Pressure and temperature effects: brittle zone onset

Snow accumulation rate and surface air temperature are primary controls on the densification of polar firn (e.g. Reference Herron and LangwayHerron and Langway, 1980; Reference Ligtenberg, Helsen and Van den BroekeLigtenberg and others, 2011), although the firnification process is not fully understood (Reference Hörhold, Kipfstuhl, Wilhelms, Freitag and FrenzelHörhold and others, 2011). Firn densification is relevant here for its role in isolating air bubbles from porous firn and because the thickness and density of the firn column controls the exact overburden pressure imposed on ice and air bubbles at depth. Overburden pressure may affect where brittle fracture of ice cores commences, as thin firn columns at warmer, higher-snow-accumulation sites should reach full ice density (920 kg m⁻³) more rapidly and thus exert more overburden pressure on ice and air bubbles at depth, and vice versa. Additionally, overburden pressure and in situ ice temperature control clathrate stability and are expected to affect the depth of the transition out of the bubbly BIZ and into the bubble-free, ductile clathrate-ice zone below, with clathrates stabilizing at lower (higher) pressures when temperatures are low (high) (e.g. Reference MillerMiller, 1969).

To quantify the influence of firn column thickness on overburden pressure at depth, while also converting BIZ depths to equivalent overburden pressures, firn-column density is modeled using the University of Washington Firn Model Intercomparison Experiment online Reference Herron and LangwayHerron and Langway (1980) model (FirnMICE, http://firny.ess.washington.edu/communityfirnmodel/). A surface snow density of 390 kg m⁻³ is assumed for all sites, and mean annual surface air temperature and snow accumulation rate (ice equivalent) are input into the model to construct 1 m resolution depth-density profiles from the surface to 300 m for each site (Fig. 3a). Constant ice density of 920 kg m⁻³ is assumed below 300 m, although for bubbly ice this may be an overestimate of several kgm⁻³. Snow density measurements from Taylor Dome, Berkner Island and WAIS Divide agree well with model data generated for these sites (density measurements provided with the FirnMICE online model, not shown). Overburden pressure at a given depth is calculated as the sum of overlying snow and ice with density prescribed by the FirnMICE model output. One cubic meter of ice (density 920 kg m⁻³) exerts 9.02 × 10⁻³ MPa overburden pressure (Fig. 3b). Modeled overburden pressure at 300 m, the depth at which the thickest modeled firn column (Dome Fuji) reaches ice density of 920 kg m⁻³, is presented in Figure 3c for all sites, plotted against modeled and observed FIT depths (830 kg m⁻³ density). The mean misfit between observed and modeled FIT depth is −4%, with minimum misfits of +1% and −2% (EPICA Dome C and WAIS Divide, respectively) and a maximum misfit of−25% (Siple Dome).

Fig. 3. Results of FirnMICE firn column density modeling (Reference Herron and LangwayHerron and Langway (1980) model) and overburden pressure calculations. (a) Density–depth profiles generated for the 18 drill-site temperature and snow accumulation regimes, shaded according to modeled FIT depth (light grey: model FIT <70 m; grey: 70 m < model FIT < 80 m; black: model FIT >80 m); bold dashed line indicates ice density (920 kg m⁻³). (b) Ice overburden pressure versus depth in the firn column (0–300 m) for the 18 drill sites (light grey: model FIT <70 m; grey: 70 m < model FIT < 80 m; black: model FIT >80 m); bold dashed line indicates overburden pressure assuming constant ice density from the surface. (c) 300 m depth overburden pressure versus FIT depth from modeled (open circles) and measured (filled circles) firn-column density data.

Across the 18 drill sites, 300 m overburden pressure varies from 2.36 MPa at Dome Fuji (modeled FIT depth 114 m, +12% misfit) to 2.56 MPa at Siple Dome (modeled FIT depth 46 m, −25% misfit; Fig. 3b and c). Poor FIT depth reproduction for these end-member sites (Fig. 3c) suggests an overestimate of the possible range of overburden pressures; more conservative is 2.4 MPa at EPICA Dome C (modeled FIT depth 101 m, +1% misfit) to 2.55 MPa at Berkner Island (modeled FIT depth 52 m, −10% misfit). This overburden pressure fluctuation at depth, a maximum difference of ∼0.15–0.2 MPa between the thickest firn column (slower firnification, lower overburden pressure at depth) and the thinnest (faster firnification, higher overburden pressure at depth), is equivalent to 16.5–22.2 m of ice overburden. If the firn column is ignored and full ice density is assumed from the surface, overburden pressure at depth (2.7 MPa at 300 m) is overestimated by 0.14–0.34 MPa or ∼15.5–37.7 m of ice (dashed line, Fig. 3a and b). This results in an underestimation of the depth for theoretical clathrate stability. The BIZ does generally occur at shallower depths where the firn/ice transition is shallower (e.g. Fig. 2) and exerts greater overburden pressure at depth. This is most clearly observed at GIS and WAIS drill sites, although with low significance, as is expected due to low precision of reported BIZ depths and complex temperature and accumulation effects on firnification and clathrate formation processes. Reported BIZ top depths begin ∼10 m deeper per 1 m thickening of the FIT at the 11 GIS and WAIS drill sites (regression of GIS and WAIS FIT depths and BIZ top depths gives R ² = 0.54, not shown). As overburden pressure alone cannot explain such a deepening, this suggests that unidentified firnification processes likely exert significant control on BIZ onset depth. Proportional deepening of BIZ onset is not observed at all EAIS drill sites, especially not those on the EAIS plateau, perhaps because of increasing tensile strength of ice at smaller grain sizes and lower temperatures (e.g. Reference ButkovichButkovich, 1954; Reference PetrovichPetrovich, 2003), as well as other unknown effects associated with extreme low-temperature and low-accumulation firn densification, grain growth and air-bubble formation.

Pressure and temperature effects: brittle zone relief

In situ ice temperature determines the depth (pressure) of clathrate stability and thus should affect the depth of the transition from bubbly, brittle ice to bubble-free, ductile ice. This is supported by observations of shallower appearance of clathrates and disappearance of air bubbles in ice cores at low-temperature sites (e.g. observations from Dye-3 and GRIP versus Vostok and Dome Fuji; Reference Ikeda-Fukazawa, Hondoh, Fukumura, Fukazawa and MaeIkeda-Fukazawa and others, 2001). While temperature effects controlling BIZ onset are less clear, it is expected that the BIZ will be relieved at shallower depths (lower pressures) where ice temperatures are lower. Using overburden pressure as calculated above, and borehole temperature data from selected drill sites (Table 4), the BIZ can be evaluated with respect to temperature and pressure and thus compared more accurately with theoretical clathrate stability (Fig. 4). BIZ onset in many cases begins at pressures (depths) where clathrates should already begin to stabilize, but brittle fracture is not relieved until ductile conditions are reached as bubble number densities become sufficiently low and clathrates dominate, strengthening ice cores. While it is tempting to suggest that BIZ bottom pressure indeed decreases with lower ice temperatures, this is not a statistically significant feature including all reported BIZ data (excluding intermediate-depth sites where clathrates do not stabilize before the ice/bed interface is reached). Certainly the BIZ in the Greenland summit ice cores and at WAIS Divide (1200–1400 m BIZ bottom depths; ∼10.5–12.5 MPa) extends several hundred meters deeper than that of Dome Fuji, Vostok, EPICA Dronning Maud Land and Talos Dome (840–1050 m BIZ bottom depths, 7.2–9.2 MPa), but the EPICA Dome C ice core remained highly fractured to 1200 m depth (10.5 MPa). Additionally, the BIZ at Camp Century, Dye-3 and Byrd Station transitions to ductile ice at shallow depths, with pressures only 1–2 MPa greater than that required for initial clathrate stability, while most sites transition out of the BIZ at pressures 3–5 MPa in excess of requirements for theoretical onset of clathrate stability (Fig. 4). This raises the interesting prospect that ice-flow advection of cold ice towards the surface at flank sites could encourage shallower formation of clathrates (e.g. Camp Century, Byrd Station; Reference Shoji and LangwayShoji and Langway, 1987). However, without disentangling the effects of imprecise BIZ records, drill performance and ice-core handling/transport, it is difficult to use reported brittle ice depths to further evaluate this correlation between clathrate stability and the bottom of the BIZ.

Fig. 4. BIZ top (inverted triangles) and bottom pressures (triangles) plotted at respective in situ ice temperature from borehole temperature measurements (Table 4). Theoretical clathrate stability curves are plotted for N₂ (solid), O₂ (dotted) and air (dashed) hydrates (Reference MillerMiller, 1969; Reference Kuhs, Klapproth, Chazallon and HondohKuhs and others, 2000). The stability curve of Reference Kuhs, Klapproth, Chazallon and HondohKuhs and others (2000) is indicated with ‘x’. Open triangles denote BIZ bottom pressures at the ice/bed interface, thus not the full transition from the bubbly, brittle-ice zone to the bubble-free, ductile ice zone below. WSD: WAIS Divide; EDC: EPICA Dome C.

Table 4. In situ ice temperature (borehole temperature; °C) at BIZ top and bottom depths (see Tables 1–3) for selected sites. Temperatures are ±0.5°C, estimated from published graphical data where original datasets could not be accessed. Temperature data for Berkner Island are modeled

To improve understanding of the mechanisms involved in BIZ onset and relief, it may prove useful to investigate ice physical properties including bubble number density, bubble size, micro-bubbles and ice fabric (e.g. grain size, crystal anisotropy). Take the anomalously shallow BIZ at Vostok, for example: Reference Uchida, Duval, Lipenkov, Hondoh, Mae and ShojiUchida and others (1994) observed a reduction of core quality caused by fractures as shallow as 100 m (likely linked to thermal drilling at Vostok), becoming heavily fractured from 250 to 750 m with progressive improvement to 900 m where core quality again became excellent. This shallowest occurrence of the BIZ may be related to a rapid increase in bubble number density (bubbles cm⁻³) observed in the Vostok ice core beginning at ∼300 m depth (increasing from ∼400 to ∼800 bubbles cm^-3; Reference Uchida, Duval, Lipenkov, Hondoh, Mae and ShojiUchida and others, 1994; Reference UeltzhöfferUeltzhöffer and others, 2010). Such an increase in bubble number density may effectively reduce bubble pressures required for fracture propagation by narrowing interstitial ice between bubble cavities. Bubble number densities reported in the BIZ at Byrd Station and WAIS Divide were relatively more constant at ∼200 and ∼450 bubbles cm⁻³, respectively (Reference GowGow, 1971; Fitzpatrick and others, in press). Micro-bubbles formed through sublimation-condensation processes may also play a role at Vostok (e.g. Reference Lipenkov and HondohLipenkov, 2000), and are observed in the EPICA Dome C and Dronning Maud Land ice cores (Reference UeltzhöfferUeltzhöffer and others, 2010). The second shallowest occurrence of the BIZ, 335 m at Taylor Dome, may be anomalously shallow due to high strain rates at this site; elongated bubbles were observed from 360 to 390 m (Reference FitzpatrickFitzpatrick, 1994).

Drilling techniques

Currently, drilling through the BIZ produces ice-core sections with several to many breaks, fractures and hairline cracks, commonly affecting more than half of recovered ice in the middle of this zone. Drilling fluid required at these depths pervades all cracks and fractures, contaminating many chemical analyses. At WAIS Divide, where BIZ core quality was the best of any recent United States-led drilling project, 1 m long core sections from 900 to 1200 m were on average between ‘good’ (containing zero to three breaks or 50% unfractured) and ‘fair’ quality (containing >10 cm of core length without fractures; Reference SouneySouney and others, 2014).

As the only step performed at in situ pressures within the ice sheet (minimum of 2.0 MPa, maximum 12.4 MPa in the BIZ), with the added benefit of damping effects associated with a liquid-filled borehole, the drilling process provides important opportunities to reduce the major component of brittle fracture in ice cores by performing mechanically severe steps that might damage cores if performed at surface pressures (∼700 hPa, 0.07 MPa). For example, drilling at WAIS Divide employed a unique strategy specifically for brittle ice, performing three core breaks per ∼2.5 m drill run before returning the drill sonde to the surface (detailed in Reference SouneySouney and others, 2014). This technique of performing several core breaks per run at in situ pressure results in ice-core sections that are ready to ship, without subjecting brittle ice cores at surface pressure to the vibration and high stresses of making circular-saw section cuts. A similar downhole technical innovation suggested by Reference UedaUeda (2002) is the development of a drill sonde that captures and seals cores in a vessel at in situ pressures, potentially allowing for a more gradual transition to atmospheric pressure than the usual minutes-to-hours-long transition as the drill sonde is brought to the surface. Slowing the pressure and temperature transition by hoisting brittle ice-drill runs slowly to the surface, reducing drill penetration rate and considering drill fluid pressure balance have also been proposed (see the discussion by J. Schwander and others in IPICS, 2005).

Maintaining temperatures within drilling and on-site storage structures similar to those at depth in the borehole is a commonly practised technique in ice drilling (e.g. Reference SouneySouney and others, 2014). At Roosevelt Island, an actively cooled storage cave kept ice at −23°C despite surface temperatures reaching as much as −5°C (personal communication from D. Mandeno, 2014). Temperature gradients may reach several tens of degrees between in situ ice temperature and surface drill structure temperatures, which will cause differential heating of ice cores brought to the surface, inducing temperature and stress gradients as the ice-core surface warms (and expands) more quickly than the interior. However, it is difficult to overstate the primary impact of the pressure gradient between the BIZ and the surface, which, as a ratio, is a minimum of 30 : 1 at the shallowest observed BIZ onset (Vostok: 2.0 MPa at 250 m, ∼0.07 MPa surface air pressure) and grows to more than a 100 : 1 ratio at most ice-core drill sites (reaching a maximum BIZ bottom-to-surface-pressure ratio of 177 : 1 at the bottom of the Greenland Ice Sheet Project 2 (GISP2) BIZ: 1400 m depth, 12.4 MPa).

Transport techniques

Reduction of mechanical shock during core transport from the drill site to laboratories has been achieved primarily by sheathing brittle ice cores in tight-fitting nylon netting immediately upon removal from the drill, and delaying shipment of brittle ice for as long as logistically possible to allow for maximum relaxation. Use of netting was innovated at Berkner Island (Reference Mulvaney, Alemany and PossentiMulvaney and others, 2007), and the same was used successfully at WAIS Divide (Reference SouneySouney and others, 2014) and Roosevelt Island (personal communication from N. Bertler, 2013). While itself not preventing initial fracture of brittle ice cores, nylon netting holds badly fragmented sections in place, preventing any loss of stratigraphic order, and protects cores from further damage due to vibration during shipment.

Relaxation of ice cores after drilling is examined thoroughly by Reference GowGow (1971), measuring significant post-drilling volume expansion (density reduction) at all depths in the Byrd Station ice core. Cores from the BIZ around 800 m depth exhibit greatest expansion: ∼0.2% (0.002 kg m⁻³ density decrease) after 8 months, ∼0.4% (0.004 kg m⁻³ decrease) after 16 months, and up to ∼0.6% (0.006 kg m⁻³ decrease) after 27 months (see fig. 2 of Reference GowGow, 1971). Cores from other depths expanded by an average of ∼0.2% (0.002 kg m⁻³ decrease) after 27 months. Nearly identical relaxation characteristics were observed in the GISP2 ice core (Reference GowGow and others, 1997). Ice cores retrieved using a hot-water drill at Siple Dome relaxed very quickly, likely due to the thermal drilling technique, but cores were also stored at relatively high surface temperatures after drilling (Reference Engelhardt, Kamb and BolseyEngelhardt and others, 2000; Reference Gow and MeeseGow and Meese, 2007). Ice cores retrieved by PICO (Polar Ice Coring Office) mechanical drilling atSiple Dome showed very little relaxation, and remain brittle to date (Reference Gow and MeeseGow and Meese, 2007; personal communication from M. Twickler, 2013).

Much of this relaxation in brittle ice is attributed to slow dilation of highly pressurized bubbles abundant in this zone. This lends support to the practice of overwintering brittle ice, most recently performed at WAIS Divide (Reference SouneySouney and others, 2014). At Roosevelt Island, ∼2 m long drill runs were sheathed in netting and stored in a refrigerated snow cave for up to 14 days before making 1 m section cuts for shipment. Improvements were noted in bandsaw cuts after even this short period of relaxation, although some propensity for fracturing remained (personal communication from N. Bertler, 2013). It is important to note that overwintering or long-term storage of brittle ice cores delays sampling and analysis of this section of ice. While this delay is less significant on the time frame of a multi-year deep drilling project, for smaller endeavors, especially at intermediate-depth coastal sites where a significant portion of the dated ice-core record lies within the BIZ, delaying shipment and/or sampling of ice cores is more challenging.

Sampling techniques

Ice-core sampling is commonly performed at −20 to −25°C, making preliminary longitudinal cuts using a horizontal bandsaw and subsequent sampling with common vertical bandsaws. While a simple tool, the bandsaw applies consistent force at the cutting teeth, especially if tracking of the saw along-core is automated to steadily move the saw blade through the ice. After sufficient relaxation – allowing for slow dilation of air bubbles long after initial violent cracking and fracturing observed immediately after drilling–brittle ice may feed through a bandsaw with little additional fracturing. Brittle ice below depths of 475 m from Roosevelt Island proved prohibitively brittle after 6 months stored at −30°C, with several instances of near-catastrophic fracturing during horizontal cutting due to vibration from the saw blade. At this depth, a conventional vertical bandsaw used to make ∼0.035 m × 0.035 m × 1.0 m rods of ice also began to add many fractures to previously flawless ice-core samples, which had immediately prior been cut into ∼0.1 m × 0.035 m × 1.0 m slabs without damage. Core sampling was halted at 500 m and the remaining ice stored at −18°C for an additional year, after which cutting proceeded without significant challenges.

Other cutting instruments may prove more conducive to processing brittle ice-core samples. Reference TisonTison (1994) describes the use of a diamond-wire saw for preparing thin sections of debris-rich or brittle ice, an option attractive for its reduced vibration levels. However, with slow cutting rates and the high cost of diamond wires, this option may require significant development before being applicable to high-volume ice-core sample processing.

Analytical techniques

Many analyses performed in polar ice-core studies depend on relatively unbroken samples to prevent contaminants from altering original paleoclimate or paleo-environmental signals. Measurements of atmospheric gases preserved in bubbles ideally require avoidance of section cuts and other breaks present in ice-core samples, in order to exclude modern atmospheric gases from analysis. Major-ion and trace chemistry, analyzed in longitudinal samples from the center of ice cores in order to avoid modern contaminants imparted during drilling, shipment and sampling, requires removal of outer sample surfaces and cleaning of any exposed surfaces including section cuts and fractures. Drilling fluid pervades all fractures in ice-core samples – especially fluids with low volatility, such as Estisol-240 (Dow Haltermann, Germany) coconut oil extract used at NEEM (North Greenland Eemian Ice Drilling) and Roosevelt Island – rendering chemical analyses extremely difficult, especially in ice from the BIZ.

Continuous-flow analysis (CFA) systems gravity-feed longitudinal ice samples through a sectioned heating plate, pumping meltwater directly into online instruments and/or fraction collectors to create discrete sub-samples (e.g. Reference Osterberg, Handley, Sneed, Mayewski and KreutzOsterberg and others, 2006; Reference Bigler, Svensson, Kettner, Vallelonga, Nielsen and SteffensenBigler and others, 2011). When analyzing highly fractured samples in CFA systems, contamination affects not only fractured core sections, but also subsequent ice as relatively high-concentration contaminants wash out of sample lines and instruments. Additionally, vertical guide systems for gravity-feeding samples struggle to accommodate ice samples containing fractures, especially high-angle longitudinal fractures, which commonly wedge against plastic guides, disrupting sample flow and accurate depth logging.

Fractured sample sections are commonly removed entirely from CFA campaigns, such that CFA for major-ion chemistry of brittle ice from the WAIS Divide ice core only processed ∼62% of ice from the depths of 577 m to 1300 m (personal communication from D. Ferris, 2014). An optical drill fluid detection system identified 27 instances of drill fluid contamination in CFA tubing while analyzing 175 m of brittle ice from the NEEM ice core (Reference Warming, Svensson, Vallelonga and BiglerWarming and others, 2013). Using this detection system, particular negative impacts were noted for dust, conductivity, ammonium, hydrogen peroxide and sulfate datasets in brittle ice from the NEEM ice core. While technically challenging, feasible solutions have been developed for analyzing brittle ice-core samples with little sample loss, such as the high-resolution CFA water stable-isotope analysis of 13 mm × 13 mm × 1.0 m rods of ice from the WAIS Divide ice core (B. Vaughn and others, unpublished information). This approach used tightly fitting square acrylic tubes to protect the fragile ice rods during shipment, and also support them vertically in a sample rack during analysis – while light vibration successfully prevented wedging against the acrylic tubing during melting, even in highly fractured brittle ice (personal communication from B. Vaughn, 2014).