Ice Physics and Physical Processes - Sea Ice: Physics and Remote Sensing - page 64

The subgrain boundaries are delineated by the rows of
brine/air pockets, noticeably in Figure 2.34a as elongated
bright objects. Many rows appear as continuous lines,
connected to the main brine channel. The relative orien-
tations of the arms of the brine channel with respect to
those of the subgrains provide evidence that the subgrain
boundaries are also strongly associated to the shape of
the arms. It can be noted that the arms of the brine chan-
nel is full with ice crystals that are distinctly different in
shape and orientation from the neighboring elongated
subgrains. This is illustrated very clearly in Figure 2.34b.
Note the clarity and shape of the crystals in the middle
of this micrograph. The shining black objects are mostly
air pockets. It also exemplifies that the channel retained
only small amount of brine pockets. Lack of large amount
of brine inclusions concentrated in localized areas also
makes the prior brine channels from straightforward,
unambiguous detections. Perhaps, this explains why sea
ice researchers observed a decrease in the number of brine
channels as the distance from the bottom ice increases.
The prior brine channels, therefore, do not disappear in
the bulk of sea ice, but they are not visually as obvious
when the ice is matured.

Due to ice permeability, some melt water can be retained
within the ice. So far, no discussions or model has been
presented to explain the permeability of sea ice and its
dependency on the interconnected grain and, particu-
larly, subgrain boundaries in the bulk of sea ice.

Brine Mobility through Subgrain Boundaries
Before going into the discussions on this mechanism,
it is necessary to describe the relevant microstructural
aspects of level sea ice in details. The structure of colum-
nar sea ice can best be described as cellular, lamellar, or
acicular. At temperatures higher than −22.8 °C, sea ice
can be visualized as a binary alloy consisting of a mixture
of two phases like alpha‐beta ( α + β ) titanium alloys used
commonly in structural components of aircraft [ ASM,
Source Book , 1982]. The microstructure of a α + β tita-
nium alloy depends on both the thermal and mechanical
processing sequences that have been imposed on it. The
primary α ‐phase of titanium is hexagonal like ice, whereas
the metastable β ‐phase is the cubic form of titanium.
Chemically both the phases are same, except for small
amounts of molybdenum and vanadium added as β stabi-
lizers. A common structure of the α + β alloy is known as
“ Widmanstätten ” in which colonies or pockets of plates
of α ‐phase become the predominant components with β ‐
phase existing along the boundaries of the plates. This
type of structure is very similar to the features com-
monly seen in columnar grained sea ice. Since the bulk
temperature of floating first‐year sea ice sheets is signifi-
cantly higher than −22.8 °C, one of the phases in the
Widmanstätten ‐like structure of sea ice is liquid. In this
case, the primary phase is pure hexagonal ice, whereas
the secondary phase is the brine. At low temperatures,
below about −22.8 °C when most of the salts are in pre-
cipitated form, sea ice becomes equivalent to a “dispersion
strengthened” superalloys used in gas turbine engines. In a
way, this is analogous to α − β alloys with Widmanstätten
structure, but the second phase exists as dispersing parti-
cles mainly in the form of cubes of slat crystals. Anyway,
the lamellar or subgrain boundaries in sea ice are in a rela-
tively higher energy state than the ice inside the platelets
(or subgrains) with their long directions in the vertical
planes. Consequently, diffusivity of the boundaries is also
higher than the lattice diffusion of the plates.
At higher temperatures, the subgrain boundaries can act
as passages allowing the brine to flow. The total surface
area of the subgrain boundaries inside the grains exceeds
the total surface area of the grain boundaries. The network
of subgrain and grain boundaries, therefore, provides very
effective passages for migration of brine. This mechanism
also explains the feeding of brine channels through the
connected brine pockets along the boundaries, leading to
the formation of tributaries of the channels. As explained
earlier, the subgrain boundaries represent locations within

Brine Flushing
The fourth commonly known mechanism of ice desali-
nation is brine flushing [ Untersteiner , 1968]. This process
is normally active after the growth season and during the
beginning of the melt season in late spring and summer.
It is mainly caused by melting of snow and ice at the top
of ice sheets. The hydrostatic pressure overhead of brine
and melt water results in percolation through the ice
sheet, mainly vertically but also horizontally [ Eicken
et al ., 2004]. In this case, salinity of the upper 0.5 m of the
ice sheet is usually reduced to less than 1‰ down from a
typical value of 4-7‰. Eicken et al . [2004] also uses
Darcy's law to calculate the scale of brine downward
flushing as a result of meltwater percolation. Notz and
Worster [2006] used a one‐dimensional model to simu-
late summer time flushing of surface melt. Most of the
salt rejection due to the brine flushing mechanism results
from the vertical percolation of surface melts. Notz and
Worster [2009] ran the model to simulate the evolution
of melt pond percolation during the course of 1 h into
0.9‐m‐thick sea ice with a bulk salinity of 10‰ (these
numbers may be too high to be realistic) and surface tem-
perature of −5 °C increasing linearly to −1.8 °C at the ice‐
ocean interface. They found that brine flushing was very
active. After 1 h, the melt pond had drained almost com-
pletely, and the bulk salinity has decreased by more than
2‰. Melt water percolation into the ice sheet brings the
ice sheet into an isothermal condition. Eicken et al . [2002]
reported that up to 25% of the melt water produced at the
surface melt of Arctic ice is retained within the ice sheet.