Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Determination of Gypsum Content in Dryland Soils

Exploiting the Gypsum–Bassanite Phase Change


I. Lebron* The presence of gypsum in soil, even in small amounts, is relevant from the genetic, taxonomic, and
Univ. of the West Indies applied points of view. Moreover, those soils having gypsum as the main component, for example
Dep. of Food Production gypseous soils, host a number of rare and endangered organisms. Determining gypsum content
St. Augustine in soils is crucial to understanding their behavior; however, current methods of determination
Trinidad and Tobago are cumbersome or imprecise. This study was conducted to develop an accurate new method
to determine soil gypsum content by reducing the inaccuracies or poor assumptions in previous
J. Herrero methods. We determined gypsum content in samples using quartz sand and clay minerals mixed
Estación Experimental de Aula Dei with gypsum reagent powder to 0 and approximately 2, 5, and 50% gypsum content; we also
CSIC studied seven gypseous soil samples. Gypsum (CaSO4⋅2H2O), after heating at 105°C, dehydrates,
PO Box 13034 losing the two water molecules of the gypsum crystal cell and becoming mostly anhydrite
50059 Zaragoza, Spain (CaSO4). Anhydrite is very hygroscopic, absorbing atmospheric water very quickly; this process
has been found to occur even in desiccators where the relative humidity (RH) is between 12 and
D.A. Robinson 20% depending on the desiccant. Anhydrite rehydrates under laboratory conditions to bassanite
Univ. of the West Indies
(CaSO4⋅0.5H2O), which is stable at RH values <100%. We propose exploiting this difference in
Dep. of Food Production
mass between gypsum and bassanite to quantify the gypsum content of soils. We found that by
St. Augustine
equilibrating the soils at constant RH, e.g., 40%, before and after heating the samples, the clay
Trinidad and Tobago
water content is constant, eliminating a major error with other thermogravimetric methods. This
simple, yet effective method determined gypsum content to within 1% in all the mineralogies
studied, whereas other oven-based methods overestimated gypsum content by as much as 36% in
a pure Ca-montmorillonite sample.

Abbreviations: ECe, electrical conductivity of a saturation extract; RH, relative humidity.

PEDOLOGY
ypsum is a major component of soils in many arid and are based on dissolution of gypsum from the soil, selective pre-
semiarid areas of the world. It has been estimated that cipitation of the gypsum by adding acetone, and dissolution
more than 100 million ha of the world’s soils are classified as of the reprecipitated gypsum with water. The solution can be
Gypsisols or soils containing >5% gypsum (FAO, 1990; Verheye analyzed by measuring electrical conductivity (EC), since there
and Boyadgiev, 1997). It has been found that Gypsisols are im- is a correlation between the amount of dissolved salts and EC
portant as a refuge for rare, vulnerable, and endangered species (U.S. Salinity Laboratory Staff, 1954), or by analyzing SO42−
like bryophytes and lichens. About 20% of the total bryophyte (Loeppert and Suarez, 1996) or Ca2+ (van Reeuwijk, 1987).
and lichen flora are rare or endemic species that live exclusively There are variations depending on the method used, but since
or almost exclusively on gypsiferous substrates (Guerra et al., we cannot be certain that either SO42− or Ca2+ came exclusive-
1995). These habitats are ranked among the most threatened ly from gypsum, especially in arid soils, and the exchangeable
ecosystems worldwide, and are considered a priority conserva- cations in the clay complex may add or remove Ca from the
tion concern (Belnap et al., 2001; Mota et al., 2003). solution, error is unavoidable.
There are a number of methods in the literature to quan- X-ray diffraction is good as a qualitative method for deter-
tify the gypsum content of soils. In general, these methods can mining the presence or absence of gypsum (Khan and Webster,
be classified into three groups: wet chemical methods, thermo- 1968). It is relatively easy to obtain measurements, provided one
gravimetric methods, and x-ray techniques. The wet chemical has access to the expensive apparatus. Quantitative determina-
methods are time consuming and have intrinsic problems; they tion is cumbersome and time consuming, however, and be-
comes especially difficult in mixed-mineralogy soils. Therefore,
Soil Sci. Soc. Am. J. 73:403-411 a number of researchers have suggested measuring the water
doi:10.2136/sssaj2008.0001 loss from gypsum as a simple, rapid, and cost-effective way of
Received 1 Jan. 2008. determining soil gypsum content.
*Corresponding author (inma_lebron@yahoo.com).
Thermogravimetric methods have been considered for
© Soil Science Society of America
a long time as a semiquantitative approach; in Loeppert and
677 S. Segoe Rd. Madison WI 53711 USA
All rights reserved. No part of this periodical may be reproduced or
Suarez (1996), the method of Nelson et al. (1978) was rec-
transmitted in any form or by any means, electronic or mechanical, ommended as a precursor to the wet chemical method to
including photocopying, recording, or any information storage and determine the dilution needed to dissolve all the gypsum in
retrieval system, without permission in writing from the publisher. the sample. This method is based on the difference between
Permission for printing and for reprinting the material contained the masses of a sample equilibrated for 48 h in a desiccator
herein has been obtained by the publisher. with silica gel and after drying in an oven at 105°C for 24 h;

SSSAJ: Volume 73: Number 2 • March–April 2009 403


reagent. Samples of Silver Hill il- Table 1. Chemical composition, pH, and electrical conductivity (EC) of the extracts from the
soils used in this study.
lite from Montana (IMt-1) and
Ca-montmorillonite from Cheto, Saturation extract chemical composition
Soil Depth Symbol pH EC
AZ (SAz-1) were obtained from the Ca2+ Mg2+ Na+ K+ SO42− Cl− NO3−
Clay Mineral Society’s Source Clay cm dS m −1 ———————mmolc L−1————————
Repository. We also used Wyoming 0–4 MT 7.6 2.56 33.2 1.73 0.96 0.51 31.6 0.77 0.14
bentonite (Aqua Technologies of Moab 4–10 MB 7.2 2.55 32.4 1.47 0.54 0.35 30.2 0.38 0.09
Wyoming, Casper), kaolinite from 0–4 SR1T 7.5 2.49 31.2 1.22 1.67 0.22 28.1 1.23 0.23
the Lamar pit (Bath, SC), and quartz SR1 4–12 SR1M 7.7 2.38 34.8 2.10 0.53 0.51 29.6 0.94 0.38
sand (Norco, CA). Reagent-grade, 12–20 SR1B 7.8 2.33 28.9 0.84 0.50 0.19 26.7 0.26 0.10
?200-μm gypsum was used as the 0–1 SR2T 7.5 2.09 31.5 4.82 1.61 0.39 33.3 0.88 0.32
control (Thomas Baker, Mumbai, SR2 1–10 SR2B 7.5 2.51 28.3 1.05 0.95 0.19 27.1 0.74 0.37
India). The SAz-1 montmorillonite Sandy soil 0–4 Sandy 7.7 0.5 9.1 1.0 1.14 0.00 7.3 0.70 0.32
was saturated with either Ca or Mg to
produce monosaturated clay samples cator with drierite. We prepared a second desiccator with a saturated
(Goldberg and Glaubig, 1987). Mixtures were prepared containing a solution of K2CO3; the RH determined by the sensor was 41 ± 3%.
mineral/gypsum air-dry ratio of ?0, 2, 5, and 50% gypsum content The advantage of using a solution of K2CO3 is that the equilibrium RH
by mass. In the case of the Ca- and Mg-saturated montmorillonite, is constant across a wide range of temperatures (Greenspan, 1977) and
only the 0, 5, and 50% gypsum content samples were prepared due to was also close to the RH in the laboratory. In this way, abrupt changes
the small quantities of the monosaturated clays available. in the RH were minimized when opening and closing the desiccator.
We collected three gypseous soils for this study. We use the term The soils and mixtures of clay minerals with gypsum were placed
gypseous, which implies that the soils are composed of gypsum and have in the desiccator with drierite with the RH sensor inside so that we
the characteristics of gypsum, rather than the term gypsiferous, which could read the RH inside the desiccator without opening it. We mea-
merely acknowledges any presence of gypsum (Gove, 1993). The sured the mass of the samples once a day for 5 d and monitored the
gypseous soils were collected from the Colorado Plateau. The first was RH inside the desiccator every day. We found that the mass of the
from Moab, UT (38°34′15.24″ N, 109°33′52.94″ W). The gypseous samples and the RH inside the desiccator was constant after 48 h,
soil occurs as an outcrop that is not mapped independently by the soil but we continued taking readings of weight and RH up to 5 d to
survey and occurs in the Thoroughfare fine sandy loam series (coarse- make sure no further changes occurred. Different sets of samples were
loamy, mixed, superactive, calcareous, mesic Typic Torrifluvents) in placed in the desiccator with K2CO3; we weighed the samples and
the map unit of the Canyonlands Area. The other two soils were monitored the RH inside the desiccator also for 5 d; we found con-
from the San Rafael Swell: San Rafael 1 (SR1, 39°3′28.74785″ N, stant mass and RH after 48 h. We then put all the samples in the
110°54′3.33300″ W) from the Robroost–Mussentuchit association oven at 105°C for 24 h. The samples that had been in the drierite
and San Rafael 2 (SR2, 39°2′31.78168″ N, 110°54′43.18264″ W) were weighed immediately after removal from the oven; the samples
from the Goblin–Gypsum Land–Mussentuchit complex; both had previously equilibrated with K2CO3 were put back in to the desicca-
gypsum as a parent material (Soil Survey Staff, 2008). The Moab soil tor with the K2CO3 and equilibrated for another 48 h. After 48 h, we
had two very contrasting colored layers in the soil profile; we took weighed the samples and calculated the gypsum percentage according
samples from each layer, 0 to 4 cm (Moab top, MT) and 4 to 12 cm to the Nelson method and the method proposed in this study.
(Moab bottom, MB). The first soil from San Rafael (SR1) was located Desiccators do not prevent rehydration of the samples; the amount
in the lower part of a valley; we sampled three depths with different of time the sample spends in the desiccator, and the initial temperature
colorations in the profile at 0 to 4, 4 to 12, and 12 to 20 cm (SR1T, at which the sample was heated affect the amount of rehydration of the
SR1M, and SR1B, respectively); the second sample (SR2) was on the samples. To quantify the amount of water taken up by our samples in
top of a plateau and the profile was very homogeneous except for the desiccator, we took soil samples (in triplicate) and one sample of
the top 1 cm (SR2T and SR2B). For comparison, we also collected gypsum reagent and heated them at 50°C in a convection oven for 24 h.
a sandy soil from the Moab area with no gypsum (Sandy), the total We weighed the samples directly from the oven, placed the samples in
number of soil samples tested came to eight. After sieving the
Table 2. Semiquantitative analysis of the mineralogy of the soils used
samples through a 2-mm screen, we prepared a saturation paste
in this study using x-ray diffraction. Calculated percentage of gypsum
for each of the soils and the saturation extract was analyzed based on weight loss at 105°C using the Nelson et al. (1978) method
for major cations and anions, electrical conductivity, and pH and the bassanite method.
(Table 1). We also sieved portions from each of the samples
X-ray diffraction Gypsum content
through a 500-μm sieve and analyzed the mineralogical compo- Soil
Gypsum Quartz Calcite Bassanite method Nelson method
sition of our soils using x-ray diffraction (Siemens P4, Siemens
——————————— % ———————————
USA, New York City), Table 2. Semiquantitative analysis
MT 40 60 49 52
was performed using the software package X’Pert HighScore
MB 100 97 98
(PANalytical, Almelo, the Netherlands).
SR1T 40 50 10 26 30
We prepared a desiccator by placing drierite (W.A. SR1M 64 27 9 75 76
Hammond Drierite Co., Xenia, OH) in an oven at 150°C SR1B 100 trace 92 91
for 3 h and transferring it to the desiccator to cool down. The SR2T 80 20 69 71
RH inside the desiccator was measured using an RH sensor SR2B 100 94 96
(Thermo-Fisher, Waltham, MA); the RH was 12% in the desic- Sandy 100 0.1 4

SSSAJ: Volume 73: Number 2 • March–April 2009 405


Table 3. Water loss of the soil samples measured directly from the oven and after leav- was achieved after 2 h but we continued the
ing the samples in the desiccator with anhydrite (12% relative humdity) for 3 h. We experiment for a total of 7 d to make sure there
used fresh samples for each temperature.
was no more water uptake.
Measurement Weight loss We measured the water potential of
Soil
condition 50°C 70°C 105°C 200°C kaolinite, illite, Wyoming bentonite, and Ca-
——————— % ————–— and Mg-montmorillonite using a WP4-T
no desiccator 0.48 ± 0.03 0.67 ± 0.02 10.41 ± 0.21 12.99 ± 0.35 Dew-point Potentiometer (Decagon Devices,
MT
desiccator 0.22 ± 0.03 0.48 ± 0.05 9.53 ± 0.28 9.94 ± 1.05 Pullman, WA). The potentiometer measures
no desiccator 1.16 ± 0.31 1.84 ± 0.54 19.80 ± 0.05 20.80 ± 0.04 the combined effect of matric and osmotic
MB
desiccator 0.03 ± 0.06 0.00 ± 0.10 17.79 ± 1.30 18.18 ± 1.09 potentials; in samples with low salt contents,
no desiccator 0.57 ± 0.04 0.76 ± 0.02 5.97 ± 0.11 8.48 ± 0.66 the potential measured is considered to re-
SR1T
desiccator 0.33 ± 0.07 0.58 ± 0.02 5.32 ± 0.13 5.87 ± 0.49 flect the matric potential only. Clays were
no desiccator 0.61 ± 0.09 0.63 ± 0.08 15.16 ± 0.03 18.94 ± 0.36 equilibrated with different water contents
SR1M
desiccator 0.00 ± 0.05 0.13 ± 0.03 13.66 ± 0.14 14.60 ± 1.22 overnight; the potential was measured using
no desiccator 0.42 ± 0.08 0.65 ± 0.05 18.12 ± 0.04 20.55 ± 1.65 the WP4-T in a constant-temperature room
SR1B
desiccator 0.00 ± 0.11 0.11 ± 0.04 16.54 ± 0.62 16.94 ± 1.59 at 22°C. The salt content in the clay samples
no desiccator 0.36 ± 0.03 0.66 ± 0.11 14.05 ± 0.15 17.18 ± 0.49 was negligible and we therefore assumed that
SR2T
desiccator 0.17 ± 0.07 0.40 ± 0.03 12.87 ± 0.29 14.10 ± 0.53 the measurement reflected the matric poten-
no desiccator 0.77 ± 0.17 1.10 ± 0.29 19.03 ± 0.02 19.91 ± 0.13 tial. The final water content was measured by
SR2B
desiccator 0.03 ± 0.04 0.00 ± 0.03 18.03 ± 0.74 18.38 ± 0.95 oven drying at 105°C, and matric potential
no desiccator 0.04 ± 0.01 0.0 ± 5 0.00 0.10 ± 0.03 0.37 ± 0.06 was transformed to RH.
Sandy
desiccator 0.02 ± 0.01 0.02 ± 0.00 0.02 ± 0.01 0.01 ± 0.00

Gypsum
no desiccator 0.51 ± 0.03 2.39 ± 1.58 19.75 ± 0.15 20.91 ± 0.10 RESULTS AND DISCUSSION
desiccator 0.04 ± 0.02 1.43 ± 1.71 18.82 ± 0.34 19.75 ± 0.64 Figure 1 presents the gravimetric
water content of the mineral samples as a
the desiccator with drierite, and allowed them to cool down for 3 h. We function of RH. In effect, this diagram demonstrates the water
weighed the soil samples again and returned them to the desiccator and release characteristic of the clay minerals as a function of RH.
reweighed the samples after 2 d. We repeated the procedure with fresh This diagram illustrates the flaw in the assumption that by plac-
samples each time and heated at 70, 105, or 200°C. We used a styro- ing the samples in a desiccator with a RH of 12 or 20%, the clay
foam block as a thermal insulator for weighing to avoid errors in the minerals will lose all their absorbed water and that any further
measurements or damage to the scale (data shown in Table 3). water loss on oven drying is due exclusively to gypsum. The
We repeated the heating process at 50, 70, 105, and 200°C with apparent success in previous work was because Na-bentonite
a new batch of soil samples, weighing the samples before and after was used as a standard (Nelson et al., 1978), which loses almost
oven drying, but this time the samples were left open to the laboratory all of its adsorbed water at 20% RH. Figure 1 shows, however,
atmosphere at 37% relative humidity and 21°C. The mass increase of that Ca- and Mg-saturated montmorillonites both have >10%
the samples was registered every 10 min for the first hour and then at gravimetric water content remaining at a RH of 20%; even
every hour and every 3 and 12 h until constant mass. Constant mass at 12% RH, the water contents are >7%. Figure 1 shows that
not only the mineralogy but also the dominant cation present
in our soils is very important, especially Ca-montmorillonite,
which is common in semiarid ecosystems. Water loss on oven
drying will be a combination of water from the clay as well as
from the gypsum. This could potentially result in large errors
in the estimation of the gypsum content, depending on the
clay percentage in the sample.
The gravimetric water content of gypsum and bassanite
are also presented in Fig. 1, they follow almost straight lines
across the wide humidity range shown. When gypsum is heat-
ed at 105°C, its water of crystallization is removed; subsequent
rehydration in the laboratory results in the formation of bas-
sanite at relative humidities below saturation (100% RH). As
bassanite has only 0.5 of a water molecule per unit cell, there is
a loss of 1.5 water molecules between the gypsum and bassan-
ite phases, representing a 15.68% reduction in mass compared
with gypsum. In our experiments, we found that we could re-
cover the original gypsum from the bassanite phase only by
physically wetting the sample.
Fig. 1. Water content relative to oven dryness for different clay Rehydration curves for four different gypsum reagent
minerals at different relative humidities. The water content for samples, following gypsum removal from the oven at 105°C,
gypsum and bassanite is shown as ×. are presented in Fig. 2. The graph shows that, left in the labora-

406 SSSAJ: Volume 73: Number 2 • March–April 2009


than 1% in all cases, regardless of whether the mineral was quartz, sand, with gypsum compositions ranging from zero to 100%
kaolinite, illite, or Ca- or Mg-saturated montmorillonite. gypsum. Table 2 shows a semiquantitative analysis of the min-
The initial measurement of the Mg-montmorillonite eral composition of the soils in this study determined by x-ray
resulted in a 2% error; this was also observed with the pure and compared with our new method. The results indicate that
(100%) Mg-montmorillonite sample. We observed that pure the quantity of gypsum determined by the x-ray method differs
Mg-montmorillonite did not recover its original mass when from the value determined using our new method. This is not
placed back in the desiccator after oven drying. We attributed unexpected and reasonably typical of semiquantitative analysis
this discrepancy to hysteretic behavior of the clay or, in other using x-ray diffraction.
words, to the fact that the amount of water in the clay at a
fixed RH differs depending on whether the equilibrium mass Considerations for the Determination of Gypsum
at the fixed humidity was reached in a saturating or desatu- Content in Soils
rating environment. It is well known that soils are hysteretic; Data reported in the literature show that, when heated
however, at low water contents we expected this to be neg- at 105°C, the gypsum crystal loses between 13 and 19% of its
ligible. Neglecting the hysteretic behavior in the case of the mass (Artieda et al., 2006) instead of the 20.91% correspond-
Mg-montmorillonite, however, resulted in a measurable error ing to the two water molecules. There are some discrepancies
in the gypsum determination. The RH in the laboratory is in the literature as to when exactly gypsum loses the two water
subject to variations. At the time of this particular experiment molecules; as mentioned above, the Nelson method is based
shown in Fig. 3, the RH of the laboratory was 60%; as a con- on total water loss at 105°C when heated for 24 h, but Lide
sequence, the clays dehydrated through a drying process when (1993) reported that total loss occurs at 163°C. The difference
placed in the desiccator (41% RH) before oven drying, while between the two temperatures can be reconciled if we consider
following removal from the oven the clays rehydrated through that dehydration of gypsum is a kinetic process (Molony and
a wetting process in the desiccator. To overcome this problem Ridge, 1968; Fatu, 2001; Robertson and Bish, 2007). Most of
of initially dehydrating to 41% RH and then rehydrating to the gypsum dehydration studies in the literature used Raman
41% RH after removal from the oven, we placed the sample in infrared spectroscopy (Razouk et al., 1960; Putnis et al., 1990;
a desiccator at a RH of 65% after removal from the oven for Sarma et al., 1998; Prasad et al., 2001; Chio et al., 2004), neu-
6 h to reproduce the laboratory conditions at the beginning tron diffraction data (Atoji and Rundle, 1958), or x-ray dif-
of the experiment and until its mass had increased beyond its fraction (Molony and Ridge, 1968; Cole and Lancucki, 1974),
initial starting mass. We then returned the sample to the des- but to follow the release of the water molecules from the crystal,
iccator with the 41% RH, allowing the sample to dehydrate. they have used fast heating devices (5 or 10 K min−1) (Prasad et
The sample was then allowed to equilibrate for 2 d, after which al., 2001; Chio et al., 2004; Sarma et al., 1998). Consequently
time we found that the mass had returned to a value similar to the temperature at which they found total loss of water is
the initial mass, resulting in an estimation error for the gypsum around 163°C, since fast heating does not allow time for the
content of <1%. water to diffuse through the crystal.
If hysteretic behavior is suspected in the samples and the In addition, the RH is also important. During oven drying,
RH of the laboratory at the time of the experiment is higher the RH is generally close to 100% but when the RH is altered,
than the one chosen for the desiccator, we recommend an ad- dehydration is affected. Robertson and Bish (2007) reported
ditional step in which the samples, after removal from the oven, total dehydration of gypsum after 5.6 h when the sample was
are equilibrated to a RH similar to the value of the atmosphere heated at 85°C in 17.6% RH, but it took 50 h to lose the
before being placed in the desiccator at 41% RH. If the RH water at the same temperature with the RH at 47%. This may
of the laboratory is lower than that of the desiccator (41%), have important environmental implications for mineral phase
no additional step is needed since the samples will equilibrate changes in natural samples in arid regions where soil surface
along the same portion of the wetting curve before and after temperatures may reach in excess of 70°C with RH values as
oven drying. low as 10 to 15%.
The soil samples used in this study were analyzed by The soils literature provides a spectrum of possible water
preparing a saturation paste and the extract was analyzed for loss percentages, besides the theoretical 20.91%, when gypsum
basic chemical composition. Table 1 shows the electrical con- is oven dried. As a consequence, the gypsum literature has ad-
ductivity values (ECe) for the saturation extracts; the values opted the term recovery factor, defined in Burns et al. (2002),
are consistent with the presence of gypsum. The ECe values to account for the discrepancy between the experimental water
were between 2 and 2.5 dS m−1, indicating that the water was loss at 105°C and the total theoretical water loss (20.91%). In
saturated with gypsum but also indicating that no significant reality, the recovery factor is a practical solution to overcome
amount of another, more soluble salt was present. No signifi- the fact that 105°C is an arbitrary temperature and we do not
cant amount of total C, organic or inorganic, was detected. know exactly how much gypsum water is lost at this tempera-
The pH values were <8, consistent with the very small amounts ture, especially because initial RH values change from labo-
of bicarbonates identified in the saturation extracts. X-ray ratory to laboratory. The fact that different researchers have
analysis of the mineral phase confirmed the presence of gyp- found such different numbers for the recovery factor, however,
sum as the main component of all the soils; in the top layers also suggests inconsistencies in laboratory procedures.
(MT, SR1T, and SR2T) quartz was found in different quanti- In our view, this inconsistency is due to the handling of
ties, and small amounts of calcite were present in the SR1 soil. the samples after they are removed from the oven and the use
The soil samples were typically mixtures of gypsum and quartz of desiccators. As shown in Fig. 2, the anhydrite, once removed

408 SSSAJ: Volume 73: Number 2 • March–April 2009


Table 4. Percentage of water, related to the oven-dry state,
remaining in clays at 12% relative humidity (RH). Loss of
water while performing the Nelson et al. (1978) method
translates to erroneous gypsum content.
Water content at Error in gypsum content
Clay
12% RH using Nelson method
———————— % ————————
Kaolinite 0.44 ± 0.04 2.28 ± 0.23
Illite 0.7 ± 0.26 3.61 ± 1.36
Na-bentonite 0.84 ± 0.40 4.33 ± 2.04
Sand 0.12 ± 0.09 0.6 ± 0.46
Ca-montmorillonite 6.24 ± 0.38 32.13 ± 1.98
Mg-montmorillonite 7.3 ± 1.07 37.61 ± 5.52

ones calculated with our method but consistently higher. The


mineralogy for these soils was predominantly quartz sand and
silt, and consequently not much water was retained; however,
we observe that the change in mass for the sandy soil was sig-
nificant enough to estimate a 4% gypsum content using the
Nelson method. The chemical analysis in Table 1 confirms the
absence of gypsum in the sample. As a result, the new method
exploiting the bassanite stability is an appreciable improvement

Fig. 5. (A) Scanning electron micrograph of the crystalline gypsum


in the Moab soil using a back-scatter detector, and (B) detail of the
gypsum crystals.

water is retained at the clay surfaces. At 12% RH, the humid-


ity in the desiccator with drierite resulted in the Ca- and Mg-
montmorillonite having 10% water content. When multiplied
by a recovery factor, this translates into an error of 38 to 52%,
depending on the recovery factor used (19–14%).
The results from the Nelson method are also presented in
Table 2 for the soils and can be compared to our new bassanite
method. The values using the Nelson method are similar to the

Fig. 7. (A) Differences between the real gypsum content and the
gypsum content calculated using the Nelson et al. (1978) method and
the bassanite method. The samples were mixtures of gypsum reagent
and different clay minerals. All values are related to oven-dry weight;
and (B) the effect of the saturating clay cation on the differences
Fig. 6. X-ray diffractogram for the Moab soil, the 12- to 20-cm SR1 between gypsum contents calculated using the Nelson method and
soil sample, and the 1- to 10-cm SR2 soil sample. the real gypsum content.

410 SSSAJ: Volume 73: Number 2 • March–April 2009


over prior methods and may help to reduce the likelihood of Eswaran, H., and G. Zi-Tong. 1991. Properties, genesis, classification, and
distribution of soils with gypsum. p. 89–119. In W.D. Nettleton (ed.)
the misclassification of soils. Occurrence, characteristics, and genesis of carbonate, gypsum, and silica
accumulations in soils. SSSA Spec. Publ. 26. SSSA, Madison, WI.
CONCLUSIONS FAO. 1990. Management of gypsiferous soils. Soils Bull. 62. FAO, Rome.
We have developed an improved, simple, and accurate Fatu, D. 2001. Kinetics of gypsum dehydration. J. Therm. Anal. Calorim.
method to determine gypsum content in soils. This method 65:213–220.
Goldberg, S.R., and R.A. Glaubig. 1987. Effect of saturating cation, pH,
considers the differences in water content between gypsum and aluminum and iron oxide on the flocculation of kaolinite and
and bassanite after the samples have been heated in an oven at montmorillonite. Clays Clay Miner. 35:220–227.
105°C. To achieve equilibrium with bassanite and to keep the Gove, P.B. (ed.). 1993. Webster’s third new international dictionary, unabridged.
clay water content constant, we equilibrated the soil samples Merriam-Webster Inc., Springfield, MA.
at the same RH before and after oven drying the samples. The Greenspan, L. 1977. Humidity fixed points of binary saturated aqueous
solutions. J. Res. Natl. Bur. Stand., Sect. A 81:89–96.
use of the desiccator with a fixed RH after oven drying allowed Guerra, J., R.M. Ros, M.J. Cano, and M. Casares. 1995. Gypsiferous outcrops
us to accomplish two objectives: to reach equilibrium with bas- in SE Spain, refuges of rare, vulnerable and endangered bryophytes and
sanite, and to return the clays to the same water content as lichens. Cryptogam. Bryol. Lichenol. 16:125–135.
before oven drying. Given Mg-saturated smectitic soils or soils Khan, S.U., and G.R. Webster. 1968. Determination of gypsum in solonetzic
exhibiting possible hysteresis, we recommend a more involved soils by an x-ray technique. Analyst 93:400–402.
Lide, D.R. (ed.). 1993. Handbook of chemistry and physics. 74th ed. CRC
method requiring two equilibration steps after oven drying. Press, Boca Raton, Florida.
This extra step is to ensure that the sample always hydrates fol- Loeppert, R.H., and D.L. Suarez. 1996. Carbonate and gypsum. p. 437–474.
lowing the same portion of the wetting curve. By standardizing In D.L. Sparks (ed.) Methods of soil analysis. Part 3. Chemical methods.
the initial and final weights of our samples, we minimized the SSSA Book Ser. 5. SSSA, Madison, WI.
major sources of error found with other weight loss methods. Molony, B., and M.J. Ridge. 1968. Kinetics of the dehydration of calcium
sulphate dihydrate in vacuo. Aust. J. Chem. 21:1063–1065.
Our new method, in the absence of hydrated salts, consistently Mota, J.F., A.J. Sola, E.D. Dana, and M.L. Jiménez-Sánchez. 2003. Plant
estimated the actual quantity of gypsum in the sample to better succession in abandoned gypsum quarries in SE Spain. Phytocoenologia
than 1% regardless of the clay mineralogy present. 33:13–28.
Nelson, R.E. 1982. Carbonate and gypsum. p. 181–197. In A.L. Page et al.
(ed.) Methods of soil analysis. Part 2. Chemical and microbiological
ACKNOWLEDGMENTS properties. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
We wish to express our thanks to colleagues at Stanford University, Nelson, R.E., L.C. Klameth, and W.D. Nettleton. 1978. Determining soil
Scott Fendorf and Rosemary Knight, for their help and support in gypsum content and expressing properties of gypsiferous soils. Soil Sci.
conducting this research, and to Scott Jones at Utah State Univ. J. Soc. Am. J. 42:659–661.
Herrero thanks the Spanish Government for the grants AGL2006- Prasad, P.S.R., A. Pradhan, and T.N. Gowd. 2001. In situ micro-Raman
01283/AGR and PR2007/0453. investigation of dehydration mechanism in natural gypsum. Curr. Sci.
80:1203–1207.
Putnis, A., B. Winkler, and L. Fernández-Díaz. 1990. In situ IR spectroscopy
REFERENCES and thermogravimetric study of the dehydration of gypsum. Mineral.
Artieda, O., J. Herrero, and P.J. Drohan. 2006. A refinement of the differential Mag. 54:123–128.
water loss method for gypsum determination in soils. Soil Sci. Soc. Am. Razouk, R.I., A.Sh. Salem, and R.Sh. Mikhail. 1960. The sorption of water
J. 70:1932–1935. vapor on dehydrated gypsum. J. Phys. Chem. 64:1350–1355.
Atoji, M., and R.E. Rundle. 1958. Neutron diffraction study of gypsum, Robertson, K., and D. Bish. 2007. The dehydration kinetics of gypsum: The
CaSO4⋅2H2O. J. Chem. Phys. 29:1306–1311. effect of relative humidity on its stability and implications in the Martian
Belnap, J., J.H. Kaltenecker, R. Rosentreter, J. Williams, S. Leonard, and D. environment. p. 1432. In Lunar and Planetary Sci. Conf., 38th, League City,
Eldridge. 2001. Biological soil crust: Ecology and management. Bur. TX. 12–16 Mar. 2007. LPI Contrib. 1338. Lunar Planet. Inst., Houston.
Land Manage. Tech. Ref. 1730-2. Natl. Sci. Technol. Ctr., Denver, CO. Sarma, L.P., P.S.R. Prasad, and N. Ravikumar. 1998. Raman spectroscopic
Burns, D.T., K. Danzer, and A. Townshend. 2002. Use of the terms “recovery” study of phase transitions in natural gypsum. J. Raman Spectrosc.
and “apparent recovery” in analytical procedures. Pure Appl. Chem. 29:851–856.
74:2201–2205. Soil Survey Staff. 2008. Web soil survey. Available at websoilsurvey.nrcs.usda.gov/
Chio, C.H., S.K. Sharma, and D.W. Muenow. 2004. Micro-Raman studies of (accessed 25 June 2008; verified 29 Nov. 2008). NRCS, Washington, DC.
gypsum in the temperature range between 9K and 373K. Am. Mineral. U.S. Salinity Laboratory Staff. 1954. Diagnosis and improvement of saline and
89:390–395. alkali soils. Agric. Handbk. 60. U.S. Gov. Print. Office, Washington, DC.
Cole, W.F., and C.J. Lancucki. 1974. A refinement of the crystal structure van Reeuwijk, L.P. (ed.). 1987. Procedures for soil analysis. Int. Soil Ref. Inf.
of gypsum CaSO4⋅2H2O. Acta Crystallogr. Sect. B: Struct. Crystallogr. Ctr., Wageningen, the Netherlands.
Cryst. Chem. 30:921–929. Verheye, W.H., and T.G. Boyadgiev. 1997. Evaluating the land use potential of
Elprince, A.M., and A.M. Turjoman. 1983. Infrared dehydration method for gypsiferous soils from field pedogenic characteristics. Soil Use Manage.
determining gypsum content of soils. Soil Sci. Soc. Am. J. 47:1089–1091. 13:97–103.

SSSAJ: Volume 73: Number 2 • March–April 2009 411


Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14350661, 1994, 6, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800060006x by University of Queensland Library, Wiley Online Library on [01/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BERIGARI & AL-ANY: GYPSUM DETERMINATION BY CONVERSION TO SODIUM SULFATE 1625

Table 1. Selected depths and properties of the Typic Gypsiorthids.


Particle-size analysis
Sample pH Electrical Organic Equivalent
no. Depth Sand sat Clay (water) conductivity C CaCO3 Gypsum
cm % dSm-' eke-
6 6
1

1.1 0-15 26 40 34 7.7 2.20 6.3 222 52.4


1.2 15-45 33 37 30 7.9 2.21 1.3 72.5 698
2.1 0-15 34 30 36 7.8 2.20 4.8 75.0 391
2.2 15-60 36 38 26 7.6 2.32 2.0 72.5 124
3.1 0-30 29 43 28 7.7 2.42 7.9 209 188
3.2 30-45 29 40 31 7.8 2.36 6.0 195 341
3.3 45-60 27 29 44 7.7 2.60 5.0 172 492

collected from the depths specified in Table 1. The samples (Table 2) was thoroughly mixed with 1.8-g samples of a
were air dried, ground, and passed through a 2-mm sieve. nongypsic soil in triplicate to give 10% soluble salt or 2.5%
The <2-mm fraction was ground further to pass through a of each sulfate salt as an extreme case. The soil-salt mixture
0.5-mm sieve for chemical analysis. The soil analysis data was shaken with 25 mL of 50% ethanol for 1.0 h on a
presented in Table 1 were determined according to convention- horizontal shaker. The supernatant solution was separated by
ally accepted methods (Page et al., 1982). Particle-size analysis centrifugation as described above and analyzed for SO^ turbi-
was obtained by the standard pipette method after removal of dimetrically. This provided a test of the extent of soluble
organic matter, gypsum, and carbonates (Day, 1965). sulfate removal from the soil samples by 50% ethanol compared
with that of the same sulfate salts in 50% ethanol but without
Proposed Method the soil sample.
Triplicate 2.0-g soil samples in 50-mL polycarbonate centri- Comparative Method
fuge tubes were washed once with 25-mL portions of 50%
ethanol to remove soluble sulfates while reducing the solubility Soil samples were extracted with water at different soil/
of gypsum..The bulk solution was centrifuged and the sediment water ratios (Nelson, 1982) and gypsum content was obtained
was suspended in 25 mL of the 0.50 M NaaCOs solution by by the Bower-Huss (1948) method and from SO4~ analysis
a 30-s Bonification to disturb formation of stable CaCO3(s) by turbidimetry (Greenberg et al., 1985).
coating on gypsum, thus enhancing its stoichiometric conver-
sion to water-soluble Na2SO4. The supernatant solution was Statistics
centrifuged at 1000 X g and 25° C, then filtered quantitatively
through Whatman no. 42 filter paper. This extraction procedure For statistical comparisons of the methods, the coefficient
was repeated until a negative test for sulfate with BaCk crystals of variation (CV) and analysis of variance of the means were
was obtained. Sulfate in the aqueous extract was measured by determined by Tukey's honestly significant difference tests
standard BaSO* gravimetry and turbidimetry after the necessary to compare the methods within each soil collection. Linear
dilutions and acidification were made to prevent precipitation regression analysis was used to depict the graphical relations
of BaCO3 and Ba3(PO4)2 (Greenberg et al., 1985). Also, the between the methods.
correction for sample color and turbidity was applied for
turbidimetric SO4~ measurement as described by Greenberg RESULTS AND DISCUSSION
et al. (1985). Gypsum content in the soil samples was obtained
from SOl~ analysis assuming that all the SOl~ in the extract The selected soils were classified as Typic Gypsior-
came from soil gypsum, CaSO4-2H2O: thids, which are the most common soils of the experiment
station. The soil samples were selected to cover a wide
gypsum (g)/soil (kg) = [SO?- (g)/soil (kg) (172/96.1)] [3]
range in gypsum content (52.4-698 g kg"1), CaC(>3
Known Addition of Gypsum equivalent (72.5-222 g kg"1), and other properties as
shown in Table 1.
The accuracy of the proposed method of gypsum extraction Soil samples were washed once with 50% ethanol to
from gypsic soils was determined by adding known quantities remove the soluble sulfates of common cations of arid-
(100 and 200 g gypsum kg~' soil) of a reference-grade gypsum zone soils (Na, Mg, K, and NRt) and possibly the
(100 ± 1 % purity) to the samples, then using the proposed organically bound SOij" prior to gypsum extraction. This
method of extraction and turbidimetric SOi~ analysis to com-
pute the relative quantity of gypsum recovered. was done to minimize the positive error, i.e., overestima-
tion of SOl" from nongypsic sources, and meanwhile
Correction Measures
Table 2. Solubilities of sulfate salts of common soil cations in 50%
Soils may contain nongypsic soluble sulfates of Mg, Na, ethanol, determined in our laboratory at 25°C.
K, and NYU but amounts seldom exceed 2% (Nelson, 1982). Single Equimolar Saturated
These soluble sulfates cause positive errors in gypsum determi- Salt salt additiont solution:):
nation from SO4~ analysis unless a correction is applied for « T -1
such SO4~ or it is removed selectively prior to gypsum extrac-
tion. (NH&SO, 95.5 10.0 40.6
Na2SO4 97.4 10.2 28.4
The solubility of each sulfate of the common soil cations K2S04 1.6 3.0 3.6
and an equimolar mixture (taking 10.0 g (NH4)2SO4 L~' solvent MgS04-6H2O 12.8 2 7.4 1.6
as the basis for calculations) in 50% ethanol was determined CaSO4-2H2O 1.0 x 10- 1.7 x 10-2 2.5 x 10-2
at 25° C. A 0.200-g soluble sulfate mixture, consisting of equal t Addition based on 10 g (NH^SO, L-> of 50% ethanol.
weights (50 mg) each of NH/, Na + , K + , and Mg2+ sulfate t Solution saturated with cosolutes simultaneously.
14350661, 1994, 6, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800060006x by University of Queensland Library, Wiley Online Library on [01/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1626 SOIL SCI. SOC. AM. J., VOL. 58, NOVEMBER-DECEMBER 1994

Table 3. Soil gypsum extraction by the proposed method and measurement by gravimetric (Method A) and turbidimetric SO2" (Method
B) analyses vs. water extraction and measurement by turbidimetric SOi" (Method C) and the Bower-Huss conductometric (Method
D) analyses. Values represent averages of three replications.
Sample no. Method A CV Method B CV Method C CV Method D CV HSDt
1 1
gkg- % gkg- % gkg- 1
% gkg- 1
% gkg- 1
1.1 52.4 2.4 47.9 2.9 45.0* 5.9 43.0* 5.9 5.4
1.2 698 1.5 705 1.6 557» 3.1 593* 3.7 42.0
2.1 391 5.1 385 2.0 355 4.0 295* 4.7 38.2
2.2 124 1.6 116 1.1 91.7* 6.6 105* 11.0 17.3
3.1 188 2.9 184 3.8 144* 4.8 139* 6.1 18.5
3.2 341 1.3 313 2.4 213* 8.2 217* 5.2 29.6
3.3 492 3.7 477 1.4 358* 4.0 326 5.3 56.4
* Significant (P = 0.05) compared with the standard method (A).
t Tukey's honestly significant difference (P = 0.05) to compare the results of four methods of gypsum determination within each soil sample.

greatly reduce the solubility of gypsum (Matushevskiy, samples (Table 3). These differences are vividly seen in
1968) thus minimizing the negative error resulting from the scattergrams where the best-fit regression lines devi-
its loss with the solvent. The efficiency of this solvent ate from the 1:1 line as the gypsum content of soil
was verified by 100% recovery of 2.5% additions of samples increases (Fig. Ib and Ic). On the other hand,
each of Na, Mg, K, and NUt sulfate as mixtures with the agreement between the best-fit regression line and
nongypsic soil samples based on turbidimetric SO2." anal- the 1:1 line for Methods C vs. D is very good (Fig.
ysis. Also, the data of Table 2 show that gypsum solubility Id), indicating that the extraction procedure was the
in 50% ethanol is very low (1.0 x 10~2 to 2.5 x 10~2 g major cause for the deviations of Methods C and D from
L"1) depending on the composition of solution. Thus, the Methods A and B. Also, it is worth noting that Methods
negative error in gypsum determination in soils caused by C and D displayed lower precision given by CV ranges
its dissolution in 25 mL of this solvent per 2-g soil (3.1-8.2 and 3.7-11.0%, respectively; Table 3).
sample is not likely to exceed 0.62 mg or 0.031 %, which In general, the gypsum contents were higher when
is negligible. Moreover, treatment of soil samples with extracted by the proposed method (N^COs solution)
50% ethanol can retard biological transformation of whether SOl~ was measured by the standard gravimetric
SO?T, thus eliminating the error that may arise from this (A) or the turbidimetric (B) method. Moreover, the soil
source. gypsum extracted by the proposed method and by distilled
The Na2CO3 solution was selected to convert soil water (Nelson, 1982) was determined from SO2," by the
gypsum to soluble Na2SO4 and CaCC>3(s) for several same turbidimetric method (B and C) yet differences
reasons: (i) it and its reaction product NaaSC^ are very exist between the two sets of data (B vs. C, Table 3),
soluble in water, (ii) it is relatively inexpensive, and which should be ascribed largely to the variation between
(iii) soil treatment with its excess will saturate the ex- the two methods of gypsum extraction. This means that
change sites with Na + . This will allow good dispersion distilled water, which has been used routinely, is a weak
of the soil suspension with Bonification, which is effective extractant for gypsum (15 mM) where a very wide water/
for rapid dissolution of gypsum since the Ca2+ displaced soil ratio with numerous extractions were required for
from the exchanger and that derived from gypsum will improved gypsum extraction from soils high in gypsum
precipitate as CaCO3(s), leaving behind SO2." in solution
that can be separated and analyzed. 800 800
Ten y = 7.28 +1.00 X y=­6.03*0.79X
The values of gypsum content of the soil samples "£,600 r = 0.98 600 r = 0.98
obtained with the two methods of extraction and three CO
methods of determinations are listed in Table 3. Method S 400 400
A is used for comparison because it is the standard .£
a*
method for SO2." analysis. The agreement between the 2 200 200
values obtained with Methods A and B is excellent (r = 0
0.98). There were no significant differences at P = 0.05 I 200 400 600, 800 200 400 600 800
between the two sets of data by analysis of variance and Method A gkg" Method A gkg"'
also as shown by the best-fit regression line, which "iai
800 £00
y =­6.77* 1.00 X
coincides with the 1:1 line (Fig. la). Methods A and B "o. 600
' CTl
£600 r = 0.99
have about the same precision measured by CV range o
.(1.3-5.1 and 1.1-3.8%, respectively). The agreement •a A 00
o
between Methods A and B used here is closer than that
found in Chesnin and Yien (1950) for SO2." extraction 1200 1 200
from a sandy soil from Nebraska. Since the turbidimetric 0 0
method (B) of SO?r analysis is simple, rapid, and accu- 200 400 600 800 200 400 600 800
rate, we recommend it for gypsum determination in soils. Method A gkg'' Method C gkg"
The differences between the values of gypsum content Fig. 1. Comparison of the methods used for gypsum determination
obtained wiJi Methods A and C were significant (P = of the soil samples: (a) turbidimetric B vs. gravimetric A, (b)
0.05) for six out of seven soil samples, whereas with turbidimetric C vs. gravimetric A, (c) conductometric D vs. gravi-
Method D the differences were significant for all the metric A, and (d) conductometric D vs. turbidimetric C.
14350661, 1994, 6, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800060006x by University of Queensland Library, Wiley Online Library on [01/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BERIGARI & AL-ANY: GYPSUM DETERMINATION BY CONVERSION TO SODIUM SULFATE 1627

Table 4. Recovery of gypsum added to 2-g soil samples as extracted by the proposed method and obtained from turbidimetric SO3 ~ analysis.
Values represent averages of triplicates, unless otherwise specified.
Gypsum Gypsum
Present Present
Sample + G,t Found CV Recovery^ + G2t Found CV Recovery:):
———— g k g - > - 1
———— gkg- -
1.1 148 144 1.3 97.3 248 239 1.0 96.4
1.2 805 798 0.5 99.1 905 890 0.6 98.3
2.1 485 478 0.9 98.6 585 572 0.8 97.8
2.2 216 211 1.1 97.7 316 306 0.7 96.8
3.1 284 282 0.8 99.3 384 377 2.3 98.2
3.2 413 406 2.9 98.3 513 508 0.3 99.0
3.3 577 579 3.0 100 677 666 1.2 98.4
Mean (n = 7) 1.5 98.6 1.0 97.8
t Gi and Gi represent 100 and 200 g additions of gypsum kg"1 sofl sample, respectively.
t Percentage recovery is based on total gypsum found divided by [gypsum present (Table 3, Column 4) plus that added] x 100.

contents. These operations are laborious and time con- its determination from turbidimetric 864 analysis ap-
suming, as have been experienced by other investigators pears to be simple, rapid, accurate, and very efficient
(Lagerwerffetal., 1965; Nelson, 1982). relative to the methods where water is used as the ex-
Table 4 shows the data of gypsum recovery from two tractant.
levels of standard additions to various soil samples using
the proposed method of Na2COs extraction with turbidi- ACKNOWLEDGMENTS
metric SOi~ analysis. The recovery of gypsum ranged
between 96.4 and 100%, with an overall mean of 98.2%, We wish to thank Lila H. Ibrahim and Ahlam H. Al-Khafaji
which is very good considering the speed and simplicity for certain phases of laboratory work and F.J. Jack for the
of the extraction and measurement steps and the complex- graphical work.
ity of soil systems.
The efficiency of the proposed method of gypsum
extraction relative to that obtained by water is shown in
Table 5. With the proposed method, two equilibrations
at a solution/soil ratio of 25:2 were sufficient to extract
gypsum completely from four soil samples selected to
bracket the range of gypsum content (Tables 1 and 3).
However, extraction with water required a wider water/
soil ratio, 250:2, and a greater number of equilibrations
even for low levels of gypsum, such as 5% in the case
of Sample no. 1.1 (Table 3). The efficiency of gypsum
extraction for the proposed method relative to that by
water as defined by Eq. [4] ranged from 9 to 28.7 for
the first equilibration, increasing with the increase in
gypsum content. The efficiency of gypsum extraction
varied depending on the initial gypsum content of the
soil sample and the sequence of extraction. Mean values
of 25.0 to 30.7 with an overall average of 27.8 were
obtained with the proposed method (Table 5).
The proposed method of soil gypsum extraction and

Table 5. Number of equilibrations needed and efficiency of the


proposed method of gypsum extraction (25:2) relative to water
extraction (250:2) based on turbidimetric SO|~ analysis.
Extraction efficiencyt of
Equilibrations needed proposed method relative to
to extract all gypsum water extraction method
Sample Proposed Water 1st 2nd
no. method method extraction extraction Mean
1.1 2 3 9.0 52.4 30.7
1.2 2 4 28.7 30.0 29.4
2.1 2 4 20.8 31.6 26.2
2.2 2 3 9.5 40.5 25.0
Mean 17.0 38.6 27.8
t Efficiency =
Na2COrextracted gypsum , water/soil sample ratio
[4]
water-extracted gypsum Na2CO5 solution/soil sample ratio

You might also like