a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/aca

An improved ow analysision chromatography method

for determination of cationic and anionic species
at trace levels in Antarctic ice cores
Andrea Morganti , Silvia Becagli, Emiliano Castellano,
Mirko Severi, Rita Traversi, Roberto Udisti
Department of Chemistry, University of Florence, Scientic Pole, via della Lastruccia 3, 50019 Sesto F.no (FI), Italy

a r t i c l e

i n f o

a b s t r a c t

Article history:

A method was developed for the quantitative determination of cations and anions in Antarc-

Received 4 July 2007

tic ice cores at g L1 and sub-g L1 levels by ion chromatography (IC), after ultra-clean

Received in revised form

decontamination procedures. Strict manipulation and decontamination procedures were

21 September 2007

used in sub-sampling, in order to minimise sample contamination. Na+ , NH4 + , K+ , Mg2+

Accepted 25 September 2007

and Ca2+ were determined by 12-min isocratic elution (H2 SO4 eluent). Contemporaneously,

Published on line 29 September 2007

in a parallel device, F , MSA (methanesulfonic acid), Cl , NO3 and SO4 2 were analysed
in a single 12-min run with multiple-step elution using Na2 CO3 /NaHCO3 as eluent. Melted

Keywords:

ice samples were pumped from their still-closed containers (polystyrene accuvettes with

Ion chromatography

polyethylene caps), shared between the two ion chromatographic systems, online ltered

Anions

(0.45 m Teon membrane) and pre-concentrated (anions and cations pre-concentration

Cations

columns) using a ow analysis system, thus avoiding uptake of contaminants from the

Antarctic ice core

laboratory atmosphere. Sensitivity, linear range, reproducibility and detection limit were

Sample handling

evaluated for each chemical species. Anion or cation detection limits ranged from 0.01 to
0.15 g L1 by using a relatively small sample volume (1.5 mL). Such values are signicantly
lower than those reported in literature for almost all the components. These methods were
successfully applied to the analysis of cations and anions at trace levels in the Dome C ice
core. The composition of the atmospheric aerosol for the last 850 kyr was reconstructed by
high-resolution continuous chemical stratigraphies. Concentration trends in the last nine
glacial-interglacial climatic cycles were shown and briey discussed.
2007 Elsevier B.V. All rights reserved.

1.

Introduction

High-resolution chemical stratigraphies from polar ice cores

allow reconstruction of the paleo-atmospheres composition and help in understanding the complex relationships
between climate forcing and environmental feedback processes. Indeed, chemical species present in the atmosphere as
particles or gases, are scavenged by wet and dry processes and

Corresponding author. Tel.: +39 055 4573350; fax: +39 055 4573385.
E-mail address: andrea.morganti@uni.it (A. Morganti).
0003-2670/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2007.09.050

trapped in the snow layers that continuously accumulate in

the polar ice sheets. In sites where regular snow accumulation
is not signicantly perturbed by post-depositional processes
(such as wind ablation or re-distribution, summer melting,
and layer mixing by irregular ice ow), chemical compounds
irreversibly xed in the snow constitute a unique continuous
archive of the atmosphere composition. In this way, changes in
concentrations or uxes of chemical species measured along

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

an ice core can be interpreted as temporal changes in emission

efciency of their source areas or as changes in transport processes carrying these components to the ice caps (including
different atmospheric pathways and/or changes in scavenging
efciency during the transport) [14].
Antarctic snow precipitation is mainly representative of
a remote oceanic area, with signicant contributions from
long-range transport of continental components. Therefore, temporal variations in primary and secondary aerosol
components are potentially able to give information about
changes of climatic conditions controlling, for instance, seaspray formation (primary marine aerosol), phytoplanktonic
productivity (biogenic secondary emissions) or hydrological cycles (dampness or humidity) of the dust source areas
[2,5].
In particular, primary marine aerosol, marked by Na+ , Cl
and Mg2+ , constitutes the main aerosol source in Antarctic
coastal areas [6,7]. Among these compounds, Na+ is the most
reliable sea-spray marker; indeed, it is a conservative species
and non-sea-spray (nss) sources (crustal, anthropogenic) in
areas surrounding Antarctica are very scarce [8,9]. Conversely,
Cl has relevant nss sources, like volcanic emissions of HCl,
formation of gaseous HCl by chemical exchange between NaCl
and H2 SO4 in the atmosphere and contribution from longrange transport [10]. Mg2+ also has signicant crustal inputs.
Phytoplanktonic oceanic productivity can be reconstructed
by methanesulfonic acid (MSA) and nssSO4 2 stratigraphies
[3]. Indeed, these S-compounds mainly originate in oceanic
remote areas from atmospheric oxidation of dimethylsulde
(DMS), a gaseous compound which is produced by phytoplankton metabolic activity [11]. DMS is reported to play a
remarkable role in controlling global climate, via positive feedbacks through the formation of cloud condensation nuclei
(CCN) [12,13].
Finally, continental dust is the main source of Ca2+ in
Antarctic snow, especially in the internal regions, where seaspray contribution is low and long-range transport prevails.
Anyway, a simple equation system allows the calculations
of the nssCa2+ fractions, by knowing the sea-salt (ss) Na+
content and the Ca2+ /Na+ ratios in seawater and in the
crust [5]. Changes in nssCa2+ can be interpreted as changes
in atmospheric transport from dust source-areas, different
life-time of dust in the atmosphere (different wet scavenging efciency during the transport) and/or changes in dust
source-areas climatic conditions [3,4]. Since southern regions
of South America (especially Pampas and Patagonia [14]) are
supposed to have been the main dust source-areas for Antarctica, nssCa2+ stratigraphies have been used to depict changes
in hydrological cycles in southern South America in glacial
or interglacial conditions (probably due to shifts of climatic
belts), not excluding a signicant contribution from continental platforms, rising out of water by decreased sea level during
the glacial periods [15].
High-resolution chemical analysis of the EDC96 and EDC99
ice cores, drilled at Dome C (East Antarctica) in the framework
of the European project for ice coring in Antarctica (EPICA),
obtained paleo-climatic and paleo-environmental information spanning the last 850 000 years (850 kyr), covering eight
glacial-interglacial cycles, and improved our knowledge in
understanding relationships between global or hemispheric

191

climate forcings and environmental feedbacks [16,17]. This

data set doubled the previous ice core record [2] and can be
used in setting predictive climate models [18].
Sub-samples from EDC96 and EDC99 ice cores have been
shared in six European laboratories for ion chromatography
determination, in the frame of the EPICA chemistry consortium. Details of ice coring activity at Dome C and of a
round-robin exercise, aimed at testing the ability of each laboratory to give congruous results, are reported in previous
papers [19]. In recent papers [20,3,4], chemical stratigraphies of
chemical species used as climatic and environmental markers have also been shown and results have been discussed
in order to correlate the temporal trend of their concentrations or uxes in the last 850 kyr to climatic and environmental
changes. Notwithstanding this, ion chromatography methods
applied to Dome C ice cores were not described in detail up to
now.
Ion chromatography with conductivity detection and ultraclean sample preparation techniques is usually carried out for
the determination of ionic species in melt water from snow
and ice samples [19,21]. Chemical components are present in
the Antarctic snow and ice at very low concentration levels
(g L1 or sub-g L1 ), asking for analytical methods with high
accuracy, reproducibility and sensitivity. Besides, much care
must to be taken in order to avoid any possible contamination
from sample manipulation, storing and analysis. A further
difculty is the small sample-volume availability, due to the
desired high resolution and the large quantity of chemical,
physical and isotopic analyses performed on each ice core section. Indeed, high-resolution measurements require one ion
chromatography measurement about every 5 cm of ice core,
corresponding to time periods ranging from about 1 year (for
the shallow part of the ice core) to more than 40 years (at
the ice core bottom). This means that the sub-sample volume
for ion chromatography is usually around 45 mL. Since the
very low concentrations of chemical compounds in Antarctic
snow require the use of large sample loops (12 mL), this volume does not allow the use of a three-ion chromatography
system for separated determinations of cations, anions and
some organic anions (mainly MSA, due to its high climatic signicance), as usually used [2224]. Besides, the huge sample
number (about 80 000 along the entire 3200-m ice core) calls
for fast separations. For these reasons, a method improvement
was desired.
Here, we describe the setup of the ion chromatography method used in the Florence laboratory (as the Italian
pole of the EPICA chemistry consortium) for the determination of anions and cations in ice cores at g L1 or
sub-g L1 levels, which is based on an integrated system
of two parallel ion chromatographs fed with a ow analysis (FA) device. The system was optimised in order to
achieve high sensitivity and reproducibility and to minimise contamination risks. The improved method was not
applied to the analysis of short-chain organic acids (i.e.
acetate, formate, propionate), though their peaks are sufciently resolved (with retention times between F and
MSA peaks). In fact, carboxylic acids need still more efciency in decontamination procedures and specic ice core
sections, at lower depth resolution, were devoted to their
determination.

192

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Fig. 1 Flow chart of the ow analysision chromatography (FAIC) system.

Finally, new ion stratigraphies obtained from the EDC99 ice

core and covering the last 850 kyr, were shortly discussed.

2.

Experimental

2.1.

Ice core sample preparation and handling

For deep ice cores, a drilling uid is required to prevent

borehole collapse caused by wall-pressure [25]. After drilling,
ice-core bars are handled during logging, packing, shipment
and storage. Since all these steps imply unavoidable contamination, ice cores need to be decontaminated before analysis.
The cleaning techniques assume that contamination affects
only the exterior of the core, while the innermost section preserves the original chemical composition. Usually, ice-core
external layers are removed by mechanical scraping with
stainless steel knives [26] or by a series of washing-baths
with ultra-pure water [27,28]. In the EPICA project, the icecore decontamination procedure was carried out in two steps.
In the rst, squared-section core strips (about 3.4 cm 3.4 cm,
100 or 110 cm long) were obtained by cutting off external layers
with an electrical saw; then, the squared-section strips were
continuously melted lengthwise by a device (melter), whose
heating head was able to separate the water ow coming from
the internal part of the strip (circular sub-section of about
2.5 cm diameter) from that obtained by melting the external
part, potentially contaminated by the saw blade and by manipulation [29]. Only the melted sample coming from the inner
melter part was collected in pre-cleaned polystyrene accuvettes (coulter counter type) by an automatic autosampler able
to perforate their polyethylene caps. Accuvettes were sealed
in a polyethylene bag, frozen just after sub-sampling, and
stored at 20 C until analysis. This method allows contemporaneous decontamination, analysis and sub-sampling, with
minimum sample manipulation. A fraction of melted sample
was directly analysed by continuous ow analysis (CFA) and
fast ion chromatography (FIC) [30,31]. In particular, FIC measurements (chromatographic run time: 1 min) [18], allowed

the reconstruction high resolution stratigraphies of chloride,

nitrate and sulfate.
In eld CFA and FIC measurements need to be conrmed
and increased by classical ion chromatographic measurements carried out on accuvettes sub-samples. In the method
here proposed, sub-samples were left to melt still closed in
their accuvettes, locked in a specically designed autosampler
tray, under a class-100 laminar-ow hood. Ion chromatographic injection was carried out by aspirating the water
sample with a stainless steel needle connected to a remotecontrolled peristaltic pump. The sample ow was split two
ways before reaching the peristaltic pump, so that uxes
toward the two ion chromatography systems could be accurately controlled by the tube-diameter and rate of peristaltic
pump in the compression side (more reproducible and constant with respect to the aspiration side). Each samples ow
was ltered on a 0.45 m Teon membrane and injected
into a pre-concentrator connected to the ion chromatography
injection valve. A personal computer controlled autosampler movements, operation times and valve shifts (including
load/injection procedure inside the ion chromatography system). Several times per day, a sample volume actually injected
in every ion chromatography system was checked by weighing
the sample. All operations and ion chromatography measurements were carried out inside a class-10 000 clean room. Fig. 1
shows the ow chart of the ow analysision chromatography
(FAIC) system.

2.2.

Ion chromatography system and standards

Two ion chromatographs (DX2000 gradient and DX1000), were

used for simultaneous determination of anions and cations.
The DX2000 ion chromatograph was equipped with a GP50
gradient pump in order to provide a step elution for anion analysis. All chromatographic components were manufactured by
Dionex (Sunnyvale, CA, USA).
Step elution was preferred to gradient elution because
the background line quickly came back at and stable after
each eluent steps (Fig. 2). Gradient background line was more

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Fig. 2 (a) Cation chromatogram [1, Na+ (28 g L1 ); 2, NH4 +

(3 g L1 ); 3, K+ (5 g L1 ); 4, Mg2+ (3 g L1 ); 5, Ca2+
(2 g L1 )] and (b) anion chromatogram [1, F (5 g L1 ); 2,
CH3 SO3 (6 g L1 ); 3, Cl (60 g L1 ); 4, NO3 (25 g L1 ); 5,
SO4 2 (50 g L1 )].

leaning and very noisy, and was not suitable for ultra-trace
determinations. Due to the fast response of the background
line to the eluent steps, this kind of separation was also faster
than the gradient separation.
Anion separation was carried out with a Dionex Ion Pac
AG17 (4 50 mm) guard column, followed by a Dionex Ion
Pac AS17 (4 250 mm) analytical column. Conductivity suppression was performed by a Dionex ASRS-Ultra II 4 mm
self-regenerating suppressor [32]. Eluent (Na2 CO3 /NaHCO3
buffer, see Fig. 2 for concentration changes during the analysis) were stored in 2 L vessels pressurised at 10 psi using
high purity nitrogen, and ow-rates were kept at 1.6 mL min1 .
Samples (1.5 mL volume) were loaded into a pre-concentration
column (low-pressure Ion Pac TAC-2, 3 35 mm) by a peri-

193

staltic pump (1 mL min1 ). Fig. 2a shows the anion separation.

The nal step for nitrate and sulfate elution was followed by
a stabilization of the system to the injection (initial) conditions. The time schedule was strictly controlled, including the
time between the injections of consecutive samples, in order
to have maximum reproducibility in the component peaks.
Cation separation was obtained with a Dionex Ion Pac
CG12A (4 50 mm) guard column and a Dionex Ion Pac CS12A
(4 250 mm) separation column (eluent: H2 SO4 22.5 mN,
1.7 mL min1 ), following a standard ion chromatography
method. Dionex CSRS-Ultra II 4 mm self-regenerating suppressor was used before conductivity cell. Samples (1.5 mL volume)
were loaded into a pre-concentration column (low-pressure
Ion Pac TCC-2, 3 35 mm) by a peristaltic pump (1 mL min1 ).
Pre-concentration was necessary for both anion and cation
analysis due to the low concentrations (few g L1 to subg L1 ) of some components (especially F , MSA and NH4 + ).
Dionex Chromelion chromatography software was used for
instrumentation control and data acquisition.
Samples were loaded into the ion chromatographic systems by a 222 XL liquid handler automated sampler (Gilson,
Middleton, WI, USA) to minimise sample handling and to continue analysis overnight.
Standard solutions for calibrations were daily prepared
in pre-cleaned polyethylene vials diluting stock standard
solutions (1000 mg L1 ) purchased from Merck (Darmstadt,
Germany), with ultra-pure water (resistivity > 18 M; Milli-Q
system, by Millipore, Billerica, MA, USA). Ultra-pure water was
also used for eluent and regenerant preparation. Daily, 5-point
calibration curves were used for quantitative determination of
each ion. Blanks were always evaluated before and after the
calibration procedure.

3.

Results and discussion

3.1.

Method performances

Anion separation was carried out by a step method, which

is based on three fast variations of the eluent concentration during the chromatographic run. Eluent changes were
obtained by diluting 4.0 mM Na2 CO3 + 2.0 mM NaHCO3 buffer
with ultra-pure water (Table 1). In every separation stage, the
Na2 CO3 /NaHCO3 ratio was kept constant (2:1). Starting conditions, reached after a pre-established stabilization carried out
at the end of the previous run, were: 5% Na2 CO3 /NaHCO3 and
95% water. Here, a good separation of F and MSA peaks was
obtained. The retention time of F is sufciently higher than

Table 1 Eluent characteristics during the anion gradient method system

Time (min)
0
0.5
2
3
4
8
12

Milli-Q water (%)

95
95
95
85
10
95
95

2 mM NaHCO3 , 4 mM Na2 CO3 (%)

5
5
5
15
90
5
5

Note
Inject
Load
Inject
1 Step
2 Step
End

194

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Table 2 Analytical performances of the ow analysision chromatography

Ion

Na+

NH4 +

Upper limit of the linear range (g L1 ) 800

14
77
82
Sensitivity (nS g1 L)
Reproducibility (R.S.D.%)
2
1
0.09
0.15
Detection limit (g L1 )

K+

Mg2+

Ca2+

1300
700
700
36
56
33
1.9
1.5
1.5
0.13
0.11
0.11

CH3 SO3

16
14
78
19
0.6
1.9
0.01
0.1

Cl

NO3

SO4 2

600
78
1.5
0.15

3000
58
0.9
0.02

2500
43
1.8
0.08

Reproducibility were calculated as percentage standard deviation of 10 replicates of a 1 (F ), 10 (MSA, NH4 + , K+ , Mg2+ , Ca2+ ), 20 (Na+ , Cl , NO3 ,
SO4 2 ) g L1 standard.
Detection limit was calculated as three times the standard deviation of 10 replicate of a 0.5 g L1 (F ), 1 g L1 (Na+ , NH4 + , K+ , Mg2+ , Ca2+ , MSA),
2 g L1 (Cl , NO3 , SO4 2 ) standard.
Linear range was calculated with the slope variation.

water-dip, so that uoride can be determined at sub-g L1

levels (usual concentrations in Antarctic precipitations). The
rst eluent step was optimised to determine chloride. The
second step sped up the elution of nitrate and sulfate. The
separation run ended at 9.0 min. A nal step (4 min) was
necessary to stabilize separation conditions at the initial
values.
Cation separation was obtained by using 22.5 mN H2 SO4
as eluent under isocratic conditions at a ow rate of
1.6 mL min1 . These conditions were the best balance between
a short run time and a suitable Na+ /NH4 + resolution.
Although cation separation was faster than that of anions
(including the stabilization time), run times were unied
(12 min), in order to operate in parallel for each sample under
automated conditions.

Blanks obtained by injection of ultra-pure water did not

show any measurable peak in the anion chromatogram;
cationic blanks showed signals higher than the detection
limit only for calcium (mean blank value: 0.15 g L1 ), but signicantly lower than the minimum concentration found in
Antarctic samples (0.8 g L1 ).
Anion and cation method performances were evaluated in
terms of sensitivity (nS g1 L), reproducibility, linear range
and detection limit (3). These values are shown in Table 2.
The linearity range was calculated with a 15-point regression curve. Iterative steps allowed the calculation of the
slope of 5-points regression lines, starting from the lowest
concentration. Step by step, a more concentrated standard
solution was added in place of the lowest standard, and a
new regression (and slope) was calculated. The standard con-

Fig. 3 Linear range of Na+ and Cl calculated with the slope variation.

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Table 3 Comparative data for analytical performances

Detection limit (g L1 )

Time (min)

CH3 SO3 Cl NO3 SO4 2

0.02

0.01

0.3
0.3f
0.1

0.1

a
b
c
d
e
f

0.5
0.3
0.21
0.4
0.15

1.5
0.8
0.31
0.02
0.02

1.5
0.7
0.38
0.07
0.08

20
16f + 22
24
4
12

Reference

[33]a
[34]b
[35]c
[36]d
This studye

Gradient NaOH, injection volume 5.0 mL.

12.5 mM Na2 CO3 , injection volume 0.3 mL;
Gradient Na2 B4 O7 , injection volume 5.0 mL.
8 mM NaOH, injection volume 0.25 mL.
Multiple-step Na2 CO3 /NaHCO3 , injection volume 1.5 mL.
12 mM NaHCO3 , injection volume 0.8 mL.

centration giving a slope 10% lower than that obtained in the

initial regression was considered the upper limit of the linearity range. Examples for Na+ and Cl are shown in Fig. 3.
Reproducibility values were calculated as percentage standard
deviation of 10 replicates of a standard solution with concentrations ranging from 1 (F ) to 20 (Na+ , Cl , NO3 , SO4 2 )
g L1 . Reproducibility was always better than 2% for all the
components.
The detection limit was calculated as three times the standard deviation of 10 replicate injections of a standard solution
with a concentration of 0.5 g L1 (F ), 1 g L1 (Na+ , NH4 + , K+ ,
Mg2+ , Ca2+ , MSA) or 2 g L1 (Cl , NO3 , SO4 2 ); these values
were chosen because they are about 10 times higher than the
expected detection limit value.
Table 3 shows a comparison between anion detection limit
recently shown in the literature for polar ice and snow analysis, together with run times [3336]. The detection limits here
obtained are signicantly lower than those reported in previously methods for simultaneous determination of MSA, Cl ,
NO3 and SO4 2 , with a run time one half lower (12 min, with
respect to more than 20 min). Detection limit values are especially low for nitrate and sulfate (at least about 15 and 5 times
lower than previous methods, respectively). This performance
is mainly due to the low run time and the effect of the last
concentration step. We have to note that Jauhiainen at al.
[34] carried out the anion separations in two isocratic runs
(inorganic and organic anions), while Curran et al. [35] used
a gradient method lasting 24 min. Moreover, sample volumes
are usually higher in other experiments. For instance, Curran
et al. [35], injected samples three times larger, obtaining similar (MSA, Cl ) or higher detection limits. Cole-Dai et al. [36] in
their CFA-IC (continuous ow analysis with ion chromatography detection) method obtained detection limits close to those
yielded by our method, but their method is not able to measure
MSA content. Since MSA was supposed to be a unique marker
of past oceanic productivity and sea-ice cover reconstructions
[1,37], and uoride is used as specic marker of volcanic signatures [38]. Cation detection limits were similar to those
obtained by Doscher et al. [39], but we used a reduced sample volume (about 1:3). Even when accounting for the lower
sample volume, the ammonium detection limit is three times
lower than that reported by Doscher et al. [39]. Due to the short
running time, the detection limits reported by Cole-Dai et al.

195

[36] for cation are lower than those here reported, but the optimisation of our method aimed to determine anions (including
F and MSA) and cations in parallel, thus a faster method for
cations determination was not necessary.

3.2.

Analysis of EDC99 ice core

In the framework of the EPICA project, two ice cores were

drilled in the East Antarctic plateau [3]. EDC ice-core was
drilled at Concordia Station (Dome C, 75 06 04 S, 123 20 52 E,
3233 m a.s.l., about 1100 km from the seacoast, Pacic-Indian
sector). EDML ice-core was drilled at Kohnen Station (Dronning Maud Land, 75 00 S, 00 04 E; 2892 m a.s.l., about 600 km
from the seacoast, Atlantic sector).
EDC ice coring started in the 1996/97 Antarctic campaign
and reached a depth of 788 m (corresponding to 45 kyr before
present). At this depth the drill got stuck. This core is labeled
EDC96. A new drilling (EDC99) was started in the 1999/2000
summer eld and reached the bottom in January 2005 (about
3270 m). EDC99 has been analysed from 770 m, to have a
sufcient overlap with the old EDC96 core. EDC96-EDC99 synchronicity was determined by volcanic signature tting.
More information about the EPICA ice cores are reported in
previous papers [3,40].
The ion chromatographic systems developed here were
applied to the determination of anions and cations of the
EDC99 samples in the depth range 7703190 m. Discussions
on climatic and environmental information coming from a
detailed examination of the temporal trends of selected chemical markers recorded in the EDC ice cores have been already
published or are in preparation for thematic journals. Here
we show the stratigraphies of the ionic compounds for the
entire EDC96EDC99 record, with a summarized discussion of
the different pattern of the chemical markers in glacial and
interglacial periods, revealing different load and composition
of paleo-atmospheres in cold and warm climatic stages.
Fig. 4 shows the concentration/depth proles of several
chemical species measured along EDC96 and EDC99 ice cores.
Glacial and interglacial periods were identied by the prole
of the deuterium isotopic ratio (dD), used as a temperature proxy-marker: even numbers identify glacial periods and
odd numbers mark interglacial stages [41]. Note that deuterium isotopic ratio prole is plotted in reversed Y-scale, in
order to better highlight the inverse relationship with chemical stratigraphies. EDC proles show nine climatic cycles
over the last 850 kyr and provide the most direct and highly
resolved records of (especially) atmospheric parameters over
this timescale.
Each chemical prole shows a mirror-like pattern with
respect to the isotopic prole: higher concentrations in snow
layers were recorded during cold periods (identied by more
negative deuterium isotopic ratio values) and low concentration levels in the warm stages (less negative deuterium
isotopic ratio values). This pattern is due to the effect of climatic and environmental conditions during the glacial periods
(among others: higher aerosol source intensity and transport efciency towards Antarctica, and higher life-time of
atmospheric species due to the lower scavenging efciency).
Glaciological effects, like the lower dilution of dry-deposited
compounds by reduced snow accumulation rates in the cold

196

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Fig. 4 Chemical parameters proles concentration/age of climatic and environmental marker.

periods, can only partially account for the observed changes

[18,42]. The glacial/interglacial waving pattern is evident for
all cycles where the 100 kyr periodicity is generally kept [2].
A major feature of the isotopic and chemical records is the
different pattern shown in climatic periods older or younger
than 430 kyr, just before the marine isotopic stage 11 (MIS 11).
Indeed, the time between MIS 12 and MIS 11 is known as Mid-

Brunhes Event (MBE) and is supposed to be a boundary line

between climatic cycles characterized by similar duration of
warm and cold period and younger cycles where glacial conditions lasted for longer time.
Fig. 5 reports the mean concentrations (expressed as g L1 )
of Na+ , Mg2+ and Cl , mainly originated from sea-spray,
calculated in interglacial (black and white striped column)

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

Fig. 5 Concentration averages for sodium, magnesium

and chloride in interglacial and glacial periods along the
whole Dome C ice core.

and glacial (solid grey column) periods, accordingly to the

deuterium isotopic ratio prole [40]. Vertical bars indicate
standard deviation values. As previously observed, glacial concentrations are higher than values measured in warm stages.
The average of the ratios between glacial and interglacial
mean concentrations along the whole record (nine cycles) was
about four for Na+ , six for Mg2+ and seven for Cl . These values
are higher than glacial/interglacial ratios of accumulation rate,
estimated to be close to 2 for each climatic cycle. This means
that differences in glacial/interglacial concentrations cannot
to be explained only by variations in accumulation rates, but
changes in source intensity and/or transport efciency played
a signicant role in increasing the primary marine aerosol in
glacial atmospheres over Antarctica. Indeed, more sea-spray
production is believed to occur in glacial conditions, mainly
caused by stronger winds and, possibly, by a larger extent of
sea ice [3]. The relationship between sea-spray production and
sea ice extent was supposed to be related to the formation of
frost ower on the surface of new sea ice [43]. Frost owers are sea salt crystals, depleted in Na2 SO4 by precipitation
of mirabilite, that form in the sea-water trapped in the seaice pockets when pack-ice becomes compact. The role of frost
ower formation in increasing sea-spray aerosol in coastal
areas has been conrmed, but its contribution to sea-spray
budget in the internal areas of Antarctic plateau is still controversial.
Glacial increase of sea-spray at Dome C could be attributed,
aside from stronger winds acting on open sea-water surface
and frost ower formation, to a longer life-time of primary
aerosol in the atmosphere, due to the reduced hydrological
cycle (less efcient scavenging by wet deposition).
The higher temperature differences between glacial and
interglacial conditions in the last four climatic cycles is
reected in the increased concentration ratios of sea-spray
species. In fact, in the four cycles after MIS 11, the mean
ratio between glacial/interglacial concentrations was ve
for Na+ and eight for Mg2+ . Such values are higher than

197

those measured in the whole 850 kyr period. On the contrary, glacial/interglacial ratios before MIS 11 (850430 kyr B.P.)
show lower mean values: three and four for Na+ and Mg2+ ,
respectively, demonstrating that the oldest interglacials experienced colder temperatures (see also deuterium isotopic ratio
proles). Unlike sodium and magnesium, the chloride concentration does not show such a marked difference of mean
glacial/interglacial ratios before and after MIS 11 (6.3 in the
four younger cycles and 5.8 for the older ones). This evidence can be explained by considering that the sea-spray
is not the only source of chloride at Dome C. Surface-snow
and aerosol analysis showed that both NaCl particles and
gas-phase HCl are scavenged at Dome C by wet and dry precipitation or by supercial uptake [9]. HCl mainly originates
by exchange reactions of NaCl with sulfuric acid in coastal
areas and follow different transport pathways than NaCl in
reaching inner regions of Antarctica. Since HCl is affected by
post-depositional remission into the atmosphere when acidic
snow content is high (as it occurs in interglacial periods) [44],
chloride concentrations in warm stages are very low. In contrast, the higher dust content in glacial times makes snow less
acidic, so that larger quantities of HCl are kept in the snow layers. As a consequence, chloride shows high glacial/interglacial
mean ratios both before and after MIS 11.

4.

Conclusions

An integrated ICFA method allowed the simultaneous determination of main and trace cations (Na+ , NH4 + , K+ , Mg2+ ,
Ca2+ ) and anions (F , MSA, Cl , NO3 , SO4 2 ) in the EDC99
ice core at g L1 or sub-g L1 levels with high sensitivity
(ranging from 19 nS g1 L for MSA to 82 nS g1 L for NH4 + ),
good reproducibility (relative standard deviations better than
2%), low detection limits (ranging from 0.01 g L1 for F to
0.15 g L1 for Cl and NH4 + ) and short run time (12 min for
both anion and cation separation). Detection limits obtained
here were lower (or much lower, for nitrate and sulfate) than
those obtained in analogous methods for ice core analysis
able to measure MSA and F together with inorganic anions.
Detection limits for anion analysis were also comparable or
better than those reported by Cole-Dai et al. [36], despite this
method is not able to measure in the same run F , MSA,
Cl , NO3 and SO4 2 . Moreover, sample contamination was
highly reduced, thanks to reduced manipulation during subsampling and analysis. Blank values were below the detection
limit for all the components with the exception of calcium.
However, Ca2+ blanks were at least one order of magnitude
lower than concentrations measured along the EDC99 ice core.
These performances were obtained by using a relatively
small sample volume (1.5 mL for each separation). Relatively
fast separations and low sample volume needs allowed for
the increase in measurements resolution, thus obtaining highresolved chemical stratigraphies along the entire ice core. The
resulting record is the longest (850 kyr) and the most resolved
(few centimetres, i.e. few years) available record of climatic
and environmental markers from an ice core to date.
The interpretation of chemical proles changes during different climatic regimes and, especially, during fast climate
variations, like glacialinterglacial transitions and stadial

198

a n a l y t i c a c h i m i c a a c t a 6 0 3 ( 2 0 0 7 ) 190198

or intestadial stages (e.g. DansgaardOeschger events in

Northern hemisphere; Antarctic Isotopic Maxima, in South
hemisphere), will help in understanding forcing factors and
feedback processes which triggered the Earths climatic system in the last one million years, in order to set and test
reliable predictive models of our climatic future.

[12]
[13]
[14]
[15]

Acknowledgements
This research was nancially supported by the MIUR-PNRA
program through a cooperation agreement among the PNRA
consortium, ENEA and Venice University, in the framework
of the 9.1 (Chemistry of Polar Environments) and 5.1 (Glaciology) programs. This work is a contribution to the European
Project for ice coring in Antarctica (EPICA), a joint ESF (European Science Foundation)/EC scientic programme, funded by
the European commission and by national contribution from
Belgium, Denmark, France, Germany, Italy, the Netherlands,
Norway, Sweden, Switzerland, and the United Kingdom. The
scientic activity at Dome C beneted from the support of the
French-Italian base Concordia Station. This is EPICA publication n 187.

[16]
[17]
[18]
[19]

[20]

[21]
[22]
[23]
[24]

references

[25]
[26]

[1] M.A. Curran, A.S. Palmer, T.D. Van Ommen, V.I. Morgan, K.L.
Phillips, A.S. Palmer, Science 302 (2003) 1203.
[2] J.R. Petit, J. Jouzel, D. Raynaud, N.I. Barkov, J.M. Barnola, I.
Basile, M. Bender, J. Chappellaz, M. Davis, G. Delaygue, M.
Delmotte, V.M. Kotlyakov, M. Legrand, V.Y. Lipenkov, C.
Lorius, L. Pepin, C. Ritz, E. Saltzman, M. Stievenard, Nature
399 (1999) 429.
[3] E.W. Wolff, H. Fischer, F. Fundel, U. Ruth, B. Twarloh, G.C.

Littot, R. Mulvaney, R. Rothlisberger,

M. De Angelis, C.F.
Boutron, M. Hansson, U. Jonsell, M.A. Hutterli, F. Lambert, P.
Kaufmann, B. Stauffer, T.F. Stocker, J.P. Steffensen, M. Bigler,
M.L. Siggaard-Andersen, R. Udisti, S. Becagli, E. Castellano,
M. Severi, D. Wagenbach, C. Barbante, P. Gabrielli, V. Gaspari,
Nature 440 (23) (2006) 491.
[4] H. Fischer, F. Fundel, U. Ruth, B. Twarloh, A. Wegner, R.
Udisti, S. Becagli, E. Castellano, A. Morganti, M. Severi, E.

Wolff, G. Littot, R. Rothlisberger,

R. Mulvaney, M.A. Hutterli,
P. Kaufmann, U. Federer, F. Lambert, M. Bigler, M. Hansson, U.
Jonsell, M. De Angelis, C. Boutron, M.L. Siggaard-Andersen,
J.P. Steffensen, C. Barbante, V. Gaspari, P. Gabrielli, D.
Wagenbach, Earth Planet. Sci. Lett. 260 (12) (2007)
354.

[5] R. Rothlisberger,
R. Mulvaney, E.W. Wolff, M.A. Hutterli, M.
Bigler, S. Sommer, J. Jouzel, Geophys. Res. Lett. 29 (20) (2002)
1963.
[6] R. Traversi, S. Becagli, E. Castellano, O. Largiuni, A. Migliori,
M. Severi, R. Udisti, Int. J. Environ. An. Ch. 84 (2004) 457.
[7] R. Udisti, S. Becagli, E. Castellano, R. Traversi, S. Vermigli, G.
Piccardi, Ann. Glaciol. 29 (1999) 77.
[8] D. Wagenbach, F. Ducroz, R. Mulvaney, L. Keck, A. Minikin,
M. Legrand, J.S. Hall, E.W. Wolff, J. Geophys. Res. 103 (D9)
(1998) 10961.
[9] S. Benassai, S. Becagli, R. Gragnani, O. Magand, M. Proposito,
I. Fattori, R. Traversi, R. Udisti, Ann. Glaciol. 41 (2005) 32.
[10] A.J. Aristarain, R.J. Delmas, Atmos. Environ. 36 (4) (2002)
765.
[11] E.S. Saltzman, Ocean/atmosphere cycling of dimethylsulde,
in: R.J. Delmas (Ed.), Ice Core Studies of Global

[27]

[28]
[29]
[30]

[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]

[42]
[43]
[44]

Biogeochemical Cycles, vol. I 30, 65, Springer-Verlag, Berlin,

Heidelberg, 1995, NATO ASI Series I.
R. Charlson, J.E. Lovelock, M.O. Andreae, S.G. Warren, Nature
326 (6114) (1987) 655.
G.P. Ayers, J.L. Gras, Nature 353 (1991) 834.
I. Basile, F.E. Grousset, M. Revel, J.R. Petit, P.E. Biscaye, N.I.
Barcov, Earth Planet. Sci. Lett. 146 (1997) 573.
R. Rothlisberger, R. Mulvaney, E.W. Wolff, M. Hutterli, M.
Bigler, S. Sommer, J. Jouzel, Geophys. Res. Lett. 29 (2002) 1963.
M. Legrand, P.A. Mayewski, Rev. Geophys. 35 (1997) 219.
M. Legrand, E.W. Wolff, D. Wagenbach, Ann. Glaciol. 29
(1999) 66.
R. Traversi, S. Becagli, E. Castellano, A. Migliori, M. Severi, R.
Udisti, Ann. Glaciol. 35 (2002) 291.

G.C. Littot, R. Mulvaney, R. Rothlisberger,

R. Udisti, E.W.
Wolff, E. Castellano, M. De Angelis, M. Hansson, S. Sommer,
J.P. Steffensen, Ann. Glaciol. 35 (2002) 299.
E. Castellano, S. Becagli, J. Jouzel, A. Migliori, M. Severi, J.P.
Steffensen, R. Traversi, R. Udisti, Global Planet. Change 42
(14) (2004) 195.
C.F. Buck, P.A. Mayewski, M.J. Spencer, S. Whitlow, M.S.
Twickler, D. Barrett, J. Chromatogr. 594 (1992) 225.
R. Udisti, S. Becagli, R. Traversi, S. Vermigli, G. Piccardi, Ann.
Glaciol. 27 (1998) 391.
R. Udisti, E. Barbolani, G. Piccardi, Anal. Chim. 81 (1991)
325.
R. Udisti, S. Bellandi, G. Piccardi, J. Fresenius, Anal. Chem.
349 (1994) 289.
P.G. Talalay, N.S. Gundestrup, Mem. Natl. Inst. Polar Res. 56
(2002) 171 (special issue).
W. Chisholm, K.J.R. Rosman, C.F. Boutron, J.P. Candelone, S.
Hong, Anal. Chim. Acta 311 (1995) 141.
J.C. Priscu, E.E. Adams, W.B. Lyons, M.A. Voytek, D.W. Mpgk,
R.L. Brown, C.P. McKay, C.D. Takacs, K.A. Welch, C.F. Wolf, J.D.
Kirshtein, R. Avci, Science 286 (1999) 2141.

D.M. Karl, D.F. Bird, K. Bjorkman,

T. Houlihan, R. Shackelford,
L. Tupas, Science 286 (1999) 2144.

A. Sigg, K. Fuhrer, M. Anklin, T. Staffelbach, D. Zurmuhle,

Environ. Sci. Technol. 28 (1994) 204.
R. Rothlisberger, M. Bigler, M. Hutterli, S. Sommer, B.
Stauffer, H.G. Junghans, D. Wagenbach, Environ. Sci.
Technol. 34 (2000) 338.
R. Udisti, S. Becagli, E. Castellano, R. Mulvaney, J. Schwander,
S. Torcini, E. Wolff, Ann. Glaciol. 30 (2000) 20.
S. Rabin, J. Stillian, V. Baretto, K. Friedman, M. Toofan, J.
Chromatogr. 640 (1993) 97.
M. Legrand, M. De Angelis, F. Maupetit, J. Chromatogr. 640
(1993) 251.
T. Jauhianinen, J. Moore, P. Peramaki, J. Derome, K. Derome,
Anal. Chim. Acta 389 (1999) 21.
M.A. Curran, A.S. Palmer, J. Chromatogr. A 919 (2001) 107.
J. Cole-Dai, D.M. Budner, D.G. Ferris, Environ. Sci. Technol. 40
(2006) 6764.
U. Jonsell, M.E. Hansson, M.L. Siggaard-Andersen, J.P.
Steffensen, J. Geophys. Res. 112 (2007) D14313.
C.U. Hammer, H.B. Clausen Jr., C.C. Langway, Climate
Change 35 (1997) 1.
A. Doscher, M. Schwikowski, H.W. Gaggeler, J. Chromatogr. A
706 (1995) 249.
Community Members EPICA, Nature 429 (2004) 623.
J. Jouzel, F. Vimeux, N. Caillon, G. Delaygue, G. Hoffmann, V.
Masson-Delmotte, F. Parrenin, J. Geophys. Res. 108 (D12)
(2003) 4361.
B. Delmonte, J.R. Petit, V. Maggi, Clim. Dynam. 18 (8) (2002)
647.
A.M. Rankin, E.W. Wolff, S. Martin, J. Geophys. Res. 107 (D23)
(2002) 4683.
P. Wagnon, R.J. Delmas, M. Legrand, J. Geophys. Res. 104 (D3)
(1999) 3423.