Nitrate

Apart from N₂O, a long-lived trace gas encapsulated in air bubbles, nitrate and, to a lesser extent, ammonium are the major nitrogen species trapped in polar ice. Nitrate ice-core data are relatively easy to analyze, so a large number of them exists, including those from deep ice cores spanning the last ice age. Nitrate in polar ice, however, remains an enigma (Reference Legrand and KirchnerLegrand and Kirchner, 1990; Reference Wolff and DelmasWolff, 1995), while its records have been proposed as a proxy of solar activity (modulated by odd nitrogen in the upper stratosphere), lower stratospheric temperatures and free tropospheric sources including lightning. Based on the similarity between the atmospheric nitrate seasonal cycle and the seasonal cycle of ²¹⁰Pb at Mawson, Reference Savoie, Prospero, Larsen and SaltzmanSavoie and others (1992) suggested a continental source for nitrate present in the boundary layer of this coastal Antarctic site. Another study at Neumayer failed to confirm this similarity between nitrate and ²¹⁰Pb (Reference Wagenbach, Wolff and BalesWagenbach, 1996). Furthermore, a strong depletion of the ¹⁵N/¹⁴N ratio is seen in atmospheric nitrate samples collected at Neumayer in early spring/summer, arguing in favor of a non-tropospheric contribution to the nitrate budget of the boundary layer at this coastal Antarctic station (Reference Wagenbach, Wolff and BalesWagenbach, 1996).

The mean seasonal nitrate pattern in the atmospheric boundary layer of coastal Antarctic sites (Fig. 2) reveals a quasi-constant background level near 10 ng m⁻³ from April to June and a seasonal maximum from spring to midsummer. Based on concurrent measurements of ⁷Be, ¹⁰Be, ²¹⁰Pb and δ¹⁵ N in aerosols at Neumayer and tritium in precipitation at Halley (Fig. 2), Reference Wagenbach, Legrand, Fischer, Pichlermayer and WolffWagenbach and others (1998) concluded that the main part of the spring/summer seasonal nitrate maximum is stratospheric in origin, with a secondary peak in winter and late summer due to sedimentation of polar stratospheric clouds (PSCs) and stratospheric air-mass intrusion, respectively. Provided that the air-snow relationship of nitrate is well understood, the possibility exists of reconstructing some aspects of the past chemistry and climatology of the stratosphere, including the nitrogen stratospheric budget and the intensity of the stratosphere-troposphere exchanges. An additional complication is the recent finding that alkyl nitrates, possibly of marine origin, dominate the odd nitrogen budget at coastal sites in summer (Jones and others, 1999). This reopens the question of possible non-stratospheric sources for material that ends up as nitrate in ice.

Fig. 2. Left: Mean annual nitrate cycles at Dumont d’Urville (DDU; 1991-94), Neumayer (NM; 1983-95), and Mawson (Maw; 1987-90) (adapted from Reference Wagenbach, Legrand, Fischer, Pichlermayer and WolffWagenbach and others, 1998). Right: Seasonal pattern of the stratospheric aerosol input at Neumayer indicated by the enhancement of the ¹⁰Be/⁷Be (1983-85) or ⁷Be/²¹⁰Pb (1983-93) ratios; PSC-related material indicated by enhanced tritium content (normalised to the individual annual means) in Halley precipitation (1983-93); and mean annual cycle of δ15 Nm nitrate (1986-92) at Neumayer (shaded area here refers to the overall range of δ15 N commonly observed in tropospheric nitrate) ( adaptedfrom Reference Wagenbach, Legrand, Fischer, Pichlermayer and WolffWagenbach and others, 1998). Vertical bars refer to the ±16 standard deviation in the interannual variability.

A previous study of nitrate present in the Vostok ice (0-2200 m depth) deposited over the last climatic cycle (Reference Legrand, Lorius, Barkov and PetrovLegrand and others, 1988) shows that nitrate concentrations, close to 20ngg⁻¹ during the warm climatic stages, were enhanced in glacial ice. High nitrate levels in glacial ice coincide with large enhancements of mineral dust content. Thus, Reference Legrand, Lorius, Barkov and PetrovLegrand and others (1988) concluded that the enhancement of the nitrate content in glacial ice may reflect efficient scavenging of HN0³ by dust particles at these times. Investigation of nitrate in shallow snow layers at Vostok revealed a rapid decrease of concentrations with depth, from 100 ngg⁻¹ close to the surface to 20 ngg⁻¹ at 2 m depth (Reference Mayewski and LegrandMayewski and Legrand, 1990). This observation strongly suggests that a large part of the nitrate initially trapped in the Vostok snow deposits as volatile nitrate like HN0₃ , and is thereafter slowly re-emitted toward the atmosphere via processes that are not yet clearly understood (see Reference Mulvaney, Wagenbach and WolffMulvaney and others, 1998 for updated discussion). Because the impurity content of the glacial ice is very different from that at the present, we can speculate that the intensity of the HN0₃ reemission may have changed at Vostok during ice ages with respect to the situation at present (Reference Wolff and DelmasWolff, 1995; Reference Legrand, Leopold, Domine, Wolff and BalesLegrand and others, 1996). Thus, an alternative explanation for the enhancement of nitrate levels in glacial ice at Vostok may be that the enhanced dust input in the ice deposited during the ice age limited the re-emission of nitrate with respect to the situation at present.

In Figure 3, we report the nitrate and non-sea-salt calcium (denoted Ca²⁺ _exc) profiles for Vostok, extended to the last two climatic cycles (0-2755m depth). Figure 3 confirms that a nitrate level close to 20 ngg⁻¹ characterised the Vostok interglacial ice and that higher nitrate contents were present in dust-rich ice (i.e. high Ca²⁺ _exc content) deposited during the two ice ages. Whatever the reason for the observed increase of nitrate in glacial ice containing a large amount of dust, it remains difficult to interpret the meaning of these dust-rich ice layers for the atmospheric nitrogen oxide budget over Antarctica in the past. Nevertheless, as seen in Figure 3, numerous Vostok ice samples deposited during the ice ages contain <20ngg⁻¹ of nitrate in spite of Ca²⁺ _exc levels higher than those observed in interglacial ice (i.e. 1.7 ± 1.9ngg⁻¹). Hence, we examine the Vostok nitrate profile by considering only samples containing a relatively low dust amount. For those samples containing <15ngg⁻¹ of Ca²⁺ _exc, we tentatively try to estimate the residual effect of the dust input. As reported in Figure 4, a linear regression of N0₃- and Ca²⁺ _exc yields [NO₃ ^–] = 4.0 + 1.542 [Ca²⁺ _exc.] (r² = 0.65 for 54 samples) in samples containing >20 ng g⁻¹ of Ca²⁺ _exc. In order to estimate the residual effect of dust on nitrate levels present in ice layers containing <15 ng g⁻¹ of Ca²⁺ _exc, we correct total nitrate content by subtracting 1.542 [Ca²⁺ _exc]. The remaining part of nitrate (denoted NO₃- _res. may be more representative of changes of the nitrogen oxide cycle in the past. As seen in Table 1, N_{O3- res.} levels are lower in ice corresponding to cold climates and low snow-accumulation rates. Reduced snow-accumulation rates during the ice age may cause a larger re-emission of nitrate after deposition. However, Reference Legrand and DelmasLegrand and Delmas (1988) reported a mean nitrate content of 18.5 ± 8.0 ng g-1 in Holocene ice deposited at Dome C, a site characterised by a mean annual snow accumulation of 34 kg m⁻². They also observed lower nitrate concentrations (–16 ng g⁻¹) in Dome C glacial ice containing <15 ng g⁻¹ of calcium. These nitrate levels in the Dome C ice are very similar to those observed at Vostok, suggesting that the nitrate content of ice deposited over the high Antarctic Plateau is not modulated by small changes of the snow-accumulation rate.

Fig. 3. Vostok depth profiles of nitrate and non-sea-salt calcium (Ca²⁺ _exc) contents along with temperature change (δT) with respect to present (depicted by the deuterium profiles from Jouzel and others, 1996) over the last two climatic cycles. Horizontal dashhed lines refer to mean Holocene levels of nitrate (20 ng g₁) and Ca²⁺ exc. (3ng g¹)

Fig. 4. Nitrate vs Ca2⁺ _exc. levels of the Vostok ice deposited over the last two climatic cycles. Triangles refer to samples being dust-rich (Ca2⁺ _exc levels >20ng g-1), circles to samples having <20 ng g-1 of Ca2⁺ exc. Dashed line refers to the linear regression between _{NO3 ^–} and Ca2⁺ exc of Vostok samples having >20 ng g-1 of Ca2₊ exc.

Thus, lowered snow-accumulation rates during the ice age cannot explain lowered N_O3 ^- res. levels observed in Vostok glacial ice. While not proofs, the preceding considerations suggest that atmospheric nitrate levels may have been reduced during the ice age with respect to present climatic conditions over central Antarctica. Such a conclusion is consistent with the expected effects of global cold climatic conditions on the stratospheric odd nitrogen budget. At present, the stratospheric nitrogen oxide reservoir is mainly built up from N₂0 surface emissions. One key parameter modulating the main natural N ₂ 0 source (soils) is related to the humidity of soils. Thus, as with methane, a decrease of this source is expected during the ice age due to drier climatic conditions over source regions (Reference CiceroneCicerone, 1989). Measurements of N₂0 in ice air bubbles indeed indicated a decrease of tropospheric concentrations during the ice age (Reference Zardini, Raynaud, Scharffe and SellerZardini and others, 1988; Reference Leuenberger and SiegenthalerLeuenberger and Siegenthaler, 1992). On the other hand, because of a decreasing effect of greenhouse trace gases such as C0₂ during the ice age, we can speculate that the lower polar stratosphere was warmer than at present. This last effect may have induced a less frequent occurrence of PSCs within the Antarctic vortex during ice ages. The water-vapor budget of the stratosphere itself may also have changed during cold climates, because of lowered methane concentrations. Less frequent occurrence of PSCs during ice ages may have weakened the exchange of odd nitrogen from the stratosphere into the polar troposphere and hence decreased nitrate concentrations of glacial ice.

When considering only samples having a moderate dust content, the Vostok glacial ice contained some 20% (down to 50%, if we correct for the dust effect) less nitrate than ice deposited during warm climates. This is consistent with a stratospheric origin of nitrate present in the Antarctic boundary layer which may have been weakened under cold climatic conditions by a lowered N₂0 stratospheric loading and/or by less efficient transport from the stratosphere due to reduced frequency of PSCs in a warmer lower stratosphere. Clearly further work is required to confirm these conclusions, which should include the study of deep ice cores with significantly different snow accumulation rates.

Ammonium

Unlike nitrate, ammonium is present in Antarctic ice at levels close to a few ng g"¹ or less. Thus, reliable ammonium ice-core data remain very rare and no geochemical information has yet been extracted by studying ammonium in deep Antarctic ice cores. However, by itself the observation of an ammonium level close to lngg⁻¹ in Antarctic ice appeared inconsistent with ammonium atmospheric levels of tens of ng m⁻³ observed at Mawson (coastal Antarctica) by Reference Savoie, Prospero, Larsen and SaltzmanSavoie and others (1992). Atmospheric measurements carried out in the framework of the CAASC programme show that very high ammonium atmospheric concentrations (several hundreds of ng m⁻³) are related to the proximity of penguin colonies from which ammonia is emitted following the bacterial decomposition of uric acid (Reference Legrand, Ducroz, Wagenbach, Mulvaney and HallLegrand and others, 1998). In Antarctic stations which are not exposed to penguin-colony emissions, such as Neumayer, mean annual ammonium concentrations are close to a few ng m⁻³, with a maximum in January-March in phase with the increase of biogenic sulphur species. Based on these Neumayer data, Reference Legrand, Ducroz, Wagenbach, Mulvaney and HallLegrand and others (1998) suggested that ammonium present in these regions may be related to ammonia oceanic emissions occurring south of 50° S in summer. Finally, the ammonia neutralisation of acidic sulphur aerosols remains very weak, as depicted by NH₄ ⁺/(MSA+SO4^2-) weight ratios close to 2.5% (i.e. a molar ratio of 0.13) observed in summer aerosols at Neumayer.

Previous studies of ammonium in the deep Vostok (Reference Legrand, Lorius, Barkov and PetrovLegrand and others, 1988) and Dome C (Reference Legrand and DelmasLegrand and Delmas, 1988) ice cores indicate that ammonium concentrations range from 1 to 3 ng g"¹. Although these two studies tend to show a slightly higher mean ammonium content during the ice age with respect to the one in interglacial ice, the ion-chromatography (IC) working conditions employed for these measurements did not guarantee the reliability of a change of 1-2 ngg⁻¹ of ammonium. In Figure 5, we report ammonium data from Vostok that were obtained using new IC working conditions that have been improved. First, all ice samples were melted just before analysis in airtight glass bottles to prevent the acidic meltwater from uptaking ammonia present in the ambient atmosphere of the laboratory. Moreover, these measurements were made using a new (CS12) separator column recently developed by the Dionex Ion Chromatography Corporation, which unlike the previously used columns efficiently separates ammonium and sodium peaks. The overlapping of ammonium and sodium peaks which occurred when using the old cation separator columns was the main cause of uncertainties because of increased sodium content in glacial ice with respect to interglacial ice. From Figure 5 which shows measurements of ammonium in Vostok ice samples, obtained using the CS12 column, we observe that ammonium levels were three times lower in interglacial ice (0.6±0.35ngg⁻¹) than in ice deposited during very cold climatic conditions (2.0 ± 0.5 ng g⁻¹). These data indicate, for the first time, that the ammonium profile of deep ice cores contains a climatic signal. Figure 5 shows that higher ammonium levels observed in glacial ice coincide with high non-sea-salt sulphate (denoted SO₄ ²- _exc) levels. The increase of SO₄ ²- ^e _xc. levels in Vostok glacial ice is accompanied by an increase of MSA level, suggesting that DMS emissions may have been strengthened during the ice age (Reference Legrand, Feniet-Saigne, Saltzman, Germain, Barkov and PetrovLegrand and others, 1991). The linear regression of NH₄ ⁺ and SO₄ ²- _exc. yields [NH₄ ⁺] = -0.14 + 0.009 [SO₄ ²- exc] (r² = 0.60 for 241 samples), indicating a mean NH₄ ⁺ /(MSA + SO₄ ²- _exc) weight ratio close to 1% in the Vostok ice deposited during both warm and cold climates (Fig. 5). Such ammonium data in the Vostok ice are consistent with a marine biogenic origin of ammonium and indicate that the degree of ammonia neutralisation of sulphur acidic species previously remained very low. If real, these preliminary ammonium levels of old ice deposited over central Antarctica suggest that continental sources do not contribute significantly to the Antarctic ammonium budget.

Fig. 5. Ammonium concentrations of Vostok ice measured by using a CS12 Dionex column (see text) as a function of ST (temperature change with respect to the mean Holocene value depicted by the deuterium profiles from Petit and others, 1999d) and non-sea-salt sulphate contents.