An Optical Dye Method for Continuous Determination of Acidity in Ice

Sep 1, 2016 - Helle Astrid Kjær†, Paul Vallelonga†, Anders Svensson†, Magnus Elleskov L. Kristensen†, Catalin Tibuleac†, Mai Winstrup†, a...
1 downloads 3 Views 2MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

An optical dye method for continuous determination of acidity in ice cores Helle Astrid Kjær, Paul Vallelonga, Anders Svensson, Magnus Elleskov L. Kristensen, Catalin Tibuleac, Mai Winstrup, and Sepp Kipfstuhl Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b00026 • Publication Date (Web): 01 Sep 2016 Downloaded from http://pubs.acs.org on September 5, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

Environmental Science & Technology

An optical dye method for continuous determination of acidity in ice cores Helle Astrid Kjær,∗,† Paul Vallelonga,† Anders Svensson,† Magnus Elleskov L. Kristensen,† Catalin Tibuleac,† Mai Winstrup,† and Sepp Kipfstuhl‡ Centre for Ice and Climate, Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark, and Alfred Wegener Institute, Bremerhaven, Germany E-mail: [email protected]

Abstract

1

2

The pH of polar ice is important for the stability and mobility of impurities in ice

3

cores and can be strongly influenced by volcanic eruptions or anthropogenic emissions.

4

We present a simple optical method for continuous determination of acidity in ice

5

cores based on spectroscopical determined color changes of two common pH-indicator

6

dyes, bromophenol blue and chlorophenol red. The sealed-system method described

7

here is not equilibrated with CO2 , making it simpler than existing methods for pH

8

determination in ice cores and offering a 10-90% peak response time of 45 sec and a

9

combined uncertainty of 9%. The method is applied to Holocene ice core sections from

10

Greenland and Antarctica and compared to standard techniques such as Electrical

11

Conductivity Measurement (ECM) conducted on the solid ice, and electrolytic melt

12

water conductivity, EMWC. Acidity measured in the Greenland NGRIP ice core shows

13

good agreement with acidity calculated from ion chromatography. Conductivity and ∗

To whom correspondence should be addressed CIC ‡ AWI †

1

ACS Paragon Plus Environment

Environmental Science & Technology

80

Page 2 of 30

°

N

Humboldt North NEEM 70

°

NEGIS NGRIP

N

GRIP 60

°

pH

3.0

3.5

4.0

DYE

N

70

GISP2

°

°

W

60 ° W

4.5

°

50 ° W

5.0

°

40 W

° 30 W

5.5

20

W

10

W

6.0

6.5

7.0

14

dye-based acidity H+ dye are found to be highly correlated in the Greenland NEGIS

15

firn core (75.38◦ N, 35.56◦ W), with all signals greater than 3σ variability coinciding

16

with either volcanic eruptions or possible wild fire activity. In contrast, the Antarctic

17

Roosevelt Island ice core (79.36 ◦ S, 161.71 ◦ W) features an anti-correlation between

18

conductivity and H+ dye , likely due to strong influence of marine salts.

19

Introduction

20

Polar ice cores provide continuous archives of past climate and atmospheric conditions, 1–6

21

with the possibility for reconstructions at high temporal resolution. Greenland deep ice

22

cores cover the last glacial-interglacial cycle, 7,8 whereas some Antarctic ice cores archive up

23

to 800,000 years of past climate. 9 Ice core impurities occur in solid, liquid and gaseous phases

24

in the ice, with the pronounced seasonality of some impurities 10 allowing the potential for

25

establishing ice core chronologies based on the counting of annual layers of deposition. 11,12

26

The detection of pH in ice cores provides valuable information on past environmental

27

processes such as volcanic activity, 13 anthropogenic pollution (particularly NOx and SO4 )

28

and Arctic haze, 14 but the pH in ice cores is also influenced by mineral dust (calcium car-

29

bonates), marine biogenic emission products such as methanesulfonic acid (MSA), as well as

30

ammonium and organic acids from terrestrial biogenic emissions and biomass burning. 15,16

31

Further pH controls both the solubility of ions and the rate of many chemical reactions. pH 2

ACS Paragon Plus Environment

Page 3 of 30

Environmental Science & Technology

32

has been invoked as the major factor controlling in-situ production of CO2 in Greenland

33

ice cores and is thus a critical barrier to reconstructing CO2 greenhouse gas records for the

34

Northern Hemisphere. 17

35

Several methods for the detection of acidity in ice cores have been developed. On solid

36

ice the Electrical Conductivity Measurement (ECM) signal is mainly dominated by strong

37

acids in the ice and provides a fast way to detect volcanic horizons. 18,19 An empirical re-

38

lationship between ECM and [H+ ] has been established. This relationship depends on the

39

ECM instrument, the instrument operator, ambient conditions, and the impurity content of

40

the ice (eg. dust, major ions, sea salts, organic and/or inorganic acids). 14,18,20–23 The ECM

41

sensitivity to alkaline ice is low.

42

Dielectric profiling (DEP) is also applied to solid ice and is mostly associated with the

43

H+ and Cl– content of the ice, but during periods of high alkalinity a relationship to NH4+

44

has also been observed. 23–25 In Greenland Holocene ice DEP is mainly controlled by the ice

45

acidity. 23 The acidity can also be determined using pH probes. 16,26

46

Previous studies report a pH range of 4.49-6 in ice cores (polar ice); with more acidic

47

values during the second half of the 20th century, 16,27 especially in ice cores located closer

48

to pollution sources as Europe and North America 14,27 and in ice containing volcanic aerosol

49

deposits. 13 pH values in the range of 5.05-5.5 were found in snow deposited between 2005-

50

2009 at the NEEM site in Northwest Greenland. 26 The seasonality of pH in this snow is

51

not very pronounced, but slightly less acidic values were found in the fall compared to the

52

spring. In spring, the higher acidity occurs when Arctic haze, polar ozone destruction and

53

transport of pollutants from mid-latitudes are peaking. 26,28,29 Similar annual variation in pH

54

has been observed in the Greenland Humboldt North record (Figure 1). 16 More alkaline pH

55

values are observed in ice with high dust concentrations eg. Greenland glacial ice. 20,30

56

Hence, methods for continuous detection of pH in ice should preferably be sensitive to

57

resolve pH at least within the Holocene range from 4.49 to 6 as well as being capable of

58

resolving seasonal variations. Such a techniques should ideally also be able to determine pH

3

ACS Paragon Plus Environment

Environmental Science & Technology

Page 4 of 30

°

°

0

15 12 ° 0 E

0° W

°

70 S

90 W

90° E

°RICE 80 S

W

90° S

60

°

60 ° W

E

°

60 ° N

15

°

70 ° N

B21 H.N. B18 NGRIP NEGIS B16

180 W

0 12

80 ° N

E



30 ° E

° 30 W

W

50 ° W

W

°

°

10

W

30

70 °

Figure 1: Location of ice cores measured (red) and mentioned (black) in this study. Left: The Greenland NEGIS firn core, the NGRIP ice core and the Humboldt North (HN) firn core 16 as well as the B16, B18 and B21 firn cores. 27 Right: The Antarctic RICE ice core. 59

values lower than 4.5 as expected in dust-laden glacial ice.

60

The polar ice core matrix is relatively pure and unsaturated in CO2 and ions. Therefore,

61

the ice will readily take up CO2 from the atmosphere when it is melted. The dissolution of

62

CO2 in the melt water poses the risk of rapidly altering the sample pH. 15 The use of bench

63

top laboratory techniques to determine pH in melted ice samples require precise monitoring

64

of ambient CO2 concentrations for corrections (as well as pressure and temperature, more

65

in the supplementary material). Alternatively, the acidity of the sample can be artificially

66

lowered to reach a level where the dissociation of carbonic acid is minimized. 15

67

Continuous Flow Analysis techniques (CFA) are often applied to ice core analysis, allow-

68

ing rapid throughput and high resolution for a range of analytes. 31–33 The Copenhagen CFA

69

system uses a sealed debubbler to separate air from melted water 33 and thus is sealed from

70

ambient air, which offers an advantage for pH determination, allowing the melted ice sample

71

to be handled and analysed soon after melting and without interaction with the ambient 4

ACS Paragon Plus Environment

Page 5 of 30

72

Environmental Science & Technology

laboratory air.

73

Recent CFA methods using pH electrodes have implemented an equilibration of the sam-

74

ple to a known concentration of CO2 . 16,34 Ensuring the complete calibration of the sample

75

with a precisely known concentration of CO2 involves complicated CO2 bubbling and de-

76

bubbling systems which introduce sample dispersion and limit the temporal resolution of

77

the analytical system. 16,34 The benefit of such a CO2 calibration is minor for the acid con-

78

centrations of H+ found in ice and segmentation with air or helium may lead to inconsistent

79

results. 34

80

Continuous pH measurements using electrodes are further limited by their relatively

81

slow response time at the low ion concentrations of polar ice. Even with the addition of

82

electrolytes, such as KCl (0.01 M) or NaCl (0.002 M) to improve the ionic strength, 16,34

83

electrodes require longer calibration times than instant dye methods. In the method by

84

Pasteris et al. 16 as much as 85 seconds was required to respond to acidic signals or 120

85

seconds to respond to alkaline signals and even longer times to return to baseline were

86

observed. While this is mainly due to the CO2 bath, the response of the electrode also adds

87

to the long response time.

88

In this study, we present a simple optical method for continuous determination of acidity

89

in polar ice cores that avoids the drawbacks associated with implementing a CO2 bath

90

in a continuous electrode-based pH analysis system. The method is adapted from that

91

proposed for oilfield formation waters by Raghuraman et al. 35 by selecting indicator dyes

92

sensitive to the pH range of polar ice and by minimizing the dye concentration to limit

93

sample dilution. The proposed method features similar rapid response times to both acid

94

and alkaline solutions. The method is sensitive over a broad range of pH values and ionic

95

strengths, can be applied to firn and ice cores equally well and while this method is not CO2

96

equilibrated it is in a sealed system and only affected by the CO2 within the ice itself giving it

97

similar performance in terms of concentration and better performance in terms of response

98

time to other continuous pH determination systems. The technique is thus compared to

5

ACS Paragon Plus Environment

Environmental Science & Technology

99

electrode-based methods particularly well suited to field-based analytical campaigns where

100

laboratory gas supplies mays not be easily obtained.

101

Materials and methods

102

Dye-based pH determination relies on the principle that the spectra of a dye will vary

103

according to the sample pH. Depending on the pH the fraction of the base and acid forms

104

will change. Due to the optical nature of the measurement, dye-based determination can be

105

affected by instabilities resulting from bleaching of the dyes, light source intensity fluctuations

106

and attenuation changes in the optical fiber. These fluctuations can be accounted for by

107

monitoring multiple wavelengths and/or by regular calibration with a set of standards. 36

108

Using a fixed concentration of two dyes, the pH can be determined as a function of the

109

optical density at two wavelengths. 35 A similar colorimetric method using dyes has also

110

recently been applied for the detection of pH in sea water. 37

111

In this study the dyes used were bromophenol blue (from yellow at pH 3.0 to purple at

112

pH 4.6), and chlorophenol red (from yellow at pH 4.8 to violet at pH 6.7). The dyes were

113

chosen based on their response to the ranges of pH expected in polar ice.

114

Experiment setup

115

A schematic of the set up is shown in Figure 2 with details provided in Table 1. Optimal

116

dye concentration was found with a sample flow rate of 0.9 ml/min and 0.15 ml/min of

117

dye mixture (see Supplemetary Figure S1). The dye mixture is made by dissolving 0.025 g

118

of bromophenol blue and 0.025 g of chlorophenol red in 500 mL of milliQ water (Millipore

119

A10 Advantage, 18.2 MΩ.cm) and adding 100 µL Brij L23 (30%w/w). The sample and dye

120

are mixed in a 1 metre long mixing coil kept in a 65◦ C heat bath, after which the mixture

121

passes through a 6 cm-long Accurel (PP S6/2, Membrana GmbH, Germany) gas permeable

122

membrane to allow the removal of excess air bubbles created during the heating. Finally

6

ACS Paragon Plus Environment

Page 6 of 30

Page 7 of 30

Environmental Science & Technology

123

detection of light intensity takes place in an absorption cell with a 2 cm path length. Brij

124

L23 was added to the dye solution to facilitate the removal of air bubbles from the absorption

125

cell, in case they were not removed via the accurel membrane.

126

A white LED (6000 mcd, 18◦ dispersion, FIA Lab Inc. USA) was used as a light source

127

and an Ocean Optics USB 2000 spectrometer was used to measure the light intensity. An

128

integration time of 6-80 ms was used depending on the age and concentration of the dye-

129

mixture. The optimum response was found at a wavelength of 589 nm, with a second peak

130

at 450 nm (see supplementary material Figure S2).

131

The sample is heated to 65◦ C to ensure temperature stability in the analytical system

132

and from a theoretical point of view a minimal content of CO2 in the sample stream. In

133

theory, CO2 uptake in water is temperature dependent and thus the effect on the closed

134

system was investigated by changing the temperature in the heat bath between 25 and 75◦ C.

135

No significant effect of temperature was however observed when measuring a similar set of

136

standards under varying temperature conditions (see supplementary material Figure S3 and

137

S4). This suggests that the pH detection system described here is not affected by ambient

138

CO2 concentrations to any observable extent, essentially the system is sealed. However we

139

can not rule out the expected CO2 -temperature effect being reduced due to the influence of

140

the Brij L23 surfactant added to the dye mixture. Surfactants have been shown to alter the

141

air-water CO2 exchange rates. 38,39

142

Standards

143

For the analysis of rainwater and other low conductivity samples, diluted standard buffers

144

are more suitable than technical buffers, because they are more representative of the low ionic

145

strength of the sample and have minimal memory effects. 40 Previous studies have found that

146

diluted HCl or H2 SO4 are the most suitable standards for calibrating a system as described

147

here. 16,40,41 For alkaline standards, however, a strong base such as NaOH does not accurately

148

reflect the properties of polar ice as it rapidly takes up CO2 from the laboratory air, yet no 7

ACS Paragon Plus Environment

Environmental Science & Technology

Table 1: pH CFA detection system parameters Flow rates: Sample Dye Reagent mixing length Reagent mixing temperature Absorption path length Absorption wavelength 1 Absorption wavelength 2 Spectrometer integration time Light source

Response time (5-95%) Analytical uncertainty (pH) 149

0.9 mL/min 0.15 mL/min 1.0 m 65◦ C 2 cm 450 nm 589 nm 6000-8000 µs white LED 6000 mcd 18 ◦ 45 s 8.7% (5.8%)

alternative for alkaline standards has been found. 16,40,41

150

The standards used in this study were prepared from standard solutions of HCl (1.001

151

M, Fluka) and NaOH (1.002 M, Fluka) diluted to concentrations of 19.4 µM, 9.8 µM and 4.9

152

µM for HCl and 19.5 µM and 9.8 µM for NaOH, with corresponding pH values of the HCl

153

and NaOH concentrations being 4.71, 5.00, 5.26, 6.80 and 6.23 pH respectively. Laboratory

154

milliQ was calculated to have a pH of 5.747 assuming a CO2 concentration of 385 ppm,

155

the laboratory atmospheric pressure of 1030 hPa and a detection temperature of 65◦ C. The

156

theoretical calculation of pH of the standards follows the procedure of Pasteris et al. 16 and

157

is shown in the supplementary material. 42

158

The stability of the standards and their potential response to changes in laboratory CO2

159

concentrations have been evaluated. The level of CO2 in the laboratory during a typical ice

160

core measurement campaign was recorded over a day and was found to be highly variable,

161

with peaks up to 1000 ppm CO2 . Such variations could theoretically have a significant impact

162

on the unbuffered HCl and NaOH standards. For standards with low (≤5µM) HCl or NaOH

163

concentrations it can alter the H+ concentrations by more than 13% (see supplementary

164

material). Such variability is greatly diminished for more concentrated HCl standards, to

165

merely 0.8% for a standard of 20µM H+ at 65◦ C. 8

ACS Paragon Plus Environment

Page 8 of 30

Page 9 of 30

Environmental Science & Technology

Figure 2: The continuous flow setup for detection of pH. Ice is melted on a melt head (MH) in a freezer kept at -20◦C. An injection valve (IV) in combination with a selection valve (SV) for standards (St) and blank (Bl) allows for running either. The sample line is de-bubbled (D) to remove the air bubbles naturally occurring in the ice or firn and the gas (GE) is used for gas extraction measurements. The water line is split into various analytical channels (DO), and the pH line which is the only one shown here. The sample (Sa) is mixed with dye then passed through a 1.0 m heated (65 ◦ C) mixing coil before an air permable membrane, an accurel (Ac), is reached to remove airbubbles formed during heating. Intensity (A) at 450 nm and 589 nm is measured in a 2 cm absorption cell after which the line goes to waste (W). Peristaltic pumps are represented by arrow boxes with flow rates (in mL/min) indicated by the number within the box. 166

As shown in Figure 3-A and in supplementary Figure S5 little variability was observed

167

in a standard over approximately 32 hours (two measurement days). The uncertainty on

168

a fit based on the standards from the two day campaign, is below 1.5 % for detection at

169

589 nm. We found that the response at wavelength 450 nm is low, while the wavelengths

170

586 nm and 589 nm, show stronger signals and higher stability (see supplementary Figure

171

S5). This is mainly due to stronger light intensity at this peak in the spectra. The 450 nm

172

signal was also more prone to drift over time due to ageing of the Bromophenol blue dye (see

173

supplementary Figure S5), confirming the importance of regular standard calibration if 450

174

nm is to be used for pH determination, as well as the need for a calibration similar to that

175

I done for absorption (-log10 I0 ), to correct for baseline drift, though strictly speaking the term

9

ACS Paragon Plus Environment

Environmental Science & Technology

176

absorption does not make sense for I≥I0. Here we are just determining the color change by

177

I . The difference between the the mathemathical expression also used for absorption -log10 I0

178

response (slope in Figure 3-B) to NaOH standards and HCl standards is most likely related

179

to different response strength in the two dyes between more acid and less acid standards.

180

Due to the possible indluence of dissolved CO2 from the ambient atmosphere, the repro-

181

ducibility and stability of acid standards has been evaluated. HCl standards were produced

182

in laboratories in Denmark and New Zealand using different water purification systems, and

183

we do not see an offset in either the firn (NEGIS, Greenland) or ice (RICE, Antarctica)core

184

data collected. These tests suggest that the effect of lab air on the standards either stayed

185

stable or was insignificant. To further test the variability of standards, similar standard

186

concentrations of H+ were made using both HCl and H2 SO4 (see supplementary Figure S7).

187

Identical results were found for the two acid standards, leading us to conclude that the acid

188

standards were made reproducibly, that fluctuations of CO2 in the laboratory were of minor

189

importance, and finally that the technique reported here is insensitive to ambient CO2 , when

190

determining Holocene ranges of polar ice.

191

Uncertainties

192

All measurements of firn and ice were calibrated using a linear fit between -log10( II0 ) and

193

standards of 19, 9.8 and 4.9 µM HCl as these were found to best represent the values found

194

in the ice. The uncertainty on the HCl concentration of the standards is ±5%. An additional

195

uncertainty on the standards relates to the fact that the CO2 concentration in the lab varies

196

over a day, however the effect of this variance on the standards is low, of the order of 0.8

197

% for a HCl concentration of 20 µM, however increasing with more alkaline samples (see

198

supplementary material). The uncertainty on the linear fit used to calibrate samples is ∼

199

0.08 ± 0.08 µM H+ (∼2.8 %) over the range of concentrations found in the NEGIS firn core.

200

Signal drift can be observed if the analytical system is not sufficiently pressure-decoupled

201

or if the peristaltic pump tube is not replaced regularly. System pressure fluctuations can 10

ACS Paragon Plus Environment

Page 10 of 30

Page 11 of 30

Environmental Science & Technology

1.6

0.12

NaOH19.5 µM

a = −0.005 NaOH U = 1.1% a

0.08 0.04 0

2.4 HCl 9.8 µM

A 0

HCl 4.9 µM

B

HCl19.4 µM

100

200

300

400 500 Time (s)

0

2

−log10(I/I )

Intensity (I) x 104

NaOH

NaOH 9.8 µM

600

700

800

900−20

a = −0.0036 HCl U = 0.7%

−0.04

a

HCl

0 + H (µ M)

20

−0.08

Figure 3: Panel A: Two examples of standards with concentrations 19.4 µM HCl, 19.5 µM NaOH, 9.8µM HCl, 9.8 µM NaOH and 4.9µM HCl ran with 7 hours interval and measured at 589 nm. The lighter colors are the second run. Panel B: Example of the -log10( II0 ) of HCl and NaOH concentrations relative to milliQ, NaOH concentrations are referred to as negative [H+ ] on the x-axis. The measurements were performed over two days using the same dye (day 1 squares, day 2 stars). Standards were prepared three times a day. A linear relation is plotted for standards of NaOH (dotted) and for standards of HCl (full) separately. The text aNaOH refers to the slope of the fit found using all NaOH standards (negative H+ ) and UaNaOH to the uncertainty of the fit, similar for HCl standards. 202

cause ∼ 0.20 ± 0.24 µM H+ (6.5%) uncertainty and is determined as the difference between

203

using the baseline before and after measurements of a standard series. As the technique is

204

sensitive to the mixing ratio of sample and dye, a steady drift can also be caused by differen-

205

tial wearing of tubing over time. Further, as mentioned earlier, the ageing of Bromophenol

206

Blue dye can cause drift primarily in the 450 nm signal. Drifts due to aging of pump tubing

207

are observed to be linear and thus can be corrected in a simple manner.

208

The total combined uncertainty from HCl standard concentration, uncertainty on fit

209

of standards and uncertainty related to baseline drift of the described pH technique thus

210

accounts to ∼ 8.7% or 5.7% if baseline drift is excluded determined by propagation of

211

uncertainty. Such a resut is very similar to the one sigma variation (∼6%) found on the acid

212

calibration slope for the 589 nm wavelength over the two month-long RICE measurement

213

campaign. For the campaign results, a higher standard deviation on the alkali calibration

214

slope was found and that may be related to faster CO2 lab air uptake in the NaOH standards 11

ACS Paragon Plus Environment

Environmental Science & Technology

215

Page 12 of 30

(∼9%) (Figure S7).

216

This uncertainty does not include the effect of different CO2 concentrations in the ice

217

itself, which though minimized by heating is not fully accounted for by this new method. The

218

difference between the CO2 used for calibration (385 ppm) and the CO2 in glacial samples

219

(180 ppm) could cause the results to be biased toward lower acidity (see Figure S3). Please

220

note that it is also possible using this method to include a CO2 bath similar to that used

221

by Pasteris et al. 16 before detection in the absorption cell, however with a trade-off between

222

accuracy and resolution. In this study we have optimized for high resolution.

223

We found that the 450 nm wavelength was not appropriate for the study of firn/ice cores

224

due to substantial level of noise at this wavelength. However, as the 450 nm and 589 nm

225

wavelengths respond inversely to pH variations (see supplementary Figure S2 and S5), but

226

identically to flow variations, we used the trend of the two wavelengths as a diagnostic tool

227

to identify unexpected flow changes. These could be attributed to the presence of air bubbles

228

in the sample line and/or absorption cell or also due to upstream pressure fluctuations in

229

the continuous flow stream.

230

Results

231

The optical dye method was tested on three different ice cores offering different types of polar

232

ice as well as different ages. The method was tested on Holocene sections of the Greenland

233

NGRIP icecore 7 to validate absolute measurements against results from an ion budget. The

234

method was used on the Greenland firn core NEGIS 43,44 to test the method for firn effects

235

and to compare results to those obtained from ECM, DEP and further to confirm that the

236

method found the 1970’s anthropogenic peak found in other recent cores. 16 Finally the dye

237

method was tested on the Antarctic coastal ice core RICE, which contrary to the central

238

Greenland ice cores has a high sea salt content. A map of the all three cores is shown in

239

Figure 1.

12

ACS Paragon Plus Environment

Page 13 of 30

Environmental Science & Technology

240

Greenland ice core (NGRIP)

241

To validate the optical dye detection method H+ dye was determined in a section of the Green-

242

land NGRIP ice core and results were compared to the H+ Ionbudget concentration calculated

243

based on the ionic budget and to the electric melt water conductivity (EMWC). H+ Ionbudget

244

was based on ion chromatography analyses using equation 1. Organic acids were however not

245

determined by IC in that part of the NGRIP ice core and is not included in the IC derived

246

H+ , but the organics are assumed to have concentration of 0.3 µeq from both formate (9.4

247

ng/g) and acetate (7.3 ng/g), which are the concentrations found in the Greenland GRIP

248

ice core between 1800 and 1980 AD, and similar values have been found throughout the

249

Holocene part of the GRIP ice core. 45,46

250

The two sections shown in Figure 4 cover the depths from 580.9-582 metres and 1309.8 to

251

1310.1 metres (∼3250 and ∼9090 yrs b2k). The deeper section includes a volcanic eruption.

252

+ The directly measured H+ dye concentrations are in good agreement with the HIonbudget for

253

high concentrations of H+ . A mean difference of about 0.8 µM is observed for both the two

254

sections shown and up to 2 µM is observed in the deeper NGRIP section between the two

255

methods on the fringe of the large volcano, while the shallower section agrees better.

256

The offset observed for low concentrations could be due to the amount of carbonic acid

257

in the water having a larger influence in more alkaline waters, but could also be caused by

258

uncertainties in the ion budget, which can vary from about 5% for SO43– to 30% for K+ and

259

Ca2+ . 47 For example it is found that the CFA determined NH4+ concentrations is lower than

260

the IC determined NH4+ concentrations by about 0.3 µeq NH4+ . This may result partly

261

from the difference in resolution, helping to explain the discrepancy for between the two

262

methods at low concentrations.

263

Further as the formate and acetate concentrations were not determined directly in these

264

samples. It is observed that a small peak at 581.7 meters depth is found by the higher

265

+ resolution EMWC and H+ dye methods, but not detectable in the IC determined Hionbudget

266

probably partly due to the lower resolution imposed by IC sampling, but it could also be 13

ACS Paragon Plus Environment

Environmental Science & Technology

Page 14 of 30

267

due to organic acids being increased at this dept. The organic acids was not included

268

in the Hion+ determination. Following the formate-NH4+ relation found by Legrand and

269

De Angelis 46 and further used by Pasteris et al. 16 , up to 0.4 µeq/L formate can be present

270

for NH4+ concentrations of 0.6µM further explaining the offset between the two methods.

271

Discrete IC derived NH4+ is also known to be effected by lab air ammonia, which the sealed

272

+ CFA derived NH4+ is not, explaining the difference between the H+ dye and the HIonbudget .

273

Further it is also worth noting that for high peak concentrations of NH4+ the H+ dye concen-

274

tration decreases slightly as would be expected from the alkaline NH4+ , while in comparison

275

the EMWC stays high. Finally, the EMWC and Hdye+ resemble each other well and the ratio

276

2− -1 found between the two is EMWC= 0.37 µS cm−1 mol−1 ·[H+ 0.91) and dye ] + 0.45 µScm (R

277

thus not far off the 0.35 µS cm -1 mol -1, exepected if H+ was the only ion present in the water

278

(see supplementary Figure S11).

279

In conclusion though differences between the ion derived H+ and the H+ concentration

280

found using the dye method presented here do exist, they are within the combined uncer-

281

tainties of the two methods, and the great resemblance between EMWC and H+ dye show that

282

likely the ion derived H+ is more influenced by such uncertainties than the H+ dye , at least for

283

this section of the NGRIP data.

+ [HIonbudget ] = [F − ] + [MSA− ] + [Cl− ] + [NO3− ] + [SO42− ] + [org − ] +

− [Na ] −

[NH4+ ]

+

2+

(1) 2+

− [K ] − [Mg ] − [Ca ]

284

Greenland firn core (NEGIS)

285

H+ dye was determined in the upper part of a firn core from the North East Greenland Ice

286

Stream (NEGIS, 75.38◦ N, 35.56◦ W) covering the period 1900 AD-2005 AD. 43 The firn core

287

was analysed on the Copenhagen CFA system, 33 optimized for high resolution measurements

288

of impurities in firn or ice samples with a cross-sectional area of 35 mm x 35 mm. Calibrations 14

ACS Paragon Plus Environment

Page 15 of 30

Environmental Science & Technology

15 + dye H+ Ion budget

+

H (µ eq/L)

10

5

EMWC

5

2.5

0

0

EMWC (µ S/cm)

H

2 0.80

0.78

0 −2 5

15

3

+

6

NH4 CFA (µ eq/L)

SO

3− 4 IC

(µ eq/L)

+

∆ H (µ eq/L)

4

1

1 581.2

581.5 Depth(m)

581.8

1309.8

1310.1

Figure 4: NGRIP data for two separate depths. Top) H+ dye determined using the optical dye + method (blue) shown together with the H ionic budget from IC measurements (red) and EMWC (black). Center) The difference (∆H+ ) between the the H+ ionic budget from IC measurements and H+ dye . The number and dotted line indicate the mean difference between the two methods for each depth section. Bottom) Discrete IC measurements of SO42 (purple) and continuously determined ammonium (green). Grey vertical lines indicate volcanic eruptions, while red lines indicate peaks in ammonium.

15

ACS Paragon Plus Environment

Environmental Science & Technology

289

were made before and after melting 3 samples of firn of 55 cm length. The core chronology

290

was established by counting annual peaks e.g. winter peaks of Na+ , 6,10,48 between volcanic

291

reference tiepoints identified by DEP and ECM analysis.

292

Acidity in the NEGIS record shows similar levels to that observed in other Greenland ice

293

core records. The mean acidity level is 2.3 µM H+ dye prior to the mid 1940’s, after which it

294

strongly increases reaching mean values of 7 µM H+ dye in the mid 1970’s, and then decreasing

295

16 to 1.2 µM H+ dye by 2004 (Figure 5). This trend is similar to that reported by Pasteris et al.

296

for the Greenland Humboldt ice core located in North-western Greenland 615 km west of

297

NEGIS. We expect contributions of 4.1-4.8 µM H+ from SO42– and 1.6 to 2.1 µM from NO3–

298

based on the maximum 1970’s concentrations of 2.08-2.39 µM SO42– and 1.6-2.1 µM NO3–

299

observed in the nearby Greenland firn cores (B16, B18, B21). 27 These increases in SO42– and

300

NO3– account for the 5 year mean increase in the H+ dye signal observed during the 1970’s in

301

the NEGIS firn core.

Page 16 of 30

302

The concentration of H+ dye in the NEGIS firn core varies throughout the year with higher

303

levels occurring in winter. As noted by Pasteris et al. 16 higher acidity levels are observed

304

during late winter/early spring when the Arctic haze phenomenon is strongest. 28,29 Similar

305

seasonality is also observed albeit weaker in NEEM snow pit samples as measured by a

306

pH electrode. 26 The seasonality of acid deposition (see supplementary material Figure S9,

307

bottom) for the period 1950 to 2000 is more pronounced than for the period 1900-1950

308

and shifts towards the spring months rather than the winter. The acidity concentrations

309

and trends reported here for NEGIS closely follow those found in the Greenland Humboldt

310

North firn core, for which a pH-electrode based detection method was used (shown in red in

311

top panel of Figure 5).

312

Most of the acidity peaks in the NEGIS firn core record can be consistently attributed to

313

volcanic activity, though acid peaks in the industrial era can not be excluded from originating

314

from anthropogenic sources. In table 2 extreme H+ dye concentrations found in the NEGIS

315

record (defined as those exceeding a running five-year 3σ (99.8%) probability interval) are

16

ACS Paragon Plus Environment

Page 17 of 30

Environmental Science & Technology

316

compared to contemporaneous volcanic eruptions or extreme wildfire events. Most of the

317

enhanced H+ dye events are linked to other volcanic eruption proxies such as ECM, DEP and/or

318

SO42– . 49–51 Volcanic eruptions that have increased NEGIS H+ dye levels above 3σ variability

319

threshold include Redoubt (Alaska, 1989 and 1967), El Chichon (Mexico, 1982), Hekla and

320

Katla (Iceland, 1913, 1918 and 1971), Raikoke (Kuril Islands, 1924) and Katmai (Alaska,

321

1912) (Figure 5). Although Agung (Indonesia, 1963) is commonly observed in Antarctic

322

records it is not prominent in Greenland due to the elevated background level resulting from

323

industrial sulphate emissions in the Northern Hemisphere. Table 2: Years of enhanced H+ dye level compared to a five year average in the NEGIS firn core for the recent century and suggested cause as well as the enhanced concentration as compared to a five year mean (∆H+ ). Event, location and date wildfire (2002) 52 Redoubt, Alaska (1989) Augustine, Alaska (1986) El Chichon, Mexico (1982) Hekla, Iceland (1970) Redoubt, Alaska (1966 or 1967) wildfire(1950) 16,53 wildfire (1943) 16 or Unknown NH Raikoke, Kuril Islands (1924) Katla, Iceland (1918) Hekla, Iceland (1913) Katmai, Alaska (1912) Ksudach, Kamchatka (1907)

Years of high H+ dye (exceeding 3 STD) 2002 (sum) 1995(spr) 1989(spr) 1986(spr) 1982(spr) 1971(spr) 1967(win) 1949(aut) 1944 (spr) 1925(spr) 1918(win) 1913(spr) 1912 (spr) 1907(sum)

∆H+ dye (µeq) 3.6 2.5 6.2 3.4 4.3 3.8 3.0 2.7 1.8 0.2 1.8 4.2 1.9 1.3

324

Ice core pH can be influenced by organic acids such as formic acid, produced by biomass

325

burning events, though in most cases the formic acid will be neutralized by the NH4+ ar-

326

riving simultaneously from the fire event. Only two NEGIS peaks exceeding 3σ variability

327

can possibly be attributed to biomass burning events dated to 1950 and 2002 AD. The

328

Chinchaga fire in 1950 (3,500,000 acres) is likely the largest northern American wildfire in

329

recent history 53 and in 2002 the RodeoChediski Fire in Arizona (467,066 acres) 54 and/or the 17

ACS Paragon Plus Environment

Environmental Science & Technology

Page 18 of 30

330

Biscuit fire in Oregon state (499,750 acres) 52 could be attributed to the acid peak. Pasteris

331

et al. 16 identified other large biomass burning events, particularly in 1943 and 1951, although

332

those events do not exceed 3σ variability in the NEGIS acidity record. However, without

333

other measurements related to fires, such as black carbon or vanilic acid, the source of these

334

particular peaks remain speculation.

335

Relating pH to ice core conductivity

336

Electrolytic melt water conductivity (EMWC 33 ), electrical conductivity measurements (ECM 18 )

337

and dielectric profiling (DEP 23 ) are standard ice core measurements that were also obtained

338

for the NEGIS firn core (Figure 5). All of those records are related to pH in some manner,

339

although it is important to distinguish that EMWC is measured on the melted sample and

340

hence is influenced by soluble ions and CO2 , whereas ECM and DEP are determined on solid

341

ice. A good correlation (r>0.6) is found between the H+ dye concentration and the EMWC,

342

ECM and DEP in the NEGIS firn core (see supplementary material Figure S8 and Table

343

S1). Furthermore, all four records show similar trends over the 20th century with a peak in

344

the 1970’s due to anthropogenic acid deposition.

345

Similar to NGRIP the NEGIS EMWC can be almost completely reproduced by a linear

346

−1 fit of EMWC− 0.32 µScm−1 mol1 ·[H+ (R2 = 0.88). Within uncertainties dye ] + 0.54 µScm

347

the slope is thus close to the expected 0.35 µScm−1 mol1 molar conductivity expected by H+

348

at 25◦ C. This suggests that the melt water conductivity in the NEGIS firn core is mainly

349

governed by the H+ content and that other ions such as sea salts are insignificant (mean

350

level of Na+ is in the order of 1.08 µM for NEGIS with peak values of 4.30 µM).

351

For lower concentrations of H+ dye the relationship to EMWC is less correlated (