See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/249956516

ASTM C 88 Test on Soundness of Aggregate Using Sodium Sulfate or Magnesium

Sulfate: A Study of the Mechanisms of Damage

Article in Journal of ASTM International · January 2005

DOI: 10.1520/JAI12517

CITATIONS READS

12 27,454

2 authors, including:

Harvey Hilbert Haynes

Haynes & Associates LLC
32 PUBLICATIONS 367 CITATIONS

SEE PROFILE

All content following this page was uploaded by Harvey Hilbert Haynes on 21 September 2017.

The user has requested enhancement of the downloaded file.

 Journal of ASTM International, January 2005, Vol. 2, No. 1
Paper ID JAI12517
Available online at www.astm.org

Harvey Haynes1

ASTM C 88 Test on Soundness of Aggregate Using Sodium

Sulfate or Magnesium Sulfate: A Study of the Mechanisms of
Damage

ABSTRACT: In this study, the mechanisms of damage to aggregate particles by sodium sulfate and
magnesium sulfate solutions were investigated. By using glass and aluminum containers to represent
aggregate pore space and confining walls, test results showed that during soaking cycles crystal growth
occurred that applied some bearing pressure on the container. This bearing pressure, however, was
secondary to more damaging pressure that occurred during the oven drying cycles. During soaking
cycles, mirabilite (Na2 SO4.10H2O) applied bearing pressures up to about 0.6 MPa. During oven drying
cycles, thenardite applied bearing pressures up to about 2.0 MPa. Magnesium sulfate salts produced
bearing pressures about the same magnitude as that of the sodium sulfate salts.
KEYWORDS: ASTM C 88, force of crystallization, physical salt attack, magnesium sulfate, salt
crystallization, salt hydration distress, salt weathering, sodium sulfate, soundness test

Introduction
In 1828, Brard [3] used sodium sulfate, magnesium sulfate, and other salts to test masonry
stone for durability against weathering. The method consisted of placing pieces of stone in
boiling solutions of saturated salt, and then allowing the stone to dry naturally. Salt
crystallization at or near the surface of the stone caused a scaling type of distress, much like that
of freezing and thawing. Stone that showed relatively large amounts of scaling were rejected for
use in structures.
Variations of the Brard test have been used by geologists to study damage to stone by salt
weathering [4,14]. Using sodium sulfate or magnesium sulfate solutions, ASTM Standard C 88
was introduced in the 1930s to test the soundness of aggregate. The mechanism of distress was
believed to be an increase in the salt volume on conversion from a lower hydrate phase to a
higher phase.
In nature, this volume increase can occur by hydration of salts exposed to changes in ambient
temperature and relative humidity, or by formation of crystals in solution. When in solution, the
term crystallization applies to both the formation of crystals by nucleation and by growth of
existing crystals.
A description of the ASTM C 88-99a [1] test follows. The test starts by using oven-dried
aggregate that has cooled to room temperature. The aggregate is sieved and weighed and then
submerged in a saturated solution of sodium sulfate or magnesium sulfate at 21 ± 1°C for a
period of 16–18 h, removed from the solution, allowed to drain, and placed in an oven at 110 ±
5°C until dry. After drying, the aggregate is cooled to room temperature, submerged again for
16–18 h, and then oven dried. This process is repeated for a number of cycles that are specified
by the purchaser (usually five cycles). The aggregate particles are washed to remove the salt and
then sieved and weighed to determine the amount of breakdown of aggregate particles.

Manuscript received 22 December 2003; accepted for publication 23 September 2004; published January 2005.
1
Principal, Haynes & Associates, Oakland, CA 94611.

Copyright © 2005 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959.
2 JOURNAL OF ASTM INTERNATIONAL

Garrity and Kriege [5] presented the conventional view of damage by the C 88 test. They
primarily studied sodium sulfate. Distress was considered the result of an accumulation of
sodium sulfate in the pores of the aggregate particles from cycles of soaking and drying.
Eventually sufficient thenardite (Na2SO4) existed in aggregate pores so that during the
subsequent soaking cycle, when mirabilite (Na2SO4·10H2O) was formed, the volume of
mirabilite would exceed the volume of the aggregate pore. Mirabilite would thus apply a bearing
pressure to the aggregate pore walls and cause damage. Calculations showed that after the third
soaking cycle, mirabilite would over-fill the pore space.2
Wuerpel [18] expressed a similar opinion regarding the mechanism of damage after
performing C 88 tests using magnesium sulfate.
ASTM C 88-99a states that when using magnesium sulfate the test is usually more severe
than with sodium sulfate. The reason magnesium sulfate is more aggressive has not been
determined. Past studies have mainly used sodium sulfate to study the distress mechanism
[5,17,14,15], likely because this salt is less complex than magnesium sulfate.

Purpose
The purpose of this investigation was to determine the mechanism, or mechanisms, of
damage caused by sodium sulfate and magnesium sulfate. These salts are known to cause
degradation of stone in the field [6], and sodium sulfate is known to cause physical salt attack to
concrete [8,9]. Few studies, if any, have produced convincing evidence of the actual mechanism
of damage.

Properties of Sodium Sulfate and Magnesium Sulfate

Table 1 summarizes various properties of sodium sulfate and magnesium sulfate.
Conversion of anhydrous or lower hydrate phases of salt to higher hydrate phases results in an
increase in crystal volume. For example, the conversion of thenardite (Na2SO4) to mirabilite
(Na2SO4·10H2O) yields an increase in crystal volume of 315 %. However, it should be
mentioned that the change in total volume of water and thenardite combining to form mirabilite
is a decrease of 5.6 %. Similarly with magnesium sulfate, the conversion of kieserite
(MgSO4·H2O) to epsomite (MgSO4·7 H2O) yields an increase in crystal volume of 159 %;
however, the change in total volume of water and kieserite combining to form epsomite is a
decrease of 10.9 %.
The solubility curves of sodium sulfate and magnesium sulfate are shown in Fig. 1, and
specific gravity curves are shown in Fig. 2. For sodium sulfate at 20°C, a saturated solution
occurs at 19.06 g anhydrous salt per 100 g H2O, and for magnesium sulfate, 33.70 g anhydrous
salt per 100 g H2O. After oven drying, thenardite remains for sodium sulfate, while kieserite
remains for magnesium sulfate, and on a volume basis, 127 % more kieserite is deposited than
thenardite. Thus, aggregate pore volume that is flooded with saturated solutions and then oven-
dried will more rapidly accumulate kieserite than thenardite.
Table 1 gives heat of crystallization data for sodium sulfate and magnesium sulfate at 18°C
and infinite dilution. Crystallization of mirabilite is exothermic at –78.5 kJ/mol, while

2
The present author found errors in Garrity and Kriege’s calculation methodology. Revised calculations were made,
which are not given in this paper, that showed aggregate void space did not become filled or over-filled with
hydrated salt regardless of the number of cycles.
 HAYNES ON ASTM C 88 TEST 3

crystallization of thenardite is endothermic at 1.9 kJ/mol. For epsomite, crystallization is

exothermic at –13.3 kJ/mol, while kieserite is endothermic at 58.6 kJ/mol [13]. The C 88 test is
conducted at saturated solution concentrations, which has the effect of reducing the heat of
crystallization. Myerson [12] provided a value of heat of crystallization of –73 kJ/mol for
mirabilite at near saturated conditions.

TABLE 1—Physical properties of sodium sulfate and magnesium sulfate.

Salt, Stable Temp. Stable Molecular Molecular Density of Heat of Coefficient of
Phases Phase, ºC Mass, g/mol Volume Crystal, Crystallization* Thermal**
[12] cm3/mol g/cm3 kJ/mol [13] Expansion,
µm/m/ºC
Sodium Sulfate
Thenardite, 32.4 to 220 142.04 53.0 2.680 + 1.9 28.8
Na2SO4
Mirabilite, 0 to 32.4 322.19 220.1 1.464 -78.5 …
Na2SO4.10H2O
Magnesium
Sulfate
Anhydrous, >180 120.36 45.3 2.657 +88.3 …
MgSO4
Kieserite, 68 to 180 138.40 56.6 2.445 -58.6 23.1
MgSO4.H2O
Hexahydrite, 48.6 to 68 228.48 129.8 1.760 +2.3 …
MgSO4.6H2O
Epsomite, 1.8 to 48.3 246.49 146.7 1.680 -13.3 …
MgSO4.7H2O
Dodecahydrate, –3.9 to 1.8 336.57 … … … …
MgSO4.12H2O
*+endothermic and – exothermic.
**Experimental values determined herein, see Appendix A.

FIG. 1—Solubility of sodium sulfate and magnesium sulfate [10].

4 JOURNAL OF ASTM INTERNATIONAL

FIG. 2—Specific gravity of sodium sulfate and magnesium sulfate [10].

Experimental Scope-of-Work
C 88 tests were conducted using glass and aluminum containers, which represented aggregate
pore space and the confining wall. The volume used for simulated pores was 150 cm3. Tests
using sodium sulfate and magnesium sulfate were conducted. Two to four duplicate tests were
made for each test condition. Thirteen total tests were conducted, where each test consisted of
4–7 cycles of soaking and oven drying.

Test Set-up
Glass and aluminum containers were instrumented with three strain gages and three
thermocouples. The strain gages were mounted on the exterior wall of the container for
tangential strain measurement and were located at distances of 7, 13, and 19 mm from the inside
bottom of the container. The thermocouples were mounted on the inside of the container wall
directly opposite a strain gage. Figure 3 shows the test set-up for the glass containers.
Strain and temperature readings were taken at 20 s intervals during the soaking cycles and
30 s during the oven drying cycles. The thermocouples provided temperature measurements that
appeared to vary randomly by about ± 1.5°C. These data were smoothed by a running average of
20 readings before temperature corrections were made to the strain readings.
Temperature corrections to strain readings were made by the following procedure. Glass and
aluminum containers were subjected to temperature cycles of 22–120°C while empty, filled with
150 cm3 of water, and filled with 150 cm3 of saturate salt solution. The raw strain versus
temperature relationships were linear and essentially equal for the empty and filled containers.
Equations representing the average curve for glass and for aluminum were determined, and these
equations were used to correct individual raw strain measurements for temperature effects during
the C 88 cycles. All strain data presented herein have been temperature corrected.
 HAYNES ON ASTM C 88 TEST 5

FIG. 3—Test set-up using glass containers. Strain gages and thermocouples were located at
7, 13, and 19 mm from the inside bottom of the container. A similar arrangement was used for
the aluminum containers.

Test Results and Discussion

Sodium Sulfate
Soaking Cycles—Soaking cycles started when saturated solution at 20°C was added to the
containers, such that the total volume of solution and oven-dried salt (thenardite) was 150 cm3. A
period of about 24 h was allotted for soaking to observe mirabilite formation. Figure 4
represents the fourth soaking cycle using a glass container, and it shows typical behavior of
strain in the container and temperature in the salt as a function of time.
Temperature Behavior—At the start of the cycle, just after the saturated solution was added
to the container, there was an immediate temperature increase. This temperature increase was
due to exothermic behavior from dissolution of thenardite and crystallization of mirabilite.
Without an increase in temperature, a thermodynamic imbalance would not have existed to
dissolve thenardite, because the added solution was already saturated at 20°C. With an increase
in temperature, the salt solution was able to hold a higher concentration of ions. Once the
temperature of the solution began to decrease, supersaturation of the solution permitted further
crystallization of mirabilite.
An important observation was that only a limited amount of oven-dried thenardite dissolved
during the soaking cycle. The conventional view of the C 88 test was that thenardite dissolved
entirely [5]. Experimentally, two methods showed the limited dissolution of thenardite. The first
method measured the height of the oven-dried salt in the container before saturated solution was
added and then again at the end of the soaking cycle. These measurements were approximate,
because the surface of the salt was rough. A small increase in height of the crystals at the end of
the soaking cycles was observed (Fig. 5). A greater height change was anticipated if all
6 JOURNAL OF ASTM INTERNATIONAL

thenardite converted to mirabilite because the size of mirabilite crystals was about four times the
size of thenardite crystals3.

FIG. 4—Strain in glass container and temperature of sodium sulfate during the fourth
soaking cycle of a simulated C 88 test.

FIG. 5—Height of sodium sulfate salt in glass container at the beginning and end of soaking
cycles.

3
Also, at the end of the soaking cycles, a thin layer of translucent salt was observed on the surface of white salt.
Thenardite is opaque and white, while mirabilite is translucent. Hence, this observation indicated that a shell of
mirabilite deposited on the surfaces of thenardite.
 HAYNES ON ASTM C 88 TEST 7

The second method measured relative temperature behavior of two forms of thenardite
during a soaking cycle. Experience working with sodium sulfate showed that when mirabilite
dried at room conditions (about 20ºC and 50 % RH), the resulting thenardite was light and fluffy
in texture. This air-dried thenardite contrasted with thenardite formed at oven drying conditions,
which resulted in a hard, dense mass of salt. (This hard mass of thenardite was strain gage
instrumented for determination of coefficient of thermal expansion. See Appendix A.) The bulk
specific gravity of air-dried thenardite was about 1.0, and oven-dried thenardite was about 2.1.
(The specific gravity of thenardite is 2.68.) Hence, the surface area of air-dried thenardite was
considerably greater than that of oven-dried thenardite.
A separate soaking-cycle test was conducted where an equal mass of air-dried thenardite
replaced oven-dried thenardite at the start of the third soaking cycle. Figure 6 shows the
temperature behavior during the initial 8 h for both the air-dried and oven-dried thenardite test.
A relatively larger amount of heat was generated by the air-dried thenardite compared to that of
the oven-dried thenardite. The additional heat showed that more air-dried thenardite dissolved
and more mirabilite crystallized. Or, in other words, only a limited portion of the oven-dried
thenardite dissolved. The reason oven-dried thenardite remained undissolved was because a shell
of mirabilite encapsulated the thenardite and prevented rapid dissolution. It was not determined
whether all air-dried thenardite dissolved; however, white salt was observable within translucent
salt, so it was assumed that some air-dried thenardite also remained at the end of the cycle.

FIG. 6—Temperature of oven-dried and air-dried sodium sulfate during a third soaking
cycle.

Interestingly, the maximum temperature in the air-dried thenardite test was about 32.4°C,
which corresponds to the phase change from mirabilite and thenardite (Fig. 1). During the test,
when the temperature exceeded 32.4°C, thenardite would form, and mirabilite would dissolve.
Both of these activities are endothermic. If the temperature decreased below 32.4°C, mirabilite
would crystallize, and thenardite would dissolve; both of these activities are exothermic. Thus,
the temperature was self-regulated at about 32.4°C for approximately 30 min before the
temperature steadily decreased to ambient room temperature.
8 JOURNAL OF ASTM INTERNATIONAL

Strain Behavior—Strain behavior of the container showed that the salt applied a bearing
pressure to the container wall (Fig. 4). The maximum strain in the wall, called the peak strain,
was about 120 µm/m.
During this fourth soaking cycle, the salt in the container ranged in height from about 15–20
mm (Fig. 5), yet the strain gage at 19 mm did not measure appreciable strain. Strain was
measured by gages at 7 and 13 mm from the bottom of the container. This indicated that the salt,
near the top of the salt mass, did not push on the container wall.
The increase in bearing pressure with time could be attributed to crystal growth at the
interface of salt mass and container wall. Becker and Day [2] demonstrated that crystal growth
can apply bearing pressure, and Taber [16] and Hansen [7] discussed the mechanism. A
necessary condition is that supersaturated solution exists at the site of crystal growth.
The solubility of sodium sulfate increases as pressure increases. For this test set-up, the top
of the salt mass had a bearing pressure of essentially zero. Hence, the top surface of the salt mass
should have grown preferentially to the wall and bottom. This would probably occur with
sufficient time; however, the test measurements were taken during significant thermodynamic
imbalance.
Table 2 summarizes the peak strains. For soaking cycles in glass containers, sodium sulfate
showed an average peak strain of 70 ± 40 µm/m, when the peak strains were averaged for cycles
#3 and higher. The maximum peak strain of all the cycles was 150 µm/m. For aluminum
containers, the average and maximum peak strains were higher because these containers had a
larger diameter and thinner wall thickness than that of the glass containers. Hence, for the same
magnitude of bearing pressure, the strain would be greater in the aluminum containers. Later in
this paper, the maximum peak strains will be used to predict bearing pressures applied by the
salts.

TABLE 2—Summary of peak strains in glass and aluminum containers during soaking and
drying cycles.

Sodium Sulfate Magnesium Sulfate

Type of No. of Tests/ Avg. of Peak Max. Peak No. of Avg. of Peak Max. Peak
Cycle and Cycles* Strains from Strain µm/m Tests/ Strains from Strain µm/m
Container Cycles µm/m Cycles* Cycles, µm/m
Soaking
Glass 4 / 18 70a ± 40 150 a 2/8 110 a ± 60 200 a
Aluminum 4 / 16 130 a ± 60 300 a 3 / 12 290 a ± 130 480 a

Drying
Glass 4 / 13 240 b ± 110 430 b 2/8 270 c ± 140 430 c
2/8 260 d ± 80 380 d
Aluminum 4 / 11 110 b ± 40 200 b 3 / 12 840 d ± 220 1200 d
*Total number of cycles used in averaging peak strains. For each test, strain data from the 1st and 2nd cycles were
not used in the averages.
a
Crystal growth of mirabilite or epsomite.
b
Coefficient of thermal expansion of thenardite.
c
Coefficient of thermal expansion of salt mass composed of kieserite, epsomite, and hexahydrite during first 2 h
of oven drying as the temperature increased to about 75ºC.
d
Crystal growth of magnesium sulfate dihydrate at temperatures above 85°C.
 HAYNES ON ASTM C 88 TEST 9

Oven Drying Cycles—Behavior of the solution–salt mass system was more complex during
oven drying than during soaking. When glass containers were used, the containers broke during
the oven drying cycles. Four tests were conducted with glass containers, and the containers broke
at cycles #3, 4, 7, and 7.4
The glass containers broke in a similar manner. Although vertical cracks were found in the
containers, it appeared that the significant crack was circumferential around the bottom of the
containers near the wall to bottom transition. This circumferential crack gave the appearance of
the wall being pushed out too far or the bottom being pushed off, or both.
Figure 7 shows strains and temperatures in a glass container during the sixth drying cycle.
The temperature of the salt showed three plateaus. The first plateau, at about 80°C, was due to
evaporative cooling. (At 9 h the abrupt increase in temperature was due to tightly closing the
oven door, which was inadvertently ajar.) The second plateau was at a temperature of about
105°C, which was caused by the formation of a salt crust that spanned the container and
decreased the effects of evaporative cooling. The third plateau, at about 110°C, was after all
H2O evaporated.
Drying cycles typically required two days to complete because evaporation was significantly
hindered by the formation of the salt crust. When a salt crust was intentionally broken, the
temperature of the salt solution decreased to about 80°C. Garrity and Kierge [5] also observed
the formation of salt crusts.

FIG. 7—Strain in glass container and temperature of sodium sulfate during the sixth oven
drying cycle.
4
Prior to this experimental program, exploratory tests were conducted on glass containers without instrumentation.
A total of 16 containers broke during oven drying cycles using sodium sulfate, magnesium sulfate, and sodium
carbonate. Containers did not break using sodium chloride.
10 JOURNAL OF ASTM INTERNATIONAL

Strain behavior basically followed temperature. During increases in temperature, the strains
in the container increased. A large portion of the salt mass at the start of oven drying cycles was
thenardite, which was surrounded by a shell of mirabilite. As the oven temperature increased
from room temperature to 32ºC, thermal expansion of this salt mass caused the salt to push
against the glass container. As the temperature rose above 32.4°C, mirabilite melted, and the
strain in the container decreased. Only thenardite remained, and with further increases in
temperature, thermal expansion caused thenardite to push against the container. The coefficient
of thermal expansion of thenardite was determined as approximately 28.8 µm/m/°C (Appendix
A). The coefficient of thermal expansion of Pyrex glass is 3.4 µm/m/ºC. Hence, expansion of
thenardite was considerably greater than the increase in the diameter of the glass container for a
given change in temperature5.
This mechanism of applying load to the container wall was verified by the tests using
aluminum containers. Thenardite applied a bearing pressure to the aluminum container that was
less than that applied to the glass container. This was shown by the peak strain data (Table 2).
The coefficient of thermal expansion of aluminum was 23.9 µm/m/°C, so the aluminum
containers expanded more than the glass containers, and the thenardite was not as highly
confined.
Table 2 shows the average peak strain for the glass container was 240 ± 110µm/m, and the
maximum peak strain was 430 µm/m. For the aluminum container, the average peak strain was
110 ± 40 µm/m, and maximum peak strain was 200 µm/m. For mineral aggregates, such as
granite, the coefficient of thermal expansion ranges from 2–12 µm/m/°C and for sandstone from
4–14 µm/m/°C. The glass container at 3.4 µm/m/°C was closer to representing aggregate than
that of the aluminum container at 23.9 µm/m/°C. Hence, damage would occur to aggregate by
thermal behavior of thenardite during oven drying.
During the first and second temperature plateaus, the strains slowly decreased. This was
likely due to creep relaxation of the mass of thenardite. During the third temperature plateau,
after all H2O had evaporated, the oven-dried mass of thenardite did not show creep relaxation.
Two glass containers broke during the temperature increase from the first to second plateau
(80–105°C), and two containers broke during the temperature increase from the second to third
plateau (105–110°C).

Magnesium Sulfate
Soaking Cycles—Figure 8 represents the third soaking cycle using a glass container, and it
shows typical behavior of strain in the container and temperature of the salt as a function of time.
At the start of the cycle, just after saturated solution was added to the container, which held
oven-dried kieserite, there was an increase in temperature of the solution. This activity was
similar to that of the sodium sulfate test, but the rate of temperature gain was slower with
magnesium sulfate. The temperature increased for the first ½ h, which was related to kieserite
dissolving and epsomite crystallizing. Both activities were exothermic. The maximum
temperature at about 31ºC was within the stable range of temperatures for epsomite (1.8–
48.3ºC).

5
If the coefficient of thermal expansion of the salt and the container material were equal, then an increase in
diameter of the salt mass would equal the increase in diameter of the container for a given change in temperature.
 HAYNES ON ASTM C 88 TEST 11

FIG. 8—Strain in glass container and temperature of magnesium sulfate during the third
soaking cycle.

The initial strains, which occurred during the temperature increase, were likely related to
thermal expansion of the mass of kieserite, which was surrounded by a shell of epsomite. As the
temperature decreased to room temperature there was a slight decrease in strain in the container,
which was related to thermal behavior. When the strains started to increase again at about 2 to 3
h, this was from slow crystal growth of epsomite from the supersaturated solution. The strains
increased to a peak strain of about 200 µm/m at 23 h., when the cycle was stopped.
Table 2 summarizes the peak strains. For the glass containers, the average peak strain was
110 ± 60 µm/m, and the maximum peak strain was 200 µm/m. For the aluminum containers, the
peak strains were greater because of the thinner wall thickness and larger diameter.

Oven-Drying Cycles—Figure 9 shows the strains and temperatures in a glass container

during the fifth drying cycle. The temperature behavior showed the same general features as that
of sodium sulfate, but not as pronounced. The first temperature plateau was at about 95°C, the
second about 108°C, and the third about 115°C.
As a reminder, at the end of the soaking cycles, the salt mass was mostly kieserite with a
shell of epsomite (and perhaps some metastable hexahydrite). During the beginning of the oven
drying cycle, the initial strains in the container were due to thermal expansion of the salt mass as
the temperature increased from room temperature to about 68°C. The coefficient of thermal
expansion of epsomite is not known; however, the coefficient for kieserite was determined to be
about 23.1 µm/m/°C (Appendix A).
12 JOURNAL OF ASTM INTERNATIONAL

FIG. 9—Strain in glass container and temperature of magnesium sulfate during the fifth oven
drying cycle.

Figure 9 shows that at a temperature of about 68°C, the strains started to decrease to zero.
Above 48°C epsomite is unstable, and above 68°C hexahydrite is unstable, so these crystals
evidently melted, and the remaining mass of kieserite was not in contact with the glass wall.
After the temperature increased above about 85°C and during the following 3.5 h (from 1.5–5.0
h), the strains increased steadily to a maximum of 320 µm/m. This increase in strain indicated
that a hydration phase of magnesium sulfate was crystallizing and applying a bearing pressure on
the container. This increase in strain was not related to thermal behavior. It appeared that the
metastable dihydrate phase was crystallizing. Evidence of magnesium sulfate dihydrate was
shown at 18 h, when the strain started decreasing to zero. The dihydrate phase lost H2O, and a
volume decrease occurred. The dihydrate phase reportedly melts at a temperature of 112°C [18],
which was the approximate temperature of the salt at the time of strain decrease. Once all
evaporable water was lost, the temperature increased slightly to 115°C (at about 26 h). At this
condition, kieserite was the remaining phase. Kieserite is stable to 180°C.
Table 2 summarizes the average and maximum peak strains that occurred during the initial
thermal expansion stage and during the crystallization stage. Strains from both stages were
similar in magnitude.
The aluminum containers showed different behavior than that of the glass containers during
the crystallization stage; here the bearing pressure of the salt caused yielding of the aluminum.
Figure 10 shows the second drying cycle in an aluminum container. The strain in the aluminum
steadily increased over a period of 55 h and reached 1100 µm/m. Again, this increase in strain
was not related to thermal behavior. Upon removal of load, much of the strain was permanent.
During subsequent oven drying cycles, the peak strain increased substantially.
 HAYNES ON ASTM C 88 TEST 13

FIG. 10—Strain in aluminum container and temperature of magnesium sulfate during the
second oven drying cycle.

Two glass containers were tested, and one container broke at the end of cycle #4 and the
other at cycle #6. Both containers broke when the temperature just started to increase from the
second to the third plateau. During exploratory tests, some glass containers with magnesium
sulfate broke earlier in the oven drying cycle, during the time of crystallization of magnesium
sulfate dihydrate.

Calculations of Salt Bearing Pressures

Empirical data are scarce on bearing pressures of salt, also called forces of crystallization.
This study calculated the bearing pressures by using the strain data from the glass and aluminum
containers. A structural model of the containers was analyzed using the general purpose finite
element program SAP 90.
The models represented one-quarter of a glass or aluminum container, where the base was
free to move horizontally, but not vertically. A salt mass in the glass container had a height of
23.1 mm, and in the aluminum container 17.8 mm. It was assumed that the salt applied a
uniform bearing pressure on the container wall. This assumption did not model actual
conditions, as the strain data showed a non-uniform distribution of bearing pressures. The effect
on the results was to underestimate the bearing pressures actually applied to the containers.
The glass containers had an interior diameter of 64.0 mm, height of 88.9 mm, and wall
thickness of 2.30 mm. The aluminum containers had an interior diameter of 71.4 mm, height of
52.1 mm, and wall thickness of 0.76 mm. The modulus of elasticity and Poisson’s ratio for glass
was 64.0 GPa and 0.20, respectively, and for aluminum 69.0 GPa and 0.33.
The calculations were made using an assumed bearing pressure of 0.7 MPa. For this loading
condition, the maximum tangential strain in the glass container was calculated as 150µm/m, and
14 JOURNAL OF ASTM INTERNATIONAL

in the aluminum container as 500 µm/m. Using the maximum peak strain in the containers, the
experimental bearing pressures of the salts were calculated by proportion.
Table 3 shows the bearing pressures of the salts, as calculated from maximum peak strains.
In summary, during soaking cycles mirabilite developed a bearing pressure of about 0.6 MPa,
and epsomite 0.8 MPa. During oven drying, thenardite and kieserite showed a bearing pressure
of about 2.0 MPa caused by thermal expansion. The bearing pressure from crystallization of
magnesium sulfate dihydrate was about 1.8 MPa in the glass containers. However, it was during
this portion of the test that the bearing pressure caused yielding of the aluminum container wall.
Hence, the greatest bearing pressure occurred, which means it was greater than 2.0 MPa, but no
estimated bearing pressure is given. Bearing pressures that developed during oven drying were
two to three times those during soaking cycles.
During soaking cycles the ambient temperature was representative of field conditions, so the
bearing pressures of 0.6 and 0.8 MPa for mirabilite and epsomite can likely occur in the field to
cause salt weathering of rocks, concrete, and other materials. However, oven drying is an
artificial condition, so the bearing pressures of 1.8 MPa or greater would not likely be
experienced in the field.

Discussion on Improvement of ASTM C 88 Test

Past studies to improve the repeatability and reproducibility of the C 88 test [5,18,14,15]
investigated parameters deemed important by the conventional view of distress, which
emphasized factors related to the soaking cycles. Much effort was misdirected because the
present study has shown that the oven drying cycles were more damaging. Improvements in
precision are possible by evaluating parameters related to oven drying.
The current C 88 method of drying to constant mass, which is defined as achieving a mass
loss of less than 0.1 % of sample mass in a 4 h drying period, does not assure complete removal
of all evaporable water. Complete drying is a significant parameter when using sodium sulfate
because the most damaging stage of the test occurred at the temperature increase related to
evaporation of all water. Without complete drying, this stage is missed. The formation of salt
crusts slows the evaporation of water, such that longer than expected time periods may be
required for oven drying.
Temperature control during oven drying is also important. Temperatures above the permitted
maximum of 115°C may cause extra damage from thermal expansion because the change in
temperature from the second to the third plateau will be larger than typical. Temperatures less
than the permitted minimum of 105°C may cause less damage. It may also be found that the
permitted temperature range of 105–15°C is too great, that is, damage caused at 115°C may be
significantly greater than that caused at 105°C.
It appears that a C 88 test using magnesium sulfate should be less sensitive to oven drying
conditions than that of sodium sulfate because the test was relatively independent of the
maximum oven temperature and the range of oven temperatures of 110 ± 5°C. Damage by
magnesium sulfate was related to thermal expansion of kieserite at low temperatures (20–75ºC)
and to crystal growth of magnesium sulfate dihydrate at temperatures above about 85°C. In fact,
ASTM C 88-99a shows better precision for magnesium sulfate than for sodium sulfate, which
supports the above reasoning.
Another parameter for investigation is the removal of oven-dried salt from the pores of
aggregate by washing with water at the end of the test. The oven-dried salts do not dissolve
quickly.
 HAYNES ON ASTM C 88 TEST 15

TABLE 3—Calculated bearing pressures by sodium sulfate and magnesium sulfate.

Cycle and Salt Container Max. Peak Strain Calculated Bearing Pressure
in Container, µm/m of Salt, MPa
Soaking
Mirabilite Glass 150 0.7
Aluminum 300 0.4
Average = 0.6a
Epsomite Glass 200 0.9
Aluminum 480 0.7
Average = 0.8b
Drying
Thenardite Glass 430 2.0c
Aluminum 200 …d
Magnesium Sulfate Glass 430 2.0c
Dihydrate Glass 380 1.8e
Aluminum 1200 …f
a
Bearing pressure due to crystal growth of mirabilite.
b
Bearing pressure due to crystal growth of epsomite.
c
Bearing pressure due to thermal expansion of salt.
d
Bearing pressure not given because aluminum container expanded only slightly more than thenardite, so
calculated bearing pressure was low.
e
Bearing pressure due to crystal growth of magnesium sulfate dihydrate.
f
Bearing pressure not calculated because aluminum container showed considerable yielding.

Summary
Experimental tests were conducted using glass and aluminum containers to represent
aggregate pore space and wall. All tests on glass containers resulted in the glass breaking during
the oven drying cycles.
This investigation found that the mechanisms of damage by the ASTM C 88 test were
different than those of the conventional view. It was shown by experiment that oven-dried salt
showed limited dissolution during soaking cycles in saturated solution, that a shell of hydrated
salt formed on the oven-dried salt during soaking cycles, and that crystal growth of hydrated salt
slowly applied a bearing pressure to the container, yet this pressure was not the primary damage
mechanism. Maximum damage occurred during the oven drying cycles. Here, sodium sulfate as
thenardite caused damage due to thermal expansion. The coefficient of thermal expansion of
thenardite was larger than that of glass and aluminum, and, for that matter, mineral aggregate.
During oven drying, when the oven temperature was constant, two stages occurred that resulted
in increases in temperature of the salt system. The first stage was formation of a salt crust that
reduced evaporative cooling effects, and the second stage was complete evaporation of water.
The resultant temperature increases caused thenardite to bear more strongly against the container
wall. For magnesium sulfate, the damaging processes were due to thermal expansion at the
beginning of the oven drying cycle when temperatures were less than 68°C, and then due to
crystallization of magnesium sulfate dihydrate at temperatures above about 85°C.
Strain measurements from the containers were used to calculate the experimental bearing
pressure of the salts. During soaking cycles, the hydrated salts showed bearing pressures that
slowly increased to about 0.6 MPa for mirabilite and 0.8 MPa for epsomite. During the oven
drying cycles, sodium sulfate and magnesium sulfate salts showed a bearing pressure of about
2.0 MPa.
16 JOURNAL OF ASTM INTERNATIONAL

Previous observations that magnesium sulfate is more damaging than sodium sulfate during
C 88 tests can be explained by these findings. First, magnesium sulfate solutions deposited a
greater volume of salt after each soaking and drying cycle than that of sodium sulfate. Hence,
the bearing pressure applied by magnesium sulfate, which was essentially similar in magnitude
to that of sodium sulfate, created more damage because the pressure acted on a larger area of the
aggregate pore walls. Second, magnesium sulfate solutions caused damage at lower temperatures
than sodium sulfate solutions. The magnesium sulfate test was less dependent on controlling
maximum oven temperature and obtaining completely dry test samples, so more repeatable test
results could be expected than from the sodium sulfate test.

Acknowledgments
Appreciation is extended to Dr. Julio Miranda of New Dynamics, Fremont, CA, for his
structural analysis of the containers, and to Mr. Perry Joiner, technician, for his assistance with
instrumentation and data reduction.

References
[1] ASTM Standard C 88-99a, Test Method for Soundness of Aggregates by Use of Sodium
Sulfate or Magnesium Sulfate, Annual Book of ASTM Standards, Vol. 4, No. 2, 2003, pp.
42–46.
[2] Becker, G. F. and Day, A. L., “The Linear Force of Growing Crystals,” Proceedings of the
Washington Academy of Sciences, Vol. VII, 1905, pp. 283–288.
[3] Brard, “On the Method Proposed for Immediate Detection of Stones Unable to Resist the
Action of Frost,” Annales de Chimie et de Physique, 38, 1828, pp. 160–192.
[4] Evans, I. S., “Salt Crystallization and Rock Weathering: A Review,” Revue de
Geomorphologue Dynamique, 19, 1970, pp. 155–177.
[5] Garrity, L. V. and Kriege, H. K., “Studies of the Accelerated Soundness Tests,” Highway
Research Board, Vol. 15, 1935, pp. 237–258.
[6] Goudie, A. and Viles, H., Salt Weathering Hazards, John Wiley & Sons, Chichester, 1997.
[7] Hansen, W. C., “Crystal Growth as a Source of Expansion in Portland Cement Concrete,”
ASTM Proceeding, Vol. 63, 1963, pp. 932–945.
[8] Haynes, H., O’Neill, R., and Mehta, P. K., “Concrete Deterioration from Physical Attack by
Salts,” Concrete International, Vol. 18, No. 1, January 1996, pp. 63–68.
[9] Hime, W., Martinek, R. A., Backus, L. A., and Marusin, S. L., “Salt Hydration Distress,”
Concrete International, Vol. 23, No. 10, October 2001, pp. 43–50.
[10] Linke, W. F., Solubilities: Inorganic and Metal-Organic Compounds, American Chemical
Society, 4th ed., 1964.
[11] Montillon, G. H. and Badger, W. L., “Rate of Growth of Crystals in Aqueous Solution,”
Industrial and Engineering Chemistry, TP 1.I6, Vol. 19, July 1927, pp. 809–816.
[12] Myerson, A. S., Handbook of Industrial Crystallization, Butterworth-Heineman, 1993, p. 4.
[13] Perry, R. H. and Green, D. W., Chemical Engineering Handbook, 6th ed., McGraw-Hill
Book Co., New York, 1984.
[14] Price, C. A., “The Use of Sodium Sulfate Crystallization Test for Determining the
Weathering Resistance of Untreated Stone,” UNESCO/Rilem International Symposium
Determination and Protection of Stone Monuments, Reilure, Paris, 3.6, 1978, 23 pp.
 HAYNES ON ASTM C 88 TEST 17

[15] Sheftick, W., “Na2SO4 Soundness Test Evaluation,” Cement, Concrete, and Aggregates,
Vol. 11, No. 1, ASTM International, West Conshohocken, PA, 1989, pp.73–79.
[16] Taber, S., “The Growth of Crystals Under External Pressure,” American Journal of Science,
Vol. XLI, 1916, pp. 532–557.
[17] Walker, S. and Proudley, C. E., “Studies of Sodium and Magnesium Sulfate Soundness
Tests,” Appendix II, Report of Committee C-9, ASTM Proceeding, Vol. 36, Part 1, 1936, pp.
327–338.
[18] Wuerpel, C. E., “Factors Affecting the Testing of Concrete Aggregate Durability,”
Appendix II, Report of Committee C-9, ASTM Proceeding, Vol. 38, 1938, pp. 327–351.

Appendix A
Coefficient of Thermal Expansion of Thenardite and Kieserite
The coefficient of thermal expansion of thenardite and kieserite was experimentally
determined by using oven-dried samples of the salts. These samples were instrumented with
strain gages and cycled though the temperature range of 22–110°C. In addition, samples of
carbon steel and glass and aluminum from the containers were instrumented and cycled along
with the salt samples. Raw strains versus temperature measurements were plotted for each
material, and linear relationships were found. A slope to each curve was obtained that
represented raw strain per °C.
The experimental coefficients of thermal expansion for the salts were determined by using a
method of differences. Steel was used as the known condition, where the coefficient of thermal
expansion for carbon steel was 11.7 µm/m/ºC. Table A shows the data and method of
differences. The experimental coefficients for glass and aluminum were near those of handbook
values. The coefficient of thermal expansion for thenardite was determined as 28.8 µm/m/°C,
and for kieserite 23.1 µm/m/˚C.

TABLE A—Determination of coefficient of thermal expansion of thenardite and kieserite.

Material Slope to Curve of Raw Strain Experimental Coefficient of Handbook Value of
Versus Temperature, µm/m/ºC Thermal Expansion, µm/m/ºC Coefficient of Thermal
Expansion, µm/m/ºC
Steel, Carbon –1.3 13.0* – 1.3 = 11.7 11.7a
Glass, Sodium –10.7 13.0 – 10.7 = 2.3 3.4b
Borosilicate
Aluminum, +10.9 13.0 + 10.9 = 23.9 19.5 – 24.2b
3003-0
Oven-Dried
Thenardite,
Sample #1 +17.4 13.0 + 17.4 = 30.4
Sample #2 +14.2 13.0 + 14.2 = 27.2
Average = 28.8
Oven-Dried +10.1 13.0 + 10.1 = 23.1
Kieserite
*This coefficient of thermal expansion represents that of the strain gage itself, and its value was selected because it
gave the proper coefficient of thermal expansion for carbon steel.
a
Manual of Steel Construction, American Institute of Steel Construction, New York, 6th ed., 1964, p. 6.11.
b
Callister, W. D., Jr, Materials Science and Engineering: An Introduction, John Wiley & Sons, 4th ed., 1997, pp.
788–789.

View publication stats