ¼ temperature of the reference point
T ¼ temperature of the sample point
a ¼ thermal diffusion factor
Temperature gradients introducing thermal fractionation are
associated with seasonal and diurnal fluctuations in air temperature rather than ice temperature, which changes far more slowly.
Surface temperature changes lead to changes in the isotopic
composition of firn air with depth as subsurface gases try to attain
the equilibrium fractionation. The diffusivity of gases in air is on
order 1 m2 day1, so gases in the firn will respond to diurnal
temperature variations down to ~1 m depth, and the annual cycle
down to ~20 m depth (Schwander et al., 1988). Gases that are isolated in the original firn of the Mullins glacier can be thermally
fractionated if close-off is shallower than 20 m depth. The same is
true for gases trapped when cracks in ice seal.
The third process affecting the elemental and isotopic composition of trapped air is the fractionation of air during gas loss
through cracks in the ice, where the fractionation of different
species is dependent primarily on the different diffusion rates of
2. Methods
2.1. Ar analytical methods
The Ar isotope analysis method is similar to that described by
Bender et al. (2008), with a few simplifications. A vacuum system is
used to extract Ar from air samples in order to limit isobaric interferences, improve linearity, and amplify the Ar signal during
spectral analysis. 700e1300 g of ice were cut depending on the total
air content of the sample, and the outer 2e5 mm shaved off to
remove core contamination associated with storage. We use an
adapted extraction and equilibration technique based on Emerson
et al. (1995). Each sample of ice was placed in a sealed, evacuated
2.5 L glass bottle and allowed to melt. According to Henry's Law, we
estimate that >98% of the sample air is in the headspace of the
sample bottle, and thus the fractionation of gases between the
headspace and liquid is negligible. In addition, to further reduce any
possible fractionation of sample air, the melt water and bottle
headspace were actively equilibrated for 4 h at room temperature.
Following equilibration, most of the melt water was drained
32
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
(Emerson et al., 1995). In cases where high debris content interfered
with properly draining the melt water, the headspace gases were
expanded into a 2.5 L evacuated glass flask at room temperature.
The air was then passed through a liquid nitrogen trap that
captured residual water and CO2. The remaining O2, N2, and noble
gases were condensed onto a mole sieve trap submerged in liquid
nitrogen. Following this transfer, the air was exposed to the first
getter (30 sections of SAES ST 101 getter at ~900 C) for 15 min. The
remaining sample air was condensed on a second mole sieve trap
submerged in liquid nitrogen. Following the second transfer, the air
was exposed to a second getter (SAES CapaciTorr) for 25 min at 7 V.
After these two successive gettering steps, which remove >99.999%
of non-noble gases, the remaining Ar was frozen out of the vacuum
line into a 12 cc stainless steel tube submerged in liquid helium. The
tube was warmed to room temperature, and then analyzed on a
Finnigan MAT 252 with collectors (Faraday cups) specific to the 3 Ar
isotopes. Each sample was analyzed for 6 blocks of 24 cycles to
reduce standard error. For modern air (which serves as our standard), the standard deviations of d40Ar/38Ar, d38Ar/36Ar, and
d40Ar/36Ar are ±0.027‰, ±0.027‰, and ±0.053‰.
For dating purposes, the term of merit is d40Ar/36Aratm,
defined as
d40 Ar=36 Aratm ¼ d40 Ar=36 Ar 2:002 * d38 Ar=36 Ar
(3)
The coefficient (2.002) corrects for gravitational fractionation,
the main cause of isotope fractionation in deep ice core samples. In
the discussion below, we show that our samples are also influenced
by thermal and gas loss fractionation. These processes would have a
slightly different coefficient but the overall fractionations are small
(see Section 3.5). Data included in the fractionation analyses are
limited to those with sample sizes between 1.2 and 3.2 V at a
resistance of 300 MU for 40Ar. Air standards were run optimally at
3 V. The standard deviation in d40Ar/36Aratm values of modern air
collected from the roof of the Princeton University Geosciences
building in New Jersey was ±0.015‰, corresponding to an age
uncertainty of ±220 kyr (1s). Five Holocene-aged ice samples
(Vostok borehole ice between 118 and 180 m depth) were also
analyzed using the same extraction technique applied to samples,
yielding an age of 100 ± 200 ka.
2.2. Procedure for the analysis of O2/N2/Ar, d15N of N2, d18O of O2,
and total air content
The d15N, dAr/N2, dO2/N2, dO2/Ar, and d18O of O2 of the trapped
air were assessed by extracting the trapped gases under vacuum in
similar fashion to the Ar extraction. In these extractions, approximately 50 g of ice were used per analysis, and the equilibrating time
of the headspace and melt water was 1 h. The sample was connected to a vacuum line via a liquid nitrogen trap, and then was
Fig. 3. Geochemical data for Cores 1, 2, and 3 with depth. In panel A, both negative and positive d15N2 values indicate thermal fractionation of gases, rather than gravitational
fractionation alone. In panel B, low total air content indicates cracking of the ice, and/or abnormal firn densification. Increasing TAC with depth suggests that pristine air may be
preserved at greater depths. In panel C, low dAr/N2 indicates gas loss from the trapped air bubbles and limited partial melting of the ice. In panels DeF, depleted dO2/N2 and dO2/Ar
values and enriched d18Oatm data indicate microbial respiration of the trapped air.
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
frozen into a 12 cc stainless steel sample tube submerged in liquid
helium. It was then warmed to room temperature and analyzed on
a Finnigan Delta Plus XP mass spectrometer with Faraday collectors
for masses 28, 29, 32, 34, and 40. Standard deviations of modern air
standards, processed through the vacuum line as for ice core
samples, for d15N is ±0.014‰, for dAr/N2 is ±0.56‰, for dO2/N2 is
±0.78‰, for dO2/Ar is ±0.36‰, and for d18O of O2 is ±0.03‰. The
paleoatmospheric d18O, d18Oatm, is equal to d18O corrected for
gravitational fractionation: d18Oatm ¼ d18O e 2.01 * d15N with a
standard deviation of ±0.03‰. A correction for thermal fractionation is not applied as the total effect is negligible and has no
impact on the subsequent interpretation. Total air content (TAC)
was estimated by comparing ion currents of sample air with ion
currents of volume-calibrated vials of air.
2.3.
17
D analyses procedure
The triple isotope composition of oxygen was determined by
processing air extracted from ~150 g of ice through an automated
vacuum line connected to a gas chromatograph as per Blunier et al.
(2002). Isotope compositions were measured on a Finnigan MAT
252 mass spectrometer. The analytical uncertainty based on the
standard deviations of modern air standards for 17D is ±11 per meg.
The mass-independent fractionation of O2 (Luz et al., 1999) is
expressed as the 17O anomaly (17D). This term is calculated by using
d15N to gravitationally correct both d17O and d18O, which are then
used in calculating the 17O anomaly from the equation:
17
17
18
Dzd Oatm 0:516 * d Oatm
(4)
The difference in coefficients for thermal fractionation (0.515)
and for gravitational fractionation (0.500) account for a small error
in 17D in the case that part of the isotopic fractionation is thermal.
However, this small error does not impact the following discussion.
3. Results and discussion
3.1. d15N and d38Ar/36Ar of the trapped air
The observed d15N values for Mullins glacier samples range
between 0.13‰ and þ0.11‰ (Fig. 3; Table 1), which is unusual as
most ice sheets and glaciers show only positive d15N values. Typical
d15N values range between 0.2‰ and 0.5‰, which is indicative of
gravitational fractionation in a firn layer with a close-off depth
between ~40 and 100 m. Negative d15N values are most likely a
result of thermal fractionation of the air, which can fractionate
Table 1a
Geochemical data core 1.
Depth
(m)
d15N
(‰)
dAr/N2
( ‰)
dO2/N2
(‰)
dO2/Ar
( ‰)
d18Oatm
( ‰)
TAC
(%)
1.6
1.6
4.3
4.3
4.9
4.9
6.6
6.6
8.6
8.6
8.6
11.4
11.4
15.1
15.1
0.03
0.05
0.06
0.12
0.01
0.04
0.05
0.05
0.06
0.08
0.03
0.06
0.09
0.05
0.00
4.75
5.97
0.31
3.54
1.59
6.05
2.89
3.00
10.02
8.83
7.08
6.76
5.87
16.88
9.98
24.91
26.17
46.29
48.21
61.35
61.02
17.92
10.25
15.77
26.89
6.14
18.77
30.14
53.40
60.83
20.27
20.32
46.02
44.85
59.77
55.32
20.76
13.21
5.69
18.11
13.12
12.09
24.44
37.16
51.42
0.83
0.41
0.49
0.63
0.64
0.75
0.41
0.35
0.86
0.10
0.18
1.02
0.71
1.01
1.17
2.0
4.6
3.3
2.7
2.1
17
25.4
28.3
17.3
2.9
Table 1b
Geochemical data core 2.
Depth
(m)
d15N
( ‰)
dAr/N2
(‰)
dO2/N2
(‰)
dO2/Ar
( ‰)
d18Oatm
(‰)
TAC
(%)
17
1.6
1.6
3.6
3.6
5
6.5
6.5
8.2
8.2
9.5
9.5
10
10
11.3
11.3
13.1
0.02
0.08
0.10
0.05
0.01
0.07
0.01
0.02
0.04
0.04
0.02
0.01
0.00
0.01
0.07
0.06
2.65
12.01
0.91
5.40
3.92
1.18
3.71
5.96
2.63
8.80
12.75
10.12
4.75
1.94
1.74
8.93
22.82
30.43
18.91
40.53
64.71
78.06
97.84
86.90
42.42
47.77
56.92
29.95
18.02
1.38
4.65
69.23
20.23
18.65
19.81
35.34
60.93
76.99
94.52
81.44
39.90
39.31
44.75
19.92
22.66
0.67
2.92
60.86
0.71
0.99
0.29
0.56
0.27
0.76
0.92
0.77
0.30
0.30
0.32
0.20
0.61
0.03
0.25
1.37
2.1
17.1
27.9
49.8
1.8
2.9
4.1
5.4
4.3
3.9
4.7
5.0
D
(per meg)
37.9
4.4
1.4
25.4
3.0
Table 1c
Geochemical data core 3.
Depth
(m)
d15N
( ‰)
dAr/N2
(‰)
dO2/N2
( ‰)
dO2/Ar
(‰)
d18Oatm
( ‰)
TAC
(%)
2.1
2.1
2.7
2.7
4.4
4.4
6.6
6.6
8.6
8.6
10.8
10.8
13.1
13.1
14.7
14.7
15.1
15.1
17.1
17.1
19.4
19.4
24
24
27.5
27.5
31.5
31.5
31.9
31.9
0.05
0.01
0.02
0.04
0.12
0.09
0.13
0.06
0.01
0.02
0.04
0.03
0.11
0.06
0.11
0.08
0.08
0.02
0.01
0.06
0.13
0.02
0.05
0.04
0.08
0.02
0.02
0.02
0.01
0.13
5.29
10.07
7.64
0.13
5.63
17.54
8.95
5.34
12.37
11.21
7.87
8.32
1.26
7.19
1.09
5.83
5.46
4.75
9.31
9.63
16.20
10.87
19.45
24.35
2.50
3.76
19.48
9.46
4.64
7.80
8.70
19.44
15.21
6.23
16.19
27.10
38.81
35.50
38.30
41.03
41.51
43.87
28.57
36.14
45.39
28.81
32.34
29.35
35.70
42.88
75.61
58.57
45.06
47.89
35.57
48.69
142.16
135.65
113.09
131.19
3.43
9.47
7.52
6.37
10.63
9.74
30.03
30.34
26.26
30.17
33.80
35.85
29.81
29.17
44.38
34.45
27.04
24.73
26.64
33.58
60.29
48.23
26.12
24.14
37.87
45.10
125.01
127.40
108.95
124.38
0.29
0.34
0.58
0.30
0.48
0.39
0.53
0.45
0.14
0.09
0.14
0.12
0.27
0.17
0.46
0.49
0.06
0.05
0.28
0.25
0.17
0.40
0.64
0.13
0.52
0.66
0.48
1.08
1.49
1.55
3.3
2.5
1.9
3.1
2.6
3.5
4.3
4.9
5.0
3.9
3.9
5.3
3.9
5.6
5.0
5.7
1.4
3.8
6.3
4.8
4.3
4.9
17
D
(per meg)
11.3
6.6
3.5
7.8
8.5
23.0
2.5
14.2
4.7
2.3
1.6
4.1
4.0
29.4
21.3
55.8
22.9
20.9
D
(per meg)
4.8
1.9
1.8
33
isotopes both negatively and positively along a temperature
gradient between the surface and the subsurface ice where air is
being trapped (Grew and Ibbs, 1952; Severinghaus et al., 2001). A
10 K temperature gradient between the ice at the trapping zone and
the surface air temperature could explain the observed range in
d15N values. Using a thermal diffusion sensitivity for 15N of U
(‰ K1) ¼ (8.656/T) e (1232/T2) at T ¼ 250 K (Severinghaus et al.,
2001), if the ice at the re-trapping zone was 10 K warmer than the
surface air temperature, then the trapped air would record a
negative d15N signal of 0.15‰, and a þ0.15‰ signal if the temperature gradient was in the opposite direction. A 10 K temperature
difference between the atmosphere and the depth at which air was
re-trapped could occur on seasonal timescales (Kowalewski et al.,
2011).
34
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
A comparison of d15N values to d38Ar/36Arcorr data (corrected for
gas loss, see Table 2) also indicates that gravitational fractionation is
not the only process fractionating air (Fig. 4). If it were, the data
would lie along the solid black line, which passes through the origin
and has a slope of 0.5 (e.g., see the point for modern Vostok ice).
Data from Mullins glacier samples do not lie firmly on an expected
line for thermal fractionation either (slope ¼ ~0.7).
Fig. 4 shows that in principle, the combined Ar and N2 isotope
data can be explained by a combination of gravitational and thermal fractionation. Gravitational fractionation would raise d15N and
d38Ar/36Ar along a line with a slope of 1:2, whereas thermal fractionation would lower them along a line with a steeper slope
(indicated by the dashed lines). The summed processes could account for points with an 38Ar enrichment greater than 2 d15N.
Unfortunately, it is not possible to quantify how much of the
isotopic fractionation is from gravitational or thermal fractionation
as firn thickness at the time of ice formation is unknown for the
region, and consequently a mixing curve cannot be generated.
3.2. Total air content and dAr/N2 of the trapped air
Further evidence for unusual entrainment and preservation of
trapped air comes from the total air content (TAC) of the ice cores.
Polar glaciers trap a volume of air that, at standard temperature and
pressure, comprises roughly 10% of the total ice volume (Raynaud
and Lebel, 1979; Herron and Langway, 1987). Lower TAC values
may be due to a different mechanism for bubble close-off as
opposed to typical firn densification processes (Herron and
Langway, 1980), such as the loss of trapped air by cracking and
resealing of glacial ice. The total air content measured in Mullins ice
Table 2a
Ar-isotope data core 1.
Depth (m)
d40Ar/36Armeas (‰)
d38Ar/36Armeas (‰)
dN2/Ar (‰)
d40Ar/36Arcorr (‰)
d38Ar/36Arcorr (‰)
Age (Ma)
1.4
1.7
4.7
6.5
11.3
11.5
15.0
0.140
0.155
0.515
0.077
0.519
0.090
0.177
0.049
0.086
0.299
0.008
0.276
0.020
0.108
4.92
4.92
4.00
2.92
6.17
6.17
13.32
0.141
0.156
0.510
0.077
0.510
0.099
0.119
0.050
0.087
0.296
0.008
0.272
0.024
0.078
0.61
0.26
1.26
0.93
0.51
0.76
0.55
Depth (m)
d40Ar/36Armeas (‰)
d38Ar/36Armeas (‰)
dN2/Ar (‰)
d40Ar/36Arcorr (‰)
d38Ar/36Arcorr (‰)
Age (Ma)
1.4
3.5
4.9
6.4
6.6
8.1
8.3
9.4
9.7
11.2
13.0
13.2
0.168
0.246
0.315
0.149
0.369
0.529
0.026
0.137
0.039
1.088
0.723
0.690
0.091
0.164
0.187
0.117
0.167
0.315
0.008
0.045
0.021
0.512
0.379
0.381
6.78
3.18
4.11
2.86
2.86
4.19
4.19
10.76
10.76
1.45
8.23
8.23
0.146
0.243
0.315
0.149
0.369
0.525
0.030
0.177
0.001
1.088
0.700
0.667
0.079
0.162
0.187
0.117
0.167
0.313
0.006
0.066
0.000
0.512
0.367
0.369
0.19
1.23
0.91
1.30
0.52
1.55
0.63
0.67
0.02
0.96
0.52
1.10
Depth (m)
d40Ar/36Armeas (‰)
d38Ar/36Armeas (‰)
dN2/Ar (‰)
d40Ar/36Arcorr (‰)
d38Ar/36Arcorr (‰)
Age (Ma)
2.0
2.3
2.6
4.3
4.6
6.5
8.5
10.7
13.0
13.2
14.5
16.9
19.3
19.5
23.9
24.1
27.6
31.8
32.0
0.032
0.200
0.473
0.018
0.139
0.516
0.151
0.112
0.114
0.145
0.121
0.227
0.195
0.377
0.283
0.162
0.435
0.185
0.102
0.036
0.142
0.241
0.004
0.058
0.267
0.122
0.071
0.061
0.100
0.050
0.061
0.079
0.221
0.174
0.133
0.251
0.123
0.017
7.38
7.38
4.13
10.47
10.47
6.75
11.84
8.10
3.26
3.26
2.14
9.10
12.79
12.79
22.95
22.95
1.00
5.51
5.51
0.014
0.182
0.463
0.060
0.098
0.504
0.103
0.090
0.109
0.140
0.121
0.256
0.249
0.322
0.157
0.036
0.435
0.180
0.107
0.027
0.133
0.236
0.026
0.037
0.261
0.097
0.060
0.058
0.097
0.050
0.076
0.107
0.193
0.108
0.068
0.251
0.121
0.019
0.62
1.27
0.14
0.12
0.37
0.28
1.38
0.45
0.12
0.84
0.33
1.58
0.52
0.96
0.91
1.50
1.03
0.95
1.04
Table 2b
Ar-isotope data core 2.
Table 2c
Ar-isotope data core 3.
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
Fig. 4. Averages of replicates for d38Ar/36Arcorr for Cores 1, 2, and 3 plotted against
averages of replicates for d15N. Expected fractionations for gravity (black line) and
temperature (dashed line) are noted. Dashed arrows indicate possible trajectories for
air with both gravitational and thermal fractionation.
is unusually low, between 1.5 and 7% (Fig. 3; Table 1). The low total
gas content, together with the anomalously low d15N values, suggests that air has been lost via cracking of the ice. Re-trapped air
was then thermally fractionated, and/or firn densification occurred
abnormally. In Core 3, an increasing trend in total air content with
depth provides some support that pristine air may be preserved at
greater depths below ice that has been cracked.
dAr/N2 of the trapped gases also acts as an indicator of core
quality and integrity. Typical pristine ice cores have dAr/N2 values
between 5‰ and 0‰, close to that of atmosphere. dAr/N2 values
below 5‰ indicate gas loss due to the preferential loss of the
smaller Ar atom (3.5 Å) relative to the larger N2 molecule (3.8 Å)
(Craig et al., 1988; Severinghaus et al., 2003; Severinghaus and
Battle, 2006; Huber et al., 2006). dAr/N2 values greater than 0‰
indicate that the ice may have melted and refrozen at some point.
Ar is approximately twice as soluble as N2 (so that the dAr/N2 of
equilibrium dissolved O2 is ~þ1000‰), hence air trapped in
refrozen melt water would have greatly enriched dAr/N2 values.
dAr/N2 values do not exceed þ7‰ in any of our cores, indicating
that very little gas is present in refrozen melt water (Fig. 3; Table 1).
On the other hand, dAr/N2 values are as low as 25‰, indicating
significant gas loss in some samples.
The gas loss reflected in the dAr/N2 values fits with the
measured low total air content, indicating that air has been lost via
cracks in the ice. Furthermore, the air that was re-trapped when the
ice annealed was thermally fractionated, according to the observed
d15N and d38Ar/36Ar data.
Fig. 5. The paleoatmospheric ratio of
Bender et al., 2008).
35
40
Ar/36Ar for the past 800 ka (adapted from
The Ar-ages for trapped air recovered from Mullins glacier in
Cores 1, 2, and 3 are presented in Fig. 6 and in Table 2. Significant
scatter exists in all cores. However, the salient point is that all three
cores contain air samples that date to at least 1 Ma, with the oldest
3.3. Ar-ages
The Ar isotope chronometer relies on the measured rate of
change of atmospheric 40Ar/36Ar (Bender et al., 2008). Concentrations of 36Ar and 38Ar have been more or less constant throughout
geologic time, but 40Ar has been slowly and steadily increasing
through Earth's history as a result of the decay of 40K. The increase
in the 40Ar/36Ar ratio over the past 1 Ma has been observed to be
0.066 ± 0.007‰/Ma, and we assume this rate has been constant
through the Pleistocene (Fig. 5; Bender et al., 2008). We precisely
measure the ratio of 40Ar/36Ar of the trapped air and determine the
age of the air based on this rate of 40Ar/36Ar increase.
Fig. 6. Ar ages with depth for Cores 1, 2, and 3. Error bars indicating analytical uncertainty are ±0.2 Ma.
36
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
individual sample as old as 1.6 ± 0.2 Ma. Core 3, the deepest core,
shows a convergence in age with depth, hinting that older air might
be preserved at greater depths. The wide scatter in gas ages suggests contamination, likely with relatively young air entrained in
re-annealed fractures formed via brittle deformation during ice
flow or thermal-contraction cracking. The plausibility of contamination is substantiated by other geochemical and isotopic properties of the trapped air (Sections 3.1e3.2), as well as visual evidence
of cracking in the ice (Fig. 2). Hence, we believe the oldest dated
sample at 1.6 ± 0.2 Ma is a minimum age for the trapped air from
this site in Mullins glacier.
It could be conjectured that rocky debris and sand veins in the
ice may be contributing some radiogenic Ar from the decay of 40K
that would force calculated ages towards the present or even to the
future. The samples analyzed contained on average roughly 0.1% by
weight dolerite. Assume that a sample had as much as 1% by weight
dolerite and that the weight percent of potassium (K) in dolerite is
0.75% (Gunn, 1966). If the K in the dolerite decayed for 4 Ma, and if
all the resulting 40Ar were lost to the ice, there would be a 0.0017‰
increase in d40Ar/36Aratm (the equivalent of ~26 kyr on the Ar
isotope chronometer). Such a small contribution of radiogenic Ar
from the decay of 40K indicates that there is very little 40Ar
contamination from sources other than air. One could also invoke
the loss of radiogenic 40Ar that had accumulated before the dolerite
particles were trapped in the ice. However, it seems very unlikely
that the escape rate of 40Ar would be greater than its current production rate in this cold environment. It thus appears that the atmosphere is the only significant Ar source for the trapped gases.
As noted in Section 3.1, if indeed the isotopes of N2 and Ar were
fractionated thermally, then there could be an error in the Arisotope chronometer used to estimate ages for the trapped air.
We calculate d40Ar/36Aratm assuming that Ar isotopes are fractionated only by gravity. Grachev and Severinghaus (2003) established
that the thermal diffusion sensitivity for the isotopic pair 40Ar/36Ar
(U ¼ 0.0411‰ K1 at 250 K) is not simply double that for 38Ar/36Ar
(U ¼ 0.0211‰ K1 at 250 K). Therefore, a thermal gradient would
affect d40Ar/36Ar and d38Ar/36Ar differently, which is not accounted
for in the calculation of d40Ar/36Aratm. Assuming that 38Ar/36Ar
fractionation is only thermal, and using a 10 K temperature
gradient in the direction that would explain the observed negative
d15N values, this would equate to a 0.0111‰ increase in
d40Ar/36Aratm, which means an underestimation of 168 kyr on the
Ar-chronometer for this case.
Gas loss also affects the isotopes of Ar, which may contribute
additional errors to the Ar-isotope chronometer. We estimate the
amount of Ar loss and consequent fractionation of Ar isotopes as
per Kobashi et al. (2010). Following convention, dAr/N2 that has
been
corrected
for
gravitational
fractionation
(dAr/N2
15
corrected ¼ dAr/N2 observed e 11.98*d N) is used to indicate the
amount of Ar loss. The coefficient (11.98) is the ratio of the mass
differences of the pairs dAr/N2 and d15N. Thermal fractionation is
ignored here as the difference between thermal coefficients for dAr/
N2 and d15N is small and contributes little to this correction
(Kobashi et al., 2010). An enrichment of 0.007‰ in d40Ar/36Ar
relative to a 1‰ increase in dAr/N2 is used to correct the measured
isotopic composition of Ar (Severinghaus et al., 2003; Kobashi et al.,
2010). While a correction for d38Ar/36Ar due to gas loss has not been
empirically demonstrated, we theoretically estimate that the gas
loss for d38Ar/36Ar is proportional to the molecular diffusion coefficient for d40Ar/36Ar, as molecular diffusion is the most likely
mechanism for fractionation during gas loss. Thus, the gas loss
fractionation of d38Ar/36Ar ¼ 0.520*d40Ar/36Ar, where the coefficient 0.520 is the natural log of the ratio of reduced masses of argon.
The largest enrichment for d40Ar/36Ar is 0.126‰ and d38Ar/36Ar is
0.065‰ (Table 2), resulting in an overestimation of at most 80 kyr
due to 40Ar enrichment from gas leakage. This small effect from gas
loss counters the effect of thermal fractionation on the Ar isotopes,
both of which whose impacts remain small when compared to the
analytical uncertainty of the Ar measurement (±220 kyr).
3.4. Cracking of surface ice
The existence of cracks in the ice has been visually confirmed
to a depth of 25 m. d40Ar/36Aratm ages younger than ages of
overlying ashes indicate that cracking continues to depths of at
least 33 m. The depth that thermal-contraction cracks can propagate vertically in ice depends mostly on the time duration of
forcing, the material properties of the ice, and the amplitude of the
thermal forcing (Lachenbruch, 1961; Maloof et al., 2002). Here, we
present a simple estimation of the depth to which these cracks
would likely propagate in Mullins glacier to determine whether
the geochemical data can be substantiated. A simplified estimation
of the maximum depth to which a thermal-contraction crack can
propagate in a semi-viscous material without regard to an overlying debris layer is adapted from the linearized approximation of
Maloof et al. (2002):
bmax z½jsz * gðuto Þ=pKC 2
bmax
j
s
E
v
a
DT
z
k
u
g(uto)
to
h
KC
¼maximum depth of crack propagation
¼geometric correction factor
¼maximum thermal stress at surface
¼E/(1v)* aDT
¼*Young's modulus
¼*Poisson's ratio
¼*thermal expansion coefficient of ice
¼temperature variation at the surface
¼skin depth of temperature variation
penetration
¼√2k/u
¼*thermal diffusivity
¼angular frequency of variation
¼time weighting function
¼(uto√2)/(√1þ u2t2o)
¼stress relaxation time
¼h/(E/(1v))
¼effective linear viscosity
¼þcritical crack edge stress intensity factor
(5)
1.12√p
9 109 Pa
0.3
5 105 K1
10e20 K
1.3 106 m2 s1
7.27 105 s1 diurnal
1.99 107 s1 annual
1012e1015 Pa s
1.5 105 Pa m1/2
*
values from Cuffey and Paterson (2010); þvalues from Fischer et al. (1995).
It is important to note that both h and KC are poorly constrained
for Mullins glacier and therefore limit the robustness of this estimation. By using reasonable estimates for KC and an effective viscosity of 1015 Pa s, we can estimate the maximum crack depth for
thermal amplitudes that are observed at the buried ice surface
under various depths of surface debris. For a diurnal forcing, thermal amplitudes at the ice surface of 2 K, 5 K, and 10 K, are consistent
with depths of surface debris 30 cm, 10 cm, and 2 cm respectively
(Kowalewski et al., 2011). This forcing could result in respective
crack depths of ~1 m, ~7 m, and ~25 m. The current debris thickness
at the core sites averages ~30 cm. Annual forcing, with a thermal
amplitude of 25 K at the ice surface (Kowalewski et al., 2011) and an
effective viscosity of 1015 Pa s, can lead to cracking deeper than
30 m. This process of thermal fracture, together with cracks and
fractures from uneven ice flow, could provide abundant conduits
for pervasive contamination with relatively young air replacing gas
initially trapped at the base of the firn column.
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
3.5. dO2/N2, dO2/Ar, and d18Oatm of the trapped air
Atmospheric O2/N2 and O2/Ar ratios have been essentially constant over the Pleistocene (Bender, 2002; Kawamura et al., 2007).
Thus, any deviations in these ratios from that of modern air would
reflect alteration of the trapped air by a biological or physical
process. The O2/N2 and O2/Ar ratio of trapped air from Mullins
glacier is lower than modern air by up to 15% (Fig. 3; Table 1). This
depletion is a result of microbial respiration (Souchez, 1997), which
consumes O2, while leaving N2 untouched, and gas loss, which
results in the preferential loss of the smaller O2 and Ar atoms (3.5 Å)
relative to the larger N2 molecule (3.8 Å) (Craig et al., 1988;
Severinghaus et al., 2003; Severinghaus and Battle, 2006; Huber
et al., 2006). The large magnitude of the O2 depletions, and the
fact that dO2/Ar and dO2/N2 are similarly depleted, suggest that
most of the fractionation signal comes from respiration and not gas
loss, as O2 and Ar atoms behave similarly in response to gas loss
fractionation due to their similar molecular sizes (Huber et al.,
2006). Likewise, trapped O2 is enriched in 18O, a consequence of
isotope fractionation when O2 is consumed by respiration, with
some small part of the signal due to fractionation during gas loss
(Fig. 3). Replication of d18Oatm, dO2/N2, and dO2/Ar analyses show
heterogeneity that is likely a result of small-scale variability in
microbial activity and bubble gas loss. While Cores 1 and 2 do not
show a clear trend of decreasing dO2/N2 and dO2/Ar with depth,
Core 3 does, and the mirror image trend is seen in the d18Oatm of the
trapped air. The apparent trends in depth in Core 3 suggests slower
exchange of air at depth at this site.
In Fig. 7, we analyze the variations of d18Oatm in terms of its
modern value, its historical changes, and its fractionation during
respiration. The d18O of O2 has been documented for the past 800
ka, ranging between 0.4‰ and 1.4‰ (Dreyfus et al., 2007). d18Oatm
is plotted v. the fraction of O2 consumed ([O2/Ar]SAMPLE/[O2/Ar]AIR).
Point 1 on the right hand axis is the current d18Oatm value, point 2 is
the lowest value of the past 800 ka, and point 3 is the highest
(Dreyfus et al., 2007). The figure shows that all the data can be
explained by invoking three simple assumptions. The first is that
the starting d18Oatm was always within the range of the past 800 ka.
The second is that during respiration, the maximum respiratory
fractionation effect is that associated with Rayleigh distillation
(isotope effect ¼ 18‰), which determines the slopes of lines A and
37
B. The third is that the minimum 18O fractionation is zero, which
would occur selectively in individual bubbles where all O2 was
consumed.
The minimum d18Oatm then corresponds to samples in which the
initial d18Oatm value is 0.4‰ (point 1), and respiration consumes
all O2 in selected bubbles, so that d18Oatm does not rise (line D). Line
C corresponds to the same scenario for a sample starting with the
modern d18Oatm. Line B corresponds to a sample starting with the
modern d18Oatm, in which O2 is Rayleigh-fractionated during
respiration (isotope effect ¼ 18‰). Line A, which gives the
maximum attainable d18Oatm at a given value of O2 consumption,
corresponds to Rayleigh fractionation in a sample starting with the
heaviest known d18Oatm. All points fall between lines A and D,
indicating that all trapped air initially had d18Oatm within the
known range, and that the only other cause of variability is respiration with an isotope effect of 18‰.
3.6.
17
D of O2 of trapped air
CO2 concentrations cannot be directly measured from the
trapped air from Mullins glacier due to microbial respiration, which
artificially increases CO2, and because CO2 can be released by CaCO3
dust in the ice when trapped gases are sampled (Anklin et al., 1995).
We use the triple isotope composition of O2 as a proxy for CO2
concentrations as its value varies with the concentration of CO2 in
the atmosphere (Luz et al., 1999; Blunier et al., 2012). The isotopic
composition of O2 depends on two sets of processes. One is massdependent, which includes respiration as well as the fractionation
of isotopes in the hydrologic cycle that affects the isotopic
composition of O2 in leaf water. In mass-dependent processes, 18O
is fractionated about twice as strongly as 17O, relative to 16O. The
second is the mass-independent isotope transfer of 17O and 18O
from O2 to CO2 in the stratosphere. This stratospheric fractionation
of oxygen produces an isotopic composition of O2 that depends on
the CO2 concentration of air and the global rate of respiration and
photosynthesis. In the stratospheric reaction, 17O is known to
fractionate 1.7 times as much as 18O (Lammerzahl et al., 2002). This
‘mass peculiar’ fractionation, measured as 17D, scales linearly with
atmospheric CO2 concentrations to a first approximation. Analyses
of 17D have been conducted on samples from the Vostok ice core,
dating back to 400 ka, illustrating empirically that 17D co-varies
with the measured CO2 concentrations with a root mean squared
error of ±21 ppm (Blunier et al., 2012, Fig. 8). Much of the noise in
this relation is due to analytical uncertainties. 17D is thus a property
that reflects atmospheric CO2, and the state of Earth's climate
system at the time air was trapped. In Mullins glacier samples, the
full range of glacial-interglacial 17D is expressed with no data points
exceeding this range. Thus, data from Cores 1, 2, and 3 (Fig. 8) show
that 17D values fall within the full range of 0e400 ka 17D variations
(Blunier et al., 2012). Due to the uncertainty of the ages of the
trapped air and the fact that the trapped air is likely a mixture of air
trapped at different times, it is not possible to be more specific
about atmospheric CO2 history.
4. Conclusions
Fig. 7. Averages for replicate samples of fraction of O2 consumed plotted against
d18Oatm. Points between lines B and C can be explained as modern air that has undergone respiration. Points within lines A and D can be explained as air with paleod18Oatm values that have undergone respiration.
Dating of air trapped in ice samples collected from Mullins
glacier gives ages as old as 1.6 Ma. However, the visual evidence for
thermal-contraction cracks and brittle failure in all three cores, as
well as (1) the observed scatter in calculated gas ages over short
depth profiles, (2) the isotopic evidence for mixed thermal and
gravitational fractionation, and (3) the low total gas contents and
gas loss measured in all three cores suggest contamination with
younger air. These geochemical data suggest that the measured gas
age of 1.6 ± 0.2 Ma is most likely a minimum age for the original
38
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
Fig. 8. 17D for the past 400 ka adapted from Blunier et al. (2012), and CO2 for the past 800 ka adapted from Luthi et al. (2008) plotted versus age. On the right side, 17D of Mullins
glacier air plotted with samples <1 Ma in gray and samples >1 Ma in white. Samples show the full range of glacial-interglacial CO2 through 1.6 Ma.
glacial air. A convergence in observed ages with increasing depth in
the deepest core analyzed (down to 33 m) hints at the possibility
that pristine air might be recovered at greater depths. The
compositional analysis of the air indicates that microbial respiration has altered the trapped gases. 17D analyses are used as a CO2
proxy and indicate that the full range of glacial-interglacial CO2
values are seen through 1.6 Ma with no data exceeding this range.
Acknowledgments
This material is based upon work supported by the National
Science Foundation, Polar Programs, under Grant ANT-0636731 to
MLB and Grants ANT-0739700 and ANT-0944702 to DRM. Logistical
support for this project in Antarctica was provided by the U.S. National Science Foundation through the U.S. Antarctic Program. And
we thank Jeffrey Severinghaus for his thoughtful counsel.
References
Anklin, M., Barnola, J.M., Schwander, J., Stauffer, B., Raynaud, D., 1995. Processes
affecting the CO2 concentrations measured in Greenland ice. Tellus Ser. B-Chem.
Phys. Meteorol. 47, 461e470.
Bender, M.L., 2002. Orbital tuning chronology for the Vostok climate record supported by trapped gas composition. Earth Planet. Sci. Lett. 204, 275e289.
Bender, M.L., Barnett, B., Dreyfus, G., Jouzel, J., Porcelli, D., 2008. The contemporary
degassing rate of Ar-40 from the solid Earth. Proc. Natl. Acad. Sci. U. S. A. 105,
8232e8237.
Blunier, T., Barnett, B., Bender, M.L., Hendricks, M.B., 2002. Biological oxygen productivity during the last 60,000 years from triple oxygen isotope measurements. Glob. Biogeochem. Cycles 16. http://dx.doi.org/10.1029/2001gb001460.
Blunier, T., Bender, M.L., Barnett, B., von Fischer, J.C., 2012. Planetary fertility during
the past 400 ka based on the triple isotope composition of O2 in trapped gases
from the Vostok ice core. Clim. Past. 8, 1509e1526.
Chapman, S., Cowling, T.G., 1970. The Mathematical Theory of Non-uniform Gases.
Cambridge University Press.
Craig, H., Horibe, Y., Sowers, T., 1988. Gravitational separation of gases and isotopes
in polar ice caps. Science 242, 1675e1678.
Cuffey, K.M., Paterson, W.S.B., 2010. The Physics of Glaciers, fourth ed. Elsevier, MA.
Dreyfus, G.B., Parrenin, F., Lemieux-Dudon, B., Durand, G., Masson-Delmotte, V.,
Jouzel, J., Barnola, J.M., Panno, L., Spahni, R., Tisserand, A., Siegenthaler, U.,
Leuenberger, M., 2007. Anomalous flow below 2700 m in the EPICA Dome C ice
core detected using delta O-18 of atmospheric oxygen measurements. Clim.
Past. 3, 341e353.
Emerson, S., Quay, P.D., Stump, C., Wilbur, D., Schudlich, R., 1995. Chemical tracers of
productivity and respiration in the subtropical pacific-ocean. J. Geophys. Res.Oceans 100, 15873e15887.
Fischer, M.P., Alley, R.B., Engelder, T., 1995. Fracture-toughness of ice and firn
determined from the modified ring test. J. Glaciol. 41, 383e394.
Fountain, A.G., Nylen, T.H., Monaghan, A., Basagic, H.J., Bromwich, D., 2010. Snow in
the McMurdo dry valleys, Antarctica. Int. J. Climatol. 30, 633e642.
Grachev, A.M., Severinghaus, J.P., 2003. Determining the thermal diffusion factor for
Ar-40/Ar-36 in air to aid paleoreconstruction of abrupt climate change. J. Phys.
Chem. A 107, 4636e4642.
Grew, K.E., Ibbs, T.L., 1952. Thermal Diffusion of Gases. Cambridge University Press.
Gunn, B.M., 1966. Modal and element variation in Antarctic tholeiites. Geochim.
Cosmochim. Acta 30, 881e920.
Herron, M.M., Langway, C.C., 1980. Firn densification e an empirical model.
J. Glaciol. 25, 373e385.
Herron, M.M., Langway, C.C., 1987. Derivation of paleoelevations from total air
content of two deep Greenland ice cores. In: The Physical Basis of Ice Sheet
Modelling, Proceedings of the Vancouver Symposium. IAHS, p. 170.
Huber, C., Beyerle, U., Leuenberger, M., Schwander, J., Kipfer, R., Spahni, R.,
Severinghaus, J.P., Weiler, K., 2006. Evidence for molecular size dependent gas
fractionation in firn air derived from noble gases, oxygen, and nitrogen measurements. Earth Planet. Sci. Lett. 243, 61e73.
Kawamura, K., Parrenin, F., Lisiecki, L., Uemura, R., Vimeux, F., Severinghaus, J.P.,
Hutterli, M.A., Nakazawa, T., Aoki, S., Jouzel, J., Raymo, M.E., Matsumoto, K.,
Nakata, H., Motoyama, H., Fujita, S., Goto-Azuma, K., Fujii, Y., Watanabe, O.,
2007. Northern hemisphere forcing of climatic cycles in Antarctica over the past
360,000 years. Nature 448, 912e916.
Kobashi, T., Severinghaus, J.P., Barnola, J.M., Kawamura, K., Carter, T., Nakaegawa, T.,
2010. Persistent multi-decadal Greenland temperature fluctuation through the
last millennium. Clim. Change 100, 733e756.
Kowalewski, D.E., Marchant, D.R., Swanger, K.M., Head, J.W., 2011. Modeling vapor
diffusion within cold and dry supraglacial tills of Antarctica: Implications for
the preservation of ancient ice. Geomorphology 126, 159e173.
Kowalewski, D.E., Marchant, D.R., Head, J.W., Jackson, D.W., 2012. A 2D model for
characterising first-order variability in sublimation of buried glacier ice,
Antarctica: assessing the influence of polygon troughs, desert pavements and
shallow subsurface salts. Permafr. Periglac. Process 23, 1e14.
Lachenbruch, A., 1961. Depth and spacing of tension cracks. J. Geophys. Res. 66,
4273e4292.
Lammerzahl, P., Rockmann, T., Brenninkmeijer, C.A.M., Krankowsky, D.,
Mauersberger, K., 2002. Oxygen isotope composition of stratospheric carbon
dioxide. Geophys. Res. Lett. 29 http://dx.doi.org/10.1029/2001gl014343.
Levy, J.S., Marchant, D.R., Head, J.W., 2006. Distribution and origin of patterned
ground on Mullins Valley debris-covered glacier, Antarctica: the roles of ice
flow and sublimation. Antarct. Sci. 18, 385e397.
Luthi, D., Le Floch, M., Bereiter, B., Blunier, T., Barnola, J.M., Siegenthaler, U.,
Raynaud, D., Jouzel, J., Fischer, H., Kawamura, K., Stocker, T.F., 2008. High-resolution carbon dioxide concentration record 650,000-800,000 years before
present. Nature 453, 379e382.
Luz, B., Barkan, E., Bender, M.L., Thiemens, M.H., Boering, K.A., 1999. Triple-isotope
composition of atmospheric oxygen as a tracer of biosphere productivity. Nature 400, 547e550.
Mackay, S.L., Marchant, D.R., Head, J.W., Lamp, J.L., 2014. Investigations of coldbased, Antarctic debris-covered glaciers: evaluating their potential as climate
archives through studies of ground penetrating radar and surface morphologic
change. J. Geophys. Res-Earth Surf. 119, 2505e2540.
Maloof, A.C., Kellogg, J.B., Anders, A.M., 2002. Neoproterozoic sand wedges: crack
formation in frozen soils under diurnal forcing during a snowball Earth. Earth
Planet. Sci. Lett. 204, 1e15.
Marchant, D.R., Lewis, A.R., Phillips, W.M., Moore, E.J., Souchez, R.A., Denton, G.H.,
Sugden, D.E., Potter, N., Landis, G.P., 2002. Formation of patterned ground and
sublimation till over Miocene glacier ice in Beacon Valley, southern Victoria
Land, Antarctica. Geol. Soc. Am. Bull. 114, 718e730.
Marchant, D.R., Phillips, W.M., Schaefer, J.M., Winckler, G., Fastook, J.L., Shean, D.E.,
Kowalewski, D.E., Head, J.W., Lewis, A.R., 2007. Establishing a Chronology for the
World's Oldest Glacier Ice. USGS Open-file Report Extended Abstract 054, p. 4.
Marchant, D.R., Head, J.W., 2007. Antarctic dry valleys: microclimate zonation,
variable geomorphic processes, and implications for assessing climate change
on Mars. Icarus 192, 187e222.
Martinerie, P., Raynaud, D., Etheridge, D.M., Barnola, J.M., Mazaudier, D., 1992.
Physical and climatic parameters which influence the air content in polar ice.
Earth Planet. Sci. Lett. 112, 1e13.
Ng, F., Hallet, B., Sletten, R.S., Stone, J.O., 2005. Fast-growing till over ancient ice in
Beacon Valley, Antarctica. Geology 33, 121e124.
Raynaud, D., Lebel, B., 1979. Total gas content and surface elevation of polar ice
A.M. Yau et al. / Quaternary Geochronology 28 (2015) 29e39
sheets. Nature 281, 289e291.
Schafer, J.M., Baur, H., Denton, G.H., Ivy-Ochs, S., Marchant, D.R., Schluchter, C.,
Wieler, R., 2000. The oldest ice on Earth in Beacon Valley, Antarctica: new
evidence from surface exposure dating. Earth Planet. Sci. Lett. 179, 91e99.
, B., Ahn, Y., Yoon, T., Shin, S.W., Huh, K.I., 2004. DEM Generation
Schenk, T., Csatho
from the Antarctic LiDAR Data: Site Report. US Geological Survey, Reston, VA.
Schorghofer, N., 2005. A physical mechanism for long-term survival of ground ice in
Beacon Valley, Antarctica. Geophys. Res. Lett. 32 http://dx.doi.org/10.1029/
2005GL023881.
Schwander, J., Stauffer, B., Sigg, A., 1988. Air mixing in firn and the age of the air at
pore close-off. Ann. Glaciol. 10, 141e145.
Schwander, J., 1989. The transformation of snow to ice and the occlusion of gases.
In: Oeschger, H., Langway, C. (Eds.), The Environmental Record in Glaciers and
Ice Sheets. John Wile, New York.
Severinghaus, J.P., Bender, M.L., Keeling, R.F., Broecker, W.S., 1996. Fractionation of
39
soil gases by diffusion of water vapor, gravitational settling, and thermal
diffusion. Geochim. Cosmochim. Acta 60, 1005e1018.
Severinghaus, J.P., Grachev, A., Battle, M., 2001. Thermal fractionation of air in polar
firn by seasonal temperature gradients. Geochem. Geophys. Geosyst. 2,
2000GC000146.
Severinghaus, J.P., Grachev, A., Luz, B., Caillon, N., 2003. A method for precise
measurement of argon 40/36 and krypton/argon ratios in trapped air in polar
ice with applications to past firn thickness and abrupt climate change in
Greenland and at Siple Dome, Antarctica. Geochim. Cosmochim. Acta 67,
325e343.
Severinghaus, J.P., Battle, M.O., 2006. Fractionation of gases in polar lee during
bubble close-off: new constraints from firn air Ne, Kr and Xe observations. Earth
Planet. Sci. Lett. 244, 474e500.
Souchez, R., 1997. The buildup of the ice sheet in central Greenland. J. Geophys. Res.Oceans 102, 26317e26323.