1 Introduction

Previous work has shown that the electrical properties of polar glacier ice are very different from those of temperate glacier ice or pure laboratory ice. In geophysical resistivity surveys, Reference RöthlisbergerRöthlisberger (1967) found that the electrical conductivity of cold polar glacier ice was orders of magnitude larger than that of temperate glacier ice. Dielectric properties of deep cores taken from polar glaciers have been studied by several authors (Reference Paren, Whalley, Jones and GoldParen 1973, Reference Maeno and KuroiwaMaeno 1974, Reference Fitzgerald and ParenFitzgerald and Paren 1975, Reference MaenoMaeno 1978). From these different studies, it was shown that relaxation times of the dielectric dispersion were shorter than those obtained with pure laboratory or temperate glacier ice. The activation energy of the principal relaxation time is around 0.23 eV in polar ice compared with around 0.58 eV in pure single crystals. Reference Fitzgerald and ParenFitzgerald and Paren (1975) found that the electrical behaviour of polar ice is similar to that observed for single crystals doped with HF. It is difficult to explain the unusual electrical properties by impurity effects alone (Reference Fitzgerald, Glen and ParenFitzgerald and others 1977). The most recent idea put forward was based on the influence of CO₂ adsorbed at the ice surface (Reference Paren and GlenParen and Glen 1978). This assumption seemed to be confirmed by a reported high CO₂ content of polar ice. However, recent measurements have clearly shown that the CO₂ content of polar ice is small compared with values given previously (Reference Delmas, Ascencio and LegrandDelmas and others 1980). By measuring the static electrical conductivity and the sulphate content of Dome C and D 10 ice samples, Reference Maccagnan, Barnola, Delmas and DuvalMaccagnan and others (1981) found a relationship between the two studied parameters, but impurities did not seem to explain entirely the electrical properties of these samples. In other respects, internal friction measurements made in the 1 Hz range on ice from Dome C have shown that anelastic properties of polar ice, just like electrical properties, are very different from those of pure ice (Reference Vassoille, Maï, Perez, Tatibouet, Duval and MaccagnanVassoille and others 1980). The aim of this paper is firstly to present the internal friction spectrum versus temperature of polar ice from D 10 (the relaxation phenomenon is observed at a particularly low temperature) and secondly to compare electrical and mechanical measurements made on the same polar ice.

2 Experimental methods

3 Experimental results

Thirteen samples cut from blocks situated between depths of 126 and 900 m from Dome C were studied: as the results have been published elsewhere (Reference Vassoille, Maï, Perez, Tatibouet, Duval and MaccagnanVassoille and others 1980), only one example of the anelastic properties of these samples is shown in Figure 1(a) (curve A). The features previously observed with lowfrequency internal friction measurements made on pure ice (Reference Vassoille, Tatibouet, Perez and GobinVassoille and others 1974) can again be seen: (i) in the high-temperature range, a peak associated with the movement of grain boundaries (Reference Perez, Maï, Tatibouet and VassoillePerez and others 1979), and (ii) at lower temperatures, a relaxation peak generally interpreted in terms of reorientation of water molecules by means of rotational defects (Reference Vassoille, Maï and PerezVassoille and others 1978).

Fig.1(a) Internal-friction spectra versus temperature; curve A: specimen from Dome C 680 m deep, curve B: pure ice.

In the case of the former feature, no difference is observed between pure and polar ice; however, in the latter, the temperature T_p of the relaxation peak in polar ice (145 K) is lower than in the case of pure ice (190 to 200 K at 1 Hz: curve B, Fig.1(a)).

To be sure that the preceding results are not caused by an experimental artefact, electrical measurements were also carried out. It is shown in Figure 1 (b) that the variations of ε^' and ε^" versus frequency of Dome C ice core (curves A) are very different from those of pure ice (curves B).

Fig.1(b) Real and imaginary parts of the permittivity ε^' and ε^" versus frequency; curves A: specimen from Dome C 680 m deep, curves B: pure ice.

To improve our knowledge of polar ice, other specimens have been studied. Thus, new results obtained on samples drilled from D 10 (46 m depth) have shown still larger differences from pure ice. In order to prevent any trouble connected with sample preparation, mechanical and electrical techniques have both been employed on the same specimens: the results are shown in Figure 2. The relaxation phenomenon is observed at a lower temperature (127 K, Fig.2(a)) than in the case of Dome C ice, and dielectric measurements confirm this low value of the relaxation time of the dominant dispersion (Fig.2(b)).

Fig.2.(a) Comparison between mechanical and dielectric relaxation: specimen from D 10 station 46 m deep. (a) internal friction spectrum (ζ) and period (P) versus temperature.

Fig.2.(b) Permittivity ε^' and ε^" versus frequency; full line T = 258 K, dashed line T = 230 K.

Actually, results concerning dielectric behaviour are more complex than those concerning internal friction as they are dependent on both relaxational effects and static electrical conductivity (SEC). For instance, the latter is higher in the case of Dome C than in the case of D 10 ice (compare Figs.1(b) and 2(b)). As this point has been discussed by Reference Maccagnan, Barnola, Delmas and DuvalMaccagnan and others (1981), more attention will be paid to the relaxation time. It is well known that a correlation does exist between the relaxation time for mechanical and dielectric relaxation: on the one hand, for a given temperature, mechanical and electrical relaxation times τ have nearly the same value, and, on the other hand, the temperature dependence of these relaxation times is similar.

Such a correlation has been experimentally verified by Reference KuroiwaKuroiwa (1964), but only in a frequency (temperature) range where the relaxational phenomenon in so-called pure ice is not sensitive to the extrinsic rotational defects. So it is interesting to note such a correlation at lower frequencies by comparing the results of Reference KawadaKawada (1978) on dielectric relaxational properties with the very recent results of Reference Tatibouet, Perez and VassoilleTatibouet and others (1981[b]) on internal friction of pure ice measured between 10⁻⁴ and 1 Hz

Figure 3 shows that the experimental points can be considered to lie along two straight lines: at high temperatures, the slope corresponds to 0.58 eV, implying rotational defects in thermodynamic equilibrium; at low temperatures, the slope is 0.25 eV (Reference Tatibouet, Perez and VassoilleTatibouet and others 1981[b]), implying rotational defects with a concentration independent of temperature. In the same figure, theoretical curves have been drawn from the simplified field relation

with E_F = 0.34 eV and E_M = 0.24 eV being the energy of formation and energy of migration of intrinsic rotational defects, respectively, and C the concentration of extrinsic rotational defects.

Fig.3. Relaxation time versus T⁻¹. C is the concentration of impurities which leads to a rotational defect.

As an example, an experimental point obtained from internal friction measurement made on ice doped with about 10⁻²‰ HF (Reference Vassoille, Maï and PerezVassoille and others 1978) is also indicated

Thus, by plotting the data obtained with polar ice in Figure 3, it appears that such a material has a high content of rotational defects; an estimation of their concentration can be made leading to 5 × 10⁻⁷to 10⁻⁶ for ice samples from Dome C, and about 10⁻⁵ for ice samples from D 10.

4 Discussion

First, the effect of impurities and/or ice history will be specified and complementary experimental results presented. Then a tentative interpretation of the origin of the properties of ice from Antarctica will be proposed.

At first glance (see Figure 3, for instance), the behaviour of Antarctic ice is very similar to that of laboratory-doped ice; the dielectric and mechanical relaxation peaks are observed at a low temperature as for HF-doped ice. So it may seem that impurities can explain all the experimental data concerning polar ice. However, chemical analysis does not support this assumption. The low value of T_p would imply 0.5 to 1 × 10⁻⁶ ionic impurity molecules giving rotational defects (30 to 60 μeql ⁻¹) with Dome C samples, and about 10⁻⁵ with the D 10 sample (600 μeql ⁻¹). Actually, values about one order of magnitude lower have been published for central areas in Antarctica (Reference Delmas and BoutronDelmas and Boutron 1980, Reference Delmas, Briat and LegrandDelmas and others in press). D 10 samples have a sea-salt burden relatively high for an Antarctic ice sample, but not more than 0.5 × 10⁻⁶ (32 μeql ⁻¹) of Cl⁻ (M Legrand personal communication). Thus, it does not seem possible to explain the low value of the relaxation time by extrinsic rotational defects.

Nevertheless, in the case of ice from Dome C, a good correlation between SEC and sulphate content has been observed (Reference Maccagnan, Barnola, Delmas and DuvalMaccagnan and others 1981). This point has been interpreted by making an assumption about continuous paths of liquid hydrated sulphuric acid in ice (Reference MaccagnanMaccagnan unpublished).

The possibility that CO₂ may be responsible has also been considered (Reference Paren and GlenParen and Glen 1978), but Reference Delmas, Ascencio and LegrandDelmas and others (1980) have now shown that the CO₂ content in ice from Antarctica is lower than that previously anticipated, leading to a difficulty in attributing the anomalous properties of polar ice to the effect of CO₂.

Mechanical relaxation

1. all the results of Reference KuroiwaKuroiwa (1964),

2. all the results of Reference SchillerSchiller (1958),

3. ∇low frequency results of Reference Tatibouet, Mai, Perez and VassoilleTatibouet (1981),

4. all the results concerning Dome C ice core

5. ▪HF-doped ice

6. ▫specimen from D 10

Dielectric relaxation:

1. ⊕specimen from D 10,

2. all the results concerning Dome C ice core,

3. •from results ofReference Kawada Kawada (1978).

Intrinsic defects connected with the history of the material should be taken into consideration. Snow crystals are created by the freezing of super-cooled water and subsequently grow by diffusion of vapour and by coalescence processes. Ice sheets are formed through the sintering of such snow grains. Ice properties may therefore depend on (i) the conditions of crystal growth, i.e. climatic conditions, and (ii) ageing effects which probably occur in the first metres of the firn due to the high specific surface area presented by the material over an extended period.

In order to specify how the origin of the properties of polar ice may be better explained in terms of either impurity or intrinsic-defect effects, other experiments have been made (Reference Vassoille, Maï, Perez, Tatibouet, Duval and MaccagnanVassoille and others 1980). These will be reviewed briefly. For example, a specimen of Dome C ice (depth equal to 680 m) has been melted and recrystallized. After recrystallization, the relaxation peak appeared in the same temperature range as for pure ice.

On the other hand, an annealing treatment at 270 K has brought all the specimens to the same structural state since T_p was between 149 and 152 K, as it was for artificial ice aged at 271 K.

As the ice from D 10 seems to have a higher content of rotational defects, it was interesting to observe the effect of annealing at 271 K. Indeed, after 1500 h at 271 K, the peak has clearly shifted from 127 to 140 K as shown in Figure 4. Such a result confirms that it is improbable that the anomalous properties of polar ice are induced by impurity effects. It seems easier to admit that the history of polar ice crystals implies that the formation of intrinsic defects in concentrations is higher than that of thermodynamic equilibrium in perfect crystals. Some fraction of these defects are then annealed when the material is kept at 271 K.

Fig.4 Internal friction spectra versus temperature of D 10 sample. The relaxation peak has shifted from 127 K to 140 K after 1500 h at 271 K.

Next, we compared the behaviour of Antarctic ice and that of aged ice studied by Reference Tatibouet, Mai, Perez and VassoilleTatibouet and others (1981[a]), who have shown that, during the ageing process, the microstructure is modified: In a freshly grown single crystal a dislocation lattice can be observed by X-ray topography; during ageing, subgrains are formed and polygonization becomes more pronounced with increased duration of annealing at 271 K. After about 1000 h, the mlcrostructural state seems constant. So observations by X-ray topography have been made with ice from Dome C: this material exhibits a fine microstructure which, in addition, varies with depth (Reference Vassoille, Maï, Perez, Tatibouet, Duval and MaccagnanVassoille and others 1980).

The ageing phenomenon was interpreted by Reference Tatibouet, Mai, Perez and VassoilleTatibouet and others (1981[a]) in terms of a sort of disaggregation of the ice crystal in order to lower its free energy, assuming that internal stresses might exist in freshly grown ice crystals. These internal stresses can be relaxed by the creation of linear and point defects.

On the basis of these results (internal friction measurements and ageing studies), some ideas will be put forward to help to solve the problem. Anomalous properties of polar ice might be due to additional point defects induced by the ageing effect during the formation of ice crystals by transformation of snow. In the case of the ice from Dome C, the ageing temperature can rise to 250 K during summer in the first few metres of the firn, but is lower than 225 K between 10 and 900 m depth; nevertheless, T_p is as low as 140 to 145 K. With the ice from D 10, T_p is still lower; however, the temperature of ageing may be higher. The specific area may not only increase the kinetics of ageing but may affect the “strength” of ageing, i.e. the smaller the size of the crystal, the more Important the effect of ageing may be. Such an assumption implies that after growth of these small crystals the memory of the ageing effect is kept. This point remains to be clarified. To sum up, it seems possible that anomalous properties of polar ice are the consequence of the modification of the concentration and distribution of intrinsic defects through ageing phenomena connected with the conditions of formation of crystalline Ice from snow.