Content uploaded by Yuya Sakai
Author content
All content in this area was uploaded by Yuya Sakai on Aug 25, 2016
Content may be subject to copyright.
SCMT4
Las Vegas, USA, August 7-11, 2016
Freezing Damage Mechanism in Model Microchannels and Vacuum
Curing on Cement Paste to Improve Frost Damage Resistance
Yuya Sakai1a, Tomohisa Kamada1b, and Toshiharu Kishi1c
1Institute of Industrial Science; the University of Tokyo – 4-6-1, Komaba, Meguro, Tokyo, Japan.
1aEmail: <ysakai@iis.u-tokyo.ac.jp>, 1bEmail: <tkamada@iis.u-tokyo.ac.jp>,
1cEmail: <kishi@iis.u-tokyo.ac.jp>.
ABSTRACT
To understand the conditions surrounding frost damage of concrete, freezing and thawing tests were
executed on microchips containing microchannels of various geometries. The straight channels remained
intact, and the channels having ink bottle geometries suffered frost damage. The channels with ink bottle
profiles avoided frost damage when air remained in their cavities. After water permeation, air remained in
channels having a large ratio of cavity to neck volume. Often, air is entrained to fresh concrete as a means
of improving frost damage resistance. The results of this study indicate the importance of air entrapment
during water permeation. Based on the results of the microchannel study, cement paste specimens were
cured under reduced pressure to investigate the hypothesized improvement in resistance to frost damage.
Freezing and thawing of hardened cement paste specimens containing AE water reducing agent resulted
in surface deterioration. On the other hand, vacuumed cement paste specimens suffered almost no
damage. These results indicate, under our experimental condition, vacuum curing as an effective means of
reducing frost damage and also support the observations and results of the microchannel study.
INTRODUCTION
It is difficult to both evaluate and ensure resistance against frost damage of concrete. Air spacing has
previously been used as a factor in assessing frost damage resistance [Marchard et al. 1994; Pigeon and
Pleau 1995]. However, a low correlation between air spacing and frost damage resistance has been
pointed out recently [Sakata et al. 2012]. Admixtures add another element of complexity; for example,
silica fume admixtures cause changes to air spacing that improve resistance to frost damage [Pigeon et al.,
1986], and fly ash admixtures reduce resistance to scaling [Gebler and Klieger 1986; Bilodeau and
Malhotra 1992]. The condition of curing is also a factor affecting frost damage resistance; some research
shows curing has little effect on frost damage resistance [Klieger and Gebler 1987; Afrani and Rogers
1994], but other research concludes that curing outdoors could accelerate damage, particularly on
concrete with low W/C [Hama et al. 2003]. Furthermore, Tabata et al. [1986] show that mild drying
reduces frost damage, while heavy drying accelerates the deterioration. These examples demonstrate the
complex nature regarding frost damage of concrete, and the challenge involved in evaluating or ensuring
resistance against it. Understanding the conditions for the damage is helpful to the eradication of frost
damage from concrete structures and the evaluation of resistance to the damage. The objectives of this
paper include understanding the factors affecting frost damage on concrete and developing a cement paste
with high frost damage resistance. To identify relevant factors, water permeation and frost damage were
studied using microchannels fabricated on glass slides. The study revealed that the larger ratios of cavity
to neck volume of the ink bottle channel type were effective in reducing frost damage. In accordance with
this finding, vacuum environments were used to cure cement paste specimens producing enlarged
cavities, and the frost damage resistance was subsequently examined.
EXPERIMENT DESIGN
Microchannel geometry
Fabrication. Microchips (Photo 1) were fabricated with glass slides (soda lime glass, 76 × 26 × 1.3 mm).
One microchip was composed of two slides; microchannels were milled on one slide and holes of 1 mm in
diameter were made on another slide. The machining was done with an NC processing machine. The two
machined glass slides were cleaned with pure water and a melamine sponge, and immersed into solution
of one part hydrogen peroxide to three parts sulfuric acid. The slides were rinsed with pure water, dried
with duster spray, and pressed against each other to adhere by hydrogen bond. The bonded glass slides
were baked at 645 °C for 5 hours and cooled slowly. The slides were bonded to each other with enough
strength to undergo punching failure rather than detachment from each other when water in the channels
was frozen. Each microchip contained five microchannels with different geometries to study the effect of
channel shape and size on frost damage. Basic designs of the five microchannels are shown in figure 1.
One channel had a straight profile and the remaining four had ink bottle profiles with different cavity
areas and neck lengths. All channels had a depth of 500 μm and the straight channel and necks of ink
bottle channels had a width of 1000 μm. The dimensions of the cavities are shown in figure 1. After
machining the glass slides, the dimensions of the channels were confirmed using a surface roughness
tester. Additionally, in some experiments, microchips containing microchannels with widths and depths of
500 μm and 250 μm were used. Lastly, in some experiments, microchips had similar geometries to those
found in figure 1 but contained only inlets, and no outlets.
Photo 1. Fabricated microchip Figure. 1. Design of microchannels
Water permeation. Microchips were immersed in water of 20 °C in a graduated cylinder, and water
permeation into the channels was observed.
Frost damage. Water was introduced to the microchannels and frozen to study the relationship between
the channel geometry and frost damage. Water was frozen with a low-temperature liquid tank. A floating
steel tray containing the water-saturated microchips was placed upon antifreezing solution in the tank.
The water in the microchips was frozen in two ways: gradually decreasing the temperature from 20 °C to
-20 °C over the course of 12 hours, and having the initial temperature of the tank be -20 °C. In both cases,
the microchips were observed one hour after the temperature reached -20 °C. If damage was not observed
via cracking of the glass, the microchip was put in a room at 20 °C for three hours to melt the ice and the
water was frozen again, up to 10 cycles.
Air bubbles in channels. Air remained in some channels after water permeation. To study the effect of
remaining air on frost damage, two types of chips, with and without air, were prepared and frozen (Photo
2). Air was introduced to the channel using a micropipette. Because the air volume was difficult to
control, the ratio of air to cavity areas (observed from the top) ranged from 10% to 22%. The water in the
channels was frozen using the same technique previously described.
Photo 2. Channels with and without air (left: with air, right: without air)
Cement paste
Specimen preparation. Cement paste of W/C = 40% with or without air entraining (AE) water reducing
agent (0.5%/cement weight) was prepared with ordinary Portland cement (density = 3.15 g/cm3). The
cement paste was placed in plastic cylindrical molds (φ50 × 100 mm) and cured under atmospheric
pressure (20 °C, 0.10 MPa) or vacuum (20 °C, 0.02 MPa) for 24 hours. Vacuum curing was achieved
using a desiccator of reduced pressure acquired via vacuum pump. The specimens were named A0, A0.5,
V0 and V0.5. A and V stand for atmospheric pressure and vacuum, respectively, and the numbers indicate
the amount of AE water reducing agent. As seen in Photo 3, V0 specimens, located at the front, expanded
and overflowed the molds. Vacuum curing was expected to reduce frost damage on the cement paste since
the study of microchannel geometry indicated that channels with a larger ratio of cavity to neck volume
reduced frost damage. Vacuum curing was expected to achieve this effect because air bubbles increase in
size under vacuuming, by a factor proportional to the initial air bubble size raised to the power of 3/2
[Kijito et al. 2009]. The specimens were demolded and immersed in water (20 °C, atmospheric pressure)
24 hours after casting. After 7 days of aging, the specimens were moved to a desiccator (20 °C, RH 60%).
After four days, the specimens were immersed in water for three days. Freezing and thawing test was
executed on the 14th day after casting.
Compressive strength and bulk density. Compressive strength of the specimens was measured at 14-
day age with a loading rate of 30 kN/min (0.25 N/mm2 s). Before the test, the height, diameter, and mass
of the specimens were measured, and the bulk density was calculated. Compressive strength and bulk
density were obtained from three specimens of each case to calculate the average value.
Porosity. The porosity of the specimens was measured with mercury intrusion porosimetry (MIP). On the
14th day after casting, the MIP sample was collected from the center of the specimens and crushed into 3
× 3 × 3 mm cubes. The cubes were immersed in acetone for 24 hours and dried under an absolute pressure
of 10 Pa for 24 hours. The samples used for MIP had masses of 1.0 g.
Freezing and thawing test. As with the microchips, freezing and thawing cycles of the specimens were
accomplished using a low-temperature liquid tank. Specimens were placed horizontally on a floating tray
in antifreeze solution. Water was poured into the tray until the distance between the tops of the specimens
and the water surface reached 3 mm. To prevent the tray from overturning, glass slides were placed on the
tray to prevent the specimens from rolling. A 12 hour cycle in which the temperature went from 20 °C to -
20 °C and back to 20 °C (figure 2) was performed 100 times. The frost damage on the samples was
evaluated by mass reduction.
Photo. 3. Cement paste under vacuum Figure 2. Temperature pattern in freezing and
thawing test
RESULTS
Microchannel geometry
Water permeation into microchannels. The channels with both an inlet and an outlet filled with water
immediately after the inlet were submersed in water. Conversely, it took several minutes to several hours
(varied depending on the channel type) for the microchannels without an outlet to form a meniscus at their
dead end. The channels filled from their dead ends (bottom of Photo 4), as opposed to their inlets.
Meniscus formation tended to take more time in longer channel geometries. The air in the channels was
pushed out (Photo 5) as the water filled them. The air was removed completely from some channels but
remained in the cavities of channels having larger ratios of cavity to neck volume. For these channel
geometries, remaining air occupied a large portion of the cavity (Photo 4). The size of the air bubble in
the cavity was consistent over the course of one week. No visible air remained in the straight channels.
Effect of channel geometry on frost damage in microchips. The microchannels were filled with tap
water and frozen. Photo 6 shows microchannels of 500 μm in depth after freezing. In the photo, each
channel is separated but their dimensions are the same as those in figure 1. Regardless of the freezing
pattern, the straight channel was not damaged and the other ink bottle channels were cracked or punched.
The channels with smaller cavities suffered small cracks in the first freezing; however, punching occurred
at the subsequent freezings. Photo 6 shows the results in which the temperature was decreased gradually
for 12 hours. Here, the microchannels of 250 μm in depth suffered only small cracks (Photo 7) after first
freezing. However, during the third freezing punching fracture was observed. The straight channels
suffered no damage even after ten cycles of freezing and thawing.
(a) 4 hours (b) 8 hours (c) 24 hours
Photo 4. Water permeation into microchannels after varying elapsed times of immersion
Photo 5. Air pushed out of channel Photo 6. Damaged channels after freezing
Effect of remaining air on frost damage of microchips. The microchips with and without remaining air
bubbles were frozen. Water-saturated channels suffered punching upon freezing but channels with air did
not, even after ten cycles of freezing and thawing (Photo 8). Photo 9 shows the magnified air bubble
before and after the water was frozen.
Meniscus
Meniscus
Remaining air
(a) Before (b) After
Photo 7. Damage in smaller Photo 8. Effect of air Photo 9. Air bubble before
channels on frost damage and after freezing
Cement paste
Compressive strength and bulk density. Compressive strength and bulk density are shown in figure 3.
Vacuum curing reduced both compressive strength and bulk density in V0 and V0.5 specimens when
compared with their A0 and A0.5 counterparts. AE water reducing agent reduced bulk density and
increased compressive strength for specimens cured under atmospheric pressure (A0 and A0.5).
Porosity. Porosity measured with MIP is shown in figure 4. Specimens that were vacuum cured or had
AE water reducing agent (V0, V0.5, A0.5) showed increased porosity compared with A0 for pores with
diameters of 100-200 nm. Vacuum cured specimen V0 showed reduced porosity for pores with diameters
of 30-40 nm compared to A0. Specimens V0 and A0.5 demonstrated similar porosity curves for pores
larger than 20 nm in diameter. A0.5 and V0 had lower total porosity than A0.
Figure 3. Compressive strength and Figure 4. Pore size distribution
bulk density
Freezing and thawing test. Figure 5 shows the mass change induced by frost damage. Over 100 cycles
of freezing and thawing, A0 collapsed completely and V0 suffered the least amount of damage (Photo
10). The specimens with AE water reducing agent (A0.5, V0.5) did not collapse, but showed surface
deterioration. A few large fragments fell off of specimen A0.5.
Figure 5. Mass change during freezing thawing test
Photo 10. Specimens after freezing and thawing test (left to right: A0, V0, A0.5, and V0.5)
DISCUSSION
Microchannel geometry
Water permeation into microchannels. The microchannels with both an inlet and an outlet filled
immediately with water after the inlet was immersed. Conversely, the microchannels with only an inlet
took a long time to fill (Photo 4). This delay is likely due to resistance from the air in the channels. These
channels filled from their dead ends, a phenomenon explained by corner flow and film flow in which fluid
travels to corners and surfaces of channels first [Harris et al. 1995]. The fluid reached the dead ends,
where a meniscus then formed. Applying this concept to water permeation of concrete, it can be assumed
that when a hardened concrete specimen is immersed in water, water permeates from all surfaces. This
condition corresponds to that of microchannels without an outlet. When hardened cement paste specimens
were immersed in water, it was observed that air was pushed from the specimens (Photo 10) as seen with
the microchannels (Photo 5). In the study with microchannels, more air remained in the channels having
larger ratios of cavity to neck volume. Air remained when a meniscus formed at the neck near the inlet
before the cavity could fill with water (Photo 4). Such an air trapping mechanism suggests that the larger
the cavity, or the smaller the neck of the ink bottle profile, the more air will become trapped in the
channel. Even a straight channel retains large amounts of air, particularly as the permeation length
increases [Han et al., 2006].
Effect of channel shape on frost damage. Freezing damaged only the microchannels with ink bottle
profiles. One possible reason is that the freezing occurred near the channel walls, and water in the neck,
which has a smaller cross sectional area than the cavity, froze completely before the water in the cavity.
Ultimately, the freezing of the water remaining in the cavity is what caused the damage, as evidenced by
the punching fractures centered within the cavities (Photo 8). More freezing and thawing cycles were
required to damage smaller channels because the smaller channels (Photo 7) had less water volume, and
therefore less expansion due to freezing, in addition to having thicker slide walls since less glass was
removed during machining. Dense concrete with low W/C typically has high damage resistance under
standard freezing and thawing testing [Hama et al. 2003], likely due to a similar mechanism as that of the
small microchannels.
Photo 11. Air pushed out from hardened cement paste
Effect of remaining air on frost damage. Water increases its volume by 9% upon freezing. In the
microchannels, the deformation of the air bubbles upon freezing indicated that the air absorbed some
volume expansion of the frozen water (Photo 9). Photo 9 shows the initial area ratio of the air to be about
12% and that the area was reduced by only slightly after freezing. This shows that the volume expansion
of freezing water was absorbed by the remaining air amongst other factors such as elastic deformation of
the glass and pushing unfrozen water out of channels.
Cement paste
Bulk density, compressive strength, and porosity. Vacuum curing reduced both compressive strength
and bulk density of the hardened cement paste specimen (figure 3), likely due to the air volume increase
associated with vacuum curing. Figure 4 shows that V0 had lower porosity than A0. It is probable that V0
had more pores larger than 10 μm, which is outside of the measurement range of MIP. Vacuum curing
increased porosity for pores with diameters of 100-200 nm. Such changes in porosity indicate that the
vacuum curing affects not only larger volumes of air such as entrapped and entrained air, but also smaller
volumes of air on the nano scale. The fact that V0 and A0.5 had similar porosity curves for pores with
diameters exceeding 20 nm demonstrates the similar effect of the AE water reducing agent and vacuum
curing on pore structure.
Freezing and thawing test. Under atmospheric pressure, the specimen without AE water reducing agent
(A0) collapsed at the 50th cycle. On the other hand, all other specimens remained intact through 100
cycles (figure 5) apart from surface deterioration of specimens A0.5 and V0.5 (Photo 10). The specimen
cured under vacuum condition without AE water reducing agent (V0) avoided even surface damage. The
above results indicate that vacuum curing is effective in improving resistance against frost damage,
including surface deterioration, although it can reduce compressive strength and bulk density (figure 3). If
vacuum curing enlarges the cavities of ink bottle geometries in cement paste specimens, and air remains
entrapped, then the effect of the remaining air is improved resistance to frost damage as seen with the
microchips. To confirm this, the relationship between remaining air volume in concrete and frost damage
resistance needs to be investigated quantitatively.
CONCLUSION
The following general conclusions can be drawn from the study described in this paper:
Model microchannels having an ink bottle geometry were damaged
Larger ratios of cavity to neck volume in ink bottle microchannels result in more air entrapment
within the cavity
Model microchannels with air inside the cavity avoided frost damage
Hardened cement paste specimens cured under reduced pressure had larger volumes of coarse pores
(100–200 nm in diameter) compared with those cured under atmospheric pressure
The hardened cement paste specimens cured under vacuum showed higher resistance to frost damage
ACKNOWLEDGEMENTS
Support for this research project was provided by LIXIL JS Foundation.
REFERENCES
Afrani, I., and Rogers, C. (1994). “The effects of different cementing materials and curing on concrete
scaling.” Cement, Concrete and Aggregates, 16(2), 132-139.
Bilodeau, A. and Malhotra, V. M. (1992). “Concretes incorporating high volumes of ASTM Class F fly
ashes: Mechanical properties and resistance to deicing salt scaling and to chloride penetration.” ACI
Special Publication, 132, 319-350.
Gebler, S. H. and Klieger, P. (1986). “Effect of fly ash on the durability of air-entrained concrete.” ACI
Special Publication, 91, 483-520.
Hama, Y., Senbu, O., and Tomosawa F. (2003). “Evaluation of frost resistance of concrete considering
drying effect and capillary suction”, Cement Science and Concrete Technology, 57, 266-272.
Han, A., Mondin, G., Hegelbach, N. G., de Rooij, N. F., Staufer, U. (1995). “Filling kinetics of liquids in
nanochannels as narrow as 27 nm by capillary force, Journal of Colloid and Interface Science, 293(1),
151-157.
Harris, W., Radke, C. J., and Morris, s. (1995) “The motion of long bubbles in polygonal capillaries. Part
1. Thin films." Journal of Fluid Mechanics, 292, 71-94.
Kizito, J., Balasubramaniam, R., Nahra, H., Agui, J., and Truong, D. (2009). “Vapor-gas bubble evolution
and growth in extremely viscous fluids under vacuum.” 47th AIAA Aerospace Sciences Meeting
including the New Horizons Forum and Aerospace Exposition, 215606.
Klieger, P. and Gebler, S.H. (1987). "Fly ash and concrete durability." ACI Special Publication, 100,
1043-1069.
Marchand, J., Pleau, R., and Gagne, R. (1995). “Deterioration of concrete due to freezing and thawing.”
Materials Science of Concrete, 4, 283-354.
Pigeon, M. and Pleau, R. (1995). “Durability of concrete in cold climates”, E&FN Spon, 94-127.
Pigeon, M., Pleau, R., and Aiitcin, P. C. (1986) Freeze-thaw durability of silica fume concrete: Scaling vs
internal microcracking.” Cement, Concrete and Aggregate, 8(2) 76-85.
Sakata, N., Sugamata, T., Hayashi, D., and Hashimoto, M. (2012). “Investigation for relationship between
frost damage resistance and air-void system in concrete.” Concrete Research and Technology, 23(1),
35-47.
Tabata M., Koh Y., and Kamada E. (1986), “Experimental study on the factorial effects: Influence of
environmental conditions on the frost damage of concrete (I).” Journal of structural and construction
engineering, 367, 8-15.
